Transcript of Breaking Boundaries: AI’s Promise in Bladder Cancer Care

Click here to listen to the podcast

Voice over:

This is Bladder Cancer Matters, the podcast for bladder cancer patients, caregivers, advocates, and medical and research professionals. It’s brought to you by the Bladder Cancer Advocacy Network, otherwise known as BCAN. BCAN works to increase public awareness about bladder cancer, advances bladder cancer research, and provides educational and support services for bladder cancer patients and their loved ones. To learn more, please visit bcan.org.

Rick Bangs:

Hi, I’m Rick Bangs, the host of Bladder Cancer Matters, a podcast for, by and about the bladder cancer community. I’m also a survivor of muscle invasive bladder cancer, the proud owner of a 2006 model year neobladder, and a patient advocate supporting cancer research at the Bladder Cancer Advocacy Network, or as many call it, BCAN, producers of this podcast.

I’m pleased to welcome today’s guest, Dr. Bishoy Faltas. Dr. Faltas is the director of Bladder Cancer Research at the Englander Institute of Precision Medicine at Weill Cornell in New York City. He’s an assistant professor of medicine, cell and developmental biology. He’s also the Gellert Family-John P. Leonard MD research scholar in hematology and medical oncology.

Dr. Faltas is a physician scientist who focuses his research and clinical practice on urothelial cancer of the bladder and upper urinary tract. His work aims to improve the lives of bladder cancer patients by translating our discoveries to early phase and first in-human clinical trials.

Dr. Faltas has authored several peer-reviewed publications and high impact journals. He has received several awards, including the ASCO Young Investigator Award and the ACR NextGen Star Award. He received funding from the NIH, the National Institutes of Health, the Department of Defense, and Star Cancer Consortium. He is also a member of the BCAN Scientific Advisory Board and he co-chaired the 2023 BCAN Think Tank.

Before we begin, I want to let our listeners know that because AI is the topic of this podcast, I walk the talk after preparing a draft set of questions and I mine to ChatGPT to formulate a list of questions for an AI and bladder cancer podcast for patients. So while I’m the one asking the questions, ChatGPT offered some refinements, and I will tell you, it was a good first experience and maybe next time I’ll start with AI.

So Dr. Faltas, with that, I want to thank you for joining our podcast.

Dr. Bishoy Faltas:

Thank you, Rick, for having me. I appreciate it and I look forward to our conversation.

Rick Bangs:

I think it’s going to be fun. So let’s start with some basics. What is artificial intelligence or AI and when did it become a thing?

Dr. Bishoy Faltas:

So let me tell you about a few words that we often commonly hear, so artificial intelligence, there are a few phrases. Artificial intelligence is the ability of a machine to imitate intelligent human behavior, and usually that’s problem solving. Machine learning is an application of artificial intelligence that allows a system to automatically learn and improve from learning from experience. And then deep learning, which is an application of machine learning that uses deep neural networks and complex algorithms to train a model to achieve very sophisticated human-like intelligent behaviors.

Rick Bangs:

Okay, so it’s taking a bunch of inputs and it’s doing something with them, digesting them, synthesizing or whatever. And we had a conversation about this, so are we actually using AI or artificial intelligence without even knowing it?

Dr. Bishoy Faltas:

Yes, absolutely. I think many of us heard of AI for the first time when there was a lot of excitement in November 2022 when ChatGPT was launched by OpenAI, and there were other models that were generative AI models, which is another important phrase, which are models that generate new content. Many of us heard of AI at that time, but AI has actually been in development for many, many years. And many of us, I would say probably all of us have used AI at some point without even recognizing that.

I’m going to give you just a few examples of applications or appliances that we use every day that involve AI in some shape or form. I think many of us have used these little automated vacuum cleaners, your Roombas if you will. They’ll scan the house and then they’ll vacuum your carpet, and those rely on some type of AI and they will also learn their environment.

We’ve all, or many of us have used voice assistance such as CRE. That’s some form of AI. You and I had a conversation about facial recognition to unlock your phone, and that’s also another form of AI. Car companies now are involving self-driving hardware in their programs for these cars to self-drive and assist with driving.

Rick Bangs:

Right, right. I think some people think it may be scary, so what are the perceived dangers and can you help us understand what the true risks may be?

Dr. Bishoy Faltas:

Yeah, so there are actual risks and there are theoretical risks. And I think a lot of us when we hear the words AI and scary together, think of AI becoming sentient and doing bad things. And I guess that’s theoretically possible, but we’re very, very far from that. And every once in a while, you’ll hear in the news about an AI model where someone will claim that the AI became self-aware, are becoming sentient, which is really the hallmark of that.

