Using Machine Learning for Risk Prediction in Multiple Sclerosis Following DMT Discontinuation: Marisa McGinley, DO

WATCH TIME: 6 minutes | Captions are auto-generated and may contain errors.

"Interestingly, if you look at one patient compared to another, the most important predictive factor differs. As clinicians, we often say, for this patient, it's really how long they've been on their medication. Or for this patient, maybe it's their age. We saw that with this model development—it helps reiterate that different factors are at play for each individual."

Recently, researchers from Cleveland Clinic observed that current guidelines in multiple sclerosis (MS) do not seem to offer an evidence-based approach for discontinuing disease-modifying therapies (DMTs). To address this gap, they recently developed a machine-learning-based tool to predict individualized risk of recurrent MS disease activity following DMT discontinuation. The model was derived from a retrospective cohort of adults with MS who had at least 2 clinic visits between January 2015 and July 2023 and was externally validated using data from the DISCOMS trial (NCT03073603). The primary outcome was MS inflammatory activity, including relapse, new T2 lesions, or gadolinium-enhancing lesions, assessed at 2 years.

The analysis, presented at the 2025 European Committee for Treatment and Research in Multiple Sclerosis (ECTRIMS) Congress, held September 24-26, in Barcelona, Spain, included a total of 1104 patients in the development cohort (discontinued, n = 184) and 259 DISCOMS participants (discontinued, n = 131). Presented by lead author Marisa McGinley, DO, the model demonstrated similar performance in both cohorts and predicted an average recurrent disease activity risk of 12.6% for discontinuers. Investigators reported key predictors including months since any inflammatory activity, duration of current DMT use, age, months since last relapse or MRI activity, and DMT efficacy.

Following the presentation, McGinley, a staff neurologist at Cleveland Clinic’s Mellen Center for MS, spoke with NeurologyLive^® to discuss how her team went about using the individualized risk prediction model in patients with MS considering DMT discontinuation. She noted that the model provided patient-specific predictions, highlighting how different factors such as age, disease duration, or treatment history could affect individual risk. Furthermore, McGinley emphasized that this tool supports shared decision-making rather than dictating treatment choices, offering clinicians concrete data to guide care discussions with their patients.

Click here for more coverage of ECTRIMS 2025.

REFERENCES
1. McGinley M, Felix C, Lapin B, Corboy J, Ontaneda D, et al. Individualized decision support tool to predict risk of recurrent disease activity after disease modifying therapy discontinuation in multiple sclerosis. Presented at ECTRIMS Congress; September 24-26, 2025; Barcelona, Spain. Abstract O119.