Host
Heather Simonsen, MA
Public Affairs Senior Manager
Huntsman Cancer Institute
Taylor Smedley
Director, Visitor Experience & Community Engagement at The Leonardo
Jonathan Tward, MD, PhD
Leader, Genitourinary Cancers Center at Huntsman Cancer Institute
Professor, Department of Radiation Oncology at the University of Utah
Episode Transcript
Topics by Timestamp
- Welcome to the future: AI's immersive story (00:51)
- AI in everyday life: From data to diagnosis (04:49)
- Precision medicine: Tailoring cancer care (07:39)
- AI's role in prostate cancer treatment (14:04)
- The power of patterns: AI's predictive potential (18:06)
- Challenges and opportunities: The future of AI in oncology (20:39)
- Looking ahead: The AI revolution in cancer care (30:21)
Welcome to the Future: AI's Immersive Story (00:51)
Heather Simonsen: Hello and welcome to delivering a cancer free frontier. I'm your host, Heather Simonsen.
We're visiting The Leonardo, a museum in downtown Salt Lake City dedicated to the intersection of science and art. We're walking through an immersive exhibit. Projected gray and white images flash on the walls and on the floor of the 10,000 square foot space. At first, it looks and feels like static, then evolves into other shapes and colors, metallic blues and silver waves. This is just the beginning of the exhibit's story, the story of AI, artificial intelligence. An AI voice appears: “darkness, then slowly light lines of code merging into something way more complex.”
Taylor Smedley: It starts with the formation of data turning into an AI that's aware of itself.
Heather Simonsen: This is Taylor Smedley, director of visitor experience and community engagement at the Leonardo.
Taylor Smedley: You’re looking in the mind of AI. So, all of the things that AI might narratively think about. “What am I?”
Heather Simonsen: There are several chapters in AI's immersive story, each about five minutes long. One explores AI and customer service, then biotech
Taylor Smedley: Through some really beautiful visuals of microbiology and cells.
Heather Simonsen: Later, the images turned dark and disturbing. This is the dystopian world where AI has taken over, but the presentation ends on a hopeful note.
Taylor Smedley: With the open-ended question of utopia and what that looks like. So, a lot of solar power, a lot of green a lot of advanced societies and systems. So, it really uses these visuals to tell a story of the potential of AI and how it might affect all of our lives in an emotional, immersive setting where you're looking at the projections on the floor and on the ceilings, completely immersed in our idea space.
Heather Simonsen: We don't know which path humanity will take with AI, and neither do the staff at the Leonardo. Taylor says what they want instead is to get the public thinking about AI's place in our lives.
Taylor Smedley: We want it to be at the forefront of these discussions around AI, and as it is evolving ever so quickly. I'm sure that there's pieces of the exhibit that are currently out of date, even though we've only had it for a few months. And a lot of the public we serve, and guests we serve, and students we serve, it is their first interaction with AI. So, it's communicating in a less formal way of maybe taking away some of the fear around it, or just providing the opportunity to engage with it and learn about it and really educate our guests on its potential.
Heather Simonsen: Outside the immersive exhibit, there's also an educational section that explains where we are with AI on our shared journey, its history, its uses, its ethics, its place in pop culture. And very soon, there will be another display.
Taylor Smedley: We are working with Huntsman Cancer Institute to provide some educational pieces to the exhibit about AI, specifically with cancer. What I've been struck by is actually how much AI has already been used in cancer research is just kind of coming to public attention. That this is a way that doctors are enhancing their ability to diagnose cancer.
Heather Simonsen: Taylor says he hopes the cancer exhibit, which will open in January 2025, helps people understand how AI is already making a huge difference in cancer care.
Taylor Smedley: Or at least to pose the question of like, have you thought about cancer and AI? Providing the exhibit as a platform to get people thinking about these big topics that affect us all in an innovative way.
AI in everyday life: From data to diagnosis (04:49)
Heather Simonsen: On today's episode of delivering a cancer free Frontier, we also want to get our listeners thinking about cancer care and artificial intelligence. It's a big, unwieldy topic. So, called in one of our physician scientists to help us. Dr. Jonathan Tward is a radiation oncologist at Huntsman Cancer Institute and professor in the Department of Radiation Oncology at the University of Utah. I wanted to start our conversation at the basic level. I asked him how he defines artificial intelligence.
Johnathan Tward: I define artificial intelligence as programming that allows a computer to reason, learn and create, and it really mimics those things. It's not a human. It is not intelligent, per se, but the underlying programming allows it to behave as if it were intelligent. and it's really fascinating how AI has penetrated our lives. In the last couple years, it's become common to hear about artificial intelligence, but people have been working on artificial intelligence for many decades.
