Tensor Decomposition of Foot Pressure for Classification of Neurodegenerative Diseases

Accurate and early detection of neurodegenerative diseases (NDDs), such as Parkinson's disease (PD), Huntington's disease (HD), and amyotrophic lateral sclerosis (ALS), is of paramount importance for timely intervention and effective management. Gait analysis offers a noninvasive and cost-effective method to identify subtle gait pattern differences associated with these conditions. This letter leveraged a publicly available gait dataset that utilizes ground reactive forces for gait analysis. The subject data is converted into tensors and decomposed to form a tabular dataset, which is then used to identify healthy subjects and the three NDD classes. A model ensemble of seven binary support vector machine models is used as the decision system. The presented approach accurately classified the healthy subjects and the three NDDs.