Development and Validation of a Predictive Model for Severe COVID-19: A Case-Control Study in China

Area of Science:

• Infectious Diseases
• Medical Informatics
• Clinical Diagnostics

Background:

• Predicting severe COVID-19 is crucial for personalized treatment and efficient resource management.
• Early identification of high-risk patients can optimize medical interventions.
• Understanding progression factors aids in developing targeted therapies.

Purpose of the Study:

• To develop and validate a predictive model for COVID-19 progression to severe disease.
• To create a user-friendly, open-source tool for clinical risk assessment.
• To identify key clinical and laboratory markers associated with severe COVID-19 outcomes.

Main Methods:

• Prospective study of 206 COVID-19 patients.
• Utilized variation analysis, LASSO, and Boruta algorithms for model derivation.
• Validated model performance using ROC curves, AIC, calibration plots, DCA, and Hosmer-Lemeshow tests.

Main Results:

• Developed a predictive model incorporating alanine aminotransferase (ALT), interleukin-6 (IL-6), expectoration, fatigue, lymphocyte ratio (LYMR), aspartate transaminase (AST), and creatinine (CREA).
• Achieved high predictive performance with AUCs of 0.9104 (derivation) and 0.8792 (validation).
• Created an open-source nomogram and online calculator for clinical application.

Conclusions:

• An open-source, free predictive calculator for COVID-19 progression was developed.
• The model effectively predicts progression to severe COVID-19 using specific biomarkers and symptoms.
• The tool supports early, personalized management and optimized medical resource allocation.