Title: Machine Learning and Surgical Decision Making

ILE Adviser: Dr. Weihua Guan

Abstract: Patients with severe disease are often at a higher risk for surgical complications; therefore, the decision to operate should be made upon careful consideration. Preoperative health data has the predictive potential to identify high-risk cases and guide physicians to make more informed care decisions. To demonstrate this ability, a dataset documenting cases of lung cancer surgeries with baseline clinical attributes and outcomes at 1-year was analyzed. Following an exploratory data analysis and dataset splitting, three distinct predictive models were developed: logistic regression, random forest, and gradient boosting. Through model development, the aim was to identify preoperative factors that best predict survival after surgery. The performance of each model was evaluated using metrics such as the area under the ROC curve and accuracy, and results were compared between models. The presented findings underscore the value of classification models to promote personalized care decisions in the clinical setting and optimize patient outcomes.