Abstract
Sudden cardiac arrest/death (SCA/SCD) is a disease that the heart cannot pump the blood effectively, so the blood flow loses rapidly. The patient may lose consciousness in an hour without appropriate treatment, and may take the patient’s life within minutes. Heart Rate Variability (HRV) is an electrocardiography (ECG) that uses QRS wave detection to calculate the R wave interval (R-R Interval, RRI), and uses the R wave interval to extract the time domain, frequency domain, and nonlinear characteristics of the heart rhythm. This work presents a neural network model algorithm based on heart rate variability for classifying patients with sudden cardiac arrest (SCA) and normal sinus rhythm (NSR). The established neural network model can achieve 87.88% accuracy, 88.89% sensitivity and 87.87% specificity by k-fold cross validation for predicting SCA 55 minutes ago. Since hardware can have a faster computing speed than software, this paper implements the established neural network model on hardware and compares the computing speed with software. The hardware is written in Verilog HDL, and Vivado 2020.2 is used for RTL simulation and verification.
