Establishing Credibility in Medical Device Simulations: A Risk-Informed Approach to Developing Regulatory-Grade Computational Evidence
Paulina Rodriguez, George Washington University
Computational modeling and simulation have the potential to inform applications ranging from low to high risk, but there is insufficient trust in the medical device space to support regulatory decision making due to device complexity and uncertainties surrounding biological systems. To address this concern, we developed an end-to-end regulatory-grade computational model of an electronic drug delivery system using computational fluid dynamics and heat transfer with the commercial numerical solver and the ASME V&V 40 Standard, which has been recognized by the U.S. Food and Drug Administration. We also employed an agile iterative approach for computational model development and credibility activities such as verification, validation, and uncertainty quantification. The verification study included a mesh convergence analysis and an error estimation. The relative error for the device geometry was 0.01%, while the error based on a simplified geometry with an analytic solution for the Hagen-Poiseuille flow was 0.0001%. For the validation study, the simulation generated 10 samples using Latin Hypercube Sampling to propagate input uncertainties, and the physical experiments provided three replicates. The comparison between the simulation and the experiments showed that agreement is sufficient to validate the model for use at a low to moderate risk level. The credibility analysis showed that the computational model was sufficiently accurate to be used as a source of evidence for a low to moderate model risk. These efforts demonstrate a risk-informed approach to developing a medical device computational model and performing credibility building activities, which can serve as a collection of evidence to inform regulatory applications.
Abstract Author(s): Paulina Rodriguez, Seyed Ahmad Reza Dibaji, Bruce Murray, Matthew Myers