In a world of increasingly complex cancer care, how can oncologists ensure they have the most accurate, up-to-date information at the exact moment of need? For years, life-saving clinical guidelines have been buried in dense, hard-to-search PDF documents, creating a significant barrier to evidence-based decision-making. This episode dives into a groundbreaking solution that uses AI for oncology guidelines to solve this existential problem. We're joined by Dr. Clifford Hudis, CEO of the American Society of Clinical Oncology (ASCO), and Aashima Gupta, Global Director of Healthcare at Google Cloud, to discuss their revolutionary partnership.
They reveal the inside story behind the ASCO Guidelines Assistant, a new tool designed to provide instant, reliable answers to complex clinical questions. Discover how this powerful tool is not just a search engine, but a sophisticated system built on the principles of trustworthy AI in medicine. This is a deep dive into how technology can transform patient care by empowering clinicians and making critical knowledge more accessible than ever before.
In this detailed discussion, Dr. Hudis and Aashima Gupta explain the core challenges that led to the ASCO and Google Cloud collaboration. For busy clinicians, the process of finding the right information within hundreds of pages of guidelines was becoming an unwieldy and time-consuming task. The episode explores how the ASCO Guidelines Assistant, powered by Google's Gemini model, addresses this by improving clinician workflow with AI. We unpack the unique technical framework behind the tool, including the use of Retrieval Augmented Generation (RAG) to ensure every answer is based only on ASCO's vetted guidelines and includes direct citations to the source material. This commitment to accuracy and transparency is crucial for building trust and ensuring safe application in a clinical setting.
The conversation also highlights the rigorous "expert in the loop" process, where a panel of oncologists continuously evaluated and refined the AI's responses to ensure clinical relevance and safety. Dr. Hudis emphasizes a key feature: the tool's built-in "humility," meaning it will explicitly state when it doesn't have an answer within the guidelines, a deliberate choice to prevent the "hallucinations" common in other generative AI models. Looking to the future, the discussion expands to cover how AI will reshape all of healthcare, from unlocking new frontiers in science with initiatives like AlphaFold to restoring the doctor-patient relationship by automating documentation. As Aashima Gupta notes, the era of AI is here, and "doctors who use AI will replace doctors who don't use AI."
About Our Guests:
Dr. Clifford Hudis is the CEO of the American Society of Clinical Oncology (ASCO), the world's leading professional organization for physicians and oncology professionals caring for people with cancer. He shares the story of why ASCO saw an existential need to innovate beyond traditional PDF guidelines and partner with a tech leader to better serve its members and their patients.
Aashima Gupta is the Global Director of Healthcare at Google Cloud. She provides insight into the technical architecture that makes the ASCO Guidelines Assistant a reliable and trustworthy tool, explaining how the collaboration exemplifies Google's vision for applying AI to solve critical challenges in medicine and improve patient outcomes.
Introduction
Dr. Trevor Royce: Welcome to HealthTech Remedy. This is the next installment of our bonus series on innovators, where we sit down with individuals who've stepped beyond the clinic to become leaders and innovators, usher in new areas of health technology, and shape the future of care delivery. I'm Trevor Royce. Join me by my co-host, Tim Showalter.
Dr. Tim Showalter: I'm Tim Showalter, and in this series, we're exploring how physicians are applying their medical knowledge in unconventional ways, from building companies and products to driving change as ecosystem leaders. And bonus for this one, we actually have a technologist with deep healthcare experience as well.
Dr. Trevor Royce: Today, we are without our third leg of the stool, Paul Gerrard, which means all our jokes can be at his expense, even more so than usual.
Dr. Tim Showalter: Well, Trevor, before we hop in for the interview, maybe do you want to give a little bit of information to our listeners about ASCO, the American Society of Clinical Oncology?
Understanding ASCO: The Heart of Oncology Practice
Dr. Trevor Royce: Yeah, definitely. This episode is a little different in that we're going to be interviewing Clifford Hudis, who's the ASCO CEO, and he has led many innovative efforts at ASCO, and they've impacted the world of oncology in the U.S. broadly and across the world as well in different ways. And so I think he's a great individual to interview on the topic of innovators and he has enormous impact.
