Prediction of Cardiac Arrest through Bio-Inspired Optimization Algorithms

Cardiac arrest is a serious medical condition that necessitates prompt and precise detection to enhance survival rates. This research suggests a novel deep learning model based on Deep Auto-encoders (DAE) and meta-heuristic optimization methods for cardiac arrest detection. Conventional models like Artificial Neural Network (ANN), Support Vector Machine (SVM), and Deep Neural Network (DNN) tend to fail due to poor hyper-parameter tuning and high computational expenses. To improve detection efficiency, we incorporate Elephant Herd Optimization (EHO), Tuna Swarm Optimization (TSO), and Harmony Search Optimization (HSO) for feature selection and parameter optimization. Among these, EHO showed the greatest accuracy, largely enhancing model performance. A data-set of cardiovascular patients was employed to train and test the proposed DAE - EHO model, which outperformed current techniques, such as BPNN, Decision Tree (DT), Random Forest (RF), and Sparse Auto-encoders (SAE), in terms of accuracy, precision, recall, and F-Score. The findings validate that EHO based optimization improves the learning efficiency and predictive accuracy of the model, which outperforms a 93% detection rate. This study highlights the promise of integrating deep learning with EHO based optimization to achieve efficient and accurate cardiac arrest prediction, which can lead to future breakthroughs in medical diagnostics.