Can artificial intelligence improve the diagnosis and prognosis of disorders of consciousness? A scoping review

Background: Artificial intelligence (AI), including machine learning (ML) and deep learning (DL), may assist clinicians in diagnosing and prognosticating disorders of consciousness (DoC), which are challenging conditions following severe acquired brain injury. This scoping review maps the use of ML/DL for DoC diagnosis and prognosis. Materials and methods: A scoping search across PubMed, Embase, Scopus, and Cochrane Library was conducted starting October 3, 2024, for peer‑reviewed articles published between 2000 and 2024. Results: The search identified 49,417 records; after de-duplication and screening, 21 studies met inclusion criteria. Included studies involved 14,683 patients and 180 healthy controls. AI approaches leveraged neurophysiological, neuroimaging, autonomic, and clinical predictors to support diagnosis (e.g., differentiating UWS vs. MCS) and prognosis (e.g., recovery potential). Conclusion: Advancing AI in DoC requires standardized data protocols and accounting for demographic and etiological variations. AI shows promise for improving diagnosis, prognosis, differentiation between consciousness states, and personalizing rehabilitation by identifying key recovery factors.

Mirjam Bonanno, Davide Cardile, Piergiuseppe Liuzzi, Antonio Celesti, Giuseppe Micali, Francesco Corallo, Angelo Quartarone, Francesco Tomaiuolo, Rocco Salvatore Calabrò