WEBVTT - Innovation in Medical Treatment and Technology 0:00:05.280 --> 0:00:09.479 Allow a seven, five or three. The pulse of medical 0:00:09.480 --> 0:00:13.360 technology has quickened in recent years, bringing forth a transformative 0:00:13.400 --> 0:00:16.479 new age in healthcare with no signs of slowing down. 0:00:17.520 --> 0:00:21.160 Forty percent of healthcare industries globally are already regularly using 0:00:21.239 --> 0:00:24.840 AI and machine language right now. An AI's stake in 0:00:24.880 --> 0:00:28.159 the healthcare market is expected to grow ninefold in the 0:00:28.200 --> 0:00:31.480 next six years, making it worth nearly one hundred and 0:00:31.640 --> 0:00:35.720 ninety billion dollars by twenty thirty. As doctors rely on 0:00:35.760 --> 0:00:39.560 technology to improve the medical experience for each and every 0:00:39.560 --> 0:00:43.720 one of their patients. Hi, how are you feeling today? 0:00:46.400 --> 0:00:48.840 After coming out of the grips of a global pandemic 0:00:49.040 --> 0:00:52.320 involving a virus the world had not seen before. Healthcare 0:00:52.360 --> 0:00:54.840 needs to be at the forefront of research and technology 0:00:55.080 --> 0:00:59.360 now more than ever. AI's role becomes not just innovative, 0:00:59.480 --> 0:01:03.840 but a cent while creating a lifeline for overburdened healthcare systems. 0:01:05.160 --> 0:01:07.880 Join us as we explore the intersection of technology and 0:01:07.920 --> 0:01:11.000 medicine and how the two are revolutionizing the way we 0:01:11.040 --> 0:01:18.320 experience healthcare today and in the future. Welcome to Technically Speaking, 0:01:18.520 --> 0:01:21.480 an Intel podcast, the show that brings you the stories 0:01:21.520 --> 0:01:26.600 and insights of AI presented by iHeartMedia's Ruby Studio and Intel. 0:01:27.840 --> 0:01:31.559 Hey there, I'm Graham class. In this episode, we're diving 0:01:31.560 --> 0:01:34.319 into the world of healthcare and medicine where AI and 0:01:34.360 --> 0:01:37.959 technology are not just changing the game, they're saving lives. 0:01:38.680 --> 0:01:41.200 We'll be joined by two experts who at the vanguard 0:01:41.240 --> 0:01:47.600 of this revolution that's introduced today's guests. Alex Flores is 0:01:47.600 --> 0:01:51.240 the General Manager of Health and Life Sciences Vertical at Intel. 0:01:51.560 --> 0:01:55.160 He will share insights into how AI is reshaping patient care. 0:01:55.760 --> 0:01:58.600 Also joining us is Peter sched, the Head of Digital 0:01:58.640 --> 0:02:02.200 Health for North America for Siemens Health and Ears, which 0:02:02.240 --> 0:02:06.760 focuses on the implementation of advanced technologies like therapeutic imaging 0:02:06.840 --> 0:02:10.919 and laboratory diagnostics to enhance patient care. Semens Health and 0:02:10.960 --> 0:02:14.000 Years work with healthcare providers to ensure that innovative new 0:02:14.040 --> 0:02:17.920 technology is working efficiently and that staff understand how to 0:02:17.960 --> 0:02:21.800 best use technology so patients can get accurate answers about 0:02:21.840 --> 0:02:26.079 their health fast than ever before. Both guests will help 0:02:26.160 --> 0:02:28.760 us understand the direction of healthcare as a whole and 0:02:28.800 --> 0:02:32.920 the AI powered diagnostics and innovations currently changing the face 0:02:32.960 --> 0:02:38.959 of medicine. Thank you both for joining me today. 