Detecting Lung Cancer Earlier With AI w/ Prashant Warier, Founder & CEO, Qure.ai

Lung cancer is often discovered too late, when treatments are expensive and survival rates are low. But what if routine chest x-rays could flag cancer early…long before symptoms appear? 

AI is transforming everyday imaging into a powerful early detection tool, reshaping screening economics and saving lives around the world. 

Prashant Warier, CEO and Founder of Qure.ai, joins CareTalk to discuss how AI enables earlier diagnosis, why chest x-rays are an untapped opportunity for detection, and what it takes to integrate AI into national health systems at scale.

🎙️⚕️ABOUT PRASHANT WARIER
Prashant is an expert in the field of Artificial Intelligence and Deep learning. He has architected and commercialized several data science solutions in his 19 year career. He is also a prolific researcher, author, and speaker on topics related to data science and machine learning. He is passionate about using deep learning to make healthcare affordable and accessible. 

🎙️⚕️ABOUT QURE.AI
Qure.ai is a health tech company that uses deep learning and artificial intelligence to make healthcare more equitable for patients worldwide. Learn more about how Qure is supporting clinicians and advancing work across the pharmaceutical and medical device industries, here.

🎙️⚕️ABOUT CARETALK
CareTalk is a weekly podcast that provides an incisive, no B.S. view of the US healthcare industry. Join co-hosts John Driscoll (Chairman, UConn Health) and David Williams (President, Health Business Group) as they debate the latest in US healthcare news, business and policy. 

GET IN TOUCH
Follow CareTalk on LinkedIn
Become a CareTalk sponsor
Guest appearance requests
Visit us on the web
Subscribe to the CareTalk Newsletter

Transcript

David: 0:00

Most lung cancers are found too late when treatment is costly and survival odds are slim. But what if a routine chest x-ray could flag tumors before symptoms appear? AI is turning everyday scans into early warning systems, and it's already changing outcomes around the world. Welcome to Cure Talk. Executive features a series where we spotlight innovative companies and leaders working to advance the healthcare field. I'm David Williams, president of Health Business Group, and my guest today is Prashant Warier. He's CEO and founder of cure.ai. His company is using artificial intelligence to detect lung cancer earlier, and they're dramatically improving survival rates and reshaping how health systems think about screening economics and public health policy. Prashant welcome to Care Talk. Great.

Prashant Warier: 0:53

Thanks for inviting me, David. Great to be here.

David: 0:56

Outstanding. Okay. Now my first question is why focus on lung cancer and what makes that such a defining use case for early detection?

Prashant Warier: 1:06

So, lung cancer is one of those conditions. I mean, see, I, I'm a, uh, my background is a data scientist, so I've been building AI and algorithms, data science algorithms for the last 25 years, and, uh, got into healthcare with cure about 10 years ago. Uh, as I was looking at this space, I realized that lung cancer is one of those conditions where early detection actually can create substantial value, right? A stage four, uh, detection of lung cancer. Stage four, diagnosis of lung cancer. Uh, five year survival is about 5%. A stage one, stage two detection, uh, or diagnosis, uh, can lead to about 70, 80% survival rates. And so you're going from 5% to, uh, 70, 80%, and it's, you can go from stage two to stage four in six months. To 18 months. So very short period of time, you could get to a late stage diagnosis. And that's where screening matters. If you're able to diagnose early, you could have extremely high survival. And uh, that's, that's sort of the problem we're trying to solve is how, and, and this is just wanted to say that lung cancer is specifically one of those conditions. It's the leading cause of death, leading cons cause of, uh, cancer death. And, uh, it is one of those conditions where early detection can create the most amount of value, and that's why we focus on this particular problem.

David: 2:18

No, that sounds, uh, that sounds exactly right. So as you're saying, you know, a lot of cases are found late, and if I understand correctly, most cases are often found at, at stage four. So it's pretty common that people hear that. First thing they hear is that, yeah, you don't have much time to go. How do you think about the cost of the delay, both in human terms and in economic terms? Now, you certainly gave me this kinda statistical difference between, you know, say 5% versus 78%, but how do you think of that, you know, beyond those numbers?