And so what do I mean by that? Well, ASCO is really the leading professional society that represents oncologists. And I think indirectly by representing oncologists, the patients that we take care of. Most folks know them of their annual meeting where it's the largest oncology conference, as I know it, in the world. They also have annual meetings in other locations and countries, but the main one is in the U.S. once a year, where it's this pilgrimage for those interested in oncology, physicians, people presenting their science, pharma partners, everyone you can think of that interacts with the oncology ecosystem comes to the annual meeting for ASCO in Chicago every year.
So that's one example of how they have impact. They also, of course, publish several different academic journals, some of the most prestigious journals on the academic side in oncology. And so if you're an oncology professional, you know ASCO and it has a lot of impact on your professional life.
Dr. Tim Showalter: It's hard to explain if you're not an oncologist, to explain how impactful ASCO is. They're the go-to society in terms of oncology research. They've got a whole arm that funds fellows grants as well as from faculty principal investigators. They're the main advocates for oncology, at least among practicing physicians. It's not just general ASCO meetings. It's also these sub-meetings for GU cancers and GI cancers and a whole bunch of others.
Not only is their journal, the Journal of Clinical Oncology, one of the top cancer journals, but they've got other ones that are focused more on clinical care and even their newsletters get a lot of readership. So I think they're really at the heart of oncology. And interestingly, we'll hit on this a little bit in our interview, but they've also made a lot of innovative tools that leverage technology to bring to bear for oncology. And in that vein, I think they're really leading the way. So it's not that surprising to see them really pushing things forward with the use of artificial intelligence as well.
Dr. Trevor Royce: Yeah, it's definitely worth spending a little time as we're doing to reflect on how ASCO impacts healthcare delivery and oncology specifically, for the technologists that are listening and may not be so familiar with ASCO. They also do a lot of advocacy work, things on the billing side, how do physicians taking care of patients with cancer get paid for their services and what policies would impact providers and their patients. So their headquarters is in the DC area and they create things like guidelines, which we'll hear about later. So it's just a very influential professional society that does a lot of good.
Dr. Tim Showalter: Yeah. And I'm really admired your humility, Trevor, because you have not mentioned this yet. However, it's important for our listeners to know that they are in the presence of a FASCO, a fellow of ASCO. So congratulations to our own Trevor Royce for that distinction as well.
Dr. Trevor Royce: Thanks, Tim. Yeah, that stands for fellow of the American Society of Clinical Oncology. And I should as a disclaimer say some prior work with ASCO when I was full-time on the academic side was as a health policy fellow where we spent a year at the office in DC working on these kind of sticky oncology policy issues facing oncologists and the patients. And for example, we worked a lot on pharmacy benefit managers or PBMs, which has become one of the hottest topics in oncology and health policy even broadly in the U.S. over the last probably five years. So they do a lot of good work on that area.
Dr. Tim Showalter: Well, great. And I think we're really excited to hear about the work that they've done in partnership with Google Cloud to really bring artificial intelligence to clinicians for accessing insights from the ASCO clinical guidelines, which are considered a real authority within the oncology space. So Trevor, before we hop over, any final words before we start the interview?
Dr. Trevor Royce: I think it's going to be a great conversation. I'm not surprised to see ASCO contributing to this new era of large language models and how they can directly impact care. So it'd be really cool to hear their perspective and see how they've tackled this problem.
Dr. Tim Showalter: Today, we're thrilled to continue the series with Clifford Hudis, CEO of ASCO, and Aashima Gupta, Global Director of Healthcare and Google Cloud.
Dr. Trevor Royce: We're really excited to have you guys on today and looking forward to the conversation. I think everyone's probably still riding high off of the ASCO annual meeting. And I think we're going to cover a lot of themes that were present there. But just to really start from the top, tell us a little bit about what's bringing us together today.
Introducing the ASCO Guidelines Assistant: A New Era for Clinical Information
What is the ASCO Guidelines Assistant? How did this come together and what problem is this new tool solving for?
Dr. Clifford Hudis: Well, thank you very much. Maybe I'll start and then Aashima can actually tell you how it got done. But I will share what led us to this collaboration. That is to say, what problem did we see from an ASCO point of view? And in order to do that, I want to go back just a little bit. For the last few years, our members have, I think, increasingly been looking for high-quality, reliable guidelines and guidance in terms of the increasingly complex treatment of cancer. We have more and more specific targeted precision therapies. We have more and more subsets of populations or tumor types, subtypes, where you really have to know what you're doing in order to get the right therapy to the right patient at the right time.
