Cognitive Cardiovascular Disease Prediction Using ECG-Crown Auto Encoder Neural Network for ECG Time Series Signals

Cognitive computing research in cardiovascular disease prediction is needed in the post-COVID-19 era. Recently, post-vaccination anomalies have been declared at certain levels throughout the world, and in rare cases, the vaccine causes Thrombosis Thrombocytopenia Syndrome, which causes blood clot formation. The core threat is that if such clots travel to heart arteries, there is a possibility of sudden cardiac arrest, cardiac arrhythmia diseases, and other heart diseases. The heart’s function can identify such threats, and the primary tool for cardiac function tests is an electrocardiogram (ECG). So, analyzing the heart rate and heartbeat status more accurately is necessary. The manual analysis for cardiac arrhythmia using ECG requires manual QRS-complex identification, which may mislead the precise disease prediction. Hence, the proposed research focuses on fast decision by enhancing the cognitive computing accuracy of ECG signals through the proposed ECG-Crown autoencoder with an LSTM sequential model. The ECG-Crown auto-encoder neural network gives fast classification results as the downsampling is executed concurrently, making the arrhythmia prediction possible even for a single heartbeat by ECG waveform R-R peak analysis. Hence, the proposed model performance is very impressive compared to existing techniques. The proposed system is tested with ECG time series signal data, and the results are very promising, with 99.7% accuracy. The model loss recorded minimum, which lowers the classification error and subsequently helps to predict a cardiovascular disease very accurately.