The Intersection of Technology and Medicine – Wharton Global Forum São Paulo

It's time now that we embark on a fascinating exploration at the intersection of technology and medicine, where groundbreaking innovations continue to redefine the landscape of health care. We are privileged to have with us today an amazing lineup of experts who are leading the charge in this transformative field. Dr. Zeke Emanuel is the Diane v.S. Levy and Robert Levy, University Professor and WhartonÕs Vice Provost for Global Initiatives. He's also a legendary oncologist and bioethicist whose contributions have shaped the future of medicine. Dr. Vijay Kumar is the Nemirovsky Family Dean of Penn Engineering and a pioneering roboticist whose work is revolutionizing health care delivery. And this panel will be moderated by none less than Mr. Alberto Duran, a Penn trustee, Wharton adviser, and one of the most engaged Brazilian alumni at Penn who will guide us through this enlightening conversation. Welcome. Okay. Good morning everyone. It's really a pleasure to be here home. Finally, after I mean, I've been involved in board meetings at both the Wharton School and lately as a trustee for the last 20 years, believe it or not. And that's all thanks to our great Odemiro Fonseca, who's likely sitting here. I wanted to recognize he's been wonderful for us since the very beginning. And let's remember this our third forum. And Sallouti, he's probably outside talking to a lot of people, but he is doing a great job, absolutely phenomenal. So we need to thank him very much and the whole team. But Odemiro had two forums, so Sallouti has more to go. Anyway, I'm very lucky to be here with two giants. The reason I wanted this panel to happen is because, since I joined as a trustee, and I'm a very proud Wharton grad, five and a half years ago when I joined the trustees, I realized that, you know what, I don't really understand the university. But since IÕm a Wharton grad, let me give you just a few figures for you to have a sense of what it is. The University of Pennsylvania has 35,000 staff, 30,000 students, 12,000 undergrads, 18,000 grad, 5500 faculty members. 12 schools, only for undergrads. But this is the really interesting thing. We have an endowment of $21 billion and we have around $15 billion annual budget. Guys, this is huge. Just Penn Medicine is around $8 billion on a yearly budget. Since Marcos Galperin was here sitting and talking about a peer in Palo Alto, let me give just a few figures about that. In Palo Alto, our peer has 17,000 students. Roughly, they have 2500 faculty, and has around eight and a half billion dollars in budget. The whole entire university is the size of Penn Medicine. So we're doing a lot of things. And we really, I personally, didn't understand. So it's important to put these facts, since we like numbers, to put it in perspective of all the things we do. So again, I'm very lucky. I'm very lucky. And that's, by the way, is a great opportunity to continue raising money because they have a much bigger but much bigger endowment than us. So I call upon you and I will continue until the end of my life, coming to you to help our great institution. There are many things that are really happening, and I would like to start with Vijay. Vijay, you're a phenomenal person, a great leader. I mean, I see you moving all over to the university for years And IÕve been very lucky to have met you a long time ago and now weÕre becoming real good friends. Can you please give us an idea of what Engineering is doing at this moment? Sure. But first let me just say how happy I am to be in S‹o Paulo. Happy to be part of the Wharton Global Forum. This is my fourth Wharton Global Forum. And I'll just say that before me, no dean of Engineering attended a Wharton Global Forum. And one of the reasons I love this is because we have so many partnerships with the Wharton School. And today Engineering is everywhere. And as you pointed out, the University of Pennsylvania is a very tightly integrated organization. So lots of opportunities for impact across campus. What's exciting about engineering to me is really the flow of talent. I mean, in technology we talk about Moore's Law, you know, stuff doubling 18 months to two years, whatever. For me, it's the amount of student interest that's grown in the last eight years. We've doubled our student body and we've grown our faculty. Now, why is that? I think three areas that I'm particularly interested in. One is, of course, the intersection between technology and medicine. I think lots of interesting opportunities there. We had an amazing session on AI led by Eric Bradlow. Of course, data science is front and center for our engineering school as well. And then lastly, this is the grand challenge we all face: climate change. So energy, sustainability, and of course agriculture. We talked about that as an important part of that. So to me, these are three big things that we should be worried about as a society. And the engineering school is, of course, deeply committed to advances in all three areas. Fantastic. Zeke, you're a master. You're absolutely phenomenal. I've seen you for many years as well. We have had many, many conversations and you have been deeply involved throughout your life in medicine, and not only medicine, but public policy as well, which is incredibly important in the day that we live. Can you please also describe about why Penn Medicine is so great? Just to give a background. Yeah. Penn Medicine. Unbeknownst to the medical school and probably to you, is really the most innovative medical school in the world by far. And I like to say, in addition to the data, we are here at Wharton, you know, the fact that we have a $8 to $10 billion budget, depending on how you want to calculate it. We have a five-hospital healthcare system that really dominates eastern Pennsylvania, New Jersey, Delaware area -- but really the nation and the world. The really