Heather Simonsen: Luckily for us, Dr Tward is one of them. He says to think of AI in terms of pattern recognition, just like we do in our everyday lives.
Johnathan Tward: When you think about how you might reason, it's similar to how software allows a computer to potentially reason. So, let's say you're trying to determine what clothes you should wear, if you should bring an umbrella when you go outside on any given day. The way you would do that is you would look at data available to you. In this case, you're using your senses, your vision, your ability to read documents and see things. And you might look outside and say, “okay, if it looks cloudy outside,” you'll make a certain judgment about the probability of rain; versus if it looks very sunny and there are no clouds in the sky, and you're taking that piece of information and drawing certain conclusions. Then you might also watch a news report, and there might be a forecast, and then you're integrating that piece of information as well to draw your conclusion. But what you're doing is you're taking lots of data in from various data sources, and you are processing that in a way that, based on your experience and past learning, allows you to conclude on whether or not you should bring an umbrella. And that is, quite frankly, very similar to how artificial intelligence software works. There's lots of different models out there, but something like deep learning, which you might hear about, is kind of an example of something like that, where you take large data sets, and you draw conclusions based on all of this information that you can process.
Precision medicine: Tailoring cancer care (07:39)
Heather Simonsen: Let's talk about AI and cancer, specifically, when did you first take interest in AI as a researcher? I mean, why did it interest you?
Johnathan Tward: I've always been interested in how computers can make our lives better, or how we can use computers and software to make our lives more efficient. And so, with regard to AI and cancer, specifically, I've always been doing manipulations and data with complex analytics that I wouldn't have called AI 10 years ago, but now we do call AI. How about that? But I think that even though I was using little statistical methods here and there on certain papers, I wasn't really focusing on what people today are calling deep learning. And when chatGPT came out, and I started paying attention to how truly remarkable and humanesque GPT was at not only interacting with me through a computer interface, but generating, you know, creative content. I thought, wow, this could really, I should focus more on this deep learning to solve cancer problems. And so, I shifted almost all of my individual research program into looking at how to make treatments, precision medicine approaches, as opposed to, let's call it, staging approaches.
The way oncology is always and still is approached is we like to lump people in boxes, and people listening this podcast might know that they're a stage one or a stage two or a stage three or stage four. And right now, even today, we still have guidelines that say, if you're a stage two, you do this thing. If you're a stage three, you do this thing. And the problem with that approach is you will over treat or under treat people, because people are individuals, they're not groups. And AI is very, very good at looking at massive amounts of data, and we accumulate massive amounts of data on people in our medical records and from other sources. And it can find patterns in those data that I hypothesized, and many others have hypothesized could lead to precision medicine approaches. And when we started turning our research efforts into using AIs to see if it could find patterns in this data that humans aren't as good at identifying and see if that could create precision medicine approaches. We learned that it's very, very good at that, and so it's very exciting.
Heather Simonsen: This is so fascinating. Can you define precision medicine for us? What is precision medicine?
Johnathan Tward: Yeah, it's interesting. I'm not sure there's any one definition, but my definition of precision medicine is looking at an individual and taking into account all the complexity of that person. And creating diagnoses or treatment strategies that you know are more likely than not to be effective for them, as opposed to looking at that individual as part of a group. A lot of medical practice is based on lumping people into groups as opposed to looking at individual features. Sometimes when people talk about precision medicine, they misconstrue that term as to looking only at genomic features, let's say or little molecular changes in their genome.
But to me, precision medicine approaches look at where do you live, what are your demographics, potentially, what are your genomics? It could be literally any piece of information you have on somebody, because those things might be relevant to how you approach them clinically, and also how they respond. We know that genetics matters. We know that social determinants of health matters. We know that your economic situation matters. We know that the amount of support you have at home matters, like all these things are related to precision medicine approaches, as far as I'm concerned.
Heather Simonsen: And so, could AI look at those layers differently than the human brain? Can it take all that information and put it together to help a physician deliver that precision medicine?
Johnathan Tward: I think so. Like what AI is particularly exceptional at is finding patterns that humans are not easily able to identify on their own, and it's because they can do so with such massive volumes of information, and they are able to also, for lack of a better term, think in dimensions that we can't think of. And dimensions are a really difficult concept to explain to somebody who is not deep into data science. But you know, we perceive our world in three dimensions. You know, there's a height, a width and a depth to our space. There's another dimension, for example, such as time, and we can characterize things in our world as being in that space. Computers can also characterize information in dimensions past that. Without getting too technical that allows computers to find patterns that are very challenging for a human mind to identify.