So far, my understanding is that we’re not quite there yet because that’s actually … We don’t fully understand what that is, but it’s actually a very complex function. And while these models, for example, ChatGPT may appear as if they are self-aware, they’re not. They’re just imitating the behavior of a self-aware being because they were trained on language from self-aware beings, meaning ourselves. That’s sort of a psychological fear.

Now, practically in terms of applications, there are lots of issues that we have to navigate, so we have to deploy it ethically, right? So a lot of AI model development depends on training the model from existing data sets. And those data sets are data sets that are also generated by humans, imperfect beings, and these data sets will carry all our biases, they’ll carry all the imperfections, and if you train an AI on that, it will continue to have these biases. And then if you were to deploy that AI model to make a decision, then those factors will carry through, and that’s something that we have to be aware of.

There are obviously privacy issues because of the huge amounts of data that are required to train these artificial intelligence models. There is new risks that are related to generative AI because as I mentioned earlier, that’s AI that generates new content, so it can generate a new video, it can generate a new audio, it can probably recreate our conversation here without it ever happening and in a persuasive way. And so it takes that to the next level where distinguishing between real and AI-generated content becomes tricky and going to be an issue in research. It’s going to be an issue in many other aspects of our lives.

Those are just some of the issues or the challenges I would say that relate to the use of artificial intelligence in general.

Rick Bangs:

That all makes sense. And as we make the transition to applying this in the bladder cancer context, I think the facial recognition that we talked about drives home the point that sample is really critical, and I think there was some discussion of African Americans being underrepresented in the sample for facial recognition and some challenges with that, but I may be wrong on that.

Dr. Bishoy Faltas:

Yeah, absolutely. And I’ve read some news reports about AI use to assist in decision making within the court system, for example, to determine the length of sentences for certain people. And it was making decisions that are based on the data sets that it was trained on, but also even in healthcare, and I think we’re going to get to that. The AI is just a reflection of the dataset that it was trained on.

We know for example, that if one was to train an AI model from a clinical trial dataset, and I know you are very involved in clinical trial design as a patient advocate, that we know that certain groups of people are not well-represented within our clinical trial cohorts, so we have to be very aware about training AI models from these clinical trial cohorts because the output is going to reflect the training dataset.

Rick Bangs:

Sure. Okay, so let’s shift to healthcare, specifically bladder cancer. And what we’re going to do now is we’re going to explore some of the possibilities that AI either already has supported or might in the future support, so let’s start with prevention. Now, this one to me seems difficult.

Dr. Bishoy Faltas:

Well, so I mean, in my mind, prevention and early detection overlap because in some ways if you were to predict or … There’s prediction and there’s early detection. Let me start with early detection. As I mentioned earlier, image recognition in general is a very advanced area of artificial intelligence.

For example, one of the ways that urologists would detect bladder cancer in someone is urine cytology. Now you could take these images from urine cytology or add other imaging technologies to that, and the output of that could be used to train a model to better predict the presence or absence of bladder cancer, so that’s one way that that could be deployed. I’m just mentioning some ideas here and I’m sure that some of these may have already been tried or are already in development, but these are just some of the ideas that come to my mind.

But another aspect is that bladder cancer, as we all know and as you know very well, is a cancer that is associated with significant environmental exposures. And if you have big data on environmental exposures, you could potentially at a population level and perhaps eventually even at the individual level, stratify patients in the sense of giving them predictions in terms of the risk of developing bladder cancer or other cancer based on how polluted is the air where they live, do they live near a shipyard or a factory, how many particles in the air are there, so I think there are a lot of potential for these types of applications to provide more precise and comprehensive predictions of the risk of developing cancer.

Rick Bangs:

Yeah. I like that risk stratification aspect. That could be very promising. Let’s talk about modeling. You mentioned it a little bit. We have things like what are described as mouse models and we could theoretically model the tumor and the treatment and see how those interplay. What kinds of things would fall into the modeling category?

Dr. Bishoy Faltas:

Yeah, so here, this is not necessarily artificial intelligence per se, but this is actually a fusion of mathematical modeling and artificial intelligence. I work with a collaborator at Cornell Ithaca. Her name is Dr. Jaehee Kim, and we are working together in developing what we call bio digital twins of bladder cancers, and those essentially are bio digital avatars of bladder cancer.