Prashant Warier: 2:48

So there are about 2,230 K cases of lung cancer diagnosed in the US every year. And more than 1 26 K, more than half of them die every year. And now again, the people who die might be people who got diagnosed last year. So not necessarily the same set of folks, but it's, I mean, there is a substantial amount of people who that don't survive. Right? And uh, that's one, I mean, that's a huge cost. Second is that the cost of treatment. Early stage versus late stages, maybe 10 to 20 XI mean, early stage, you can have a lung resection surgery and uh, you could treat it at a very low cost that could go much, much higher, uh, when you're looking at late stage diagnosis. Um, and I mean, I think I, I mean, if I look at it, I mean, these two, uh, are the most common. I mean, I would say, I mean the cost of, um, uh, I mean the, the, uh, mortality due to lung cancer. And the cost of treatment. I think these are the two areas where we could improve from a cost perspective.

David: 3:46

Got it. Now there's this create trial. I always love the names with these acronyms for the, the trials, that that always sounds very positive. It's drawn a lot of attention, uh, really worldwide. Why, what's unique about this study that's causing it to get so much attention?

Prashant Warier: 4:01

So the, uh, basic thing is that, uh, look at the problem this way, right? Why is lung cancer diagnosed late? Because patients don't have symptoms. Uh, when, uh, lung cancer is early in the lungs, it's small nodules, small mass, uh, you don't have cough anything, any symptom at all, right? And by the time you start getting symptoms, it's already pretty late. Uh, and it's already a stage three, stage four diagnosis. So the way to diagnose early is through screening, right? And typical screening protocols are for, uh, are using CT scans. Are for smokers and people who are either 50 or 55 and beyond. So heavy smokers and people who are beyond a certain age limit. So a lot of times people, young people don't get screened, uh, and non-smokers don't get screened. And what we have seen is that all of them have a probability of getting lung cancer and a significant probability of getting lung cancer. So what we did is we said, let's, I mean, and the other problem that we didn't sort of, we also identified is the fact that. CT scan as a modality. I mean, in the US it is widely available, but outside of the us especially in countries like India, uh, or Southeast Asia, uh, you don't have enough CT scans to do screening. So there is no screening happening there. Right? So one of the things we looked at is there are 1.3 billion chest x-rays taken around the world. That's about 20% of the population. These are people who are getting a chest x-ray for routine reasons. I mean, you could take a chest x-ray for an annual medical checkup, you could take a chest x-ray for an infectious condition that you have pneumonia or cough or flu or whatever else, right? And we said if we were to create a risk score on these routine x-rays, and then identify the right set of patients who could be then triaged for a ct, that could create, create enormous value. And, and that is what we did. And we, we built what is called. A nodule detection algorithm and a malignancy risk scoring for that nodule. Uh, and now the point is that that score is only useful if you're only identifying the right set of patients who need to be examined further. If you say out of out of a hundred patients, 10 patients need, need further examination, there is not enough capacity to do that. Right? So what we wanted to check was what is the accuracy level of this particular tool? Was it accurate enough to detect the right lung cancer cases? What we found is that on a chest x-ray, if we identify a high-risk nodule, that patient is 54% likely to have a high-risk nodule on a subsequent CT that is performed. And a chest x-ray is a very basic examination. First of all, these are routine exams. So again, you're not going into get a chest x-ray for screening. For lung cancer. Screening is something that you did for something else already, for another condition, right? And so we were able to show that 54% of patients. Who came in for a routine exam and were identified as high risk by cure, then actually ended up positive on a chest ct. So that was one of the key findings that came out of the create study. And the sort of side effect of this is the fact that the people who are getting these chest x-rays are people who may not necessarily qualify for low dose CD scan screening. These are people who are, uh, non-smokers. People who are below 50 years of age. So we are also helping diagnose lung cancer in patients who would not be diagnosed through the, the typical screening based pathway.

