Putting Regulatory Science to Practice for Medical Devices

ACTP01What is the opportunity?

Demands for medical device evidence development addressing patient safety, therapeutic efficacy, and cost-benefit determinations are increasing. These evidence demands are well-documented drivers of medical device clinical trial size, complexity, timelines and costs. At the same time, and seemingly paradoxically, patient and provider demands for faster treatment access, and innovative treatment options are increasing. Given the legitimacy of these stakeholders’ data demands, combined with the realities of resource (budget and time) limitations, historical data acquisition techniques will soon no longer meet patient and provider needs.

While numerous solutions have been hypothesized in the literature, adoption of clinical trial solutions which both make clinical trials more efficient, and provide the data necessary to show safety and efficacy, has been scarce.

Integration of VP data within the clinical trial simplification framework has the power to revolutionize how device companies conduct clinical trials, sustain this paradigm shift, and satisfy patient and provider demands for safety, efficacy and improved access.


The VP working group is made up of MDIC members including FDA, nonprofits, software and device manufactures who volunteered their engineers, statisticians, regulatory professionals, and medical doctors to help addresses barriers by combining simulations and experiments to inform clinical trials.

Leveraging FDA guidance for the use of Bayesian statistics in medical device clinical trials, the working group created a framework to augment clinical trial design with virtual patients.

This group has conducted retrospective clinical trial case studies leveraging the VP Model, developed and launched several methods, tools, and resources to facilitate VP Model implementation, and has developed and released the VP Model’s code in R, a free to the public repository for statistical code.  In addition, ecosystem stakeholders have begun highlighting the VP Model independently of MDIC activities.  This growth in activity and communication related to the VP Model is reflective of MDIC’s commitment, and that of our medical device ecosystem partners, to maximize clinical trial efficiency and improving patient access to cutting-edge medical technologies.

The working group is conducting a mock submission, demonstrating how to implement this framework in an example trial design and IDE submission.

The framework creates the potential for smaller, shorter, and more cost-efficient clinical trials. Incorporation of prior knowledge in clinical trial design, most commonly through the use of Bayesian statistics, has benefits of decreased sample size and trial length while maintaining the same study endpoints. A common source of prior information is historical data from similar trials, such as outside-the-US device use, or field performance of similar therapies. When validated models that can predict safety and effectiveness outcomes are available, these models can be a superior source of prior information. Use of a model as this prior knowledge can increase the power of a clinical trial and reduce its size, duration, and cost.

There are two important features of this framework:

  1. First, the computational model must predict safety and/or effectiveness outcomes as well as uncertainty in the prediction. We call this model a virtual patient, and the model is most likely a model of the device and anatomy impacted by the device.
  2. Second, the number of virtual patients used to enhance the trial is based on the agreement between the model and real patient results. Better agreement, more virtual patients.

Additional things to consider.

  • Predictive models require patient data to model anatomy, physiology, and disease progression. Access to data is a barrier.
  • This framework can enrich clinical trials for small populations like pediatrics, how do we encourage and support the creation of validated models for these under represented populations?
  • How do we engage patients to understand their support for this type of clinical trial?
  1. Haddad T, Himes A, Thompson L. Incorporation of stochastic engineering models as prior information in Bayesian medical device trials. Journal of Biopharmaceutical Statistics 2017; http://dx.doi.org/10.1080/10543406.2017.1300907
  2. October 17th, 2016: Innovation in Clinical Trial Design to Reduce Trial Size.  AdvaMed 2016 The MedTech Conference.
  3. February 20th, 2017: Leveraging Existing Information for Future Studies:  The Case for Bayesian Methods.  Cardiovascular Research Technologies (CRT) 2017.
  4. Faris O, Shuren J. An FDA Viewpoint on Unique Considerations for Medical-Device Clinical Trials. The New England Journal of Medicine; 376:1350-1357; April 6, 2017; DOI: 10.1056/NEJMra1512592
  5. April 11th, 2017: AdvaMed Innovation Summit, Innovation in Clinical Evidence Generation, Synthesis and Appraisal to Advance Regulatory Science for the Total Product Life Cycle
  6. April 17th, 2017: MDICx – Leveraging Existing Information for Future Studies: The Case for Bayesian Methods
  7. April 27th, 2017: Bayesian and Adaptive Designs. 10th Annual FDA/AdvaMed Medical Devices & Diagnostics Statistical Issues Conference.
  8. May 18th, 2017: Augmenting a Clinical Study with Virtual Patient Models: FDA and Industry Collaboration on a Mock Submission Sponsored by MDIC.  BMES/FDA Frontiers in Medical Device Conference.
  9. May 18th, 2017: MDICx – Virtual Patient R Code Access and Use
  10. May 18th, 2017: MDICx – Virtual Patient R Code Release
  11. August 1st, 2017: Joint Statistical Meeting (JSM2017), Improving the Efficiency of Medical Device Clinical Trials by Combining Simulations and Experiments — Invited Papers Section on Medical Devices and Diagnostics
  12. Bayesian Loss Function – Operating Characteristics (R shiny app)