Multidisciplinary Optimization Design of pVADs Using Analytical Target Cascading-Guided Genetic Algorithm

The purpose of this article is to address the limitations of inconsistency between the impeller and motor when designing Percutaneous ventricular assist devices for different clinical scenarios, such as cardiogenic shock and high-risk percutaneous coronary intervention surgery. This inefficiency results in the waste of computational and experimental resources. This work presents a novel optimization method for systematically designing the pVAD.

The system-level optimization framework combines artificial neural networks (ANN), analytical target cascading (ATC), and NSGA-II. ATC ensures coordinated parameter matching between the motor and blade, NSGA-II refines specific design parameters, and ANN reduces computational overhead. The approach streamlines key design parameters.

The integrated optimization method successfully achieves balanced performance between cardiac output and motor power demand. The prototype achieves a pressure head of > 80 mmHg at 5 L/min with a hemolysis index < 0.02. The total efficiency of the pVAD reaches 33.65%, compared to the baseline design's 7.59%; the algorithm significantly improves the design.

The proposed framework resolves the trade-off between blade performance and motor power in pVAD design, enabling more efficient and feasible device optimization. The results also revealed that the proper design of blades is vital. Proper inlet and outlet angles will yield optimal hydraulic performance, while a shorter blade chord may reduce the risk of hemolysis. The two aspects are not independent, and design points should be chosen comprehensively.