Abstract:Objective To investigate the predictive value of a nomogram prediction model based on the extreme gradient boosting(XGBoost) algorithm for the risk of readmission due to acute exacerbation in elderly patients with stable chronic obstructive pulmonary disease(COPD). Methods A retrospective cohort study included 348 elderly patients with stable COPD hospitalized at the Affiliated Hospital of Chengde Medical College from January 2021 to June 2024. The participants were randomly assigned to a training set(n=244, used for model development) and a validation set(n=104, used for internal validation) in a 7∶3 ratio. First, univariate analysis was performed to screen potential predictors of readmission due to acute exacerbation in stable COPD patients. Subsequently, the XGBoost algorithm was applied to rank candidate variables by importance and perform feature selection. Finally, a secondary screening was conducted using multivariate logistic regression analysis to generate a visualizable nomogram prediction model whose performance was evaluated. SPSS statistics 22.0 and R(4.2.3) were used for statistical analysis. Between-group comparisons were performed using t test orχ2 test depending on data type. Results Compared with conventional full-variable multivariate logistic regression models, the nomogram prediction model constructed in this study based on XGBoost algorithm and multivariate logistic regression screening achieved an area under the curve value of 0.936(95%CI 0.897-0.963) and 0.923(95%CI 0.853-0.949) in the training and validation sets, respectively, for predicting the risk of acute exacerbation readmission in elderly patients with stable COPD. The model also demonstrated high clinical net benefit within the threshold probability range of 0.1-0.7. Conclusion The nomogram prediction model based on the XGBoost algorithm has good discrimination, calibration, and clinical utility, and can effectively predict the risk of acute exacerbation readmission in elderly patients with stable COPD.