Observational study abstract reports AI RNA sequencing diagnostic accuracy in multiple sclerosis and neuromyelitis opticaArtificial Intelligence identifies early Multiple Sclerosis and separates it from others

This publication is an abstract reporting an observational diagnostic accuracy study. The study population included 540 Multiple Sclerosis patients, 221 Neuromyelitis Optica patients, and 149 healthy controls, totaling 997 participants. The intervention involved PBMC RNA sequencing analyzed with AI ensemble models. Comparators included healthy controls, Neuromyelitis Optica patients, and Progressive MS patients. The primary outcome focused on diagnostic accuracy measured by AUC for early MS discrimination, MS versus NMO differential diagnosis, and RRMS versus progressive MS subtyping.

Regarding performance metrics, the model achieved 74% AUC at 100% coverage for discrimination of early MS from healthy individuals. For differential diagnosis of MS from NMO, the result was 91% AUC at 80% coverage. Subtyping RRMS from progressive MS yielded 79% AUC at 80% coverage. These effect sizes indicate positive diagnostic accuracy across the specified comparisons. No p-values or confidence intervals were reported in the provided text.

Authors note that this is an observational diagnostic accuracy study, meaning no causal inference regarding disease progression is supported. While authors state results may have immediate impact on clinical management of MS patients, the author claim of immediate impact is an assertion not supported by external validation data in this text. Follow-up duration was not reported. Adverse events were not reported. Clinicians should interpret these findings as preliminary until further validation occurs.

Practice relevance remains theoretical pending external validation. The study does not describe a prior molecular test for these tasks as an author claim. Clinicians should recognize the observational nature limits generalizability.