important thing is, you know, we are mainly committed to developing new treatments for unsolved problems. Since 2000 -- which is the end of the Human Genome Project, Bill Clinton announced the end -- until today, there have been four major advances in medicine that are really transformative. One is CRISPR. You've heard of gene splicing. One is gene therapy, being able to cure a genetic disease. And now we're curing more genetic diseases. One is CAR-T therapy for cancer, taking people who are literally on their deathbed and saving their lives and curing them of their cancer. And one is mRNA that brought us the COVID vaccine and now increasingly more vaccines. Of those four three happened at Penn, three. Now most of the people at Penn, not to mention here in Brazil don't know that. They didn't happen at Harvard, they didn't happen at Stanford, they didn't happen at Wash U or Johns Hopkins. They happened at Penn, and Penn Medicine has a unique culture. It's very creative, but it also has this one enormous advantage, which is we sit on the same campus as Engineering, as Wharton. And the collaborations that happen not just on these three, but hand transplants pioneered at Penn. Lots of health policy, all developed at Penn. The testing out of new treatments and the evaluation of whether they're actually promoting quality, that happens at Penn, not just at the med school, but in a larger environment that includes Wharton and Engineering, as well as the College and the departments there. So it really is the most innovative medical school in the world, and it's really making amazing advances. Who in the audience, by the way, took the Pfizer or the Moderna vaccine. Can you please raise your hands? Well, you guys are carrying, we are carrying Penn inside our systems. Who knows we won the Nobel Prize last year because of that? Yeah. We have two more Nobel Prizes to come for CAR-T therapy and gene therapy. And the committee in Sweden is just very, very slow. It really is. There's no explanation why you give a Nobel Prize for, okay, we have Neanderthal genes. That's really important. Except, you know, what are we doing to make our situation better? Oh, gene therapy. That's one of the things that will cure things like sickle cell anemia. In the case of Penn, it was blindness, you know, or CAR-T therapy, curing cancers that no one thought were curable. Those are those are real changes. It's very interesting you mentioned the Swedes, because I actually believe that the FDA moves very slow. So development of new treatment, anything in medicine takes forever. So then engineering is extremely important. Can you please tell us a few advances that we're doing with technology in AI to speed up all the development of technology that will go into our lives in the time to come? And by the way, I'm a believer that AI, now, we've seen it like a little box. When I was just starting my businesses in internet, we saw them as little boxes as well. Guess what? In my son's time itÕs just going to be around him and he's just not going to realize it. So it's not a little box that we're going to start learning about. We have to live with it. And itÕs going to start bringing our thinking towards hopefully a better truth. Vijay? Yeah, I think I mean, AI, of course, has lots and lots of applications and potential to change our lives as we've seen before. But to me, the really exciting thing about generative AI is its ability to model proteins at a level that we've never been able to do. And more importantly, synthesize new proteins. So everybody's familiar with Chat GPT. The idea of you have some tokens, this abstract latent space that suddenly spews out stuff that looks very, very language-like, human language like. So imagine the same thing for the world of antibiotics. Now, we probably know and understand a tiny fraction of the potential antibiotics that are available on the market. And of course, as you know, now, bacterial resistance has just grown. Drug resistance has grown in bacteria. But imagine trying to synthesize antibiotics like the ones we know. Imagine going back to extinct species, actually going back to the Neanderthal man and trying to understand what proteins are encoded in that DNA that we don't have today. And then try to imagine synthesizing proteins like that, because those peptides have never been seen by bacteria before. So they are actually pretty bacteria-resistant, or the other way around. Bacteria have never seen or have never been trained on these peptides. So the potential for synthesizing things that you've never seen before is just huge. And we're seeing that in collaboration in engineering and medicine. The second thing is this sort of high-throughput experimentation. So I can now synthesize lots and lots of peptides that sound viable, but then I can combinatorial and test them with lots and lots of cells or proteins, proteins that occur, for example, in tumor cells and see what works and what doesn't work. So thereÕs this two-phase process that I think will dramatically accelerate the development of different therapeutics. One is you generate new therapeutics very similar to the ones that we've seen before, or we think that'll work, but actually very different in how they function. And the second is this combinatorial testing. And this is important because, maybe, already some of you already know this, but if you look at the cost of drug development today, it's actually doubling again, this exponential Effect. If you go back to 1950 from today, $1,000,000,000 in 1950s would probably get you about 50 drugs. $1,000,000,000 today will get you one drug, and that's how expensive things are. So technologies like this can really accelerate the development of new therapeutics. And you just mentioned mRNA vaccine. Again, lots of interesting applications there too. At the end of the day, it's about finding the right proteins. So the same technology can be used now to synthesize