And those are essentially two parts. There is the biological part, which is a tumor organoid that is derived from a patient, so that’s a piece of tumor tissue that we are growing in the lab and we’re able to test with drugs, we’re able to test many different scenarios, so essentially what we call a co-clinical trial that we’re doing on the tumor outside of the patient’s body while sparing them having to go through all of that.

And we are recreating essentially a digital replica of that, a mathematical model that is recapitulating the behavior of these cancer cells under the effects of these drugs. And by going back and forth iteratively between the model, the mathematical model and the lab, we are able to understand a lot more about the behavior of the actual tumor.

We can also model things because there’s no limit to the experiments that we can do in the mathematical model or what we call in silico, so we can have the computer simulate so many scenarios. And where the artificial intelligence comes in actually is that if you were to generate a very complex mathematical equation of something like a cancer, it becomes unsolvable.

Where the artificial intelligence comes in is that Jaehee has developed a very interesting way where the artificial intelligence model can not necessarily solve the equation, but predict the space of equations that would give you the right answer and therefore really narrow down the answer, which would enable us to then really model very complex behaviors and get to answers quite quickly that wouldn’t be doable except with a supercomputer that would run for three days or something like that.

We were able to do this now and our computers in a way that is still fairly accurate. Again, this is in principle similar, for example, to how ChatGPT predicts what’s the right next word. It’s not exactly an equation that it solves, but it predicts based on a statistical model the most likely next word. So that’s what we’re trying to do with this, and we’re hoping that we will use these bio digital avatars to solve some very important clinical problems.

Rick Bangs:

I mean, that’s just fascinating. I mean, you’re not interacting specifically with a patient and you’re still garnering information. That’s just wonderful. Talk to me about imaging. That seems like a rich area.

Dr. Bishoy Faltas:

Yes. Imaging has been and will be on the forefront of the application of artificial intelligence technologies in medicine because as I mentioned, any type of image, whether that’s a retinal scan or a radiograph, a CT scan, an MRI image or it’s a cystoscopy image or it’s a histopathology, all of these are really amenable to having artificial intelligence decipher the data encoded in these images.

For example, we just presented some work that we did at ASCO GU at the ASCO Genitourinary Symposium that was done in collaboration with Fei Wang here at Weill Cornell and Olivia Elemento, where we developed a model and applied it to a clinical trial dataset that integrates imaging RNA expression and the spatial information, which is another type of imaging output that we get from these pathology slides from the patient’s tumor, and integrated all of that into three different models that we then fuse to predict whether that patient is going to have complete pathologic response, which means that there would be no cancer left after treatment with chemotherapy. We can predict that just from feeding the model the images from the histopathology slide and the RNA expression data.

Rick Bangs:

Wow.

Dr. Bishoy Faltas:

And it works fairly well and we are validating this in additional cohorts. That was something that is something that we’re very excited about expanding to other trials and other treatments as well.

Rick Bangs:

Just a clarification. So you talked about spatial, and so is that 3D kind of thing rather than just flat?

Dr. Bishoy Faltas:

It’s looking at the difference. If you think about a cancer, we focus on the cancer cells, but the cancer is not just cancer cells, it’s cancer cells and many other supporting types of cells that are normal cells or altered cells, but not quite cancer cells. We call that the stroma or the matrix or the micro environment. And that involves immune cells, it involves fibroblasts, it involves other blood vessels, cells that form blood vessels, and all of these play an important role in the biology of the cancer in its clinical behavior.

You can see these on a regular histopathology image, and we have trained a model that based on the morphology, just how the cells look like or their nuclei, the central part that contains the DNA, looks like on a regular histopathology slide, the model can determine the type of the cell that’s there and it can make a map of that on the histopathology slide. So it could pinpoint where the cancer cells are, where the fibroblasts are, where the immune cells are. And by studying the relationships between these different components in the tumor microenvironment, we’re able to find some signals and generate some hypothesis about the interactions between them that are important for response to chemotherapy. And that’s actually something that came out of this model and we’re working on validating that in different data sets as well.

Rick Bangs:

Wow. Much more comprehensive information, which is what we need. Let’s talk about diagnostics. And I want to look at this from both the initial diagnosis as well as recurrence. What about that area? We talked a little bit about imaging, and I’m sure that will play in, but this can be much broader as you’ve already mentioned, versus imaging and cystoscopies and urine and other kinds of biospecimens and so forth. So what about the diagnostic space?

