Dr. Gunjan Bhardwaj, author of Inside the Cockpit, writes about the modern system of drug discovery and development—how it is broken, and how it can potentially be fixed and made exponentially better with AI and blockchain.
Gunjan is the CEO of Innoplex and Cancer Coin. He and his team are working to bring lifesaving drugs to market faster and cheaper. His hope is that the innovative combination of blockchain technology and AI will change the incentives for all the players in the drug development ecosystem.
Gunjan is also the author or coauthor of 23 patent applications. He currently serves on multiple advisory boards, including the Center for Human and Machine Intelligence and the Forbes Technology Council. He also lives in Germany, and actually did this recording in a Swiss restaurant on the outskirts of Basel. It’s a great conversation. If you care about the future of healthcare, this is the episode for you.
Gunjan Bhardwaj: I had a non-eventful childhood–I went to the best schools, I got good grades, I went to the best university in India. I got a scholarship, came to Germany, and finished fast. As a senior, I got a chance to work in a consulting firm with a sponsor. So, everything went okay, fantastic cool parents, everything goes well. My sponsors were my only family in the strange country I came to, where everybody said I would need five times the time I would need in the US or UK to get to somewhere because Germany is an extremely conservative country.
My sponsor worked like crazy–typically our update calls would happen at two in the morning when he would be driving back from the office to his home and he would ask me, “Hey G, how was your day?” He called me one day and told me that he was diagnosed with cancer. That was the moment that kind of shook me for the first time in my life. Subsequent to that, I spent a number of days working out of the hospital where he was being treated.
I saw his family relatives–my only family at that point in time in that country–looking hopeless, sometimes helpless, and withquestions on their faces. Hetold me that the head oncologist told him that there was aspecific therapy. I said, “Why do you believe all of what he says? He’s not Albert Einstein. He just studied medicine. What are the alternatives to what he says? Where can you get access to those alternatives? Who are the real experts for this specific kind of disease?”
Of course, when I spoke to this oncologist, he said, “Well son, we are not complete idiots. We have atumor board in this country for the specifics to be discussed with other experts.”
I said, “Experts need to be qualified–presented. Do all of these experts have experience with this specific kind of tumor? Probably not.”
So, I went to the German Cancer Society, talked to my fantastic mentors in big Pharma, and I found that the questions that I had, which all of you would have in such a situation, were profound questions. Not only layman like myself, or the friends and family of someone who is facing grueling rounds of chemo would have those questions, but also medical practitioners, because healthcare is all about data.
Data which is exploding–some studies say, doubling every 73 days.
There are mountains of data, and there is no way to sift through all of this data to make sense of it.
Some of these questions I had in my mind, when this good friend of mine was facing this. Thankfully, the friend of mine is doing amazing, but this question remains. This question has confronted me a couple of times again after this experience. I’m sure this has confronted many of the audience.
Structure the Data
Charlie Hoehn: If you had to pick one, what would you say is the core idea in this book that you want listeners and readers to take away?
Gunjan Bhardwaj: To do any analytics or to answer any questions through machines, the data needs to be structured. Machines can work with structured data in a context. Unless we provide the context to a machine and structure the data that needs to be analyzed, machines cannot answer questions.
For AI to be successful, for creating better healthcare outcomes, to really fix the broken drug discovery and development system, we need to structure the published universe, we need to make it searchable in a specific context, and we need to mobilize data stuck in silos in the unpublished universe.
Unless we do this, we will remain in the Middle Ages in drug discovery.
Identify the Problem
Charlie Hoehn: I have no industry knowledge in this, but from an outsider’s perspective, it looks bad. How bad are things really?
Gunjan Bhardwaj: It’s bad, because the individuals that are in the system, be it in the big Pharma innovative drug discovery model or individuals in the science, they all mean well, they all want to do something good, that i swhy they probably choose those careers–at least the majority of them. But it’s bad because the system in itself, the design of the system, the incentives, they are broken in such a way that it makes it impossible to really transform drug discovery in a way that would get better healthcare outcomes, as is desired for patients awaiting these therapies.
If you look at some of the things happening, more than 65% or 60% of all drugs that are approved by USFD have been fast tracked. People said, “Europeans were too fast and USFD was too slow.” Now it is the other way around. Therapies that are not really conclusive in studies and evidence also get passed, because there is so much public pressure. The public says, “We need the drugs, and we need them to be very cost effective. Big Pharma is crooked because they just care about making money.”
This is not the case, because if you want to get a good drug out in the present system, you spend from 1.5 to 4 billion dollars, which gives you one product which will give you more than one billion dollars revenue per year. You need this, because if you want to be commercially viable, you also need to pay for your R&D investments.