And so, we can exploit the capacity of these computers to do that, to identify patterns that we couldn't otherwise identify. And one of the things that's problematic, by the way, about computers ability to identify patterns that we can't routinely identify, is that we can't always describe what these things are to a human. And this frustrates people, because humans like to have an understanding of their world and what they're seeing. And a lot of what AI does is considered, for lack of better term, a quote “black box,” where we know it's finding patterns that we can't even describe to ourselves, and all we can say is we know that they're accurate at prognosticating something in the future or identifying something well.
AI's role in prostate cancer treatment (14:04)
Heather Simonsen: And you recently worked on a project using AI as a predictive model for prostate cancer treatment. Could you break that down for us explain how it works and its outcomes?
Johnathan Tward: Yes, I collaborated with a research group that had access to digital images of prostate biopsy tissue. When a man is diagnosed with prostate cancer, they are diagnosed by usually having small needles placed into their prostate gland, and those little pieces of tissue are then put on a slide. And conventionally, a pathologist would look under a microscope and make a determination based on those images whether or not someone has cancer. So usually, diagnostic pathology images are used for diagnosis.
However, these particular slides were archived from men who had enrolled on prospectively randomized trials that had been done over the last 20 years from a group called the RTOG, which is the Radiation Therapy Oncology Group. And there were 1000s and 1000s of men who volunteered for clinical trials where they were looking at treatment A versus treatment B to see which treatments were more effective. And there were numerous trials, and we had this tissue available.
When these digital pathology slides, or these slides were digitized, and we knew from these clinical trials what the outcomes of the patients were, the oncologic outcomes. We knew, for example, whether or not they were cured. We knew whether or not after certain treatments, they developed further spread of disease, we knew whether or not they died of prostate cancer. So, because we had this 20 plus years of data, we were interested in seeing if the AI could simply look at these biopsy samples and correlate what it saw on these biopsy samples with outcomes. And much to all of our surprise, it did an exceptional job of looking at this biopsy tissue, and it can give extremely reliable estimates of how probabilistic it would be after treatment like surgery, radiation, let's say, specifically radiation in this case, how likely it would be that someone would progress to spread after that in spite of that, or how likely someone would potentially die a cancer in spite of that.
And the benefit of knowing that prognosis is that you can now up front, before you render treatment, decide if the amount of treatment you're going to render is sufficient or not. Should you add more things? Should you de-intensify? And an additional element of the research, because you mentioned the word predictive, is that a common treatment that is given to men with localized prostate cancer is the combination of androgen deprivation therapy with radiation therapy. And I will assure you, men do not like receiving androgen deprivation therapy. It is basically chemical castration making a man's testosterone level zero through the use of drug therapy. And one of the things the AI determined was whether or not using that particular drug would literally be futile or not.
And so, the ultimate summary of this research collaboration that I was involved in is that something like half of men who are routinely recommended to get androgen deprivation therapy can safely avoid it, because they simply won't benefit from it. And in addition, we can ponder additional kinds of treatments if we thought that their risk of metastasis or spread after treatment were significantly high enough, we know not to use that particular drug.
Heather Simonsen: Gosh, then you're not putting patients through things that aren't likely to help anyway, saving them time and suffering.
The power of patterns: AI's predictive potential (18:06)
Johnathan Tward: That is clearly the precision medicine approach. Like if you knew up front that a drug would simply work or not, that is a hugely powerful tool in your arsenal, and that exact example is what we found with the AI's ability to evaluate the digital pathology images. It's actually a very remarkable achievement, and there are almost no instances or other contexts of oncology where we have these truly predictive markers that say you know this drug will work or this drug won't work.
You know, we were talking about treating people like groups, as opposed to precision. You know, today, most persons with a metastatic cancer diagnosis are generally offered the same treatment path, like start with this kind of chemo and start with this kind of surgery. But if you could select the drug or select the treatment up front to the one that you knew would work, you would always choose that.
Heather Simonsen: And then you can say to the patient with greater clarity. You know this, this is the course we recommend, and this is why.
Johnathan Tward: Correct, and not just greater clarity. I mean, I could give them precise risks that it'll work or not. I mean, that's a very interesting thing. Normally, you go to a physician and it's, it's sort of challenging for them to say, you Mr. Jones, have precisely a 5% risk of recurrence, as opposed to, let's say, a 27% risk of recurrence. And, you know, traditional grouping in one way, like staging generally allows doctors to do that.
