ML algorithms for the assessment of prescribed physical exercises

Home-based physical therapies are specially effective if the prescribed exercises are correctly executed. That is specially important for older adults who can easily forget the guidelines given by therapists. Inertial Measurement Units (IMUs) are commonly used for tracking exercise execution giving information of patients' motion data. In this work, we propose the use of Machine Learning (ML) techniques to asses whether a given exercise is properly executed using data from IMUs. We evaluate the performance of four ML classifiers in the context of binary classification of the performance of a given exercise. We apply our proposal to a set of 7 exercises of the upper-and lower-limbs frequently proposed in physical therapy routines, carried out by 14 volunteers. The findings of this study support the possibility of automatically evaluate exercises in a physical therapy routine, with a misclassification error of 0.5% with the best evaluated algorithm, the support vector machine with a polynomial kernel. Sensitivity and specificity achieve values over 99% in the detection of wrongly performed motions.