David: 7:12

So, one kind of technical question. So beyond the insight of using a chest x-ray, there's also the necessity to be able to see some very small, uh, nodules on the chest x-ray that perhaps wouldn't be visible. Under routine examination, are you applying some technology in order to be able to find those small nodules?

Prashant Warier: 7:34

So what we do is we apply deep learning. I mean, we have millions of scans or tens of millions of scans that have gone into the algorithm to train the algorithm on, uh, what a nodule looks like, right? Uh, plus the fact that we, we. Have corresponding ct. So on a chest x-ray I can see an nodule. Now I can also, or maybe the nodule is not clearly visible on a chest x-ray, but I have a corresponding CT where it is clearly visible. So I train the algorithm using, using corresponding CT plus I also have biopsy data where I can say that this nodule is actually a malignant nodule. So I am linking my X-ray data with CT data and biopsy data, and that allows the algorithm to. Now learn what a nodule looks like. I mean, and it can detect really small nodules, hidden nodules behind the heart, shadow behind ribs, uh, and also identify which kind of nodules are likely to be, uh, malignant, which is again, something that on a chest x-ray, uh, a radiologist find hard to do. So we do both these things and, and, and the algorithm works because it's seen, uh, a lot of x-ray data and it's seen a lot of paired x-ray CT biopsy data as well.

David: 8:44

I wanna go back to something you just said before and, uh, kind of emphasize it, which is the population where you're actually finding these lung cancers. Now, most people think about lung cancer as something you get from smoking and smoking certainly causes lung cancer. And so after a lifetime of smoking, you know, people get lung cancer. That's, that's got a traditional view and there are a lot of people like that. But let me know if this is right. So it sounds like, um, that. You detected cancers, 59% of the people where you detected cancers were actually non-smokers, and almost 10% of them were under 50. Now those people, I don't think would get screened, including not just in a low resource area like Southeast Asia, but maybe in the US as well. So is that right? And then what does that tell us about how we should be thinking about the screening guidelines?

Prashant Warier: 9:34

Exactly. Exactly right. And in fact, I mean this is something that we are now scaling up across the health systems in the US where incidental detection on chest x-rays is becoming more and more common as we deploy our solutions across the systems. Right. And the data that you presented is exactly correct. Right? Out of the number of cancers that we detect, 50% of the cancers are detected early. Typical numbers are 80, 20, 80% are detected, late 20% are detected early. We are talking about 50 50, 50% early, 50% late. Right? 59% of the cases are non-smokers. I mean, which would not, they would not go through, uh, they would not go through screening at all. Uh, around 10% are under 50. Again, these are people who would not, uh, go through screening. And, um, this, I think the, the reality and coming back to the question that you asked about screening guidelines, right? Screening guidelines are set in a way that you have limited CT capacity. You can only CT scan so many more people because this is not a person, this is, this is an elective procedure, right? This is not something where somebody needs to get a ct. I mean, this is something that you have to do on an annual basis going forward. I mean, so, uh, you don't have enough capacity to, let's say if, if you decrease that, that number 50 to 40. There might be like, let's say five x more people who need to be screened, which you may not be able to do, uh, which any, any health system may not be able to do. So those screening guidelines are set in a way where there is enough capacity to perform, uh, the screening. Uh, but that said, I think what we are trying to do as cure is to look at routine exams, whether it be a chest x-ray or a chest ct, or even look at electronic medical record data and see if there, if there are risks that we identified that can be flagged. For that patient to then get a ct. So now you're not, you're not screening somebody who has got no symptoms or no risk of lung cancer. You're actually screening a high risk population versus regular screening is okay. Anybody above 50, anybody who's got 25 pack years of smoking goes and gets a ct. Now you're changing the paradigm to say that I will load, I will basically look at people who are high risk and get them a ct. Right. And this might, I mean, again, we are early in this, but this could potentially. Change screening guidelines over the next several years, but right now, screening protocols are all low dose C scan driven, and for the near future, I see them remaining that way.

