Short-term predictor for COVID-19 severity from a longitudinal multi-omics study for practical application in...

Area of Science:

• Biomarker discovery
• Intensive care medicine
• Pandemic preparedness

Background:

• COVID-19 pandemic strained ICU resources globally.
• Existing predictors focus on long-term outcomes, not short-term ICU planning.
• Need for tools to predict acute disease progression for better resource allocation.

Purpose of the Study:

• To establish a workflow for identifying short-term predictors of COVID-19 progression in ICUs.
• To develop a multi-omics based predictor for clinical decision-making.
• To aid in managing resources and medication during acute phases of critical illness.

Main Methods:

• Recruited 32 SARS-CoV-2 patients admitted to ICU.
• Collected serial plasma samples for proteomic, metabolomic, and lipidomic analysis.
• Utilized statistical analysis and machine learning to identify progression predictors.

Main Results:

• Identified two proteins (CCL7, CA14) and one lipid (hexosylceramide) linked to disease worsening.
• Developed a predictor with 79-84% accuracy for forecasting deterioration 3-5 days ahead.
• Predictor performance complemented clinician assessments.

Conclusions:

• A workflow for omics-based biomarker discovery was established.
• A short-term predictor for COVID-19 symptom aggravation was successfully developed.
• Methods are adaptable for future pandemic preparedness and small cohort studies.