Abstract The study used cluster analysis and decision tree modeling to construct a multivariate data fusion physical fitness evaluation model, and also built a physical fitness core literacy evaluation model through factor analysis and fuzzy comprehensive evaluation method. These two evaluation models were subsequently applied to physical education and health teaching in colleges and universities to measure the quality of physical education teaching in colleges and universities and students’ physical health status. By analyzing the practice of physical education teaching in colleges and universities, it was found that the pull-up program had the worst performance in students’ physical fitness test, with a score of 287. In addition, the lung capacity and 50-meter dash programs contributed the most to students’ total performance, with scores of 1,291 and 1,245, respectively. From the perspective of physical education and health teaching quality, the scores of teaching situation, teaching management and biological education school effectiveness were 4.68, 4.01 and 4.52, respectively, and all of these evaluation grades were good. This study provided an accurate and scientific analysis of the overall quality of physical education and health teaching in universities and the health status of students.