Classificação de ritmos cardíacos em tempo real aplicando tecnologias embarcadas

• Autor(a) principal: Hübner, Lucas Guilherme
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações do UNIOESTE
• Texto Completo: http://tede.unioeste.br/handle/tede/5114
• topic: Eletrocardiograma; Processamento de sinais; Arritmias; Monitoração; Electrocardiogram; Signal processing; Arrhythmia; Monitoring; BIOENGENHARIA::PROCESSAMENTO DE SINAIS BIOLOGICOS

The aging process makes changes in the cardiovascular structure, increasing the risk of cardiovascular diseases and making people more dependent and vulnerable. The National Institute of Social Security points out that every year around 250 thousand new cases of stroke occur in Brazil, with approximately 40 % of retirement requests because of strokes and heart attacks. Heart disease is one of the most prominent causes of death in the world. Most cases of sudden death occur without previous symptoms, while some non-lethal arrhythmias such as ventricular extrasystoles precede others directly related to sudden death. In this sense, it is advisable to monitor individuals at high risk on a daily basis who are not hospitalized. In addition, considering the aging of the population and the increase in the number of people living alone, it is important that remote monitoring systems for various types of biomedical signals. For this reason a Monitoring System of the Human Body, to work in real time, is being developed at UNIOESTE to helps the monitoring of elderly patients or with cardiac risk. This work is part of the Monitoring System of the Human Body and The main objective is to implement a method for classifying QRS complexes and send alarms to a hospital, doctor and / or guardian. The classifier was chosen comparing 3 methods using the same database. To choose the classifier, we used several approaches, including entropy, fractal dimension and statistical measurements using the same database comparing the performance obtained on each approach. The algorithms tested were: J48, Multilayer Perceptron and Random Forests.