KMAN Publication Inc. Quality of Life and Health Sciences 1 Serial Number 1 2025 07 01 Predicting Quality of Life Among Patients With Chronic Obstructive Pulmonary Disease Using Random Forest and Gradient Boosting Algorithms Predicting Quality of Life Among Patients With Chronic Obstructive Pulmonary Disease Using Random Forest and Gradient Boosting Algorithms 1 15 10.61838/59906r95 EN Journal Article 2025 03 15 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> This study aimed to predict quality of life among patients with chronic obstructive pulmonary disease using Random Forest and Gradient Boosting algorithms and to identify the most important clinical, respiratory, psychological, and comorbidity-related predictors of health-related quality of life.</p> <p><strong>Methods and Materials: </strong>This descriptive, cross-sectional, predictive study was conducted among 342 patients with chronic obstructive pulmonary disease who were selected from pulmonary clinics and hospital-affiliated respiratory care units in Arak, Iran. Eligible participants had a confirmed diagnosis of chronic obstructive pulmonary disease and completed standardized clinical and patient-reported measures. Health-related quality of life was assessed using the St. George’s Respiratory Questionnaire. Predictor variables included demographic characteristics, disease duration, spirometric indices, exacerbation and hospitalization history, COPD Assessment Test score, modified Medical Research Council dyspnea score, anxiety and depression scores, body mass index, and comorbidity burden. Data were analyzed using descriptive statistics, correlation analysis, and supervised machine learning regression. The dataset was divided into training and testing subsets, and Random Forest and Gradient Boosting models were developed using cross-validation and hyperparameter tuning.</p> <p><strong>Findings: </strong>Significant associations were found between poorer quality of life and higher COPD Assessment Test score, dyspnea severity, exacerbation frequency, hospitalization history, anxiety, depression, disease duration, body mass index, and comorbidity burden. Pulmonary function indices, including FEV1 percentage predicted, FVC percentage predicted, and FEV1/FVC ratio, were negatively associated with total quality-of-life score. In the independent testing set, the Random Forest model achieved a mean absolute error of 7.62, root mean square error of 9.81, and coefficient of determination of 0.64. The Gradient Boosting model demonstrated superior performance, with a mean absolute error of 6.93, root mean square error of 9.12, and coefficient of determination of 0.68. The most important predictors were COPD Assessment Test score, dyspnea severity, exacerbation frequency, depression, and FEV1 percentage predicted.</p> <p><strong>Conclusion: </strong>The findings indicate that machine learning algorithms, particularly Gradient Boosting, can predict quality of life among patients with chronic obstructive pulmonary disease with acceptable accuracy. Quality-of-life impairment was best predicted through a multidimensional profile combining symptom burden, dyspnea, exacerbation history, psychological distress, and pulmonary function.</p> </td> </tr> </tbody> </table> <p> </p> https://journals.kmanpub.com/index.php/qlhs/article/download/5753/10904