The Future of Medicine, From a Leader in Venture Capital

Eric J. Topol, MD: Hello. I am Eric Topol, editor-in-chief of Medscape. I have a thoroughly delightful chance today to interview Vinod Khosla, who many years ago started Khosla Ventures, one of the most successful venture capital firms in the world.

Vinod, I think you started out in engineering in India at one of the most prestigious institutes of technology, then you went to Carnegie Mellon, and then Stanford. But engineering wasn't where you landed long-term, because at some point you started Sun Microsystems, right?

Vinod Khosla, MS, MBA: Yes. When I started out, there wasn't a thing called computer science. That tells you how old I am.

In those days, you could pursue any new area if you created your own program because there were not a lot of programs. At the Indian Institute of Technology, we started the computer science program, and then I moved to biomedical engineering. I went on to get a master's degree in biomedical engineering at Carnegie Mellon. That was also a tiny program in the basement.

Dr Topol: When you started Sun Microsystems—and that turned out to be one of the four horsemen of the Internet— did you know where that was headed? That was extraordinary.

Mr Khosla: We were pretty excited about the idea of distributing computing. That was a fundamentally new idea. It is funny in retrospect; the networking piece is the one piece no one wanted, so we kept getting requests to delete it. It's similar to the data piece that no one wants in medicine now.

When you start something, often you know it is a fundamental idea that should be developed, but you do not actually know how it will play out. That's what's interesting about all of these new spaces. You do a lot of discovery. It's like playing video games and discovering new puzzles to solve along the way.

In healthcare, the equivalent would be to say that there is no billing code for this or that. What is fundamentally new that should be valuable?

Dr Topol: You are an avid reader. When we first got together, you had read my book, The Creative Destruction of Medicine.^[1] I believe that we realized that a lot of our ideas were in alignment because you were also seeing this big disruption. Since that time, you have been shaking it up and getting behind many new exciting technologies.

But I want to set the record straight, because your ideas about doctors have been misconstrued. They have these quotes [attributed to you]: "Khosla says that 80% of doctors are going to be replaced." You wrote a white paper^[2] on that. Let's get that clarified.

Mr Khosla: Fundamentally, doctors spend a lot of time doing things that others can do for them. We have technology similar to a doctor's assistant, which can do many of the things that doctors do today.

My view is that doctors should be involved in the most human elements of care. A very large percentage of what doctors do can be done with technology, which would free the doctor to do other things. No doctor spends enough time with the typical patient. The median patient visit is so short.

Dr Topol: We will get back to that. But first, what are you involved in that you find the most exciting?

Mr Khosla: Medicine is better than it ever has been. Medicine today is better than it was 10 years ago, or 20 years ago, or 40 years ago. It has done pretty damn well. In every stage of the practice of medicine, we create the next small improvements.

I believe that we are beginning an era where we will go from the practice of medicine to the science of medicine. That science of medicine will be much more quantitative. The patient won't get three different recommendations depending upon the physician; the patient will get the exact same recommendation and it will be probabilistically the best recommendation for them.

Dr Topol: It will come from a machine?

Mr Khosla: It will come from a machine, and it will be quite multidimensional. Part of the reason the practice of medicine needs to change is that it is based on things we were able to measure a very long time ago—200 years ago. You could take someone's pulse a long time ago. You could put a cuff around their arm and measure their blood pressure or measure their temperature. These simple ideas evolved because it was the only thing technology could do. On that base was built a series of iterations. We have become more sophisticated, and we can look for cell-free DNA in blood and look for cancer mutations. But by and large, these have been incremental improvements.

One other fundamental thing is that medicine almost always has been based on symptoms. I feel something, I report that, then you doctors go on a detective chase and you mostly come up with the right answers. Mostly. If you rethink medicine, how would you do it now? There are a few thousand metabolic pathways in the body. Every part of your body is doing some physiologic thing that could be measured. If you could measure all of the things all the time, how would you diagnose people?

Dr Topol: You are saying that by digitizing a human being, this may in fact be a quicker or more efficient way than dealing with subjective symptoms, correct? Is that what you're getting at?

Mr Khosla: That is absolutely true but I'm going further. As an example, diabetes is a continuum; your parameters get worse and worse. It is not as though at a single point in time it is diabetes, after a certain A1c measurement or whatever metric you want to use. Rather, it is a continuum of shifts.

