Supporting public health policy decisions through live birth predictions for health regions of Goiás with machine learning

• Ano de defesa: 2023
• Autor(a) principal: Vitória, Arthur Ricardo de Sousa
• Orientador(a): Galvão Filho, Arlindo Rodrigues
• Banca de defesa: Galvão Filho, Arlindo Rodrigues, Coelho, Clarimar José, Soares, Anderson da Silva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Programa de Pós-graduação em Ciência da Computação (INF)
• Departamento: Instituto de Informática - INF (RMG)
• País: Brasil
• Palavras-chave em Português: Aprendizado de máquina; Predição de nascidos vivos; Séries temporais univariadas; Séries temporais multivariadas
• Palavras-chave em Inglês: Machine learning; Live births prediction; Univariate time series; Multivariate Time Series
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO
• Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tede/12860

The use of forecasting models is becoming even more common in healthcare and administration applications because they can be reliable decision support tools. The live birth rate is a health index that is directly linked with maternal and newborn health, and its prediction can assist health managers to anticipate resources destined for obstetric and pediatric services. Thus, the objective of this work is to forecast the number of live births in the state of Goiás (Brazil) for a 24-month horizon, providing useful information to support the planning and implementation of public policies. This study investigates two distinct approaches: univariate and multivariate, allowing a better understanding and management of the Brazilian territorial hierarchy. Both approaches are evaluated with data provided by the information system on live births of the information department of the single health system (SINASC-DATASUS). The dataset is composed of 252 monthly records of the number of live births for the 18 health regions of Goiás. The results were measured in prediction ability by Mean Absolute Percentual Error (MAPE) and Mean Absolute Error (MAE). For the univariate approach using a LMU, the average MAPE and MAE achieved were 6.4614 and 19.9136, respectively. The multivariate approach was combined with the K-means method for clustering similar time series using a dynamic time warping measure, generating an average result of 5.5985 and 18.1360 for MAPE and MAE, respectively.