ASCO for years has generated what I think are widely respected, high quality, painstakingly constructed, evidence-based guidelines. And they are put out into the world historically in the form of essentially a PDF. They are publications in the Journal of Clinical Oncology most of the time. And they're highly cited and they're very, very popular.
They are not easily accessible, and they're really not easily accessible in the moment when a busy clinician needs to just get an answer. So that's the backdrop. And I didn't say it plainly, but the problem we're really trying to solve was easy access to high-quality, trustworthy information at the moment of need.
And with the revolution that the last couple of years have brought in the form of generative AI and new tools being pushed out broadly across our society, it became, I think, painfully obvious that an immediate solution for us was some kind of use of those new tools to bring forward trustworthy, reliable, accurate information for our members. And that's really what brought us to Google Cloud, because they were, from our point of view, the people who, A, clearly had the skills to do what we needed, B, understood the problem we were trying to solve, and C, were, I think, able to not just do it, but to lead us as more than partners and collaborators at getting this done in a timely way.
We'll talk, I'm sure, a little more about some of the special features that we built in to accomplish this, but that's really the backstory. I'll sum it up more simply by saying to me, in the end, this became almost existential. I don't think historical guidelines like we would generate will be relevant too much more into the future. The world is going to demand something like what we have started to deploy.
Dr. Trevor Royce: That makes a lot of sense. And I'd love to hear from Aashima on this, but another follow-up question is, how do you think that physicians access these guidelines now? You described it as burdensome or becoming unwieldy. What does the past look like before we get deep into the future?
Dr. Clifford Hudis: The past was that you went to the JCO or you went to your favorite search engine, let's just say Google, and you look for the document and then you pull down the PDF and then you search through it for the subset and section that you thought was relevant to your acute issue. That's how we did it in the past.
Dr. Tim Showalter: I was reflecting on how quickly things have changed over the years. And I finished residency in 2009. And I actually remember the days of going to find the guidelines in the stacks of the library and photocopying them and having a notebook of guidelines that I would refer back to and bringing JCO and so that all the articles are electronically available and then ASCO having a great website that is very member friendly and having everything organized has been a step forward. And then now I think it's amazing to see how this technology advance can move us even further forward and maybe that's a good point to pass it over to Aashima a little bit.
I'm curious to hear about maybe starting with the first reach out from ASCO or how this project started to materialize. And would love to hear about how this particular use case fits into your vision for Google's healthcare contributions and maybe a little bit of the story of this collaboration coming together.
Aashima Gupta: It's a pleasure to be here and this collaboration is a direct outcome of a shared strategic vision. From a technology perspective, Google Cloud, we have a strong belief that technologies like cloud, like AI, tooling, if given to the hands of experts around the globe, can actually lead to much better patient outcomes. So this is a deep partnership for what was really exciting was when we met with Dr. Hudis, he was explaining how these guidelines are in PDFs. These are hundreds of pages long guidelines. This is a problem that AI is actually good at. And for us, it was exciting because it was an incredibly significant use case. It serves as a practical example of how our technologies like Google Cloud or Vertex AI and Gemini model can be applied to a highly sensitive and critical areas like oncology.
There are many, many use cases in the world of AI and healthcare. And the examples like this, the practical examples like this, means that where AI will be really, really impactful is creating that space for oncologists to take care of their patients. If you can take some of that rote task, some of those things that take away the joy, like searching for guidelines, reading 100 pages of documents, it's hard. It's hard for anyone. But imagine you're hard-pressed for time as an oncologist. And so this is one of the consistent themes we see in healthcare, the efficiencies lead to better outcomes. This is a strong belief we have.
And from the technology perspective, it's always fun to unpack the engineering foundation behind tools like ASCO Guidelines Assistant. So from the story perspective, it's a really amazing use case. It was ASCO's leadership in creating those guidelines. They have hundreds of them, and they are in PDF. Can you make them findable? Can you lead them to our Gemini model, but do it in a safe, reliable, and trusted fashion? This is where Google Cloud shines in. And I'll share the technology underpinning behind it. But to me, that was a great combination of leading cancer organizations, leading leadership with ASCO combined with deep technical. That's where the magic happens.
