Transcription
When you have anything from a little surgery or discomfort to something major, you start meeting people and the system's pretty efficient. Your friends, your family say, you know, we knew someone that had this. Why don't you talk to them? All of those stories and go into AI. It's a phenomenal place that we're at in terms of transformation.
Hi, my name is Mika Newton and welcome to AI and Healthcare. I am super excited to be here today with my great friend and colleague, Bob Battista. Bob is so accomplished. For example, Bob is an advisor to the Institute for Health Metrics and Evaluation. He's a board member to Every Cure. He's been the CEO and founder of a number of startups across the space. And I just couldn't be more excited to have you here, Bob. Thanks for coming on.
Thank you so much, Mika. Thanks for having me and it's going to be a great pleasure to have our discussion today. You know, I know you spent a lot of time working in drug development, drug discovery, and particularly the intersection of AI and trying to understand the evidence to support, um, drug development. Um, you know, I know you've been working on drug repurposing, uh, along with the Every Cure folks. Can you tell me, what do you see as the biggest hurdle in drug repurposing today?
Yeah, it's a terrific question. I think that the biggest hurdle related to repurposing is really the sharing of knowledge from the pharmaceutical industry across the regulatory firewall to the research and development side of the house. I think that there's about three thousand drugs indicated in the market for about twenty thousand indications. And the knowledge inside of the pharmaceutical organizations is really all we need to know about drug repurposing. So we need to process that knowledge, but there's regulatory restrictions to doing so. So I think that innovation really needs to happen on the policy side of the house versus the technology side of the house. The technology today is phenomenal to identify repurposing candidates. The problem is that a lot of that knowledge is already locked up inside of what you would call really market access issues around regulatory restrictions for pursuing new drugs and the expensive nature of clinical trials in relation to getting those drugs approved.
I guess what you're saying then is we know what the drugs are and we have the technology to identify probably additional indications if the biopharma companies don't already know where else they would go. But there's this issue with basically every drug needs to go through the full process. And so you have to go through this long, expensive regulatory cascade essentially every single time in order to get something new to the market and approved.
Yes, to break it down, I mean, you have a 17 year patent life. The average drug gets to market in about the ten year mark. You have seven years and it's actually maybe ten to eleven years now. So you have six to seven years of active marketing before it goes generic. Once it goes generic, everything changes. The price comes way down, manufacturing gets done by a whole bunch of companies and generics and other things. So companies oftentimes will have a molecule that gets to market that has many possible solutions, but they have to pursue one of them, being profit driven, much like pharmaceutical companies are, they'll generally pursue the biggest indication that a drug can have an effect on the population or quite frankly, sometimes the most meaningful like in oncology. So you might have a molecule, rheumatoid molecules have a huge spectrum of indications, is a good example, actually. And oftentimes they'll get pursued for a large indication and there's a lot of small indications that they just never get marketed for. So all of that potential sits on the shelf essentially in the pharmaceutical company. So we need to find a way to loosen that up and bring it to market.
I can talk about some ideas for that. Bob I'd love to talk about that. You know, one of the things that also strikes me is and I guess this may be a question and as you think about the efforts here in this space, do we really need to go down that pathway or are there, particularly as we start thinking about maybe some of the other generic molecules out there, are we able to learn from things like the real world data or the other experiences of organizations enough that we would actually be able to get some of these molecules to the right people who need them. And so it feels to me like we just have more and more knowledge and there's a very formal set piece process that people go through. And then there's the, hey, the drug's been in the market, maybe it's even cheap, but maybe it can save somebody's life. Like, how do we identify that opportunity as well?
For sure. There's two other huge repositories of knowledge. One is physician knowledge. So 65% in the U.S. of all drugs that get prescribed or prescribe what they call off-label, which means they're not indicated for the things they got approved for. So that means that physicians know that they're effective for certain things. And then also patient knowledge, patients do phenomenal things with drugs because they know what works. So oftentimes they know what dosages work and they know what certain drugs are good for other things. All of that knowledge is really unleveraged at this point in time. I should say that physician off-label prescribing knowledge for instance most famously drugs like Viagra and Cialis, they were heart medication drugs. And one of the side effects was erectile. So they got indicated for erectile dysfunction. So that's a perfect example, but that indication was big enough that the pharmaceutical companies were able to pursue it. So that's how that class of drugs really got discovered. It's a great example.
