Construction of a Personalized Prediction Model for Diabetic Lung Infection Based on Machine Learning.

Objective: This study aimed to develop and validate a clinical feature-based nomogram to predict the risk of lung infection in diabetic patients.

Methods: A total of 168 patients diagnosed with pulmonary infections at our hospital-comprising both diabetic and Nondiabetic individuals-were retrospectively enrolled and divided into a training cohort and an internal validation cohort. Feature selection was performed using the least absolute shrinkage and selection operator (LASSO) method, followed by multivariate logistic regression analysis to construct the predictive nomogram. Model performance was evaluated through calibration curves, receiver operating characteristic (ROC) analysis, and decision curve analysis (DCA) to assess predictive accuracy, calibration, and clinical utility, respectively.

Results: Multivariate analysis identified advanced age, male sex, abnormal neutrophil count, elevated glycated hemoglobin (HbA1c), and fasting plasma glucose (FPG) levels as independent risk factors for diabetic lung infection. A nomogram incorporating these variables and other clinically relevant predictors was constructed. The area under the ROC curve (AUC) was 0.919 (95%CI: 0.825-0.937) in the training set and 0.862 (95% CI: 0.819-0.912) in the validation set, indicating strong discriminative ability. Calibration curves demonstrated good agreement between predicted and observed outcomes. DCA confirmed the nomogram's clinical value across a wide range of threshold probabilities.

Conclusion: We developed a robust and clinically applicable nomogram for predicting the risk of pneumonia in diabetic patients with pulmonary infections. This model exhibits high accuracy and may assist clinicians in identifying high-risk individuals who could benefit from early preventive measures and timely interventions.