Dr. Bishoy Faltas:

Even the diagnostic space, I mean, I think we’ve alluded to that earlier with looking at urine cytology, but even urine cytology has a lot of limitations. Now, there are new technologies not necessarily related to AI, but can serve as data sets for training AI models that are quite promising. For example, urine cell-free DNA. So we know that when there are cancer cells in the urine, they release DNA into the urine. And there are several colleagues of mine and many others and several commercial entities as well that are developing models for detecting bladder cancer with high sensitivity and specificity from urine cell-free DNA.

And that’s quite the message in all of these is that the interesting thing about artificial intelligence in my opinion is that you are able to fuse or utilize several types of data into a single output, so that’s what we’ve done in the work that I described earlier. And you can certainly do that in a diagnostic setting, where for example, you can have a cytology layer, cell-free DNA layer from imaging, and you can fuse all of that into a much more robust diagnostic.

And if you think about it, that is how your urologist or your oncologist operates. They don’t just take the cytology or the imaging. The human experience or the doctor’s experience, clinical experience and acumen is to take all of that and then provide that prediction or call on what they think is the right thing to do.

Rick Bangs:

Yeah, it’s almost like we’ve been living in a two-dimensional world and now it’s three-dimensional and this is helping us connect some of the dots, which is what we’ve been trying to do as humans, but this is on a much more accelerated basis.

Let’s talk about metastasis. So is there some possibilities for AI in terms of predicting metastasis or diagnosing it?

Dr. Bishoy Faltas:

Yeah, so similar to what I was just describing, so the way we detect metastasis now is by looking at images, so CT scans, MRI scans and other imaging modalities, PET scans, to find areas that look like they are metastatic. Recently, there is a very exciting technology called circulating tumor DNA, which is very similar to what I described in your end, but it’s essentially cell-free DNA, but circulating in the blood, and we can detect that with a blood test.

And one could easily, and in fact, we’re actually already working on that, to integrate circulating tumor DNA into some of these models because it is a quantifiable output. You can detect specific individual mutations and you can measure how these mutations are changing over time, which is a reflection of the clonal structure, if you will, meaning the different cancer cells that all have different genetic makeups within the parent tumor or the metastatic tumor.

I think we would be able to detect metastasis better earlier, which is critical, and then track their evolution during the treatment with drugs. And that generates a lot of data, and even clinicians now are not entirely sure what to always do about this flood of new data types that is coming our way, and I think artificial intelligence would hopefully give us some decision-making aids to help us make decisions based on all these different data inputs.

Rick Bangs:

Yeah. Absolutely, absolutely. Okay, so the next question, ChatGPT suggested a question about biomarkers. And we may have covered some of this already. You’ve mentioned some, but are there any additional perspectives you want to provide on the identifying of biomarkers? And you may want to define a biomarker is so that we can kind of set the stage.

Dr. Bishoy Faltas:

Yeah, so a biomarker is a test that helps us make a clinical decision. And there are different types of biomarkers and different levels of rigor and validation that are required for a biomarker before it becomes a clinical test.

And the interesting thing about artificial intelligence, and I’ve alluded to that already, is that for example, some of our work that I mentioned earlier, fusing together images and RNA expression and spatial data can be developed into a clinical test into a biomarker that we can use in clinic. So you would have a package, essentially, of these three components with a model that is run in a certain way and gives reliable outputs, and oh, there’s a lot of standardization that needs to happen in terms of what happens in the lab behind the scenes and in terms of reporting and testing and performance before that becomes a clinical test.

But we are finally at a point where we can integrate different data types into a single biomarker, which in the past has been quite challenging. And usually in the past, our biomarkers have been a single protein, for example, like HER2 expression or a single or FGFR3 mutation in bladder cancer, which is a single gene, so mutations in a single gene. What the artificial intelligence technology allows us to do is to incorporate many different data types that may be very different in their structure and then be able to still combine all of that into a clinical test.

Rick Bangs:

Awesome. ChatGPT had another suggestion, which I thought was a really good one. What about side effects? Is there a place for AI in the side effects arena?

Dr. Bishoy Faltas:

Yeah, so I think one thing we haven’t talked about is patient education and interaction between what role would artificial intelligence play in terms of assisting doctors communicating with patients. But I’m sure many of us or many of my patients would go on Google before they see me and they would do the research, as they should.

And now I’m sure there’s a lot of asking ChatGPT, and I think that will continue to evolve as ChatGPT gets better and better, and I’m sure there are already communication assistance artificial intelligence platforms that are being developed to try to help doctors and patients communicate better and everybody else in the health team. So I think there is a huge opportunity here for improving communication.