If we have all the technology, so we have the AI, we have the real world data, we have different statistical methods for doing analysis, meta analysis, etc. so we can pick up where the drugs work. We have biological information, which is being stored in the pharma companies from their own R&D efforts, their own exploration of the molecules itself. How do we share this information? Is that part of the problem here? Just like nobody, to your point, I think you're saying it's kind of just locked up and siloed away. So it's just not general knowledge.
Yes, we need to have some safe harbor provisions, if you will, in the policy area where pharmaceutical companies can share the data to do both regulators and, quite frankly, to each other. I don't want to get the number wrong but I think it's two or three drugs have been actually invented inside of pharmaceutical companies. With GLP-1s it might be actually more like a half dozen now. So most of the stuff that comes to market, comes through bench research, but once it gets acquired by a pharmaceutical company, you stop having open access data information. Because the bench research is basically open access. Most funded in the US by the NIH. So we need to find a way to let pharmaceutical companies share knowledge between each other and also to regulators and to other individuals that want to pursue niche indications. And I think we need to have a financial instrument where we let pharmaceutical companies still carry the R&D and the development costs of their drugs somehow on their balance sheet. So when they let a niche indication go and let someone else pursue it for drug repurposing, it won't erode their main market, if you will.
Yeah, that's very interesting. I was talking to another friend of mine about this and he was actually trying to license drugs that he thought had different purposes from pharmaceutical companies, where they would find something and they would know who holds the IP and they would go and try and license it. And he was telling me that even when it's failed, there was often a lot of hesitance to even try to bring it back off the shelf. So it's essentially sitting there as a wasted asset. But it was primarily due to the cost or exposure that someone would encounter around that. They didn't seem to be a financial incentive, to your point. To bring those back to market.
That's right. There's actually a financial disincentive, I can give you another just classic industry example. It's the Avastin - Lucentis issue. Avastin was a very famous and incredibly successful oncology drug for breast cancer. And the vial size, it was about $1200 for oncologists to buy. They were oftentimes put it into a chemo cocktail and and the exact same drug was also very effective for macular degeneration related to diabetes and to, patients that had macular degeneration and the various causes of it. So they released it for ophthalmologists and and retinal docs for injection into the eye half the vial size, and it was $50 a vial. So what was happening, and by the way, cured macular degeneration for the most part. A couple of shots in the eye I think per year, I'm not sure exactly the dosage regiment, but it worked. And they did a great job. It was actually Genentech and Roche. And what happened is the oncologist always wanted a different vial size than the one $1200 vial size. But that was the one that got approved through the regulatory channels and all the other things that I talked about and the true expense of getting it to market right. We give a lot of criticism for drug development. Without pharmaceutical companies, you wouldn't have a lot of innovation because they take the risk of manufacturing and regulatory and trials and all the other things. So what happened is oncologists were buying two of the vials for macular degeneration and they weren't buying the $1200 Avastin vial size. So what happened is the macular degeneration one got pulled off market. It was a terrible thing. I guess patients that could have afforded the $1200 vial size and by the way, it doesn't get billed in through insurance if it's not approved. So probably still did it. But by and large, most of the patients were not eligible for the shot. So it's one of the great travesties of this regulatory problem that I'm talking about. It was really driven by the differential reimbursement for the indication use, where the oncology indication was super high value because it's cancer versus the macular degeneration, which is still valuable. It just isn’t comparable really.
That's right, niche disease versus a larger disease. Yeah. So this is really interesting. So we have all these drugs, right, that are out there now. We have ideas about new drugs, right? We're talking about what does it take to bring some of these molecules to market. One of the points you made a little bit earlier was really about the knowledge that patients and physicians have. And one of the things that I get asked a lot about, particularly as it relates to AI, is how is You know, A. I. And data, particularly data around the evidence to support the use of different drugs. How is it actually changing health care decision? So as this information we're talking about becomes there's more out there, right? How are patients and doctors now able to leverage this decision makers? It's just like everybody just asked chat. GPT. What it thinks are the best drugs for you? Or do you see this kind of technology? Revolution happening around clinical decision making in a different way. How is AI changing healthcare decision making?