0:02:39.480 --> 0:02:41.800 Graham and Peter, thank you for having me today. Really 0:02:41.800 --> 0:02:45.200 excited about this conversation. Excited to be here today, Graham, 0:02:45.280 --> 0:02:46.680 great to talk to you and Alex. 0:02:47.480 --> 0:02:51.640 Recently, I read an interesting survey conducted in August last 0:02:51.680 --> 0:02:55.240 here of onenty twenty seven people, which found that sixty 0:02:55.240 --> 0:02:58.840 four percent of people would prefern Ai system over a 0:02:58.880 --> 0:03:02.560 human doctor. For gen Z, that number rises to eighty 0:03:02.600 --> 0:03:06.080 two percent that would prefer Ai over humans. I'd like 0:03:06.120 --> 0:03:10.320 to get your general thoughts about that. I'll start with Alex. 0:03:10.960 --> 0:03:13.680 Yeah, it's a really interesting topic. I've heard a lot 0:03:13.720 --> 0:03:17.080 about this too. I think what fascinates me most is 0:03:17.280 --> 0:03:19.119 in a lot of surveys, a lot of data that's 0:03:19.160 --> 0:03:24.720 out there, patients are often more honest with virtual assistance 0:03:24.840 --> 0:03:28.680 with chatbots, so I find that very fascinating. What's also 0:03:28.760 --> 0:03:33.840 interesting is oftentimes a chatbot, for example, can also show 0:03:33.880 --> 0:03:38.280 more empathy. You know, chatbots don't get tired of, for example, 0:03:38.320 --> 0:03:42.000 answering the same question over and over again. But the 0:03:42.160 --> 0:03:44.440 other thing that's really interesting, kind of the flip side 0:03:44.480 --> 0:03:49.119 of this is accountability. So, for example, people are more 0:03:49.120 --> 0:03:53.440 accountable to other people, to other humans, specifically doctors, So 0:03:53.520 --> 0:03:56.320 I find that another really interesting area in terms of 0:03:56.960 --> 0:04:00.840 you know, maybe people do prefer chatbots or control assistance 0:04:00.920 --> 0:04:04.120 for some areas, but there's always that need for human 0:04:04.160 --> 0:04:05.480 touch beta. 0:04:06.160 --> 0:04:09.480 Yeah. Maybe just to add to what Alex was saying, Graham, 0:04:09.520 --> 0:04:13.000 I think a lot of us don't maybe even realize 0:04:13.040 --> 0:04:17.280 that AI is already playing a role today within their healthcare. 0:04:17.600 --> 0:04:21.400 So patients who are going to go get a diagnostic test, 0:04:21.440 --> 0:04:25.120 for example, to get a MRI of their knee, or 0:04:25.160 --> 0:04:27.560 maybe they've got something bothering them in their chests so 0:04:27.600 --> 0:04:30.760 they get a chest CT scan or whatnot. When that 0:04:30.800 --> 0:04:33.560 patient lays down on the table to get that diagnostic 0:04:33.600 --> 0:04:37.040 scan on that MRI, the MRI actually in some cases 0:04:37.120 --> 0:04:40.360 is already looking at the patient's anatomy and is able 0:04:40.360 --> 0:04:43.080 to identify and recognize, oh, this is the patient's knee. 0:04:43.120 --> 0:04:46.080 So I'm going to position the patient within that diagnostic 0:04:46.120 --> 0:04:49.240 scan to the most optimal position so that they can 0:04:49.279 --> 0:04:52.160 actually get a good visualization of that knee. So all 0:04:52.240 --> 0:04:54.839 that is actually being done not just by a human 0:04:54.880 --> 0:04:57.839 but also by artificial intelligence. It's actually built into that 0:04:58.000 --> 0:05:01.560 MRI scanner that's already helping create that optimal position for 0:05:01.600 --> 0:05:05.320 that patient. So AI is already being utilized in many 0:05:05.400 --> 0:05:08.479 aspects of healthcare, and again, patients may not actually even 0:05:08.520 --> 0:05:11.000 realize that they're getting some of the benefits from AI. 0:05:11.839 --> 0:05:16.039 So in a sense, it's