The Credibility of Computational Models Program in the FDA's Center for Devices and Radiological Health (CDRH) conducts regulatory science research to help ensure the credibility of computational models used in medical device development and regulatory submissions. This is one of 20 research programs in CDRH's Office of Science and Engineering Laboratories (OSEL).
On November 16, 2023, the FDA issued the final guidance, Assessing the Credibility of Computational Modeling and Simulation in Medical Device Submissions. The guidance provides a framework that manufacturers can use to show that computational modeling and simulation (CM&S) models to support regulatory submissions are credible. The guidance applies to CM&S models that are physics-based or mechanistic. It does not apply to standalone machine learning or artificial intelligence-based models.
The guidance is intended to help improve the consistency and transparency of the review of CM&S, to increase confidence in the use of this evidence in regulatory submissions, and to facilitate improved interpretation of this evidence submitted in regulatory submissions reviewed by FDA staff.
Computational Models and Medical Devices
For many years, medical device manufactures have used computational simulations of their medical devices as supporting evidence for safety or effectiveness of the device in regulatory submissions to the FDA. Now, medical device manufacturers are also embedding computational models in software platforms and using them as clinical decision support tools.
Computational Modeling and Simulation (CM&S) has been identified as a priority by the FDA. The FDA promotes the use of in silico clinical trials using Computational Modeling and Simulation (CM&S), in which a device is tested on a cohort of virtual patients, which is anticipated to replace or supplement clinical trials. However, insufficient or inconsistent information on the credibility of CM&S currently hampers their use to support regulatory submissions.
As a regulatory organization, the FDA is responsible for ensuring that CM&S used to support regulatory submissions is reliable. The credibility of a computational models is defined as the trust, based on all available evidence, in the predictive capability of the model.
Regulatory Science Gaps and Challenges
Major regulatory science gaps and challenges that drive the Credibility of Computational Models Program are:
- Unknown or low credibility of existing models: Many established or proposed models have never been rigorously evaluated and therefore have unknown credibility or suffer from known deficiencies which limit model credibility.
- Insufficient data: Quality experimental or clinical data for model development and validation, especially human physiological data under in vivo conditions and data on physiological variability, is scarce.
- Insufficient analytic methods: These methods include test problems in code verification, appropriate validation metrics, as well methods to evaluate acceptability of individual members of a virtual cohort.
- Lack of established CM&S best practices: These include good simulation practices and full end-to-end examples of the entire credibility assessment process across the wide range of applications process.
- Lack of credibility assessment tools: These tools include those associated with performing code verification, calculation verification, identifiability analysis, and so forth, as well as lack of decision-making frameworks for overall credibility assessment.
Credibility of Computational Models Program Activities
The Credibility of Computational Models Program focuses on regulatory science research in these areas:
- New credibility assessment frameworks and application of existing frameworks (for example, evaluation of methods for patient-specific modeling and in silico clinical trials for cardiovascular applications) relevant to the wide range of modeling disciplines, clinical domains, and regulatory submissions in which the FDA receives modeling for review.
- Domain-specific research related to credibility of computational models (for example, hierarchical validation of interlaboratory simulations of compression-bending testing of spinal rods) relevant to current and expected regulatory submissions so that the FDA has complete understanding of model capability when they are used in regulatory submissions.
For more information, email OSEL_credibilityofmodels@fda.hhs.gov.