It's a great question. A little complicated in that I think a big part of the answer is that the knowledge that I’m talking about i s not in the LLMs. So what AI does extremely well in your example of access and people having access to asking those questions, they're asking it in front of the firewall, if you will. So all that knowledge that I mentioned, like if we had the knowledge that sits inside of pharmaceutical companies, arguably the big asset of a drug, is the 20 years of call center and patient-reported outcomes and all that sits inside of the firewall pharmaceutical companies. So I think we need to get access to that data and again, have regulatory mechanisms where we can because it would answer a lot of the questions you're talking about. Today unfortunately, I don't think those questions could be answered through AI where AI is really being helpful in all this is the traditional science part of it, looking at it all, creating all the knowledge graphs with drugs and mechanisms of actions and what they call pathway, how effective a drug is with the clinical pathway that it has its gene expression and all the other things. So that's very valuable. But that's more for deep researchers and organizations like Every Cure that are doing incredible work on identifying new drug repurposing candidates, but they're starting from scratch, we don't have to start from scratch. We have all this incredible knowledge that AI could process if we were able to make it available into LLMs.
So it's interesting when you talk about that, one of the things we worked on together in a previous project we're working on was really guidelines and evidence generation. So how do we look at, right, looking overall at healthcare guidelines, And what are the best practices for care? One of the things that happened to me in this current xCures I was working on was like, we took a long look at how complex oncology guidelines were getting first. And so, like there's hundreds of pages being added to NCCN guidelines on an annual basis and there's just like exploding all over the place. Wouldn’t some of this information actually help us maybe deconvolute this space? Because it's not like there's a prescriptive treatment path for every single person, but there's some good understanding of what best practices are.
It's exactly right. You bring up another phenomenal point in that regard, which is really guidelines and they have what's called the living guidelines. Actually, the CHEST, American College of Chest Physicians really started the living guideline model probably 15 years ago now. And we did a lot of work in our prior company, Dr.Evidence on this, where we made dynamic guidelines that got updated with clinical evidence. This is where AI could be really, really effective. Guidelines should be tailored for the individual. So most clinical care has a major branch in the beginning of it, which is do you have a prior condition or do you have an at risk condition? So do you have a family history of risk for what you're being treated for? And that's where AI could really be effective, is it could look literally read your patient chart on the fly, look at the clinical evidence that exists. If you added the firewall issue I mentioned, you don't even need that for that because we have a ton of amazing data in published clinical evidence. Which I really haven't talked about and given respect to. There's over 40 million clinical studies probably in the market now that represent a billion pages of clinical data and charts and all kinds of other things. So table data that's really, really relevant for this stuff. So you can actually read that today and you can create a dynamic guideline for individual patients. By the way, if you do this, you would save tons of money because for instance, in the U.S., we end up doing the right thing, but we don't do it in what's called the earliest instance of care. So systems like in the UK, NICE, they work really hard to make sure they're covering the right treatment on their reimbursement protocols in the earliest instance of care because they know they don't have to pay for it eventually. In the U.S., that's not exactly the case because you have patient churn from insurance plan to insurance plan. So we don't think about owning the patient holistically.
Yeah, that's an interesting point. You know, one of the things. We ran into, and I see this all the time in oncology, right, is the standard of care, right, that we know that doesn't work, right, so you can't go and do anything that's more advanced care, that we would think of like second line or third line, until you've completed the first line, This is something that makes me kind of Generally you have to fail, right, you fail. You have to fail two or three things, you're, you're in, you know, you're in the lab essentially, you're a lab animal, in many cases until you get to the right instance of care. Right. And those initial things that we're doing, some of them actually make the disease worse. Right. So again, kind of using the cancer example, you're exposing your cancer, which now has never seen any drugs, right? So your cancer is like, well, it's like a baby. It's got no immune system, right? It's growing like crazy. And then the first thing you show it, right. Is a chemical that it, you know, is bad for it, but you know, it's going to figure out how to resist. Right. So what do you do? You're actually training the cancer to be strong. And this is, again, I'm not a doctor. This is my lay person view, right? We've just trained that cancer to be even stronger. Before we bring the things that we actually know are going to happen, um, are going to work to that. And I think that's, that's pretty, so your, your point here about the sequencing of care and the costs associated with it, I think is just really, really relevant. Do you think there's a, a large cost op, you, you talked a little bit about how much money we could save, right, if we just optimized care for individuals. Is that kind of true across the board? For sure. You mean across treatment conditions? Different conditions, different populations. Yeah. Yeah. The, the, the best example of this is, is really, um, in oncology, but it's really in diagnostic versus management. So a great example is that you generally, um, there's certain tests that are very expensive. Um, and they're, um, they're not as readily available. So the most classic example is like a pet CT. Um, you oftentimes need to go for earlier diagnostic screenings. And if If in the period of, of moving to the right test, because it's expensive, um, is nine months, you've managed nine months of management. Now some, on some aggressive cancers, nine months could be fatal, right? Like, pancreatic cancer is a two months sentence. If you can maybe stage it earlier And this is where gene testing is very helpful, particularly with AI. And we're doing a lot of great things on identifying, you know, gene sequencing with diseases. So I think that body of work hasn't fully yet hit the market yet. Um, there's some great work being done there by, uh, Callum labs at Brigham women's for instance. But we, um, we. If, if we staged earlier, we'd be able to manage better. And oftentimes staging can be a little bit expensive. So we need to get better. We, we have what's called defensive medicine where they say, you know, we're going to try this test, this test, this test. If that doesn't work, we're going to send you for the right test. It's a perfect example of what you're talking about.
