@prefix airo: . @prefix dcterms: . @prefix dpv: . @prefix eli: . @prefix mailo: . @prefix owl: . @prefix rdfs: . @prefix skos: . @prefix xsd: . a owl:Ontology ; dcterms:abstract "MAILO (Medical AI Legal Ontology) is a machine-executable knowledge graph encoding EU regulatory obligations for medical artificial intelligence systems. It formalises eight EU legislative instruments -- the GDPR, MDR, AI Act, Product Liability Directive, EHDS, Data Governance Act, Data Act, and Trade Secrets Directive -- as OWL 2 classes and SHACL shape constraints, grounded in three CJEU judgments on automated decision-making (Ligue des droits humains, SCHUFA Holding, and Dun & Bradstreet Austria). MAILO functions as a deterministic, auditable compliance-verification layer: SHACL NodeShapes validate instance data describing a medical AI system against the encoded legal constraints and produce structured, traceable conformance reports."@en ; dcterms:created "2026-03-27"^^xsd:date ; dcterms:description """\r Ontology capturing the legal landscape governing AI-driven automated\r decision-making in medical contexts. Covers CJEU precedents on GDPR\r Art.22, UPC/EPO patent litigation in MedTech, the EU AI Act high-risk\r regime, MDR 2017/745, PLD 2024/2853, EHDS 2025/327, DGA 2022/868,\r and Data Act 2023/2854 — with full precedent chains, SHACL runtime\r constraints, and compliance obligation mappings.\r Imports: W3C DPV v2.0 (data privacy vocabulary), AIRO (AI risk ontology),\r ELI (European Legislation Identifier), ECLI (European Case Law Identifier).\r """@en ; dcterms:language "en" ; dcterms:modified "2026-08-12"^^xsd:date ; dcterms:title "Medical AI Legal Ontology (MAILO)"@en ; owl:imports , , . mailo: a owl:Ontology ; dcterms:created "2026-04-30"^^xsd:date ; dcterms:creator "Victor Kao" ; dcterms:description "OWL ontology mapping EU regulatory frameworks (GDPR, AI Act, MDR, PLD, EHDS, DGA, Data Act) to medical AI automated decision-making, CJEU case law, UPC patent litigation, and compliance obligations." ; dcterms:license ; dcterms:modified "2026-09-17"^^xsd:date ; dcterms:publisher "KU Leuven, Faculty of Engineering Science" ; dcterms:title "Medical AI Legal Ontology (MAILO)" ; rdfs:comment "Version 5.0.2 (patch): corrected the ECLI recorded for C-203/22 (Dun & Bradstreet Austria) from the erroneous ECLI:EU:C:2025:145 to the correct ECLI:EU:C:2025:117."@en, "Version 5.1.0: corrective release consolidating four rounds of work, none previously deployed. (1) All nine CJEU paragraph citations were re-checked and all nine corrected; two substantive mischaracterisations were removed (the health-status exclusion is categorical under PNR Directive Art.6(4), not curable by a GDPR Art.9(2) basis; the strict-necessity test does not entail invalidity, the Court having upheld the Directive by Charter-conforming interpretation). (2) The patent, EPO and UPC instances were verified: T 1191/19 re-dated and its holding rewritten, T 0161/18 corrected to a refused application, the citesPrecedent graph repaired, litigation statuses refreshed. (3) Legislative citations were audited: all four EHDS articles carried proposal numbering and were renumbered, Art.73 was added, PLD Art.11 was corrected from the causation presumption to exemption from liability, DGA Art.31 was corrected to Art.29, and AI Act Art.50 was relabelled. (4) MAILO's boundaries were made explicit: every case now declares an InclusionBasis and every principle an EncodingBasis, separating what a court held from what the ontology designer inferred or operationalised. Correcting the hierarchy that let the boundary errors pass, AutomatedDecisionSystem is no longer a subclass of MedicalAISystem — the entailment ran backwards, making every automated decision system medical. Parts IV and V of the SHACL shapes now enforce citation integrity and boundary declaration on the graph itself, and each shape has been tested against the error it was written for."@en, """Version 5.2.0 — second corrective release, 10 September 2026. Continues the audit begun in v5.1.0, which covered the CJEU paragraph citations, the patent and EPO instances, an initial set of legislative citations, and the inclusion and encoding boundaries. This release audits everything that remained: the cross-framework tensions, the principles and cases outside the trilogy, the GDPR, AI Act, MDR, Data Act and Trade Secrets Directive provisions, and the SHACL shapes. Coverage rises from 48 to 71 articles. The recurring defect is not miscitation but misattribution of normative character. Two provisions had their effect inverted and two pairs were transposed. CROSS-FRAMEWORK TENSIONS. None of the ten was sound. Four cited the wrong provision: DGA neutrality is Art.12(1)(a) not Art.10; Data Act trade-secret safeguards are Arts.4(6)-(8) and 5(9)-(11) not Art.15; the EHDS opt-out tension still carried proposal numbering; and the PLD tension attributed the defect presumption to Art.11 rather than Art.10(4). Two cited C-203/22 at paras 60-70 instead of 59-62. Seven instances used conflictType with hasResolutionStatus while three used conflictsWith with involvesFramework, so any cross-tension query silently omitted a third of the set; the model is harmonised. Tension_BlackBox_PatentProtection carries a warning that it conflates patent protection with secrecy — a patent requires disclosure, as T 161/18 and T 1191/19 both demonstrate. GDPR. Art.22(1) had been recorded as a flat prohibition, pre-committing to one side of a contested question C-634/21 did not settle, and omitting profiling. Art.22(3) had been described as applying whenever a derogation applies; it attaches only to points (a) and (c) of Art.22(2), which matters for medical AI running on a national legal basis. Art.22(4) had been called 'enhanced protection' for special category data when it is a prohibition admitting only Art.9(2)(a) or (g) — consequential because health data sits squarely in Art.9(1). Art.15(1)(h) had omitted the existence limb and the cross-reference scoping it to Art.22(1) and (4). AI ACT. Art.86's right to explanation is confined to Annex III systems. Medical devices reach high-risk status by the Annex I route under Art.6(1), so Art.86 does not apply to a diagnostic or imaging AI device, and neither does the Art.26(11) duty to inform affected persons; for that population GDPR Arts.15(1)(h) and 22 carry the explanation duty alone. A HighRiskRoute vocabulary encodes the distinction, and MDR Annex VIII Rule 11 — the classification rule that produces it, previously absent — is linked through triggersRoute. Art.13 addresses DEPLOYERS, not affected persons, which is the pivot of the trade-secret argument. Art.10(3) had dropped the qualifier 'to the best extent possible'. OTHER INSTRUMENTS. TSD Art.5 had been read as disapplying secrecy where information rights are exercised; Art.5(a) is the freedom of expression of Charter Art.11, and the provision that preserves regulatory disclosure duties is Art.1(2)(b), now added along with Art.3 on reverse engineering. PLD Arts.4 and 7 were transposed and Art.5, the strict-liability rule itself, was missing. Data Act Art.3 is a design duty and Art.4 the access right, with different application dates; Art.17 is the request procedure, not public-emergency sharing. Six EHDS provisions were added, of which Art.53(1)(e)(ii) matters most: it expressly permits secondary use for training, testing and evaluating algorithms in medical devices and AI systems, the legal pathway on which medical-AI training under the EHDS rests. EHDS Art.52, governing IP and trade secrets in secondary use, was also absent. CASES AND SHAPES. G 1/19 is a COMVIK inventive-step ruling, and the Enlarged Board held that whether the simulated system is technical is NOT decisive; all three EPO instances were attributed to EPC Art.52 and are re-anchored to Arts.56, 83 and 84. C-307/22 (FT v DW) involves no AI. Four cases are rescoped, leaving two of sixteen with a verified medical-AI technology claim. Correcting the class hierarchy in v5.1.0 inverted what shape target classes mean, since targeting AutomatedDecisionSystem was narrower than MedicalAISystem before and is broader after; three C-817/19 shapes had drifted and all five now target MedicalAISystem. MedicalAIComplianceShape records that its C-817/19 constraints are transposed rather than held, and that its conjunctive structure is a modelling choice: conformance evidences alignment with MAILO's model, not compliance with law. NOT VERIFIED. Three CJEU cases lack ECLIs, deliberately not supplied from memory. Three UPC instances retain unverified substantive assertions, flagged on the instances. No finding in either corrective release has been reviewed by a second EU data-protection specialist."""@en, "Version 5.2.1 (patch): factual consistency repair, no schema change. (1) The AI Act had been represented by two unconnected identifiers, mailo:AIAct carrying the article-level assertions and mailo:AIAct_2024_1689 carrying the temporal versions, so no query could reach a regulation version from an article; all assertions were migrated to mailo:AIAct and the second identifier was deprecated. (2) The AI Act timeline was still triggered by the provisional agreement of 7 May 2026 and described a conditional standards-availability mechanism, while another part of the graph already recorded the adopted Regulation (EU) 2026/1744; the operative triggers now point to the adopted Regulation and the provisional agreement is retained as history alone. (3) A version starting 2 August 2028 was flagged as the current version; the current version is now the post-Omnibus text in force from 27 July 2026, which records the amended text rather than the applicability of the high-risk obligations. (4) The Article 27 FRIA date was corrected from 2 August 2027 to 2 December 2027 and its scope limited to Article 6(2) deployers. (5) Chapters I and II were dated to their application date of 2 February 2025 rather than the entry into force of 1 August 2024. (6) The original Annex III and Annex I high-risk dates, conflated in a single version, were separated to 2 August 2026 and 2 August 2027. (7) dcterms:date was corrected on the AI Act and the PLD and added to the MDR, so that every EU instrument records the date of the act. (8) The same identifier split was found in the GDPR representation and repaired the same way. No SHACL shape, no CJEU assertion and no class or property definition was changed."@en, """Version 5.2.2 (patch): one substantive post-Omnibus correction, the modelling items v5.2.1 deferred, and a set of provenance and metadata repairs. UNLIKE v5.2.1, THIS RELEASE DOES ADD A CLASS AND NEW INDIVIDUALS: one class, mailo:EPOProceeding; two HighRiskRoute individuals for the Annex I Section A and Section B split; and five AI Act article individuals (Arts.4a, 27, 43, 53, 55). No object property and no datatype property was added: the counts stand at 43 and 39, unchanged from v5.2.1. HEADLINE. Article 1(9) of Regulation (EU) 2026/1744 deleted AI Act Art.10(5) and Article 1(6) inserted a new Art.4a. This is not a relocation: the legal basis for processing special categories of personal data for bias detection and correction was extracted from the high-risk training-data article into Chapter I, and its personal scope was broadened from providers of high-risk systems to deployers of high-risk systems and to providers and deployers of other AI systems and AI models, on five cumulative conditions and with no obligation to perform bias detection. Art.4a is encoded as an article node and Art.10 records the deletion, so any constraint anchored to Art.10(5) is visibly without a source. As an insertion effective 27 July 2026 it cannot operate retroactively. DEFERRED ITEMS NOW CLOSED. (a) Annex I is no longer treated as a single set: AnnexISectionARoute and AnnexISectionBRoute are declared, the MDR and IVDR are recorded as Section A legislation, and MDR Annex VIII Rule 11 now triggers the Section A route. The distinction is load-bearing for timing, since Chapter III Sections 1-3 apply to Section A from 2 August 2028, a one-year deferral rather than the two years granted to Annex III. (b) Art.43 is encoded with the notified-body transitional period: the second subparagraph of Art.43(3) sets 28 January 2028, while recital 18 describes eighteen months from 27 July 2026, which would be 27 January 2028; the operative date is the Article's and both are recorded because the divergence is in the published text. (c) Art.27 is encoded as a provision, the graph having previously carried its FRIA date without the article itself. (d) Arts.53 and 55 are added for temporal completeness; no MAILO constraint depends on the GPAI regime. METADATA AND PROVENANCE. The ECLIs for C-201/14, C-311/18 and C-511/18 were withheld in v5.2.0 rather than supplied from memory, since asserting an identifier from recall is the error class that produced the original C-203/22 defect; they are now added, each verified against InfoCuria or EUR-Lex and an independent secondary source, and they pass the citation-chronology constraint. The misspelled class mailo:EPOProceding is deprecated in favour of mailo:EPOProceeding, declared equivalent, with instances typed against both; renaming a published IRI would break every reference to it. P_634_004 is marked unanchored, and the art22ExceptionApplied comment, which had presented that reading as a holding of C-634/21, is reworded. Two triggeredBy edges left doubled by the v5.2.1 simplification were removed with their orphaned amendment nodes, and three dangling nodes were connected: Sanofi and Amgen as parties to UPC_Sanofi_Amgen, and EP 3 435 866 to UPC_Dexcom_Abbott. STILL OPEN. The unverified substantive assertions on UPC_10x_NanoString, UPC_Dexcom_Abbott and UPC_Sanofi_Amgen; the holding recorded for P_Bara_001; the conceptual confusion flagged on Tension_BlackBox_PatentProtection, which treats patent protection as an incentive to secrecy where a patent in fact requires disclosure; and independent expert review, which no finding in this or either preceding corrective release has received."""@en, "Version 5.2.3 (patch): canonical serialisation and residual consistency repair. AI Act RegulationVersion intervals were corrected to the exclusive-end convention [start, end) used elsewhere in the graph: AIAct_v1_ProhibitedPractices now ends 2025-08-02 rather than 2025-08-01, and AIAct_v2_GPAIModelObligations now ends 2026-07-27 rather than 2026-07-31, so the three AI Act intervals abut exactly at [2025-02-02, 2025-08-02), [2025-08-02, 2026-07-27) and [2026-07-27, ...). The end values had been recorded as inclusive last-days, which left a one-day and a four-day gap against the following start dates. The withdrawal of the AI Liability Directive proposal was recorded as 2025-02-10; February 2025 is when the Commission Work Programme listed the intended withdrawal, while formal withdrawal followed on 6 OCTOBER 2025, which is the operative date and is now recorded as such. The four root serialisations were regenerated from the canonical Turtle and synchronised to their docs/ copies. No class, property, individual or SHACL constraint was added, removed or altered, and mailo_shacl_shapes.ttl keeps its own version lineage."@en ; owl:priorVersion ; owl:versionIRI ; owl:versionInfo "5.2.3" . mailo:AIAct_2024_1689 a mailo:EURegulation ; rdfs:label "AI Act 2024/1689 (deprecated identifier)"@en ; rdfs:comment "Deprecated duplicate identifier for Regulation (EU) 2024/1689, retained for backward compatibility with v5.2.0 and earlier. Up to v5.2.0 the AI Act's article-level assertions were attached to mailo:AIAct while its temporal versions were attached to this identifier, and the two carried no relation, so no query could traverse from an article to the applicable regulation version. All assertions were migrated to mailo:AIAct in v5.2.1. Use mailo:AIAct."@en ; owl:deprecated true ; owl:sameAs mailo:AIAct . mailo:AnnexISectionBRoute a mailo:HighRiskRoute ; rdfs:label "Annex I Section B route — harmonisation legislation without third-party conformity assessment"@en ; rdfs:comment "The Art.6(1) route for products under the Annex I SECTION B legislation (civil aviation security, two- and three-wheel vehicles, agricultural vehicles, marine equipment, rail, motor vehicles), which does not carry the Section A third-party assessment requirement and is therefore not a medical-device route. Recorded to keep Section A identifiable rather than because MAILO models Section B products. Added in v5.2.2."@en . mailo:CJEU a mailo:Court ; rdfs:label "Court of Justice of the EU (CJEU)"@en . mailo:C_203_22 a mailo:CJEUCase ; rdfs:label "C-203/22 Dun & Bradstreet Austria"@en ; dcterms:date "2025-02-27"^^xsd:date ; dcterms:identifier "C-203/22", "ECLI:EU:C:2025:117" ; rdfs:comment "Inclusion basis (TransposedPrinciple): Telecoms credit scoring. The Art.15(1)(h) explanation standard is transposed to medical AI explanations."@en ; skos:notation "ECLI:EU:C:2025:117" ; mailo:citesPrecedent mailo:C_634_21 ; mailo:decidedBy mailo:CJEU_FirstChamber ; mailo:establishesPrinciple mailo:P_203_001 ; mailo:hasInclusionBasis mailo:TransposedPrinciple ; mailo:hasOutcome "Actionable explanations required (not algorithm formulas); explanation must enable effective exercise of Art.22(3) remedies."^^xsd:string ; mailo:hasStatus "Final — binding precedent"^^xsd:string ; mailo:interpretesArticle mailo:GDPR_Art15_1h, mailo:GDPR_Art22 . mailo:C_307_22 a mailo:CJEUCase ; rdfs:label "C-307/22 FT v DW (Medical Records Access)"@en ; dcterms:date "2023-10-26"^^xsd:date ; dcterms:identifier "C-307/22", "ECLI:EU:C:2023:811" ; rdfs:comment "Inclusion basis (DirectSubjectMatter): Access to medical records under GDPR Art.15 in a medical context."@en, "SCOPE CORRECTION (v5.1.0): this case involved NO artificial intelligence. It concerned dental records and the cost of a subject access request. The instance previously asserted involvesTechnology MedicalAISystem, an interpretesArticle link to GDPR Art.15(1)(h) (the logic of automated decision-making, not at issue here), and an outcome mentioning AI-generated diagnostic data. All three are removed. Note that this error survived the Part V technology constraint because the case was declared DirectSubjectMatter: that shape tests whether a medical AI claim is consistent with the declared basis, not whether the declared basis is itself correct."@en ; skos:notation "ECLI:EU:C:2023:811" ; mailo:decidedBy mailo:CJEU_FirstChamber ; mailo:establishesPrinciple mailo:P_307_001 ; mailo:hasInclusionBasis mailo:DomainContext ; mailo:hasOutcome "First Chamber, 26 October 2023 (FT v DW): a dentist refused a patient a free first copy of his medical records, relying on German law requiring patients to bear the cost. The Court held that Arts.12(5) and 15(1) and (3) GDPR oblige the controller to supply a first copy free of charge irrespective of the requester's reasons, that Art.23(1)(i) does not permit national law to shift that cost to protect the controller's economic interests, and that in a doctor-patient relationship the copy must be a faithful and intelligible reproduction, extending to full documents where necessary to understand the data." ; mailo:hasStatus "Final — binding precedent"^^xsd:string ; mailo:interpretesArticle mailo:GDPR_Art12_5, mailo:GDPR_Art15, mailo:GDPR_Art15_3, mailo:GDPR_Art23 . mailo:C_311_18 a mailo:CJEUCase ; rdfs:label "C-311/18 Schrems II"@en ; dcterms:date "2020-07-16"^^xsd:date ; dcterms:identifier "C-311/18", "ECLI:EU:C:2020:559" ; rdfs:comment "ECLI added in v5.2.2, verified against InfoCuria or EUR-Lex plus an independent secondary source rather than supplied from memory."@en, "Inclusion basis (DomainContext): Schrems II. Background for cross-border health data transfer under EHDS."@en ; skos:notation "ECLI:EU:C:2020:559" ; mailo:hasInclusionBasis mailo:DomainContext ; mailo:hasStatus "Final"^^xsd:string . mailo:ComplianceMatrix a owl:Class ; rdfs:label "ComplianceMatrix" ; rdfs:comment "A cross-reference mapping between frameworks" . mailo:ConstitutionalPrinciple a owl:Class ; rdfs:label "ConstitutionalPrinciple", "Constitutional Principle"@en ; rdfs:comment "Principle derived from EU Charter of Fundamental Rights", "Principle derived from the EU Charter of Fundamental Rights."@en ; rdfs:subClassOf mailo:LegalPrinciple . mailo:DataSubject a owl:Class ; rdfs:label "DataSubject", "Data Subject"@en ; rdfs:comment "An individual whose personal data is processed" ; rdfs:subClassOf mailo:LegalEntity . mailo:EPO_G0001_19 a mailo:EPOProceding, mailo:EPOProceeding ; rdfs:label "EPO G 0001/19 — Patentability of Computer-Implemented Simulations"@en ; dcterms:date "2021-03-10"^^xsd:date ; dcterms:identifier "G 0001/19" ; rdfs:comment "CORRECTED in v5.1.0. The earlier record stated that 'computer-implemented simulation of a technical process is not excluded from patentability (Art.52(2))'. Two errors. First, the Enlarged Board held that whether the SIMULATED system or process is technical is NOT decisive, so framing the holding around simulation of a technical process inverts the point. Second, the decision is an INVENTIVE STEP ruling under the COMVIK approach, not an Art.52(2) exclusion ruling: the eligibility hurdle is passed by any computer implementation and the analysis happens at Art.56. interpretesArticle is corrected from Art.52 to Art.56. Note this is the third instance of the same pattern in MAILO's EPO material — T 0161/18 and T 1191/19 were both attributed to Art.52 when they turn on Arts.83 and 56."@en, "Inclusion basis (TransposedPrinciple): Computer-implemented simulation generally. MAILO applies the technical-character reasoning to clinical simulation; the Enlarged Board did not rule on medical AI."@en ; mailo:decidedBy mailo:EPO_EnlargedBoA ; mailo:hasInclusionBasis mailo:TransposedPrinciple ; mailo:hasOutcome "Enlarged Board of Appeal, 10 March 2021 (application EP03793825.5, simulation of pedestrian crowd movement through a building). The COMVIK approach of T 641/00 applies to computer-implemented simulations, which are assessed by the same criteria as any other computer-implemented invention. A technical effect going beyond the normal electrical interactions within the computer — a further technical effect — is required to contribute to inventive step, and any further technical use of the simulation output must be at least implied by the claim. Question 2B was answered in the negative: it is NEITHER SUFFICIENT NOR NECESSARY that the simulation is based on technical principles underlying the simulated system." ; mailo:interpretesArticle mailo:EPC_Art56 . mailo:EPO_T1191_19 a mailo:EPOProceding, mailo:EPOProceeding ; rdfs:label "EPO T 1191/19 — Neuronal plasticity/INSTITUT GUTTMANN: meta-learning, inventive step and training-data sufficiency"@en ; dcterms:date "2022-04-01"^^xsd:date ; dcterms:identifier "T 1191/19" ; rdfs:comment "Inclusion basis (DirectSubjectMatter): Meta-learning for personalised neurorehabilitation interventions."