more of an augmentation to 0:05:16.080 --> 0:05:20.520 help doctors and medical practitioners to make better diagnosis. Yeah. 0:05:20.600 --> 0:05:23.640 Absolutely, I think as Alex pointed out, certainly we seemen's 0:05:23.640 --> 0:05:28.120 health in years here, we also value the relationship and 0:05:28.320 --> 0:05:31.760 acknowledge the relationship that the patient has with their physician, 0:05:31.960 --> 0:05:34.280 and we want to make sure that that relationship isn't 0:05:34.279 --> 0:05:37.360 disturbed by artificial intelligence. But as you said, Graham, really 0:05:37.400 --> 0:05:40.839 augmented by AI, so that physician, that doctor, he or 0:05:40.880 --> 0:05:44.600 she can make a more informed diagnostic decision or maybe 0:05:44.600 --> 0:05:48.600 a more personalized therapeutic decision for that patient, backed up 0:05:48.600 --> 0:05:50.280 by what the AI is helping with. 0:05:50.960 --> 0:05:53.240 If you don't mind, just to add what Peter was saying, 0:05:53.640 --> 0:05:55.960 I really like to use the analogy of a pilot 0:05:55.960 --> 0:06:00.520 and copilot. So airlines have been using artificial intelligence for many, 0:06:00.560 --> 0:06:04.320 many decades now, but the need for a pilot and 0:06:04.400 --> 0:06:08.120 a co pilot has never gone away. Even when the 0:06:08.240 --> 0:06:11.720 plane is an autopilot, there's still a need for a 0:06:11.760 --> 0:06:13.800 pilot and a co pilot. So the way I see 0:06:14.120 --> 0:06:17.960 artificial intelligence is really more that co pilot for that 0:06:18.040 --> 0:06:21.000 physician who happens to be the pilot. And at the 0:06:21.080 --> 0:06:23.240 end of the day, it's really about the patient. What 0:06:23.440 --> 0:06:28.760 can artificial intelligence do to help enable better patient outcomes 0:06:28.839 --> 0:06:30.359 and so forth for the patient. 0:06:31.120 --> 0:06:33.640 Yeah, that copilot concept. I mean I use that for 0:06:34.640 --> 0:06:38.000 my coding and it's helped me tremendously. But I'd like 0:06:38.040 --> 0:06:41.640 to sort of turn towards maybe your personal stories about 0:06:41.720 --> 0:06:45.520 what makes you so passionate about this intersection between technology 0:06:45.520 --> 0:06:48.360 and healthcare. I'll start with Peter. Do you have a 0:06:48.400 --> 0:06:49.479 story that you could share? 0:06:50.120 --> 0:06:52.080 Oh, I don't know if it's a story or not, 0:06:52.279 --> 0:06:54.479 but this is an area that I've grown into and loved. 0:06:54.520 --> 0:06:56.479 I mean, it's an area that I've been part of 0:06:56.560 --> 0:06:59.800 for over twenty five years. Outside of healthcare and our 0:06:59.800 --> 0:07:02.440 p lives, we embrace technology. We're always looking at the 0:07:02.480 --> 0:07:05.440 latest and greatest and technology standpoint. How do you take 0:07:05.480 --> 0:07:08.200 that same comfort level, that passion and bring that now 0:07:08.240 --> 0:07:12.240 into a space like healthcare, Because at least from my perspective, 0:07:12.360 --> 0:07:15.560 I see the opportunity for so much benefits for the 0:07:15.600 --> 0:07:18.760 patient here, so certainly not just for the clinician in 0:07:18.840 --> 0:07:21.840 terms of efficiencies and time savings and everything, but really 0:07:22.080 --> 0:07:24.960 remarkable benefits for the patient in terms of being able 0:07:25.040 --> 0:07:29.920 to diagnose ailments earlier, find more personalized treatments for those patients, 0:07:30.160 --> 0:07:34.680 potentially saving patients' lives or detecting diseases earlier and treating 0:07:34.680 --> 0:07:38.840 those diseases earlier because of technology. To me, that's super exciting, 0:07:39.240 --> 0:07:41.560 super interesting in why I love being in this space. 