Morpheus - Ambiente de coleta, processamento e pré-diagnóstico do bruxismo com base em inteligência computacional

Abstract

The diagnosis of sleep disorders is usually made using polysomnography. This article focuses on improving the conditions of this analysis for bruxism. He develops a system for collecting, processing and identifying events related to bruxism, which serves as a pre-diagnosis to help analysts confirm the pathology. An initial hardware was created using a microcontroller and piezoelectrics for measurement. Diagnostics uses a multilayer perceptron neural network and steps to prepare and handle input and output data. The classifier achieved an accuracy of 93.75% in identifying bruxism events in a separate dataset from the creation of the classifier. These results show the initial effectiveness of the system, allowing further testing for improvements and possible commercial development of the device.