So as these health systems, and I think maybe it's the insurers, right, are doing here. Is there an opportunity with technology now to actually prove this point or is this just so, you know, intrinsically built into our, our system now that we're just kind of stuck with, like, how do we change this, Bob? Like what, what has to happen to kind of change that dynamic? Because it feels to me like if we were saving money and people were getting better care at the same time, that that's kind of why we all started doing this in the first place, right? So move to it fast, right?
Yeah, yeah. Uh, it's, it's the big question. It's, it's, it's a tough one to answer. Um, I think what helps this along is the idea of, um, in the U. S. what you call like these, uh, integrated delivery networks or accountable care organizations. This is some of the historical stuff. where you do more preventive medicine, right? So Kaiser Permanente is nine and a half million or so patients in a system. A lot of their practices and things like, uh, systems like Intermountain Health, and there's lots of other systems. I just don't have time to mention, uh, North Shore in Chicago. There's a lot of phenomenal systems, Geisinger, where they take a little bit more of a holistic approach, Mayo, Cleveland Clinic to care, so perhaps. It's maybe studying those systems a little bit and, and moving them over to, um, some of the other large systems. I'm going to approximate some numbers. I'm not sure. I haven't tuned these up in the last year or so, but about in the U. S., about 50 percent of the health care dollar is really Medicare and, and, uh, approximately 35 percent is self insured plans. So companies like Microsoft, CVS. They actually carry their insurance risk. They hire what's called a third party administrator, and it might be a little more like 37%. The rest is what you call traditional insurance. So if you look, you know, Walmart's self insured, a lot of companies are self insured. If you looked at the practices of those organizations. Maybe it's sharing best practices out across the, you know, bringing it over to Medicare. One of the great travesties, um, is, uh, you know, Medicare, and it's changed now, and I don't want to be political, but Medicare hasn't been able to negotiate price. So if it gets approved by the FDA, they have to cover it at cost, um, or at price. If there were just better collaborations between reimbursement organizations and, um, and, and maybe pharmaceutical companies, you might get some good innovation. But I think studying what's happening in the ACOs and the systems that have to really, uh, endure the total cost of care and bringing those practices over, maybe that's a way to find good reform.
That's a great point. So I'm going to switch the topic up a little bit, uh, on you here. When we think about patients, like I as a patient, you know, I know one of the things that you, you worked on a lot was I'm a patient. I think I need this drug. My doctor thinks that I need this drug, right? But I'm having a hard time getting it, you know, covered by my insurance, right? What is the evidence that I need, right? In order to do that, to get that approved, like, how do I show things like medical necessity, etc. Is, is that world changing or is it still very much the way that it was before?
