Wrist Movements Estimation Based on Limited Training Dataset

This paper presents a novel approach to decode multi-DOF wrist movements using a single-DOF training dataset. The main machine learning model employed is the support vector regression (SVR), with R2 as the evaluation index. Through a thorough comparison of slide window lengths, model parameters, and different machine learning models, we demonstrate the effectiveness of our proposed approach. Longer slide window lengths enhance decoding accuracy but come with increased computational complexity. Fine-tuning SVR model parameters significantly improves performance. Our findings highlight the feasibility of reducing dataset size and alleviating patient burden, offering potential advancements in assistive technologies for individuals with motor disabilities.