David: 11:58

Now you've imp, you've implemented this AI lung screening in a couple of places in Goa, in India, and also in El Salvador, and I'm wondering what kind of lessons have you learned from doing these rollouts?

Prashant Warier: 12:12

I think the, the key couple of lessons and multiple lessons that we have, uh, learned, right? One is that in Goa and El Salvador, what we have done is basically every X-ray machine, every x-ray that is taken automatically comes to cure. So one is that you basically saturate the system. You, you, anything that, uh, is, uh, any, any scan that is taken there is an ambient, think about it as an ambient ai, right? It's just there. Looking at everything, every scan that is performed, whether it be a chest x-ray or a chest cd, right? And making sure that only the right set of patients get flagged for further examination. So one is that the ambient AI use case, uh, needs saturation, needs to work across, uh, all of the hospitals in a system. Second thing that we learned, uh, very important is the fact that I can flag as many high-risk patients as I want, but unless. If there is a follow up performed, there is somebody who's ensuring that there is a CT that is done for that patient. It's not very effective, right? So follow follow up is super, super critical. And follow up requires two things. One is that we have to integrate with the health system, that their IT ecosystem, their EMRs, uh, their pacs, to ensure that there is a follow-up, uh, created within the. IT systems. And second is that patient navigators are super critical. I mean, there are nurse navigators who are calling those patients, getting them back, uh, for a ct. And what we realized is that we are building AI for the doctors, but we are also building AI for the nurses. And we need to ensure that they have all the right information to make sure that the patient knows that they have to come back for a follow-up examination, which is a ct. After the, uh, high risk xray that we identified. So, so learning learnings, again, one is the saturation across systems. Uh, second is integration with, uh, the, uh, the IT systems in the, the hospital. And third is, um, ensuring that we build an AI for the nurses as well, not just for the physicians. And that's an area where, uh, I think a lot of, not a lot of companies are focused on, on that particular problem.

David: 14:14

Prashant, I was reading an article the other day and it was talking about how. Uh, Western drug companies in particular were developing products and a lot of the trials were done in lower income places like India, like El Salvador, but then that, uh, a pretty small percentage of the therapies once they got on the market were actually available there. So they were developed in these places and those were the insights were generated and the evidence, but then they weren't available there later. Are you following a similar path or should we expect something different?

Prashant Warier: 14:46

Our products are available across 105 countries. We are actually deployed across about 4,800 hospitals in 105 countries. And there is, uh, like you mentioned, I mean Goa, state of Goa, which is in India, uh, El Salvador as a whole country. Uh, we are in every Sub-Saharan African country. We are in, uh, all countries in Southeast Asia. Uh, we are in uk. We are in many countries in continental Europe. We are, uh, in US and Canada, Latin America. So, uh, I think. The good thing about AI is that you can deliver it anywhere, and I can deploy it in any, any hospital, anywhere in the world in a matter of a couple of days, uh, or in actually even a matter of a couple of hours. Right. So AI that way is something that can be easily deployed and skilled, and our technology is, uh, available universally.

David: 15:35

Great. So if you are advising a, a health minister, let's say, for a country that hasn't yet. A region that hasn't yet implemented, uh, this technology, what do you tell'em or what would you tell them about how to be most successful in the rollout?

Prashant Warier: 15:50

I think, um, my, my, uh, and having, I mean, having sort of worked on deploying CURE solutions for the last five, six years, uh, across health systems, uh, in many, many parts of the world, uh, I think my advice to a health minister would be that look at. Look at the end-to-end, um, workflow that is required to deploy ai. Uh, AI itself will not be successful unless and until you think about what the follow on effects of that are. So, for example, in the lung cancer case, if we say that this patient needs a follow up ct, now you need to have the capacity to do that ct. You need to identify the, the machine or the hospital where they're gonna get that CT scan done. Uh, you need to have a mechanism. To arrange that, that, um, schedule their appointment, you need to have a nurse navigator who's following up with that patient to get them back for ct. So, uh, that whole, um, ecosystem needs to be enabled to deploy, to make the AI successful. Uh, AI itself is great technology. It can add a lot of value, but you need to get all the other pieces also, uh, figured out.

