Detecção e previsão da apneia obstrutiva do sono através da variabilidade da frequência cardíaca

• Ano de defesa: 2023
• Autor(a) principal: Rocha, Jônatas de Lira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Aprendizado de máquina; Apneia obstrutiva do sono; Detecção e previsão; Eletrocardiograma; Variabilidade da frequência cardíaca; subject.br-rjbn; Área(s) do conhecimento do documento (Tabela CNPq)
• Link de acesso: http://repositorio.ufes.br/handle/10/17391

Obstructive Sleep Apnea (OSA), characterized by temporary pauses in breathing during sleep, poses a significant challenge in the field of healthcare. This research addresses the complexity of OSA, exploring not only its clinical manifestations but also investigating the details of associated biomedical signals, such as Electrocardiogram (ECG) and Heart Rate Variability (HRV). By incorporating machine learning techniques, the aim is to enhance the diagnosis and prediction of OSA, providing a more comprehensive understanding and treatment of this multifaceted medical challenge. In the detection process, classifiers such as Support Vector Machine (SVM), k-Nearest Neighbors (KNN), and Neural Network (NN) are employed. The results indicate that NN stood out in training, achieving an accuracy of 84.9%, while in testing, SVM recorded 82.9%. NN demonstrated effectiveness with 73.7% specificity in detecting normal breathing, contrasting with SVM's sensitivity of 94.7% in detecting apnea. Despite slightly lower performance in testing, KNN maintained equivalent levels of specificity compared to SVM. In the prediction phase, the implementation of Nonlinear AutoRegressive with eXogenous inputs (NARX) neural networks achieved 95% accuracy in training and 94.37% in testing. The sensitivity of 94.74% and specificity of 93.94% in the test highlighted the effectiveness of this approach in predicting moments of apnea and normal breathing. These results are valuable for advancing the detection and prediction of sleep apnea, underscoring the effectiveness of machine learning techniques and neural networks in this challenging clinical scenario.