Dr Daniel Kraft on the Art of the Possible

Well, I'm not 30 yet, so I'll take that under advisement — I keep saying yes to too many things. But I like the quote; I've never heard it before.

A lot of folks in medical training want to get there fast, and I think there's something to be said for slowing down and getting other life experiences, especially in your 20s, that will make you a better clinician and human. I was glad I did the six-year med school plan. That was a real hard decision for me — I had six months of activities to do, from the Australian flying doctors to NASA research to digital health in its early stages — and I'm glad I had that time now.

So follow your bliss and explore. Particularly if you take an extra year during medical training: many folks are very happy to have a senior medical student come and do a rotation and help them out in almost any field around the world. It's a great time to explore. You're kind of like a junior officer — you don't have a lot of responsibility, but you get to have the most fun. So I'd encourage people to take that gap year, as they sometimes call it, before or during med school, and not be afraid to stretch it out a bit. It's a pretty magical time to explore and ideate.

Cold calling never hurts. It's good to have some chutzpah. I just tried to do things that seemed interesting to me at the time. I was always interested in technology and different fields, and I liked things that overlapped. I had this space and flying passion — I learned to fly in college, ran the Brown University Flying Club, got involved in some NASA research, was almost an astronaut, all that good stuff. And that led to other misadventures. I went to the International Space University program between my first and second years of medical school, and that led to other adventures — and later Singularity University, co-founded by some of the same folks. So you can't always predict where things might be two, five, ten years later.

You might also meet someone early who crosses paths with you later. I grew up in the pre-Facebook and LinkedIn era — I wonder how things would be different now — but it's great to make friends in your youth, and to know many kinds of books and many kinds of people, because that leads to interesting collaborations and synergies later that you can't predict. A lot of folks in medicine get very siloed and focused: I'm going to be a neurosurgeon, and that's all they do. Sometimes it's useful to open your aperture — it'll make you a better neurosurgeon, or psychiatrist, or pharmacist, or whatever you end up doing — because you'll have other life experiences, particularly if you get out of your own silo and the place where you grew up.

It depends what lens you put on something. I like shiny objects and gadgets and gizmos. But if you look at something and go, "this is a toy today, it's not very good," and then ask, "how might this be useful in healthcare?" — that's one way to look at early wearables, or early elements of AI, or robotics, or 3D printing.

Take the lesson of not just Moore's Law — technology accelerating, getting faster, smaller, cheaper, better — but also Amara's Law. Amara's Law is roughly that we tend to overestimate what will happen in a couple of years and underestimate what will happen in a decade. The classic example: ten or twelve years ago, IBM Watson — "we're going to have AI doctors soon, and it'll be fabulous in two or three years" — even though I think IBM spent more on marketing than on the technology. Fast-forward a decade, and AI really is hitting the groove in radiology and pathology and dermatology and many other elements of healthcare, and it's only going to accelerate. Or early wearables — they weren't great. It was a big deal ten years ago to get heart rate on a watch; now that's passé, and soon we'll have blood pressure and probably non-invasive blood glucose. What could you do with that?

So if you're in the healthcare realm, some of the fun technologies start kind of kludgy. But you need to hold onto an understanding of how exponentials work, and of human mindsets and shifts — while also realizing that in healthcare, things happen more incrementally. We have more regulatory and reimbursement challenges, and often culture challenges, in terms of what gets adopted and what doesn't.

It's fun to be an early adopter. One example: when I was a medical resident at Mass General — actually even in medical school — I had the first pocket computer, the HP 200LX. I was doing an emergency medicine rotation at Harbor-UCLA, and there was one other geek there, a resident who had one: John Halamka, who's now a famous healthcare IT guru — he was at Harvard, and now heads up platforms at Mayo. We got to play with these things early, share our notes, find other birds of a feather. And I'm still collaborating with those folks 20-something years later.

The fourth industrial age, in a nutshell, is where the magic of AI and big data come together — Netflix knows what you want to watch before you've watched it, your digital banking is easy and Uber-ified. When we say healthcare is stuck in the third or second, it's because, at least in the US, a lot of care is still driven by fax machines. To get data from UCSF to Stanford, until a couple of years ago, you had to burn it onto a CD-ROM and drive it down the peninsula. We still have people filling out the same paper forms, or lots of available data that isn't actionable or used during the clinician's workflow.

So even though we're in this exciting digital age of medicine — and still in its early stages — there's a lot of room for improvement, and we can take lessons from other fields that have been changed and disrupted. It's often not the technology piece; it's the underlying architecture. In some places in the world they don't have that legacy architecture, and things move to the fourth industrial age faster.

Great question. Who do we choose for medical school? People who do well on their MCATs in the US, or organic chemistry and physics, or A-levels in the UK. They've learned the game of academic excellence, which is certainly important. But are those all the qualities you need in a good physician? We don't just need good bedside manner now — maybe good website manner. We may need different ways of interacting with patients in this metaverse — the medaverse, I'm going to trademark that one. And we need all types of folks, including people with computer science degrees. So yes, we still need to be agreeable and have good EQ as well as IQ, but who and how we select for medical school — let alone other healthcare fields — might need to shift a bit.

And what we put in the curriculum: should we have big data and AI alongside pathology and anatomy? I've helped Shafi Ahmed teach a course at Barts a couple of times, in person and virtually. There's no real digital health course yet for most medical students that I've seen. So we need to think not only about who we select and their traits, but how we train them for the next few decades.

Sure. You don't want to be Mark Zuckerberg and move fast and break things in healthcare, because lives are at stake — and we've seen some of the implications of that behavior at Facebook. By the way, about six years ago I spent a summer at Facebook helping them think about what they could do with healthcare, given that health is social. Their missteps with Cambridge Analytica and others squandered their ability to delve into that space, given the trust issues. So there's a need to be a little slower in healthcare.

But we can do a better job of clinical trials — digital and virtualized trials accelerated during the pandemic; look at how quickly we got vaccines to market. The internet of medical things is accelerating how we learn. And yet even when something's published as new knowledge, there's that old quote — I need to look up the reference — that it takes about 17 years from publication to standard of care. I think that one was about treating ulcers without surgery. That gap needs to be narrowed. We need to go from big exponential data — our wearables, our digitome, our genome, our exposome, our proteome — to actionable insights that you, as a clinician, can use in the ER or the specialty clinic or the ICU. And then narrow the gap between knowing those insights and what you can actually do at the bedside. So yes, we need to be a bit slower, but there's still tremendous opportunity to accelerate health and medicine across the care paradigm.

I'll start with something that's talked about a lot but still hasn't had a big impact: personal genomics. We're now down to a $200-$300 genome, as announced by companies like Illumina, and Ultima claims a $100 genome within a year or two. We've had genomics and personal genomics — 23andMe-type data — for more than a decade, but it's not leveraged into clinical care. Something as common as pharmacogenomics: picking the right statin so a patient doesn't get myopathy, or knowing whether you're giving someone aspirin when they're actually aspirin-resistant. We're not yet in this omics age, but we're going to see it explode when your complete genome can be done at the price of an X-ray or a complete blood count, and then layered with your proteome and metabolome — that idea of the digital twin.

And quantum computing is arguably on its way. When you apply that lens to that amount of data, some of the insights will be super surprising. There's already work underway on drug development with DeepMind and others. I always like to say the future of medicine isn't being driven by any one field — AI, robotics, 3D printing, nanotech, genomics, CRISPR, blockchain — it's the convergence of them. The surprising things happen when you mash them up. If I go back 20-plus years to when I was a Stanford medical student, we didn't have AI meets radiology, or digital therapeutics, or surgical robotics, or computational biology, or CRISPR-based editing, or even the human genome. So there are tremendous opportunities when you put things together.

