Método e sistema para diagnóstico de arritmias cardíacas baseado em coeficientes de predição linear e rede neural convolucional 1D implementado em plataforma ESP32

Abstract

Cardiovascular diseases (CVDs) are one of the major problems of modern society. Besides reducing people’s quality of life, resulting in high costs for the healthcare system and loss of economic productivity, they are the leading cause of death in Brazil and worldwide. Early diagnosis and treatment are the best measures to minimize the damage and costs caused by them. To this end, the use of Artificial Intelligence (AI) technologies such as 1-Dimensional Convolutional Neural Networks (1D-CNNs) for detecting and classifying patterns in electrocardiogram (ECG) signals and the use of remote technologies have been recommended in the literature. This work aimed to evaluate the planning prediction of an arrhythmia card identification (ARC) routine based on a Linear Prediction Coefficients (LPC) algorithm for determining heart rate and a 1D-CNN on the ESP32 microprocessor platform. The overall accuracy achieved was 90,1% with an algorithm processing time of 1835 ms.