0:07:42.280 --> 0:07:46.560 Alex, Yes, very similar to Peter. I'm not a clinician. 0:07:46.760 --> 0:07:49.920 I'm an engineer by training, and you know, I have 0:07:50.120 --> 0:07:55.080 the honor to manage some of the brightest engineers today 0:07:55.240 --> 0:07:57.920 on my team. And really the way we show up 0:07:58.040 --> 0:08:01.760 is we look at this from an engineering perspective and 0:08:01.800 --> 0:08:06.160 a technology perspective. So being able to sit down with clinicians, 0:08:06.240 --> 0:08:10.680 with nurses, with practitioners and so forth, and really understand 0:08:10.760 --> 0:08:14.320 what are their problems, what are their challenges, and then 0:08:14.400 --> 0:08:16.600 being able to step back and look at it from 0:08:16.640 --> 0:08:20.320 a technology lens and seeing how we can apply that technology. 0:08:20.880 --> 0:08:23.520 For me, that's what's most exciting is being able to 0:08:23.560 --> 0:08:27.160 work across the ecosystem, being able to work with different 0:08:27.200 --> 0:08:30.320 partners and really look at it in terms of how 0:08:30.360 --> 0:08:36.239 can technology be seamless and help clinicians ultimately deliver better care. 0:08:37.559 --> 0:08:40.320 For our regular listeners of technically speaking, you know that 0:08:40.400 --> 0:08:43.400 in season one we covered some of the challenges surrounding 0:08:43.440 --> 0:08:47.080 adoption of this innovative technology in a variety of professions. 0:08:47.840 --> 0:08:50.920 There has always been some tension when advancements in technology 0:08:51.040 --> 0:08:56.160 drive major changes in an industry, be it transportation, manufacturing, retail, 0:08:56.440 --> 0:08:59.760 or security, and that's certainly true in the field of healthcare. 0:09:00.200 --> 0:09:04.320 With game changing technology like AI runs into regulations and 0:09:04.400 --> 0:09:08.800 red tape that might slow its adoption. Well, perhaps patients 0:09:08.800 --> 0:09:16.200 are simply unfamiliar with how this new technology can help them. 0:09:16.360 --> 0:09:19.199 You know, I think we all acknowledge the great possibilities 0:09:19.240 --> 0:09:22.800 of a technology like artificial intelligence, for example, but really, 0:09:22.840 --> 0:09:26.200 how do you drive adoption of this technology within the 0:09:26.240 --> 0:09:29.480 healthcare space, And certainly there's different ways to do it. 0:09:29.600 --> 0:09:32.400 We talk about this trust that the patient has with 0:09:32.480 --> 0:09:36.320 the clinician and this valued relationship there. We've got to 0:09:36.400 --> 0:09:39.960 also help the clinician build trust with the technology and 0:09:40.200 --> 0:09:43.559 trust with artificial intelligence. What we do see though, is 0:09:43.600 --> 0:09:46.760 also making sure that as you develop these AI algorithms 0:09:47.080 --> 0:09:51.280 that they're really developed based on the patient population that 0:09:51.280 --> 0:09:53.760 they're going to be applied towards. We live in a 0:09:53.800 --> 0:09:56.040 diverse world here, and we need to make sure again 0:09:56.080 --> 0:10:00.520 those AI algorithms are appropriately fine. Took think the second 0:10:00.559 --> 0:10:03.360 thing is to really help again the clinician get comfortable 0:10:03.360 --> 0:10:06.600 with this technology. We've got to be able to educate 0:10:06.600 --> 0:10:10.440 the clinician on why the AI algorithm has made the 0:10:10.440 --> 0:10:13.880 clinical conclusions that it has made. Remove this veil of 0:10:13.880 --> 0:10:17.320 a black box that the AI algorithm is helping that 0:10:17.360 --> 0:10:21.319 clinician understand why is the computer coming to this particular conclusion. 