Bioespectroscopia de fluorescência em matriz de excitação-emissão (EEM) combinada com algoritmos de classificação de 2ª ordem para o diagnóstico da doença de Alzheimer

• Ano de defesa: 2023
• Autor(a) principal: Santos, Ricardo Fernandes dos
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Rio Grande do Norte; Brasil; UFRN; PROGRAMA DE PÓS-GRADUAÇÃO EM QUÍMICA
• Palavras-chave em Português: Química; Alzheimer's disease; EEM fluorescence spectroscopy; Multivariate classification; PARAFAC-QDA; Tucker3-QDA; CNPQ::CIENCIAS EXATAS E DA TERRA::QUIMICA
• Link de acesso: https://repositorio.ufrn.br/handle/123456789/51947

Alzheimer’s disease (AD) is a neurodegenerative disease responsible for almost 70% of cases of dementia. Dementia, in turn, is the 7th leading cause of death in the world. In recent years there have been significant advances in research to identify AD, however, the methods traditionally used for diagnosis remain invasive, time-consuming, and expensive. Studies with biospectroscopic techniques combined with chemometrics have shown promising results in AD diagnosis, with the possibility of offering a minimally invasive, rapid, and inexpensive method. This thesis presents a new methodological approach for the diagnosis of AD through the analysis of blood plasma from 230 subjects (83 AD and 147 healthy controls) by excitation-emission matrix (EEM) fluorescence spectroscopy combined with second-order classification algorithms. The classification models were validated through the calculation of figures of merit commonly used in clinical studies (sensitivity, specificity and accuracy) and figures of merit that take into account the sample unbalance and the discriminatory power of the models (F2 - score (F2), Matthews correlation coefficient (MCC) and test effectiveness (δ)). The classification models performed in this study were Parallel Factors Analysis with Quadratic Discriminant Analysis (PARAFAC-QDA) and Tucker3 – QDA. The PARAFAC – QDA model obtained 83.33% sensitivity, 100% specificity and 86.21% F2. While the Tucker3- QDA model obtained 91.67% sensitivity, 95.45% specificity and 91.67% F2. Both models showed high overall performance with 94.12% accuracy and 0.87 MCC. The classifiers also showed high efficiency with the mean scores of the classes separated by three or more standard deviations. The wavelength values from both loading profiles of models were used to suggest potential plasma AD biomarkers. Future studies may correlate these wavelengths and the PARAFAC spectral profiles with plasma AD biomarkers and confirm or not our suggestion. The results achieved with the proposed new methodological approach point to a high-performance, blood-based method for the diagnosis/screening of Alzheimer's disease. This method has the advantages of being minimally invasive, fast, inexpensive, non-destructive, and label-free. In addition, it requires a small aliquot of blood plasma and is performed in easy-to-operate equipment.