Eric Dishman on Paul Allen, Andy Grove, and All of Us
March 23, 2021 | TRENDS FROM THE TRENCHES—It’s been a little over a year since Eric Dishman stepped down as CEO of the National Institutes of Health’s All of Us program and took on the role of Chief Innovation Officer. Before his time building the All of Us research program, Dishman served as an Intel fellow and vice president of the Health and Life Sciences Group at Intel Corporation.
But when he sat down to catch up with Stan Gloss, founding partner at the BioTeam, he credited much of his innovative thinking to even earlier experiences: his own 23-year journey as a cancer patient—finally cured thanks to precision medicine; his internship at Paul Allen’s think tank, Interval; and his early training in acting and improvisation.
“My only skill, actually,” Dishman told Gloss, is, “just playing universal translator when I come into forums with all these teams.”
Gloss and Dishman talked about the impacts of his early training and the roles both Paul Allen of Microsoft and Andy Grove of Intel still play in his decisions. Bio-IT World was invited to listen in.
Editor’s Note: Trends from the Trenches is a regular column from the BioTeam, offering a peek behind the curtain of some of their most interesting case studies and projects at the intersection of science and technology. The transcript below has been edited for length and clarity. For more from BioTeam, see the Trends from Trenches newsletter: https://bioteam.net/newsletter/
Stan Gloss: Eric, thanks so much for agreeing to hang out with me and talk a little bit about All of Us. It's been a while since we had a chance to talk. Before we get into the All of Us program, tell us a little bit about yourself and how you got here?
Eric Dishman: One of the first times I met Andy Grove from Intel, he said, "You're a social scientist! What the hell's a social scientist doing at Intel?" He said it in the kindest way, but I have always been a fish out of water, but in a fun and pretty exciting way. If cancer had not taken me at age 19, my desire in life would have been to be a rock star and a musician and an actor, which is what I was training to do. I was good at being in school, I kept getting scholarships to go to undergraduate to masters, masters to PhD. But the point at which you get that diagnosis, you realize that you need health insurance.
I'm a social scientist by training—a really obscure area of the communication field called interaction and conversation analysis. Once I became a patient at age 19, I saw how awful communication was amongst the health team. There was no team because people weren't talking to each other. So, I also studied organizational communication, and doctor-patient interaction. If it were not for an internship for Paul Allen's think tank, Interval, I probably would have been a professor in this obscure field.
I was working on a PhD at University of Texas at Austin, and I got a summer internship to get out of the heat of Texas. I didn't know who Paul Allen was, but I certainly knew Microsoft. I was supposed to stay for three or four months.
I was still acting some, and Interval was hiring artists and social scientists to study what computing was. It was mostly a piece of office equipment, but they wanted people like me to come in and do ethnographic studies of where there was no computing. So I studied a hair salon, and I studied a nursing home, neither of which had any computing or anything like that, to help them in those days. I ended up using theater and acting to embody future scenarios of what technology really needed to do to fit people's lives.
The first thing that we did—which really made my career in Silicon Valley, it's really funny—was something called Marcia's Hairworks. This was the name of this tiny salon on the coast that we had studied. We reimagined it where I played Marcia in some future with embedded technologies. We imagined tablets that you could talk to, and a mirror interface that she could interact with, where she didn't have to leave her mirror and her workstation cutting hair to go answer phones for the next appointments. We made these prototypes that were so compelling that the engineers that worked at Paul Allen's think tank thought we actually had gesture recognition and voice recognition and tablets with this motion all figured out, but it was all Wizard-of-Oz prototypes. Someone was clicking a mouse behind the curtain as I was supposedly cutting the hair of a designer friend of mine who helped me make all these prototypes.
Some people in life know me as a guy who brought theater and improvisation as a way to do great product design, combined with ethnographic studies. It's still a powerful method. That's part of what I brought to the National Institutes of Health. How do we bring user-centered methods to the design of a large cohort program like the All of Us research program?
So, yeah, would-be artist, social scientist, gets an internship in Silicon Valley, ends up staying seven years and running labs on RFID. My very last project was multiple-year project with Paul Allen working on tablets that didn't exist in the world at that point. We had real working tablets with a lot of the gestures that you see in common tablets today. The biggest problems back then were heat dissipation, getting computing and that small form factor dissipation, and getting glass and the screens cheap enough. So, it was just before its time, basically.
So you were with Paul Allen’s think tank Interval for almost eight years, then you moved to Intel to work with Andy Grove. None of us have mentors like that; you're so fortunate! How did Grove and Allen influence your work?
Both of them are very present with me right now. I can't believe we've lost both of them in the not-too-distant past. I'm working on a book about why innovation worked and why it failed in some places through all those years, and I've been going back to my journal and the notes, thinking about both Andy and Paul.
From Paul there were two things that I tried to incorporate when I started the All of Us research program at NIH, one of them was mixed discipline teams. I don't just mean an engineer and a marketing person. I mean, what can an artist see? Can you bring participants or users to the table to actually co-invent with you from multiple perspectives? The invention and the brilliance and the breakthrough that you can have with incredibly mixed discipline teams is powerful.
In the early days of RFID, we saw the potential for having an elegant world where the world of paper and the digital world could work seamlessly. I worked on this for almost two and a half years during Paul’s think tank. One of the big challenges in those days for RFID tags—it’s still a little bit of a challenge today—was if they were too close to one another, the antenna couldn't differentiate which tag was which.
I had a project that I called the 52-card deck pickup. I challenged our engineers to build an RFID reader—an antenna system and some tagging system—that I could embed in a paper card and not notice the tag in the card. Then, pick the 52 cards up, shuffle them together, put them in the deck and have the reader tell me what order the card is in, based on the tag that's in it. This was such a huge technical challenge. We were really struggling to get the antenna to be able to read these cards that were so close together.
It was a biologist who solved it. A biologist had studied eptesicus bats, and the way that sonar works in eptesicus bats. A bat has this radar, and this particular species of bat can only eat one or two of the kinds of bugs that are flying around it, and there might be 50 different species of bugs flying around it. So, how does it pick out that little bug from all of the others in a really crowded space? We ended up modeling our collision detection algorithms for RFID tags and the reader based on the models of eptesicus bats. And we did it.
The other thing that Paul had an insight about, and that we incorporated in the think tank, was that Paul never allowed an individual to work on only one project. He was convinced that if your job was one project—if that was your whole identity and your whole paycheck—you would not kill the project, or make it fail or stop it if it wasn't working. We had a point system, and everybody had to spend a certain number of points. That ended up meaning you are on at least three projects at any one time. It really did make people more critical of the project that they were on. They knew that the project was not their livelihood, so they would critique it. They would say, "You know what, this isn't working, let's let it fail. Let's stop and let's move on to something else." The psychology that he had figured out about having multiple people working on different projects at one time was really powerful. I've tried to incorporate some of that at times. It's easier to do in a think tank than it is coming to a government agency.
And what about Andy Grove of Intel?
So much wisdom from Andy. Andy and I met when we were both dealing with cancer, and the first thing that we learned together and helped teach each other how to do, was to own our own healthcare, and own our own healthcare data. It’s a fight that I've been on, trying to give people access to their own EHR data, working on policies, and Affordable Care Act and all those kinds of things for decades now. Andy and I were on a mission: to get people to wake up and realize that sometimes there wasn't science about your illness, and you had to manage your own path. You had a bunch of specialists, especially if you had complex healthcare problems like we did, but there was no one else to play orchestra conductor for all these different instruments that don't talk to one another. You had to make music out of your own healthcare system, which is a huge burden, right? You're the patient suffering!
So we started trying to build tools, and I sometimes joined and lectured in his class at Stanford that we called the unmedical school, because we were trying to get people to stop thinking in terms of just the specialties that they learned and just this body part, but to look back at the whole body as a system, and to treat the whole person. That's still a fight that that is worth fighting. Look at COVID right now; we're discovering that COVID is not just a respiratory illness. We're seeing it have all kinds of implications for the rest of the system.
Andy and I knew that we were suffering because there was a lack of data about our illnesses or lack of science. We knew that there was a lack of data about us, and it was not being shared amongst our teams. And we knew that the tools that were being used to measure health in an ongoing way, well, they weren't ongoing. You were going to get your blood pressure once a year and a doctor was making claims about what meds you should be on based on that one blood pressure shot in the dark. Andy and I were hell bent to reinvent the healthcare system together.
When I started working for Intel, I was building wearable sensors to try to monitor health from a distance, to look at changes in steps and changes in speech patterns as indicators of Alzheimer's or disease. When Andy Grove got Parkinson's, he asked, "Can you help detect subtle changes in my Parkinson's that may help my doctors adjust my medications?" We did that.
I cringed because Andy wasn't really supposed to be using an Alzheimer’s prototype with his doctor to titrate his meds day to day. This is the ultimate in personalized medicine. But he was, and it worked for him and a small number of other patients. Our goal was to build a system that was intelligent enough to work for you on the days that your disease is not having a bad impact—to know how to intervene at just the right moment when you need help, but not do the task for you because you want your mind to stay fertile. These were problems that were handy.
A lot of what Andy and I worked on together was how to teach other people to do strategic planning. Intel does strategic planning that looks 10, 15, 20 years out. They had to figure out: 20 years from now, what is computing going to need? How do we know how many chips and how powerful the chips are going to be and how many fabs [fabrication facilities] we're going to have to build, these multi-billion dollar holes in the ground.