David: 16:57

Prashant, that sounds, uh, interesting in terms of the Health Minister cadet's. Good. Uh, any, any other key to success for them?

Prashant Warier: 17:04

I think one, one thing which, uh, I also want to add is the fact that, like I said earlier, there are 1.3 billion chest x-rays taken routinely around the world, and that is, these are exams where you could actually find the risk of lung cancer. And a lot of times, uh, a lot of health ministers, especially in uh, emerging markets, think that they're not, there is not enough cts, and so you cannot have early detection of lung cancer because you don't have enough CT machines or capacity to do low dose CT scans. But this is a data set that is already available that is already being taken. For so many reasons. And you could now with our technology, you could be diagnosing lung cancer early. You could be creating programs. I mean, not necessarily, I mean calling it screening is probably not the right word, but you could, you could have early detection mechanisms, uh, using an x-ray based pathway.

David: 17:51

I think that's very helpful. Uh, and I'm, I hope that people will take it to heart. We discussed before, uh, the importance of. The logistics, especially the follow up and making sure people have access to that, uh, so that the rollout is successful. And you mentioned that with AI it can be deployed, you know, very quickly at scale. Now the challenge, one of the challenges is about, uh, the, the role of the clinician and what trust they may have in the AI for themselves and also for their, their patients. And we've seen in some places ar around the world, um, a concern about, Hey, I don't like ai, or I think it's going to. I don't trust it or I don't understand it, it's a black box or it's gonna take my job, or the patient doesn't like it. Is that a factor? And how do you think about earning the trust of the clinicians when you're introducing AI into these workflows?

Prashant Warier: 18:41

So, it's a great question actually. And this is something that we have been, uh, figuring out. Actually this is one, one of our biggest learnings over the last 10 years, right? Uh, and the first thing is that, I mean, you have to. Understand what you are promising. I mean, trust is created if you promise something and deliver that consistently over time. Right? And a lot of times when people think of ai, they think of something that is gonna be a hundred percent accurate all the time. That's not the case. I mean, you know that AI is playing a role in this, right? It's, it's flagging those patients who may not have been flagged further for examination. So you might have some false negatives, but you, you want to reduce the number of false positives here, right? In this particular example of, uh, lung cancer screening. So. Uh, we are, uh, we, we know how to, how to position the AI to the, uh, physician so that they can get, extract the maximum value out of it, whether that be saving them time, whether that be identifying the right set of patients who need a CT examination or finding something which, uh, they would've missed or they, they don't want to do. I mean, for example, quantifying, quantifying lesions in a CT scan is an extremely cumbersome process, time taking process. And we can do that automatically in a very short amount of time. AI can do that automatically so that that becomes extremely valuable for a radiologist, for example. So, uh, I think identifying the right value proposition, uh, pitching that and delivering that consistently over time, uh, that creates trust. And once, once there is trust, then things, I mean, work, work very well.

David: 20:14

We've been, uh, discussing a lot here, global impact and the impact on national or regional health systems, which is amazing because it's something that can impact a lot of people. But of course, each one of these, uh, diagnoses comes down to a particular. Individual and their family. Uh, you've highlighted a patient named Diane in the UK and she had an early diagnosis. It came from an AI alert on a routine x-ray, which is the kind of thing that we're talking about here, but help to personalize that story a little bit. What does that story tell you about, you know, the potential to do things differently in the real world?

