CIÊNCIA DE DADOS NA PREDIÇÃO DE EVOLUÇÃO DA COVID-19 EM PACIENTES VIA PLATAFORMA DIGITAL E DESENVOLVIMENTO DE NANOBIOSENSOR IN SILICO

• Autor(a) principal: Schultz, Júlia Vaz
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional Universidade Franciscana
• Texto Completo: http://www.tede.universidadefranciscana.edu.br:8080/handle/UFN-BDTD/1115
• topic: Ab Initio; Docagem Molecular; Inteligência Artificial; COVID-19; Telessaúde.; Ab Initio; Molecular Docking; Artificial Intelligence; COVID-19; Telehealth.; Interdisciplinar

The pandemic of the new coronavirus and the frequent identification of new variants of concern highlighted several weaknesses regarding the methods of containing infection, bringing with it a significant number of deaths accounted for the disease. Several studies point to the importance of identifying risk groups for the development of more severe forms of the disease, in addition to the need for strategies to prevent the collapse of the health system due to the massive number of cases. This work applied different methodologies to develop an artificial intelligence system using machine learning as a proposal for telescreening patients. In parallel, the proposal of a nanobiosensor of the coronavirus at molecular level. This study was based on public data available on the website of the Rio Grande do Sul – Brazil health department to identify trends in cases of recovery and death from coronavirus. A decision tree was built to predict the evolution of the disease from the treated data with 97% accuracy, proving to be superior to the models found in the literature. Thus, the model developed is positive for use in telescreening. In parallel, three phenolic compounds (apigenin, p-coumaric acid, and orientin) associated with three lipid structures (caprylic acid, lauric acid, oleic acid) were used as a proposal for the nanobiosensor. The interaction between the structures was theoretically studied by the ab initio computational methodology. The results show binding energies ranging from -0.16 to -2.70 eV, overlapping energy orbital bands, charge transfer between structures less than 0.2 e-, and the charge density is mostly accumulated in the phenolic compound. Thus, the interactions were stable and resulted in weak binding energies, characteristics of interaction via physical adsorption. Linked to this, the molecular docking methodology studied the interaction of the ligands with the receptor region of the spike protein binding of the coronavirus in the Delta and Ômicron variants. With binding affinity results ranging from -3.88 to -8.63 kcal/mol, all systems studied indicate an inhibitory activity of the spike protein of both variants. Thus, the results are positive for the COVID-19 nanobiosensor proposition.