Identifying prototype model patients in Amyotrophic Lateral Sclerosis patients at diagnosis through Archetypal Analysis

The clinical heterogeneity that characterises Amyotrophic Lateral Sclerosis (ALS) makes its diagnosis, prognosis, and care difficult. In this context, characterising patients based on their clinical features or patterns of progression becomes crucial, allowing a deeper understanding of the disease and the planning of more effective treatments. In this work, we employ Archetypal Analysis for studying a real-world ALS population based on their characteristics at diagnosis. First, we derive a set of extreme clinical types (archetypes) whose combination describes the study population, and we analyse their differences in terms of attributes. Then, we cluster patients according to their similarity to the archetypes and we investigate how the so-obtained groups differ in terms of time to life-support interventions and survival.