Integrating Deep Learning and Reinforcement ...

Integrating Deep Learning and Reinforcement Learning for Adaptive Multimodal Cardiovascular Classification

Author Name : Islam D. S. Aabdalla, D. Vasumathi

DOI: https://doi.org/10.56025/IJARESM.2025.1304251296

ABSTRACT Cardiovascular diseases (CVDs) remain the leading cause of mortality worldwide, highlighting the urgent need for improved diagnostic methods. This paper presents an innovative hybrid framework that combines deep learning and reinforcement learning(RL) to enhance CVD classification using multimodal signals from electrocardiograms (ECG) and phonocardiograms(PCG). The proposed architecture incorporates convolutional neural networks (CNNs) for spatial feature extraction, recurrent neural networks(RNNs) for temporal sequence modeling, and RL for adaptive decision-making, optimizing prediction accuracy. Multimodal fusion techniques, including early, late, and hybrid fusion, are systematically employed to improve integration of the signals. The framework is evaluated on two benchmark datasets, the PhysioNet 2016 Challenge and the EPHNOGRAM dataset, achieving a peak accuracy of 94%. The hybrid LSTM-BiLSTM model outperforms existing state-of-the art methods, demonstrating its potential for precise CVD diagnosis. These results provide a robust framework for automated cardiovascular diagnostics and offer a foundation for future advancements in combining deep learning with RL in healthcare applications.