Development and validation of a machine learning nomogram model for differentiating mycoplasma pneumoniae coinfection with other pathogens in children

Current diagnostic methods, such as serological tests and polymerase chain reaction (PCR), are limited by slow turnaround time and sample quality, making early diagnosis challenging. Although mNGS offers higher sensitivity, its expense and specialized equipment limit clinical applicability. These limitations underscore the need for more reliable, rapid, and cost-effective diagnostic tools. Between June 2023 and March 2024, 191 consecutive pediatric patients were enrolled. The most recent laboratory test results prior to bronchoalveolar lavage (BAL) were included. After LASSO screening, seven machine learning classifiers (LR, SVM, KNN, RF, ET, XGBoost, LightGBM) were tested, and the optimal one was selected for model construction. The nomogram model combined the radiomics (rad) signature and the clinical signature. The ROC curves were drawn to evaluate the diagnostic efficacy of different models. The calibration efficiency of the nomogram was evaluated by drawing calibration curves, and the Hosmer-Lemeshow test was used to evaluate the calibration ability of the models. Decision curve analysis (DCA) was utilized to evaluate the clinical utility of the models. A p -value < 0.05 was considered statistically significant. A total of 1,834 handcrafted radiomics features were extracted, including 360 first-order features, 14 shape features, and texture features. The LR classifier achieved the best AUC, reaching 0.922 and 0.867 for distinguishing Co-MPP from MPP in the training and test cohorts, respectively. For building the clinical signature, LR was selected as the base model. The univariate analysis results of all clinical laboratory and CT imaging features showed that only reticulation and bronchial lumen occlusion were significantly different between MP and Co-MPP patients ( p = 0.011 and p < 0.001, respectively). The clinical signature achieved AUC values of 0.729 and 0.706 in the training and test cohorts, respectively. A nomogram based on the LR algorithm was constructed to combine the clinical signature and the Rad signature. The DeLong test showed that the performance of the nomogram and the Rad signature was significantly higher than that of the clinical signature ( p < 0.05), whereas no significant difference was observed between the nomogram and the Rad signature. Both the Rad signature and the nomogram demonstrated substantial clinical benefit. Our findings indicate that machine learning can assist clinicians in distinguishing Co-MPP from MPP in children. Furthermore, both the Rad signature and nomogram model showed higher clinical benefit than the clinical signature.