But if you look at something like a patient with prostate cancer who has what we call intermediate risk prostate cancer, which is kind of similar to someone who's like a stage two, all I can tell that person at that grouping level is that their risk of spread of cancer after treatment is somewhere between zero and 40%. You know, it's like a huge range. But with AI based biomarkers, or other precision medicine biomarkers, I could say actually for you, it's precisely 7%. And that's very interesting, to help the patient, you know, sort of plan for their future, as well as the physician to determine if they need to do more or less based on those estimates.
Heather Simonsen: I mean, that's also, to me, revolutionary. I mean, to be able to pinpoint and take that to such a precise level that can really change cancer care.
Challenges and opportunities: The future of AI in oncology (20:39)
Johnathan Tward: Yeah, it's precision medicine at its finest. And you know, AI is excellent at this, but I don't want to suggest that there aren't other methods that are also good at this. For example, we use genomic profiling to do something very similar, and so you can look at a genomic profile of a tumor and take in some clinical information and also get to a similar conclusion. But what's kind of interesting is running genomic profile tests might cost thousands of dollars, and so even though you can arise to a similar precision medicine conclusion with a lot of expensive molecular testing.
What is so remarkable about the AI, especially this particular example we've been discussing, is that theoretically, the cost to run the test is whatever it costs to take a digital image of a slide and the amount of electricity to run the computer algorithm. So I like to think of these AI based approaches that use pattern recognitions on existing data as one that can really democratize cancer care, because now people from around the world, maybe from lower resource countries, or people who don't have a lot of assets, or even where distance is a disparity, and it's very challenging for them to come to the cancer center. You know, they might literally be able to just upload things from their computer. And you know, even though I've spent my own research career looking at the use of AI in prostate cancer, specifically, you can utilize this technology in any kind of cancer.
There's nothing particular to prostate cancer that is lends itself to artificial intelligence use what really matters is the availability of data, and so it doesn't matter. You could be a breast cancer patient, a head and neck cancer patient, any kind of cancer patient. We could utilize these same techniques that we have validated in the prostate cancer space to now extend to almost any cancer on the planet. In fact, that is an area that we are actively working on. My collaborators who originally published this research in the prostate cancer space now have access to lots of other clinical trials in different disease settings, and so we're going to see this kind of personalization percolate all the way through the entire spectrum of the various cancers out there.
Heather Simonsen: Was there any resistance when you presented this research? I mean, did people have their doubts?
Johnathan Tward: When I was first presenting this research to an expert community. Actually, I remember I was at an RTOG meeting, which is now called the NRG cooperative group. It wasn't that there was resistance to the research. You could see that it was so new, that people were struggling with understanding it, and the first questions people routinely asked were, what is it that the AI is seeing? And we had to explain to them that you cannot explain to a human all the patterns it is seeing. And that created a lot of sort of confusion on people's faces and concern like people like to know what is behind the scenes. They don't like black boxes. So that created, I don't know about resistance so much as concern.
And the other interesting thing, thinking back at that initial meeting, was when we were telling the scientists that we didn't lead the AI at all. In other words, we just gave it digital images and outcomes. We didn't point to the cancer in the biopsies. We didn't try to say, this is the ugly part. Like, this was what's called an unsupervised learning you know, people kept coming back and saying, well, what if you were to show it what you thought was the worst part of the cancer. Would it do better? Or what if you were to exclude the pieces of the biopsy tissue that you knew were irrelevant to which, I'd say the whole point is, is that what we're calling irrelevant, the computer is finding relevance in.
So, it is challenging our perception of what is relevant and what is irrelevant. These AIs are outstanding at identifying things that we were discarding as irrelevant. And in fact, my most recent research, looking at prognostic models with AI, has taught me that there are lots of easily obtainable pieces of clinical information that we are discarding that are hugely relevant to people's prognosis. These are free pieces of information that we collect on people routinely in the medical record. And there will be some surprise at how important these things are.
Heather Simonsen: Well, the future is here, but I can imagine there's much more to come with AI and cancer care and what do you think that looks like?
Johnathan Tward: It's so revolutionary a technology that it is almost impossible to forecast where we will be in the next 20 years. It's such a fundamental change. Almost like when you when you go back at the course of human evolution, really, and you start thinking about like, what were massive shifts in our ability to become civilized and advanced. You have things like the Bronze Age, the Iron Age. You know, these things used to take millennia to shift, and then Industrial Revolution, and then computers. But what we're now doing is compressing the pace of discovery from millennia to 100 years to years, to advances that are now coming in weeks to months, because the technology is so powerful.