@en ; mailo:citesPrecedent mailo:EPO_T0161_18 ; mailo:decidedBy mailo:EPO_BoA ; mailo:hasInclusionBasis mailo:DirectSubjectMatter ; mailo:hasOutcome "Technical Board of Appeal 3.5.05, application 08877672.9 (Fundacio Institut Guttmann): computer-implemented meta-learning method predicting personalised neurorehabilitation interventions. Appeal dismissed. Art.56: the mere application of a known machine learning technique to problems in a particular field is a general trend in technology (following T 161/18 point 3.6) and cannot be inventive as such. Art.83: insufficient disclosure - no example training or validation data, no minimum patient number for a meaningful prediction, and no disclosure of the artificial neural networks' topology, activation functions, end conditions or learning mechanism (following T 161/18 point 2); the Board described the disclosure as more like an invitation to a research programme. Separately, late-filed documents AX2-AX8 were not admitted under RPBA Art.12(4) - an evidential admissibility ruling, NOT a plausibility holding."^^xsd:string ; mailo:interpretesArticle mailo:EPC_Art56, mailo:EPC_Art83, mailo:EPC_Art84 . mailo:EUCharter a mailo:InternationalTreaty, mailo:RegulatoryFramework ; rdfs:label "EU Charter of Fundamental Rights"@en . mailo:GDPR_2016_679 a mailo:EURegulation ; rdfs:label "GDPR 2016/679 (deprecated identifier)"@en ; rdfs:comment "Deprecated duplicate identifier for Regulation (EU) 2016/679, retained for backward compatibility with v5.2.0 and earlier. The same representational split corrected for the AI Act in v5.2.1 affected the GDPR: article-level assertions were attached to mailo:GDPR and temporal versions to this identifier. All assertions were migrated to mailo:GDPR. Use mailo:GDPR."@en ; owl:deprecated true ; owl:sameAs mailo:GDPR . mailo:LegalTest a owl:Class ; rdfs:label "LegalTest", "Legal Test"@en ; rdfs:comment "Multi-factor test established by case law", "A multi-factor test established by case law to assess a legal question."@en . mailo:NationalAuthority a owl:Class ; rdfs:label "NationalAuthority", "National Authority"@en ; rdfs:comment "A national regulatory or supervisory body" ; rdfs:subClassOf mailo:LegalEntity . mailo:NationalCourtCase a owl:Class ; rdfs:label "NationalCourtCase", "National Court Case"@en ; rdfs:comment "Proceeding before a national court" ; rdfs:subClassOf mailo:Case . mailo:NotifiedBody a owl:Class ; rdfs:label "NotifiedBody", "Notified Body"@en ; rdfs:comment "A conformity assessment body notified under MDR/AI Act" ; rdfs:subClassOf mailo:LegalEntity . mailo:PNRDirective a mailo:EUDirective, mailo:RegulatoryFramework ; rdfs:label "PNR Directive 2016/681"@en ; dcterms:identifier "2016/681/EU" . mailo:PatentClaim a owl:Class ; rdfs:label "PatentClaim", "Patent Claim"@en ; rdfs:comment "A claim within a patent specification" . mailo:Risk_TradeSecretTransparencyConflict a mailo:ComplianceRisk ; rdfs:label "Trade Secret Shield vs. AI Act/GDPR Transparency Mandate (TSD Art.2 Conflict)"@en ; rdfs:comment "Medical AI operators relying on TSD Art.2 to protect model weights face: (a) mandatory AI Act Art.13 disclosure to deployers; (b) GDPR Art.15(1)(h) actionable explanation obligations to data subjects; (c) mandatory supervisory authority adjudication when TSD is claimed (C-203/22). Hybrid strategy required: patent final outputs, trade secret the AI platform, use ZKP/ZKML for auditor verification without weight disclosure."@en ; mailo:mitigatedBy mailo:ExplainabilityObligation ; mailo:riskLevel "HIGH"^^xsd:string . mailo:SoftwareAsMedicalDevice a owl:Class ; rdfs:label "SoftwareAsMedicalDevice", "Software as Medical Device (SaMD)"@en ; rdfs:comment "Software qualifying as medical device under MDR 2017/745", "Software qualifying as a medical device under MDR 2017/745."@en ; rdfs:seeAlso airo:HighRiskAISystem ; rdfs:subClassOf mailo:MedicalAISystem ; mailo:hasRisk mailo:Risk_AIActQMS_MDR_Overlap, mailo:Risk_SEP_FRAND . mailo:Tension_BlackBox_PatentProtection a mailo:CrossFrameworkTension ; rdfs:label "Black-Box Patent Protection vs. EU AI Act Transparency Obligation"@en ; rdfs:comment """\r Deep-learning medical AI is inherently opaque (black-box). Patent protection incentivises\r secrecy over model weights. EU AI Act Art.13 and C-817/19 black-box prohibition\r require transparency for high-risk AI systems, creating structural tension with\r commercial IP protection strategies. TSD Art.2 reinforces the secrecy interest;\r TSD Art.5 carve-outs are the only legislative bridge, but their scope is unresolved.\r """@en, "SCOPE NOTE added in v5.1.0: the label and comment conflate patent protection with secrecy. A patent REQUIRES disclosure sufficient to rework the invention (EPC Art.83; see T 161/18 and T 1191/19, which refused AI applications for failing exactly that). It is trade-secret protection, not patent protection, that incentivises withholding model weights. The genuine tension here is between keeping a model opaque and the disclosure that patenting it would require. Needs rewriting before it is cited."@en ; mailo:conflictType mailo:TradeSecretConflict ; mailo:conflictsWith mailo:TransparencyObligation ; mailo:hasResolutionStatus "unresolved" ; mailo:involvesFramework mailo:AIAct, mailo:EPC, mailo:TSD . mailo:Tension_DGA_Neutrality_DataAcquisition a mailo:CrossFrameworkTension ; rdfs:label "DGA Art.12(1)(a) neutrality obligation vs. AI Act Art.10 data governance"@en ; rdfs:comment "DGA Art.12(1)(a) prohibits a data intermediation services provider from using the data it intermediates for purposes other than putting it at the disposal of data users, and requires provision through a separate legal person. Integrated AI development models in which intermediary and developer are related entities therefore require structural separation, which no guidance resolves."@en ; mailo:conflictType mailo:DataGovernanceConflict ; mailo:hasResolutionStatus "unresolved"^^xsd:string ; mailo:involvesFramework mailo:AIAct, mailo:DGA . mailo:Tension_DataAct_TradeSecret_AITransparency a mailo:CrossFrameworkTension ; rdfs:label "Data Act Arts.4(6)-(8) / 5(9)-(11) trade secret vs. AI Act Art.13 + GDPR Art.15(1)(h) transparency"@en ; rdfs:comment "Article links added in v5.1.0: DataAct Art.4(6)-(8) and Art.5(9)-(11) for the trade-secret limb, AI Act Art.13 and GDPR Art.15(1)(h) for the transparency limb. Note the asymmetry now visible in the graph: Art.13's addressee is the deployer, not the affected person, and Art.86's right to explanation does not reach medical devices on the Annex I route, so on the transparency side only GDPR Art.15(1)(h) actually runs to the individual."@en, "C-203/22 Dun & Bradstreet §§59-62 limits the trade-secret defence: a controller cannot refuse all explanation on trade-secret grounds, and must describe the procedure and principles actually applied so the data subject understands which personal data were used and in what way. The precise boundary between permitted limitation and impermissible derogation is determined case by case, before a supervisory authority or court. NOTE: the Data Act limb of this tension was previously anchored to Art.15; the trade-secret safeguards are at Arts.4(6)-(8) and 5(9)-(11)."@en, "Re-anchored in v5.1.0 to TSD Art.1(2)(b), which preserves rules requiring disclosure of trade secrets to administrative or judicial authorities. The Art.5 exceptions previously relied on concern freedom of expression and whistleblowing, not data-subject or regulatory disclosure."@en ; mailo:conflictType mailo:TradeSecretConflict ; mailo:hasResolutionStatus "partial"^^xsd:string ; mailo:involvesArticle mailo:AIAct_Art13, mailo:DataAct_Art4, mailo:DataAct_Art5, mailo:GDPR_Art15_1h, mailo:TSD_Art1_2 ; mailo:involvesFramework mailo:AIAct, mailo:DataAct, mailo:GDPR . mailo:Tension_EHDS_GDPR_OptOut a mailo:CrossFrameworkTension ; rdfs:label "EHDS Art.71 opt-out vs. AI Act Art.10(3) representativeness"@en ; rdfs:comment "EHDS Art.71 grants natural persons a right to opt out of secondary use of their personal electronic health data, enforced by GDPR supervisory authorities. (Renumbered in v5.1.0 from Art.35, which was COM(2022) 197 proposal numbering. The earlier description of the right as 'unconditional' is not verified against the final text and has been removed.) Where opt-out rates are elevated in a clinically significant subpopulation, the resulting training dataset violates AI Act Art.10(3) representativeness. No legislative safe harbour bridges these positions."@en ; mailo:conflictType mailo:DataGovernanceConflict ; mailo:hasResolutionStatus "unresolved"^^xsd:string ; mailo:involvesFramework mailo:AIAct, mailo:EHDS, mailo:GDPR . mailo:Tension_EPO_GDPR_Disclosure a mailo:CrossFrameworkTension ; rdfs:label "EPC Art.83 Sufficiency Disclosure vs. GDPR Art.15(1)(h) + TSD Art.2 Secrecy"@en ; rdfs:comment """\r EPC Art.83 requires public disclosure of training data structure to maintain patent validity.\r GDPR Art.15(1)(h) + C-203/22 require actionable explanations to data subjects.\r The EPO disclosure may inadvertently establish a baseline that strengthens GDPR access rights,\r eroding TSD Art.2 trade secret shields that might otherwise limit GDPR Art.15 responses.\r TSD Art.9 judicial confidentiality measures do not apply to EPO prosecution, leaving no\r mechanism to contain the disclosure baseline once it enters the public patent record.\r """@en, "Re-anchored in v5.1.0 to TSD Art.1(2)(b), which preserves rules requiring disclosure of trade secrets to administrative or judicial authorities. The Art.5 exceptions previously relied on concern freedom of expression and whistleblowing, not data-subject or regulatory disclosure."@en ; mailo:conflictType mailo:TradeSecretConflict ; mailo:conflictsWith mailo:DisclosureObligation ; mailo:hasResolutionStatus "unresolved" ; mailo:involvesArticle mailo:TSD_Art1_2 ; mailo:involvesFramework mailo:EPC, mailo:GDPR, mailo:TSD . mailo:Tension_EPO_MDR_SemanticScope a mailo:CrossFrameworkTension ; rdfs:label "EPO G-II §3.3 broad claim scope vs. MDR Annex II §1.1 + AI Act Art.9/13 intended purpose"@en ; rdfs:comment "EPO patent prosecution optimises for broad technical-effect claims (wider scope = stronger IP protection); MDR Annex II requires precise intended-purpose specification (narrower scope = easier clinical evaluation); AI Act Art.9/13 requires intermediate characterisation covering intended purpose plus reasonably foreseeable misuse. The same SaMD must be described at opposing granularity levels across three institutional knowledge spaces. No EPO, MDCG, or AI Office guidance addresses this triangular scope alignment problem."@en ; mailo:conflictType mailo:SemanticGranularityConflict ; mailo:hasResolutionStatus "unresolved"^^xsd:string ; mailo:involvesArticle mailo:MDR_AnnexII ; mailo:involvesFramework mailo:AIAct, mailo:EPC, mailo:MDR . mailo:Tension_MDR_AIAct_DualTrack a mailo:CrossFrameworkTension ; rdfs:label "MDR Annex IX conformity vs. AI Act Art.43 conformity assessment"@en ; rdfs:comment "Two separate conformity assessment pathways for the same SaMD. MDCG AIB 2025-1 provides a procedural joint-assessment mechanism but does not harmonise the substantive requirement sets."@en ; mailo:conflictType mailo:ConformityGapConflict ; mailo:hasResolutionStatus "procedural_only"^^xsd:string ; mailo:involvesArticle mailo:MDR_AnnexIX ; mailo:involvesFramework mailo:AIAct, mailo:MDR . mailo:Tension_PLD_AIAct_Liability a mailo:CrossFrameworkTension ; rdfs:label "PLD Strict Liability vs. AI Act Conformity Defence"@en ; rdfs:comment """\r PLD Art.4 imposes strict liability; AI Act conformity assessment is not a\r defence to PLD claims. A CE-marked AI-SaMD that passes AI Act conformity\r assessment may still be 'defective' under PLD if it fails in real-world\r deployment. The two regimes do not create a safety harbour from each other.\r """@en, "OVERLAP NOTE added in v5.1.0: this tension and Tension_PLD_AIAct_LiabilityGap both concern PLD against AI Act conformity. They are arguably distinct — this one is that conformity assessment is no defence to strict liability (Art.4), the other that risk-management documentation may rebut the Art.10(4) presumption — but the labels do not make the distinction legible and the pair should be merged or renamed before the count of tensions is cited as a contribution."@en, "Re-anchored in v5.1.0 from PLD Art.4 to Arts.5 and 7. Strict liability is Art.5 and the defectiveness test is Art.7; Art.4 is the definitions provision. The substance of the tension is unaffected — AI Act conformity is not a defence to a PLD claim — but it had cited the wrong provisions."@en ; mailo:conflictType mailo:LiabilityConflict ; mailo:hasResolutionStatus "unresolved" ; mailo:involvesArticle mailo:PLD_Art5, mailo:PLD_Art7 ; mailo:involvesFramework mailo:AIAct, mailo:PLD . mailo:Tension_PLD_AIAct_LiabilityGap a mailo:CrossFrameworkTension ; rdfs:label "PLD Art.10(4) defect and causation presumption vs. AI Act Art.9 risk management"@en ; rdfs:comment "PLD Art.10(4) presumes defectiveness and/or the causal link where the claimant faces excessive difficulty owing to technical or scientific complexity and has demonstrated that the defect or link is probable; Art.10(5) preserves the defendant's right to rebut. AI Act Art.9-compliant risk-management documentation may serve to rebut, but no legislative safe harbour confirms this. CORRECTED in v5.1.0: the presumption was previously attributed to Art.11, which is exemption from liability."@en ; mailo:conflictType mailo:LiabilityConflict ; mailo:hasResolutionStatus "partial"^^xsd:string ; mailo:involvesFramework mailo:AIAct, mailo:PLD . mailo:Tension_TSD_AIAct_Transparency a mailo:CrossFrameworkTension ; rdfs:label "TSD Art.2 Trade Secret vs. AI Act Art.13 + GDPR Art.15(1)(h) — Core Transparency Tension"@en ; rdfs:comment "EHDS Art.52 is now encoded and provides the medical-domain resolution route for this tension: disclosure into a controlled environment under HDAB-imposed safeguards rather than to the requester. Added in v5.1.0."@en, "Re-anchored in v5.1.0 to TSD Art.1(2)(b), which preserves rules requiring disclosure of trade secrets to administrative or judicial authorities. The Art.5 exceptions previously relied on concern freedom of expression and whistleblowing, not data-subject or regulatory disclosure."@en, "TSD Art.2 protects model weights, training data schemas, and architecture as trade secrets. AI Act Art.13 requires deployers to receive sufficient information about high-risk AI system capabilities and limitations. C-203/22 §§59-62 sets the mandatory floor: controller cannot invoke TSD Art.2 to deny all causal explanation of an automated decision — disputed information must be produced to the competent supervisory authority or court for balance. AI Act Art.78 ensures national authority confidentiality but does not disapply Art.13 disclosure to deployers. TSD Art.5 is the only legislative bridge but its scope in the medical AI context remains unresolved pending a CJEU reference on Art.22 and clinical AI."@en ; mailo:conflictType mailo:TradeSecretConflict ; mailo:conflictsWith mailo:ExplainabilityObligation, mailo:TransparencyObligation ; mailo:hasResolutionStatus "partial"^^xsd:string ; mailo:involvesArticle mailo:TSD_Art1_2 ; mailo:involvesFramework mailo:AIAct, mailo:GDPR, mailo:TSD . mailo:TimelineEvent a owl:Class ; rdfs:label "TimelineEvent" ; rdfs:comment "A regulatory obligation activation date or deadline" . mailo:UPCA a mailo:InternationalTreaty, mailo:RegulatoryFramework ; rdfs:label "UPC Agreement (UPCA)"@en ; dcterms:date "2023-06-01"^^xsd:date ; dcterms:identifier "OJ C 175, 20.6.2013" ; rdfs:comment "Agreement on a Unified Patent Court (in force 2023-06-01). Establishes the UPC as a supranational court for infringement and revocation of European patents across contracting states. Key for medtech AI: opt-out strategy, PI threshold ('more likely than not' invalidity standard from 10x v. NanoString CoA), EPO-UPC divergence risk."@en . mailo:UPC_10x_NanoString a mailo:UPCCase ; rdfs:label "10x Genomics v. NanoString — Spatial Biology AI"@en ; rdfs:comment "Inclusion basis (DirectSubjectMatter): Spatial genomic analysis instrumentation and methods."@en, "SCOPE CORRECTION and UNVERIFIED (v5.1.0). The label 'Spatial Biology AI' and the tag involvesTechnology GenomicDiagnosticAI are not supported: the dispute concerns spatial genomics instrumentation and in-situ detection chemistry, with no AI element identified in the asserted claims. The tag is removed and the inclusion basis moved from DirectSubjectMatter to DomainContext. This is the same failure mode as Hologic v Siemens and Microsoft v Suinno. Separately, no case number is recorded — the only UPC instance without one — and the specific claims about the Court of Appeal's standard of proof and a 2025 settlement have not been verified. Do not cite until checked."@en ; mailo:decidedBy mailo:UPC_CoA, mailo:UPC_LD_Munich ; mailo:hasInclusionBasis mailo:DomainContext ; mailo:hasOutcome "Munich LD granted PI; CoA first substantive ruling reversed PI on invalidity grounds ('more likely than not' standard). Case settled 2025."^^xsd:string ; mailo:hasStatus "Settled 2025"^^xsd:string . mailo:UPC_Dexcom_Abbott a mailo:UPCCase ; rdfs:label "Dexcom v. Abbott — Continuous Glucose Monitoring"@en ; dcterms:identifier "UPC_CFI_230/2023 (ACT_546446/2023)" ; rdfs:comment "CORRECTED in v5.1.0. (1) The earlier record said the Paris LD revoked both EP 3 435 866 and EP 3 831 282. Only EP 866 was revoked in this decision; EP 282 was the subject of separate actions. (2) The recorded identifier ACT_583778/2023 is a different Paris action and does not correspond to the decision described; the correct reference is UPC_CFI_230/2023 (ACT_546446/2023). (3) decidedBy Munich LD is removed: Munich heard separate Dexcom actions, not this one. (4) involvesTechnology RiskScoringSystem is removed — a continuous glucose monitor is a sensing and measurement device, not a risk-scoring system. Whether any AI element is claimed has not been established, so no replacement technology is asserted."@en, "Inclusion basis (DirectSubjectMatter): Continuous glucose monitoring with algorithmic estimation."@en ; mailo:decidedBy mailo:UPC_LD_Paris ; mailo:hasDefendant mailo:Abbott ; mailo:hasInclusionBasis mailo:DirectSubjectMatter ; mailo:hasOutcome "Paris Local Division, 3 July 2024: on Dexcom's infringement action and Abbott's revocation counterclaim, EP 3 435 866 (remote monitoring of blood glucose) was revoked in its entirety for lack of inventive step, with effect across all UPC states. The patent had previously been maintained by the EPO in opposition on substantially the same grounds and prior art, making this a documented EPO-UPC divergence. The parties settled all patent disputes worldwide on 23 December 2024." ; mailo:hasPlaintiff mailo:Dexcom ; mailo:hasStatus "Settled globally 23 December 2024; all pending cases dismissed (verified 2026-09-10)" ; mailo:involvesPatent mailo:EP_3435866 . mailo:UPC_GXD_Myriad a mailo:UPCCase ; rdfs:label "GXD-Bio v. Myriad Genetics — Genomic Diagnostic AI"@en ; dcterms:identifier "UPC_CFI_437/2024" ; rdfs:comment "Inclusion basis (DirectSubjectMatter): Genomic diagnostic signature patent (EndoPredict)."@en ; mailo:decidedBy mailo:UPC_LD_Munich ; mailo:hasDefendant mailo:MyriadGenetics ; mailo:hasInclusionBasis mailo:DirectSubjectMatter ; mailo:hasOutcome "Munich Local Division, December 2025: EP 3 346 403 revoked for added matter and the infringement claims against Myriad's EndoPredict test dismissed. First UPC biotech NPE action."^^xsd:string ; mailo:hasPlaintiff mailo:GXDBio ; mailo:hasStatus "Under appeal — GXD-Bio appealed in February 2026 (verified 2026-09-01)"^^xsd:string ; mailo:involvesPatent mailo:EP_3346403 ; mailo:involvesTechnology mailo:GenomicDiagnosticAI . mailo:UPC_Hologic_Siemens a mailo:UPCCase ; rdfs:label "Hologic v. Siemens Healthineers — Medical Imaging AI"@en ; dcterms:identifier "UPC_CFI_758/2024" ; rdfs:comment "Inclusion basis (DomainContext): Mammography X-ray tube hardware, with no AI element in the asserted claim. Retained only for the EPO/MDR semantic-scope tension, where the point is precisely that patent claim scope and MDR intended purpose describe the same device at different granularities."@en, "SCOPE WARNING: this is medical imaging HARDWARE (X-ray tube focal spot control), not medical imaging AI. The asserted claim contains no AI or machine-learning element. Retained for the EPO/MDR/AI Act semantic-scope tension analysis, but it should not be counted as a medical-AI patent case."@en ; mailo:decidedBy mailo:UPC_LD_Dusseldorf ; mailo:hasDefendant mailo:SiemensHealthineers ; mailo:hasInclusionBasis mailo:DomainContext ; mailo:hasOutcome "Dusseldorf Local Division, decision of 10 June 2026 (with counterclaim UPC_CFI_259/2025): permanent injunction granted to Hologic in three member states, with recall and destruction orders against several Siemens Healthineers entities. The patent concerns the 'Flying Focal Spot' X-ray tube feature of the MAMMOMAT B.brilliant mammography system. NOTE: the item previously recorded here as a decision of 2026-03-09 was a procedural order on the further exchange of written pleadings, not the judgment."^^xsd:string ; mailo:hasPlaintiff mailo:Hologic ; mailo:hasStatus "Decided at first instance 2026-06-10 (verified 2026-09-01)"^^xsd:string ; mailo:involvesPatent mailo:EP_2352431 . mailo:UPC_Microsoft_Suinno a mailo:UPCCase ; rdfs:label "Microsoft v. Suinno — AI Method Patent"@en ; dcterms:identifier "UPC_CFI_433/2024" ; rdfs:comment "Inclusion basis (ProceduralPrecedent): EP 2 671 173 is a walking-and-browsing patent, neither medical nor AI. Retained for UPC practice only: Rule 30 RoP auxiliary requests and the representation requirement. Originally captured because the party name contains 'AI'."@en, "SCOPE WARNING: EP 2 671 173 is a 'method and means for browsing by walking' - discovering web search results by physical proximity while walking. It is neither a medical nor an AI patent. The party name (Suinno Mobile & AI Technologies Licensing Oy) contains 'AI', which is the likely reason this case was captured. It should not be counted as a medical-AI patent case."@en ; mailo:decidedBy mailo:UPC_CD_Paris ; mailo:hasDefendant mailo:Microsoft ; mailo:hasInclusionBasis mailo:ProceduralPrecedent ; mailo:hasOutcome "Paris Central Division, 7 January 2026: EP 2 671 173 revoked in its entirety on the counterclaim for revocation, with extensive guidance on the admissibility and presentation of auxiliary requests under Rule 30 RoP. Separately, the Court of Appeal held that Suinno was not validly represented, its managing director and main shareholder being precluded from acting as its UPC representative despite being qualified. NOTE: the ground previously recorded here as 'added matter' has not been confirmed against the decision text."^^xsd:string ; mailo:hasStatus "Revoked at first instance 2026-01-07; appealed by Suinno (verified 2026-09-01)"^^xsd:string ; mailo:involvesPatent mailo:EP_2671173 . mailo:UPC_Sanofi_Amgen a mailo:UPCCase ; rdfs:label "Sanofi & Regeneron v. Amgen — PCSK9 Medical Use"@en ; dcterms:identifier "UPC_CFI_505/2024" ; rdfs:comment "Inclusion basis (DirectSubjectMatter): Antibody claims; biologic rather than AI, retained for claim-scope reasoning."