If we could track that continuum before someone develops a chronic disease, we would detect these things early, whether heart disease or diabetes or other conditions. The engine temperature on your car is not "good" or "bad," but you know it's going up if it's going up. I think the human body needs to be monitored that way, so that we know what's going on.

Dr Topol: That is a reasonable assertion since we monitor our cars far better than our bodies. We have a long way to go just to catch up to that.

Mr Khosla: This pains me. There are, perhaps, a few hundred sensors in the typical car and none in the body. A single ad shown to you on Facebook has way more computing power applied to it than a $10,000 medical decision you have to make. It blows my mind. Even though medicine has gotten much better than it ever has been, I think a quantum leap is possible. We can go from reactive to proactive by monitoring all of the symptoms. We should have those check-engine lights.

Dr Topol: Sure, for all different systems of our bodies. That is absolutely true.

Mr Khosla: I see another, related opportunity. As I said, we started with all of the things we could monitor a few hundred years ago: blood pressure, temperature, heart rate. There are thousands of things we can monitor.

One of the fundamental underlying foundations of this new medicine will be starting to measure things humans do not yet know how to address. If there's a pattern of expression of 19 genes that says chemotherapy will or will not help in your cancer, that is not something a human can eyeball. But data science can solve that problem and say, your probability of benefiting from this is X. One of the things we have to admit is that we have to collect data that humans cannot use yet, whether it is 3000 proteomic markers or more spectral data in your ECG. All of these data that humans cannot understand can be reduced to actionable things and make that check-engine light go on very early.

Dr Topol: I am sure you are right about this. Eventually that's where we are going. But there are a lot of obstacles. The medical community is not so high on change. What do you think it's going to take to get this actualized?

Mr Khosla: I am a fan of Obamacare, but I think it screwed up a few things. I actually thought all of the uninsured would be perfect people to use all of this stuff, because they did not have access to medicine in the first place.

Movements like this start at the fringes. Sometimes this means someone who has no other access to healthcare. In India, probably half a billion people have never seen a doctor in their life. That is a pretty good place to start with primary care or an oncology specialist artificial intelligence (AI) system. At the other end are expensive efforts, such as those aimed at human longevity. That then starts to scale, becomes cheaper and more accessible. You need to find that anchor on the fringe. I believe that that is what will happen.

No large change comes from an organization that is at the center of the system, for two reasons: First, they tend to think linearly; and second, they really don't have an incentive to cause disruption because "disruption" sounds like a great word but it is painful for some people. If you have been disrupted, it is not fun.

Dr Topol: The medical community particularly does not like to be shaken up.

Mr Khosla: Let me give you an analogy. Walmart didn't change retail; Amazon did and it made it pretty damn uncomfortable for all other retailers. Boeing and Lockheed didn't change space; SpaceX did. General Motors and Volkswagen didn't change electric cars or self-driving cars; it was Waymo and Tesla. NBC and CBS didn't change media; it was Twitter, Facebook, and YouTube. I am hard-pressed to find one large change that came from an institutional source in that area. If you do find one, tell me.

Why should healthcare be different?

Dr Topol: I cannot believe that it would be.

Mr Khosla: That is why I believe that it will start at the fringes. It will be something that feels small and interesting but inconsequential to the majority of the healthcare field. It will be small but exponentially growing.

Dr Topol: Do you think it will come from tech titans? Is it going to be the insurers? Is it going to be the outcry of the public? Where will this really gel?

Mr Khosla: One of the things about large change is that it is very hard to predict. When we started talking about Sun Microsystems and distributing computing, it was obvious that distributing computing should happen, but it was not at all clear which form it should take.

I had a presentation on cellphones in 2004. I called my presentation "The Device That Used to Be a Phone." People looked at it and said, what do you mean? It is a phone. I said, what do you use a phone for? They said, for talking. I said, what if 95% of the time the device was not used for talking but for other things? They said, that can't be.

In fact, Nokia and Motorola went under because they insisted that phones should have keypads because you use a phone to key in a number.

Apple imagined it differently. They had never built a phone. They started by asking how they could use a phone to do a bunch of things that were not just talking. And I wrote a presentation then with the idea that the device would be used not for talking but for other things, mostly navigation.

At that time, you couldn't imagine those other uses—GPS, Twitter, news sources. I was fundamentally right, but every single example I came up with of how it might be used was wrong.

