Abstract: Disclosed is a finger kneading rating method based on intelligent model processing. The method includes the following steps: acquiring and sending finger kneading piezoelectric data meeting a preset pressure value; receiving the piezoelectric data meeting the preset pressure value, calculating and processing the piezoelectric data by using a data model, and outputting finger effective data; receiving the finger effective data, and inputting a preset training model for training to obtain a rating model; outputting the finger effective data and displaying the finger kneading evaluation result. Through the steps above, the effective times of finger kneading could be easily obtained, and the results of inaccurate counting and scoring in a short time could be avoided, thus greatly ensuring good accuracy and reliability of the test results of this project; meanwhile, the application provides powerful evidence for early identification, early treatment and treatment detection of Parkinson's disease.