Neural networks based system for the supervision of therapeutic exercises

Abstract

Present contribution describes application of the neural networks based models to detect incorrectly performed therapeutic exercises within the frameworks of wearable supervision system. Electronic accelerometers and gyroscopes attached to the human upper and lower limbs gather information about performed exercise in real time. Trained, on the data describing correctly done exercises, neural network based dynamic model of the limb is used to find the difference between the actual and "ideal" performances and judge if exercises are performed in a correct way or not. © 2012 Springer-Verlag.