AI unlocks personalized MS treatments, identifies two disease subtypes

Scientists using artificial intelligence have discovered two new biological subtypes of multiple sclerosis (MS), paving the way for personalized treatments tailored to each patient's biology rather than relying on one-size-fits-all approaches. This strategy aims to predict disease progression early, enabling timely intervention and preventing nerve damage.

Until now, MS treatments worldwide were largely based on external symptoms, which did not always align with the patient's underlying biological changes. A new study led by University College London (UCL), involving 600 patients, goes deeper into the disease's biology.

Researchers combined levels of the protein sNfL, which signals nerve cell damage, with advanced brain scans. Using an AI model called SuStaIn, they identified two disease subtypes. In the "early protein increase" subtype, nerve damage rises rapidly at the disease's onset, causing serious damage in specific brain regions, indicating a more aggressive course. The second subtype progresses slowly, with structural brain changes occurring silently before protein levels rise.

The AI allows doctors to identify each patient's subtype early, selecting the most effective treatment before permanent damage occurs. Aggressive cases can receive stronger medications, while slower-progressing cases can be treated with neuroprotective therapies.

By combining digital AI tools with traditional clinical exams, this breakthrough shifts MS treatment from symptom-based care to biology-based care, offering hope for millions of patients to halt disease progression.