OnTEPT: a multimodal OWL ontology for post-traumatic stress disorder.

Abstract

Background: Post-traumatic stress disorder (PTSD) is a heterogeneous neuropsychiatric condition that develops after exposure to traumatic events. Its diagnosis is currently governed by two major international classification systems, Diagnostic and Statistical Manual of Mental Disorders (DSM) and International Classification of Diseases) ICD, which differ in structure, granularity, and diagnostic thresholds. Moreover, PTSD assessment is also multimodal, spanning structured interviews, self-report psychometric instruments, and increasing neurophysiological recordings such as electroencephalography (EEG).

Objective: This study presents OnTEPT, an OWL-based multimodal ontology designed to formalize DSM-5 and ICD-11 diagnostic criteria, integrate psychometric instruments and EEG-derived features, and support automated reasoning over individual patient instances.

Results: OnTEPT was developed following Ontology Development 101 (OD101) in Protégé 5. A hybrid strategy was used, combining top-down and bottom-up approaches and reusing concepts from health and mental health ontologies. The current version includes 112 classes, 18 object properties, and 113 data properties. DSM-5 (A-H) and ICD-11 (1-6) criteria are formalized as OWL equivalence axioms. These use constraints on cardinality, duration, clinical significance, and exclusions. Dissociative and complex PTSD are also represented. Four instruments are modeled at the item-level with cutoff thresholds, and two of them are linked to diagnostic criteria through annotations. The ontology was populated with psychometric data from 304 patients, EEG data from 41 participants, and 30 structured synthetic instances generated with Owlready2. 36 quantitative EEG features are encoded as typed data properties. Validation with HermiT, Pellet, and FaCT + + confirmed logical consistency and class satisfiability. Four intentionally inconsistent instances were correctly rejected due to classification issues. Query-based validation retrieved 113 patients above screening thresholds. It also found three with full DSM-5 profiles, twelve under ICD-11, two with dissociative PTSD, and one with complex PTSD.

Conclusions: OnTEPT provides a computable semantic framework that harmonizes PTSD diagnosis and its subtypes between DSM-5 and ICD-11. It integrates psychometric and EEG information within a single ontology. Its main contribution is an inferential layer that enables the automated classification of patient instances using both diagnostic systems. The architecture is extensible and ready to include biomarker-based rules, longitudinal data, and semantic workflows to support clinical decision-making.