A Hybrid Evaluation System for Facial Paralysis Rehabilitation based on Machine Learning and Doctor Experience

In the facial paralysis rehabilitation training progresses, evaluation processes waste huge efforts from both patients and doctors. In the meantime, doctors’ subjective opinions cause inaccuracy result. Therefore, it is urgent to construct an objective grading system to acknowledge doctors of their patients’ exact situation. In this paper, a method is present to construct a hybrid grading system based considering both machine learning results and doctor experience. Firstly, an online marking system based on the web is designed to collect and analyze samples. Then, evaluation models are constructed through TensorFlow from the samples from doctor to grade patients. Finally, various models are mixed to construct the hybrid evaluation system. Results are achieved, which can preliminarily evaluate the patients for their recovery conditions. Although the accuracy is not satisfied enough, it can be seen that the method is effective in facial paralysis rehabilitation training. With improvements in models, an automatic evaluation system will be applied to the rehabilitation of patients.