A machine-learning approach to EMG-based classification of gait phases and sub-phases.

In the present study, a recurrent neural network approach is proposed for the classification of four gait sub phases of the gait cycle (Heel contact, flat foot contact, push off and swing phase) and the consequent prediction of gait events (Heel Strike, Mid foot strike, heel rise and toe off event), considering only sEMG signals as input to the model. To this, sEMG signals from Tibialis Anterior, Gastrocnemius Lateralis, Rectus Femoris, Vastus Lateralis, and Hamstring muscles were acquired from both legs of 30 healthy subjects walking barefoot on the floor for about 5 min at their own pace following an eight-shaped path on the ground.