@en, "SCOPE CORRECTION and UNVERIFIED (v5.1.0). The instance was declared DirectSubjectMatter while its own note concedes the case is 'biologic rather than AI' — an internal contradiction. PCSK9 antibody claims involve no AI, so the basis is moved to DomainContext. The substantive assertions — a Dusseldorf LD finding of validity without infringement, a 'dual-element test', Court of Appeal restoration of EP'797 after Munich CD revocation, and an EPO Board hearing in April 2026 — are separately checkable and none has been verified. The status date has now passed. This is also the only UPC instance recording neither plaintiff nor defendant. Do not cite until checked."@en ; mailo:decidedBy mailo:UPC_LD_Dusseldorf ; mailo:hasDefendant mailo:Amgen ; mailo:hasInclusionBasis mailo:DomainContext ; mailo:hasOutcome "Düsseldorf LD: patent valid, no infringement (dual-element test established). CoA: EP'797 restored after Munich CD revocation. EPO-UPC divergence on inventive step."^^xsd:string ; mailo:hasPlaintiff mailo:Sanofi ; mailo:hasStatus "Active — EPO BoA hearing scheduled 2026-04"^^xsd:string . mailo:amendmentDescription a owl:DatatypeProperty ; rdfs:comment "Human-readable description of what this legislative event changes" ; rdfs:range xsd:string . mailo:amendmentProcedureRef a owl:DatatypeProperty ; rdfs:label "amendment procedure reference"@en ; rdfs:comment "The interinstitutional procedure file number, which is the stable handle for tracking status."@en ; rdfs:domain mailo:PendingAmendment ; rdfs:range xsd:string . mailo:amendmentProposalRef a owl:DatatypeProperty ; rdfs:label "amendment proposal reference"@en ; rdfs:comment "The Commission document reference of the proposal."@en ; rdfs:domain mailo:PendingAmendment ; rdfs:range xsd:string . mailo:applicablePhase a owl:DatatypeProperty ; rdfs:comment "Named applicability phase label (e.g. 'high_risk_full_obligations')" ; rdfs:range xsd:string . mailo:applicationDate a owl:DatatypeProperty ; rdfs:comment "Date from which the act applies" ; rdfs:range xsd:date . mailo:appliesTo a owl:ObjectProperty ; rdfs:label "applies to"@en ; rdfs:comment "Principle applies to a technology" ; rdfs:domain mailo:LegalPrinciple ; rdfs:range mailo:Technology . mailo:appliesToRoute a owl:ObjectProperty ; rdfs:label "applies to route"@en ; rdfs:comment "Restricts a provision to systems that became high-risk by a particular route. Art.86 is so restricted; most other high-risk obligations are not."@en . mailo:art22ExceptionApplied a owl:DatatypeProperty ; rdfs:label "Art.22 exception applied"@en ; rdfs:comment "contract_necessity | explicit_consent | member_state_law", "The GDPR Art.22(2) exception relied on: 'contract_necessity' | 'explicit_consent' | 'member_state_law'. Must be stated. Whether Art.22(1) then operates as a prohibition of its own force, absent a stated exception, is MAILO's reading (P_634_004) and not a holding of C-634/21; see the note on that principle. Reworded in v5.2.2."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:string . mailo:caseDate a owl:DatatypeProperty ; rdfs:comment "Date of judgment or decision" ; rdfs:range xsd:date . mailo:celex a owl:DatatypeProperty ; rdfs:comment "CELEX number for EU legislation" ; rdfs:range xsd:string . mailo:chamberType a owl:DatatypeProperty ; rdfs:comment "Grand Chamber, Chamber, etc." ; rdfs:range xsd:string . mailo:cites a owl:ObjectProperty ; rdfs:comment "Cites another CJEU case (alias)" ; rdfs:domain mailo:CJEUCase ; rdfs:range mailo:CJEUCase . mailo:citesPrecedent a owl:ObjectProperty ; rdfs:label "cites precedent"@en ; rdfs:comment "Case relies on earlier case as precedent", "A case relies on an earlier case as precedent."@en ; rdfs:domain mailo:Case ; rdfs:range mailo:Case . mailo:complementedBy a owl:ObjectProperty, owl:SymmetricProperty ; rdfs:label "complemented by"@en ; rdfs:comment "Article is complemented by another" ; rdfs:domain mailo:LegalArticle ; rdfs:range mailo:LegalArticle . mailo:conflictType a owl:ObjectProperty ; rdfs:label "conflict type"@en ; rdfs:comment "Relates a CrossFrameworkTension to the kind of conflict it exhibits. Declared in v5.1.0."@en ; rdfs:domain mailo:CrossFrameworkTension ; rdfs:range mailo:ConflictType . mailo:conflictsWith a owl:ObjectProperty, owl:SymmetricProperty ; rdfs:label "conflicts with"@en ; rdfs:comment "Article conflicts with another", "Two legal obligations or principles that are in tension."@en ; rdfs:domain mailo:LegalArticle ; rdfs:range mailo:LegalArticle . mailo:coversArticle a owl:ObjectProperty ; rdfs:label "covers article"@en ; rdfs:comment "Framework contains an article" ; rdfs:domain mailo:RegulatoryFramework ; rdfs:range mailo:LegalArticle . mailo:decidedBy a owl:ObjectProperty ; rdfs:label "decided by"@en ; rdfs:comment "Case was decided by a court" ; rdfs:domain mailo:Case ; rdfs:range mailo:Court . mailo:discusses a owl:ObjectProperty ; rdfs:comment "Source discusses a legal principle" ; rdfs:domain mailo:LiteratureSource ; rdfs:range mailo:LegalPrinciple . mailo:doi a owl:DatatypeProperty ; rdfs:comment "Digital Object Identifier" ; rdfs:range xsd:string . mailo:ecliIdentifier a owl:DatatypeProperty ; rdfs:comment "ECLI case identifier" ; rdfs:range xsd:string . mailo:enforcedBy a owl:ObjectProperty ; rdfs:comment "Framework enforced by authority" ; rdfs:domain mailo:RegulatoryFramework ; rdfs:range mailo:RegulatoryAuthority . mailo:entryIntoForce a owl:DatatypeProperty ; rdfs:comment "Date of entry into force" ; rdfs:range xsd:date . mailo:establishesPrinciple a owl:ObjectProperty ; rdfs:label "establishes principle"@en ; rdfs:comment "Case establishes a legal principle", "A case establishes a legal principle."@en ; rdfs:domain mailo:Case ; rdfs:range mailo:LegalPrinciple . mailo:governedBy a owl:ObjectProperty ; rdfs:comment "Governed by a regulatory framework" ; rdfs:domain mailo:Technology ; rdfs:range mailo:RegulatoryFramework . mailo:grantsRight a owl:ObjectProperty ; rdfs:label "grants right"@en ; rdfs:comment "Article grants a right" ; rdfs:domain mailo:LegalArticle ; rdfs:range mailo:Right . mailo:hasConformityRecord a owl:ObjectProperty ; rdfs:label "has conformity record"@en ; rdfs:comment "Links system to its conformity record" ; rdfs:domain mailo:MedicalAISystem ; rdfs:range mailo:ConformityRecord . mailo:hasDPIA a owl:ObjectProperty ; rdfs:label "has DPIA"@en ; rdfs:comment "Links system to its DPIA record", "Links a medical AI system to its DPIA record. Required for all high-risk automated processing (GDPR Art.35; C-817/19 P_817_005)."@en ; rdfs:domain mailo:MedicalAISystem ; rdfs:range mailo:DPIARecord . mailo:hasDefendant a owl:ObjectProperty ; rdfs:label "has defendant"@en ; rdfs:comment "Defendant in the case" ; rdfs:domain mailo:Case ; rdfs:range mailo:LegalEntity . mailo:hasDefinedScope a owl:DatatypeProperty ; rdfs:label "has defined scope"@en ; rdfs:comment "Tri-level proportionality (a): targeted scope (C-817/19 P_817_003)", "Tri-level proportionality test (a): targeted scope. C-817/19 P_817_003."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:boolean . mailo:hasEncodingBasis a owl:ObjectProperty ; rdfs:label "has encoding basis"@en ; rdfs:comment "Distinguishes a court's holding from the designer's operationalisation or transposition of it. Required on every encoded principle."@en ; rdfs:domain mailo:LegalPrinciple ; rdfs:range mailo:EncodingBasis . mailo:hasExplanation a owl:ObjectProperty ; rdfs:label "has explanation"@en ; rdfs:comment "Links system to its explanation for data subjects", "Links an automated decision system to its actionable explanation for data subjects (C-203/22 P_203_001; GDPR Art.15(1)(h))."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range mailo:Explanation . mailo:hasFormation a owl:DatatypeProperty ; rdfs:label "court formation"@en ; rdfs:domain mailo:CJEUCase ; rdfs:range xsd:string . mailo:hasHighRiskRoute a owl:ObjectProperty ; rdfs:label "has high-risk route"@en ; rdfs:comment "Records how a system qualifies as high-risk. Determines the availability of the Art.86 right to explanation and the Art.26(11) notification duty, neither of which reaches Annex I systems."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range mailo:HighRiskRoute . mailo:hasHumanOversightRecord a owl:ObjectProperty ; rdfs:label "has human oversight record"@en ; rdfs:comment "Links system to its oversight documentation", "Links an automated decision system to its human oversight documentation. At least one record is required (C-634/21 P_634_002)."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range mailo:HumanOversightRecord . mailo:hasInclusionBasis a owl:ObjectProperty ; rdfs:label "has inclusion basis"@en ; rdfs:comment "Records why a case belongs in MAILO. Required on every Case; see mailo:InclusionBasisShape."@en ; rdfs:domain mailo:Case ; rdfs:range mailo:InclusionBasis . mailo:hasIndividualNonAutomatedReview a owl:DatatypeProperty ; rdfs:label "has individual non-automated review"@en ; rdfs:comment "True = individual non-automated review before any measure (C-817/19 P_817_001)", "True if every positive automated match is subject to individual non-automated human review before any measure is taken. C-817/19 P_817_001 constitutional minimum."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:boolean . mailo:hasObligation a owl:ObjectProperty ; rdfs:comment "Framework imposes an obligation" ; rdfs:domain mailo:RegulatoryFramework ; rdfs:range mailo:ComplianceObligation . mailo:hasOutcome a owl:DatatypeProperty ; rdfs:label "outcome"@en ; rdfs:comment "Outcome or operative part of judgment" ; rdfs:domain mailo:Case ; rdfs:range xsd:string . mailo:hasPendingAmendment a owl:ObjectProperty ; rdfs:label "has pending amendment"@en ; rdfs:comment "Links a LegalArticle whose numbering or content is currently in force to a proposed amendment that has not been adopted. Asserting this property makes no claim that the amendment will pass."@en ; rdfs:domain mailo:LegalArticle ; rdfs:range mailo:PendingAmendment . mailo:hasPlaintiff a owl:ObjectProperty ; rdfs:label "has plaintiff"@en ; rdfs:comment "Plaintiff in the case" ; rdfs:domain mailo:Case ; rdfs:range mailo:LegalEntity . mailo:hasPrecedenceLevel a owl:DatatypeProperty ; rdfs:label "precedence level"@en ; rdfs:comment "binding_ratio | persuasive_obiter | constitutional_grand_chamber | authorial_transposition", "binding_ratio | persuasive_obiter | constitutional_grand_chamber | authorial_transposition. The first three record how much interpretive weight a holding claims. The fourth marks a mapping the ontology designer has drawn across instruments or contexts, which is NOT a holding of the Court and must not be cited as one. This is an analytical metadata model adopted by MAILO, not a formal precedent hierarchy operated by the CJEU."@en ; rdfs:domain mailo:LegalPrinciple ; rdfs:range xsd:string . mailo:hasPredeterminedCriteria a owl:DatatypeProperty ; rdfs:label "has predetermined criteria"@en ; rdfs:comment "True = pre-determined explainable criteria; false = self-learning black-box", "True if the system's assessment criteria are pre-determined and not modified by self-learning without human review. Required to avoid black-box prohibition (C-817/19 P_817_002)."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:boolean . mailo:hasPrinciple a owl:ObjectProperty ; rdfs:comment "Establishes a legal principle (alias)" ; rdfs:domain mailo:CJEUCase ; rdfs:range mailo:LegalPrinciple . mailo:hasPurposeLimitation a owl:DatatypeProperty ; rdfs:label "has purpose limitation"@en ; rdfs:comment "Tri-level proportionality (b): purpose limitation", "Tri-level proportionality test (b): documented purpose limitation. C-817/19 P_817_003."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:boolean . mailo:hasResolutionStatus a owl:DatatypeProperty ; rdfs:label "has resolution status"@en ; rdfs:comment "Records whether the tension between two frameworks has been resolved by legislation, case law or authoritative guidance. Values in use: 'unresolved' (no authority addresses the conflict), 'partial' (addressed in part, or addressed for an analogous but not identical situation), and 'procedural_only' (an authority routes the conflict to a forum without fixing a substantive outcome, as in C-203/22). This records the state of the law, not the ontology designer's confidence. Declared in v5.1.0 Round 5: the property was in use on tension instances but had never been typed or defined, so its value set was undocumented and unqueryable."@en ; rdfs:domain mailo:CrossFrameworkTension ; rdfs:range xsd:string . mailo:hasRisk a owl:ObjectProperty ; rdfs:label "has compliance risk"@en ; rdfs:comment "Technology carries a compliance risk" ; rdfs:domain mailo:Technology ; rdfs:range mailo:ComplianceRisk . mailo:hasStatus a owl:DatatypeProperty ; rdfs:label "status"@en ; rdfs:comment "Procedural status of the case" ; rdfs:domain mailo:Case ; rdfs:range xsd:string . mailo:hasTemporalLimitation a owl:DatatypeProperty ; rdfs:label "has temporal limitation"@en ; rdfs:comment "Tri-level proportionality (c): temporal limitation or review schedule", "Tri-level proportionality test (c): temporal limitation or mandatory periodic review. C-817/19 P_817_003."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:boolean . mailo:hasTension a owl:ObjectProperty ; rdfs:comment "Tension involves a framework" ; rdfs:domain mailo:CrossFrameworkTension ; rdfs:range mailo:RegulatoryFramework . mailo:hasVersion a owl:ObjectProperty ; rdfs:comment "Regulation has a temporal applicability version" ; rdfs:domain mailo:EURegulation ; rdfs:range mailo:RegulationVersion . mailo:identifiesInputFactors a owl:DatatypeProperty ; rdfs:label "identifies input factors"@en ; rdfs:comment "Explanation identifies which inputs affected the output", "True if the explanation identifies which specific input data affected the automated output and the main reasoning path (C-203/22 P_203_001; GDPR Art.15(1)(h))."@en ; rdfs:domain mailo:Explanation ; rdfs:range xsd:boolean . mailo:interpretesArticle a owl:ObjectProperty ; rdfs:label "interprets article"@en ; rdfs:domain mailo:Case ; rdfs:range mailo:LegalArticle . mailo:interpretsArticle a owl:ObjectProperty ; rdfs:comment "Case interprets a specific article" ; rdfs:domain mailo:Case ; rdfs:range mailo:LegalArticle . mailo:involvesArticle a owl:ObjectProperty ; rdfs:label "involves article"@en ; rdfs:comment "Links a CrossFrameworkTension to a provision it turns on. Added in v5.1.0: tensions cited annexes in prose that the ontology did not encode, so nothing connected the claim to the source."@en . mailo:involvesFramework a owl:ObjectProperty ; rdfs:label "involves framework"@en ; rdfs:comment "A cross-framework tension involves a regulatory framework." ; rdfs:domain mailo:CrossFrameworkTension ; rdfs:range mailo:RegulatoryFramework . mailo:involvesPatent a owl:ObjectProperty ; rdfs:label "involves patent"@en ; rdfs:comment "UPC case involves a patent" ; rdfs:domain mailo:UPCCase ; rdfs:range mailo:Patent . mailo:involvesTechnology a owl:ObjectProperty ; rdfs:label "involves technology"@en ; rdfs:comment "Case involves a technology class" ; rdfs:domain mailo:Case ; rdfs:range mailo:Technology . mailo:ipcClass a owl:DatatypeProperty ; rdfs:comment "IPC/CPC patent classification code" ; rdfs:range xsd:string . mailo:isActionable a owl:DatatypeProperty ; rdfs:label "is actionable"@en ; rdfs:comment "Explanation enables effective Art.22(3) remedies (C-203/22 P_203_001)", "True if the explanation enables effective exercise of GDPR Art.22(3) remedies (C-203/22 P_203_001). Abstract mathematical formulas are insufficient."@en ; rdfs:domain mailo:Explanation ; rdfs:range xsd:boolean . mailo:isCurrentVersion a owl:DatatypeProperty ; rdfs:comment "True if this is the currently applicable version as of the query date" ; rdfs:range xsd:boolean . mailo:isHighRiskAIAct a owl:DatatypeProperty ; rdfs:label "is high-risk under AI Act"@en ; rdfs:comment "True if system falls under AI Act Annex III high-risk categories", "True if the system falls within AI Act Annex III high-risk categories (medical device AI, biometric, critical infrastructure)."@en ; rdfs:domain mailo:MedicalAISystem ; rdfs:range xsd:boolean . mailo:isNonTechnical a owl:DatatypeProperty ; rdfs:label "is non-technical"@en ; rdfs:comment "Explanation expressed in non-technical language", "True if the explanation is expressed in language understandable to the data subject without specialist knowledge (C-203/22 P_203_001)."@en ; rdfs:domain mailo:Explanation ; rdfs:range xsd:boolean . mailo:legislativeStatus a owl:DatatypeProperty ; rdfs:label "legislative status"@en ; rdfs:comment "Stage reached in the ordinary legislative procedure at the date given by statusVerifiedAsAt. Values used: 'proposed', 'council_position', 'parliament_position', 'trilogue', 'provisional_agreement', 'adopted'."@en ; rdfs:domain mailo:PendingAmendment ; rdfs:range xsd:string . mailo:lit_Andrade_F_Rebelo_D_M_2024_Schufa_s_Case_C-634_21_on a mailo:LiteratureSource ; rdfs:label "Andrade, F., Rebelo, D. M. (2024). Schufa's Case C-634/21 on ADM: the 'Lenders' Quest' for GDPR-friendly Scoring has not been settled yet! SSRN 4882806." ; dcterms:date "2024" ; dcterms:description "Challenges the breadth of SCHUFA holding. Informs MAILO's functional-solely rule encoding." ; dcterms:source ; mailo:relevanceScore 4 . mailo:lit_Communications_Medicine_Nature_2025_Neuro-symbolic_AI a mailo:LiteratureSource ; rdfs:label "Communications Medicine (Nature). (2025). Neuro-symbolic AI for auditable cognitive information extraction from medical reports." ; dcterms:date "2025" ; dcterms:description "LLM extracts candidate facts → symbolic verifier checks against medical rules → traceable, deterministic labels. Empirical demonstration of the harness pattern MAILO advocates, applied to clinical text." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_Conflict-tolerant_Deontic_Traditional_Scheme_for_the_Semanti a mailo:LiteratureSource ; rdfs:label "Conflict-tolerant Deontic Traditional Scheme for the Semantic Web. (2025). Journal of Logic and Computation 35(8), Oxford." ; dcterms:date "2025" ; dcterms:description "Addresses cross-framework normative conflict resolution that MAILO will face (e.g. GDPR Art.22 vs AI Act Art.86). Provides RDF-based conflict-tolerant deontic logic." ; dcterms:source ; mailo:relevanceScore 4 . mailo:lit_DAOnt_A_Formal_Ontology_for_EU_Data_Act_Compliance_2024_2 a mailo:LiteratureSource ; rdfs:label "DAOnt: A Formal Ontology for EU Data Act Compliance. (2024/2026). arXiv:2604.16386." ; dcterms:date "2024" ; dcterms:description "Reuses LKIF-Core + ODRL + DPV. Single-framework analogue of MAILO's multi-framework target." ; dcterms:source ; mailo:relevanceScore 4 . mailo:lit_EHJ_Digital_Health_Oxford_2025_Medicine_healthcare_an a mailo:LiteratureSource ; rdfs:label "EHJ Digital Health (Oxford). (2025). Medicine, healthcare and the AI act: gaps, challenges and future implications." ; dcterms:date "2025" ; dcterms:description "Explicit multi-framework gap analysis. Core citation for MAILO Chapter 2." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_Esteves_B_Rodr_guez-Doncel_V_et_al_2024_Analysis_of a mailo:LiteratureSource ; rdfs:label "Esteves, B., Rodríguez-Doncel, V. et al. (2024). Analysis of ontologies and policy languages to represent information flows in GDPR. Semantic Web Journal." ; dcterms:date "2024" ; dcterms:description "Comparative table of ontology and policy languages. Positions DPV + ODRL + LegalRuleML as Pareto-front. ODRL+LegalRuleML+DPV+GDPRtEXT can model 39 of 57 informational items." ; dcterms:source ; mailo:relevanceScore 4 . mailo:lit_Frontiers_Digital_Health_2025_Ontologies_as_the_semantic a mailo:LiteratureSource ; rdfs:label "Frontiers Digital Health. (2025). Ontologies as the semantic bridge between artificial intelligence and healthcare." ; dcterms:date "2025" ; dcterms:description "OWL2 + SWRL semantic reasoning enables auditable inference consistent with domain expertise. Ontologies encode explicit rules and traceable data relationships required by AI Act. Doctrinal scaffold for MAILO." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_Golpayegani_D_Pandit_H_J_Lewis_D_2022_2025_AIRO a mailo:LiteratureSource ; rdfs:label "Golpayegani, D., Pandit, H. J., Lewis, D. (2022, 2025). AIRO: An Ontology for Representing AI Risks Based on the Proposed EU AI Act. SEMANTiCS + arXiv:2501.10391." ; dcterms:date "2025" ; dcterms:description "OWL ontology + SHACL shapes implementing AI Act Annex III high-risk classifier + SPARQL ASK queries; integrated into W3C DPVCG. Most direct competitor to MAILO. AIRO already does for the AI Act what MAILO proposes for the medical-AI multi-framework intersection. Differentiator: MAILO adds medical-domain specificity, MDR+PLD integration, CJEU precedent chain, UPC patent overlay." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_Hybrid_Neuro-Symbolic_Models_for_Ethical_AI_in_Risk-Sensitiv a mailo:LiteratureSource ; rdfs:label "Hybrid Neuro-Symbolic Models for Ethical AI in Risk-Sensitive Domains. (2025). arXiv:2511.17644." ; dcterms:date "2025" ; dcterms:description "Risk-sensitive domains demand transparency + regulatory compliance. Neural pattern recognition + symbolic reasoning yields verifiable explanations. Healthcare among the case studies." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_JMIR_2026_The_Right_to_Understand_in_Health_Care_AI a mailo:LiteratureSource ; rdfs:label "JMIR. (2026). The Right to Understand in Health Care AI." ; dcterms:date "2026" ; dcterms:description "Argues neither AI Act nor GDPR specifies what a meaningful clinical explanation is. Identifies the gap MAILO fills via its semantic schema." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_Kang_M_Li_B_et_al_2025_R_-Guard_Robust_Reasoning_E a mailo:LiteratureSource ; rdfs:label "Kang, M., Li, B. et al. (2025). R²-Guard: Robust Reasoning Enabled LLM Guardrail via Knowledge-Enhanced Logical Reasoning. ICLR 2025." ; dcterms:date "2025" ; dcterms:description "Markov logic networks + probabilistic circuits over a safety reasoning graph. +30.2% over Llama Guard on ToxicChat; +59.5% under jailbreak attacks. Direct precedent for logic-graph-as-runtime-safety-filter pattern that MAILO advances, but for general LLM safety, not medical regulation." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_MDCG_2025-6_2025_European_Commission_Joint_Guidance_on_M a mailo:LiteratureSource ; rdfs:label "MDCG 2025-6. (2025). European Commission Joint Guidance on MDR/IVDR × AI Act." ; dcterms:date "2025" ; dcterms:description "Official EU-level mapping document. MAILO instances must align with it." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_MDPI_Information_2024_A_SHACL-Based_Approach_for_Enhanci a mailo:LiteratureSource ; rdfs:label "MDPI Information. (2024). A SHACL-Based Approach for Enhancing Automated Compliance Checking with RDF Data, 15(12):759." ; dcterms:date "2024" ; dcterms:description "Operationalises the validation layer that MAILO needs for runtime SHACL checks." ; dcterms:source ; mailo:relevanceScore 4 . mailo:lit_Medical_Law_Review_2024_A_health-conformant_reading_of_t a mailo:LiteratureSource ; rdfs:label "Medical Law Review. (2024). A health-conformant reading of the GDPR's right not to be subject to automated decision-making, 32(3), 373." ; dcterms:date "2024" ; dcterms:description "Only article addressing the exact intersection of GDPR Art.22 + medical AI. Doctrinal foundation for MAILO's medical-AI ADM rules." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_Pandit_H_J_et_al_2024_Data_Privacy_Vocabulary_DPV a mailo:LiteratureSource ; rdfs:label "Pandit, H. J. et al. (2024). Data Privacy Vocabulary (DPV) – Version 2.0. Springer / W3C DPVCG." ; dcterms:date "2024" ; dcterms:description "RDFS/SKOS + OWL2 alt; 911 concepts; serialised in TTL/JSON-LD/RDF-XML; namespace http://w3id.org/dpv#. Foundational vocabulary the thesis should reuse. Recommendation: import DPV namespace into MAILO; align mailo:hasObligation with dpv:hasLegalBasis." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_Rebedea_T_et_al_2023_NeMo_Guardrails_A_Toolkit_for_Co a mailo:LiteratureSource ; rdfs:label "Rebedea, T. et al. (2023). NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails. EMNLP Demo. arXiv:2310.10501." ; dcterms:date "2023" ; dcterms:description "Colang DSL + dialogue-state engine; rules user-defined and interpretable. 2025 update: NIM micro-services with ~0.5s added latency for ~50% protection lift. Establishes programmable-rail paradigm; uses ad-hoc DSL not formal ontology." ; dcterms:source ; mailo:relevanceScore 4 . mailo:lit_Robaldo_L_Bartolini_C_Palmirani_M_et_al_2020_Fo a mailo:LiteratureSource ; rdfs:label "Robaldo, L., Bartolini, C., Palmirani, M., et al. (2020). Formalizing GDPR Provisions in Reified I/O Logic: The DAPRECO Knowledge Base. Journal of Logic, Language and Information." ; dcterms:date "2020" ; dcterms:description "LegalRuleML + reified Input/Output logic + SPINDle defeasible reasoner. The canonical precedent for formal GDPR reasoning. Limitations addressed by MAILO: GDPR-only; no medical AI integration; no CJEU case-law layer; no runtime guardrail integration." ; dcterms:source ; mailo:relevanceScore 5 . mailo:lit_npj_Health_Systems_2026_Data-driven_medical_devices_and a mailo:LiteratureSource ; rdfs:label "npj Health Systems. (2026). Data-driven medical devices and the EU MDR: mapping gaps in standards for regulatory compliance." ; dcterms:date "2026" ; dcterms:description "ISO/IEC 27001/27010 + ISO/IEC 5338, 25059, BS 30440 are insufficient for AI medical safety/transparency — exactly the gap MAILO addresses." ; dcterms:source ; mailo:relevanceScore 4 . mailo:mitigatedBy a owl:ObjectProperty ; rdfs:label "mitigated by"@en ; rdfs:comment "Risk mitigated by obligation" ; rdfs:domain mailo:ComplianceRisk ; rdfs:range mailo:ComplianceObligation . mailo:overrules a owl:ObjectProperty ; rdfs:comment "Overrules or modifies a prior principle" ; rdfs:domain mailo:CJEUCase ; rdfs:range mailo:LegalPrinciple . mailo:oversightIsSubstantive a owl:DatatypeProperty ; rdfs:label "oversight is substantive"@en ; rdfs:comment "True = genuine review; false = rubber-stamping (C-634/21 P_634_002)", "True if the human reviewer exercised independent judgment; false if the review was a mere formality (rubber-stamping). C-634/21 P_634_002 requires true."@en ; rdfs:domain mailo:HumanOversightRecord ; rdfs:range xsd:boolean . mailo:producesSignificantImpact a owl:DatatypeProperty ; rdfs:label "produces significant impact"@en ; rdfs:comment "True if outputs have legal/similarly significant effects on individuals", "True if the system's outputs have legal or similarly significant effects on individuals (C-634/21 P_634_001). Triggers full Art.22 obligation chain."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:boolean . mailo:proposedEffect a owl:DatatypeProperty ; rdfs:label "proposed effect"@en ; rdfs:comment "What the proposal would do to the provision, in the drafter's own terms."@en ; rdfs:domain mailo:PendingAmendment ; rdfs:range xsd:string . mailo:proposedRenumberingTo a owl:DatatypeProperty ; rdfs:label "proposed renumbering to"@en ; rdfs:comment "The article designation the provision would carry if the proposal is adopted as drafted. A string, not an IRI: the target provision does not exist in law and must not be resolvable as though it did."@en ; rdfs:domain mailo:PendingAmendment ; rdfs:range xsd:string . mailo:relatedPatent a owl:ObjectProperty ; rdfs:comment "Involves patent class" ; rdfs:domain mailo:UPCCase ; rdfs:range mailo:PatentClassification . mailo:relevanceScore a owl:DatatypeProperty ; rdfs:comment "Relevance score 1-5" ; rdfs:range xsd:integer . mailo:requiresAssessment a owl:ObjectProperty ; rdfs:comment "Technology requires a conformity path" ; rdfs:domain mailo:Technology ; rdfs:range mailo:ConformityAssessment . mailo:riskLevel a owl:DatatypeProperty ; rdfs:label "risk level"@en ; rdfs:comment "CRITICAL | HIGH | MEDIUM | LOW" ; rdfs:domain mailo:ComplianceRisk ; rdfs:range xsd:string . mailo:statusVerifiedAsAt a owl:DatatypeProperty ; rdfs:label "status verified as at"@en ; rdfs:comment "The date on which the legislative status recorded here was checked against a source. A status assertion without this date is not auditable and should not be relied on."@en ; rdfs:domain mailo:PendingAmendment ; rdfs:range xsd:date . mailo:triggeredBy a owl:ObjectProperty ; rdfs:comment "Version created or activated by a legislative event" ; rdfs:domain mailo:RegulationVersion ; rdfs:range mailo:LegislativeAmendment . mailo:triggersObligation a owl:ObjectProperty ; rdfs:label "triggers obligation"@en ; rdfs:comment "Principle triggers a compliance obligation" ; rdfs:domain mailo:LegalPrinciple ; rdfs:range mailo:ComplianceObligation . mailo:triggersRoute a owl:ObjectProperty ; rdfs:label "triggers high-risk route"@en ; rdfs:comment "Links a classification rule to the AI Act high-risk route it produces. MDR Annex VIII Rule 11 places most diagnostic and decision-support software at Class IIa or above, requiring notified-body assessment, which satisfies AI Act Art.6(1) and puts the system on the Annex I route — with the consequence that the Art.86 right to explanation does not apply. Encoded in v5.1.0 so the chain is queryable rather than only described in prose."@en . mailo:usesHealthStatusAsCriterion a owl:DatatypeProperty ; rdfs:label "uses health status as criterion"@en ; rdfs:comment "Must be false — health status excluded from automated criteria (C-817/19 P_817_004)", "True if the system uses an individual's health status as an automated classification criterion. Must be false (C-817/19 P_817_004 Grand Chamber prohibition)."@en ; rdfs:domain mailo:AutomatedDecisionSystem ; rdfs:range xsd:boolean . mailo:versionEndDate a owl:DatatypeProperty ; rdfs:comment "Date until which this version applies (exclusive; absent = open-ended)" ; rdfs:range xsd:date . mailo:versionOf a owl:ObjectProperty ; rdfs:comment "Version belongs to a parent regulation" ; rdfs:domain mailo:RegulationVersion ; rdfs:range mailo:EURegulation . mailo:versionStartDate a owl:DatatypeProperty ; rdfs:comment "Date from which this regulation version/expression applies" ; rdfs:range xsd:date . mailo:AIAct_Art10 a mailo:LegalArticle ; rdfs:label "AI Act Art.10 — Data and data governance"@en ; rdfs:comment "AMENDED (v5.2.2): Art.1(9) of Regulation (EU) 2026/1744 DELETED Art.10(5), which had carried the legal basis for processing special categories for bias detection by providers of high-risk systems. That basis is now Art.4a, with a wider addressee set. Any constraint or argument anchored to Art.10(5) is without a source in the text in force and must be re-anchored to Art.4a or retired."@en, "Art.10(3): training, validation and testing data sets shall be relevant, sufficiently representative, and, TO THE BEST EXTENT POSSIBLE, free of errors and complete in view of the intended purpose. CORRECTED in v5.1.0: the earlier comment listed 'freedom from errors, completeness' as flat criteria, dropping the qualifier the final text added precisely because absolute error-freedom is unattainable. The distinction matters for any claim that a dataset defect breaches Art.10(3) — including Tension_EHDS_GDPR_OptOut, which asserts that elevated opt-out rates in a clinically significant subpopulation put a dataset in violation. Under the qualified standard that conclusion requires showing the shortfall was avoidable, not merely that it exists."@en . mailo:AIAct_Art17 a mailo:LegalArticle ; rdfs:label "AI Act Art.17 — Quality Management System"@en ; rdfs:comment "High-risk AI providers must implement a QMS covering design, development, post-market monitoring. Aligned with MDR Annex IX QMS requirements to avoid double burden."@en . mailo:AIAct_Art26 a mailo:LegalArticle ; rdfs:label "AI Act Art.26 — Obligations of deployers of high-risk AI systems"@en ; rdfs:comment "Deployers, for example a hospital using a third-party system, must take appropriate technical and organisational measures to ensure use in accordance with the instructions (26(1)); assign human oversight to natural persons with the necessary competence, training and authority (26(2)); ensure input data is relevant and sufficiently representative for the intended purpose so far as they control it (26(4)); monitor operation and inform the provider of risks or serious incidents (26(5)); keep automatically generated logs for at least six months (26(6)); inform workers before putting a system into use in the workplace (26(7)); and, where a decision concerning a natural person is taken on the basis of an Annex III system's output, inform the affected person that they are subject to it (26(11)). Note that 26(11), like Art.86, is scoped to Annex III."@en ; mailo:complementedBy mailo:AIAct_Art14 . mailo:AIAct_Art27 a mailo:LegalArticle ; rdfs:label "AI Act Art.27 — Fundamental rights impact assessment"@en ; rdfs:comment "Deployers that are bodies governed by public law, private entities providing public services, or deployers of systems under Annex III points 5(b) and (c) must perform a FRIA before first use. Scoped to Art.6(2) deployers, with application from 2 December 2027 following Regulation (EU) 2026/1744. The date was corrected in v5.2.1; this entry adds the provision itself, which the graph referenced temporally without encoding. Added in v5.2.2."@en . mailo:AIAct_Art43 a mailo:LegalArticle ; rdfs:label "AI Act Art.43 — Conformity assessment (notified-body transitional period as amended)"@en ; rdfs:comment "As amended by Regulation (EU) 2026/1744. The second subparagraph of Art.43(3) sets 28 JANUARY 2028 as the deadline within which notified bodies already designated under the Annex I Section A legislation may continue to act for AI systems falling under that legislation. Recital 18 characterises the same period as eighteen months from 27 July 2026, which would fall on 27 January 2028; the operative date is the one in the Article. Both are recorded because the divergence is in the published text and a reader checking the recital against the Article will otherwise think one of them is a transcription error. Added in v5.2.2."@en . mailo:AIAct_Art4a a mailo:LegalArticle ; rdfs:label "AI Act Art.4a — Processing of special categories of personal data for bias detection and correction"@en ; rdfs:comment """Inserted into Chapter I by Art.1(6) of Regulation (EU) 2026/1744, in force 27 July 2026. Art.4a(1) permits, exceptionally, processing of GDPR Art.9(1) special categories to detect and correct bias, on five cumulative conditions: it cannot effectively be achieved with other data including synthetic or anonymised data; technical limits on re-use plus state-of-the-art security and privacy-preserving measures including pseudonymisation; strict access controls, documentation and confidentiality; no transmission, transfer or access by other parties; and deletion once bias is corrected or the retention period ends, whichever is first. Art.4a(2) extends the basis beyond providers of high-risk systems to deployers of high-risk systems and to providers and deployers of other AI systems and AI models, and states expressly that it creates NO obligation to perform bias detection. This is not a relocation of the deleted Art.10(5). The basis was extracted from the high-risk training-data article into Chapter I with a broadened personal scope, so the addressee set changes. Art.4a sits in Chapter I, covered by Art.113 third subparagraph point (a) (2 February 2025), but as an insertion effective 27 July 2026 it cannot operate retroactively; processing before that date is assessed on its own footing. Added in v5.2.2."""@en . mailo:AIAct_Art50 a mailo:LegalArticle ; rdfs:label "AI Act Art.50 — Transparency obligations for certain AI systems"@en ; rdfs:comment "Transparency duties on providers and deployers of AI systems that interact with natural persons or generate synthetic audio, image, video or text, including machine-readable marking and disclosure of artificially generated content. Applies to AI-generated radiology reports and diagnostic summaries. Relabelled in v5.1.0: this is NOT the GPAI regime, which is Chapter V (Arts.51-56) and addresses providers of general-purpose AI models."@en . mailo:AIAct_Art53 a mailo:LegalArticle ; rdfs:label "AI Act Art.53 — Obligations for providers of general-purpose AI models"@en ; rdfs:comment "Technical documentation, information for downstream providers, copyright policy and a training-content summary. Applicable from 2 August 2025; models placed on the market before that date have until 2 August 2027 under Art.111(3). Recorded for temporal completeness: GPAI obligations are not a medical-device route and no MAILO constraint depends on them. Added in v5.2.2."@en . mailo:AIAct_Art55 a mailo:LegalArticle ; rdfs:label "AI Act Art.55 — Obligations for providers of GPAI models with systemic risk"@en ; rdfs:comment "Model evaluation, adversarial testing, systemic-risk assessment and mitigation, serious-incident reporting and cybersecurity protection, in addition to Art.53. Same application dates as Art.53. Recorded for temporal completeness. Added in v5.2.2."@en . mailo:AIAct_Art86 a mailo:LegalArticle ; rdfs:label "AI Act Art.86 — Right to explanation of individual decision-making (Annex III only)"@en ; rdfs:comment """An affected person subject to a decision taken by a deployer on the basis of the output of a high-risk AI system LISTED IN ANNEX III, excluding systems under point 2, which produces legal effects or similarly significantly affects them adversely in health, safety or fundamental rights, may obtain from the deployer clear and meaningful explanations of the role of the AI system in the decision-making procedure and of the main elements of the decision. Art.86(2) disapplies it where Union or national law provides exceptions; Art.86(3) makes it residual to rights available elsewhere in Union law. SCOPE WARNING added in v5.1.0, and consequential for MAILO's domain: AI-based medical devices are high-risk by the ANNEX I route under Art.6(1), because MDR and IVDR are listed in Annex I Section A and Class IIa/IIb/III devices require third-party conformity assessment. Annex I is not Annex III. Art.86 therefore does NOT confer a right to explanation in respect of a diagnostic or imaging AI device that is high-risk solely as a regulated medical device. It can apply to healthcare AI reaching high-risk status through Annex III point 5, such as systems determining eligibility for publicly funded healthcare services. For the Annex I population, GDPR Art.15(1)(h) and Art.22 carry the explanation duty alone. The previous complementedBy link to GDPR Art.15(1)(h) implied the two operate together across medical AI; it is retained but qualified by this note."""@en ; mailo:appliesToRoute mailo:AnnexIIIRoute ; mailo:complementedBy mailo:GDPR_Art15_1h . mailo:AIAct_Art9 a mailo:LegalArticle ; rdfs:label "AI Act Art.9 — Risk management system"@en ; rdfs:comment "A risk management system shall be established, implemented, documented and maintained for high-risk AI systems as a CONTINUOUS ITERATIVE PROCESS planned and run throughout the entire lifecycle, requiring regular systematic review and updating. It comprises identification and analysis of known and reasonably foreseeable risks, estimation of risks in intended use and reasonably foreseeable misuse, and adoption of appropriate targeted risk-management measures. Art.9(10) permits integration with procedures required by other Union harmonisation law, which is the basis for combining it with MDR risk management rather than duplicating it."@en ; mailo:complementedBy mailo:AIAct_Art14 . mailo:AIAct_v1_ProhibitedPractices a mailo:RegulationVersion ; rdfs:label "Chapters I and II apply (general provisions, AI literacy, prohibited practices)" ; rdfs:comment "Article 113(a) of Regulation (EU) 2024/1689 makes Chapters I and II applicable from 2 February 2025. The Regulation entered into force on 1 August 2024; no provision applied between 1 August 2024 and 1 February 2025, so no RegulationVersion covers that interval."@en ; mailo:applicablePhase "prohibited_practices_and_general" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:Amend_AIAct_v1_ProhibitedPractices ; mailo:versionEndDate "2025-08-02"^^xsd:date ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2025-02-02"^^xsd:date . mailo:AIAct_v2_GPAIModelObligations a mailo:RegulationVersion ; rdfs:label "GPAI model provider obligations apply (Title VIII, Arts.51–56)" ; mailo:applicablePhase "gpai_model_obligations" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:Amend_AIAct_v2_GPAIModelObligations ; mailo:versionEndDate "2026-07-27"^^xsd:date ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2025-08-02"^^xsd:date . mailo:AIAct_v2b_PostOmnibusTransitional a mailo:RegulationVersion ; rdfs:label "AI Act as amended by Regulation (EU) 2026/1744 — current legal text" ; rdfs:comment "The applicable phase from 27 July 2026, when Regulation (EU) 2026/1744 entered into force and its amendments became part of the AI Act. This version records that the amended text is the one in force. It does not assert that the Chapter III Sections 1 to 3 obligations for high-risk systems apply; those apply from 2 December 2027 and 2 August 2028 by route."@en ; mailo:applicablePhase "post_omnibus_transitional" ; mailo:isCurrentVersion true ; mailo:triggeredBy mailo:Amend_AIAct_v2b_PostOmnibusTransitional ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2026-07-27"^^xsd:date . mailo:AIAct_v3_OriginalHighRisk a mailo:RegulationVersion ; rdfs:label "Original Article 6(2) / Annex III high-risk application date (superseded)" ; rdfs:comment "Original Article 113 set 2 August 2026 for Article 6(2) Annex III systems and 2 August 2027 for Article 6(1) Annex I systems; the two were distinct from the outset and are represented separately from v5.2.1. This phase never became applicable: Regulation (EU) 2026/1744 entered into force on 27 July 2026, six days before this date."@en ; mailo:applicablePhase "high_risk_annex_iii_original" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:Amend_AIAct_v3_OriginalHighRisk ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2026-08-02"^^xsd:date . mailo:AIAct_v3a_Omnibus_AnnexIII_Standalone a mailo:RegulationVersion ; rdfs:label "Omnibus-revised Annex III standalone high-risk deadline (biometrics, employment, education, law enforcement, critical infrastructure, border management)" ; mailo:applicablePhase "high_risk_annex_iii_omnibus" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:OmnibusAI_RegEU2026_1744 ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2027-12-02"^^xsd:date . mailo:AIAct_v3b_Omnibus_AnnexI_SaMD a mailo:RegulationVersion ; rdfs:label "Omnibus-revised Annex I deadline — SaMD under MDR 2017/745 and IVDR 2017/746 (primary compliance target for MAILO)" ; mailo:applicablePhase "high_risk_annex_i_samd_omnibus" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:OmnibusAI_RegEU2026_1744 ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2028-08-02"^^xsd:date . mailo:AIAct_v3c_OriginalAnnexIHighRisk a mailo:RegulationVersion ; rdfs:label "Original Article 6(1) / Annex I high-risk application date (superseded)" ; rdfs:comment "Original Article 113 set 2 August 2027 for Article 6(1) Annex I high-risk systems, a year later than the Annex III date. Superseded by Regulation (EU) 2026/1744, which moved it to 2 August 2028."@en ; mailo:applicablePhase "high_risk_annex_i_original" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:Amend_AIAct_v3c_OriginalAnnexIHighRisk ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2027-08-02"^^xsd:date . mailo:AIAct_v4_FRIA_PublicBodies a mailo:RegulationVersion ; rdfs:label "FRIA obligation applies (Art.27)" ; rdfs:comment "Article 27 forms part of Chapter III Section 3 and binds deployers of high-risk AI systems referred to in Article 6(2). Under amended Article 113(c)(i) that Section applies from 2 December 2027. Article 27 does not reach Article 6(1) / Annex I systems, so it does not apply to SaMD taking the Annex I route."@en ; mailo:applicablePhase "fria_public_bodies" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:Amend_AIAct_v4_FRIA_PublicBodies ; mailo:versionOf mailo:AIAct ; mailo:versionStartDate "2027-12-02"^^xsd:date . mailo:Abbott a mailo:Company ; rdfs:label "Abbott Laboratories"@en . mailo:Amend_AIAct_v1_ProhibitedPractices a mailo:LegislativeAmendment ; rdfs:label "Activation event: prohibited_practices_and_general" ; mailo:amendmentDescription "Article 113(a) — Chapters I and II of Regulation (EU) 2024/1689 apply from 2 February 2025." ; mailo:versionStartDate "2025-02-02"^^xsd:date . mailo:Amend_AIAct_v2_GPAIModelObligations a mailo:LegislativeAmendment ; rdfs:label "Activation event: gpai_model_obligations" ; mailo:amendmentDescription "12-month anniversary — GPAI model chapter becomes applicable" ; mailo:versionStartDate "2025-08-02"^^xsd:date . mailo:Amend_AIAct_v2b_PostOmnibusTransitional a mailo:LegislativeAmendment ; rdfs:label "Activation event: post_omnibus_transitional" ; mailo:amendmentDescription "Regulation (EU) 2026/1744 entered into force on 27 July 2026 and its amendments became part of Regulation (EU) 2024/1689. Articles 102 to 110 are expressly applicable from that date under the new point (d) of the third paragraph of Article 113." ; mailo:versionStartDate "2026-07-27"^^xsd:date . mailo:Amend_AIAct_v3_OriginalHighRisk a mailo:LegislativeAmendment ; rdfs:label "Activation event: high_risk_original_deadline" ; mailo:amendmentDescription "Original Article 113 date for Article 6(2) / Annex III high-risk systems. Superseded by Regulation (EU) 2026/1744 before it took effect." ; mailo:versionStartDate "2026-08-02"^^xsd:date . mailo:Amend_AIAct_v3c_OriginalAnnexIHighRisk a mailo:LegislativeAmendment ; rdfs:label "Activation event: high_risk_annex_i_original" ; mailo:amendmentDescription "Original Article 113 date for Article 6(1) / Annex I high-risk systems. Superseded by Regulation (EU) 2026/1744." ; mailo:versionStartDate "2027-08-02"^^xsd:date . mailo:Amend_AIAct_v4_FRIA_PublicBodies a mailo:LegislativeAmendment ; rdfs:label "Activation event: fria_public_bodies" ; mailo:amendmentDescription "Regulation (EU) 2026/1744, amended Article 113(c)(i) — Article 27 fundamental rights impact assessment applies to deployers of Article 6(2) / Annex III high-risk systems from 2 December 2027." ; mailo:versionStartDate "2027-12-02"^^xsd:date . mailo:Amgen a mailo:Company ; rdfs:label "Amgen Inc."