Desenvolvimento de um modelo com Inteligência Artificial para predição de Síndrome Respiratória Aguda Grave utilizando Redes Neurais Artificiais na plataforma e-SUS

• Ano de defesa: 2022
• Autor(a) principal: Araújo, Kleyber Dantas Tôrres de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Ciências Exatas e da Saúde; Programa de Pós-Graduação em Modelos de Decisão e Saúde; UFPB
• Palavras-chave em Português: SRAG; Inteligência Artificial; Rede Neural Artificial; Bioestatística; Artificial Intelligence; Artificial Neural Network; Biostatistics; CNPQ::CIENCIAS DA SAUDE::SAUDE COLETIVA
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/123456789/24468

The impact caused by the Covid-19 pandemic provided the integration of technologies and the search for new solutions to mitigate the effect of the crisis on health. The insertion of Artificial Intelligence in the systems used in Primary Care is an ally in the fight against the pandemic. This study aims to present a decision model based on Artificial Neural Networks - Multilayer Perceptron, on the architecture and data on the e-SUS platform for classifying cases of severe acute respiratory syndromes caused by Covid-19, in the municipality of Cabedelo in the state from Paraíba. The solution was developed using the Citizen's Electronic Medical Record database and the Simplified Data Collection of Primary Health Care in the municipality of Cabedelo. A database of 82,827 attendance records for the year 2020 was considered, and a sample of 1,287 SARS cases was collected. The decision model proposed in this thesis was the result of the adjustments of a multilayer neural network model, with three layers and 33 input variables with an array composed of 8, 8 and 2 neurons respectively. The topology generated by this method resulted in an Overall Test Accuracy of 99.87%. This algorithm based on RNA-MLP, integrated with e-SUS, using data from the PEC e-SUS APS, for the classification of SARS cases, allows the insertion of technologies with artificial intelligence in the health management and surveillance system to combat to the SARS.