An OCSVM Based Assessment Method for Robotic Rehabilitation

The automatic assessment of rehabilitation level is an important part of the automatic rehabilitation training system. In order to carry out an accurate rehabilitation grade assessment for patients with ankle dyskinesia, this paper proposes a grade-variable rehabilitation assessment strategy. Firstly, the plantar pressure insoles are used to collect normal people's plantar pressure data samples as training set of the model and a decision boundary is obtained through the One-Class Support Vector Machine (OCSVM) model for the detection of abnormal data. Then, the patient's plantar pressure data is used as a test set to output the abnormal degree of the current ankle dyskinesia. The abnormal degree is introduced into the control process of the ankle rehabilitation robot, corresponding to realize the rehabilitation strategy of variable grades. Experiments verify that the model established in this paper is a concise, staged and graded periodical rehabilitation assessment method.