proteins of interest. You back off, figure out what the mRNAs are, and then you deliver those mRNAs to the human body, and then that becomes part of our bloodstream, so to speak. Talking to, a lot of faculty in the university throughout the years, I mean, I just get overwhelmed by the thought that the real aim here, one of the real aims, is to try to model the human body inside computers. At the moment that we're capable, and we're probably not too far away to do That, is a moment that we're going to be able to develop technology and develop new cures in a very fast way. And hopefully we can persuade the FDA to start approving all these things. How do you see, Zeke, this AI introduction into medicine in general, not only the university, the faculty, but inside the population, the whole system? So I think of AI as sort of one component of digital medicine or virtual medicine. And, you know, during COVID, we had what I like to call digital medicine 1.0, which was basically, you know, the internet helping an in-person doctor and an in-person patient or an in-person clinician connect with each other, not coming face to face. And that is very useful for lots of people, not traveling to a facility that's far away or in the middle of the night getting real advice. But that's really 1.0. I think where we have to go is 2.0 and 3.0. 2.0 is where we're using AI but other technologies to really help us with diagnoses. And we see that, you know, in many ways AI and other technologies can help us solve problems that human beings weren't able to solve. So I'll tell you about a company that I'm working with that sort of embodies this. It's a company that has trained algorithms on lots of patients to look for drug-drug interactions. The older you are, the more health problems you have and typically the more drugs you take. So we call it polypharmacy. Lots of drugs and different drugs have different interactions. Some of them should not be used on people over 65. They cause too many problems. Some of them interact with other drugs. This company looks at patients in the hospital and identifies the 5% at highest risk of having an adverse complication from their drug-drug interaction. And they can do that in seconds. And then it scans the whole web, and can identify what is the appropriate substitution or other intervention that should be done and can give all that information to the doctor as well as the pharmacist so they can prevent this. It's now being deployed in hospitals -- not Penn, I'm working on that -- to reduce costs and to prevent people from having complications that land them in the hospital. That's how it's reducing costs. And, you know, humans can't do that. We can't scan all the thousands of drugs and looking for those, you know, millions of interactions. So that's one place, thatÕs 2.0. And then 3.0, digital medicine 3.0, which we're just beginning to scratch the surface of is using a AI to both diagnose, but in real time have the computer or the phone interact with the patient and give them therapeutic maneuvers to actually improve things. Now, the place, it's easiest -- and again, I'm working with a company, a different company that's doing this is in physical therapy. There are a lot of companies out there and certainly in the United States called Hinge or Sword that are using technology to give people physical activity. So between going to the physical therapist on Monday and coming back on Monday, they always tell you to, you know, five of these repetitions every day. And what happens when you go home? You don't do it, right? One of the brilliant things about these programs is they actually interact with you. They kind of gamify it, so that you do do it. But the next generation is they give you immediate feedback. They don't wait for you to get to the therapist again, the next Monday. They give you real-time immediate feedback on how they're doing because they're watching you through the phone or through a camera with no sensors on and seeing how you move. So, for example, if you have to do this kind of exercise for your shoulder, they see whether you're crossing the midline properly or not, whether you're moving the other side of the body, that's 3.0. That's where we've got to get. That is going to increase dramatically the productivity of every therapist and clinician. It's also going to get people doing important health before they get to the clinician. Right? I don't know what the waiting lists are like in Brazil. I can tell you what they are in the United States. You know, my daughter just tried to get to a neurologist. They said April 2025, what is that? You know, a year from now? So this is true of almost every specialty, especially among children. So you can actually do therapies before they get to the doctor or the therapist, in between, and then after, so that they don't lose the advantages that they've gotten. And I think this is where we're going to really begin. They're going to improve access, they're going to improve, make you recover faster, they're going to lower costs. And I think they're going to relieve doctors and clinicians of the thing that we hate the most, which is administrative paperwork, because they can generate like Eric Bradlow, so they can generate the note, they can actually record how you're doing and show the progress quantitatively in a way a therapist couldn't do. So not only they're going to be just as good as a therapist in some ways they're going to be better than a therapist. They're going to actually be able to measure you. So I think that's what we're going we're now at 1.0, maybe 1.5. in some areas. We're beginning to get into 2.0 and some advanced companies are scratching 3.0. But that's where we're going to go. Now, there is one thing. You mentioned the FDA. You know, AI and Chat GPT are being embraced by finance. you know, they think they're going to have their workforce. Medicine, itÕs very different, at least in the United States, very hesitant. We see every problem it can create. Oh, you know, if you're African American, it doesn't recognize you as well. And those are all serious problems, But it could have the potential to slow up our adoption. Just like medicine was the last with electronic health records to adopt computerization, I'm worried that we're going to be slow. We're going to find every single problem we can before we actually embrace it. And we're going to create a negative mentality. I think that would be a mistake because the alternative to embracing these technologies is not a great health care system. The alternative is many patients aren't getting care, and they're not getting the best care. And that has to be our goal: getting as many patients as possible the absolute best care in the world. And that's only going to be possible in the future with new advanced technologies like AI. So I like to say the future is you're either going to be a doctor or a nurse or a therapist alone. Or, youÕre going to be a doctor, a therapist with AI. And the first part, they're dinosaurs, Neanderthal man. They're just not going to exist. You're going to have to use AI in creative ways. And I think that's going to happen over the next 5 to 10 years, the way I see it. Hopefully Eric will turbocharge that. Guys, look at the business opportunities all over the place, itÕs just phenomenal. There's so much to do, itÕs exciting. Vijay, your whole life has been in robotics. How do you see robotics going, and nanotech going into this field? Well, I'm just staying with the theme of medicine and technology. So if you think about robotics being applied to medicine and to just follow up and what Zeke said, the one new class of robots we're seeing are these so-called therapy robots. And so you don't the therapist doesn't have to come to your house. You don't have to go to the therapist. But indeed, those machines can be programed remotely and then they can also adapt to your physical characteristics. And that's here today. Yes. Those machines are at the hospitals. You have to go there, but it's just a question of time before they find their way into your houses. Hopefully we're not going to tax them. Yes. So that's sort of I don't know, it's 1.0 or 2.0, but that's already happening. I think the real power is if you combine this idea that you have novel algorithms that can activate machines, but those machines are living inside your body. That to me is the true next frontier. So just one story I'll tell you about. A colleague at Penn who was working with neuroscientists across the street in creation of these nano robots that can be injected inside your body. So one of the therapeutics for spinal cord injuries is you take these nerve cells and you want them to reattach. So it turns out that if you, this sounds, terribly naive, but if you take these nerve cells and stretch them so that you provide junctions for them to reattach, that actually accelerates the growth of the nerve cells. So for us to have robots that actually grab onto this is not hard. And then to pull and stretch at the level that you need to do exerting the right amount of forces, that's also not hard. So the key challenge is to try to figure out how to drive the robot to the to the relevant part of the human body and then how to program it, how to get it to do the right things. And that's where the generative AI piece comes in. So by the way, I'm not a big fan of the use of the word ÒAIÓ because we collectively, although we do it, we collectively don't understand what intelligence is. We understand artificial, but it's not cool to call it ÒAÓ -- youÕve got to call it ÒAI.Ó But the real way these things manifest themselves is that they augment our intelligence. So I prefer to use the term ÒIA,Ó intelligence augmentation. So here are things that the human, the surgeon cannot do. The surgeon is smart, but the robots are smart in a different way, and you can get them to augment the surgeon's intelligence. So all the therapeutics we are looking at, even maybe analyzing radiographs, X-rays, pathological samples. All of that has been accelerated by AI. But really, it's a human being that's making the decision. But the human being then makes fewer false positives, fewer false negatives. And that's really where everything is playing out now. But again, to Zeke's point, our engineering school, we built a facility for nanotechnology and this was about 11 years ago. No one dreamt that that technology would be used to make nanorobots that live inside the human body, lipid nanoparticles that would essentially deliver the mRNA vaccines that we all have have come to know and love. So that's how fast things are going. Ten years ago, we did not predict where we would be today. And it's just truly remarkable that all of this is happening. Oftentimes the thing I have the best job on campus being the dean of the engineering school, No disrespect to our Wharton colleagues here. But it's just it's just a great thing. So can I pick up on one thing Vijay said and also bring it back to something that you keep mentioning about the FDA. So one of the studies we've just concluded is to look at AI technology that can scan electronic health records and see if a patient is eligible for a clinical trial. The slowest part of doing a clinical trial, and one of the reasons it costs $1,000,000,000 for drug company development, is to enroll patients in the trial, find that patient who has the right lung cancer with the right markers for this drug. Very complicated. You have to go through hundreds, many times, thousands to find the one patient. So we were paired up through a connection I had with a company, where the company has natural language processing that can go through the medical record and address a bunch of questions about our requirements, eligibility requirements for a clinical trial. And it turns out we ran the AI; we ran the humans alone in the old style, the way I used to do it when I was a cancer doctor; and then we ran AI with humans. So the AI scanned it, the humans reviewed it and reviewed any place that was identified as ambiguous. It's quite clear that just AI alone is the fastest, but it's also not the most accurate. But the AI combined with the humans -- different skill sets was way faster than just humans alone because the AI went through and prescreened and then the humans just had to check and really identify the few things that are very hard to identify in an electronic health record for a computer. Like was the chemotherapy given or just considered as one of the options? Just as an example, did they get the full dose or not the full dose? Things that are very hard, it turns out, to get out of a medical record for natural language processing. Not easily given in a structured format. Much faster. So part of what we're now in the process of doing, one of the reasons we did this study is, now, to go back to the FDA and say, listen, we can do this faster and we can do it more accurately. Machine and humans. And you should permit that for enrollment in clinical trials and that might dramatically speed up the enrollment in the clinical trial by identifying, prescreening and identifying patients. So that's something we did at the Abramson Cancer Center in combination with the Health Care Management group at Wharton. So I think these are the kinds of ways you can have AI augment humans and then, come full circle, be able to more rapidly develop drugs. More rapidly we're hoping to convince the regulators that this is just a much better way for everyone to go. It'll reduce costs and make trials much better for patients. Unfortunately, we're running out of time Ð itÕs red already. But you have worked in the White House, you have worked with government for a long time. You both have. How do we keep government outside of our noses in allowing us to do the right thing, yet let us progress? Because the Chinese are actually doing that, believe it or not. I mean, I just came back from China and it's a whole revolution. Chat GPT, they have their own version and itÕs really big. So we do live in a competitive world. We should be always aware of that, and I want our side to just be successful. How do we do that? Well, I think the first thing is we're not going to live without government. We don't want to live without government. We need regulation there. People do do bad things. We also don't want approved drugs or devices or robots that are going to be more harmful than beneficial. So there is a very important role for regulation. I think your point is right: we have to be able to do it faster. We also have to be able to do it with less red tape and extraneous considerations. I think one of the benefits, there aren't many benefits to COVID, but one of the benefits of COVID is it did show that you can turbo charge a system and even the review system, you can conditionally approve things like vaccines, things like drugs while you have a positive indication that it's going to work, not the proof you want and collect the data as you progress. And either pull it off the market if it's not working or if it's causing more harm or approve it rapidly. Very important advance. Now, what does that take? Like most things, it takes money, investment. It does take people to review the data. And you know, you probably don't know this, but the FDA, I think the last estimate I saw touches 60% of the American GDP between drugs and medicine and foods and cosmetics. But, you know, it has about 4000 or 5000 people working for it. It's an absurd ratio. And its total budget is under $10 billion. It's absurd. I mean, we have to have government. And this is a case, in my humble opinion, of where more will be better because you'll have more people so they can concentrate and process the applications for drugs and devices much more quickly. But you have to do that in a very intelligent way. And I think, you know, unfortunately in the United States at this moment, we're more polarized than we are focused on delivering the best product for the American public in all sorts of ways. I'll just say one more thing, which is, yes, we want government out of our lives, but this is an area where industry is outpacing academic research. So the gap between what we as academics know -- and that's important because what we know, we publish. And we get it vetted by our peers, and everything is transparent. On the other hand, you look at all these large language models, how they operate are still a black box. We're trying to figure out these black boxes work. So it's vital that we have some kind of an oversight capability to look into it. But rather than have governments stick their nose, we should stick a nose into what they're doing and be more proactive. And that's one of the things we are thinking about at Penn. Yeah. So, I mean, one of the things that. All right, just to bring us home and I'm sorry to go over and eat into your lunch, literally one of the things that our interim president has committed to is doing more in this area and reviving a center we have in Washington called Penn Washington and working on issues, domestic issues in America, but also global issues. And one of the issues that Eric Bradlow and I have been talking about as a result of this conference is can Penn be the place that really thinks through global regulation of AI and related technologies? It's really important and you just can't polarize it. No regulation or, you know, we're going to shut it down by a variety of procedures. We have to think smarter, and we have to think smarter ironically, as you point out, Alberto, in a very competitive world environment. I think that's something that Penn can lead on and that we're hoping to create a major initiative around. So if you're interested, love to talk to you. Please help me out for a big round of applause. These two giants.