The Long Run Is Now: How FDA is Advancing Digital Tools for Medical Product Development October 25, 2018

Speech | In Person

• Speech by: Lauren Silvis, JD ; Chief of Staff - Office of the Commissioner

Keynote by Lauren Silvis
FDA Chief of Staff
Health Industries CSO Summit 2018

(Remarks as prepared for delivery)

Good afternoon. I’m happy to be here today.

Our longstanding goal for medical care is to ensure that the right drug or device is delivered to the right patient at the right time.

This vision is increasingly achievable with the innovative products and technologies that are becoming available—including next generation sequencing platforms, 3D-printed medical devices, regenerative medicines, and gene and cell-based therapies.

These approaches are making health care interventions more precise and personalized to meet patient needs and preferences.

In some cases, we’re seeing transformative applications that can offer cures for chronic illnesses that previously required a lifetime of medical treatment.

But new technologies are also challenging the FDA to modernize our programs and regulations for product development and evaluation.

The FDA is meeting that challenge by developing regulatory frameworks that maintain our gold standard for safety and effectiveness, are risk-based, and tailored to the type of technology and disease state under consideration.

Digital platforms and tools are playing a significant role in the FDA’s effort to modernize for innovation. We’re harnessing these opportunities to incorporate more efficient approaches throughout the medical product life-cycle — from digital pre-market IND submissions, in silico modeling of digital patient avatars, and post-market surveillance using mobile health (mHealth) technologies.

These approaches are still in their early stages, but they are already helping facilitate a modernized ecosystem for innovation that can help us better meet our mission of protecting and promoting public health.

Consistent with the 21st Century Cures Act, which excludes certain categories of low risk software from FDA regulation as a device, FDA has created a risk-based approach to digital health. This means, for example, that FDA does not intend to focus its regulatory oversight on certain low risk software, such as software that automates simple health care tasks for providers, or helps consumers track and organize their medical information.

Risk-based regulation allows us to focus more resources and attention on developing our novel model for the pre-market review of software that functions as a medical device: The Pre-Cert program.

Since launching the Pre-Cert pilot with nine pilot participants last September, we’ve built on our initial strategy. We released the second draft of the Pre-Cert working model in June. Our digital health team continues to work with participants and other stakeholders to assemble the essential building blocks of this unique and innovative program.

The goal of the Pre-Cert program is to develop a tailored and pragmatic regulatory framework that considers the excellence of organizations, but also continually verifies the safety, effectiveness, and performance of their devices, tailoring the regulatory approach to the software development cycle while maintaining regulatory standards of safety and effectiveness.

We hope to leverage a sponsor’s culture of quality and consistency as part of its evidence of organizational excellence, supplemented by real world data, to ensure software’s ongoing performance on a firm by firm basis, rather than regulating individual software platforms, which are frequently updated.

Pre-Cert has benefited tremendously from robust stakeholder feedback and engagement over the last year. During the test phase of the pilot, scheduled for next year, FDA intends to use our existing authority to review applications, beginning with leveraging the De Novo pathway, where appropriate.

This will allow us to pilot the digital health concept going forward, and learn from that approach, including informing any future changes to law that might be required to fully implement the underlying concepts.

In the coming months, we are going to propose a framework to further harmonize our approach to digital health tools across the agency. We’ll take a similar listen-and-learn approach and make sure we get input from software developers, patients, and providers.

Specifically, we’ll seek early input from stakeholders as we consider how to regulate software applications that are disseminated by drug sponsors for use with their prescription drug products.

We haven’t directly addressed this before in a comprehensive way. We’ll be considering when it may be appropriate to include in the FDA-required labeling software that is intended to be used with a prescription drug—and when it may not be appropriate.

In most cases, we believe the output of patient-oriented apps can be shared with FDA under our rules for promotional labeling, rather than premarket review. We’ll clarify where we think the regulatory lines are in an appropriate risk based framework that accounts for the approaches we’ve already employed to regulate digital health tools.

It’s hard to overstate how quickly the mHealth field is changing.

In just the past 3 years, more than 150,000 mobile health apps have been launched for smart devices.

We recognize the enormous opportunities apps present for increased use of digital technology with drugs and biologics to support patient health and well-being, improve compliance, and help patients share accurate data with caregivers and providers.

We are seeing apps that remind patients to take and track a sponsor’s drug, with the opportunity to simultaneously notify caregivers or providers. Or track symptoms and potential adverse events.

Moreover, an app could be customized to individual patients and be programmed by a provider to give diabetic patients information on how to adjust specific insulin doses based on blood glucose levels through a sliding electronic scale.

We’ve even approved a product where a sensor embedded in a drug sends a signal to communicate with a device, allowing the app to automatically record when a tablet has been ingested.

These approaches could be particularly useful for medications indicated to treat addiction, where facilitating patient adherence to a treatment plan can be particularly challenging, and integration of digital technology into team-based supportive care can help improve outcomes.

Digital health technologies also present important opportunities to modernize clinical trials. Greater efficiency is needed, as clinical trials are becoming more costly and complex to administer.

Earlier this year, we issued guidance for sponsors, clinical investigators, contract research organizations, institutional review boards, and other interested parties on the use of electronic health record data in FDA-regulated clinical investigations.

We’re trying to help modernize and streamline clinical investigations through the use of EHR data and the inclusion of real world data. We’re encouraging sponsors and health care organizations to work with EHR and electronic data capture system vendors. We want to help advance the interoperability and integration of EHR and EDC systems.

We hope that expanding FDA’s access to and use of sources of patient data will further enable our evaluations of safety and effectiveness. Our goal is to continually build on our understanding of medical products throughout their lifecycle.

