Electromyographid (EMG) sihave never been used to predict muscle forces of dynamicaily contracting muscles across subjects. The purpose of this studywas to pndict dynamic muscle force hm processed EMG, knee and ankle angles, and knee and ankle angularvelocities in the cat gastrocnemius and soleus during locomotion. Here. we use an artificial neural network IANN) approach to first derive an EMG-force relationship of skeletal muscle;​ second. use this relationship to predict individual muscle forces for different dynamic tasks within and across subjects;​ and third, validate the predicted muscle forces against the corresponding forces which were experinientally recorded. Our within-subject results were bettei than those published previously. even though we did not incorporate a muscle mode1 or instantaneous contractile conditions into the force predictions. The across-subject results were considered excellent.

We conclude that ANNs represent a powerful tool to capture the essential features of EMG-force relatlonships of dynamicaUy contracting muscle. and that ANNs might be used to predict muscle forces within and across subjects accurately from the corresponding EMG signals.