Objective: 
Coronary artery geometry heavily influences local hemodynamics, potentially leading to atherosclerosis. Consequently, the unique geometrical configuration of an individual by birth can be associated with future risk of atherosclerosis. Thus, this study aims to identify the order of influence of the geometrical features through systematic experiments, which can reveal the dominant geometrical feature for future risk assessment
Methods: 
According to Taguchi’s method of design of experiment (DoE), the left main stem (LMS) length (lLMS), curvature (kLMS), diameter (dLMS) and the bifurcation angle between left anterior descending (LAD) and left circumflex (LCx) artery (αLAD-LCx) of two reconstructed patient-specific left coronary arteries (LCA) were varied in three levels to create L9 orthogonal array. Computational fluid dynamic (CFD) simulations with physiological boundary conditions were performed on the resulting eighteen LCA models. 
Results:
The proximal LAD was identified to be the most atheroprone region of the left coronary artery due to large relative atheroprone area (RAA) of helicity intensity (h2), time averaged wall shear stress (TAWSS < 0.4 Pa), oscillatory shear index (OSI ~ 0.5), relative residence time (RRT > 4.17 Pa-1). In both patient specific cases, based on h2 and TAWSS, dlms is the dominant geometric parameter while based on OSI and RRT, αLAD-LCx is the dominant one influencing hemodynamic condition in proximal LAD (p < 0.05). Based on RRT, the rank of the geometrical factors is: αLAD-LCx > dLMS > lLMS > kLMS, indicating that αLAD-LCx is the most dominant geometrical factor influencing hemodynamics at proximal LAD which may lead to atherosclerosis.
Conclusion:
The proposed identification of the rank of geometrical features of LCA and the dominant feature may assist clinicians in predicting the possibility of atherosclerosis, of an individual, long before it will occur.