It, uh, it's not changing fast enough. And we've done, um, we've done many, many cases, hundreds, if not, you know, into the thousands, where We helped patients give medical evidence that prove that the treatment they're desiring is usually different than the reimbursement guideline, but it's in the best interest of the insurer. And, um, I would say that on I approximating, I mean, guidelines take a long time to, we did a lot of guideline work, evidence-based medicine guideline work, you and I and others. Um. You have to form an evidence review board of, of top, um, doctors in the space and academicians. That takes about nine months. You've got to then look at the evidence. So it takes about two to three years to update a guideline. Um, so in a world where we have rapid movement of innovation, the, you know, the science has gotten phenomenal. You, um, there's, there's about a three, three year lag, you know, from best practices, uh, and successful. patient care where you've got longer survivability, you've got progression free survival, you've got long term survival, um, until they die, you know, natural, other natural causes, uh, the, there's a disparity between the evidence and the guidelines, so we've, we've provided individualized, um, informatics. That help people get best care, which again is in the reimbursement interest of, of the, of the insurer. The three big categories I mentioned, you know, self insured, uh, the government with Medicare or, or traditional insurance. And obviously is in the best, um, interest of the patient. So, uh, it's something we should chat more about.
So, you know, part of what's been underlying this whole discussion, Bob, and it's come up a couple of times now, is really thinking about, you know, the, um, The policy that governance, right? So just going back to our conversation started. You start talking about governance and policy around data safe harbor, right? For pharmaceutical companies to share information. Um, I've spent some time recently looking at the risk associated with P. H. I. And P. H. I. Disclosure around medical records. right? We have insurers and what they do or don't have to cover. What are they allowed to like these topics, right? They can go on and on and on. If you think about maybe the big main areas, if we just try to break it down to like the pillars of governance, right, that are out there, where do you see the big opportunities? And let's kind of dive into some of those and kind of talk about why they might be areas. Certainly if of potential. Right. So, um, give me a little bit more color on that. You mean the, the, um, So we're talking the U. S. market for now. So the U. S. market, um, uh, we could talk about it globally, too. I think some of the same things happen. So let me give you an example, just the one I've been working on. It's like privacy, right? So here in the U. S., we use HIPAA, and HIPAA is our privacy guideline. We use GDPR in, you know, in the EU, basically. There are different flavors of healthcare privacy. all across the world. I actually think that, for instance, HIPAA as a law is quite old now. It didn't contemplate much of the technology, right, or innovation that we have today. Um, it's good that we have it, right? It's a foundation, but that to me round around individual privacy, access to things like my own data, my right to, um, do things. So that's, I think, one example. I think the other example, just to kind of, you know, contrast it to something would be what you were talking about with pharmaceutical data sharing, like should pharma companies have to put all the data on their asset into, maybe it's not the public domain, but at least a protected domain so that we can all learn from all the information, those, those types of decks.
That's right. And by the way, I think they would like to, I think they'd like to share information when very classically Novartis and GSK. Did a big BD swap. I think of the respiratory portfolio in oncology the The, all the molecular IP moved, but the data, like the call center to hit, it didn't move arguably. It's the most, some of the most valuable data in those 20, 30 year old franchises, um, much to the chagrin of both companies, they wish that it didn't move, but there's just, again, not good mechanisms to enable that to happen. So, uh, HIPAA is a great example. I, we can maybe focus in on that just for, uh, for a couple minutes. It used to become, and we all know why HIPAA started, right? Cause there was some of the early, it was really around prior conditions. And if you had a prior condition, you couldn't get covered. Um, again, apolitically things like Obamacare, you know, prevented that from happening. Right. And once that kind of got put into the market, the economic effect of that all sifted out and I'm sure for the, for the insurers. Um, so you can't not cover someone now if they have a prior condition, but it became so difficult to get your patient record that people just wouldn't do it. Companies like Microsoft started HealthVault, um, over a decade ago. They even gave up, you know, one of the richest companies in the world. Uh, Google tried it. And so, let alone individual patients just trying to move their record from one physician to another. Today, and I think some of the technology actually you pioneered. You can, which was, which was, you know, I'm, I'm here in Santa Monica. We've had these horrible fires, over 10, 000 families displaced. Um, a lot of those families had CDs and medical records in their house, they're all gone. Um, so, today, I think you can permission getting your record and you can amass them from all the systems you've been visiting, you know, in your, in your local, uh, community or even nationally in minutes, if not, you know, within an hour. That's a phenomenal change. HIPAA prevented that early on. And, and it was a. It's probably a good law at the time, but it is certainly antiquated. So I think being able to democratize that knowledge that you can actually get all your records today, because I don't think people realize you can, and then we can use AI to actually process all the nuance in them. Look at the historical year over year tests of your blood work or anything else and you can actually set Thermometers on your records to say I want a functional medicine approach. I want a traditional medicine approach. I want a Prescriptive medicine approach whatever it might be. That's where AI is really gonna change things radically I think for people.
Yeah, I think it brings to me the idea of risk So when I think about what under HIPAA really freaks people out Right. And particularly health care providers, it's the risk that they have inadvertently disclosed private information about you to somebody that they shouldn't have given it to, and that they're going to be held liable and suffer the financial penalties. of that. And as soon as you start thinking about that, it's like, you know, if my hospital got hacked, right? Like what would happen to them? So there's just this incredible level of probably well meaning paranoia is the only way I can think of it, right? Like it's truly meant well. And I think a little bit like the pharmaceutical example you got gave where there's like no safe harbor. It's very difficult for these larger institutions, in my opinion, to actually move that risk. So if I say, I want my records, right, that should be my risk now. Once I get them, that's on, on me, but this hasn't all been processed or litigated, right, or gone through. And so there's just these, I think, open ended questions about kind of who holds the bag, right, at the end of the day on all of this risk with that.
The, you're right. And the other interesting part of that, and I think in oncology, something like, you would know this, but 90 percent of the record is actually digitized, where if you look at the other end of the spectrum, like in primary care or, you know, your general, your GP, most of that is handwritten. We all joked about not being able to read, you know, doctors, they, they went to many years to figure out how to do that writing, but you know, the hieroglyphics, you know, And, um, but you can actually read all of those charts with AI now, which is a phenomenal change because most physicians would have told you in the past, I don't want to get all your medical charts. Just give me the summary reports. Um, today you can actually move all those charts, all that data over on yourself over. And, you know, let your physician stand on the shoulders of all the physicians that came before them to give you better care. That is a huge change today that didn't exist in the past, which is great. And that's where AI is really, I think, going to radically, radically shift things. How do we think about the governance of AI? I think that's a really big open question. So we have all the risk associated with medical records, privacy risk associated, and we're going to give it to a machine computer that's going to develop some set of information that then going to be interpreted by a physician. You still need a licensed healthcare practitioner out there. But it leads to a really interesting question. And I think there's a corollary to this in the self-driving car. And we have them here in the city in San Francisco. They're driving around on their own. There's nobody in the car, but it's offering a taxi service. So when something happens with one of those cars, I still think there's a big question of who's actually liable? Is it the manufacturer of the automobile, that didn't anticipate some edge case? You're kind of trying to figure out where the risks stops. And I think we're going to have a same issue with AI innovation in healthcare. Which is who's really responsible.
Yeah, it's a great corollary. I'm going to run with it and I'm going to say that that's the big sea change that never was able to happen. The self-driving car in medicine is you, We need to build AIs that work for the patient. Today we built AIs, we’re so early that are institutionally oriented. So really I want an AI that's kind of a virtual doctor and that I can think with. Because my doctor, I happen to be in a zip code in a part of the country where the care is a little bit different. But by and large, doctors get 4 to 5 minutes in a ten by ten room with a patient and that's it. And maybe a phone call with their back office. By the way, no fault of the doctor. They're incredibly busy. But today you can actually use AI to say, what if I wanted to manage my hypothyroidism with holistic care or functional medicine versus starting with pharmaceutical medicine or what if I wanted to pursue this pharmaceutical over that pharmaceutical, and that's where patients should be able to self-drive. So the first part of the problem was that they never had access to their own data. Companies like yours have solved that problem. That problem is now solved. You can actually get access to your own data, which is a phenomenal change. The second part of the problem is being able to process the data and that problem is solved with AI. So your self-driving example is a great one. In terms of the liability. I think liability goes away because when, oftentimes liability exists, when you give agency to a steering wheel that's moving in the car and nobody's behind it. That's where liability happens. And I think we're close to putting the agency back into the patient. So patients could have their own agency that didn't exist before. And that's really what the most exciting part of this whole time is that we're living in.
