CFD-Based Hemolysis Study of Fontan Cavopulmonary Assist Device Using Kriging Surrogate Modeling

Area of Science:

• Biomedical Engineering
• Medical Devices
• Fluid Dynamics

Background:

• Numerical prediction of hemolysis using power-law models often shows high variability.
• Idealized coefficients from simplified devices lead to inaccurate hemolysis index predictions.
• A computational fluid dynamics (CFD)-based Kriging surrogate modeling approach was developed to generate device-specific hemolysis coefficients for Fontan cavopulmonary assist devices (CPADs).

Purpose of the Study:

• To apply a CFD-based Kriging surrogate modeling approach to a CPAD.
• To generate device-specific hemolysis power-law coefficients for CPADs.
• To predict the hemolysis index of a CPAD using CFD simulations and device-specific coefficients.

Main Methods:

• Hemolysis index of a CPAD was measured through mock loop tests and simulated using CFD.
• Kriging surrogate modeling was employed for Lagrangian and Eulerian formulations of the stress-based hemolysis power-law model.
• CPAD-specific power-law coefficients from one design were used to predict the Modified Index of Hemolysis (MIH) for an alternate design.

Main Results:

• CFD predictions for MIH using CPAD-specific coefficients showed improved accuracy compared to experimental results for an alternate design (16%-20% deviation for Eulerian, 7%-15% for Lagrangian).
• This approach significantly outperforms idealized empirical coefficients, which result in MIH prediction variations up to two orders of magnitude.
• The study demonstrates good correlation between CFD predictions and experimental tests for CPAD design modifications.

Conclusions:

• The power-law approach using CFD and device-specific coefficients effectively predicts MIH for CPAD design modifications.
• The obtained hemolysis power-law coefficients are potentially valuable for predicting hemolysis in similar rotary blood pumps.
• This method enhances the accuracy of numerical hemolysis prediction in medical devices.