And those are just tools to address the challenges in healthcare — the grand challenges around cancer (which is a misnomer, because there's no one cancer), mental health, social disparities, new ways of doing public health. There are so many big pain points, and lots of micro-challenges too — the ones you see in the clinic every day, or when you're visiting an older parent. You can use those as pain points to think about solutions, with the technology of today or what's coming next, or from solutions outside healthcare that can be applied. I just launched a seed-stage, digital-health-focused venture fund, with some really interesting companies. We just invested in one leveraging video games for mental health — not in a narrow form like Akili for ADHD, but massive gaming platforms built by people who've done huge gaming titles. You can use video games to detect neurologic and mental health issues, and as a way to sneak in some therapy too. So lots of opportunities for those who want to become problem-solvers and even entrepreneurs.

It's hard to say yes or no — it's a broad field, from quantified self to quantified health. I know you're wearing the Oura ring; I usually have a few more on, but my Apple Watch is charging. Very few patients' digital exhaust makes it to their clinician today. Most clinicians don't want the data from their patients, because they can't manage it or don't want to feel liable for it. And it's not about sending the data — which is maybe the new oil — it's about the relevant insights. My little wearable ring gives me insights about my sleep. I don't need to look at my REM and deep and light; it gives me an integrated score. It's maybe not perfect, but if I had a good night's sleep I can see the difference between a 95 and a 65 in how I feel. And of course you might have all the data about your poor sleep and still not change it.

So you need to meld the data and the insights — and how it's presented to the individual or the clinician — in a way that really matches them. What changes your behavior might be different from what changes mine, based on age, culture, education, how you frame it. What's your favorite color? Do you like badges, social cred, or dollars as an incentive? Some insurance companies give people a lower life-insurance premium if they can show they're walking 10,000 steps a day. That might work for some but not others. Hopefully, just as we talk about precision medicine and precision personalized care, we'll get to an era of precision digital health — where the data flow is presented to you in a way that's digestible, actionable, aligned with your own incentives, and maybe with the payment model and workflow of the clinician.

Take a common issue: hypertension. It's already built into some watches with a cuff, and we'll have cuffless patches and watches doing live continuous blood pressure — even camera-based versions on your laptop or phone that can give a reading. So what do you do with that for all the hypertensive patients? Or real-time blood sugar, as you can already do with a CGM patch and soon in your watch? Those can drive better short- and long-term decisions, giving us more consistent feedback loops. Many people know they're supposed to exercise more — that's the wellness side. But for sicker patients, the data feedback loops in improving artificial pancreases, now being programmed by patients and published in the New England Journal, can improve outcomes. Or when you send someone home after a total hip replacement or a COVID ICU stay: are they walking more or less, based on simple Fitbit data? That can help you find the patient and intervene before they have a fall or another complication. So it's a huge "it depends." We need to study these — there are journals from Nature and others doing digital health now, studying real clinical examples — but it's still very early days, and the connection between all these data-driving wearables and the clinician rarely exists yet.

I don't know if I'm a good storyteller — I'm certainly an accidental public speaker. The style I developed came from the fact that, especially in healthcare and technology, there's so much going on. I get excited and I can't give a slow talk covering three points in 30 minutes; I cover three points in three slides. That's my style, and some people resonate with it while others say I really need to slow down — and there is something to slowing down. My style is more slides with simple images and points, rather than lots of fine print and PhD-level diagrams. It depends who your audience is and what you're trying to communicate.

There's definitely a space for slow talks with three points over 30 minutes. And sometimes constraints help — the last TED talk I gave, they said it needs to be 12 minutes, not 18, and you can't use 100 slides, you can use 15. That made me slow down, and they gave me a coach to run through it, because they really want you to memorize it. I never memorize — I have it in my flow, and I look at my slides and that triggers what I want to say. So my advice is: develop your own style. There's no one way to do it. If you can think about compelling images, and share things in not-the-standard-academic-chart form, that can really spark people's imagination and retention.

That's a great phrase — I've heard it somewhere before — and it's somewhat true. Who you make friends with, and the unlikely bedfellows, can really enrich your life. Not always to get ahead or get a better grade, but to spark creativity, collaboration and adventure.

