Interpretable Random Forest-Based Heart Disease Prediction with Clinical Feature Importance Analysis

Heart disease prediction from heterogeneous clinical attributes requires ML models capable of capturing nonlinear relationships among physiological and diagnostic variables. This research presents a Random Forest-based methodology for classification of binary heart disease data utilizing the publicly available Heart Disease dataset which includes 1,025 samples with 13 input features. The proposed approach includes data quality verification, categorical feature encoding, target generation, 80-20 training-validation split, ensemble classification, probability generation, and feature importance analysis. The RF model trained on 100 decision trees and Gini impurity index is measured for accuracy, precision, recall, F1-score, ROC-AUC, and confusion matrix analysis. This model achieves an accuracy of 92.7%, along with a ROC-AUC score of 0.94, demonstrating consistent classification performance on the independent testing subset. Analysis for feature importance reveals that some of the significant features include blood pressure, cholesterol, age, and heart rate. This model provides an explainable computational approach for heart disease risk classification.