@en . mailo:AnnexIIIRoute a mailo:HighRiskRoute ; rdfs:label "Annex III route (Art.6(2)) — listed standalone use case"@en ; rdfs:comment "The system falls within one of the Annex III use cases. Healthcare relevance arises mainly at point 5, covering evaluation of eligibility for essential public assistance benefits and services including healthcare. Art.86 applies here, except to systems under point 2. Obligations for this route apply from 2 December 2027."@en . mailo:AnnexIRoute a mailo:HighRiskRoute ; rdfs:label "Annex I route (Art.6(1)) — product or safety component"@en ; rdfs:comment "SUPERSEDED IN PART (v5.2.2): retained for continuity, but Annex I is not a single set for timing or conformity purposes. Use AnnexISectionARoute for products requiring third-party conformity assessment, which is the branch the MDR and IVDR sit in and the one every medical-device instance in this graph takes."@en, "The system is a product, or a safety component of a product, covered by Union harmonisation legislation listed in Annex I and required to undergo third-party conformity assessment. MDR and IVDR are in Annex I Section A, so Class IIa, IIb and III medical devices incorporating AI take this route automatically. Art.86 and Art.26(11), both scoped to Annex III, DO NOT APPLY. Obligations for this route apply from 2 August 2028."@en . mailo:AnnexISectionARoute a mailo:HighRiskRoute ; rdfs:label "Annex I Section A route — harmonisation legislation requiring third-party conformity assessment"@en ; rdfs:comment "The Art.6(1) route taken by products covered by the Union harmonisation legislation listed in Annex I SECTION A, which requires third-party conformity assessment. The MDR and the IVDR are in Section A, so AI-based medical devices at Class IIa or above under MDR Annex VIII Rule 11 take this route. Two things must be kept apart here. Applicability is set for the Art.6(1)/Annex I branch AS A WHOLE: Art.113 third subparagraph point (c) as amended by Regulation (EU) 2026/1744 brings Chapter III Sections 1-3 to AI systems classified as high-risk under Art.6(1) and Annex I from 2 AUGUST 2028, a one-year deferral from the original 2 August 2027 rather than the two years granted to Annex III; the provision does not key that date to Section A. What Section A status separately determines is the CONFORMITY-ASSESSMENT ARCHITECTURE, since it is the Section A legislation that requires third-party assessment and it is that requirement which Art.43 and its notified-body transition address. Art.86 and Art.26(11), confined to Annex III, do not reach systems on this route. Added in v5.2.2: the previous AnnexIRoute treated Annex I as a single set, which the Omnibus timeline makes untenable."@en . mailo:CJEU_GrandChamber a mailo:Court ; rdfs:label "CJEU Grand Chamber"@en ; rdfs:comment "Highest formation of CJEU; judgments carry maximum precedential weight."@en . mailo:C_201_14 a mailo:CJEUCase ; rdfs:label "C-201/14 Bara"@en ; dcterms:date "2015-10-01"^^xsd:date ; dcterms:identifier "C-201/14", "ECLI:EU:C:2015:638" ; rdfs:comment "ECLI added in v5.2.2, verified against InfoCuria or EUR-Lex plus an independent secondary source rather than supplied from memory."@en, "Inclusion basis (TransposedPrinciple): Health data transferred between public bodies without information; earliest link in the retained chain."@en ; skos:notation "ECLI:EU:C:2015:638" ; mailo:establishesPrinciple mailo:P_Bara_001 ; mailo:hasInclusionBasis mailo:TransposedPrinciple ; mailo:hasStatus "Final"^^xsd:string . mailo:C_511_18 a mailo:CJEUCase ; rdfs:label "C-511/18 La Quadrature du Net"@en ; dcterms:date "2020-10-06"^^xsd:date ; dcterms:identifier "C-511/18", "ECLI:EU:C:2020:791" ; rdfs:comment "ECLI added in v5.2.2, verified against InfoCuria or EUR-Lex plus an independent secondary source rather than supplied from memory."@en, "Inclusion basis (TransposedPrinciple): Bulk data retention. Retained for the proportionality line feeding C-817/19."@en ; skos:notation "ECLI:EU:C:2020:791" ; mailo:citesPrecedent mailo:C_201_14 ; mailo:hasInclusionBasis mailo:TransposedPrinciple ; mailo:hasStatus "Final"^^xsd:string . mailo:C_634_21 a mailo:CJEUCase ; rdfs:label "C-634/21 SCHUFA Holding (Scoring)"@en ; dcterms:date "2023-12-07"^^xsd:date ; dcterms:identifier "C-634/21", "ECLI:EU:C:2023:957" ; rdfs:comment "Inclusion basis (TransposedPrinciple): Consumer credit scoring. The Art.22(1) analysis is transposed to clinical risk-scoring systems."@en ; skos:notation "ECLI:EU:C:2023:957" ; mailo:citesPrecedent mailo:C_817_19 ; mailo:decidedBy mailo:CJEU_FirstChamber ; mailo:establishesPrinciple mailo:P_634_001, mailo:P_634_002, mailo:P_634_003, mailo:P_634_004 ; mailo:hasInclusionBasis mailo:TransposedPrinciple ; mailo:hasOutcome "GDPR Art.22(1) constitutes a prohibition; automated scoring by credit agency = 'decision' triggering Art.22; 'solely' test is functional not formal."^^xsd:string ; mailo:hasStatus "Final — binding precedent"^^xsd:string ; mailo:interpretesArticle mailo:GDPR_Art15_1h, mailo:GDPR_Art22, mailo:GDPR_Art22_1, mailo:GDPR_Art22_2 ; mailo:involvesTechnology mailo:RiskScoringSystem . mailo:C_817_19 a mailo:CJEUCase ; rdfs:label "C-817/19 Ligue des droits humains v. Conseil des ministres"@en ; dcterms:date "2022-06-21"^^xsd:date ; dcterms:identifier "C-817/19", "ECLI:EU:C:2022:491" ; rdfs:comment "A citesPrecedent link to C-634/21 (SCHUFA) was removed in v5.2.0 as temporally impossible: this judgment is of 21 June 2022 and SCHUFA was decided on 7 December 2023. The influence runs the other way."@en, "Inclusion basis (TransposedPrinciple): PNR / counter-terrorism. MAILO transposes the Grand Chamber's automated-decision safeguards to medical AI by structural analogy."@en ; skos:notation "ECLI:EU:C:2022:491" ; mailo:citesPrecedent mailo:C_511_18 ; mailo:decidedBy mailo:CJEU_GrandChamber ; mailo:establishesPrinciple mailo:P_817_001, mailo:P_817_002, mailo:P_817_003, mailo:P_817_004, mailo:P_817_005 ; mailo:hasInclusionBasis mailo:TransposedPrinciple ; mailo:hasOutcome "PNR Directive valid only with strict necessity; black-box ML systems prohibited; mandatory non-automated individual review; health status excluded from automated criteria."^^xsd:string ; mailo:hasStatus "Final — Grand Chamber binding precedent"^^xsd:string ; mailo:interpretesArticle mailo:Charter_Art7, mailo:Charter_Art8 ; mailo:involvesTechnology mailo:AutomatedDecisionSystem, mailo:BlackBoxMLSystem . mailo:Charter_Art7 a mailo:LegalArticle ; rdfs:label "EU Charter Art.7 — Respect for private and family life"@en ; rdfs:comment "Everyone has the right to respect for his or her private and family life, home and communications. Together with Art.8 the constitutional basis against which the Court assessed the PNR Directive in C-817/19 and the retention regimes in C-511/18."@en . mailo:Charter_Art8 a mailo:LegalArticle ; rdfs:label "EU Charter Art.8 — Protection of personal data"@en ; rdfs:comment "Everyone has the right to the protection of personal data concerning him or her; such data must be processed fairly, for specified purposes and on a legitimate basis laid down by law; everyone has the right of access to data collected concerning them and the right to have it rectified; and compliance is subject to control by an independent authority. Art.8(2) is the Charter root of the GDPR Art.15 access right that C-203/22 and C-307/22 interpret; Art.8(3) is the independent-supervision requirement echoed in C-817/19 paras 222-227."@en . mailo:ConformityAssessment a owl:Class ; rdfs:label "ConformityAssessment" ; rdfs:comment "A conformity assessment path under MDR or AI Act" . mailo:ConformityGapConflict a mailo:ConflictType ; rdfs:label "Conformity gap"@en ; rdfs:comment "Two frameworks impose parallel conformity-assessment routes that neither recognises, so satisfying one does not discharge the other (e.g. MDR notified-body assessment alongside AI Act high-risk conformity assessment)."@en . mailo:ConformityRecord a owl:Class ; rdfs:label "ConformityRecord", "Conformity Record"@en ; rdfs:comment "Conformity assessment / technical documentation record", "A conformity assessment or technical documentation record under AI Act Annex IV / MDR Annex II."@en . mailo:DGA a mailo:EURegulation, mailo:RegulatoryFramework ; rdfs:label "DGA 2022/868"@en ; eli:id_local "32022R0868" ; dcterms:date "2022-05-30"^^xsd:date ; dcterms:identifier "2022/868/EU" ; rdfs:comment "Data Governance Act. Applies from 2023-09-24. Creates data intermediaries, data altruism organisations, and re-use frameworks for public sector health data."@en ; mailo:coversArticle mailo:DGA_Art10, mailo:DGA_Art12, mailo:DGA_Art16, mailo:DGA_Art2, mailo:DGA_Art29, mailo:DGA_Art3 . mailo:DGA_Art10 a mailo:LegalArticle ; rdfs:label "DGA Art.10 — Data intermediation services (scope)"@en ; rdfs:comment "Defines which services fall within the data intermediation regime. CORRECTED in v5.1.0: the neutrality obligation previously described here belongs to Art.12; notification is Art.11."@en . mailo:DGA_Art12 a mailo:LegalArticle ; rdfs:label "DGA Art.12 — Conditions for providing data intermediation services (neutrality)"@en ; rdfs:comment "Art.12(1)(a): the provider shall not use the data for which it provides intermediation services for purposes other than putting them at the disposal of data users, and shall provide the service through a separate legal person. Art.12(1)(c) restricts use of activity metadata to service development. Added in v5.1.0 to carry the neutrality obligation previously misattributed to Art.10."@en . mailo:DGA_Art16 a mailo:LegalArticle ; rdfs:label "DGA Art.16 — National arrangements for data altruism"@en ; rdfs:comment "Member States may put in place organisational or technical arrangements to facilitate data altruism. Registration of recognised data altruism organisations, which may collect health data voluntarily shared for AI research, is governed by Arts.18-19; such entities are separate from commercial data intermediation services. Scope clarified in v5.1.0: the earlier label attributed the registration regime to Art.16."@en . mailo:DGA_Art2 a mailo:LegalArticle ; rdfs:label "DGA Art.2 — Definitions"@en ; rdfs:comment "Defines the terms used throughout the Regulation, including 'data', 'data holder', 'data user', 'data intermediation service' at Art.2(11) and 'data altruism'. CORRECTED in v5.1.0: previously labelled 'Re-use of Public Sector Health Data' and described as the Chapter II framework. Chapter II on re-use of protected public sector data runs from Art.3, with the categories at Art.3 and the conditions for re-use at Art.5."@en . mailo:DGA_Art29 a mailo:LegalArticle ; rdfs:label "DGA Art.29 — European Data Innovation Board"@en ; rdfs:comment "Expert group established by the Commission, comprising the competent authorities for data intermediation services and for data altruism registration from all Member States, the EDPB, the EDPS, ENISA and others. Its tasks are set out in Art.30. Includes guidance on interoperability between health data repositories used for AI training. Renumbered in v5.1.0: previously recorded as Art.31, which governs international access and transfer (Chapter VII)."@en . mailo:DGA_Art3 a mailo:LegalArticle ; rdfs:label "DGA Art.3 — Categories of protected public sector data eligible for re-use"@en ; rdfs:comment "Opens Chapter II (Arts 3-9) by fixing the categories of protected data held by public sector bodies whose re-use the Regulation conditions, including data subject to statistical or commercial confidentiality and personal data. The substantive re-use conditions, including anonymisation and the secure processing environment, are in Art.5. Applies to AI training use cases at hospitals and research institutes. Corrected in v5.1.0 Round 5: previously recorded as Art.2, the Definitions article."@en . mailo:DPIARecord a owl:Class ; rdfs:label "DPIARecord", "DPIA Record"@en ; rdfs:comment "Data Protection Impact Assessment document (GDPR Art.35; C-817/19)", "A Data Protection Impact Assessment document required for high-risk automated processing (GDPR Art.35; C-817/19 P_817_005)."@en . mailo:DataAct a mailo:EURegulation, mailo:RegulatoryFramework ; rdfs:label "Data Act 2023/2854"@en ; eli:id_local "32023R2854" ; dcterms:date "2023-12-13"^^xsd:date ; dcterms:identifier "2023/2854/EU" ; rdfs:comment "Data Act. Applies from 2025-09-12. Governs data sharing obligations for IoT-generated data (medical devices, wearables) and B2G/B2B data access for AI development."@en ; mailo:coversArticle mailo:DataAct_Art14, mailo:DataAct_Art15, mailo:DataAct_Art17, mailo:DataAct_Art21, mailo:DataAct_Art23, mailo:DataAct_Art3, mailo:DataAct_Art4, mailo:DataAct_Art5 . mailo:DataAct_Art14 a mailo:LegalArticle ; rdfs:label "Data Act Art.14 — Obligation to make data available on the basis of exceptional need"@en ; rdfs:comment "Where a public sector body, the Commission, the ECB or a Union body demonstrates an exceptional need under Art.15, the data holder shall make the data available. The obligation Art.15 qualifies and Art.17 procedurally implements. Added in v5.1.0: MAILO recorded the Chapter V machinery without the provision that creates the duty."@en ; mailo:hasPendingAmendment mailo:DigitalOmnibus_DataAct_ChapterV . mailo:DataAct_Art15 a mailo:LegalArticle ; rdfs:label "Data Act Art.15 — Exceptional need to use data (B2G)"@en ; rdfs:comment "Defines when a public sector body has an exceptional need justifying a request for data under Chapter V. CORRECTED in v5.1.0: this individual was labelled 'Trade Secret Protection'. The Data Act's trade-secret safeguards are in Art.4(6)-(8) (data holder to user) and Art.5(9)-(11) (user to third party), not Art.15."@en . mailo:DataAct_Art17 a mailo:LegalArticle ; rdfs:label "Data Act Art.17 — Requests by public sector bodies for data to be made available"@en ; rdfs:comment """The procedural requirements for a request under Chapter V: what the request must specify, and the cross-border route under Art.17(3) by which a public sector body intending to request data from a holder established in another Member State must first notify that Member State's competent authority designated under Art.37, which then examines the request against Art.17. CORRECTED in v5.1.0: the earlier entry described Art.17 as 'B2G Data Sharing (Public Emergency)', asserting that governments may require medical AI companies to share health data in a public health emergency. Public emergency is a ground of exceptional need under Art.15, not the subject of Art.17. Art.14 states the obligation itself. The substantive claim about epidemic surveillance AI therefore rested on the wrong provision."""@en . mailo:DataAct_Art21 a mailo:LegalArticle ; rdfs:label "Data Act Art.21 — Sharing of data obtained on exceptional need with research organisations"@en ; rdfs:comment "A public sector body, the Commission, the ECB or a Union body may share data received under Chapter V with individuals or organisations conducting scientific research or analytics compatible with the purpose of the request, or with national statistical institutes and Eurostat. Recipients assume the obligations in Art.17(3) and Art.19, and organisations under significant commercial influence likely to result in preferential access to results are excluded. Relevant to medical-AI research using data obtained through the exceptional-need route. Added in v5.1.0."@en . mailo:DataAct_Art23 a mailo:LegalArticle ; rdfs:label "Data Act Art.23 — Switching Cloud Services"@en ; rdfs:comment "Medical AI providers using cloud infrastructure must enable switching between cloud providers without data lock-in. Relevant for hospital AI SaaS deployment models."@en . mailo:DataAct_Art3 a mailo:LegalArticle ; rdfs:label "Data Act Art.3 — Obligation to make product data accessible BY DESIGN"@en ; rdfs:comment """Connected products shall be designed and manufactured, and connected services designed and provided, so that product data and related service data are, by default, directly accessible to the user in a simple, secure, free-of-charge, comprehensive, structured, commonly used and machine-readable format. Art.3(2)-(3) add pre-contractual information duties covering the nature, format and estimated volume of the data, whether it is generated in real time, and where it is stored. Applies to products placed on the market from 12 September 2026. CORRECTED in v5.1.0: the earlier entry described Art.3 as an 'IoT Data Access Right ... enforceable against manufacturers'. That is Art.4. Art.3 is a design and by-default-accessibility obligation; the user's right to obtain data from the data holder is separate and has applied since 12 September 2025. For connected medical devices the distinction is practical, not merely formal: a device placed on the market before September 2026 is outside the design duty while its holder is already subject to the access duty."""@en . mailo:DesignerTransposition a mailo:EncodingBasis ; rdfs:label "Designer transposition"@en ; rdfs:comment "The proposition is inferred across instruments or contexts by the ontology designer and is not held by any court. Must never be cited as case law. Example: the bridge from C-817/19 independent prior review to a GDPR Art.35 DPIA duty."@en . mailo:Dexcom a mailo:Company ; rdfs:label "Dexcom Inc."@en . mailo:DigitalOmnibus_DataAct_ChapterV a mailo:PendingAmendment ; rdfs:label "Digital Omnibus — proposed consolidation of Data Act Chapter V into a new Art.15a"@en ; mailo:amendmentProcedureRef "2025/0360(COD)"^^xsd:string ; mailo:amendmentProposalRef "COM(2025) 837 final"^^xsd:string ; mailo:legislativeStatus "council_position"^^xsd:string ; mailo:proposedEffect "Deletes Data Act Arts 14 and 15 and consolidates the business-to-government regime into a new Art.15a, narrowing the trigger from 'exceptional need' to public emergencies only. Not adopted as at the verification date: the AI limb of the Digital Omnibus was severed and adopted separately as Regulation (EU) 2026/1744, while the data limb remains in negotiation. MAILO therefore continues to encode the numbering of Regulation (EU) 2023/2854 as published."^^xsd:string ; mailo:proposedRenumberingTo "Data Act Art.15a"^^xsd:string ; mailo:statusVerifiedAsAt "2026-09-05"^^xsd:date . mailo:EHDS a mailo:EURegulation, mailo:RegulatoryFramework ; rdfs:label "EHDS Regulation 2025/327"@en ; eli:id_local "32025R0327" ; dcterms:date "2025-02-11"^^xsd:date ; dcterms:identifier "2025/327/EU" ; rdfs:comment "European Health Data Space. Governs primary use (MyHealth@EU, Art.23) and secondary use (HealthData@EU, Art.75) of health data for research, innovation and AI training. Adopted 11 February 2025, published in the OJ on 5 March 2025, in force 26 March 2025; most secondary-use obligations apply from 26 March 2029, with genetic and clinical-trial data from 26 March 2031. All article references were renumbered in v5.1.0 from COM(2022) 197 proposal numbering to the published text."@en ; mailo:coversArticle mailo:EHDS_Art23, mailo:EHDS_Art51, mailo:EHDS_Art52, mailo:EHDS_Art53, mailo:EHDS_Art54, mailo:EHDS_Art57, mailo:EHDS_Art68, mailo:EHDS_Art69, mailo:EHDS_Art71, mailo:EHDS_Art73, mailo:EHDS_Art75, mailo:EHDS_Art92 . mailo:EHDS_Art23 a mailo:LegalArticle ; rdfs:label "EHDS Art.23 — MyHealth@EU (cross-border infrastructure for PRIMARY use)"@en ; rdfs:comment "The cross-border infrastructure for primary use, to be distinguished from HealthData@EU at Art.75, which serves secondary use. Added in v5.1.0 because the two were previously conflatable: the superseded Art.50 entry described HealthData@EU using proposal numbering with no primary-use counterpart recorded."@en . mailo:EHDS_Art51 a mailo:LegalArticle ; rdfs:label "EHDS Art.51 — Minimum categories of electronic health data for secondary use"@en ; rdfs:comment "Lists what health data holders must make available: EHR data, data on factors influencing health, pathogen data, healthcare-related administrative and claims data, human genetic, epigenomic and genomic data, other molecular omics data, automatically generated personal EHD from medical devices, data from wellness applications, health-professional and health-facility data, population-based health-data registries, health-data registries and clinical-trial data. Art.51(h) and (n) in particular govern device-generated data, where industry commentary argues the scope should reach only processed human-readable information rather than raw machine data used internally to derive diagnostic scores. Added in v5.1.0."@en . mailo:EHDS_Art52 a mailo:LegalArticle ; rdfs:label "EHDS Art.52 — IP rights, trade secrets and regulatory data protection in secondary use"@en ; rdfs:comment "The provision governing the trade-secret / access conflict in the health domain, and the one most directly relevant to MAILO's transparency-secrecy tension. Datasets protected by IP rights, trade secrets or regulatory data protection may still be made available for secondary use provided the health data holder identifies and substantiates the protected elements. The health data access body must then take all specific, appropriate and proportionate legal, organisational and technical measures, which may include contractual arrangements, and may refuse access where a serious risk of infringement cannot be addressed satisfactorily. Art.52(3) covers technical measures. The decision chain it leaves to be operationalised runs: protected element, substantiated claim, access scenario, infringement pathway, safeguard, residual risk, permit or conditional permit or refusal. Added in v5.1.0: MAILO had encoded five EHDS articles and omitted this one."@en . mailo:EHDS_Art53 a mailo:LegalArticle ; rdfs:label "EHDS Art.53 — Permitted purposes for secondary use"@en ; rdfs:comment """Health data access bodies may grant access to Art.51 data only where processing is necessary for one of the listed purposes: public interest in public and occupational health; policy-making and regulatory activities; official statistics; education and training in the health or care sector; and scientific research related to health or care that contributes to public health or health technology assessment or ensures high quality and safety of healthcare, medicinal products or medical devices. Art.53(1)(e)(ii) is of first-order importance to MAILO and was previously absent: it expressly names 'training, testing and evaluation of algorithms, INCLUDING IN MEDICAL DEVICES, IN VITRO DIAGNOSTIC MEDICAL DEVICES, AI SYSTEMS AND DIGITAL HEALTH APPLICATIONS' as a permitted secondary-use purpose. This is the express legal pathway for training medical AI on EHDS data, and the point at which the AI Act Art.10 data-governance duties and the EHDS access regime meet. Added in v5.1.0."""