Gait parameters identification for the differentiation of neurodegenerative diseases using classifiers

Degenerative nerve diseases affect the neuromusculoskeletal system, which has implications for human gait. The study of the pathological gaits using gait analysis has been performed for the disease characteristics extraction and disease classification. This paper presents a statistical analysis of gait parameters of healthy volunteers, patients with Parkinson's disease, Amyotrophic Lateral Sclerosis and Huntington. Using the statistical analysis, a set of descriptive variables was selected. Descriptors such as the mean of the temporal parameters of gait, the energy cost, and some correlated variables such as gender, age and body mass index were used. As classification methods Discriminant Analysis and fuzzy C-means were implemented. The Linear Discriminant Analysis classifies 100% of the Parkinson's disease cases, 90% of Amyotrophic Lateral Sclerosis, 87.5% of healthy and 89.47% of Huntington patients. The fuzzy c-means achieved a lower classification of 62.5% of Parkinson 60%, 60% of ALS and 42.1% Huntington; however, the information from the membership matrix gives important information to give a final clinical diagnosis. The energy cost in the stance phase increased, and the gait speed decreased for critical conditions. Depending on the disease type, the cadence decreases. The variability of the data of Amyotrophic Lateral Sclerosis and Huntington's involves an intra-subject analysis.