It forced Intel to do two things: plan way in advance but also to build an ecosystem of fellow travelers. To this day, people get confused. Intel—they make computers, right? No, they make the chips and other technologies that go inside of the computers. So we had to make sure that an HP existed, a Dell existed. When we were working on multimedia, in the first codecs that allowed you to do CD ROMs, we had to give a lot of that software that was developed at Intel to Microsoft to bake into the operating system. We didn't control the user experience or the operating system.
Now, I'm teaching NIH people how to do strategic planning in the Andy Grove way—really teaching an evidence-based strategic planning that is action-oriented. Andy was all about “Strategy is Action”. What he meant was, the strategy is going to be how your people spend their time. It's not some abstract vapid mission statement. Strategic planning is a specific plan, and why we believe that is the right path out of all of them. It's amazing. I've taught a lot of organizations—small, not for profit, large ones—how to do strategic planning. People don't realize that a lot of the way that good organizations do strategic planning owes a lot to Intel and to Andy Grove specifically.
The other thing about Andy—I can just hear it right now, it just evokes him for me in a pleasant but also bittersweet way—Andy was always about focus, focus, focus. You did your strategic plan to set your direction, and then you ruthlessly focused on those plans, unless you had new evidence that that was not the right strategic direction to go on. Almost anytime I called Andy for advice about anything, he'd be, "Eric, you're trying to tackle too many things at once." "You're an excited, intellectually curious person, but you've got to focus." I've been trying to teach a lot of other people that focus matters. Learn to say no to other ideas that you have, to all these other things that will come in and try to distract you from the mission. Andy was ruthless about it in a really positive way.
So while you were at Intel, you were invited to join President Obama’s Personalized Medicine Initiative, and a report came out of that effort. Then about five years ago you left Intel and moved to NIH to run the All of Us program. How did that transition go?
They reached out, and strongly encouraged me to apply for the position and I thought, "I'm not going to leave Silicon Valley and move to Bethesda, Maryland." But let's just say the administration and Dr. Collins were very persistent. They kept saying, "You need to do this for your country; you're kind of uniquely situated. You're one of the first people on the planet whose life was saved; you understand it. As a former English major and actor, turned social scientist, you will be able to translate between the physician and biology folks and the technical folks and everyday users." I think, really, that's my only skill actually: just playing universal translator when I come into forums with all these teams.
How is the All of Us program different compared to other initiatives that have been done previously or currently being done?
There are other million person cohort programs underway around the world right now. We partner with them. The UK Biobank has been a fantastic helper to us. Genomics England with this large-scale project has been helpful to us. There are so many cohort studies. The folks from the Framingham Heart Study are some of the folks who originally did it and are working on it to this day, but Framingham is very small, actually, in the great scheme of things. The scale of a million is still new, and most people are struggling to achieve it. The VA's Million Veterans Program existed before us. They're partners with us and they've been wonderful to help teach us so that we don't make the mistakes that they made.
But certainly that's the other thing that Andy Grove and working at Intel taught me: how to do things at scale, right.
There are three or four things that are very differentiating about the All of Us research program. The most important word around almost all of those differences is diversity. From the beginning, and it's in the name, the All of Us research program was designed to fill the gap in science in the US and, thus, in medical breakthroughs of the so many kinds of people have been left behind by the science.
Most science on the planet has been done on white males. Then we try to extrapolate from them and hope that what we did for white males works for women, or that works for people who come from a Southeast Asian descent, or works for African Americans. We have many cases now where the whitewashed science has led to whitewashed medicine that has, in some cases, killed people and certainly harmed people. This was known long before I came to NIH. Some people think I drove the diversity agenda, but no. This was inherent in the original report.
Though, I did set that more aggressive goals than everybody imagined. I made up the initial goals and said, "Let's have 75% of the one million people, be diverse." I was at a keynote, right when I started the job, when I put that stake in the ground and some people gasped. I thought to myself, I'm in trouble. I didn't know that the best success people had had was in the 8% to 10% range, and here I was saying 75%. This is another thing Andy Grove taught me, set aggressive goals, bring brilliant people together to try to figure out how you can actually achieve them, and then focus on it and incentivize them to achieve those goals.
We're at 81% diverse rate now of the 360,000 people who have joined. So we've done it quite well.
It’s the diversity of people, and it's also the diversity of the data. There are many studies that are just EHR data studies, and even though they may have the scale of a million or tens of millions, they don't have any other data about those participants. No survey data. They're not tracking them for a long period of time like we are. They don't even have permission to recontact those people to see if they want to be part of a clinical trial for African-American women living in the Southeast.
