Metabolic signature of COVID-19 progression: potential prognostic markers for severity and outcome

Project Description

The study addresses the challenge of predicting which COVID-19 patients may develop severe symptoms or face a fatal outcome.
Researchers used untargeted ¹H NMR-based metabolomics to analyze 240 serum samples from a Danish cohort, including 106 COVID-19 patients with varying disease severity and patients who died from other causes.
The goal was to investigate how metabolomic changes reflect disease progression, severity, and outcome in COVID-19.

The analysis revealed distinct metabolic patterns in serum that can differentiate between mild vs. severe COVID-19 cases and survivors vs. non-survivors.
Using ROC analysis and four machine learning models (Random Forest, SVM, PLS-DA, Logistic Regression), the team identified two biomarker sets predictive of disease severity and outcome.
These biomarkers include inflammatory markers, amino acids, fluid balance indicators, ketone bodies, glycolysis-related metabolites, lipoproteins, and fatty acids.

Developed a predictive model for classifying COVID-19 disease severity and outcome based on serum metabolomics.
Identified prognostic biomarkers with relevant biological roles that may support early clinical decisions.
Demonstrated the potential of routine metabolic profiling to guide risk stratification and reduce mortality in future outbreaks.

Explore the full research here