Prashant Warier: 20:49

So, uh, the, the key of that, that story, right? That's of course. I mean, that's one data 0.1, one story, right? One story means a lot to that person, to that family. Right. And you could, I mean, what we are seeing at Cure is we have now hundreds of such stories that are coming out. Patients that are diagnosed in El Salvador or in India, uh, or in the us. Um, and uh, I think from a data standpoint, I think the key is that in the UK the average, average turnaround time for lung cancer diagnosis about 65 days right. To start treatment. And we were able to bring it down to 25 days with our technology. And, uh, this time, the 25 to 65 days, a lot of it is scheduling, A lot of it is not getting a CT scan on time. And by triaging that this patient needs a ct, we are ensuring that that delay, uh, is not happening. Right. And, uh, we are ensuring that nurse navigators are able to prioritize this patient, get them, uh, triaged on time. And Diane is one example. But yeah, you could have hundreds of thousands of such patients who could be diagnosed early. If you had, uh, used QAI on on scans that were taken for routine reasons and our goal is to prevent a hundred thousand cancer deaths, uh, in the next five years.

David: 22:10

So sometimes sort of bringing it back to an individual's story, you know, you go to the doctor's office, they give you a good report or they say, gee, there's something a little concerning to be able to look at. And one of the concerns in the past has been perhaps if you've got. Something and you, you're worried about it and it turns out well, I, it wasn't anything I needed to worry about after all. And so it was all that for nothing. Or even if you do a more invasive test, like a biopsy that you might have, uh, pain, even some serious side effects and all that. So I'm just thinking about kind of like the overall rhythm of it. It sounds like what you're describing here is routine chest x-ray that you were doing anyway for some other, some other reason. And then it's almost like. Incidental finding, except you're doing it in a systematic way. And then the follow up, as you've emphasized a couple times, is a low dose ct. Yeah. So like not a big deal. Um, and then you get the, you get the turnaround. How, how quickly do you get the turnaround from the read of the, of the ct?

Prashant Warier: 23:11

So, uh, I wanted to answer your first question, which is super important question, right? That this is exactly the point of what, what the create study that you asked about that is what we're trying to prove, right? Create a lot of false red flags, right? And somebody doesn't have cancer. If you have 10, 10 people that you flag, and I mean all of that, nine people don't have cancer or don't have a risk of cancer, they're unnecessary scaring them. So your, your technology has to be super accurate, and that is where the 54% success rate on, uh, that we got in the create study is super important. That out of more than one out of two patients are actually positive on the ct. Right? Uh, so that, that sort of, that number is super critical when you look at. Uh, screening or locate identifying the right set of patients for the low dose CT scan. Now, the low dose CT scan turnaround, honestly, uh, it's, it's something which depends on the system. Uh, you can get a report in one day to one week. Uh, but system to system, it might differ. Uh, yeah, I mean, radiologists will typically report in a day, uh, uh, typic. I mean, ideal numbers should be about a day or two

David: 24:16

prashan. At the, the start of the podcast, it started off talking about, you know, why lung cancer, and it was pretty clear. It's, uh, fairly common. A deadly disease where if you find it early, you can do something about it as opposed to if you don't, uh, it's, uh, it's a problem people already have. Uh, they're getting chest x-rays fairly routinely. That turns out to be a good, uh, modality for being able to pick things up. Now, it's been a few years since you did that and pick that out. Technology has changed. Treatments have changed. And so if you look today, is it still lung cancer as like a singular. Cause or would there be other conditions where you'd also be looking at this AI driven, uh, early detection? What's next? In other words?

