An embedded musculo-skeletal injury detection system

Continuous field monitoring of load carriage has proven difficult. An algorithm is proposed for estimating load from a single body-worn accelerometer. The accelerometer used is ADXL335. Accelerometer is interfaced with an Arduino Uno. Platform for the algorithm is Arduino Script and MATLAB. The algorithm has three different steps that characterizes torso movement dynamics. The three different steps are feature extraction, walking detection and multivariate regression. The system maps the extracted dynamics features to load estimates using a machine learning multivariate regression technique. The algorithm is applied, using leave-one-subject-out cross validation, to two field collections of soldiers and civilians walking with varying loads.