@en . mailo:EHDS_Art54 a mailo:LegalArticle ; rdfs:label "EHDS Art.54 — Prohibited secondary use"@en ; rdfs:comment "Secondary use may not be granted for purposes adverse to the natural persons whose data it is, including decisions detrimental to them or to groups, advertising and marketing, developing products or services that may harm individuals or public health, and decisions on insurance exclusion or premium adjustment. The counterpart to Art.53 and the boundary against which any medical-AI training purpose must also be tested. Added in v5.1.0."@en . mailo:EHDS_Art57 a mailo:LegalArticle ; rdfs:label "EHDS Art.57 — Tasks of health data access bodies"@en ; rdfs:comment "Sets out the HDAB's functions, including deciding on data access applications, issuing and withdrawing data permits, and, under Art.57(c), balancing the rights and interests of health data holders against those of health data users. Art.57(c) is the counterpart to the Art.52 protection duty and is where the residual-risk judgment is institutionally located. Added in v5.1.0."@en . mailo:EHDS_Art68 a mailo:LegalArticle ; rdfs:label "EHDS Art.68 — Data permit (secondary use authorisation)"@en ; rdfs:comment "Health data access bodies issue data permits authorising secondary use of electronic health data for research, innovation and policy. Access is granted within a secure processing environment (Art.73) from which raw data cannot be downloaded; a health data request under Art.69 yields only an anonymised statistical answer. Renumbered in v5.1.0: previously recorded as Art.33, which is COM(2022) 197 proposal numbering."@en . mailo:EHDS_Art69 a mailo:LegalArticle ; rdfs:label "EHDS Art.69 — Health data request (anonymised statistical answer)"@en ; rdfs:comment "Where a health data user seeks only an answer in anonymised statistical form, a health data request may be made instead of a data permit; the HDAB supplies the result rather than access to the underlying data. The least intrusive route in the access hierarchy and a candidate safeguard where a protected element under Art.52 cannot be adequately mitigated within a secure processing environment. Added in v5.1.0."@en . mailo:EHDS_Art71 a mailo:LegalArticle ; rdfs:label "EHDS Art.71 — Right to opt out of secondary use"@en ; rdfs:comment "Natural persons may opt out of the processing of their personal electronic health data for secondary use. GDPR supervisory authorities are competent for monitoring and enforcing this right. Creates tension with AI Act Art.10 data governance obligations, since opt-out reduces dataset representativeness. Renumbered in v5.1.0: previously recorded as Art.35, which is COM(2022) 197 proposal numbering."@en . mailo:EHDS_Art73 a mailo:LegalArticle ; rdfs:label "EHDS Art.73 — Secure processing environment"@en ; rdfs:comment "Secondary use of electronic health data must take place within a secure processing environment under the control of a health data access body, from which raw data cannot be downloaded. This is the medical-domain reconciliation route for the trade-secret / transparency tension: disclosure to a controlled environment rather than to the requester. Implementing acts under Art.73(5) apply from 26 March 2029. Added in v5.1.0."@en . mailo:EHDS_Art75 a mailo:LegalArticle ; rdfs:label "EHDS Art.75 — HealthData@EU (cross-border infrastructure for secondary use)"@en ; rdfs:comment "HealthData@EU connects national contact points for secondary use, enabling EU-wide access to datasets through a single channel without raw data transfer. Not to be confused with MyHealth@EU (Art.23), the corresponding infrastructure for primary use. Third-country participation becomes possible under Art.75(5) from 26 March 2035. Renumbered in v5.1.0: previously recorded as Art.50, which is COM(2022) 197 proposal numbering."@en . mailo:EHDS_Art92 a mailo:LegalArticle ; rdfs:label "EHDS Art.92 — European Health Data Space Board (EHDS Board)"@en ; rdfs:comment "Independent advisory body facilitating cooperation between Member States and the Commission, with two representatives per Member State (one for primary and one for secondary use). Its remit is consultative. Renumbered in v5.1.0: previously recorded as Art.68 — a particularly hazardous error, since Art.68 exists but governs the data permit, so the citation resolved to a real provision about something else."@en . mailo:EPC_Art52 a mailo:LegalArticle ; rdfs:label "EPC Art.52 — Patentable inventions"@en ; rdfs:comment """Art.52(1): European patents are granted for any inventions in all fields of technology, provided they are new, involve an inventive step and are susceptible of industrial application. Art.52(2) excludes, among others, mathematical methods and programs for computers; Art.52(3) limits that exclusion to such subject matter 'AS SUCH'. SCOPE NOTE added in v5.1.0: no MAILO case now interprets this article. All three EPO instances were re-anchored to Arts.56, 83 and 84 after the audit found the Art.52 attribution wrong in each. The provision is retained because the 'as such' limitation is the first hurdle any computer-implemented invention clears — G 1/19 confirms it is passed by any technical means — but it should be understood as background rather than as a provision MAILO's case law turns on."""@en . mailo:EPC_Art84 a mailo:LegalArticle ; rdfs:label "EPC Article 84"@en ; rdfs:comment "Claims — clarity and support by the description."@en . mailo:EPO_EnlargedBoA a mailo:Court ; rdfs:label "EPO Enlarged Board of Appeal"@en ; rdfs:comment "Ensures uniform application of the EPC and decides points of law referred to it. Declared in v5.1.0: it was already in use as the decidedBy value of G 1/19 but had never been typed."@en . mailo:EPO_T0161_18 a mailo:EPOProceding, mailo:EPOProceeding ; rdfs:label "EPO T 0161/18 — AI Medical Device Disclosure"@en ; dcterms:date "2020-05-12"^^xsd:date ; dcterms:identifier "T 0161/18" ; rdfs:comment "Inclusion basis (DirectSubjectMatter): Artificial neural network deriving cardiac output from an arterial pressure curve."@en ; mailo:decidedBy mailo:EPO_BoA ; mailo:hasInclusionBasis mailo:DirectSubjectMatter ; mailo:hasOutcome "Technical Board of Appeal 3.5.05, 12 May 2020 (Aequivalenter Aortendruck/ARC SEIBERSDORF), application 06804383.5: method of determining cardiac output from a peripheral arterial blood pressure curve using an artificial neural network. The APPLICATION WAS REFUSED (not a granted patent revoked) because the training of the network could not be reworked by the skilled person - neither suitable input data nor at least one suitable training data set was disclosed (Art.83); the claim was also held to lack inventive step for the same reason (Art.56). Leading EPO authority on training-data disclosure for AI inventions, reflected in the 2024 EPO Guidelines and followed in T 1191/19."^^xsd:string ; mailo:interpretesArticle mailo:EPC_Art56, mailo:EPC_Art83 ; mailo:involvesTechnology mailo:MedicalAISystem . mailo:EP_2352431 a mailo:EuropeanPatent ; rdfs:label "EP 2 352 431 B1"@en ; rdfs:comment "Hologic breast imaging AI-assisted reading technology."@en . mailo:EP_2671173 a mailo:EuropeanPatent ; rdfs:label "EP 2 671 173"@en ; rdfs:comment "Suinno AI-based pedestrian navigation method."@en . mailo:EP_3346403 a mailo:EuropeanPatent ; rdfs:label "EP 3 346 403"@en ; rdfs:comment "Breast cancer recurrence prediction genomic expression test."@en . mailo:EP_3435866 a mailo:EuropeanPatent ; rdfs:label "EP 3 435 866"@en ; rdfs:comment "Dexcom continuous glucose monitoring system."@en . mailo:GDPR_Art22_4 a mailo:LegalArticle ; rdfs:label "GDPR Art.22(4) — Prohibition on basing such decisions on special category data"@en ; rdfs:comment "Decisions referred to in Art.22(2) SHALL NOT be based on special categories of personal data referred to in Art.9(1) — which include health data — unless point (a) (explicit consent) or point (g) (substantial public interest) of Art.9(2) applies AND suitable measures to safeguard rights, freedoms and legitimate interests are in place. CORRECTED in v5.1.0: the earlier text described this as 'enhanced protection', which understates a prohibition with two narrow gateways. Given that health data is squarely within Art.9(1), this is one of the most consequential provisions in MAILO's scope, and the earlier phrasing repeated the error class corrected in P_817_004: a prohibition recorded as a graduated safeguard."@en . mailo:GDPR_Art35 a mailo:LegalArticle ; rdfs:label "GDPR Article 35"@en ; rdfs:comment "Data Protection Impact Assessment (DPIA)."@en . mailo:GDPR_Art9 a mailo:LegalArticle ; rdfs:label "GDPR Art.9 — Processing of special categories of personal data"@en ; rdfs:comment "Art.9(1) PROHIBITS processing of special categories of personal data, including data concerning health. Art.9(2) sets out the exhaustive exceptions, of which (a) explicit consent and (g) substantial public interest are the two that can also unlock Art.22(4). Scope clarified in v5.1.0: the earlier comment described the article neutrally without recording that Art.9(1) is a prohibition."@en ; rdfs:seeAlso dpv:SpecialCategoryData . mailo:GDPR_v1_FullyApplicable a mailo:RegulationVersion ; rdfs:label "GDPR fully applicable — all provisions in force" ; mailo:applicablePhase "fully_applicable" ; mailo:isCurrentVersion true ; mailo:versionOf mailo:GDPR ; mailo:versionStartDate "2018-05-25"^^xsd:date . mailo:GDPR_v2_OmnibusProposed_Art22 a mailo:RegulationVersion ; rdfs:label "GDPR Art.22 Omnibus-proposed version — permission list framing (PROSPECTIVE, NOT ADOPTED)" ; mailo:applicablePhase "art22_omnibus_proposed_NOT_ADOPTED" ; mailo:isCurrentVersion false ; mailo:triggeredBy mailo:OmnibusGDPR_Art22_Proposal_20251119 ; mailo:versionOf mailo:GDPR . mailo:GXDBio a mailo:NPE ; rdfs:label "GXD-Bio Corporation (NPE)"@en . mailo:Hologic a mailo:Company ; rdfs:label "Hologic Inc."@en . mailo:MDR_AnnexI a mailo:LegalArticle ; rdfs:label "MDR Annex I — General safety and performance requirements"@en ; rdfs:comment "The GSPRs a device must meet. Chapter II section 17 governs electronic programmable systems and software: devices incorporating such systems shall be designed to ensure repeatability, reliability and performance in line with their intended use, software shall be developed and manufactured in accordance with the state of the art taking account of the development lifecycle, risk management including information security, verification and validation. The provision against which AI-specific validation expectations are read. Added in v5.1.0."@en . mailo:MDR_AnnexVIII_Rule11 a mailo:LegalArticle ; rdfs:label "MDR Annex VIII Rule 11 — Classification of software"@en ; rdfs:comment """Software intended to provide information used to take decisions with diagnosis or therapeutic purposes is Class IIa, rising to Class III where such decisions may cause death or an irreversible deterioration of health, and to Class IIb where they may cause a serious deterioration or surgical intervention. Software intended to monitor physiological processes is Class IIa, or IIb where monitoring vital parameters whose variation could result in immediate danger. All other software is Class I. This is the linchpin connecting MDR to the AI Act and was absent from MAILO. Rule 11 places most diagnostic and decision-support software at Class IIa or above, which requires notified-body conformity assessment, which in turn makes the system high-risk under AI Act Art.6(1) by the ANNEX I route — and therefore outside the Art.86 right to explanation. The dual-track burden and the Art.86 exclusion both follow from this classification rule. Added in v5.1.0."""@en ; mailo:triggersRoute mailo:AnnexIRoute, mailo:AnnexISectionARoute . mailo:MDR_Art15 a mailo:LegalArticle ; rdfs:label "MDR Art.15 — Person Responsible for Regulatory Compliance"@en ; rdfs:comment "MAILO OPERATIONALISATION: the requirement that the PRRC hold expertise in medical AI regulation is the designer's extension. Art.15 specifies qualification by diploma and experience in regulatory affairs or quality management for medical devices, without an AI-specific criterion. Recorded as such in v5.1.0."@en, "Manufacturers must designate a person responsible for regulatory compliance (PRRC). For AI-SaMD companies this person must have expertise in medical AI regulation."@en . mailo:MDR_Art61 a mailo:LegalArticle ; rdfs:label "MDR Art.61 — Clinical Evaluation"@en ; rdfs:comment "MAILO OPERATIONALISATION: the requirement of AI-specific performance validation covering bias, drift and edge cases is the ontology designer's reading of what clinical evaluation entails for AI-SaMD, not language found in Art.61. Recorded as such in v5.1.0."@en, "Medical device manufacturers must conduct clinical evaluation demonstrating safety and performance. For AI-SaMD, AI-specific performance validation (bias, drift, edge cases) is required."@en . mailo:MDR_Art83 a mailo:LegalArticle ; rdfs:label "MDR Art.83 — Post-Market Surveillance"@en ; rdfs:comment "MAILO OPERATIONALISATION: continuous monitoring of model drift and performance degradation in deployment populations is the designer's reading of what PMS entails for AI-SaMD. Art.83 requires a post-market surveillance system proportionate to risk class and device type; it does not name drift. Recorded as such in v5.1.0."@en, "PMS system must collect real-world performance data. For AI-SaMD: continuous monitoring of model drift and performance degradation in deployment populations."@en . mailo:MedicalImagingAI a owl:Class ; rdfs:label "MedicalImagingAI", "Medical Imaging AI"@en ; rdfs:comment "AI system for medical image analysis" ; rdfs:subClassOf mailo:MedicalAISystem ; mailo:hasRisk mailo:Risk_DataActIoT, mailo:Risk_EPOUPC_Divergence . mailo:Microsoft a mailo:Company ; rdfs:label "Microsoft Corporation"@en . mailo:MyriadGenetics a mailo:Company ; rdfs:label "Myriad Genetics Inc."@en . mailo:NPE a owl:Class ; rdfs:label "NPE", "Non-Practising Entity"@en ; rdfs:comment "Non-Practising Entity — patent assertion vehicle" ; rdfs:subClassOf mailo:Company . mailo:ObiterDictum a owl:Class ; rdfs:label "ObiterDictum", "Obiter Dictum"@en ; rdfs:comment "Persuasive (non-binding) judicial observation", "Persuasive (non-binding) judicial observation."@en ; rdfs:subClassOf mailo:LegalPrinciple . mailo:OmnibusAI_ProvisionalAgreement_20260507 a mailo:LegislativeAmendment ; rdfs:label "Digital Omnibus on AI — provisional agreement of 7 May 2026 (superseded)"@en ; rdfs:comment "Retained as legislative history. This node no longer triggers any operative regulation version; the operative basis is mailo:OmnibusAI_RegEU2026_1744."@en ; mailo:amendmentDescription "Provisional political agreement of 7 May 2026 on the Digital Omnibus on AI, reached before formal adoption. Its conditional standards-availability mechanism is not the adopted rule and has no operative effect. Superseded by Regulation (EU) 2026/1744." ; mailo:versionStartDate "2026-05-07"^^xsd:date . mailo:OmnibusGDPR_Art22_Proposal_20251119 a mailo:LegislativeAmendment ; rdfs:label "Digital Omnibus Regulation — GDPR Art.22 Reframing Proposal (NOT YET ADOPTED)" ; mailo:amendmentDescription "Commission proposal (2025-11-19) to replace GDPR Art.22(1)-(2) prohibition structure with an exhaustive positive permission list for solely automated decision-making. EDPB-EDPS Joint Opinion 2/2026 (2026-02-11) strongly opposes the reframing, citing CJEU confirmation of Art.22 as a prohibition in principle (C-634/21, C-817/19). Status: UNADOPTED — in trilogue negotiation as of 2026-05-10. MAILO SHACL shapes remain grounded in current prohibition structure. If adopted, NodeShapes C634_Art22ProhibitionShape and related shapes require reformulation to encode positive permission conditions rather than absence of exception." ; mailo:versionStartDate "2025-11-19"^^xsd:date . mailo:PLD_Art10 a mailo:LegalArticle ; rdfs:label "PLD Art.10 — Burden of proof (presumptions of defectiveness AND of causal link)"@en ; rdfs:comment "Art.10 carries BOTH presumptions. Art.10(2): defectiveness is presumed where the defendant fails to disclose evidence, the product does not comply with mandatory safety requirements, or an obvious malfunction occurred. Art.10(4): defectiveness and/or the causal link are presumed where the claimant faces excessive difficulty owing to technical or scientific complexity and has demonstrated that the defect or the link is probable — the limb of most consequence for medical AI, expressly contemplated at recital 48. Art.10(5) preserves the defendant's right to rebut."@en . mailo:PLD_Art11 a mailo:LegalArticle ; rdfs:label "PLD Art.11 — Exemption from liability"@en ; rdfs:comment "Grounds on which an economic operator escapes liability, the burden resting on the operator: the product was not placed on the market by them, the defect probably did not exist when the product was placed on the market, or the defect is due to compliance with mandatory legal requirements. The state-of-the-art (development risk) defence sits at Art.11(1)(e), and Member States may derogate from it under Art.18(2). CORRECTED in v5.1.0: this individual previously carried the causation presumption, which is Art.10(4), not Art.11."@en . mailo:PLD_Art4 a mailo:LegalArticle ; rdfs:label "PLD Art.4 — Definitions ('product' includes software and AI systems)"@en ; rdfs:comment """Defines 'product' as all movables, even if integrated into or interconnected with another movable or immovable, including electricity, digital manufacturing files, raw materials and SOFTWARE. The provision that brings standalone software and AI systems within the strict-liability regime, and the basis on which SaMD under the MDR is simultaneously a product under the PLD. Free and open-source software developed or supplied outside a commercial activity is excluded. CORRECTED in v5.1.0: this individual previously carried the label 'Strict Liability for Defective Products' and the defectiveness test. Strict liability and the right to compensation are Art.5; the test that a product is defective where it does not provide the safety a person is entitled to expect is Art.7. The software-as-product content previously recorded under Art.7 belongs here. The two entries were in effect transposed."""@en . mailo:PLD_Art6 a mailo:LegalArticle ; rdfs:label "PLD Art.6 — Damage"@en ; rdfs:comment "The recoverable heads of damage: death or personal injury including medically recognised harm to psychological health; damage to or destruction of property other than the defective product itself, excluding property used exclusively for professional purposes; and destruction or corruption of data not used for professional purposes. Added in v5.1.0."@en . mailo:PLD_Art9 a mailo:LegalArticle ; rdfs:label "PLD Art.9 — Disclosure Obligations (Evidence Facilitation)"@en ; rdfs:comment "Courts may order defendants to disclose evidence about high-risk AI systems; creates procedural discovery mechanism absent in AILD."@en . mailo:P_203_001 a mailo:RatioDecidendi ; rdfs:label "P-203/22-001: Meaningful Explainability Standard"@en ; rdfs:comment "Art.15(1)(h) requires actionable explanations in non-technical language, identifying which inputs affected the output and the main reasoning path. Abstract mathematical formulas are insufficient."@en, "Encoding basis (DesignerOperationalisation): Para 61 requires the data subject to understand which personal data were used and in what way. identifiesInputFactors is MAILO's predicate for that standard."@en ; mailo:appliesTo mailo:AutomatedDecisionSystem, mailo:MedicalAISystem ; mailo:hasEncodingBasis mailo:DesignerOperationalisation ; mailo:hasPrecedenceLevel "binding_ratio"^^xsd:string ; mailo:triggersObligation mailo:ExplainabilityObligation . mailo:P_307_001 a mailo:RatioDecidendi ; rdfs:label "P-307/22-001: Free first copy of medical records under GDPR Art.15(3)"@en ; rdfs:comment "A patient is entitled to a first copy of his or her medical records free of charge, without having to give reasons, and to a faithful and intelligible reproduction extending to full documents where that is necessary to understand the personal data they contain (Arts.12(5), 15(1) and 15(3) GDPR). CORRECTED in v5.1.0: the earlier text described an 'unconditional' right covering 'all AI-generated diagnostic outputs'. The judgment concerns neither AI nor an unconditional entitlement — Art.15(3) allows a reasonable fee for further copies, and Art.15(4) preserves the rights of others."@en ; mailo:appliesTo mailo:AutomatedDecisionSystem ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "binding_ratio"^^xsd:string ; mailo:triggersObligation mailo:DisclosureObligation, mailo:TransparencyObligation . mailo:P_634_001 a mailo:RatioDecidendi ; rdfs:label "P-634/21-001: Broad 'Decision' Concept"@en ; rdfs:comment "A 'decision' under Art.22 encompasses any act producing legal or similarly significant effects. Automated scoring creating probabilities relied upon by third parties constitutes a 'decision'."@en ; mailo:appliesTo mailo:AutomatedDecisionSystem, mailo:ClinicalDecisionSupport, mailo:RiskScoringSystem ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "binding_ratio"^^xsd:string . mailo:P_634_002 a mailo:RatioDecidendi ; rdfs:label "P-634/21-002: Functional 'Solely' Test"@en ; rdfs:comment "C-634/21 paras 49-50: where an automated output plays a determining role in the resulting decision, establishing that output is itself a decision under Art.22(1). The Court did not set out a rubber-stamping or substantive-review standard; MAILO operationalises the determining-role test through a human-asserted substantive-oversight proxy, which is the ontology designer's construction."@en, "Encoding basis (DesignerOperationalisation): Paras 49-50 state a determining-role test. The substantive-oversight predicate is MAILO's proxy for it; the Court set no rubber-stamping standard."@en ; mailo:appliesTo mailo:AutomatedDecisionSystem, mailo:ClinicalDecisionSupport, mailo:MedicalImagingAI ; mailo:hasEncodingBasis mailo:DesignerOperationalisation ; mailo:hasPrecedenceLevel "binding_ratio"^^xsd:string ; mailo:triggersObligation mailo:HumanOversightObligation . mailo:P_634_003 a mailo:RatioDecidendi ; rdfs:label "P-634/21-003: Multi-Actor ADM Chain Gap-Filling"@en ; rdfs:comment "The prohibition in Art.22 cannot be circumvented by splitting the automated decision across multiple actors (scoring agency + final decision maker). Both may bear responsibility."@en ; mailo:appliesTo mailo:AutomatedDecisionSystem ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "binding_ratio"^^xsd:string . mailo:P_634_004 a mailo:RatioDecidendi ; rdfs:label "P-634/21-004: Art.22 = Prohibition Rule"@en ; rdfs:comment "Art.22(1) is a prohibition rule, not merely a right to object. Exceptions in Art.22(2) are exhaustive; safeguards in Art.22(3) are mandatory when an exception applies."