Cardiac Diagnosis System Using ECG and PCG Signals with a Multi-Scale Pooling-Based Deep Neural Network (MSP-DNN) Classifier

Globally, heart disease (HD) is one of the main causes of death, and survival rates are greatly increased by early identification. This paper proposes a Multi-Scale Pooling-based Deep Neural Network (MSP-DNN) for HD detection using ECG and PCG signals. The system involves pre-processing steps, including noise removal for ECG and denoising for PCG, feature extraction in both time-domain and frequency-domain for ECG, and amplitude and frequency analysis for PCG. These features are then analyzed for correlations, which are input into the MSP-DNN model for classification. A TMF-FIS is used to assess survival rates. According to experimental results, the proposed model performs better than existing models regarding f1-score, accuracy, precision, recall, sensitivity, and specificity.