Integração de análises ômicas e Inteligência Artificial no estudo da Doença Renal Crônica

• Ano de defesa: 2024
• Autor(a) principal: Picolo, Bianca Uliana
• Tipo de documento: Tese
• Tipo de acesso: Acesso embargado
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Genética e Bioquímica
• Palavras-chave em Português: Doença Renal Crônica; Chronic Kidney Disease; Biomarcadores; Biomarkers; Proteômica; Proteomics; Metabolômica; Metabolomics; Inteligência Artificial; Artificial Intelligence; CNPQ::CIENCIAS BIOLOGICAS; Genética; ODS::ODS 3. Saúde e bem-estar - Assegurar uma vida saudável e promover o bem-estar para todos, em todas as idades.
• Link de acesso: https://repositorio.ufu.br/handle/123456789/44395; http://doi.org/10.14393/ufu.te.2024.5077

Chronic Kidney Disease (CKD) is a progressive and multifactorial condition that affects millions of people worldwide, being recognized as a significant public health issue due to its high prevalence, morbidity, and mortality. Characterized by the gradual and irreversible loss of renal function, CKD often remains underdiagnosed, particularly in its early stages, due to the absence of specific symptoms and the limitations of conventional diagnostic methods. Underdiagnosis prevents early interventions, promoting the progression of the disease to advanced stages and increasing the risk of cardiovascular, metabolic, and inflammatory complications. In this context, the study of biomarkers has emerged as a crucial tool to overcome these limitations, enabling early diagnosis, monitoring of disease progression, and the personalization of therapeutic strategies. This study investigated the application of omics methodologies, such as proteomics and metabolomics, integrated with machine learning algorithms, with the aim of identifying differential biomarkers and developing predictive models for the diagnosis, prognosis, and monitoring of CKD using biological matrices such as saliva and urine. Proteomic analysis of saliva identified differential proteins such as API-5, PI-PLC, and Sgsm2, with potential for early diagnosis and monitoring of CKD. Additionally, metabolomic analysis in saliva and urine revealed differential metabolites that played a central role in the predictive models developed. These models, based on algorithms such as Support Vector Machines (SVM), Random Forest, Gradient Boosting, and Neural Networks, demonstrated high performance, as evidenced by metrics such as elevated accuracy, sensitivity, and specificity. Lipids stood out as relevant markers, reinforcing their importance in the disease's pathophysiology and in the construction of diagnostic and prognostic models. The combination of artificial intelligence with omics data proved to be a powerful tool for the analysis of complex datasets, enabling significant advances in the understanding and clinical management of CKD. The results demonstrate that the use of salivary proteomics and the integration of metabolomic approaches with machine learning can be considered innovative and promising strategies for the diagnosis and monitoring of CKD, promoting accessibility and contributing to personalized medicine. This work represents a significant contribution to the field of nephrology, highlighting the potential of new technologies to transform clinical practice and improve patients' quality of life.