Output details

An instance based algorithm with auxiliary similarity information, for the prediction of gait kinematics from wearable sensors

Wearable human movement measurement systems are popular in capturing movement data used in assessing musculoskeletal and neurological pathologies and evaluating surgical and therapeutic interventions. A new General Regression Neural Networks approach, using auxiliary similarity information, is proposed for estimation of gait kinematics from inexpensive motion sensors which is robust to model and sensor location is proposed. It exploits gait space characteristics to allow a generic estimation approach that learns from examples. Applications to estimate hip, knee, and ankle joint motions using foot and shank motion inputs and demonstrated robustness in sensor location, subject speed and intra-subject estimation.