The other thing that we're really pushing for is for our resource to open. A lot of these large cohort programs are closed to a very select number of scientists and researchers. Just as we're trying to open it up to all participants in the United States, we're trying to open the resource up to all researchers. We're working on tools that will eventually make it even for citizen scientists, community colleges, and others. Don't just let the Harvards and the Yales and the Vanderbilts continue to do all the science! Build tools—complex big data tools in the cloud—that bring more brain power per problem. That's what I preach to the team all the time. We want to increase the brain power per problem.
Think about what the All of Us platform is. On day one, you can go to our public data browser and start seeing how many people we have with this condition. Oh, maybe I can just recruit through the All of Us research program. And, I can get really specific, I want women who grew up in the Southeast, but now live on the West Coast.
The power of a million lets you start to do things like that. You don't have to go through your own big data analytic tools and platform, we're handing you one with a lot of other people in the community that are ready to teach you and train you on how to use it. So, we're more open than most, far more diverse than most. And we collect EHR, plus genetics, plus lifestyle, plus wearable, plus surveys, over a long period of time.
It's back to what Andy and I were struggling with in the health care system: how do we understand health as a system and an interaction of lots of things, not just a single variable? That's what we're setting up with the All of Us research program.
Everybody's trying to do a lot of work you're doing but on a smaller scale. What lessons have you learned for these ‘fellow travelers’, and what advice do you have to putting it together on an organizational level?
We have a ways to go with AI and natural language processing and all of that to automate a lot of the curation that today people spend of hours on. First lesson would be, don't underestimate your curation budget. Even if you do all the right things—drive standards, use common data models—the idiosyncrasies of where that data come from means you do have to harmonize it and the investment in harmonization is not small.
But you probably know this from your own work, and I've certainly seen it in the almost 30 years I've been working in Silicon Valley: the people problems are the most challenging part of it. We have gone through a lot of work to build trust with our participants and trust with our consortium partners. It takes trust building to develop the security and data use agreements. Those kinds of things are far more time consuming than a lot of the technical things.
Have use cases that help these people understand what they're buying into, scenarios in plain English, that aren't too techie, so that people can understand the future of what they're trying to do. Building those bonds of trust and listen to concerns. Even a lot of the researchers and the physicians themselves want to share their data, but they're sitting in IT environments where people are fired if there's a data breach and the data breach is going to be in the news. You understand why those CIOs and CTOs of those hospitals and clinics are so sensitive about this. They should be!
Understanding the players that you're trying to bring together to do these collaborations, and then carefully building the legal and ethical and rules of behavior, will go a long way to speeding things up later.
There's this book out, Slow Down to Speed Up, right? Sorting through those issues and deeply understanding those issues, will save you an enormous amount of time later, because you've built a foundation of trust, so that when inevitable mistakes or problems occur, people don't interpret it as, "See, I'm pulling out. I'm not staying in the program." Instead, they're, "Oh, yeah, we knew those kinds of things were coming, we have a remediation plan for it, and we've all agreed on how to do that." We spent a lot of time doing that, and we tweak those agreements, and those kinds of things as we go through.
So, use standards, use common data models, invest in the curation and then the harmonization, because it's not cheap. But start with a platform of trust that's really user-centered. Legitimately incorporate their issues into whatever agreements that you're trying to do.
I know from that first summer internship that I got with Paul Allen the power of user-centered design. It's been my key to success for my whole 30 years of career. How did a social scientist, would-be artist and musician, become a Vice President and an Intel fellow? It's because by using user-centered design the ideas and the prototypes and eventually the products that we built, were of incredible high quality. Everybody's like, "God, where'd you come up with these ideas?" From the very people that we were designing for.
I'm biased, because that's my training in the field, but I think of it as my secret weapon. I think of it as my magic, right? Everybody asks, "How does Dishman do this?" Here’s the answer: A repeatable set of strategic planning, and user-centered design product development methods that just work if you do them.
I stepped down as CEO of the All of Us research program at my own request in January 2020. Those commutes back and forth between Oregon to see my wife on her farm and Washington, DC were becoming painful. Thank God, I stepped down right before COVID hit. I got lucky in that regard. I'm now Chief Innovation Officer, and I've realized that the first powerful thing I could do for the program was be its inaugural CEO and get it set up, right. I feel like I did that.
But the lasting impact I want to have—and I don't need credit for it, people will never remember 20 years from now, who did this—is to teach them a set of tried-and-true organizational innovation and strategic planning methods that are rooted in user-centered design. Then I know the program will be successful for the 100 to 150 years that we need to do it.
So, when I said to you at the beginning of this interview, that the words of Andy Grove and the lessons from Andy and Paul were in me, they're so in me, and so back to the surface right now, because I've just gone into teacher mode. I hope that I am doing a decent job at carrying their legacy and it's something that they would be proud of.