I’ll tell you one of the applications that physicians are very interested in and excited about is the ability to have artificial intelligence help us with our clinical documentation. I’ve recently learned of an application that would be enabled when a doctor and a patient are having a conversation in a room and it would essentially transcribe all of this information and provide it in a format to the doctor after the visit in the format of a clinical note that they would be able to review and sign.

The advantage of that is that the doctor is not looking at the computer screen, they’re looking at the patient, they’re having a good conversation with them without having to worry about typing or documenting things, but the documentation still happens. So I thought was a brilliant way of improving doctor patient communication with technology that we already have.

Rick Bangs:

Yeah, that would be outstanding because I know there’s a lot of burden on the doctor side of the house, so that would be great. And it seems like there’d be an opportunity here to proactively know before a side effect has become critical. It’s kind of the early part of that after surgery or whatever, knowing before it becomes a critical emergency room kind of issue.

Dr. Bishoy Faltas:

Absolutely. I think triaging, answering questions that you would get answers to more readily without having to wait on the phone for two hours for someone to call you back to make a decision, I think that would be wonderful.

Rick Bangs:

Yeah. Okay. What about treatment, and ChatGPT is suggesting treatment plans, which is a related issue, so treatment and treatment plans?

Dr. Bishoy Faltas:

Yeah, I think treatment and treatment plans is … I think one note of caution here, and that’s interesting. I have not tried that, but my understanding is that if you were to ask ChatGPT to suggest a treatment, it will give you the disclaimer that it’s not a doctor.

Rick Bangs:

As it should say.

Dr. Bishoy Faltas:

As it should, as it should. I think it’s important to understand, and I think all of everybody in your audience appreciates that, that medicine is still quite complicated and there is definitely the part of art and science when dealing with patients and human beings and the special nature of the doctor-patient relationship.

I would just add a note of caution here that it’s important to get informed, to stay informed and do your research, but also to be able to communicate that openly and directly with your physician and understand the rationale for why they’re making the decisions that they’re making.

Rick Bangs:

Right, right. I think it’s just a constant reminder that we’re not at a point where artificial intelligence doesn’t need some kind of editing or supervision. It all makes sense.

Surgery. So you talked a little bit about making it easier to document notes and so forth on the clinician side, but it seems like there might be some interesting possibilities, particularly as we have things like robotic surgery. What about the surgeons?

Dr. Bishoy Faltas:

I’m not a surgeon, so that’s a disclaimer, but I think with the robotic surgery, there is definitely in the future or currently ongoing, I know that there are ongoing programs for developing essentially autonomous surgical robots. Talk about trust here. But these are probably initially going to be deployed in extreme environments where there is no human surgeon that is available or maybe on the battlefield or maybe in space or maybe where there would be the opportunity for an autonomous or a semi-autonomous robot to perform some surgical procedures, and that would require a complex level of artificial intelligence. You can’t really program everything in the robot. It will need to have the ability to respond to its environment and to the different anatomy and the changes in the patient’s condition and so on and so forth, so I think we’re probably going to see that in our lifetimes. I am watching that closely.

Rick Bangs:

Again, really fascinating. Let’s talk about navigation. We all know, and you and I both know that it’s very difficult for patients to navigate through the cancer journey. We’re going to talk about clinical trials next, but let’s talk more broadly about the topic of navigation.

Dr. Bishoy Faltas:

Navigation is a really important area and I think again, I see tremendous opportunities there because as you very astutely pointed out, it’s just our healthcare system can be quite overwhelming and very complex to navigate. And I think there are definitely opportunities for improving communication, improving navigation, explaining some of the terminology when you come to the hospital because I’m sure it sounds like people are speaking a foreign language when you go into the hospital because of all the medical terminology that not everybody takes the time to explain. I think again, AI could really play an important role here in patient education in helping the patients, our patients navigate the healthcare system to have more satisfaction.

Rick Bangs:

Definitely agree there. All right, let’s talk about one of our favorite topics, which is clinical trials. What about there?

Dr. Bishoy Faltas:

Yeah, so one of the areas that I’m really excited about is, so lots of areas, everything that we’ve talked about so far virtually can and will be done in the context of a clinical trial. But another very interesting area, and that’s something that we’re thinking about as well, is these in silico clinical trials. Can we simulate clinical trials if we have enough data from clinical trials that would enable us to predict the outcome of other clinical trials without ever having to do them because we can run a clinical trial in a million patients in silico, virtual patients, models of patients, or at least bio digital twins in silico without having to randomize 700 patients to get drug A versus drug B or drug A versus placebo.

