Machine learning based predictive modeling and risk factors for prolonged SARS-CoV-2 shedding

Area of Science:

• Infectious Diseases
• Epidemiology
• Machine Learning in Healthcare

Background:

• The COVID-19 pandemic highlights challenges in managing prolonged severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) shedding.
• Identifying risk factors for delayed SARS-CoV-2 clearance is crucial for effective prevention and treatment strategies.

Purpose of the Study:

• To develop a predictive model for prolonged SARS-CoV-2 viral shedding.
• To identify significant risk factors associated with extended SARS-CoV-2 infection duration in hospitalized patients.

Main Methods:

• Retrospective analysis of 56,878 hospitalized patients.
• Utilized the extreme gradient boosting (XGBoost) machine learning algorithm to build a predictive model.
• Employed Shapley Additive Explanations (SHAP) for detailed risk factor analysis.

Main Results:

• Six features significantly impacted prolonged SARS-CoV-2 shedding: vaccination, hypertension, admission cycle threshold (Ct) value, gender, age, and family accompaniment.
• The XGBoost model effectively predicted prolonged viral shedding based on these identified risk factors.
• Analysis confirmed the association of these six factors with extended SARS-CoV-2 infection duration.

Conclusions:

• A predictive model for prolonged SARS-CoV-2 shedding was successfully developed.
• Six key risk factors were identified, aiding in screening individuals at risk for long-term infection.
• Findings offer valuable insights for preventive control, resource allocation, and personal protection strategies against SARS-CoV-2.