Random Forests Are Able to Identify Differences in Clotting Dynamics from Kinetic Models of Thrombin Generation

Area of Science:

• Biochemistry
• Computational Biology
• Medical Diagnostics

Background:

• Current methods for differentiating acute coronary syndromes (ACS) from stable coronary artery disease (CAD) rely on limited analysis of thrombin formation kinetics.
• Existing approaches often use simple concentration quantifiers (e.g., peak concentration, area under the curve) for well-established species like thrombin.

Purpose of the Study:

• To develop an improved classification method for distinguishing ACS from stable CAD.
• To identify novel biomarkers and assay strategies for enhanced diagnostic accuracy.

Main Methods:

• Utilized a comprehensive 34-protein factor clotting cascade model to simulate thrombin formation.
• Transformed simulation data into a high-dimensional feature set (approx. 19,000 features) using a piecewise cubic polynomial fit.
• Applied the Random Forests statistical learning technique to identify key differentiating features.

Main Results:

• Identified specific combinations of features that effectively distinguish ACS from stable CAD.
• Found that concentrations of active alpha-thrombin, tissue factor-factor VIIa-factor Xa ternary complex, and intrinsic tenase complex with factor X, within specific time windows, are highly indicative of ACS.
• Achieved an accuracy of approximately 87.2% in classifying ACS using this combination of factors.

Conclusions:

• A combination of three specific coagulation factors at particular time points can accurately classify acute coronary syndromes.
• This finding suggests a more efficient and accurate approach to assaying the coagulation system for diagnosing ACS.
• The developed method offers a significant improvement over current diagnostic techniques for differentiating heart attacks from stable coronary artery disease.