And I think if we can get to that level … Currently, we cannot do that, but I think it’s an important goal to be able to maybe predict the outcomes of certain clinical trials and then maybe run a smaller trial or a smarter trial or a faster trial. I think we would all agree that these are all desirable outcomes to minimize the risk to the patients while still getting the answers to our clinical questions and improving outcomes for all the patients that come after that trial is conducted.

Rick Bangs:

Yeah, absolutely. All right, I’m going to merge. There’s a question I wanted to ask about survivorship. Typically, we define survivorship as anything that happens after treatment. And ChapGPT wanted to come at it a little differently, but it’s the same topic, which is managing post-treatment recovery and monitoring long-term health. I think we can kind of blend those into one question.

Dr. Bishoy Faltas:

Yeah, I think that, again, comes back to education, but also realizing, for example, early signs of a relapse. But also, I am sure, and I think there have been studies on that, that anyone who goes through a cancer-related experience will have a certain level of psychological consequences and anxiety about … We often talk about scan anxiety, which is the anxiety of the patient and the doctor, by the way, waiting for the scan results for many years after bladder cancer treatment.

I think there might be an opportunity there for helping the patients with the psychological effects in terms of counseling, in terms of just talking about and managing and navigating these feelings of anxiety, uncertainty and so on. And I think that’s quite important now with the shortage of qualified mental health care professionals. I know that there might be an opportunity for artificial intelligence to help with this.

Rick Bangs:

Excellent. You’ve talked a little bit about education and support for survivors, families, partners, so forth. Are there any other education and support topics that you want to put on the table?

Dr. Bishoy Faltas:

Yeah. I think you just mentioned partners as well, so not just the patients, but the patient’s family who oftentimes will carry a significant part of the burden and need support, and that’s something I think we don’t do well nowadays, but I think we could always improve there. Any opportunities that artificial intelligence would bring in terms of trying to help with that would be, I think, most welcome.

Rick Bangs:

Absolutely. All right. Any other work that you find extremely exciting that we haven’t talked about already? It could be in bladder cancer or medicine or beyond. What other work is kind of exciting for you here?

Dr. Bishoy Faltas:

I think it’s important … There is so much exciting work in artificial intelligence. I run a research lab that studies bladder cancer, and a couple of years ago, there was a publication by a group that developed a artificial intelligence machine learning model called Alphafold that predicts the structures of proteins of all the known proteins, and this is a very interesting area that’s continuing to grow where we are able to essentially make predictions of the structures, the folding of many hundreds or thousands of proteins that we have, and then the effects of thousands or millions of mutations or alterations on the folding of these proteins. That has a lot of implications for drug development, for example, to target specific mutant forms of the protein. That’s an area that is very exciting, developing very quickly and I’m watching that area very closely and I think we’ll see a lot more in this area.

Rick Bangs:

Excellent. Okay. Any final thoughts?

Dr. Bishoy Faltas:

Final thoughts. This is a very exciting time for machine learning and artificial intelligence in bladder cancer and a very exciting time for bladder cancer research in general. I think it’s really fortunate and privileged to participate in some of this research and I hope that we can use all of this to cure more patients.

Rick Bangs:

Absolutely. Okay, so thank you, Dr. Faltas. You’ve given us a fascinating look at the promise and clarified some of the pitfalls of artificial intelligence in the bladder cancer context. If you’d like more information on bladder cancer, please visit the BCAN website, www.bcan.org. In case people wanted to get in touch with you, could you share perhaps a Twitter handle or an email address, whatever you’d like to share?

Dr. Bishoy Faltas:

Sure. I welcome anybody who would want to get in touch with me. My Twitter handle is @faltaslab, and my email address is [email protected].

Rick Bangs:

Okay, thank you. Just a reminder, if you’d like more information about bladder cancer, you can contact the Bladder Cancer Advocacy Network at 1-888-901-2226. That’s all the time we have today. Be sure to comment and subscribe to this podcast so we have your feedback. Thank you for listening, and we’ll be back soon with another interesting episode of Bladder Cancer Matters. Thanks again, Dr. Faltas.

Dr. Bishoy Faltas:

Thank you Rick.

Voice over:

Thank you for listening to Bladder Cancer Matters, a podcast by the Bladder Cancer Advocacy Network, or BCAN. BCAN works to increase public awareness about bladder cancer, advance bladder cancer research, and provide educational and support services for bladder cancer patients. For more information about this podcast and additional information about bladder cancer, please visit bcan.org.