Individualized prediction nomograms for disease progression in mild COVID-19

Area of Science:

• Infectious Diseases
• Public Health
• Medical Informatics

Background:

• The COVID-19 pandemic necessitates effective management of mild cases to prevent progression.
• Identifying risk factors for severe outcomes in mild coronavirus disease 2019 (COVID-19) is vital for resource allocation and patient care.

Purpose of the Study:

• To develop and validate predictive models for disease progression in patients with mild COVID-19.
• To identify key risk factors associated with the need for oxygen support in mild coronavirus disease 2019 (COVID-19) cases.

Main Methods:

• Retrospective analysis of 344 mild COVID-19 patients.
• Development of two multivariate logistic regression models to predict oxygen support requirement.
• Nomogram visualization created using R software for model interpretation.

Main Results:

• Model 1 (excluding lab data) identified diabetes, coronary heart disease, fever (≥38.5°C), and sputum as risk factors.
• Model 2 (including blood routine tests) identified coronary heart disease, fever (≥38.5°C), and neutrophil-to-lymphocyte ratio as predictors.
• Model 2 demonstrated superior predictive performance (AUC 0.872 vs. 0.849) with high negative predictive values (>96%) for both models.

Conclusions:

• Two validated models incorporating clinical symptoms, comorbidities, and blood test results can predict oxygen support needs in mild COVID-19.
• These models serve as effective tools for ruling out disease progression in mild coronavirus disease 2019 (COVID-19) patients.
• Early identification of progressive cases allows for timely intervention and optimized management strategies.