A Data Driven Approach to Predicting Hemodynamic Factors of Coronary Stenosis Severity

Abstract: In this study, a data driven approach to predicting hemodynamics based diagnostic factors for ischemic severity of stenotic lesion of coronary by a machine learning technique was proposed. For a training dataset, we generated total 1,116 coronary vessel models with various geometric features of a stenosis and conducted 3D-0D coupled blood flow dynamics simulations. We employed a fully connected deep neural network model with four hidden layers and a sigmoidal activation function. This novel approach produced a promising outcome for near-real time assessment of coronary lesion severity with reasonable accuracy.