Precision Diagnosis of Coronary Artery Disease with OTLGBM
Coronary Artery Disease (CAD) causes the highest number of deaths worldwide, determining the early need for accurate diagnostic methods. This paper proposes an improved method for diagnosing CAD using Machine Learning models. The methodology aims to enhance diagnosis with the Optimal Tuned Light Gradient Boosting Machine (OT-LGBM) using Bayesian optimization. In addition to the optimization, the feature selection with the LGBM is incorporated into the framework for improving the model's prediction accuracy. The selected important features are trained with the model, and the optimization model tunes the LGBM hyperparameters, significantly reducing computational load and the risk of model fitting issues. The experimental results of the proposed OT-LGBM model outperform Bayesian and standard LGBM models along with other standard ML algorithms with the performance metrics scores of 98.64% in Area Under the Curve and 96.82% in accuracy. Our proposed OT-LGBM model also exhibits superior performance in CAD diagnosis compared with past research. By improving the early diagnosis of CAD, the OT-LGBM model could significantly impact clinical practices, leading to better patient outcomes and potentially saving lives.