Towards Data Augmentation for Parkinson’s Disease Gait Data Using Neuromusculoskeletal Simulations

Machine learning models are powerful tools for applications such as disease classification; however, their effectiveness is often limited by the availability of data. To address this issue, we propose a novel data augmentation method that utilizes neuromusculoskeletal simulations. Parameters fitted to the gait data from patients with Parkinson’s disease were estimated using a linear regression model derived from clinical scales, enabling the generation of augmented data. The proposed method successfully produced parameters similar to those obtained from actual data, and the neuromusculoskeletal model using these parameters was able to simulate the gait. This suggests that the proposed method has the potential for data augmentation.