Building Trustworthy AI: The Technology Behind the Guidelines Assistant
Dr. Tim Showalter: Just a follow-up question. I'm really curious. I think all of us have used Gemini to plan vacations or general use cases. I've certainly even gone to Gemini's general search function for medical queries as well, but I'm really interested about this particular example. Many of us are familiar with what's available to the lay public. Can you help unpack for us a little bit what are the unique technical challenges to standing up something like this that is really solely based on the ASCO guidelines and it seems like it's a more cordoned off feature.
Aashima Gupta: So the first and one of the most persistent concern we heard from the ASCO team was, there are a lot of answers out there. But what ASCO stands for is the most reliable, most trusted information. So for us, it was to, I call it in the AI is a black box problem. It directly affects the trust, the accuracy. Inclusions are actually cautious about accepting recommendation without understanding the underlying rationale. To design that, we need to really harden off, as you said. And there's techniques in technology like retrieval, augmented generation.
So what we first ingested, we worked with Dr. Hudis and his team in ingesting ASCO's rigorously vetted guidelines. That's the start. It's a gold standard. It's a vetted guideline. We converted them to vector representation. Vector representation is what Gemini model represents. It directs the model to generate the response based on this trusted source. So every suggestion includes a clear citation so clinicians can see exactly where the information came from, understand the reasoning, and verify themselves. I think this was one of the key requirements from the ASCO leadership team. Dr. Hudis, I still remember the day when we were showing you and you had said, do I need the yellow highlight in the guideline assistant? If you've seen the two, you will see as the guidelines are pulled, they're shown on the right inside as highlighted.
Dr. Clifford Hudis: Yeah, I would pick up on that and say one of the unique things about this, it's not the same experience as using Gemini in the commercial public marketplace. And the reason for that is that I think that our community, A, our clinicians are held responsible, rightly so, for every decision they take. And so a generative answer, we know, at least in the first few years of this era, sometimes will have mistakes in it. And the clinician, I think, rightly has to be careful about that.
So in use, I think something that for people that haven't used it, they'd have to go see it, is the fact that 100% of the citations, the references in an answer in the guideline assistant have a reference tag, and that tag has, as Aashima points out, highlighted text. And I would say it's a clinician's job to look at that right-hand panel, to see that source, and make sure what it's saying there is concordant with what the summary, if you will, the AI generated answer is.
And I think this is important because if you really go beneath the surface, this era, I believe, is an era where everybody will be expected to do better. In other words, everybody has access to all this information. Everybody should know it. And I think this is really a continuous learning opportunity for people to be reading up-to-date guidelines, seeing the source material, and then applying that to their individual patient. So it's a little different than the typical ChatGPT answer that people are probably getting on their iPhones right now. It's more sophisticated. It requires a moment more attention, but it gets you to the source. That's the point.
Dr. Tim Showalter: That's a really interesting direction because I think what you're touching on is how is all this technology going to improve patient care? And maybe actually level the playing field for medical knowledge among oncologists. But before we do that, I had one more question. So I logged in and used it on my desktop, through the ASCO member portal. But maybe, Aashima, before we step away from some of the technical considerations, I'm curious about how you're envisioning clinicians accessing this information. Is it also going to be available in, say, mobile format or what else is available?
Aashima Gupta: So absolutely, we are starting with the web and there's a mobile app as well. I think from the technology perspectives, to me, those are surfaces. The magic happens beneath the surface. So one is the interface where clinicians can access through a website, through a mobile. But imagine in building that where we have invested significantly both from the ASCO side and the Google side are a couple of other technical concepts that I want to bring forward. And this is what was so unique and incredibly inspiring about this. We call it expert in the loop framework. Any answer that is being generated by this garden of Gemini, we had to build the evaluation framework. It didn't exist. So we worked very closely with Dr. Hudis and his team, and they actually gave us access to oncologists.
The whole panel, who were evaluating each and every response in multiple different dimensions. So it is about that close collaboration. It was not just about building a tool and walking away. This is rigorous feedback and refinement process where oncology is actually related each and every response. Like, what is it missing? And every Friday, we would have a meeting with Dr. Hudis, and it was very inspiring for me to see even Dr. Hudis being so hands-on. And we call it expert in the loop. In the machine learning, we call it human in the loop, reinforcement learning. But this was actual experts telling us what is the answer missing so this is the ongoing cycle of testing, learning, and improving, and oncologist and technologists side by side that's what made this possible from that expert in the loop from the evaluation framework perspective.
