FDA looks to modeling and simulation tools to streamline product reviews

The US Food and Drug Administration (FDA) recently published a report on how it uses modeling and simulation (M&S), highlighting the untapped opportunities to improve efficiency and predictability at the agency. Some of the key opportunities identified in the report are the need to develop guidelines for using such tools in the premarket setting and to harmonize how they are used among its product centers, international regulators and industry.

Modeling and simulation are computational tools that can, as the name implies, be used to model and simulate how products and patients will perform. While FDA and industry use in vitro and clinical testing to understand products it reviews, M&S can also play an important role in evaluating products.

“In the last decade, M&S has become firmly established as a regulatory science priority at FDA, which has coincided with the explosive growth in data science and model-based technologies,” the agency wrote.

In November, FDA published a report, Successes and Opportunities in Modeling & Simulation for FDA, which not only provides an overview of how M&S is used across the agency but serves as a gap analysis to identify how the agency can better use such tools in the future. The agency also surveyed its staff twice, in 2019 and again in 2021, to get a better understanding of how M&S is used in different regulatory areas.

One key opportunity FDA found is to accelerate the use of modeling during product development and premarket review. The agency said that modeling may have a “substantial public health impact” and lessons learned from developing reliable M&S approaches at those stages may also help develop solutions for other regulatory problems.

However, the agency said that a lack of “Good Simulation Practices” may account for why M&S hasn’t been adopted more broadly by the agency.

“Creating these guidelines is a key opportunity for the agency to have an important leadership role,” FDA wrote.

The modeling practices FDA explored in the report vary, and the agency said it does not have information on how specific M&S disciplines are not being used by individual centers. The agency added that no single center uses every discipline and there may be good reasons for that, but for now, that may point to missed opportunities that the centers may want to explore.

“Fully understanding these gaps is another important opportunity for the agency,” said FDA. “This could further support key regulatory science efforts, as indicated in the FDA Strategic Plan for
Regulatory Science.”

The report also identified two opportunities to use M&S for FDA scientists. The first is to strengthen internal networks for sharing resources and modeling techniques; hosting training sessions to enhance hands-on experience with the resources, techniques and relevant software platforms; and to promote better understanding and harmonization between FDA stakeholders.

The second is to develop a Good Simulation Practice guide to help harmonize the use of M&S techniques across the centers and between other international regulatory agencies.

“Establishing best practice and quality control principles to ensure more harmonized standards for model development, model use and validation, would strengthen our current modeling and simulation practices,” said FDA. “It is also critical to develop a common set of expectations or guidelines
for model verification, validation, credibility assessment and maintenance between industry and regulators, as well as between regulatory scientists/modelers and reviewers within the FDA.

“Further publication and/or usage of relevant guidance … will promote better alignment on best practices and expectations between stakeholders,” the agency added.

The report lists several other opportunities and takeaways; one of the most important being that M&S is increasingly being used by the medical product industry in process analysis and improvement. The agency said this means there is a “great potential” for the agency to use the same tools to enhance its submission process and workload prediction, which could improve its efficiency.

“For example, natural language processing and machine learning approaches have been used to predict FDA review time of devices submitted under the 510(k) pathway,” FDA noted.

Report

FDA looks to modeling and simulation tools to streamline product reviews

Related topics

Recommended content

Welcome to the new RAPS Digital Experience