Early Prediction of Sepsis From Clinical Data Using Single Light-GBM Model

Abstract

Sepsis is a severe medical condition caused by body’s extreme response to an infection leading to tissue damage, organ failure, and even death. The emergence of advanced technologies such as Artificial Intelligence and machine learning, allowed faster exploration of advanced way to recognize sepsis cases. In this paper, we present two main approaches that have been tested using the clinical data. The first method is the combination of survival analysis and neural networks, and the second one is based on booting trees. Our team participated under the name of BERCLAB UND. The proposed model obtained 0.172 on holdout set and 0.005 on the full test set with ranking of 69.