But nobody sees that, of course. The patient waiting for treatment for a specific tumor in his brain, looks at therapy and of course would complain if the therapy cost upwards of $200,000 a year. People say, “Well, the science is doing a great job, but the Big Pharma and the industry is not doing a fantastic job.” However, they need to come out with asuccessful blockbuster because they need to have a very commercially viable model to fund their R&D investments.
Capital markets are cruel. If an AI entrepreneuror tech entrepreneur invested 200 million dollars in finding a new way of looking at a specific cancer therapy and he fails, he would be applauded, “Wow, great try, fantastic guy, visionary.” But if a big Pharma CEO invested 200 million dollars in a project and it failed miserably, the capital market would punish them.
Markets have a different perspective than traditional industries, and it frustrates the leadership of this industry. Look at science. Do you think scientists in the system really care about patients? No. They don’t give a damn. What they care about is publishing in highly ranked journals because that’s what secures their tenures.
Do scientists share their failures? All journals bask in the glory of successes and that’s what they publish–successful experiments. Nobody publishes failures. The failure of one scientist could be extremely valuable for another or extremely valuable for the industry.
When we talk about knowledge sharing, all peer reviewed literature requires 18 to 24 months, so there is a blind spot in science for part of that time. What people share afterwards are these published papers. Whereas the currency for AI, for machines to analyze, is theexperimental data. Even journals sometimes do not get access to this data.
If you look at all these aspects, this shows that it’s not just the industry, it’s also science which is not designed in a way to get the best therapies in the fastest possible manner to patients that are waiting. Why should a scientist give away all of his or her data? Is he incentivized to do that? What are the incentive structures–publish more in good journals and you get a tenure?
From a point of view, yes, regulatory agencies are doing a lot, but is it really optimal? Probably not. So, both the design and the incentives for the industry and science are not really aligned to create a drug discovery and development system that really looks at the patient, even though there is a lot of noise about it.
Multitudes of Improvement
Charlie Hoehn: Can you paint apicture of what you believe science and drug development could look like using AI and blockchain? How much better could things get? Could it be two times better, ten times better? A hundred times better?
Gunjan Bhardwaj: It would be multitudes of improvement, not as two, three, four, or ten times–hundreds of times better.
We talked about this when we talked about science. How do you set up an experiment? You look at the literature, and you look at the scientific gaps and based on those gaps, you design theexperiment, you collect data, and then you try to establish the science covering those gaps.
In doing so, you would have maybe a couple of hypotheses. Imagine if you were able to have access to all the evidence that is out there. If you are a machine, you could look at all the permutations and combinations of hypothesis that are possible and at the same time look at proprietary data from the unpublished universe that you mobilized securely via blockchain.
Scientists could broadcast their hypotheses on blockchain and test those hypotheses for plausibility. You would do away with a lot of innovation redundancy, which is out there in the system. It would speed up a lot of the failures and this false sense of serendipity that we have and that I talk about in my book in science. Eventually, we would have patients get access to therapies faster, which is also very cost effective.
The Difference in Patient Experience
Charlie Hoehn: For the patient, how different would their experience be if we have AI and blockchain working in tandem in the background helping scientists make these discoveries?
Gunjan Bhardwaj: We are all talking about precision medicine. If you look at reality, of course, many people are trying with their heart and mind in the right place, but there are simply things that are broken. If somebody is prescribed a blood thinning agent, they get prescribed Warfarin. You know that for people who have a specific mutation, Warfarin doesn’t even metabolize in their body.
That means those patients should not be given Warfarin. The entire medical system is based on a spray and pray approach–whatever works for 80% of the patients should work for 100%. It does not work. There are many studies about it–misdiagnosis is the biggest driver for cost. Imagine if you could have all of this data structured and searchable at the fingertips of a physician making those decisions.
Of course, everybody talks about oncology. The lion’s share of all investment goes into oncology, which is palliative. So, if you go wrong at first line treatment, you are screwed. Proper diagnosis is so critical in oncology. It is also true for neurodegenerative diseases. What therapy would exactly work for you if you could get the right recommendation in the first steps of treatment? That decides your quality of life. It is absolutely critical, and I say that at many places in the book.
What is AI? AI is computational power, it is algorithms, it is people, but eventually, the most important element for AI is data. Unless you structure this data, make it available in a context,and mobilize all of thisdata that is trapped in silos,itisnot going to work.
I feel extremely sad when some of these entrepreneurs complain. Creating large data pools is the way to go, because that is how you can train AI models. Why should scientists give up their data? Why should patients give up their data to a center that bundles this data into a company which sells this data for billions of dollars to Big Pharma? That is ridiculous. Why shouldn’tthe owners of the data have the right to decide what’s done with this data?
