Unsupervised machine learning strategy and Shapley Additive Explanation to distinguish gait abnormalities through IMU-based gait analysis in movement disorders.

Distinct, though related, groupings of neurodegenerative illnesses frequently share common symptoms, implying that the underlying pathogenetic mechanism is likely to overlap. Machine learning (ML) automates the analysis of complex datasets to identify patterns that would otherwise be difficult to find using traditional qualitative methods. When paired with gait analysis utilizing IMU, automated classification of gait disorders using Machine Learning algorithms can enable fast and clinically significant assessment of gait anomalies in persons with gait disorders [1]. Unsupervised methods can be used to classify gait patterns of people suffering from various neurological illnesses. Clustering algorithms were assessed in this study to detect patterns and produce insights on sensor data from people with movement problems and healthy subjects.