Construction of a prediction model for elderly patients with chronic obstructive pulmonary disease complicated by frailty based on decision tree algorithm

LIU Hong; LU Yun

doi:10.12102/j.issn.2095-8668.2025.14.016

Abstract：Objective: To analyze the risk factors for frailty in elderly patients with chronic obstructive pulmonary disease (COPD) and to construct a decision tree prediction model. Methods: A total of 200 elderly COPD patients admitted to the Health Care Department of Hefei First People's Hospital from September 2021 to September 2022 were selected. Univariate and multivariate analyses were used to screen the main risk factors for frailty, and SPSS Modeler software was used to build a decision tree model for predicting frailty in elderly COPD patients. The receiver operating characteristic (ROC) curve was drawn to evaluate the predictive performance of the model. Results: Among the 200 elderly COPD patients included, 136 patients had no frailty (68.00%) [67 patients (33.50%) were normal, and 69 patients (34.50%) were pre-frail], while 64 patients (32.00%) had frailty. Univariate analysis showed significant differences between the frailty group and the non-frailty group in marital status, polypharmacy, Geriatric Depression Scale-15 (GDS-15) scores, Mini-Nutritional Assessment Short Form (MNA-SF) scores, and Pittsburgh Sleep Quality Index (PSQI) scores (P < 0.05). Logistic regression analysis showed that polypharmacy, being divorced or widowed, high GDS-15 scores, and high PSQI scores were risk factors for frailty in elderly COPD patients, while high MNA-SF scores were protective factors (P < 0.05). The decision tree model selected GDS-15 score, MNA-SF score, PSQI score, and polypharmacy as nodes, with GDS-15 score being the most important predictor. The area under the ROC curve (AUC) of the decision tree model for predicting frailty in elderly COPD patients was 0.940 (95% CI: 0.893–0.986), while that of the logistic regression model was 0.864 (95% CI: 0.813–0.916). The Delong test between the two models showed Z = 2.143, P = 0.032, indicating that the AUC of the decision tree model was better than that of the logistic regression model (P < 0.05). Conclusion: Being divorced or widowed, polypharmacy, high GDS-15 scores, and high PSQI scores are risk factors for frailty in elderly COPD patients, while low MNA-SF scores are protective factors. The decision tree model established in this study has high accuracy and reliability in predicting the risk of frailty in elderly COPD patients, providing personalized nursing strategies that help improve patients' quality of life and prognosis.