Early identification of exercise-induced respiratory symptoms in competitive athletes: establishment and validation of a predictive model

This is the authors' abstract. We don't add key points for this paper.

Exercise-induced respiratory symptoms (EIRS) are common among athletes, but cultural factors may deter some athletes from reporting these. Thus, we have developed a predictive model to anticipate the occurrence of EIRS. A survey of athletes from five centers in Chongqing, China, was conducted from December 2021 to December 2022. Risk factors for EIRS were identified using univariate and multiple logistic regression analyses, and a predictive model was constructed. The model’s performance was assessed for discrimination and calibration, and its clinical utility was evaluated with Decision curve analysis (DCA). Model validation was done through 10-fold cross-validation. A total of 147 athletes were included in the analysis, resulting in the establishment of a predictive model consisting of five variables: “sex, type of sport, history of allergy, Gastroesophageal Reflux Disease (GERD), allergic rhinitis” and visualized using a nomogram. Bootstrap ROC analysis showed good model discrimination (AUC, 0.806; 95% CI, 0.731–0.878). Calibration analysis further affirmed the accuracy of the model. 10-fold cross-validation suggested moderate internal predictive performance. DCA indicated its practical applicability. We developed and internally validated a nomogram-based predictive model incorporating five clinical variables to estimate the probability of EIRS in athletes. The model showed satisfactory discrimination and calibration, and Decision curve analysis supported its potential clinical utility. These findings warrant further external validation in larger and more diverse athlete populations before broader clinical application.