Prediction of Multiple Sclerosis Disease Progression Using Machine Learning Techniques on Datasets

Multiple sclerosis (MS) is a disease of the central nervous system that causes deterioration of nerves. The purpose of this study is to explore the use of different biomarkers and machine learning methods to predict MS progression, as well as analyzing the data over time instead of simply predicting the current severity of MS in a patient.Two datasets were analyzed in this study: one from Kaggle and one from G-Node Open Data. The Kaggle dataset contained a variety of medical features, whereas the G-Node dataset contained only medical features from motor evoked potential tests. Two machine learning models (scikit-learn Random Forest Regressor and XGBoost Regressor) were developed on the datasets to predict the final Expanded Disability Status Scale (EDSS) score for each patient given various medical and demographic features, as well as their initial EDSS score.Both datasets showed potential for use in prediction of MS disease progression, although the models trained on the Kaggle dataset performed better than the G-Node dataset, suggesting that a variety of biomarkers may be more effective in MS prediction rather than only one biomarker.