And one final one is when you're putting a solution like this of a clinical domain and this was high pressure, 40,000 plus oncologists accessing that and it went live. One of the things that we had to pay close attention to was the adversarial training techniques. So we dedicated enough time, we call it break the solution, ask it tricky questions so we had to have this new engineering behind. It's not new, but it's called adversarial testing data sets where you take model and not just ask the real simple questions, you ask it complicated questions and the focus is around robustness and safety.
We intentionally try to break the model by integrating these adversarial data sets, meaning we didn't just train on clean representative data, but we also fed the model pre-perturbed adversarial inputs and designed to similar real-world edge use cases. And see, can it still perform? Because safety is critical. So not just from the RAC technique we talked about, but this adversarial testing and this evaluation framework. It takes that level of collaboration, deep collaboration to make something like this, which has the kind of implication for oncologists around the world to bring it to life.
Dr. Trevor Royce: That's super interesting. Maybe take a step back and just make sure we don't lose sight of the forest, because there's some real key themes worth highlighting that we've heard so far. One, as you just touched on, was the trustworthiness of information and really how you guys have gone above and beyond to make sure these models are bringing accurate, trustworthy information. That's number one. Number two is the speed of access, which we heard earlier. We've gone from an era where you either had to go to a library, get a physical journal or have a printed out PDF. And now you're going to have this basically at your fingertips.
Three was the timeliness of information. And I feel like we've almost undercovered that in a sense, because for me, when you're thinking about many different tumor types and you pick your guideline of choice, but you go and you look it up and all of a sudden you see that there was an update last month and you had no idea. And then all of a sudden the guideline looks a little different than it did before, but this kind of cuts through all that noise and it'll bring an accurate up-to-date guideline right to the clinic. And then the evidence-based decision-making, I think it was probably the last core theme that we've heard about where you've got direct citations that are supporting these guidelines and how they were developed. So maybe I'll just pause there, see if there are any kind of core themes that we've missed or if you want to elaborate on any of those before we go on to some other topics.
Dr. Clifford Hudis: I want to just pick up on that because it's a chance to highlight a very specific feature, which we learned as we went along. When we started, we thought if we had a tool, we had to be able to answer questions about how to treat patients. But we quickly realized, or not so quickly, realized that really what we needed to prize was trustworthiness. And that meant building in humility. So a unique feature of this interaction, I think, for most people is that it will plenty of times tell you, ASCO guidelines do not have an answer to your question. As opposed to creating an answer or reaching out or taking lower quality sources. And that's the safety ring fencing that I think required a little bit of a shift in view. So we're not going to answer every question. We were only going to answer the questions that are answered within ASCO guidelines. And I just think that's a critical aspect of both reliability and building trust. And therefore, your use of it is a little different.
Dr. Trevor Royce: It reminds me of the old adage where you can't believe everything you read on the internet. And I think about that all the time with commercially available LLMs, but here maybe actually you can believe what you're reading on the internet because of all the effort that you guys put towards that and the humility, as you put it.
Dr. Clifford Hudis: Well, we tune it towards facts and away from creativity.
Aashima Gupta: It took a very deliberate step. I still remember that meeting that we were discussing, should we answer the question, even if the model knew, but hallucinations are a big concern. So we take a very deliberate decision and that humility, we will say no, or this guideline tool will simply respond and, to me this is where the leadership matters like how advanced are you going to push that these tools are super capable these are general purpose tools so what are the guardrails we will put on these solutions so that they can be deployed widely. I think that there's that thoughtfulness that matters.
Early Reception and the Drive for Continuous Improvement
Dr. Tim Showalter: Something that Dr. Hudis said earlier about the new demands that are expected from physicians in terms of really using these tools and understanding but and also checking the references. And I want to save that because I really want to spend some time thinking about the future. And Aashima, I would love to hear your perspective on this too, in terms of how you as a technologist see this evolving. But before we leave the present, I would love to ask what you're hearing so far from ASCO members. What's the reception then? What are you hearing back for the impact that it's making in the clinic?
Dr. Clifford Hudis: Our community is rightly cautious about new tools. And I often remind people that when I started at ASCO as a volunteer, I remember having to convince the Internet Services Committee and then the meetings people that we should have Wi-Fi at the annual meeting. There was a fear that if we had Wi-Fi, people would be distracted, believe it or not. And I think it goes to something we touched on earlier, which is ultimately, even in this modern era, the clinician with the real or virtual pen in their hand bears full responsibility for the decisions made, the treatments given, the interventions deployed. And I think that gives people a real pause.
