Automatic Classification of Amyotrophic Lateral Sclerosis through Gait Dynamics

Amyotrophic Lateral Sclerosis (ALS) is a neurode-generative disease characterized by the progressive and specific loss of motor neurons in the brain, causing a variety of symptoms, including weakness of muscles and changes in gait. Currently, there is no cure for ALS, nor there is a definitive diagnostic test that can detect whether someone has ALS. Therefore, there is still a need for alternative and non-invasive methods to aid the diagnosis of ALS. This article proposes an automatic method to aid the diagnosis of ALS. A feature extraction technique based on metrics of fluctuation magnitude and fluctuation dynamics, followed by a machine learning algorithm to separate subjects with ALS from healthy ones, was used. The results showed that the proposed approach is comparable to others in the literature even though a simpler and smaller feature set was considered. Five different machine learning classifiers were compared and evaluated using the leave-one-out cross-validation method. A comparison and discussion of the results based on the foot from which the data were extracted and the phases of the gait were also carried out.