@en, "Encoding basis (DesignerOperationalisation): The characterisation of Art.22(1) as a prohibition rather than a right to be invoked is not anchored to its own paragraphs and remains contested in the literature. Recorded as the designer's reading pending verification."@en, "UNANCHORED (v5.2.2): the characterisation of GDPR Art.22(1) as a prohibition operating of its own force, rather than a right the data subject must invoke, is not tied to identified paragraphs of C-634/21 and was not settled by that judgment; the reading remains contested. Recorded as the ontology designer's position, not as a holding."@en ; mailo:hasEncodingBasis mailo:DesignerOperationalisation ; mailo:hasPrecedenceLevel "persuasive_obiter"^^xsd:string . mailo:P_817_001 a mailo:RatioDecidendi ; rdfs:label "P-817/19-001: Mandatory Non-Automated Individual Review"@en ; rdfs:comment "Any positive match from automated PNR analysis must be subject to individual non-automated review before any measure is taken. Establishes constitutional minimum for meaningful human intervention."@en ; mailo:appliesTo mailo:AutomatedDecisionSystem, mailo:MedicalAISystem ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "constitutional_grand_chamber"^^xsd:string ; mailo:triggersObligation mailo:HumanOversightObligation . mailo:P_817_002 a mailo:RatioDecidendi ; rdfs:label "P-817/19-002: Black-Box ML Prohibition"@en ; rdfs:comment "Non-transparent, self-learning ML risk assessment systems may not be used for decisions with significant impact. Pre-determined criteria and regular review are required."@en ; mailo:appliesTo mailo:BlackBoxMLSystem, mailo:MedicalAISystem ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "constitutional_grand_chamber"^^xsd:string ; mailo:triggersObligation mailo:ExplainabilityObligation, mailo:TransparencyObligation . mailo:P_817_003 a mailo:RatioDecidendi ; rdfs:label "P-817/19-003: Strict Necessity Tri-Level Proportionality Test"@en ; rdfs:comment "Large-scale automated processing must satisfy: (a) targeted scope (C-817/19 paras 167-168, 184, 191), (b) purpose limitation (para 191), (c) temporal limitation (paras 255-256, 262). The Court upheld the PNR Directive subject to Charter-conforming restrictive interpretation; non-conformity at a limb is flagged for review and does not entail invalidity."@en ; mailo:appliesTo mailo:AutomatedDecisionSystem ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "constitutional_grand_chamber"^^xsd:string . mailo:P_817_004 a mailo:RatioDecidendi ; rdfs:label "P-817/19-004: Non-Discrimination — Health Status Excluded"@en ; rdfs:comment "Pre-determined assessment criteria may in no circumstances be based on health (PNR Directive Art.6(4); C-817/19 paras 105, 196-197) — a categorical prohibition, not a special-category rule curable by a GDPR Art.9(2) basis. MAILO transposes this to medical AI as a bar on using health condition as a discriminatory classification or risk-scoring criterion, which is narrower than any processing of health data."@en ; mailo:appliesTo mailo:GenomicDiagnosticAI, mailo:MedicalAISystem ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "constitutional_grand_chamber"^^xsd:string . mailo:P_817_005 a mailo:ObiterDictum ; rdfs:label "P-817/19-005: Independent Prior Review (and MAILO's DPIA bridge)"@en ; rdfs:comment "C-817/19 paras 222-227 (PNR Directive Art.12(3)(b)): disclosure after the initial period requires prior review by a court or independent administrative authority, which must be objective, impartial, free from external influence, and distinct from the requesting authority or the PIU. The Court said nothing about GDPR Art.35 or DPIAs; the link from independent prior review to a DPIA obligation for medical AI is an authorial cross-instrument interpretation by the ontology designer, not a holding. The DPIA requirement stands on GDPR Art.35 alone."@en, "Encoding basis (DesignerTransposition): C-817/19 paras 222-227 concern prior review of PNR disclosure requests. The bridge to a GDPR Art.35 DPIA duty is the designer's; the DPIA constraint stands on Art.35 alone."@en ; mailo:hasEncodingBasis mailo:DesignerTransposition ; mailo:hasPrecedenceLevel "authorial_transposition"^^xsd:string ; mailo:triggersObligation mailo:DPIAObligation . mailo:P_Bara_001 a mailo:RatioDecidendi ; rdfs:label "Transparency Rights Foundation"@en ; rdfs:comment "C-201/14 Bara: personal data transferred between public administrative bodies and further processed require that data subjects be informed of the transfer and of the categories of data concerned. UNVERIFIED (v5.1.0): no paragraph anchor and no ECLI are recorded, and the holding has not been checked against the judgment text. Do not cite until verified."@en ; mailo:hasEncodingBasis mailo:CourtHolding ; mailo:hasPrecedenceLevel "binding_ratio"^^xsd:string . mailo:PatentClassification a owl:Class ; rdfs:label "PatentClassification" ; rdfs:comment "An IPC/CPC patent classification code" . mailo:PatientRight a owl:Class ; rdfs:label "PatientRight", "Patient Right"@en ; rdfs:comment "A patient-specific data or health right" ; rdfs:subClassOf mailo:DataSubjectRight . mailo:ProceduralPrecedent a mailo:InclusionBasis ; rdfs:label "Procedural precedent"@en ; rdfs:comment "The case is relevant for procedure or forum practice rather than substance. Carries no implication about the technology at issue."@en . mailo:RegulatoryAuthority a owl:Class ; rdfs:label "RegulatoryAuthority" ; rdfs:comment "An EU or national regulatory or supervisory body" . mailo:Right_Contest a mailo:DataSubjectRight ; rdfs:label "Right to Contest Decision (Art.22(3)(c))"@en . mailo:Right_Explanation a mailo:DataSubjectRight ; rdfs:label "Right to Meaningful Explanation (Art.15(1)(h), C-203/22)"@en . mailo:Right_ExpressViews a mailo:DataSubjectRight ; rdfs:label "Right to Express Views (Art.22(3)(b))"@en . mailo:Right_HumanIntervention a mailo:DataSubjectRight ; rdfs:label "Right to Human Intervention (Art.22(3)(a))"@en . mailo:Right_MedicalRecordAccess a mailo:PatientRight ; rdfs:label "Right to Access Medical Records (C-307/22)"@en . mailo:Risk_AIActQMS_MDR_Overlap a mailo:ComplianceRisk ; rdfs:label "AI Act Art.17 QMS / MDR Annex IX Double Burden"@en ; rdfs:comment "AI-SaMD must comply with both AI Act QMS (Art.17) and MDR QMS (Annex IX). Without regulatory alignment, manufacturers face duplicative audit requirements."@en ; mailo:riskLevel "HIGH"^^xsd:string . mailo:Risk_BlackBox a mailo:ComplianceRisk ; rdfs:label "Black-Box ML System (C-817/19 Prohibition)"@en ; mailo:mitigatedBy mailo:ExplainabilityObligation, mailo:TransparencyObligation ; mailo:riskLevel "HIGH"^^xsd:string . mailo:Risk_DataActIoT a mailo:ComplianceRisk ; rdfs:label "Data Act Art.3/5 IoT Data Sharing Obligation"@en ; rdfs:comment "Medical device manufacturers must share IoT-generated health data with users/third parties. Conflicts with trade secret protection (Art.15) and competitive AI training data strategies."@en ; mailo:riskLevel "MEDIUM"^^xsd:string . mailo:Risk_EHDSOptOut a mailo:ComplianceRisk ; rdfs:label "EHDS Art.35 Opt-Out Erodes Training Dataset"@en ; rdfs:comment "Widespread exercise of opt-out rights under EHDS Art.35 may substantially reduce available health data for AI training, creating representativeness risk under AI Act Art.10."@en ; mailo:riskLevel "MEDIUM"^^xsd:string . mailo:Risk_EPOUPC_Divergence a mailo:ComplianceRisk ; rdfs:label "EPO-UPC Divergent Validity Rulings (Opt-Out Strategy)"@en ; mailo:riskLevel "HIGH"^^xsd:string . mailo:Risk_HealthDataDoubleLayer a mailo:ComplianceRisk ; rdfs:label "Health Data Double Protection Layer (Art.9 + Art.22(4))"@en ; mailo:mitigatedBy mailo:DPIAObligation ; mailo:riskLevel "CRITICAL"^^xsd:string . mailo:Risk_InsufficientDisclosure a mailo:ComplianceRisk ; rdfs:label "Insufficient Training Data Disclosure (EPO T0161/18)"@en ; mailo:mitigatedBy mailo:DisclosureObligation ; mailo:riskLevel "HIGH"^^xsd:string . mailo:Risk_NPE_Revocation a mailo:ComplianceRisk ; rdfs:label "NPE-Initiated Revocation Attack"@en ; mailo:riskLevel "MEDIUM"^^xsd:string . mailo:Risk_PLDStrictLiability a mailo:ComplianceRisk ; rdfs:label "PLD 2024 Strict Liability for AI-SaMD"@en ; rdfs:comment "Medical AI systems qualifying as 'defective products' under PLD Art.4 expose manufacturers to strict liability without fault. Presumptions in Art.10 and Art.11 shift burden to defendants."@en ; mailo:mitigatedBy mailo:HumanOversightObligation, mailo:TransparencyObligation ; mailo:riskLevel "HIGH"^^xsd:string . mailo:Risk_SEP_FRAND a mailo:ComplianceRisk ; rdfs:label "SEP/FRAND Obligations (HL7 FHIR, DICOM AI)"@en ; mailo:riskLevel "MEDIUM"^^xsd:string . mailo:Risk_SolelyTest a mailo:ComplianceRisk ; rdfs:label "GDPR Art.22 'Solely' Test Failure (Rubber-Stamping)"@en ; mailo:mitigatedBy mailo:HumanOversightObligation ; mailo:riskLevel "CRITICAL"^^xsd:string . mailo:Sanofi a mailo:Company ; rdfs:label "Sanofi S.A."@en . mailo:SemanticGranularityConflict a mailo:ConflictType ; rdfs:label "Semantic granularity conflict"@en ; rdfs:comment "Frameworks describe the same artefact at incompatible levels of specificity, so a statement that satisfies one reads as over- or under-inclusive under another (e.g. broad EPO claim scope against narrow MDR intended purpose against intermediate AI Act foreseeable use)."@en . mailo:SiemensHealthineers a mailo:Company ; rdfs:label "Siemens Healthineers AG"@en . mailo:TSD_Art2 a mailo:LegalArticle ; rdfs:label "TSD Article 2 — Definition of Trade Secret"@en ; rdfs:comment "Three-part cumulative test: information must be (1) secret (not generally known or readily accessible), (2) have commercial value because it is secret, and (3) be subject to reasonable steps to keep it secret. All three must hold for ML model weights or training data to qualify."@en . mailo:TSD_Art3 a mailo:LegalArticle ; rdfs:label "TSD Art.3 — Lawful acquisition, use and disclosure"@en ; rdfs:comment "Acquisition is lawful where obtained by independent discovery or creation, or by observation, study, disassembly or testing of a product or object that has been made available to the public or is lawfully in the possession of the acquirer who is free of any legally valid duty to limit the acquisition — that is, REVERSE ENGINEERING is lawful acquisition unless contractually excluded. Directly relevant to medical AI: analysing a deployed model's behaviour through observation and testing is not an unlawful acquisition of the trade secrets it embodies, which bears on both explainability research and the practical reach of secrecy over deployed systems. Added in v5.1.0."@en . mailo:TSD_Art5 a mailo:LegalArticle ; rdfs:label "TSD Art.5 — Exceptions (freedom of expression, whistleblowing, workers' representatives, legitimate interest)"@en ; rdfs:comment """An application for the measures, procedures and remedies is dismissed where the acquisition, use or disclosure was carried out (a) for exercising the right to freedom of expression and information as set out in the Charter, including respect for the freedom and pluralism of the media; (b) for revealing misconduct, wrongdoing or illegal activity in order to protect the general public interest; (c) by workers to their representatives in the legitimate exercise of representative functions; or (d) for the purpose of protecting a legitimate interest recognised by Union or national law. CORRECTED in v5.1.0, and this was a substantive error rather than a citation slip. The earlier entry read that trade secret protection does not apply 'where disclosure is required for the exercise of the right to information', and treated Art.5 as the provision engaged when a data subject invokes GDPR Art.15(1)(h). Art.5(a) is the freedom of expression and information of Charter Art.11 — media freedom, investigative journalism, protection of journalistic sources per recital 19. A patient seeking an explanation of an automated medical decision is not exercising that right. Art.5(d) might be argued to cover data-protection rights as a legitimate interest recognised by Union law, but that is an argument to be made rather than a plain reading, and the CJEU did not take that route in C-203/22: the Court resolved the trade-secret defence within the GDPR itself, requiring disclosure to the supervisory authority or court for balancing. The provision that actually preserves regulatory disclosure duties is Art.1(2)(b), now encoded separately."""@en . mailo:TSD_Art9 a mailo:LegalArticle ; rdfs:label "TSD Article 9 — Preservation of Confidentiality in Legal Proceedings"@en ; rdfs:comment "Courts may impose confidentiality measures to protect trade secrets during litigation. Relevant to UPC discovery proceedings where a party seeks disclosure of AI model architecture or training data as part of infringement or validity analysis."@en . mailo:UPC a mailo:Court ; rdfs:label "Unified Patent Court (UPC)"@en . mailo:UPC_CD_Paris a mailo:UPCDivision ; rdfs:label "UPC Central Division — Paris"@en . mailo:UPC_CoA a mailo:Court ; rdfs:label "UPC Court of Appeal"@en ; rdfs:subClassOf mailo:UPC . mailo:UPC_LD_Paris a mailo:UPCDivision ; rdfs:label "UPC Local Division — Paris"@en . mailo:AIAct_Art13 a mailo:LegalArticle ; rdfs:label "AI Act Art.13 — Transparency and provision of information TO DEPLOYERS"@en ; rdfs:comment """Art.13(1): high-risk AI systems shall be designed and developed so that their operation is sufficiently transparent to enable DEPLOYERS to interpret the output and use it appropriately. Art.13(2)-(3): instructions for use must accompany the system, specifying provider identity, characteristics, capabilities and limitations of performance including accuracy, robustness and cybersecurity metrics, known foreseeable risks, human oversight measures and expected lifetime. SCOPE NOTE added in v5.1.0: the addressee is the DEPLOYER, not the affected person. Art.13 confers nothing on a patient or data subject. This is the pivot of the trade-secret debate — a provider may satisfy Art.13 in full while an affected individual receives nothing, which is why the explanation route for individuals runs through GDPR Art.15(1)(h) and Art.22, and, in the health domain, through EHDS Art.52's controlled-access mechanism. The earlier one-line comment did not record the addressee distinction on which that argument depends."""@en . mailo:BlackBoxMLSystem a owl:Class ; rdfs:label "BlackBoxMLSystem", "Black-Box ML System"@en ; rdfs:comment "Non-transparent self-learning ML system (C-817/19 prohibition)", "A non-transparent, self-learning system whose decision logic cannot be explained. Prohibited under C-817/19 for decisions with significant impact."@en, "Corrected in v5.1.0: previously declared a subclass of MedicalAISystem. Opacity is a property of a system's decision logic, not of its clinical application; the PNR system in C-817/19 was not medical."@en ; rdfs:seeAlso airo:OpaqueSystem ; rdfs:subClassOf mailo:AutomatedDecisionSystem . mailo:ClinicalDecisionSupport a owl:Class ; rdfs:label "ClinicalDecisionSupport", "Clinical Decision Support System"@en ; rdfs:comment "Clinical decision support system (CDSS)" ; rdfs:subClassOf mailo:MedicalAISystem . mailo:DPIAObligation a owl:Class ; rdfs:label "DPIAObligation", "DPIA Obligation"@en ; rdfs:comment "Obligation to carry out a DPIA under GDPR Art.35" ; rdfs:subClassOf mailo:ComplianceObligation ; owl:equivalentClass dpv:DPIA . mailo:DataAct_Art4 a mailo:LegalArticle ; rdfs:label "Data Act Art.4 — Rights and obligations of users and data holders regarding access and use"@en ; rdfs:comment """Art.4(1): where data is not directly accessible from the connected product, data holders shall make readily available data, and the relevant metadata, available to the user without undue delay, in the same quality as it is available to the data holder, easily, securely, free of charge and in a machine-readable format. Art.4(4): data holders shall not unreasonably impede the user's choices or rights through the structure, design, function or operation of a digital user interface. Art.4(6)-(8) carry the trade-secret safeguards: trade secrets are disclosed to the user only where the holder and user agree the measures necessary to preserve confidentiality; the holder may withhold or suspend sharing where it demonstrates that it is highly likely to suffer serious economic damage, with notification to the competent authority. This is the Data Act limb of MAILO's trade-secret-against-transparency tension, and it is a qualified withholding right rather than a veto. Scope broadened in v5.1.0: the entry previously covered only paragraphs (6)-(8)."""@en . mailo:DataAct_Art5 a mailo:LegalArticle ; rdfs:label "Data Act Art.5 — Data Sharing with Third Parties"@en ; rdfs:comment "Data holders must share IoT-generated health data with third parties at user request. Creates AI training data supply chain implications for medical device manufacturers."@en . mailo:DataGovernanceConflict a mailo:ConflictType ; rdfs:label "Data governance conflict"@en ; rdfs:comment "Frameworks impose incompatible conditions on the acquisition, sharing or reuse of data (e.g. an EHDS opt-out reducing dataset representativeness against AI Act Art.10 data-quality duties)."@en . mailo:EPO_BoA a mailo:Court ; rdfs:label "EPO Board of Appeal"@en . mailo:GDPR_Art12_5 a mailo:LegalArticle ; rdfs:label "GDPR Art.12(5) — Exercise of data subject rights free of charge"@en ; rdfs:comment "Information and action under Arts.15-22 are provided free of charge; a reasonable fee or refusal is possible only for manifestly unfounded or excessive requests, and the controller bears the burden of showing this. Basis, with Art.15(3), for the first-copy holding in C-307/22. Added in v5.1.0."@en . mailo:GDPR_Art15 a mailo:LegalArticle ; rdfs:label "GDPR Article 15"@en ; rdfs:comment "Right of access by the data subject."@en ; mailo:grantsRight mailo:Right_MedicalRecordAccess . mailo:GDPR_Art15_3 a mailo:LegalArticle ; rdfs:label "GDPR Art.15(3) — Right to a copy"@en ; rdfs:comment "The controller shall provide a copy of the personal data undergoing processing; a reasonable fee based on administrative costs may be charged for further copies. Interpreted in C-307/22 as requiring a faithful and intelligible reproduction, extending to full documents where necessary to understand the data. Added in v5.1.0."@en . mailo:GDPR_Art22_1 a mailo:LegalArticle ; rdfs:label "GDPR Art.22(1) — Right not to be subject to a solely automated decision"@en ; rdfs:comment "The data subject shall have the RIGHT NOT TO BE SUBJECT to a decision based solely on automated processing, INCLUDING PROFILING, which produces legal effects concerning him or her or similarly significantly affects him or her. CORRECTED in v5.1.0: the earlier text read 'individuals shall not be subject to', stating a flat prohibition. Whether Art.22(1) is a prohibition applying of its own force or a right the data subject must invoke is contested and was not settled by C-634/21 — see the note on P_634_004. The encoding now follows the wording of the provision rather than pre-committing to one reading. The earlier text also omitted profiling and compressed the two-limb effects test."@en . mailo:GDPR_Art22_2 a mailo:LegalArticle ; rdfs:label "GDPR Art.22(2) — Derogations"@en ; rdfs:comment "Art.22(1) does not apply where the decision is (a) necessary for entering into or performance of a contract between data subject and controller, (b) authorised by Union or Member State law to which the controller is subject and which lays down suitable measures to safeguard rights, freedoms and legitimate interests, or (c) based on the data subject's explicit consent."@en . mailo:GDPR_Art22_3 a mailo:LegalArticle ; rdfs:label "GDPR Art.22(3) — Safeguards, applying only to points (a) and (c) of Art.22(2)"@en ; rdfs:comment "In the cases referred to in POINTS (a) AND (c) of Art.22(2), the controller shall implement suitable measures to safeguard the data subject's rights, freedoms and legitimate interests, at least the right to obtain human intervention on the controller's part, to express his or her point of view and to contest the decision. CORRECTED in v5.1.0: the earlier text said these safeguards apply whenever an exception applies. They do NOT attach to point (b): where the basis is Union or Member State law, that law must itself lay down suitable measures. The distinction is material for medical AI operating on a national legal basis, where the Art.22(3) triad cannot simply be assumed."@en ; mailo:grantsRight mailo:Right_Contest, mailo:Right_ExpressViews, mailo:Right_HumanIntervention . mailo:GDPR_Art23 a mailo:LegalArticle ; rdfs:label "GDPR Art.23 — Restrictions"@en ; rdfs:comment "Union or Member State law may restrict the scope of Arts.12-22 where the restriction respects the essence of fundamental rights and is necessary and proportionate. Held in C-307/22 not to permit national law to shift the cost of a first copy onto the data subject in order to protect the controller's economic interests. Added in v5.1.0."@en . mailo:GenomicDiagnosticAI a owl:Class ; rdfs:label "GenomicDiagnosticAI", "Genomic Diagnostic AI"@en ; rdfs:comment "AI system for genomic/genetic diagnostics" ; rdfs:subClassOf mailo:MedicalAISystem ; mailo:hasRisk mailo:Risk_EHDSOptOut, mailo:Risk_NPE_Revocation . mailo:HumanOversightRecord a owl:Class ; rdfs:label "HumanOversightRecord", "Human Oversight Record"@en ; rdfs:comment "Documented record of substantive human review (C-634/21 P_634_002)", "A documented record of substantive human review of an automated decision (C-634/21 P_634_002: rubber-stamping is insufficient)."@en . mailo:LiabilityConflict a mailo:ConflictType ; rdfs:label "Liability gap"@en ; rdfs:comment "Compliance with one framework does not resolve, and may not even bear on, exposure under another liability regime (e.g. AI Act conformity against the revised Product Liability Directive)."@en . mailo:MDR a mailo:EURegulation, mailo:RegulatoryFramework ; rdfs:label "MDR 2017/745"@en ; eli:id_local "32017R0745" ; dcterms:date "2017-04-05"^^xsd:date ; dcterms:identifier "2017/745/EU" ; rdfs:comment "Listed in AI Act Annex I Section A, which is why a Rule 11 Class IIa-or-above device is high-risk under Art.6(1) by the Section A route. Recorded in v5.2.2."@en ; mailo:coversArticle mailo:MDR_AnnexI, mailo:MDR_AnnexII, mailo:MDR_AnnexIX, mailo:MDR_AnnexVIII_Rule11, mailo:MDR_Art15, mailo:MDR_Art61, mailo:MDR_Art83 . mailo:MDR_AnnexII a mailo:LegalArticle ; rdfs:label "MDR Annex II — Technical documentation"@en ; rdfs:comment "The technical documentation manufacturers must prepare, beginning at section 1.1 with the device description and specification including intended purpose, intended users and clinical indications. Referenced by Tension_EPO_MDR_SemanticScope, which contrasts the precision MDR demands of the intended purpose with the breadth patent prosecution seeks in claim scope. Added in v5.1.0: the tension cited this annex while the ontology did not encode it."