I think that we are going to have tremendous tools that help us, you know, really tailor therapies to the patient, as we've kind of discussed, but the same time, we'll introduce a lot of efficiencies and inefficiencies. And we're going to struggle with how we deal with all the additional information that an AI can lead us to ponder. But on balance, even though there's always good and bad to all new forms of technology, I'm highly optimistic that there is far more good, especially in the world of cancer care and AI, that will come than the downsides.
Heather Simonsen: And let's address that. What are the potential drawbacks?
Johnathan Tward: Yeah, I think the drawbacks of AI and cancer care is that the conclusions an AI can draw are really only as good as the data it is trained on. And just like people, it can be highly biased. And so we have to be very careful with trusting the AI implicitly. So, I think it is still critical that humans are ultimately the people who are watching over this technology and ensuring that the conclusions AIs are drawing are accurate. I think that the AIs, although, can create a lot of efficiencies, can also create a lot of inefficiencies.
So, for example, a lot of attention has been placed on using AI to help radiologists interpret imaging studies like CAT scans and MRIs. And at first someone might think, well, they're really good at identifying diagnoses based on imaging. Does this mean we don't need as many radiologists? I think it's the opposite, which is that AIs can be so good at identifying possibilities, including rare ones, that it might suggest, for instance, to radiologists to consider additional diagnoses that they hadn't thought of before. And now that's going to require more time and energy for that human radiologist to really go through the data and do things that, like look through the medical record that the AI may not have access to, to kind of ensure that they're coming to the correct conclusion.
Heather Simonsen: Well, and then the ability with more oncologists and AI to reach more people. There's really a shortage of care currently, so I could see the potential there being great.
Johnathan Tward: There's so much enthusiasm for that right now. And this kind of dovetails into other technologies, like people are constantly using their phones, and the phones themselves are gathering data. And it's kind of interesting to ponder, is there unlocked potential on just certain behaviors that people routinely do when they're using their phones that maybe we can use to help in the oncology space? Now, of course, that brings up a lot of privacy issues. Or what about wearables, you know, like people are wearing, you know, smart watches that can monitor their pulse rate and oxygen saturation levels.
And so, the availability of all that data is very enticing to AI researchers in medicine, because we think that we can use all of that data to potentially head off problems the past with people's health. But there's so much that we have to work toward, especially with the privacy issues and things of that nature. Before we can really exploit that data instead of using all this data to market to you, you know, we'd like to explore using this data to actually make you healthier.
Looking ahead: The AI revolution in cancer care (30:21)
Heather Simonsen: What is your parting thought about AI and cancer? What do you want our listeners to consider that maybe they hadn't before?
Johnathan Tward: I think that the most important thing to consider is that we are going to see an absolute revolution over the next few years in how we are truly able to tailor our therapies to the individual based on this AI technology. And hopefully our patients won't fear it. I think some might be fearful of it, but it is such an amazing opportunity for all of us. And again, the pace of discovery is just so rapid. I think one of the biggest struggles for us in the future is actually, how do we move from a whole paradigm of treating people like stages or groups to one where everybody is an individual. That is the biggest challenge, but it's also the biggest opportunity for the future.
It's interesting because, you know, treatment guidelines are based on lumping people in boxes and saying this category of person with those features should be treated a certain way. But now, with the explosion of AI and the personalization, it'll be interesting to see how we deal with the complexity that is this patient gets this, that other patient gets that. And also how insurers and payers will be able to deal with it, because, you know, they have certain rules about what they will or won't pay for. And I think that there's going to be a lot of catch up to do, to kind of deal with this massive paradigm shift.
Heather Simonsen: I really enjoyed our conversation today. Thank you so much for being with us, Dr. Tward.
Johnathan Tward: You're welcome. It's such a pleasure to talk to you, and I'm looking forward to seeing where we are two or three years from now. It'd be interesting to update this conversation and see what's changed.
Heather Simonsen: Thank you.
Johnathan Tward: You're welcome.
Heather Simonsen: Thank you to Dr Jonathan Tward and Taylor Smedley for joining us today. The Leonardo Huntsman Cancer Institute exhibit on AI and cancer will open in January 2025. To our dedicated listeners, we are thankful for your support. For additional resources, be sure to check out our show notes. And if you want to stay connected with us and be the first to know about upcoming episodes, subscribe on your favorite podcast platform. Please log on to Apple podcasts and leave us a five-star review. This helps other people like you find this podcast. If you have questions, comments, suggestions for future episodes or a personal story you'd like to share, please visit our website, huntsmancancer.org. Mixing by Trent Cell theme music composed by Mix at 6 Studios, additional music from Art List, a special thanks to The Huntsman Cancer Institute Communications and Public Affairs team.