Integrated feature selection and ensemble learning for heart disease detection: a 2-tier approach with ALAN and ET-ABDF machine learning model

The findings of this investigation give a novel approach to the forecasting of heart disease. For the purpose of determining significant features, it is a 2-tier procedure that uses a combination of the analysis of variance (ANOVA) and Least Absolute Shrinkage and Selection Operator (lasso) regression known as ALAN. In addition to this, it uses the Ensemble Technique with Adaptive Boosted Decision Fusion (ET-ABDF) in order to enhance its accuracy. An improvement in the detection of critical properties for predictive modelling is achieved through the use of the ALAN technique. With an accuracy of 88.0%, precision of 89.81%, recall of 89.80%, and an F1 Score of 89.00%, the ET-ABDF model shows outstanding results in the identification of cardiac disease. The outstanding result of 96.21% for the area under the curve (AUC) shows its strong ability to differentiate across several groups. It has been established through the comparative analysis that the model that has been provided displays better results in comparison to other algorithms. It is possible that in the future, undertaking more research on new techniques for choosing characteristics and expanding studies to encompass a larger spectrum of cardiovascular problems will have the potential to increase the accuracy and practicability of heart disease prediction models in real-world healthcare settings.