AIdentifyAGE Ontology Standardizes Forensic Dental AI

🔑 Enhanced Key Takeaways

• •AIdentifyAGE ontology, published on arXiv in February 2026, standardizes forensic dental age estimation for 12-25 year olds using tooth development stages from panoramic radiographs, adhering to FAIR data principles for AI interoperability.
• •It integrates with existing ontologies like SNOMED CT, Dental Ontology, and ML-specific ones such as ML-Schema, enabling traceable provenance from raw images to legal reports in forensic workflows.
• •Developed collaboratively by forensic odontologists, AI researchers, and legal experts from institutions including the University of Zurich and INTERPOL, addressing rising demand for age verification in migration and crime cases.
• •Supports both manual expert assessments and AI/ML models (e.g., CNN-based tooth segmentation), with emphasis on explainability to meet judicial standards amid EU AI Act regulations.
• •Promotes reproducibility by modeling full pipeline: data acquisition, feature extraction (e.g., Demirjian stages), uncertainty quantification, and outcome linking to court decisions.

🛠️ Technical Deep Dive

• •Ontology built using OWL 2 DL, with 150+ classes, 200 object properties, and 50 data properties covering entities like ToothDevelopmentStage, RadiographicImage, AgeIntervalEstimate.
• •Key modules: Clinical (Demirjian/Willems methods), Forensic (chain-of-custody tracking), Legal (EvidenceAdmissibility), Imaging (DICOM/PNG formats), ML (ModelCard, PredictionSet).
• •Interoperability via alignments to OBO Foundry ontologies (e.g., Uberon for anatomy) and W3C standards; uses SKOS for semantic annotations.
• •FAIR compliance: F (persistent IRI identifiers), A (RDF serialization), I (SPARQL endpoints), R (licensing under CC-BY 4.0).
• •Implementation example: Protégé editor validation, GitHub repo with SHACL shapes for data validation, and demo Reasoner queries for age range inference.

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline