Machine learning models for predicting severe COVID-19 outcomes in hospitals

Philipp Wendland; Vanessa Schmitt; Jörg Zimmermann; Lukas Häger; Siri Göpel; Christof Schenkel-Häger; Maik Kschischo

doi:10.1016/j.imu.2023.101188

Machine learning models for predicting severe COVID-19 outcomes in hospitals

Inform Med Unlocked. 2023:37:101188. doi: 10.1016/j.imu.2023.101188. Epub 2023 Feb 1.

Authors

Philipp Wendland¹, Vanessa Schmitt¹, Jörg Zimmermann¹, Lukas Häger², Siri Göpel², Christof Schenkel-Häger³, Maik Kschischo¹

Affiliations

¹ University of Applied Sciences Koblenz, Department of Mathematics and Technology, Remagen, DE, Germany.
² University Clinic Tübingen, Department of Internal Medicine 1, Tübingen, DE, Germany.
³ University of Applied Sciences Koblenz, Department of Economics and Social Care, Remagen, DE, Germany.

Abstract

The aim of this observational retrospective study is to improve early risk stratification of hospitalized Covid-19 patients by predicting in-hospital mortality, transfer to intensive care unit (ICU) and mechanical ventilation from electronic health record data of the first 24 h after admission. Our machine learning model predicts in-hospital mortality (AUC = 0.918), transfer to ICU (AUC = 0.821) and the need for mechanical ventilation (AUC = 0.654) from a few laboratory data of the first 24 h after admission. Models based on dichotomous features indicating whether a laboratory value exceeds or falls below a threshold perform nearly as good as models based on numerical features. We devise completely data-driven and interpretable machine-learning models for the prediction of in-hospital mortality, transfer to ICU and mechanical ventilation for hospitalized Covid-19 patients within 24 h after admission. Numerical values of. CRP and blood sugar and dichotomous indicators for increased partial thromboplastin time (PTT) and glutamic oxaloacetic transaminase (GOT) are amongst the best predictors.

Keywords: Clinical decision support; Covid-19; Machine learning; Predictive modelling.