@en . mailo:MDR_AnnexIX a mailo:LegalArticle ; rdfs:label "MDR Annex IX — Conformity assessment based on a QMS and technical documentation"@en ; rdfs:comment "The conformity assessment route involving notified-body assessment of the manufacturer's quality management system and of the technical documentation. Referenced by Tension_MDR_AIAct_DualTrack, which sets it against AI Act Art.43. Added in v5.1.0: the tension cited this annex while the ontology did not encode it."@en . mailo:OmnibusAI_RegEU2026_1744 a mailo:LegislativeAmendment ; rdfs:label "Digital Omnibus on AI — Regulation (EU) 2026/1744"@en ; eli:id_local "32026R1744" ; dcterms:date "2026-07-08"^^xsd:date ; dcterms:replaces mailo:OmnibusAI_ProvisionalAgreement_20260507 ; dcterms:source ; mailo:amendmentDescription "Regulation (EU) 2026/1744 of the European Parliament and of the Council of 8 July 2026 amending Regulations (EU) 2024/1689, (EU) 2018/1139 and (EU) 2023/1230 as regards the simplification of the implementation of harmonised rules on artificial intelligence (Digital Omnibus on AI). Published in the Official Journal on 24 July 2026 and in force from 27 July 2026. Chapter III Sections 1 to 3 apply from 2 December 2027 for Article 6(2) / Annex III systems and from 2 August 2028 for Article 6(1) / Annex I systems." ; mailo:versionStartDate "2026-07-27"^^xsd:date . mailo:PLD a mailo:EUDirective, mailo:RegulatoryFramework ; rdfs:label "PLD 2024/2853"@en ; eli:id_local "32024L2853" ; dcterms:date "2024-10-23"^^xsd:date ; dcterms:identifier "2024/2853/EU" ; rdfs:comment "Revised Product Liability Directive. The AI Liability Directive proposal was listed for intended withdrawal in the Commission Work Programme of February 2025 and formally withdrawn on 6 October 2025; the operative withdrawal date is 2025-10-06. PLD 2024 is therefore the EU strict-liability instrument for AI systems as 'products'."@en ; mailo:coversArticle mailo:PLD_Art10, mailo:PLD_Art11, mailo:PLD_Art4, mailo:PLD_Art5, mailo:PLD_Art6, mailo:PLD_Art7, mailo:PLD_Art9 . mailo:PLD_Art5 a mailo:LegalArticle ; rdfs:label "PLD Art.5 — Right to compensation"@en ; rdfs:comment "Member States shall ensure that any natural person who suffers damage caused by a defective product is entitled to compensation. The strict-liability rule itself: fault need not be shown, only damage, defectiveness and the causal link. Added in v5.1.0, the strict-liability content having previously been attributed to Art.4."@en . mailo:PLD_Art7 a mailo:LegalArticle ; rdfs:label "PLD Art.7 — Defectiveness"@en ; rdfs:comment """A product is defective where it does not provide the safety that a person is entitled to expect or that is required under Union or national law. The assessment takes account of all circumstances, including presentation and instructions, reasonably foreseeable use and misuse, the effect of the ability to continue to learn or of software updates after placing on the market, the moment in time at which the product was placed on the market or remained within the manufacturer's control, and relevant product safety requirements including cybersecurity. CORRECTED in v5.1.0: previously labelled 'AI and Software as Products' and describing the product definition, which is Art.4. The continued-learning limb recorded here is the provision on which MAILO's model-drift arguments actually rest."""@en . mailo:Patent a owl:Class ; rdfs:label "Patent", "Patent"@en ; rdfs:comment "A patent grant" . mailo:Right a owl:Class ; rdfs:label "Right", "Right"@en ; rdfs:comment "A legal right" . mailo:RiskScoringSystem a owl:Class ; rdfs:label "RiskScoringSystem", "Risk Scoring System"@en ; rdfs:comment "Automated risk stratification or scoring system" ; rdfs:subClassOf mailo:AutomatedDecisionSystem . mailo:UPC_LD_Dusseldorf a mailo:UPCDivision ; rdfs:label "UPC Local Division — Düsseldorf"@en . mailo:UPC_LD_Munich a mailo:UPCDivision ; rdfs:label "UPC Local Division — Munich"@en . mailo:AIAct_Art14 a mailo:LegalArticle ; rdfs:label "AI Act Art.14 — Human oversight"@en ; rdfs:comment "High-risk AI systems shall be designed and developed so that they can be effectively overseen by natural persons during use, including through appropriate human-machine interface tools. Oversight measures aim to prevent or minimise risks to health, safety or fundamental rights, and must enable the person to understand capacities and limitations, remain aware of automation bias, correctly interpret output, disregard or override it, and intervene or stop the system. The complementedBy link to GDPR Art.22(3) should be read subject to the note on that article: the Art.22(3) safeguards attach only to Art.22(2)(a) and (c), so oversight under Art.14 and the data-protection safeguards do not automatically coincide."@en ; mailo:complementedBy mailo:GDPR_Art22_3 . mailo:CJEU_FirstChamber a mailo:Court ; rdfs:label "CJEU First Chamber"@en . mailo:DesignerOperationalisation a mailo:EncodingBasis ; rdfs:label "Designer operationalisation"@en ; rdfs:comment "The court's test is open-textured and the constraint encodes a proxy for it. The paragraphs are correct; the operative predicate is MAILO's. Example: C-634/21 states a determining-role test; the substantive-oversight predicate is a proxy for it."@en . mailo:DisclosureObligation a owl:Class ; rdfs:label "DisclosureObligation", "Disclosure Obligation"@en ; rdfs:comment "Obligation to disclose information to data subjects" ; rdfs:seeAlso dpv:ProvideInformation ; rdfs:subClassOf mailo:ComplianceObligation . mailo:EPC a mailo:InternationalTreaty, mailo:RegulatoryFramework ; rdfs:label "European Patent Convention"@en ; dcterms:date "1977-10-07"^^xsd:date ; rdfs:comment "European Patent Convention (in force 1977-10-07). Governs patentability of AI/medical-AI inventions (Art.52 exclusions), sufficiency of disclosure for ML models (Art.83), and EPO opposition/appeal proceedings. Administered by EPO; distinct from EU law but intersects via UPC."@en ; mailo:coversArticle mailo:EPC_Art52, mailo:EPC_Art83 . mailo:EPC_Art56 a mailo:LegalArticle ; rdfs:label "EPC Article 56"@en ; rdfs:comment "Inventive step. Basis for the line of authority holding that merely applying a known machine-learning technique to a particular field follows a general technological trend and is not inventive as such (T 161/18 point 3.6; followed in T 1191/19)."@en . mailo:EPC_Art83 a mailo:LegalArticle ; rdfs:label "EPC Article 83"@en ; rdfs:comment "Disclosure of the invention — sufficiency of disclosure. For AI inventions, requires disclosure of suitable training and validation data and of the neural network's topology, activation functions, end conditions and learning mechanism (T 161/18; T 1191/19)."@en . mailo:EPOProceding a owl:Class ; rdfs:label "EPOProceding", "EPO Proceeding"@en ; rdfs:comment "EPO opposition, appeal, or examination", "DEPRECATED in v5.2.2: misspelling, one 'e' missing. Use mailo:EPOProceeding."@en, "Opposition, appeal, or examination proceeding before the EPO."@en ; rdfs:subClassOf mailo:Case ; owl:deprecated true ; owl:equivalentClass mailo:EPOProceeding . mailo:EUDirective a owl:Class ; rdfs:label "EUDirective", "EU Directive"@en ; rdfs:comment "EU Directive (requires transposition)" ; rdfs:subClassOf mailo:LegalInstrument . mailo:GDPR_Art22 a mailo:LegalArticle ; rdfs:label "GDPR Article 22"@en ; rdfs:comment "Prohibition on solely automated individual decision-making."@en . mailo:InternationalTreaty a owl:Class ; rdfs:label "InternationalTreaty", "International Treaty"@en ; rdfs:comment "International treaty or convention" ; rdfs:subClassOf mailo:LegalInstrument . mailo:LegalInstrument a owl:Class ; rdfs:label "LegalInstrument", "Legal Instrument"@en ; rdfs:comment "Any binding or persuasive legal document", "Any binding or persuasive legal document (regulation, directive, treaty, standard)."@en ; rdfs:subClassOf eli:LegalResource . mailo:TSD a mailo:EUDirective, mailo:RegulatoryFramework ; rdfs:label "Trade Secrets Directive 2016/943"@en ; eli:id_local "32016L0943" ; dcterms:date "2016-06-08"^^xsd:date ; dcterms:identifier "2016/943/EU" ; rdfs:comment "EU Trade Secrets Directive. Harmonises civil-law protection of undisclosed know-how and business information. Key for medical AI: model weights, training data schemas, and proprietary algorithms qualify as trade secrets under the Art.2 three-part test. Art.5 carve-outs limit the secrecy shield when exercise of the right to information is at stake — directly engaged by GDPR Art.15(1)(h) and AI Act Art.13 transparency obligations."@en ; mailo:coversArticle mailo:TSD_Art1_2, mailo:TSD_Art2, mailo:TSD_Art3, mailo:TSD_Art5, mailo:TSD_Art9 . mailo:DirectSubjectMatter a mailo:InclusionBasis ; rdfs:label "Direct subject matter"@en ; rdfs:comment "The case is itself about a medical AI or medical software system. Only such cases may carry a medical-AI value for mailo:involvesTechnology."@en . mailo:EPOProceeding a owl:Class ; rdfs:label "EPOProceeding"@en ; rdfs:comment "Correctly spelled replacement for mailo:EPOProceding, added in v5.2.2. The misspelled class is deprecated rather than renamed, because renaming a published IRI breaks every reference to it; instances are typed against both during the transition."@en ; rdfs:subClassOf mailo:Case . mailo:EncodingBasis a owl:Class ; rdfs:label "EncodingBasis"@en ; rdfs:comment """Whether an encoded principle states what a court held, or a construct of the ontology designer built on top of it. hasPrecedenceLevel records how much weight a holding claims. It cannot answer the prior question of whether the encoded proposition is a holding at all. Keeping the two apart means a reader can query for everything the graph asserts on the designer's own authority."""@en . mailo:EuropeanPatent a owl:Class ; rdfs:label "EuropeanPatent", "European Patent (EP)"@en ; rdfs:comment "A European patent granted by the EPO" ; rdfs:subClassOf mailo:Patent . mailo:Explanation a owl:Class ; rdfs:label "Explanation", "Explanation"@en ; rdfs:comment "Actionable non-technical explanation for data subjects (C-203/22 P_203_001)", "An actionable, non-technical explanation of the logic and main input factors of an automated decision (C-203/22 P_203_001; GDPR Art.15(1)(h))."@en . mailo:HumanOversightObligation a owl:Class ; rdfs:label "HumanOversightObligation", "Human Oversight Obligation"@en ; rdfs:comment "Obligation to ensure substantive human review" ; rdfs:seeAlso airo:HumanOversight ; rdfs:subClassOf mailo:ComplianceObligation ; owl:equivalentClass dpv:HumanInvolvement . mailo:TSD_Art1_2 a mailo:LegalArticle ; rdfs:label "TSD Art.1(2) — Matters the Directive does not affect"@en ; rdfs:comment """The Directive shall not affect: (a) the exercise of the right to freedom of expression and information as set out in the Charter, including media freedom and pluralism; (b) THE APPLICATION OF UNION OR NATIONAL RULES REQUIRING TRADE SECRET HOLDERS TO DISCLOSE, FOR REASONS OF PUBLIC INTEREST, INFORMATION INCLUDING TRADE SECRETS, TO THE PUBLIC OR TO ADMINISTRATIVE OR JUDICIAL AUTHORITIES FOR THE PERFORMANCE OF THE DUTIES OF THOSE AUTHORITIES; (c) the application of rules requiring or allowing Union institutions and bodies or national public authorities to disclose information submitted by businesses; (d) the autonomy of social partners. Art.1(2)(b) is the correct doctrinal anchor for the transparency-secrecy interface and had been absent from MAILO, its function having been wrongly attributed to Art.5. It maps directly onto the structure the CJEU adopted in C-203/22 — disclosure to a supervisory authority or court, which then balances — and onto EHDS Art.52, where protected elements are disclosed into an HDAB-controlled environment rather than to the requester. Note the limit it also implies: Art.1(2)(b) preserves disclosure duties owed to authorities and, where the rule so provides, to the public. It does not by itself create a disclosure duty owed to an individual. Added in v5.1.0."""@en . mailo:TradeSecretConflict a mailo:ConflictType ; rdfs:label "Trade secret against transparency"@en ; rdfs:comment "A disclosure duty under one framework collides with a legitimate secrecy interest protected by another (TSD Art.2 against AI Act Art.13 and GDPR Art.15(1)(h))."@en . mailo:UPCDivision a owl:Class ; rdfs:label "UPC Division"@en ; rdfs:subClassOf mailo:Court . mailo:DataSubjectRight a owl:Class ; rdfs:label "DataSubjectRight", "Data Subject Right"@en ; rdfs:comment "A right of an individual under GDPR" ; rdfs:seeAlso dpv:DataSubjectRight ; rdfs:subClassOf mailo:Right . mailo:DomainContext a mailo:InclusionBasis ; rdfs:label "Domain context"@en ; rdfs:comment "Retained as background for a cross-framework tension rather than for any principle it establishes."@en . mailo:ExplainabilityObligation a owl:Class ; rdfs:label "ExplainabilityObligation", "Explainability Obligation"@en ; rdfs:comment "Obligation to provide actionable explanations" ; rdfs:seeAlso dpv:ExplainabilityObligation ; rdfs:subClassOf mailo:ComplianceObligation . mailo:GDPR_Art15_1h a mailo:LegalArticle ; rdfs:label "GDPR Art.15(1)(h) — Automated decision-making: existence, logic, significance and consequences"@en ; rdfs:comment "The data subject has the right to be informed of THE EXISTENCE of automated decision-making, including profiling, REFERRED TO IN ART.22(1) AND (4) and, at least in those cases, meaningful information about the logic involved as well as the significance and the envisaged consequences of such processing. Interpreted in C-203/22 paras 59-62. CORRECTED in v5.1.0: the earlier text omitted both the existence limb and the cross-reference scoping this right to Art.22(1) and (4) cases, which is what ties the explanation duty to the automated-decision regime rather than to processing at large."@en ; mailo:grantsRight mailo:Right_Explanation . mailo:HighRiskRoute a owl:Class ; rdfs:label "HighRiskRoute"@en ; rdfs:comment "The route by which an AI system becomes high-risk under the AI Act. Encoded in v5.1.0 because the distinction determines whether Art.86 and Art.26(11) apply at all, and MAILO had recorded it nowhere."@en . mailo:InclusionBasis a owl:Class ; rdfs:label "InclusionBasis"@en ; rdfs:comment """Why a case is present in MAILO at all. Added in v5.1.0 after an audit found two cases carrying medical-AI technology tags they did not warrant. The underlying defect was not the tags: it was that no instance recorded WHY it had been included, so twenty implicit inclusion judgements sat unstated across the graph and could not be reviewed. Note that MAILO's method is transposition. None of the three CJEU automated-decision cases is a medical-AI dispute: C-817/19 concerns passenger name records, C-634/21 and C-203/22 concern credit scoring. A criterion requiring direct medical-AI subject matter would exclude the trilogy the ontology is built on. The criterion therefore records the BASIS for relevance rather than demanding a single subject matter."""@en . mailo:ConflictType a owl:Class ; rdfs:label "ConflictType"@en ; rdfs:comment "Classification of the kind of friction a CrossFrameworkTension represents. Declared in v5.1.0: the five instances below were in use as values of mailo:conflictType but had never been typed."@en . mailo:TransparencyObligation a owl:Class ; rdfs:label "TransparencyObligation", "Transparency Obligation"@en ; rdfs:comment "Obligation to be transparent about AI processing" ; rdfs:subClassOf mailo:ComplianceObligation ; owl:equivalentClass dpv:TransparencyObligation . mailo:TransposedPrinciple a mailo:InclusionBasis ; rdfs:label "Transposed principle"@en ; rdfs:comment "The case establishes a principle in another domain that MAILO applies to medical AI by structural analogy. The transposition is the ontology designer's step, not the court's, and the affected shapes say so."@en . mailo:GDPR a mailo:EURegulation, mailo:RegulatoryFramework ; rdfs:label "GDPR 2016/679"@en ; eli:id_local "32016R0679" ; dcterms:date "2016-04-27"^^xsd:date ; dcterms:identifier "2016/679/EU" ; rdfs:comment "General Data Protection Regulation. Core data protection framework; Art.22 governs solely automated decision-making, Art.15/15(1)(h) the right of access and meaningful information, Art.9 special category health data, Art.35 DPIA obligation. Canonical identifier for the GDPR; mailo:GDPR_2016_679 is a deprecated alias."@en ; mailo:coversArticle mailo:GDPR_Art12_5, mailo:GDPR_Art15, mailo:GDPR_Art15_1h, mailo:GDPR_Art15_3, mailo:GDPR_Art22, mailo:GDPR_Art22_1, mailo:GDPR_Art22_2, mailo:GDPR_Art22_3, mailo:GDPR_Art22_4, mailo:GDPR_Art23, mailo:GDPR_Art35, mailo:GDPR_Art9 ; mailo:hasVersion mailo:GDPR_v1_FullyApplicable, mailo:GDPR_v2_OmnibusProposed_Art22 . mailo:LegalEntity a owl:Class ; rdfs:label "LegalEntity", "Legal Entity"@en ; rdfs:comment "Any natural or legal person with legal standing" . mailo:Technology a owl:Class ; rdfs:label "Technology", "Technology"@en ; rdfs:comment "A medical AI technology class" . mailo:ComplianceObligation a owl:Class ; rdfs:label "ComplianceObligation", "Compliance Obligation"@en ; rdfs:comment "A concrete obligation arising from regulation" . mailo:CourtHolding a mailo:EncodingBasis ; rdfs:label "Court holding"@en ; rdfs:comment "The encoded proposition is what the court decided, at the paragraphs cited."@en . mailo:PendingAmendment a owl:Class ; rdfs:label "Pending Amendment"@en ; rdfs:comment "A proposed but not yet adopted amendment to a legislative provision encoded in MAILO. MAILO encodes the numbering of the text currently in force; a PendingAmendment records, separately and explicitly, that a legislative proposal would change or delete that provision. The two are kept apart deliberately: encoding proposal numbering as though it were in force is the failure mode that produced the EHDS renumbering errors corrected in v5.1.0 Round 3."@en . mailo:UPCCase a owl:Class ; rdfs:label "UPCCase", "UPC Case"@en ; rdfs:comment "Case before the Unified Patent Court", "Infringement, revocation, or PI proceeding before the Unified Patent Court."@en ; rdfs:subClassOf mailo:Case . mailo:Company a owl:Class ; rdfs:label "Company", "Company"@en ; rdfs:comment "A corporate entity" ; rdfs:subClassOf mailo:LegalEntity . mailo:Court a owl:Class ; rdfs:label "Court", "Court"@en ; rdfs:comment "A judicial body" ; rdfs:subClassOf mailo:LegalEntity . mailo:LegislativeAmendment a owl:Class ; rdfs:label "LegislativeAmendment" ; rdfs:comment "A legislative event that creates or modifies a regulation version" . mailo:EURegulation a owl:Class ; rdfs:label "EURegulation", "EU Regulation"@en ; rdfs:comment "EU Regulation (directly applicable)" ; rdfs:seeAlso eli:LegalResource ; rdfs:subClassOf mailo:LegalInstrument . mailo:LegalPrinciple a owl:Class ; rdfs:label "LegalPrinciple", "Legal Principle"@en ; rdfs:comment "A binding or persuasive rule extracted from case law", "A binding or persuasive rule extracted from case law or legislation."@en . mailo:RatioDecidendi a owl:Class ; rdfs:label "RatioDecidendi", "Ratio Decidendi"@en ; rdfs:comment "Binding legal reasoning upon which decision rests", "The binding legal reasoning upon which the decision rests."@en ; rdfs:subClassOf mailo:LegalPrinciple . mailo:CJEUCase a owl:Class ; rdfs:label "CJEUCase", "CJEU Case"@en ; rdfs:comment "Judgment of the Court of Justice of the EU", "Preliminary ruling or direct action before the Court of Justice of the EU."@en ; rdfs:subClassOf mailo:Case . mailo:MedicalAISystem a owl:Class ; rdfs:label "MedicalAISystem", "Medical AI System"@en ; rdfs:comment "An AI-based system used in clinical contexts", "An AI-based system used in clinical diagnosis, treatment recommendation, or patient risk stratification."@en ; rdfs:seeAlso airo:AISystem ; rdfs:subClassOf mailo:AutomatedDecisionSystem, mailo:Technology ; mailo:hasRisk mailo:Risk_BlackBox, mailo:Risk_HealthDataDoubleLayer, mailo:Risk_InsufficientDisclosure, mailo:Risk_PLDStrictLiability, mailo:Risk_SolelyTest . mailo:RegulationVersion a owl:Class ; rdfs:label "RegulationVersion" ; rdfs:comment "A versioned expression of an EU regulation at a specific applicability phase" ; rdfs:subClassOf mailo:EURegulation . mailo:CrossFrameworkTension a owl:Class ; rdfs:label "CrossFrameworkTension" ; rdfs:comment "A conflict or gap between two regulatory frameworks" . mailo:ComplianceRisk a owl:Class ; rdfs:label "ComplianceRisk", "Compliance Risk"@en ; rdfs:comment "A compliance risk for medical AI developers" ; rdfs:seeAlso airo:Risk . mailo:Case a owl:Class ; rdfs:label "Case", "Case"@en ; rdfs:comment "Any judicial or quasi-judicial proceeding", "Any judicial or quasi-judicial proceeding."@en . mailo:AIAct a mailo:EURegulation, mailo:RegulatoryFramework ; rdfs:label "EU AI Act 2024/1689"@en ; eli:id_local "32024R1689" ; dcterms:date "2024-06-13"^^xsd:date ; dcterms:identifier "2024/1689/EU" ; rdfs:comment "Regulation (EU) 2024/1689 of 13 June 2024, published in the Official Journal on 12 July 2024 and in force from 1 August 2024. dcterms:date records the date of the act, following the convention used for the other EU instruments in MAILO. Canonical identifier for the AI Act; mailo:AIAct_2024_1689 is a deprecated alias."@en ; mailo:coversArticle mailo:AIAct_Art10, mailo:AIAct_Art13, mailo:AIAct_Art14, mailo:AIAct_Art17, mailo:AIAct_Art26, mailo:AIAct_Art27, mailo:AIAct_Art43, mailo:AIAct_Art4a, mailo:AIAct_Art50, mailo:AIAct_Art53, mailo:AIAct_Art55, mailo:AIAct_Art86, mailo:AIAct_Art9 ; mailo:hasVersion mailo:AIAct_v1_ProhibitedPractices, mailo:AIAct_v2_GPAIModelObligations, mailo:AIAct_v2b_PostOmnibusTransitional, mailo:AIAct_v3_OriginalHighRisk, mailo:AIAct_v3a_Omnibus_AnnexIII_Standalone, mailo:AIAct_v3b_Omnibus_AnnexI_SaMD, mailo:AIAct_v3c_OriginalAnnexIHighRisk, mailo:AIAct_v4_FRIA_PublicBodies . mailo:RegulatoryFramework a owl:Class ; rdfs:label "RegulatoryFramework", "Regulatory Framework"@en ; rdfs:comment "A coherent body of rules governing a domain", "A coherent body of rules governing a specific domain."@en . mailo:LiteratureSource a owl:Class ; rdfs:label "LiteratureSource" ; rdfs:comment "An academic paper, commentary, or official guidance" . mailo:AutomatedDecisionSystem a owl:Class ; rdfs:label "AutomatedDecisionSystem", "Automated Decision System"@en ; rdfs:comment "System processing personal data without meaningful human review", "A system that processes personal data to produce outputs affecting individuals without meaningful human review."@en, "Corrected in v5.1.0: this class was declared rdfs:subClassOf mailo:MedicalAISystem, which entailed that every automated decision system is a medical AI system. The credit-scoring systems in C-634/21 and C-203/22 are counter-examples in MAILO's own graph. The relation runs the other way: a medical AI system producing decisions about individuals IS an automated decision system, which is why MedicalAISystem is now the subclass."@en ; rdfs:seeAlso dpv:DataProcessing ; rdfs:subClassOf mailo:Technology ; owl:equivalentClass dpv:AutomatedDecisionMaking . mailo:LegalArticle a owl:Class ; rdfs:label "LegalArticle", "Legal Article"@en ; rdfs:comment "A specific provision within a legal instrument", "A specific provision within a legal instrument."@en ; rdfs:subClassOf eli:LegalExpression .