In this paper, a Logistic regression model is established for the integration of dance and mental health in colleges and universities, and a decision tree is selected as the analysis algorithm. To further improve the model effect of the Logistic regression model, Logistic regression is combined with Noisy-or model to establish a fusion model based on LRANO model, which realizes the improvement of the model accuracy and the adaptation to the changing data. Finally, we set up a comparison experiment with or without college dance course study to explore the integration of college dance courses and college students’ mental health education with the help of the fusion model. After 12 weeks of dance course study, the six body shape indicators such as weight, thigh circumference, and body fat percentage of the subject college students changed with significant differences (P<0.05). Improvement was seen in all physical fitness indicators, including sit-ups with a 58.6% increase. Among the 15 positive psychological quality factors of the subject college students, there was a significant increase in the cooperativeness, faith, and hope factors from 3.26 and 3.5 to 4.02 and 3.95.