A Novel Multimodal Framework for Early Detection of Congestive Heart Failure using Ensemble Learning based Fusion Approach

In recent years, critical heart conditions like myocardial infarction (MI), congestive heart failure (CHF), pulmonary diseases, and non-communicable diseases have become increasingly prevalent. CHF, in particular, poses a significant threat with high mortality rates. It is defined by the heart’s incapacity to pump blood effectively, which causes the body to retain fluid. Various studies have focused on early diagnosis of heart failure using machine learning (ML) algorithms such as Random Forest (RF), Support Vector Machine (SVM), XGBoost, and Deep Learning (DL) models like Tiny Unet. However, these methods heavily rely on signal processing and manual analysis of ECG data, resulting in complexity and reduced accuracy. In order to tackle these challenges, we suggest a multimodal framework for CHF classification using ECG data. This framework extracts discriminatory features from pre-processed data and integrates the accuracy of ensemble learning models. Our approach achieves ${9 8. 5 2 \%}$ accuracy, indicating its potential for clinical use. Future research should explore additional features and modalities to enhance predictive modeling and extend ensemble learning for better MI detection, thereby improving cardiovascular health monitoring in clinical settings.