Citation: Kim, J.K.; Ahn, W.; Park, S.;
Lee, S.-H.; Kim, L. Early Prediction of
Sepsis Onset Using Neural
Architecture Search Based on Genetic
Algorithms. Int. J. Environ. Res. Public
Health 2022, 19, 2349. https://
doi.org/10.3390/ijerph19042349
Academic Editors: Keun Ho Ryu and
Nipon Theera-Umpon
Received: 13 January 2022
Accepted: 14 February 2022
Published: 18 February 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
International Journal of
Environmental Research
and Public Health
Article
Early Prediction of Sepsis Onset Using Neural Architecture
Search Based on Genetic Algorithms
Jae Kwan Kim
1,2
, Wonbin Ahn
3
, Sangin Park
1
, Soo-Hong Lee
2,
* and Laehyun Kim
1,4,
*
1
Center for Bionics, Korea Institute of Science and Technology, Seoul 02792, Korea; kimjk@kist.re.kr (J.K.K.);
sipark@kist.re.kr (S.P.)
2
School of Mechanical Engineering, Yonsei University, Seoul 03722, Korea
3
Applied AI Research Lab, LG AI Research, Seoul 07796, Korea; wonbin.ahn@lgresearch.ai
4
Department of HY-KIST Bio-Convergence, Hanyang University, Seoul 04763, Korea
* Correspondence: shlee@yonsei.ac.kr (S.-H.L.); laehyunk@kist.re.kr (L.K.)
Abstract:
Sepsis is a life-threatening condition with a high mortality rate. Early prediction and
treatment are the most effective strategies for increasing survival rates. This paper proposes a neural
architecture search (NAS) model to predict the onset of sepsis with a low computational cost and high
search performance by applying a genetic algorithm (GA). The proposed model shares the weights
of all possible connection nodes internally within the neural network. Externally, the search cost is
reduced through the weight-sharing effect between the genotypes of the GA. A predictive analysis
was performed using the Medical Information Mart for Intensive Care III (MIMIC-III), a medical
time-series dataset, with the primary objective of predicting sepsis onset 3 h before occurrence. In
addition, experiments were conducted under various prediction times (0–12 h) for comparison. The
proposed model exhibited an area under the receiver operating characteristic curve (AUROC) score
of 0.94 (95% CI: 0.92–0.96) for 3 h, which is 0.31–0.26 higher than the scores obtained using the
Sequential Organ Failure Assessment (SOFA), quick SOFA (qSOFA), and Simplified Acute Physiology
Score (SAPS) II scoring systems. Furthermore, the proposed model exhibited a 12% improvement
in the AUROC value over a simple model based on the long short-term memory neural network.
Additionally, it is not only optimally searchable for sepsis onset prediction, but also outperforms
conventional models that use similar predictive purposes and datasets. Notably, it is sufficiently
robust to shape changes in the input data and has less structural dependence.
Keywords: genetic algorithm; intensive care unit; neural architecture search; sepsis
1. Introduction
Sepsis is a condition in which inflammatory reactions occur all over the body in
response to infection. Severe sepsis can lead to sepsis shock or even death; it results
in tissue and organ damage, is a major cause of death worldwide, and has a very high
mortality rate [
1
]. Approximately 270,000 people die of sepsis annually in the United States.
There has been a steady global increase in the number of sepsis-related incidents, with
approximately 30 million cases worldwide and 6 million deaths [
2
]; however, no precise
treatment has been developed. Early prediction and active treatment through diagnosis
continue to be the most effective strategies for reducing mortality [
3
]. Moreover, if sepsis
can be prevented by predicting in advance, it is possible to reduce the consumption of
medical resources.
A method capable of objectively diagnosing sepsis is required to accurately predict
its onset. Sepsis is a condition that presents a systemic inflammatory response rather than
originating from a specific pathogen. Thus, the definition and diagnostic criteria for this
condition keep changing. The purpose of these changes has been to determine and treat
suspected sepsis more quickly. Sepsis-3, defined in 2016, is the latest definition and is
simpler than earlier ones [
1
]. Although Sepsis-3 has become more practical, it is associated
Int. J. Environ. Res. Public Health 2022, 19, 2349. https://doi.org/10.3390/ijerph19042349 https://www.mdpi.com/journal/ijerph