My world opened up when I was a fellow doing hematology-oncology and bone marrow transplant at Stanford. Someone I knew from Boston, where I'd done residency, told me about TED — "Daniel, it's perfect for you, you need to go." At the time it was super expensive — it's super expensive now, but it was expensive then too. I got a partial scholarship and showed up in Monterey. And I'm in the hallway talking to Paul Berg, the Nobel laureate who taught one of my medical school biochemistry classes — so I actually knew someone there. I'm talking to Paul Berg, and up come Larry and Sergey, who you may have heard of. I'm like, "oh, hey guys," and we're talking about biochemistry and genomics. Then Dean Kamen rolls in — this is my first hour at the TED conference — and Dean Kamen is one of my heroes; I hadn't met him until then. He says, "oh, we're doing lunch, come along." So I follow Dean Kamen to lunch and sit down at a table with Larry Page, Matt Groening, Vinod Khosla and other interesting folks — not realizing I'd crashed the brain-trust lunch for TED by mistake. This was TED 2006, before anyone had heard of TED Talks. I didn't know who I was in the mix with. And that didn't just open up a certain network, but a mindset — now I'm in a mix of folks from technology, entertainment, design and some healthcare. That really catalyzed my thinking about the art of the possible, and the power of having networks and collaborators and friends from all sorts of different fields. That's where the real fun and magic happens.

Well, I'm not saying I was somebody now, and I certainly wasn't then — I was just a postdoc at Stanford. But I wasn't shy, and that was an environment where everyone's friendly and interested. There are always networking events; some are overdone. But put yourself in environments and mixes of people you might not always run with — that's where you might grow and learn. There's that old phrase that it's only when you get out of your comfort zone that the real growth happens. Sometimes that means getting out of your normal circle of friends and colleagues. Especially in healthcare, I think that benefits the entire field.

I was lucky to be a space and science nerd and to get to do some research at NASA Ames. The guy I worked with is now an astronaut, and might be one of the ones going to the moon. I almost got to be an astronaut myself — I spent time at Johnson Space Center as a medical student. So I joined the Air National Guard as a flight surgeon — basically joined the military as a Guard officer — and got to be, not the pilot, but the doctor for the pilots flying F-16s and F-15s, and to see that operational world: how much it takes, mentally, physically and operationally, to fly fighter jets. We did one deployment over Iraq during the no-fly-zone mission. So I got an appreciation of systems at scale, and super-complex machinery combined with humans in the cockpit, with the target of bombs on target — not always the best end goal, but there are lessons.

One is the art of the possible. My grandmother was born in Salzburg, Austria in 1903 — the year the Wright brothers flew, I think the same month — and she lived to see us focused on the moon. Aviation is such a great example of how quickly things can emerge and evolve. Another lesson, more recent: I started flying again. I took a few years off with young kids, and during the pandemic I wasn't flying anywhere, so I went to the local flying club, found a flight instructor — a former F-18 pilot — and started flying again. It was hard to go back after the Air National Guard, flying the F-16 and F-15. I have a couple hundred hours of stick time in the back seat of fighter jets, so it's no harder to go to putt-putt planes — but the big difference is that now I'm flying a plane with a full glass cockpit, where I grew up flying with the round-dial analog instruments. That's the same analog-to-digital shift we've seen in some elements of care. There are good and bad things about it, but now you have a moving map with GPS; you know where the airplanes around you are. I just bought half an airplane with a touchscreen — I fly the whole plane, I just own half of it — and you can touch the airport, or the weather, or the traffic. All this new situational awareness — that's what we need to bring to healthcare.

There are all these new data flows, but the user interface is still broken. These Epic-fail EMR-type systems aren't a way to grok and get a radar or map of where you and your patient are. So one of the lessons is to use better mapping and integration and user interfaces from aviation — not just the checklist, but how you present information, and how you can change your screen to see completely different elements. That makes flying safer and more fuel-efficient. And it comes back to the lesson of crowdsourcing. I overuse the Google Maps and Waze example, but if we could see all the traffic around us and get to the hospital faster, that's great — and what if we could see all the patients like us around us, with similar journeys and trajectories, and build better health maps for each of us? Build the GPS for healthcare. Just like it changed driving and aviation, that would have huge advantages, because we're not trying to synthesize things and hold the map in our head — it's moving and updated constantly.