I was directionally right but really embarrassingly wrong on the specifics. I suspect that I will be the same with healthcare.

Dr Topol: This reminds me of a conference at the beginning of 2007 in San Diego, where there were smartphones; it was at Qualcomm. Someone got up and said that they were thinking about putting a camera on a smartphone. And everyone asked, Why would you want to do that? We already have point-and-shoot cameras.

When you have a fresh start, which is hard, of course, in the healthcare world, that's when innovation happens. That is what you're getting at.

Mr Khosla: Keep in mind that today most phones have two cameras. In fact, the latest iPhone has this sensor to detect your face. It's really a third camera, a laser-based thing that actually maps your face. It is mind-blowing to think about.

Dr Topol: It is mind-blowing, including the fact that it will be doing that to every person you may come across.

A lot of people are concerned that, as we get into these advanced technologies, we then will make the inequities worse. There are already very serious problems with gaps in care. Are you concerned about that?

Mr Khosla: I am very concerned about that. AI especially will be a driver of that. In 2014, I wrote a very long essay for Forbes, called, "The Next Technology Revolution Will Drive Abundance and Income Disparity."^[3] I believe that we are reaching a point where many of the things that require human decision-making—not human caring, but human decision-making—will be done better by machines.

Today, this is obvious in radiology. The sensitivity and specificity, the accuracy with which a machine can read an MRI or a CT scan or an x-ray, is way better than what a human can do. In fact, it's sort of criminal to have radiologists doing this, because in the process they are killing people. It is very important that we realize this.

Dr Topol: That hasn't been tapped yet. It has been written up in papers, but it isn't the practice.

Mr Khosla: Two weeks ago I gave a talk at the National Bureau of Economic Research meeting. I took the top 20 employment categories in the United States and asked, Will most of the jobs in most of these categories be replaced by AI? The answer was very clearly "yes." It is hard to predict timelines. It probably will not happen in 10 years; it will be much longer than that. But is it 20 or 40 or 50 years? It is difficult to speculate. Directionally, you are right.

I believe that the single largest problem we will face is income disparity and disparity of resources. There's good news to come with that. The good news is, if we have great abundance, people may not have a lot of income but they will have a personal AI physician to answer their questions. And that will be free. Just like Google Maps is free.

Dr Topol: Absolutely. I believe that the fact that you can make that available universally is going to be just extraordinary.

You have mentioned that, with deep learning, AI can read scans, assess skin lesions for cancer, check for diabetic retinopathy, and on and on and on. This obviously has legs. Then fast-forward: Where do we go?

From the outset, our conversation was that we could put the care back in healthcare, that the human factor could then be emphasized, since we are lacking that today. We don't have time, we don't have care, the humanity of the interaction is diminished. Do you really believe that, ultimately, when this is integrated into the future, that we will get back to the real relationship with the human-to-human bond?

Mr Khosla: I hope so. This is all speculation, as I try to repeat again and again. I do not predict anything; I speculate. It's very hard to say exactly what path this will follow.

Three years ago, I was having a conversation with Jeff Flier, who was then the dean of Harvard Medical School. I said, You are selecting students for IQ. It is so hard to get into Stanford Medical School or Harvard Medical School; you are selecting for IQ. I suggested that they change the admissions criteria to be the same as those of the USC Film School. Why? Because that is where they select for mirror neurons, empathy, all of the characteristics that, at least for now, lead human beings to respond to other human beings.

I hope that happens. It is clear that emotional elements are an area, for now, without a clear, obvious path for machines to do as well as humans can. Frankly, humans should be trained more in the human-to-human interactions.

Dr Topol: I agree with you. There hasn't been a significant change in medical education for 100 years, since The Flexner Report in 1910. The idea is quite novel: Instead of relying on MCAT scores, you rely on the person-to-person connection; the master diagnostician will likely be supplanted by algorithms.

Mr Khosla: Dr House will be an AI [app], and everybody will have a free version.

Dr Topol: Right. I would love to talk to you for hours. Thank you, Vinod. This has been a fascinating conversation. I want to thank everyone who has tuned in to our One-on-One conversation on Medscape. We have gotten some speculative ideas from a brilliant fellow, Vinod Khosla. I will look forward to many more interactions with you in the future. Thank you.