This study focused on feedback FES control of foot movements of hemiplegic gait. In order to create target movement trajectory, prediction of initial contact (IC) timing would be useful. In this paper, a method for predicting the IC timing using LSTM network was tested. The prediction was based on short-term historical signal data of an inertial sensor mounted on the foot. The prediction test were performed for measured data with 2 healthy individuals. The results showed that the mean and standard deviation of the errors tended to decrease as the percentage of the gait cycle increased, with a significant decrease after heel-off. The proposed method was suggested to be applicable in IC timing prediction.

Abnormal foot movements seen in hemiplegic patients include drop foot and clubfoot. This study aimed to detect and evaluate abnormal foot movements during early stance phase of hemiplegic patients using inertial sensors. Two inclination angles associated with abnormal foot movement were calculated from acceleration and angular velocity obtained from inertial sensors, and reference range of healthy subjects was created by defining a two-dimensional plane with these two angles. In addition, the angle distribution was divided into 3 groups based on the stride length, and the reference range was created for each group. Measured data of paralyzed side of a hemiplegic subject was suggested to be abnormal foot movements by comparing to the reference range. Therefore, the reference ranges of healthy subject for different stride lengths are expected to be useful for the detection and evaluation of abnormal movements of hemiplegic subjects.