Sepsis Patient Outcome Prediction Using Machine Learning

Author(s)

de Alencar Saraiva, Jose Lucas ; Victor Junior, Marcus Henrique ; Becker Junior, Otavio Monteiro ; Kadirkamanathan, Visakan ; Silva, Eliezer ; Kienitz, Karl Heinz ; CostaFelix, R ; Machado, JC ; Alvarenga, AV

• Document type: Journal article
• Source: XXVI BRAZILIAN CONGRESS ON BIOMEDICAL ENGINEERING, CBEB 2018, VOL 1; v. 70, n. 1, p. 5-pg., 2019-01-01.

Abstract

Sepsis is a life-threatening response of the body to infection that often leads to death. In Brazil, hundreds of thousands of deaths occur every year, with a mortality rate higher than the world average. In this context, it is important to develop tools for decision support and training of healthcare professionals. This work discusses a modelling for sepsis that may be instrumental for achieving such goals. Machine learning is used to train a model for prognosis of outcome of septic patients. The inputs of the model include information about the patient sepsis, previous clinical records and treatment variables. The results show that high accuracy can be achieved and that the network is able to really map patients in two highly distinct groups of risk (survival and fatal outcome), under certain conditions. (AU)

• FAPESP's process: 17/11272-2 - Sepsis prediction: modelling using neural networks and its robustness
• Grantee: José Lucas de Alencar Saraiva
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• FAPESP's process: 17/25497-6 - Artificial Intelligence for personalized sepsis patient outcome prediction
• Grantee: José Lucas de Alencar Saraiva
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation