Inteligência Artificial aplicada a análise de Eletrocardiograma

Abstract

Artificial Intelligence offers mechanisms and automation for predictions in the field of health, benefiting the population and the quality of life of each individual who is being assisted by it. Together, Internet of Things devices are fundamental allies for this system, since they are essential for data extraction and monitoring. However, the challenge arises of reconciling algorithms with smart devices in order to detect and distinguish arrhythmias, cardiac problems with the highest cause of death in the world. The objective of this work is to build a prediction model of cardiac arrhythmias, specifically Arterial Fibrillation and Sinus Bradycardia from data collected from the MP-IoT multiclinical device. An Artificial Intelligence was trained combining the techniques of Bidirectional Long Short-Term Memory, Convolutional Neural Network and Dense Neural Networks. The necessary data for this training were obtained from the Physionet platform, totaling 66,989 data with arrhythmias and normal. The results obtained are romising for predicting arrhythmias with the MP-IoT device, given the evaluation metrics of 96.91%, 96.91%, 96.90%, 96.90% and 98.45% of accuracy, precision, Recall, F1-score and specificity.