Objective:To develop and validate a clinical diagnosis and prediction model for Mycoplasma pneumoniae infection in paediatric outpatient clinics. Methods:Initial outpatient medical records of 516 children aged 5 to 10 years (239 infected and 277 uninfected cases) have been retrospectively reviewed. The prediction model was constructed using the LASSO-Logistic regression method, and the model was then validated by bootstrapping and 153 cases of different age groups. The area under the subject working curve, the consistency curve, the Brier coefficient and the clinical decision curve were used to evaluate the model’s effectiveness. Results:Fever days, cough days, and white blood cell count(WBC) were finally included to build the predictive model. Internal verification of the bootstrap showed that the area under the subject’s working curve was 0.718, the consistency curve showed that the actual curve was close to the ideal curve, and the Brier coefficient was 0.214. The clinical decision curve showed that the clinical benefit was > 0 when the threshold probability was 0.10-0.58. Verification in different age groups showed that the AUC under the subjects’ working curve was 0.717, the consistency curve was close to the ideal curve, the Brier coefficient was 0.177, and the clinical decision curve was 0.20-0.65 with clinical benefit > 0. Conclusion:In this study, a simple predictive model of Mycoplasma pneumoniae infection in paediatric outpatient clinics was established for the first time, which may provide a reference for early diagnosis and treatment of Mycoplasma pneumoniae infection.