AI model recognizes transition to progressive MS

Wed 30 April 2025
AI
News

For proper treatment of multiple sclerosis (MS), it is important to know when the disease changes from relapsing-remitting to secondary progressive, a transition that is currently recognized an average of three years too late. Researchers at Uppsala University have now developed an AI model that can determine with 90 percent certainty which variant the patient has. The model increases the chances of starting the right treatment in time to slow the progression of the disease.

MS is a chronic, inflammatory disease of the central nervous system. More than 36,000 people are estimated to be living with MS in the Netherlands. Most patients begin with the relapsing-remitting form (RRMS), which is characterized by episodes of deterioration with intervening periods of stability. Over time, many people progress to secondary progressive MS (SPMS), in which their symptoms actually get progressively worse, with no apparent breaks.

Choosing the right treatment in time

Identifying this transition is important because the two different forms of MS require different treatments. Currently, they are diagnosed on average three years after the onset of the transition, which can lead to patients receiving medications that are no longer effective.

The new AI model summarizes clinical data from more than 22,000 patients in the Swedish MS registry. The model is based on data already collected during regular visits to physicians. That's data from neurological tests, MRI scans and ongoing treatments. “By recognizing patterns from previous patients, the model can determine whether a patient has the relapsing-remitting form or whether the disease has progressed to secondary progressive MS. A unique feature of the model is that it also indicates how much confidence it has in each individual assessment. This means the physician knows how reliable the conclusion is and how much confidence the AI has in its assessment,” said Kim Kultima, who led the study.

High accuracy

In the study, now published in the journal npj Digital Medicine, the model identified the transition to secondary progressive MS correctly or earlier than documented in the patient's medical record in nearly 87 percent of cases, with an overall accuracy of about 90 percent. For patients, this means that the diagnosis can be made earlier, allowing timely adjustments to the patient's treatment and slowing the progression of the disease. This also reduces the risk of patients receiving medications that are no longer effective.

“In the long term, the model could also be used to identify suitable participants for clinical trials, which could contribute to more effective and individualized treatment strategies,” Kultima concludes. An open, anonymized version of the model is now available to researchers at this link.

Exploring the potential that AI and big data can offer for the diagnosis and treatment of MS was also recommended several years ago by a European collaborative of scientists, MAGNIMS. At the time, they were already pledging to improve the diagnosis of this disease by using big data and AI in MRI examinations. At that time, the scientists named three concrete possibilities by which MRI measurements in MS, and thus MS diagnostics, could be improved.