Prashant Warier: 24:58

So, uh, this is something that we have been thinking about all the time, David. We process about, uh, 20 million scans, uh, across the world, 20 million chest x-rays, uh, on, uh, on an annual basis. And, uh, what we have been thinking about is what else can we detect early on these routinely taken chest takes space. And, uh, a few things that we identified. One is, uh, chronic ulcerative pulmonary disease. Uh, it's again, one of those conditions which gets diagnosed late and early detection can have, uh, substantial value. So, uh, that is something that we built out a detection of COPD, just like our lung cancer product. Um, and that is something we are scaling up now. We built out the technology. Uh, we are scaling that up now. Uh, there are a few other, uh, areas, I mean, looking at. Detecting heart failure on a chest x-ray. So you can see some signs of that. Uh, and we have some technology around that. We are still sort of testing that, doing the equivalent of the create study, uh, that we spoke about, uh, to see what the value of that would be. Uh, and then there are some more, uh, conditions that we are working on, uh, and we'll probably announce, uh, in the next couple of months. But asq, our focus is on using. Ambient AI on routinely taken exams, whether it be a chest x-ray or a chest cd to diagnose disease early, and that that has been our focus, that will continue to remain our focus, uh, for the next couple of years.

David: 26:22

That's a great theme to be pursuing. And of course, the, uh, the economics of doing that do differ by condition and by country. So with lung cancers, you're saying if you don't detect something early, it could proceed pretty quickly to death. COPD, just even the first word of it, is chronic. So it usually, you know, a few months, uh, delay in, in, uh. In diagnosis, they're not gonna be fatal. However, there are some good drugs for COPD and it's underdiagnosed. And so, at least in the us, which I can speak to, uh, you know, more COP diagnoses also has a good economic outcome for a good clinical outcome for the patient, but also potentially good economic outcome for some other other players, and that's gonna vary around the world. What is your vision? You know, I've talked about the, the rollout of, uh, the lung cancer. Screening, uh, what's your vision for how you integrate AI into national health systems more broadly?

Prashant Warier: 27:17

One of the biggest issues that I see, uh, with integrating our nationals health systems is, uh, that the nations also have to think about how they are adopting ai. And there is a lot of different kinds of AI that is available right now. And, uh, I think having nations need to have clear strategies on what kind of AI that. They're deploying and, uh, how they're deploying it directly with companies like your or through platforms. Uh, and what is, like I mentioned earlier also, what is the follow up action from that and how do you enable the follow up action that comes from deploying AI solutions here? So I'm seeing that, for example, UK NHS has, uh, a very, uh, well thought out AI strategy. Uh, I, I see that, uh, in the us I mean, not, not from a national perspective, but a lot of. Larger health systems have thought out, uh, very, uh, well, well thought through, uh, AI strategies. I think many nations are still figuring that out. And I would say, uh, that's, that's the primary one is to figure out what is, what is the strategy to deploy AI in the health system. Uh, we are always happy to work together if, if there is an opportunity,

David: 28:26

Prashant, I know as a patient it would be great to. You'll get this early detection and if I, if I had cancer, to find out about it and do something about it. But one of the problems with healthcare around the world is that you add things, maybe it's good for the patient, but it adds costs and it's already kind of unaffordable. I mean, is there a business case for doing this or is it just sort of one more cost in the health system?

Prashant Warier: 28:52

Again, a great question. I think one of the things that I wanted to add is that we are not adding screening costs, right? The kind of technologies that we're deploying, these are ambient in nature, running on routine scans, so there is no additional x-ray or a CT taken. There is no additional cost there. Second is the cost of treatment. Again, what we are seeing is that detection, early detection actually reduces the cost of treatment, does not increase the cost of treatment, and it actually increases the lifespan of the patient. So it is good for the patient. Cost-wise, it is good for the government or the, uh, health system, uh, and it's not adding more, uh, x-ray or c or more investigational cost, more X-ray C costs. So overall, I think, uh, early detection of lung cancer or some of the other conditions, uh, adds positive value to the health system.

David: 29:38

That's it for another episode of Care Talk Executive Features. My guest today has been Prashant Warier. He's CEO and Founder of Cure. His company is using artificial intelligence to detect lung cancer earlier, dramatically improving survival rates without breaking the bank. I'm David Williams, president of Health Business Group. If you like what you heard, please subscribe on your favorite podcast platform and thank you Prashant.