Inflammatory biomarkers in infective endocarditis: machine learning to predict mortality

Area of Science:

• Cardiology
• Immunology
• Biomarker Discovery

Background:

• Infective endocarditis (IE) has high mortality rates.
• Improved patient management requires novel biomarkers for risk stratification.
• Current prognostic tools may not fully capture IE patient risk.

Purpose of the Study:

• To investigate if cytokines, chemokines, and growth factors at diagnosis predict mortality in IE patients.
• To identify novel biomarkers that enhance risk prediction beyond C-reactive protein (CRP).
• To develop a predictive model for IE patient outcomes using machine learning.

Main Methods:

• Analysis of 27 cytokines, chemokines, and growth factors using Luminex assay in 69 IE patients.
• Application of machine learning techniques to predict mortality.
• Development of a decision tree incorporating biomarker levels and CRP for risk stratification.

Main Results:

• In-hospital mortality was 26%.
• Interleukin-15 (IL-15) and C-C motif chemokine ligand (CCL4) significantly predicted death.
• A decision tree model achieved 91% accuracy in outcome prediction.
• High-risk group (elevated CRP, IL-15, CCL4) had 88% mortality; low-risk had 8% mortality.

Conclusions:

• Cytokines IL-15 and CCL4 are valuable predictors of mortality in IE.
• These biomarkers offer prognostic value beyond CRP levels.
• Assessing cytokines holds potential for clinical risk stratification and monitoring of IE patients.