So the uptake for all of this is, I think, still relatively in the experimental range. People are trying it out. They're playing with it. They're seeing how satisfactory it is. The experts that know everything about the disease are asking it their high-level expert questions. And we're getting mostly positive reviews. We're getting, and I think, Aashima, you referred to this before, the system is built with essentially continuous feedback. Every answer, every time we're asking people, please tell us what we got right, what we got wrong. We're not suggesting that this is a final version. This is an iterative learning experience for us. But they're mostly positive.
And if I'll pivot a little bit and say the most actionable feedback for us is one I realize you alluded to before, but I didn't follow up on. And that is that the timeliness of the content is now under scrutiny. Because what this shows very quickly is that maybe there's a guideline that we haven't updated since 2019 or 2020. And it might have been just sitting there read occasionally before, but now it's being surfaced and presented to people in 2025. And they're saying, well, wait a second, why aren't you on top of this? And so for us, the good internal challenge that this generates is a requirement for us to update and revamp our entire approach to generating and maintaining guidelines. And that's going to be a service to the community as well.
Dr. Tim Showalter: That's really interesting. And just for context for the listeners, I don't know exactly what ASCO's policies are about iteratively updating guidelines, but it's typical for professional societies that issue guidelines or guidelines panels to update on a cadence that's measured in years. It's usually not, sometimes it's annual, but oftentimes it takes a lot of work and you have to get a lot of people together and you have to have a meaningful change in the evidence to issue an update. So sometimes it's every two to four years or these are reasonable time intervals. And so that's really fascinating to your point about the new expectation that things are updated in real time and people are probably checking this a little bit more. Well, maybe we should transition to the future then.
I'm really curious to hear your thoughts first, just at the maybe ASCO and Google collaboration level. Now that you've got this program together and you're seeing some success from this particular tool, are there other things that are in scope potentially to take on for new contributions to the field?
Dr. Clifford Hudis: The next step for us, as opposed to taking our existing guidelines and making them available through the tool, is to expand the reach and improve the timeliness of those guidelines. That's basically the challenge. And we are continuing our collaboration along with our publisher, Wolters Kluwer, looking at how we can modernize guidelines generation using some of these same techniques and approaches. And I think in many ways, that's going to be the really biggest part of that.
Aashima Gupta: Well, another fun thing we have talked about is we have technology like NotebookLM, I don't know if you've used it. It creates a podcast conversational voice of any artifact. So give it a PDF or a long guideline. Can that be a much more acceptable format for you to listen to the guideline? So we're exploring multiple different ways how this contact can be absorbed. So most of that will come from the phase two thinking.
The Broader Vision: How AI Will Reshape the Future of Healthcare
Just by taking a step back from where, and I know, Trevor, was that your question about the future, how we are thinking about applying AI in healthcare. We see three big ways in where AI, there are a lot of hype, a lot of use cases, but to me, three big things stand out for us. Number one is access. We all, as Dr. Hudis talked about, global health equity and access is a big concern. And we believe AI will be a real game changer here because one major area is expanding access to even late patients, not just in the context of the ASCO, but think of this, you're diagnosed with some condition, you want to find out helpful response. And many times, it's a very broken, fragmented experience.
So imagine can we help transplant patients navigate their complex care journeys? Cancer is complex and emotionally overwhelming. So imagine conversational health 24 by 7 health companions, where this is one example of that is with American Cancer Society. They launched Anna, which was an assistant. You can ask, oh, I was diagnosed with melanoma. Can you tell me more about it. It's multimodal. So you can converse about your complex condition. So we believe AI will get to the point where each one of us will have a companion, not like today, but three to five years from now.
Imagine having that same level of concierge 24 by 7 health that can help you navigate. The second big area, and Dr. Hudis touched on that, the speed of access. It is critical in all clinical workflows. There are much rote documentation. We hear that for every one hour of patient care, there's two hours of clinical documentation. So when you talk about affordability, can we leverage AI in those rote tasks, eliminate some of the rote work the clinicians are doing today, like documentation, writing, searching?