0:10:21.760 --> 0:10:24.400 Having that type of education I think is really important 0:10:24.440 --> 0:10:27.400 in terms of driving that overall adoption of a technology 0:10:27.440 --> 0:10:27.880 like AI. 0:10:28.840 --> 0:10:31.679 And Alex you know, I'm pleased to hear that you're 0:10:31.679 --> 0:10:34.760 an engineer as well. We deal with challenges and problems 0:10:34.800 --> 0:10:37.240 all the time. What are some of the key challenges 0:10:37.480 --> 0:10:41.359 that you face getting this technology into healthcare systems? 0:10:41.840 --> 0:10:43.760 Yeah, I think there's two areas that I would want 0:10:43.840 --> 0:10:48.640 to add. One is around transparency. There needs to be 0:10:48.679 --> 0:10:51.600 a bigger focus in terms of transparency in terms of 0:10:52.320 --> 0:10:57.320 educating doctors, nurses, and so forth on when AI is 0:10:57.360 --> 0:11:00.680 actually being used so they understand it, they know that 0:11:00.800 --> 0:11:05.120 it's there and hopefully it is actually helping them solve 0:11:05.120 --> 0:11:09.560 their problems. So that transparency and understanding when it's being applied, 0:11:10.040 --> 0:11:12.600 why it's being applied, and how it should be applied, 0:11:12.640 --> 0:11:15.720 I think is very important. I think the second thing 0:11:16.080 --> 0:11:19.160 that the industry hasn't been talking enough about, and that's 0:11:19.280 --> 0:11:23.720 around validation, and specifically what I mean by validation is 0:11:23.840 --> 0:11:28.880 once those algorithms are out there, going back and really understanding, Okay, 0:11:28.920 --> 0:11:32.040 are they doing what they were supposed to do? And 0:11:32.120 --> 0:11:35.440 if they are, what is their effectiveness? But if they're 0:11:35.480 --> 0:11:38.400 not doing what they're supposed to be doing, then what 0:11:38.559 --> 0:11:41.480 can be done to actually augment them to make them 0:11:41.480 --> 0:11:43.960 better and so forth? And a lot of times that 0:11:44.080 --> 0:11:47.680 has to go back to the target population that's using 0:11:47.679 --> 0:11:50.080 them and really understand how we can make that better 0:11:50.120 --> 0:11:53.960 and ultimately get solutions out there that are impacting the 0:11:54.040 --> 0:11:55.400 right way in. 0:11:55.440 --> 0:11:58.880 Terms of the intellence, seemens healthy as partnership and the 0:11:58.920 --> 0:12:01.840 way you work. Do you have any specific projects or 0:12:01.840 --> 0:12:06.160 examples that you could share where some of these either 0:12:06.240 --> 0:12:10.400 AI or technology driven solutions that actually made a difference 0:12:10.480 --> 0:12:11.640 in a healthcare outcome. 0:12:12.280 --> 0:12:16.359 Yeah, it's so great to partner with a similar innovative 0:12:16.480 --> 0:12:20.000 company like Intel here to deliver our solutions to the 0:12:20.040 --> 0:12:23.040 healthcare professional seem as Healthy Ears has one of the 0:12:23.160 --> 0:12:27.720 unique distinctions of being the only medical technology company capable 0:12:27.760 --> 0:12:31.320 of end to end cancer care, so from diagnosis to screening, 0:12:31.360 --> 0:12:34.680 to treatment to survivorship. This is something that we cover 0:12:35.120 --> 0:12:37.720 to take care of the patient. And one aspect of 0:12:37.760 --> 0:12:41.560 that is during the treatment of cancer patients, especially during 0:12:41.640 --> 0:12:46.240 radiation therapy, they might have had a cancer identified in 0:12:46.280 --> 0:12:48.560 some portion of their anatomy and now we've got to 0:12:48.720 --> 0:12:52.280 apply radiation to kill that cancer. There's a tedious task 0:12:52.360 --> 0:12:54.319 that has to be done to make sure that we 0:12:54.400 --> 0:12:58.800 target that radiation towards the cancer but not the healthy 0:12:58.880 --> 0:13:02.960 tissue around the cancer. So what's typically done a clinician 0:13:03.040 --> 0:13:07.000 will sit down and they'll actually manually draw out where 0:13:07.040 --> 0:13:11.360 the cancer is and the anatomical structures around that cancer, 0:13:11.800 --> 0:13:15.840 so that they can feed that plan to radiation therapy 0:13:15.920 --> 0:13:18.520 machine so that the machine knows where to target the 0:13:18.600 --> 0:13:23.160 radiation on that patient. So for clinicians, that actually takes 0:13:23.200 --> 0:13:27.040 sometimes hours on end and actually in some cases delays 0:13:27.120 --> 0:13:30.120 the treatment for patients because of this kind of very 0:13:30.280 --> 0:13:32.880 tedious step we had seen in Healthy aers. We actually 0:13:32.880 --> 0:13:36.000 created an AI algorithm that helps kind of automate some 0:13:36.120 --> 0:13:40.080 of that tracing. But because of the complexities of three 0:13:40.160 --> 0:13:44.440 D objects and human anatomical structures, no two tracing is 0:13:44.440 --> 0:13:47.959 alike here, so we actually have to have really high 0:13:47.960 --> 0:13:51.600 powered computing that's really accessible to the clinician to be 0:13:51.679 --> 0:13:57.120 able to accurately trace out these malignant cancer abnormalities and 0:13:57.120 --> 0:13:59.920 then making sure that healthy tissue is protected here. So 0:14:00.559 --> 0:14:02.760 with the help of Intel, we've actually been able to 0:14:03.080 --> 0:14:08.040 accelerate tracings of tumors where instead of taking hours, it 0:14:08.120 --> 0:14:11.480 takes literally minutes now, So what that translates to is 0:14:11.480 --> 0:14:14.400 for patients, they can actually schedule their treatments quicker in 0:14:14.480 --> 0:14:17.360 advance and in rapid succession to be able to get 0:14:17.400 --> 0:14:20.760 rid of that cancer. So we're actually seeing direct patient 0:14:20.880 --> 0:14:24.040 benefit because of this relationship that we have with our 0:14:24.040 --> 0:14:25.440 technology partner at Intel. 0:14:26.080 --> 0:14:29.000 Yeah, I was actually gonna ask a question about the 0:14:29.120 --> 0:14:31.680 radiation side of things, So it's great that you are 0:14:31.680 --> 0:14:34.640 able to expand on that. In terms of the actual 0:14:34.680 --> 0:14:38.320 cost of these sorts of systems being implemented or slotted 0:14:38.360 --> 0:14:41.680 into the existing workflow, what are your thoughts on the 0:14:41.720 --> 0:14:45.920 cost models or the ability for hospitals and maybe even 0:14:46.080 --> 0:14:50.800 smaller practitioners to get this sort of technology into their practice. 