EMG-informed neuromusculoskeletal models accurately predict knee loading measured using instrumented implants

We investigated three different methods for simulating neuromusculoskeletal (NMS) control to generate estimates of knee joint loading which were compared to in-vivo measured loads. The major contributions of this work to the literature are in generalizing EMG-informed and probabilistic methods for modelling NMS control.

A single calibration function for EMG-informed NMS modelling was identified which accurately estimated knee loads for multiple people across multiple trials. Using a stochastic approach to NMS modelling, we investigated the range of possible solutions for knee joint loading during walking, showing the method's generalizability and capability to generate solutions which encompassed the measured knee loads. Through this stochastic approach, we were able to show that a single degree of freedom planar knee is suited to estimating total knee loading, but is insufficient for estimating the directional components of load.