So, we believe as the rising cancer rates, so growing number of older cancer survivors and the explosion of really new treatment options, while we're facing the shortage of clinicians worldwide, there's a 10 million shortage by 2030. So how do you create more capacity? So can AI be that assistant, a helper to a clinician for better care? Navigation for, I think that's where the efficiency will come in. So for affordability, much of that can be really argument clinical workflows with this technology.
Access. And third is science. We are investing in the future of science. Initiatives like AlphaFold and AI co-scientist are unlocking new frontiers in drug discovery. So just to give you a point on alpha fold it protects the structure for protein normally takes a scientist four years for one protein structure these are building blocks many of the drugs and new therapies are protein based as well now our team in deep mind they took 200 million protein structures and made them open source so imagine this four years for one protein now we already have that 200 million protein structures available for research, for science. We have large AI co-scientist. This is, again, augmentation of scientists or researchers. Imagine them having a PhD-level assistant, not one, not 10, maybe 100, maybe 1,000, helping them create hypotheses, doing testing. So these are the three areas of affordability, access, and sciences where we believe this next three to five years will be truly defining in healthcare.
Dr. Trevor Royce: Yeah, that's amazing and inspiring. AlphaFold has been such an incredible example of how Google has been able to push the field forward. I think it's safe to say that basically every aspect of cancer care will be touched by AI at some point in the future. That's the key takeaway of what you described there. On the ASCO side, I feel like ASCO has been an innovator in oncology for a long time in a lot of different areas. Tim and I both have experience in real-world data, and ASCO was an early innovator there as well with CancerLinQ. There was the TAPER trial, mCODE for data standards, and then quality standards. Also on the research side, of course, with things like the Conquer Cancer Foundation. Dr. Hudis, any other areas that the membership should keep an eye on in terms of innovations in oncology that are on top of your mind for ASCO?
Dr. Clifford Hudis: Well, first of all, you've read out a great list of places where I think we were fortunate to see where things might be going and what the needs of our membership might be. We built CancerLinQ with a big Q at the end because of the emphasis ultimately on improving quality of care. And we're excited by the fact that it has been purchased, as you know, and is a commercial entity now running and I think delivering nicely on that original promise. TAPER isn't in the past tense. TAPER is very much thriving to this day. And the key point there is a subtle one, and it is that the TAPER trial puts into action many of the real-world efficiencies that ASCO has lobbied for going back at least a decade, meaning efficient clinical trials that are much less expensive to run and can get answers faster, lower cost for the whole system.
And that, of course, is a broad benefit. It also minimizes and simplifies data collection. The mCODE data standards, which are expanding, I think represent an important contribution that really came out of our experience with CancerLinQ and the hodgepodge that real-world data was. And I know you're both very familiar with that problem.
Going forward, I think that there is an opportunity not fully realized yet, leveraging the electronic medical record and increasing either standardization or interoperability to make quality tools all digital, all simple, all cheap, and more relevant and up-to-date as they are replaced by new ones. I think that's going to be an area for great growth. But to answer your question about the membership, I think the key thing for us, and I've touched on it a couple of times is we need to support our members and educate them and help them take those first and early steps into this inevitable new world.
And they are rightly cautious. A lot is riding on it. And I think our opportunity, which is exemplified beautifully in this initial rollout of the guidelines assistant, is to show them, in a sense, why what we're developing and what is being presented to them can and should be trusted. Because I think, and Aashima described this beautifully, we are absolutely dependent on gaining significant efficiencies over the next few years for our sustainability. And the only way we're going to do that is by a complete embrace of these new tools.
Advice for the Next Generation: Embracing AI in Medicine
Dr. Tim Showalter: I just wanted to ask a couple advice and future looking questions, maybe one for each of you. Cliff, maybe we could start with you. I'm thinking about as you were talking about the use of these guidelines among members, I'm sort of picturing a medical oncology fellow, studying for the boards or something and downloading the information quickly. For the fellows out there who are in their training right now and who have their whole career ahead of them in the world of AI, what sort of advice or vision would you offer to them, guidance for entering their careers?
Dr. Clifford Hudis: So I'm going to wax philosophical here because I think all of this is peripheral to the job of doctoring.
There is a beauty in the relationship that we are privileged to form with patients and families. I've been often asked as an old man why I went into oncology. And yes, the science and the opportunity to be on the leading edge of a number of developments, that was exciting. But the real motivation, which I think is enduring for people, is this privilege, like I said, of helping people cross one of the scariest chasms they face in their life. The moment somebody's diagnosed with cancer, getting to the place where they can have hope about the future, either for cure or for a high quality of life. That's what we do across oncology.
