Heart Disease Prediction Using Different Boosting Models

Nowadays, heart disease (HD) is one of the leading causes of death. The detection of HD is required at an early stage in order to diagnose it beforehand. The traditional method for the detection of HD is becoming more challenging and time-consuming. It is also not available in many healthcare facilities. To overcome this issue, machine learning is an alternative and sophisticated tool. For this purpose, we used five boosting methods, namely Adaptive Boosting (AB), Gradient Boosting (GB), Categorical Boosting (CB), Extreme Gradient Boosting (XGB), and Light Gradient Boosting Machine (LGBM) to predict HD on the UCI Cleveland HD dataset to get the best possible result. After the comparative study of five models, we found that the AB and XGB models are best in accuracy (≈ 93 percent), precision (0.94), recall (0.93), and Fl-Scores (0.93). The accuracy value is equal to that of the recently reported ensemble technique and greater than the accuracy of 91.8 percent of a stacking classifier in the same dataset.