Day 2: Thursday March 23, 2017

Moderators: Donna Lochner (FDA)

Andrew Rossi (NIMH)

Speakers:

8:20 - 8:40 am: Adam Himes

Incorporating Virtual Patients into Clinical Studies

Advances in numerical modeling have made it possible to directly simulate virtual patient outcomes.  Incorporating virtual patients into a clinical study has the potential to improve the clinical decision process while exposing fewer patients to clinical trials.  However, there is a need for a rigorous and credible statistical methodology to perform the analysis of a clinical trial that includes virtual patients.  This talk will cover distinguishing features of virtual patient models compared to traditional simulations, detail the statitstical methodology developed by an FDA/industry Medical Device Innovation Consortium (MDIC) working group, and review a mock submission that demonstrated the use of virtual patients in a hypothetical clinical study design.

8:40 - 9:00 am: Leonardo Angelone

Use of computational modeling for assessment of electromagnetic fields and medical devices

Our group supports the FDA regulatory and guidance role by advancing our knowledge on the complex interactions of electromagnetic (EM) fields and the human body. The research combines anatomically precise computational models and experimental measurements applied to several areas of clinical significance, including the: 1) analysis of radiofrequency (RF)-induced heating in patients with passive implanted medical devices who undergo magnetic resonance imaging (MRI); 2) analysis of the safety and effectiveness of MR Conditional active implants (e.g., deep brain stimulators and pacemakers) during MRI; 3) RF safety of human subjects during interventional MRI; and 4) analysis of patient safety with respect to gradient-induced heating and unintended nerve stimulation undergoing MRI.  These projects are conducted with active collaborations between several researchers, within the FDA and worldwide, at leading academic research institutes and industry organizations.  There is a direct impact to the regulatory mission of the Agency, as the tools developed by our research are extensively used by industry in pre-market evaluation for the safety and effectiveness of medical devices. 

Webpage:  Electromagnetic Modeling at FDA

9:00 - 9:20 am: Eugene Civillico

MSM in SPARC: Plans and Stretch Goals

9:20 - 9:40 am: Panel discussion with presenters
 

The objectives of this session are to:

• Describe and illustrate examples of computational methods that are used in medical device develoment or evaluation
• Describe data and other tools that improve understanding of physiology and pathophysiology relevant to medical devices
• Discuss collaborative approaches to translate models, often developed in academia, to the clinic in order to improve clinical outcomes for patients needing medical devices

The specific charge to you as a speaker are to address in your talk or in the panel discussion is:

• Discuss your work and its relevance to medical devices
• What are the challenges and opportunities for modeling for medical devices? What are shortcomings of current approaches?
• For a given application, are there models that do a particularly good job?
• How can industry, academia and government work collaboratively to improve the data and validation strategizes for modeling for medical devices?

Comments/Questions (please identify yourself):

Ching-Long Lin (Iowa)

Adam Himes presentation slide 2, “Yet we still run clinical studies the same as 20 years ago”.

This is not a true statement. For example, the multi-center trial studies, such as Severe Asthma Research Program (SARP) (which started in 2001) and SPIROMICS (COPD) sponsored by NIH NHLBI, have been advancing precision medicine rapidly. These studies have been collecting rich sets of diverse data both cross-sectionally and longitudinally, including genetic, biological, imaging, and clinical data over large populations.  Different machine learning techniques have been applied to identify molecular clusters, clinical clusters, sputum clusters, and imaging clusters, and different simple clustering has been proposed for classification. There are a lot of experience that we can be learned from these studies.