I love the patient centricity. That's like near and dear to everything I've worked on. I love that idea like I'm the self-driving vehicle. If you imagine that we don't have enough doctors, we don't have enough nurses, we don't have enough hospitals, clinics, any, like we have a healthcare system that doesn't actually have the physical resources required to treat the population. And we have a runaway cost train like completely runaway cost train, which we just have to like, you don't even need the analytics to look at it. You just got to look at a bill that you get, which comes back to your house. If we could take a significant portion, I want to say large, but just a significant portion of the regular day to day healthcare interactions that are routine, that are just reading what we were talking earlier about a guideline like just what's in the guideline, how do I do this thing, which most people would need to call their doctor, talk to this office staff or nurse with and allow an AI just to provide the routine, preventive educational care that we all need by the way, I have questions all the time, I had to call my doctor. I'm like, I don't really know what to do. You're only there for the thing you absolutely need them for. So I can also imagine how boring it must be to be a trained physician and answered the same, because we all think we're unique, I think I'm the most special person in the world, honestly. Right. So I think whatever's on my mind today is the most important thing. But I guarantee that the same question comes up over and over and over again. And at some point, these poor healthcare practitioners have just got to be like, I don't ever want to hear that question again, I just don't hear it. This is why I came in. I really worked hard in this industry, starting in the first ever specialty pharmacy, TheraCom that came out of the Cystic Fibrosis Foundation in the mid nineties and moved over to Dr.Evidence and other things that I've done where what you're talking about is the learning capital. I want to say two big things about this. One is the problem you just described is the learning capital, we found that health literacy is completely nonlinear to regular literacy. That is, that people that have actually the more literacy, this is an anecdotal fact, but the more literacy patients tend to have, the more they turn over agency to the system. So you find that people that don't have multiple degrees or advanced degrees oftentimes their health literacy, that is the literacy, their disease, what works, what doesn't work is generally very strong. So you might think about that as learning capital. That learning capital doesn't get shared patient to patient. That's kind of circles back to our early discussion, whereas now you can have AI do that. Every time someone gets exhibited to a disease or gets diagnosed, they immediately would benefit from meeting tons of patients like them. You could do that with AI because you can start at a lot higher rate, think about it as S curves, problem curves. Everyone starts when they get diagnosed, they all learn the same thing early on in their disease. What if we lived in a world where you could start with the best knowledge in that disease? That's what AI enables us to do, number one. Technology really has three big phases duplication, enhancement and transformation. So take electronic medical records. We used to write them in a chart, it was called the One-Write System, where a doctor's office would have these medical things or these metal things, and they'd write it and they'd go down to three or four papers. We have electronic medical records, so that was a huge transformation. Then we were able to read the record a little bit, get patient notices, reminders, medication notices, show up at the office a certain time. That would be more like, you're getting to enhancement. These are small examples of enhancement. Transformation would be, hey, you're being treated for this. There's this other treatment that has phenomenal outcomes. So it's constantly looking at data. And imagine if we were able to access all that data inside of pharmaceutical firewalls that I mentioned. It would be completely transformative. And that's the stage we're not at yet. That's the stage we're going to get to. We'll get there five, six, ten years. We should really get there in a metric of months instead of years. Because we have everything in front of us to get there. When Avatar was made it was said that that he wanted to make that movie 15 years before. But the tools didn't exist to make it. So the vision existed 15 years before the movie was made. And then finally, you had all the processing capability to make that phenomenal movie and it was made. So that's where we are, I think, in medicine right now.
So you just answered, and maybe we'll just end with this here. The question that I get, which I'm still waiting for somebody to be able to produce the system that answers. And I hear this in various forms from every single patient that I've ever talked to, myself included. And it's a really simple question, it’s really obvious, which is what did people like me do in the same situation and didn’t work? It just didn’t work. So tell me what other people did. And it's like a restaurant review, or anything else. I just want to know, did this lead to a good outcome or not? Because I need to make that decision. And I agree with you. We're so close, Bob. I think we can see all the pieces now. Now it's just a question of putting it all in place and making it a reality.
That's right. And it is perfectly said, I mean, when you have anything from little surgery or discomfort to something major, you start meeting people and the system is pretty efficient. Your friends, your family say, you know, we knew someone that had this, why don't you talk to them? All of those stories can go into AI. It's a phenomenal place that we're at in terms of transformation. Primary way that if you look at patient advocacy, navigation groups and all of that other stuff is patients just sharing anecdotally their experiences with things like side effect management, best doctor to go to, who knows what, like all of this stuff. And we have a chance to do it.