I've been involved for many years with XPRIZE. I helped come up with the Tricorder XPRIZE — synthesizing diagnostics into your hand — and, more recently, one around fast, frequent, cheap and easy COVID testing, since that was such a problem early in the pandemic. But the first XPRIZE was the Ansari XPRIZE: a $10 million prize to get a non-NASA, non-ESA rocket ship to 100 kilometers with two people and return, and do it twice in a week. It was audacious but achievable. It was finally won by Burt Rutan, I think in 2004. I had a chance to visit that hangar and meet Burt Rutan and the team a few months before they won, and it was like being in the Wright brothers' hangar — "wow, this is history." It opened up the art of the possible: it's not just the NASAs and ESAs and the Chinese that can get to space. And now, look at what SpaceX has done in just the last decade — they're basically running laps around some of what NASA is doing.

So I think you still need the blend — it can't be either/or. In healthcare, at the government level you still set the regulations and often the reimbursement, and that has to meld with technology if we're going to align incentives and move from sick care to healthcare, not just fee-for-service. But there's a role for larger governmental bodies. You work with the NHS, which has some great qualities: you can scale things quickly, you have maybe better data, and everyone has a base level of care. Then we have these private elements in the US and elsewhere that might work for the fortunate but don't advance health equity or access. So it's not either/or — we can learn lessons from both. And now that we're more nimble with technology — what SpaceX has up its sleeve, and others — it shows how quickly industries can move even in really challenging environments, like space. So watch that space. A lot of healthcare has been catalyzed by space technologies. When Buzz Aldrin and Neil Armstrong landed on the moon in 1969, they were live-streaming the EKG — and it took us until about 2018 to be able to do that here on Earth with patients. So sometimes in challenging environments we innovate faster, and those things translate back to the ground.

Plenty of bad habits. But habits that helped me — maybe not being too afraid to take the path less traveled. I did the six-year plan at Stanford Medical School; they made it easy and encouraged it, but most people did four or five. In that last year I thought, should I be in a hurry to get to residency? So, taking a bit of a chance and trusting your intuition about what might be a good match for you — not just what your mother, or your advisor, or your friends going into some other field tell you. Don't be afraid to take some risks outside the norm. That's where the growth and the opportunity lie — just like on the exponential curve, there's the linear line and there's the exponential, and the most interesting things happen on the exponential. If I've had any such habit, it's to look a little downstream at where the puck is going, and try to connect some of those dots — people, community, technologies, ideas. Some of those turn into patents and products and impact.

Maybe things always look like they worked out. But you've seen that classic image — you think it's a straight line to success, but really it's a wiggly line, everything squirreling around in between. There's no standard path, and you can only learn so much from other people's prior paths, because those happened in a different time, with different pressures.

For me it was a blend. I was doing the traditional physician-academic thing — I had an NIH grant, I was on faculty at Stanford, attending at UCSF — and then I got pulled into this interesting Singularity University world. It was just going to be summer programs, and then it kept expanding, and that felt like, "wow, now I've found my niche." Not Singularity University per se, but this idea of mixing with people from unlikely areas who knew things I didn't, who were trying to solve big, hairy, audacious challenges in fun and empowering ways, with community and energy. That was where I fit. I never quite felt like the right round peg in the round hole in academics — I was fine doing mouse research, published in Nature and Science, but that wasn't really getting me jazzed, and writing grants was not my forte. I like doing patient care — I'm still licensed — but I didn't see myself doing just that every day.

So for everyone it's going to be a bit different, and that might mean trying different elements and realms. Particularly as a medical student, you have the opportunity to try different rotations — though it's hard, because everything can be fun for a month. How do you get a bit more depth and figure it out? That's where it might make sense to take an extra six months or a year to explore before you make your final leap into a specialty, or change fields entirely. Many people now are also getting MBAs or law degrees and melding them with an interest in entrepreneurship. It's a really exciting time for anybody in healthcare — to be a great clinician, but also, if they have the energy and appetite, to think about shaping the future of medicine, whether as early adopters of these new technologies or by collaborating to solve for them with the whole ecosystem that's hopefully going to accelerate us into this new health age.