Transforming acute exacerbations of chronic obstructivepulmonary disease (AECOPD) risk assessment: Amulti-algorithm machine learning approach for preciseclinical phenotyping

Acute exacerbations of chronic obstructive pulmonary disease (AECOPD) sig-nificantly impact patient outcomes and quality of life. Accurately predictingAECOPD occurrence remains essential for optimizing disease management,yet existing predictive models have notable limitations. A retrospective analy-sis was conducted on 878 patients with chronic obstructive pulmonary disease(COPD) at Zhujiang Hospital, encompassing comprehensive clinical data includ-ing demographic, biochemical, and pulmonary function parameters. Potentialpredictors were identified through univariate Cox regression, and the datasetwas split into 7:3 training-test sets. Ninety-one machine learning algorithmswere constructed to predict AECOPD, with performance compared via con-cordance index (C-index) metrics. Model performance was evaluated usingreceiver operating characteristic curves, k-fold cross-validation, and subgroupanalyses based on disease severity, age, and gender. Five biochemical indicators(including fibrinogen and prothrombin time), six demographic characteristics(including smoking status and age), and three pulmonary function parameters were significantly associated with AECOPD risk. The integrated machinelearning model, which combined stepwise Cox regression and random survivalforest algorithms, exhibited superior predictive performance compared to tradi-tional models (p < .05). Area under the curve, calibration curves, and decisioncurve analyses consistently confirmed the model’s excellent predictive capac-ity. A high-performance AECOPD risk score (AECOPD-RS) prediction modelintegrating multidimensional clinical features was developed. The findingsdemonstrate that multi-algorithm machine learning techniques significantlyenhance AECOPD prediction accuracy and stability. Future validation throughmulticenter prospective studies and incorporation of additional biomarkerscould further optimize individualized COPD management strategies