Artificial intelligence for predicting mortality in hospitalized COVID-19 patients

Abstract

Background: The global demographic situation has been significantly impacted by the COVID-19 pandemic. The objective of this study was to develop a model that predicts the risk of COVID-associated mortality using clinical and laboratory data collected within 72 h of hospital admission. Materials and methods: = 189) of patients had a fatal outcome. Five machine learning models and the Boruta-SHAP feature selection method were utilized to assess the risk of mortality during COVID-19 hospitalization. Results: All methods demonstrated high efficacy, with ROC AUC (Receiver Operating Characteristic Area Under the Curve) values exceeding 80%. The selected Boruta-SHAP features, when incorporated into the random forest model, achieved an ROC AUC of 93.1% in the validation. Conclusion: Throughout the study, close collaboration with healthcare professionals ensured that the developed tool met their practical needs. The success of our model validates the potential of machine learning techniques as decision support systems in clinical practice.

Ähnliche Arbeiten