Accurate Prediction of Sudden Cardiac Death Based on Heart Rate Variability Analysis Using Convolutional Neural Network
Citations Over TimeTop 10% of 2023 papers
Abstract
Sudden cardiac death (SCD) is a significant global health issue that affects individuals with and without a history of heart disease. Early identification of SCD risk factors is crucial in reducing mortality rates. This study aims to utilize electrocardiogram (ECG) tools, specifically focusing on heart rate variability (HRV), to detect early SCD risk factors. In this study, we expand the comparison group dataset to include five groups: Normal Sinus Rhythm (NSR), coronary artery disease (CAD), Congestive Heart Failure (CHF), Ventricular Tachycardia (VT), and SCD. ECG signals were recorded for 30 min and segmented into 5 min intervals, following the recommended HRV feature analysis guidelines. We introduce an innovative approach to HRV signal analysis by utilizing Convolutional Neural Networks (CNN). The CNN model was optimized by tuning hyperparameters such as the number of layers, learning rate, and batch size, significantly impacting the prediction accuracy. The findings demonstrate that the HRV approach, in conjunction with linear features and the DL method, achieved a higher accuracy rate, averaging 99.30%, reaching 97% sensitivity, 99.60% specificity, and 97.87% precision. Future research should focus on further exploring and refining DL methods in the context of HRV analysis to improve SCD prediction.
Related Papers
- → Evolutionary-based Hyperparameter Tuning in Machine Learning Models for Condition Monitoring in Wind Turbines – A Survey(2021)13 cited
- → Estimation of Sea State Parameters From Measured Ship Responses: The Bayesian Approach With Fixed Hyperparameters(2010)6 cited
- → A Novel Extreme Value Theory Based approach to Hyperparameter Optimization(2023)4 cited
- → No More Pesky Hyperparameters: Offline Hyperparameter Tuning for RL(2022)3 cited
- → Practical Differentially Private Hyperparameter Tuning with Subsampling(2023)2 cited