Every day, millions of EHRs are updated with data about medical interventions used in routine clinical care. Our guidance is intended to promote the use of EHR data in clinical investigations. And to help incorporate data collected in routine care settings into clinical trials. By harnessing the real-world data being captured in electronic health records, clinical investigators may be able to collect data from routine medical care that could generate scientific evidence that’s appropriate for regulatory decision making.

Our work to modernize clinical trials should not only improve our understanding of medical product benefits and risks, but could also help lower the costs of drug development and facilitate bringing new treatments to patients. New trial designs incorporating digital tools and real-world evidence have the potential to help make trials more efficient, more reflective of our diverse population, and more accessible to patients where they live and work.

For instance, FDA’s Center for Devices and Radiological Health (CDRH) is working with the Medical Device Innovation Consortium (MDIC) to advance the use of real world evidence (RWE). Robust discussion of how best to translate real world data into real world evidence that could support regulatory decision making has the potential to lead to earlier patient access to safe and effective technologies, while increasing the value and use of real world evidence to address the needs of all stakeholders.

To help support the integration of RWE throughout the total product life cycle for medical devices, we’ve also awarded a contract to MDIC to manage the National Evaluation System for health Technology coordinating center (NESTcc).

This is a key effort for the FDA.

As we’ve outlined in the FDA’s Medical Device Safety Action Plan, one important role for NEST is to serve as a robust patient safety net for med tech in the U.S. through active surveillance.

The NESTcc has established relationships with 11 data partners, 150 hospitals, and thousands of outpatient clinics. Taken together, they represent nearly 470 million patient records. NEST’s data partners will conduct testing to assess their system’s capabilities for addressing critical RWE questions across the total product life cycle.

Computer modeling and simulation is another approach with the potential to advance clinical trials and product development by developing in silico models that balance the desire for certainty in evaluating device performance with limiting associated delays in patient access.

To advance these goals, the MDIC Virtual Patient project has already demonstrated savings by limiting the size of a clinical trial that may be needed to demonstrate safety and effectiveness.

It uses computational modeling to create virtual patients to be used as Bayesian priors. A retrospective analysis showed a potential reduction in clinical study size of 15 to 50% across different device and diagnostic areas using in silico patient models.

To help the broader community gain familiarity with these models, the MDIC Virtual Patient team sent a mock submission to FDA to get official feedback on the proposed method. They posted their entire submission and FDA review memos online.

The model was also published in peer-reviewed literature to explain the statistical methodology behind the Virtual Patient, and the underlying software is publicly available.

As the frontier of medical innovation advances, the agency is also playing a growing role in curating standards for medical technologies can help advance innovation in areas that may lack consensus standards now.

One example is increased use of predictive algorithms for guiding and managing patient care. These software tools are becoming more sophisticated, enabling a broader set of opportunities to analyze large, complex data sets to identify patterns that wouldn’t be apparent to the unassisted human mind.

Artificial intelligence (AI), for example, holds enormous promise for the future of medicine. CDRH is actively developing a new regulatory framework to promote innovation in this space and encourage development of transparent benchmarks for the performance of AI-based technologies.

So, as we pilot our Pre-Cert program  — where we will focus on a firm’s underlying quality in assuring software products meet safety and effectiveness standards — we’ll consider how to account for one of the greatest benefits of machine learning, how algorithms like convolutional neural networks can continue to learn and improve as they take in more information.

We know that to support the widespread adoption of AI tools in the clinic, we need patients and providers to understand the connection between decision-making in traditional health care settings and the use of these advanced technologies.

One specific area that we’re exploring with stakeholders is how we can benchmark the performance of AI technologies in the field of radio-genomics, where AI algorithms can be taught to correlate imaging features on a PET or MRI scan with the genomic features of tumors, as well as matched pairs of treatment and outcomes data gleaned from EHRs.

AI based diagnostic imaging approaches provide an opportunity to improve patient prognosis, identify digital biomarkers of early response to treatment, or develop novel clinic-genomic phenotypes that could be used to triage patients for routine scans based on their likelihood of eventually developing an aggressive cancer.

Toward these goals, the FDA is exploring the use of a neutral third party to collect large annotated imaging data sets, for example highly annotated radiology scans used in a variety of clinical trials for specific disease indications.

This database could be used to provide an objective benchmark for the performance of a novel AI algorithm for a proposed indication, and help providers and payors compare AI systems with the best human standard of care.

The FDA is also one of many stakeholders deeply interested in advancing the assessment and quantification of symptom and functional outcomes in cancer patients through clinical outcome assessments (COAs).

COAs, in layman’s terms, are measures that describe or reflect how a patient feels, functions, or survives. Several technological advances hold promise to revolutionize how we can capture patient-centered clinical outcomes in controlled trial and real-world settings. One traditional COA is a survey that collects patient reported outcomes (PROs) through a questionnaire.

Electronic capture of PRO data (ePRO) is also becoming standard, providing a rich pipeline of structured clinical data. In addition to ePRO, mobile wearable technologies can complement traditional PRO surveys by generating objective, continuous activity and physiologic data.

Obtaining reliable wearable device data on activity level, coupled with direct patient report on their ability to carry out important day to day activities, can provide information on physical function — a digital biomarker that is directly relevant and important to the quality of life of patients with diseases like cancer, depression/anxiety, or multiple sclerosis.

Medical products are becoming increasingly sophisticated.

The advent of advanced computing and new digital health platforms will continue to help make health care more personalized, as connected technologies break down barriers between clinical research and real-world patient care.

Clinical trials will continue to incorporate more of these technologies, allowing researchers and companies to reduce the time it takes generate reliable insights from the clinic to better inform product development, and to improve our ability to translate scientific advances into more impactful diagnostics and treatments.

We look forward to working with you to seize these opportunities and deliver their full promise to improve and save lives.

Thank you.