Análise Preditiva de Sepse em Unidades de Terapia Intensiva utilizando Machine Learning
Resumo
A sepse é a principal causa de morte evitável no mundo, com taxas de mortalidade no Brasil chegando a 60%. Este artigo apresenta um trabalho baseado em dados reais da UTI da Santa Casa de Misericórdia de Belo Horizonte na identificação precoce da sepse. Nossa abordagem integra dados de sinais vitais e o protocolo NEWS com modelos de Machine Learning para prever a deterioração clínica. Resultados preliminares utilizando modelos de Ensemble mostram ser promissores, como o Gradient Boosting com F1-Score de 89% para predição da sepse com 6h de antecedência, 88% para 12h e 85% para 24h, equilibrando o trade-off entre detecção e precisão diagnóstica.
Referências
Bomrah, S. et al. (2024). A scoping review of machine learning for prediction of sepsis and septic shock. Critical Care, 28:180.
Covino, M. et al. (2023). Predicting icu admission and death in the emergency department: A comparison of six early warning scores. Resuscitation, 190:109876.
ILAS (2024). Instituto latino-americano de sepse - dados e estatísticas. Acesso em: 21 dez. 2025.
Li, X. et al. (2023). Machine learning algorithm to predict mortality in critically ill patients with sepsis-associated acute kidney injury. Scientific Reports, 13:5223.
Ren, J. et al. (2024). Inverse set estimation and inversion of simultaneous confidence intervals. Journal of the Royal Statistical Society Series C.
Singer, M., Deutschman, C. S., Seymour, C. W., et al. (2016). The third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA, 315(8):801–810.
Tan, K. M. et al. (2025). Advancing multi-organ disease care: A hierarchical multi-agent reinforcement learning framework. Preprint.
Yin, C. et al. (2024). Sepsislab: Early sepsis prediction with uncertainty quantification and active sensing. In Proceedings of the 30th ACM SIGKDD Conference, pages 459–461.
Zhao, C. et al. (2025). Diffmv: A unified diffusion framework for healthcare predictions with random missing views and view laziness. In Proceedings of KDD.
Covino, M. et al. (2023). Predicting icu admission and death in the emergency department: A comparison of six early warning scores. Resuscitation, 190:109876.
ILAS (2024). Instituto latino-americano de sepse - dados e estatísticas. Acesso em: 21 dez. 2025.
Li, X. et al. (2023). Machine learning algorithm to predict mortality in critically ill patients with sepsis-associated acute kidney injury. Scientific Reports, 13:5223.
Ren, J. et al. (2024). Inverse set estimation and inversion of simultaneous confidence intervals. Journal of the Royal Statistical Society Series C.
Singer, M., Deutschman, C. S., Seymour, C. W., et al. (2016). The third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA, 315(8):801–810.
Tan, K. M. et al. (2025). Advancing multi-organ disease care: A hierarchical multi-agent reinforcement learning framework. Preprint.
Yin, C. et al. (2024). Sepsislab: Early sepsis prediction with uncertainty quantification and active sensing. In Proceedings of the 30th ACM SIGKDD Conference, pages 459–461.
Zhao, C. et al. (2025). Diffmv: A unified diffusion framework for healthcare predictions with random missing views and view laziness. In Proceedings of KDD.
Publicado
01/06/2026
Como Citar
SANTOS, Ana F.; ALVES, Gustavo A.; TAVARES, Mariana F.; ANDRADE, Michelle.
Análise Preditiva de Sepse em Unidades de Terapia Intensiva utilizando Machine Learning. In: SIMPÓSIO BRASILEIRO DE COMPUTAÇÃO APLICADA À SAÚDE (SBCAS), 26. , 2026, Ouro Preto/MG.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 1451-1456.
ISSN 2763-8952.
DOI: https://doi.org/10.5753/sbcas.2026.21629.
