Scientists have used artificial intelligence (AI) to uncover two previously unrecognized biological subtypes of multiple sclerosis (MS), in research that could significantly improve diagnosis and treatment. The study analysed data from hundreds of people with MS, including advanced brain scans and blood markers that indicate nerve cell damage. By combining these datasets, the AI model was able to detect patterns that traditional symptom‑based classifications miss, revealing distinct disease trajectories.
The two new subtypes — described as “early‑sNfL” and “late‑sNfL” — are defined by when certain biological changes occur in the disease. In the early‑sNfL group, levels of a protein associated with nerve damage rise quickly soon after disease onset, and this is linked with more aggressive brain injury in key regions. In contrast, the late‑sNfL subtype is characterised by brain tissue loss before the blood marker rises, suggesting a slower but steadily progressing disease course.
Researchers say that current MS categories, which are largely based on clinical symptoms like relapses or progression, don’t always reflect what’s happening biologically in the nervous system, making it hard to tailor treatments effectively. The new AI‑informed subtypes provide a more nuanced biological understanding of MS, potentially allowing doctors to predict how the disease will evolve in each patient and choose therapies that more precisely target individual disease mechanisms.
Experts believe this discovery could transform MS care by enabling personalised monitoring and treatment plans. Instead of relying solely on observable symptoms, clinicians may soon be able to use AI‑supported tools to interpret both imaging and molecular data, improving early intervention and reducing long‑term disability. It’s an important step toward shifting MS classification from broad clinical labels to biologically meaningful categories that guide more effective medical decisions.
