Baseline Machine Learning Approaches To Predict Multiple Sclerosis Disease Progression

Multiple Sclerosis (MS) is a chronic disease that causes the disruption of the ability of the nervous system to transmit signals thus resulting in a progressive neurologic impairment. Unfortunately, MS onset and progression are extremely heterogeneous across patients that tend to have significantly different management and treatment needs. This heterogeneity in disease progression is among the main aspects of MS that hinder efforts to assess the efficacy of developmental treatments designed to delay disease progression, improve the patient’s quality of life, and prolong survival. Consequently, the prediction of disease progression has vastly gained interest among researchers in recent years, with the main aims of deriving new relevant insights into disease mechanisms and manifestations and enabling better treatment development. However, this crucial point has not yet been sufficiently addressed mostly due to insufficient access to rich clinical datasets and effective methodologies. Developed in the context of the iDPP@CLEF 2023 challenge, this work aims at developing different machine-learning approaches to predict a worsening in patient disability caused by MS using a shared dataset provided by the challenge organisers. Results were modest (C-Index and AUROC ∼ 0.6) and employing non-linear methods did not lead to a discernible advantage with respect to the well-known Cox proportional hazard model. Exploring alternative, more sophisticated, machine learning techniques or improving data pre-processing to obtain more relevant input features may help in augmenting model discrimination and obtaining satisfactory results.