An Edge AI Enabled IoT Cardiac Decision Support System Using RF Fuzzy Feature Selection and Efficient Deep Neural Networks

Cardiovascular disease (CVD) is the leading cause of global mortality, responsible for approximately 17.9 million deaths each year. This underscores the need for early, accurate, and scalable diagnostic solutions. As IoT-enabled healthcare and edge computing become more prevalent, there is growing interest in intelligent cardiac decision support systems that can operate under resource constraints. This study introduces an Edge-AI–assisted cardiac diagnosis framework that combines Random Forest (RF)–based feature selection, fuzzy logic–driven feature optimization, and an efficient Deep Neural Network (DNN) classifier. The framework is validated using the UCI Heart Disease dataset. RF selects the most relevant clinical features, reducing dimensionality and computational cost while preserving essential diagnostic information. Gaussian fuzzy membership functions, developed with clinical expertise, address uncertainty in physiological data and optimize the selected features. A lightweight DNN enables low-latency, energy-efficient real-time inference at the edge. The system achieves 99.4% accuracy, 98.7% precision, 97.9% recall, and an AUC of 0.998. This approach is well suited for IoT healthcare applications, including wearable sensors, WBANs, and edge computing environments, providing a scalable, interpretable, and practical solution for resource-limited healthcare systems.