I think technology is going to restore some of that because I think the last 20 years, if nothing else, have been marked by an increasing intrusion of distracting technology and intrusive technology that got between the doctor and the patient. And the stereotypical image of the doc with the nose in the screen and the fingers on the keyboard back to the patient filling out an online checklist, I think we're right at the point where that's going to be flipped. And I think that ambient listening and scribes and all these tools that are just coming into the clinic now hold great promise to allow the patient or rather the doctor to roll that chair around.
Face the patient, hold a hand and really be the emotional support as well as technical guide through this scary thing. And I think optimistically that we're going to be able to do our jobs better than ever before because of this. I don't want to sound like a Pollyanna. There are lots and lots of challenges in healthcare now, but I think I've oversimplified this and I'm sorry to carry on, but I watched all this carefully. I don't know if any of you are old enough to remember the blue screen of death.
But in the early era of standalone PCs, when we all got them, we all faced the loss of our work, the hours wasted and all of that. And really, I may be oversimplifying the story, but the world was revolutionized ultimately by the LAN and networking. And the long-sought productivity gain that wasn't described in the first decade or so of the PC era suddenly materialized. Nobody talks about that anymore. I think that's a model for the moment that we're probably in right now. I really do think we're finally turning the corner. And I think this technology is a reason for a young trainee to be excited and optimistic. I'll close by saying not only will they be able to actually spend more quality time with their patients, the care they're going to be able to deliver is going to be higher quality in technical ways than ever was imaginable before.
Dr. Tim Showalter: Thanks so much. And I think so much of what's exciting about the positive changes for healthcare really are derived directly from technology. And Aashima, you've got a tremendous technology background, but you've also got years and years of healthcare experience. And so you've got a ton of knowledge in that space. And I'm curious what you would offer to maybe a technologist or a young person entering the world of health technology in terms of how you see the future unfolding in the next, say, 20 or 30 years, how someone might plan their career out to make an impact?
Aashima Gupta: That's a great question. I think I'll have advice for both physician and technologist, because this is where the both, in the next three to five years, there'll be much, much tighter collaboration. So from the physician perspective, I, in my role, I have a privilege to work with many clinicians. So my one advice for them would be just embrace AI. It's here, it's not a fad, it's a truly step change in technology so if you're not experimenting with AI because with your involvement will this technology become better so please embrace it these tools are to augment you, to not replace your clinical expertise.
So leverage them to scale your expertise and scale your reach. And one thing we do say is doctors who use AI will replace doctors who don't use AI. So it's not AI is going to replace the doctors, but the doctors or clinicians, no matter what your practice is, who use AI, will replace doctors who do not use AI. And also, I think this is calling. We are running the most transformational era of our time. I'm a computer scientist. I've never seen the pace like I'm seeing now. This is truly different. And to me, their feedback will be essential to ensure AI continues to support and not disrupt the practice of medicine. So this is not the time to be on the sidelines. Please embrace AI.
And from the technologists, same thing. Technology is an enabler tool. It's not the be-all and end-all. So working with clinicians, doing it with them and not to them. We have several examples in healthcare where some respond people come together and we truly being in the shoes of clinicians. And I think that with them and not to them is the mantra to make any healthcare technology more useful in the clinical setting. Clinical workflows are complex. So getting that and working alongside this combination is needed.
Dr. Trevor Royce: Absolutely. And we've heard this from both of you, but sometimes it's nice to be just totally explicit. But from the technologist's side, there is no greater mission than helping people that are sick or have this terrible diagnosis of cancer. And there are a lot of opportunities, a lot of different fields that are benefiting from AI. But if you're thinking about impact, it's hard to beat what you guys are working on. So thank you guys so much for both for joining us. And it's really been incredible to hear about this cutting edge collaboration and the new world of oncology and AI and how they interface. And it'll be fun to see what the future holds.
Dr. Clifford Hudis: Thank you very much.
Aashima Gupta: Thank you for having us.
Dr. Tim Showalter: Well, that's it for this episode of HealthTech Remedy. Be sure to subscribe and follow us on your favorite podcast platform. And if you know a physician innovator or a company that we should feature, reach out. We'd love to hear their story.
