Using Bayesian Networks to Explore Risk Factors for Sports Injuries in Chinese Adult Women Who Exercise Regularly: A Nationwide Study

Objective Given the influence of women's unique joint structures and hormonal levels, sports injuries among females have become a focus in sports research. This study aimed to use Logistic regression and Bayesian networks (BNs) models to explore factors associated with sports injuries in Chinese adult females who exercise regularly. Methods This was a cross - sectional study. From October to November 2021, data on sports - injury - related factors were collected through online questionnaires from adult females aged 18 and above who exercised regularly in 336 cities across 34 provinces nationwide. Logistic regression and BNs models were used to explore factors associated with sports injuries in Chinese adult females with regular exercise. Results A total of 6,912 valid questionnaires were included, with a median age of 34.00 (31.00–39.00) years. Among the participants, 4,265 (61.70%) had experienced sports injuries. Logistic regression indicated that age grouping, body mass index (BMI), the most frequent daily exercise time, learning of specialized movement, insufficient energy to complete daily tasks, fatigue or illness status, sleep quality, and awareness of sports injury risks were all risk factors for sports injuries. BNs revealed that age, sleep quality, the most frequent daily exercise time, and fatigue or illness status were directly correlated with sports injuries. Moreover, exercise venue type and learning of specialized movement were indirectly associated with sports injuries through the mediating variable of the most frequent daily exercise time. Conclusion BNs can identify both direct and indirect correlates of sports injuries, and Bayesian risk inference enables risk prediction for sports injuries. BNs serve as a complementary method to logistic regression, providing deeper insights into complex risk factor interactions.