A Learning-Based Model to Evaluate Hospitalization Priority in COVID-19 Pandemics

Area of Science:

• Medical research
• Infectious diseases
• Health systems management

Background:

• The COVID-19 pandemic has strained healthcare systems globally.
• Accurate assessment of disease severity is crucial for patient management and resource allocation.
• Distinguishing non-severe from severe COVID-19 cases upon admission is a significant clinical challenge.

Purpose of the Study:

• To develop and validate a precise model for assessing COVID-19 severity at admission.
• To identify patients at high risk of rapid deterioration requiring hospitalization.
• To provide a practical tool for healthcare administrators to manage hospitalization resources efficiently.

Main Methods:

• Development of a four-variable assessment model incorporating lymphocyte, lactate dehydrogenase, C-reactive protein, and neutrophil levels.
• Validation of the model using the XGBoost machine learning algorithm.
• Evaluation of the model's predictive performance in identifying severe COVID-19 cases.

Main Results:

• The four-variable model demonstrated high accuracy in predicting severe COVID-19 progression.
• Achieved a sensitivity of 84.6% and a specificity of 84.6% in identifying severe cases.
• Attained 100% accuracy in predicting disease progression toward rapid deterioration.

Conclusions:

• The established XGBoost-validated model is effective for precise COVID-19 severity assessment.
• This clinical measure-based formula aids in prioritizing hospital admissions during epidemics and pandemics.
• Efficient resource distribution is achievable through this predictive tool for critical care needs.