Predicting Multiple Sclerosis Relapses Using Patient Exposure Trajectories

Multiple Sclerosis (MS) is a prevalent autoimmune neurodegenerative disease characterized by progressive nerve inflammation, leading to increasingly severe symptoms. Approximately 85% of MS cases exhibit a relapse-remitting pattern, where sudden symptom exacerbations (relapses) are followed by periods of improvement (remissions). Previous research has shown that MS progression is influenced by external factors such as weather and air pollution. In this paper, we present a Machine Learning-based approach to predict the timing of MS relapses based on environmental exposure. This work was conducted as part of the Intelligent Disease Progression Prediction (iDPP) CLEF 2024 Challenge, which focused on the impact of environmental exposure on MS progression using retrospective data. Specifically, we utilized two anonymized datasets from clinical institutions in Pavia and Turin, Italy, containing real patient data. We employed Topological Data Analysis to compute personal exposure trajectories and used two predictive approaches, one based on the application of Linear Regression, Random Forest, and Extreme Gradient Boosting models to the last follow-up data, and one based on the application of Mixed-Effects modeling on longitudinal data from the first to the last follow-up. Results suggest that integrating environmental variables yields valuable insights for predicting MS relapses, emphasizing the need for improved methods of calculating personal pollution exposure patterns to enhance the accuracy of MS progression predictions.