The purpose of this study was to develop simulation models for the planning of the stimulation timing and to evaluate the possibility of detecting swing-though gait cycle using an accelerometer and a machine learning technique (Neural Network). Two non-disabled adult males volunteered for this study. One 2-axis accelerometer, heel sensor and infrared rays sensor were used for the sensors. For Neural Network training, acceleration data was processed with input data, and the infrared rays data and heel data were processed with the target data. The microcomputer produced output signals using the Neural Network program. The accuracy of the microcomputer output data was compared with the motion analysis data. The Neural Network detector could correspondingly detect predict the beginning of gait cycles. The present system has a potential to access the reconstruction of FES assisted swing-through gait with free-knees in paraplegic patient.
