Cardiovascular Adverse Events After Definitive Chemoradiotherapy for Lung Cancer in an Appalachian Population: Incidence and Machine Learning Based Prediction

Background Cardiovascular adverse events (CVAEs) after chemoradiotherapy (CRT) for lung cancer are major concerns in Appalachia due to high rates of smoking and pre-existing cardiovascular diseases (CVD). The objectives of this study were to characterize the incidence of CVAEs in this population and evaluate machine learning (ML) models for CVAEs risk stratification and mortality prediction. Methods A retrospective study was conducted among Appalachian patients with lung cancer treated with definitive CRT at a single institution between 2013 and 2025. Baseline clinical variables, including demographics, smoking status, pre-existing CVD, and post-CRT CVAEs were collected. Heart dosimetric parameters were also obtained. ML models [Random Forest (RF), Gradient Boosting (GBM), Support Vector Machine (SVM), Logistic Regression (LR)] were trained using 5 fold cross validation and evaluated using AUC, sensitivity, specificity, and F1 score. Feature importance was assessed using permutation analysis. Wilcoxon and Chi-squared tests were used for descriptive comparisons. Results Eighty-six patients (mean age 66 years, 47% male) were included. At diagnosis, 80% (n=69) had NSCLC and 20% (n=17) had LS-SCLC. CVAEs occurred in 51 patients (59%). The most frequent events were NSTEMI (n=15, 29.4%), pericardial disease (n=15, 29.4%), and arrhythmia (n=8, 15.7%). Mean heart dose was higher in the CVAE group (13.4 vs 9.4 Gy, p=0.27). For CVAE prediction, GBM achieved the highest AUC (0.55, 95% CI 0.44-0.69) and sensitivity (75%), while RF showed the highest sensitivity (80%, 95% CI 69-90%). Key predictors included age and cardiac dosimetrists (Heart V20, V40, V50, and mean heart dose). For mortality prediction, RF achieved the highest discrimination (AUC = 0.63, 95% CI 0.496-0.750). Age, cardiac dosimetry, disease stage, and cardiovascular comorbidity were the most influential predictors. Conclusion High incidence of CVAEs occurred among patients with lung cancer treated with CRT in this Appalachian cohort. While ML models demonstrated modest predictive performance, tree-based approaches demonstrated high sensitivity for identifying patients at risk for CVAEs and mortality. Age and cardiac radiation dose metrics consistently emerged as key predictors, highlighting the importance of cardiac dose optimization and ML-based risk stratification for cardio-oncology surveillance.