Accepted for/Published in: JMIR Public Health and Surveillance

Date Submitted:

Open Peer Review Period: -

Date Accepted:

Date Submitted to PubMed:

closed for review but you can still tweet
  • Chunshui C, Jiang T, Li J, Yonghui Y, Liang H, Jian L
  • Machine Learning-Based Mortality Risk Prediction Model for Sepsis: A Tool for Clinical Decision Support
  • JMIR Public Health and Surveillance
  • DOI: 10.2196/11848
  • PMID: 30303485
  • PMCID: 6352016
Preprints Accepted Manuscript

Machine Learning-Based Mortality Risk Prediction Model for Sepsis: A Tool for Clinical Decision Support

Abstract

The precise prediction of mortality risk is essential for managing patients with sepsis. This study aimed to create a machine learning (ML) system that could accurately predict the likelihood of in-hospital death in critically ill patients with sepsis. This single-center retrospective study included 352 patients with sepsis and septic shock who presented to a tertiary hospital between 2010 and 2015. The Boruta technique was used for feature selection. Several machine learning models, including logistic regression, XGBoost, LightGBM, Random Forest, Support Vector Machine, and Gaussian NB, have been developed to predict quality metrics. The performance of the model was compared with traditionally utilized scores. The Shapley additive explanation (SHAP) was used to evaluate the value of features. The in-hospital mortality rate of patients with sepsis was 48.3%. The model creation included the use of a set of 26 variables. The model with RF had the greatest predictive potential among the six models, with an area under curve (AUC) value of 0.785 (0.633–0.935). The development of a public online prediction platform has enhanced the clinical feasibility of risk-prediction models. The RF model created in this study is a precise indicator of mortality among patients with sepsis in hospitals.

As per the author’s request the PDF is not available.
Copyright

© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it’s website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.

Citation

Please cite as:

Chunshui C, Jiang T, Li J, Yonghui Y, Liang H, Jian L

Machine Learning-Based Mortality Risk Prediction Model for Sepsis: A Tool for Clinical Decision Support

JMIR Public Health and Surveillance

DOI: 10.2196/14886

URL: https://preprints.jmir.org/preprint/11848

PMID: 31789598

PMCID: 6352016

Machine Learning-Based Mortality Risk Prediction Model for Sepsis: A Tool for Clinical Decision Support

Current Preprint Settings

(as selected by the authors)

  1. Allow access to the preprint PDF upon submission to:

    (a) Open peer-review purposes

    (b) Logged-in users only

    (c) Anybody, anytime

    (d) Nobody

  2. When the manuscript is accepted, display the accepted manuscript PDF to:

    (a) Logged-in users only

    (b) Anybody, anytime

    (c) Anybody, anytime, and the author has chosen to deposit the ahead-of-print metadata to PubMed

    (d) Nobody

  3. When a final paper is published in a JMIR journal, display the preprint as follows:

    (a) Allow download

    (b) Show abstract only

    (c) Do not display anything

  4. If the paper is rejected from JMIR journals, display the preprint to:

    (a) Logged-in users only

    (b) Anybody, anytime

    (c) Nobody