The right way is to have the proper incentive to share this data. Since I am in Europe, I would say that the European way is the AI and blockchain way, and that’s exactly how the future should be built, especially for life science. Because you don’t want your behavioral aspects, your genetic information, or your medical history to be shared with everyone. You can’t give a blank check to a company who just gets access to this large treatment center data.
You want complete control over this information, and blockchain is this fantastic disruptive technology that enables exactly that. Combine this with AI, and you can train models from data coming in from these different silos and take healthcare into what I call the third wave of AI.
Vision of the Future
Charlie Hoehn: Talk to me about that third wave of AI is, when blockchain and AI are working together in drug discovery. What would that start to look like? Paint a vision of the future for us.
Gunjan Bhardwaj: So, imagine everything that is published out there is crawled, extracted, and structured into a large searchable platform. Then you can search everything for relevance. So, it is kind of a Google Plus for healthcare. I say Google Plus because Google does not necessarily structure all of their data, and unless you structure, you can’t do an analysis. Imagine you wanted to compare the prices of all iPhones in all US cities. You can’t just get all of this data. You would have to crawl all of this information.
So, all of the published universe, which is exploding, is available in real time, in a structured format at your fingertips to do any kind of analysis with theinformation. At the same time, you have all of these data silos, be it medical records, experiments, sitting in a university, or clinical data sitting with some of the CROs. All of this data is mobilized. That complete security is incentivizing it being done fairly and appropriately.
All of this data is flowing into one large domain-specific AI, that speaks and understands the language of healthcare and life science. That AI processes this, analyzes and tries to find answers to a lot of questions, some of them I referred to previously. That is the future we are envisioning. A virtual life science paradigm–if investors are going to invest in a life science company and a biotech company, they will come to this platform.
If Pharma, a biotech, or an emerging bio Pharma wants to invest inanything, wants to know what novel targets or novel biomarkers, they will have to come to this. If patients want answers to questions, they will have to come to such a platform. And then, Big Pharma will not be built in a research lab. It will be built in such a platform.
AI at Scale
Charlie Hoehn: You said if things don’t change to become what you just described,we could stay stuck in the Middle Ages. I had a bit of resistance to thephrase, the Middle Ages, because I think of scientists and the drug industry as being cutting edge. There is so much money in it and they are doing hard research, but as you just described, the way things currently are really is the Middle Ages.
Gunjan Bhardwaj: It indeed is. What transformational, what architectural innovation have we done? I always feel in healthcare and in life science, we are solving point problems. Somebody will have a great idea, do a small research grant, buy all the data that is available, and try to solve that problem.
However, if he wants to solve this problem at scale, we need to solve the data problem, because if you solve that problem, then many problems can be solved at once. Some of the gurus in AI call it AI at scale. AI at scale is not a generic AI. I believe it is completely non-sensible to think of an AI predicting prices of baby diapers and the same AI being leveraged to find the next kinds of drugs.
All of us from various domains know very well people in those domains do not speak English, French or German—they speak medical French, medical German. Journalists have their own lingo, and bureaucrats have their own lingo. We all have a domain specific intelligence that helps provide context. Unless we create that, we can never get scale in solving the healthcare challenge that is in front of all of us. If you look at any country, one of the top problems is healthcare.
If you look at the solutions they are looking at, they are all point problems that they are tryingto solve, and this approach can provide a scale to solve many of those problems. We indeed are in the Middle Ages, if the Middle Ages stand for not being transformative, not having the courage to do the right thing at the scale that will really, truly benefit mankind.
A Challenge from Gunjan Bhardwaj
Charlie Hoehn: Let’s say you only have 100 copies of your book, Inside the Cockpit. Who do you most want to read your book and why?
Gunjan Bhardwaj: Everybody who truly believes healthcare needs to be transformed. Why? Because it is the purpose and the idea that conquers the world, that changes minds. An independent India started with Mahatma Gandhi being thrown out of a train. I would look at it from an industry perspective, maybe I would say the top peoplein medicine, and all the CEO of Big Pharma, but I would look at people who believe it is their purpose to transform healthcare, no matter what they do.
Charlie Hoehn: What is the best way for our listeners to either follow your journey and the work that you are doing, or to potentially connect with you?
Gunjan Bhardwaj: They should read my book and reach out to me at email@example.com.
Charlie Hoehn: Finally, give our listeners a challenge. What is the one thing that they can do this week related to your book that will have a positive impact on their life?
Gunjan Bhardwaj: Look at anybody in their friends or family that has suffered a near death experience related to a specific disease and think about what would change if what I described in my book were to become a reality.