Hybrid Deep Learning CNN with Optimization for Improved ECG Signal Classification

IOT has greatly enhanced healthcare by linking smart devices and facilitating more efficient communication. This study aims to classify ECG signals using a hybrid optimised deep learning technique that makes use of Convolutional Neural Networks (CNNs) to help in the diagnosing of arrhythmias at early stage. Current deep learning algorithms often only handle singular or deterministic fluctuations of medical factors; this study aims to rectify that. To improve classification accuracy and system performance, the suggested method instead uses a hybrid of convolutional neural networks (CNNs) and Radial Basis Functions (RFDs) and Particle Swarm Optimisation (PSO). The study found that dependable IoT networking is crucial for data aggregation. We show a new hybrid route selection algorithm called REL-RFD-PSO. Its goal is to make data transmission more efficient, reduce packet loss and latency, and replace more traditional methods like Dynamic Source Routing (DSR) and Routing Protocol based on Link Quality (REL) and Energy. The initiative's overarching objective is to enhance healthcare results and system dependability via the development of IoT-based Healthcare Monitoring Systems (HMS). These systems will provide non-invasive, continuous monitoring as well as fast alarms for early arrhythmia diagnosis. The CNN-RFD-PSO hybrid model has the maximum accuracy of 95.15%, 3.23% higher than the CNN model. Integrating RFD and PSO with CNN improves accuracy over other methods. The hybrid CNN-RFD-PSO model has the highest sensitivity, specificity, and F1-Score across all classes, improving by up to 3.43%, 2.1%, and 3.21% above CNN-PSO and CNN-RFD.