Aplicação do método de quantis no estudo e diagnóstico do Transtorno do Espectro Autista através de sinais de EEG

• Ano de defesa: 2024
• Autor(a) principal: Maximo, Mariane Vieira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Transtorno do espectro autista; Redes complexas; Quantis
• Link de acesso: https://hdl.handle.net/11449/255506

Autism Spectrum Disorder (ASD) is a complex condition that affects neurodevelopment, involving areas such as social behavior, communication, and language. Due to its variability of symptoms, diagnosing ASD is challenging and prone to errors, potentially significantly affecting short and long-term quality of life. Early identification and effective therapeutic interventions can improve prognosis. Electroencephalography (EEG), a technique that records brain electrical activity, is a valuable tool for investigating ASD. In research, various computational techniques for analyzing EEG signals have been explored in the literature to automatically detect ASD, such as Entropy, Wavelet Transform, Independent Component Analysis (ICA), among others. However, new studies and methods are needed to deepen understanding of the underlying mechanisms of this condition. In this study, we employed the quantile method, an innovative approach, to analyze EEG time series from patients with ASD and neurotypical individuals from two distinct databases. Our goal was to perform a comparison between the results obtained. Although novel for ASD, the quantile method has already demonstrated efficacy in classifying other conditions, such as Epilepsy and Alzheimer’s disease. Mapping using the quantile method, combined with topological descrip-tors, proved effective in classifying ASD. The use of combinations of topological descriptors in classifiers, such as SVM (Support Vector Machine), with three different descriptors, also demonstrated efficacy in both databases analyzed in this study. Additionally, we identified two brain regions of interest: the occipital lobe and a specific electrode in the frontal lobe. Analysis of the collaboration of brain waves in identifying the disorder revealed that alpha frequency waves yielded the best results in differentiating groups for both databases, further enhancing understanding of the mechanisms underlying ASD. In summary, this study demonstrates the promising application of time series analysis techniques, such as the quantile method, in ASD investigation through EEG. These approaches have the potential to significantly contribute to a deeper understanding of this complex condition and to the development of more accurate diagnostic methods and more effective therapeutic interventions.