Immune-Based Prediction of COVID-19 Severity and Chronicity Decoded Using Machine Learning

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• CCR5 and its ligands are implicated in COVID-19 pathogenesis.
• Targeting CCR5 is a potential therapeutic strategy for COVID-19.
• Understanding the immunologic spectrum of COVID-19 is crucial for effective treatment.

Purpose of the Study:

• To explore the role of CCR5 and its ligands across the COVID-19 immunologic spectrum.
• To use bioinformatics to model COVID-19 immunologic phases for treatment strategy development.
• To identify immune cell subsets and cytokine profiles associated with different COVID-19 disease severities.

Main Methods:

• Investigated 224 individuals (healthy controls, mild-moderate, severe COVID-19, and PASC patients).
• Assessed plasma and peripheral blood mononuclear cells (PBMCs) using immune subset profiling and a 14-plex cytokine panel.
• Employed bioinformatics, including SMOTE for data balancing and random forest classifiers, to develop predictive models and disease scores.

Main Results:

• Elevated B-cells and CCR5+ monocytic subsets observed in COVID-19 patients.
• Decreased CD4+, CD8+ T-cells expressing PD-1, and T-regulatory cells compared to controls.
• Significant alterations in cytokine levels (e.g., elevated IFN-γ, IL-2, IL-6, IL-10, VEGF; decreased GM-CSF, CCL4).
• Developed predictive models and disease scores (S1 for PASC, S2 for severe COVID-19) based on key cytokines and immune cells.

Conclusions:

• Severe COVID-19 is characterized by excessive inflammation and dysregulated T-cell responses.
• PASC is associated with profiles that activate effector T-cells but lack proper recruitment signals.
• The identified immune signatures and predictive scores offer insights for targeted COVID-19 therapies and patient monitoring.