<![CDATA[Lung disease is a health issue associated with mortality risk; therefore, an analysis is needed to identify factors associated with the risk of death and to model patient survival risk. This study aims to identify factors associated with mortality risk in patients with lung disease using Cox Proportional Hazards Regression and to develop a predictive model using Gradient Boosting Survival. The study utilized medical records from 1,119 inpatients with lung disease at Pertamina Hospital in Balikpapan during the 2024–2025 period. Analyses were conducted using Kaplan-Meier, Cox Proportional Hazards Regression, and Gradient Boosting Survival. Hyperparameter optimization for Gradient Boosting Survival was performed using Grid Search with 5-fold stratified cross-validation on the training data. The results of the Cox Proportional Hazards Regression showed that BMI, diastolic blood pressure, and Chronic Obstructive Pulmonary Disease (COPD) were significantly associated with the risk of mortality, with a C-index of 0.6500. Gradient Boosting Survival yielded the optimal hyperparameters: a learning rate of 0.10, n_estimators of 200, max_depth of 2, and min_samples_leaf of 10, with a C-index of 0.6576 on the training data and 0.5575 on the testing data. Feature importance analysis shows that BMI, diastolic blood pressure, systolic blood pressure, SpO₂, and COPD make the largest contributions to the model. Cox Proportional Hazards Regression provides information on the direction and magnitude of the relationship between risk factors through the hazard ratio, while Gradient Boosting Survival provides information on the relative contributions of variables through feature importance. Thus, both approaches provide complementary information in modeling the risk of mortality in patients with lung disease.]]>
