Baseline Machine Learning Approaches To Predict Amyotrophic Lateral Sclerosis Disease Progression

Amyotrophic lateral sclerosis (ALS) is a rapidly progressive neurodegenerative disease that typically leads to death within 3-5 years, characterised by a heterogeneous progression across the patient population. This heterogeneity has hindered efforts to assess the efficacy of developmental treatments designed to delay disease progression and prolong survival. As such, prediction of disease progression has been a long-standing interest in the field as a means of enabling better drug development using cheaper, more accurate clinical trials, as well as deriving new and important insights into disease mechanisms and manifestations. So far, this critical point has not yet been sufficiently addressed due to limited access to patient-level data and sophisticated computational tools. This contribution aims at comparing the performance of different baseline machine learning approaches on a common dataset obtained via the integration of different datasets from different countries provided by the challenge organisers. Results show that the ability of different methods across different subtasks to discriminate among subjects at risk and to predict the time of adverse events improves as dynamic variables, monitoring the first six months of patient follow-up, are included as possible predictors.