Machine learning insights into thrombo-ischemic risks and bleeding events through platelet lysophospholipids and...

Area of Science:

• Cardiovascular Medicine
• Lipidomics
• Machine Learning

Background:

• Coronary artery disease (CAD) poses a significant burden, with traditional risk factors insufficient for comprehensive cardiovascular (CV) risk assessment.
• Platelets play a role in atheroprogression, and their lipid profiles may indicate disease severity in CAD patients.

Purpose of the Study:

• To investigate the association between platelet lipid species and cardiovascular risk in CAD patients.
• To evaluate the utility of machine learning models integrating platelet lipidomics and clinical data for improved CV risk stratification.

Main Methods:

• Lipidomics data were obtained via mass spectrometry from a cohort of 595 CAD patients.
• Machine learning models were trained using CV risk measurements and clinical parameters to phenotype risk groups.
• Platelet lipids were analyzed in conjunction with conventional risk factors to assess diagnostic accuracy for adverse CV events.

Main Results:

• Specific platelet lipids were identified as independent predictors of increased CV or bleeding risk and adverse events.
• Integrating platelet lipid data with conventional risk factors significantly enhanced the diagnostic accuracy for adverse CV events.
• Machine learning models combining platelet lipidome and clinical data showed increased diagnostic value for risk discrimination in CAD.

Conclusions:

• Aberrant platelet lipid signatures and functions are linked to an elevated risk of adverse CV events in CAD patients.
• Machine learning approaches integrating platelet lipidomics offer improved early risk discrimination and classification for CV events in CAD.
• Platelet lipid analysis represents a promising biomarker for refining cardiovascular risk assessment in coronary artery disease.