Foot movement classification based on signals from accelerometer

The assessment of achieved function is essential for both the follow up of therapy and feedback for assistive system. In this paper we present one method of the data processing for signals coming from the inertial sensor. Specifically, we address the assessment of the foot movement; we applied k nearest neighbors classification technique for extracting information regarding qualitative description of the movement. The ankle joint was chosen because it plays a major role in walking, and it needs to be controlled very often in persons after stroke. We investigated influence of foot movement dynamics on classification error. Results suggest a high level of accuracy, which is in the classification. The methods are presented by using the measurements in healthy volunteers. Healthy volunteers were selected in order to create the benchmark that will be later used for assessment of the function in patients with gait abnormalities (e.g., stroke patients).