The modern physical education teaching cannot satisfy the requirements of social advancement anymore. It has prevented the realization of the Health China Strategy due to its boring format of instructions and poor quality of teaching. With a focus on the current relevance of the Healthy China Strategy, the present paper presents the LacePose network as the way of collecting sports action posture data and storing the collected data in the form of a sample library and dataset. Due to the benefits of machine learning approaches to sports learning, logistic regression is chosen to create a sports action evaluation system and further develop an intelligent physical education teaching system. The model and the system are then examined and analyzed through the use of the research data of the present study. The p-value of the logistic regression algorithm is 0.7-0.95 and that of the random forest algorithm is 0.6-0.7, which confirms the efficacy of the suggested algorithm. Also, the reliability coefficient of the given system is 0.9 -0.95, which is significantly higher than that of the other two sports teaching systems. This implies that the proposed system has good performance and can offer better support in the teaching of physical education in colleges and universities. Through the search of a novel development direction of physical education, the education of physical education in colleges and universities will be able to better facilitate the execution of the Healthy China Strategy.
