Visualização, interação e análise de dados do COVID-19 nos sistemas prisionais

• Ano de defesa: 2023
• Autor(a) principal: Nunes, Lucas dos Santos
• Orientador(a): Ordonez, Edward David Moreno
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Programa de Pós-Graduação: Pós-Graduação em Ciência da Computação
• Palavras-chave em Português: Prisões; Algorítmos computacionais; Aprendizado do computador; COVID–19 (doença); Sistema prisional; Dados abertos; Mapeamento sistemático
• Palavras-chave em Inglês: Prison system; Machine learning; Open data; Systematic mapping
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://ri.ufs.br/jspui/handle/riufs/19476

Context: The newly identified coronavirus pandemic, later named COVID-19, is highly transmissible and pathogenic. An additional concern in this context refers to the entry and spread of the disease in Brazilian prisons, whose conditions of incarceration in the country are overcrowded and poorly ventilated cells, which makes these environments extremely susceptible to the rapid spread of the disease. Objective: To analyze COVID-19 data in the prison system to characterize its use in terms of sources, purpose, and data availability. Methods: To present the elaboration of a systematic mapping, to identify publications related to COVID-19 in the prison system that use records from the Department of Informatics of the Unified Health System (DATASUS) and how they are being treated. Next, information is collected on COVID-19 and other diseases classified as cardiac, respiratory, infectious, and mental in the state of Bahia, Brazil, where they will be processed and stored. The intention is to convert these pathological data into a visual panel, called a dashboard, that is, a representation of the pandemic through easily understood visual attributes, such as graphs, histograms, and geographic maps. The panel will be developed with the collaboration of Nielsen heuristics and usability tests with invited participants, so that they can help interested researchers and organizations, mainly in the area of health and computing. Finally, Machine Learning (ML) algorithms were used to predict the spread of these diseases, to help those responsible to take emerging actions. Results: 125 publications were reviewed, of which 29 were identified as relevant to the objectives of the systematic mapping carried out. Then, 8 more studies were found using the Forward Snowballing (FS) technique, totaling 37 studies. As for the experiments with ML, the Polynomial Regression models obtained the best measurements with DATASUS data. The results point to trends and need for research on the subject, as the idea is to help reduce the mortality rate from COVID-19 and other diseases. The contributions made by test participants have become important for the development of the dashboard, so that we can better understand how it works, in addition to the possibility of improving it even further.