One Giant Leap for Precision Medicine: One million research cohort to help millions of Americans who suffer from disease

By Eric Dishman, Intel Fellow and General Manager for Health and Life Sciences for Intel

On January 20th, President Obama launched the Precision Medicine Initiative (PMI) to accelerate our understanding of individual variability and its effect on disease onset, progression, prevention, and treatment.  One of the key components in bringing this initiative to life was the creation of a large research cohort that could help researchers discover and quantify factors that contribute to illness, and then test approaches that can preserve health and treat disease.

To create the cohort, the National Institutes of Health (NIH) created a team of experts in precision medicine and large clinical research studies to articulate a vision for building a national participant group that would advance research and – be essentially in a word – Big.  Big in its number of participants – one million or more Americans.  Big in its diversity of the cohort.  Big in the size and new types of data that would be collected.  Big in its impact for America.  Precision healthcare is what many are referring to as the next global space race, and this is a big stake in the ground on America’s competitiveness in healthcare.

As a technology executive in the private sector leading Intel’s Health and Life Sciences Group, I was honored to be invited to participate in this working group.  I wanted to give it a try.  And I was blown away.

I learned more about advancing precision medicine in the last 5 months working with these brilliant committee members, the experts at the public workshops, and the listening sessions across the country than I did in the last five years combined.  The leadership shown all the way from NIH Director Dr. Francis S. Collins, challenging everyone to think and act differently, was bold and inspiring.  Witnessing so many great minds across the working group and from different parts of HHS including Rob Califf from the Food and Drug Administration (FDA) and Dr. Karen DeSalvo and Jon White from the Office of the National Coordinator for Health Information Technology (ONC) inspired me to work nights and weekends to make our recommendations as impactful for the American people as possible.  Today September 17, during a meeting of the Advisory Committee to the NIH Director, Dr. Francis Collins, the PMI Working Group report was made public.  Here are 5 reasons why I’m so very proud to be associated with the proposals that we collectively made in this first report.

  1. True Partnership between Participant and Researcher: As a cancer survivor, I know what it’s like to submit fluids or samples and have no idea what is being done with it.  I also know the frustration of information flowing from patient to researcher and not being able to shape the ideas of things I’d like them to study.  If participants can’t access and understand data about themselves, or understand conclusions made in aggregate about them and others like them, then they can’t own their own health.  And if they can’t own their own health—in partnership with their team of clinicians, friends, and family —then behavior change and preventive care paradigms that are critical to sustainable healthcare will not happen.  This was one of the things I shared with the US Senate Committee on Health Education, Labor and Pension yesterday.   In the PMI cohort, patient engagement and true partnership between participants and researchers will be the norm and not the exception.  In a Foundation of National Institutes of Health (FNIH) survey, 71% of Americans agreed that researchers and participants should be equal partners.  Simply put, the expectations are high for proactive participant engagement to manage their own health and the health of their family members.  This will be an interactive participation model.
  2. The data and insights go back to the Participant: In the same survey described above, 90% of Americans surveyed said they would be motivated to participate in this if they thought they’d get health information that could help them or their family members.  For any participant that wants it, we recommended giving back both individual data about themselves and aggregated data about the scientific findings that emerge from their data over time.  And that would include counseling, like genetic counseling to help participants make sense of their individual genomic data so they can appropriately use/interpret that knowledge.
  3. Participant pool will be large and diverse: One million or more Americans, completely healthy and very ill, that span across all ages, all socio economic status, all diseases and disease states.  Every American is eligible.  Reaching people who are of all education and economic backgrounds, including those who are incarcerated or decisionally impaired who have not been served by many studies in the past is hugely challenging.  But if we don’t then we’re not reflecting the realities of the country.  Many minority populations are simply under-studied, and therefore we recommended over-recruiting them.  If we want to assemble rich data to really personalize treatments based on all aspects of the American population, whether that’s race/ethnicity, location/environment, or access to care – we need diverse participation.  Imagine what researchers will be able to discover from a public resource that is this large and diverse.   Large scale research that for any given individual research institution could never have been possible otherwise.
  4. The data types will be diverse: Electronic Health Record (EHR) data is foundational to this study, collected via the participants themselves or their HPOs.  But true leadership in precision medicine will require capturing data and driving analytical insights from all data types.  Clinical/claims data, diagnostic/lab/medical device data, along with two new emerging data sets:  genomic data and consumer generated data.  Today, individual researchers can’t create sample sizes big enough to validate whether their genomic analytics tool is truly valid or even really useful.  An application for consumer health is meaningless if the sensor embedded into a smart phone isn’t picking up data that will be clinically useful.  With one million participants having their diverse data collected, we can start to help move the clinical and scientific validation of the new data types forward. I was particularly heartened and excited to see a consensus recommendation to move in this direction.  We also felt the need for patients to contextualize their data was important.  In all of our Intel studies for people with congestive heart failure or other diseases that use our remote patient monitoring technologies, we include a log/diary that lets participants add things about the data they captured for researchers.  For example “Oh, I just ran up the stairs right before I did this test” to help explain a sudden change in heart rate helps care teams know what the data they just captured meant.
  5. The Analytics Model will be leading edge: This means employing a hybrid model. Collecting participant data in a central, secure repository whenever possible is one of our goals.  But realistically scaling to one million or more participants and adding massively large datasets like imaging and genomics, while streaming mHealth data will require new computational models and hybrid architectures of central and distributed data.  And that’s why we’re proposing a hybrid approach.  Whole genome sequencing of just the one million participant’s normal cells (not cancerous tumor cells) would result in 300 petabytes of data alone.  That’s 300 + 15 more zeros.  One petabyte of average MP3-encoded songs would require 2000 years to play.  Just moving 1 petabyte of patient data from a state on the west coast to the east coast would take about 21 days when you include the time it takes to copy data to disk, pack the disks and ship.  That’s not going to work when researchers coming from multiple sites want to run distributed analytics on large amounts of genomic data for a given project.  As I talked about recently in a different blog, we’ll clearly need architectures that can support collaborative analytics of large data sets securely, across different sites.   

The report and these five proposals are bold on paper, but now we need to maintain its boldness in practice.  This will require the NIH and its research grantees to embrace new methods, new technologies, new measures, and most challengingly – new relationships between what it means to be a researcher and research participants.  One of the reasons I love this PMI cohort project so much is that it will be the proving ground to scale precision medicine to all the other 319M Americans.  The bold proposals and actions we’ll take will force us to learn how to deploy cutting edge technologies and address challenging policies that will be required in clinical practice of the future.  Not unlike the space program stimulating innovation in other industries, we’re going to solve problems here that will help advance the entire healthcare system.

Comments are closed.