0:14:51.520 --> 0:14:54.320 Yeah, you know, certainly cost comes into play here, and 0:14:54.760 --> 0:14:57.080 one of the challenges that we're seeing with the overall 0:14:57.160 --> 0:14:59.760 adoption here is that, you know, it becomes a challenge 0:14:59.760 --> 0:15:02.560 for are some providers to be able to make an 0:15:02.600 --> 0:15:06.200 investment in these type of technologies because of the uncertainty 0:15:06.480 --> 0:15:09.120 around not just the cost, but making sure that they 0:15:09.160 --> 0:15:13.600 get reimbursed for those costs. Unfortunately, with the way the 0:15:13.720 --> 0:15:18.000 landscape stands today and how AI is continuously evolving, our 0:15:18.040 --> 0:15:21.400 current setups for payment for these types of services haven't 0:15:21.440 --> 0:15:25.360 evolved this quickly. So you have today over seven hundred 0:15:25.400 --> 0:15:28.680 different AI algorithms that have been approved by the FDA 0:15:28.760 --> 0:15:32.920 here in the United States, but merely a handful and 0:15:32.920 --> 0:15:35.080 when I say handful, like literally you can count them 0:15:35.080 --> 0:15:38.280 on the fingers of your hands are actually reimbursed for 0:15:38.360 --> 0:15:41.520 that technology, and some of them are not even reimbursed 0:15:41.560 --> 0:15:43.960 at the same level that it costs for those technologies. 0:15:44.160 --> 0:15:47.840 So if you're a larger organization that maybe has some 0:15:48.000 --> 0:15:51.200 financial flexibility, maybe you can take that risk and make 0:15:51.200 --> 0:15:54.440 that investment. But certainly if you go to let's say 0:15:54.600 --> 0:15:59.520 rural communities or the underserved populations where that financial flexibility 0:15:59.560 --> 0:16:02.480 isn't there, it becomes a very difficult decision for the 0:16:02.520 --> 0:16:05.200 provider is to make that investment. And I think that's 0:16:05.200 --> 0:16:07.560 where we're seeing some of the shortfalls with adopting this 0:16:07.680 --> 0:16:10.400 technology and why we at Semen's Health in years we've 0:16:10.400 --> 0:16:14.200 been advocating to folks in Washington that we need to 0:16:14.240 --> 0:16:19.000 have a consistent and predictable reimbursement associated with artificial intelligence, 0:16:19.080 --> 0:16:21.520 not just to make sure that hey, everybody gets paid 0:16:21.600 --> 0:16:24.480 on it, but more importantly for us to be able 0:16:24.600 --> 0:16:28.280 to see what is the downstream benefit of this technology 0:16:28.400 --> 0:16:31.080 to the healthcare system and to the patients. 0:16:31.760 --> 0:16:33.480 One of the things that we like to help you 0:16:33.720 --> 0:16:37.400 scale this adoption of artificial intelligence and this new technology 0:16:37.800 --> 0:16:40.640 is really showing how hospital systems can deploy on their 0:16:40.720 --> 0:16:44.000 existing infrastructure. We want them to know that they don't 0:16:44.000 --> 0:16:48.000 need to rip and replace their existing infrastructure. What they 0:16:48.040 --> 0:16:50.760 can do is with partners like Semen's Health in EARS, 0:16:50.760 --> 0:16:53.200 we can show them how to deploy on their existing 0:16:53.240 --> 0:16:56.600 assets and then from there they can really derive the 0:16:56.640 --> 0:17:01.040 benefits of that technology. From there, they can determine Okay, 0:17:01.120 --> 0:17:03.720 how do I scale this? And again we can work 0:17:03.760 --> 0:17:06.760 with them very closely to determine. Okay, in the future, 0:17:06.800 --> 0:17:10.240 what are your needs from a compute standpoint that's going 0:17:10.280 --> 0:17:13.320 to allow you to really scale this new innovation, these 0:17:13.359 --> 0:17:17.359 new AI algorithms without really having to break the bank. 0:17:20.640 --> 0:17:23.760 Coming up next on Technically Speaking and Intel Podcast. 0:17:24.920 --> 0:17:27.400 I don't want to see healthcare just become a solution 0:17:27.840 --> 0:17:31.240 for rich people. I want AI to really be able 0:17:31.320 --> 0:17:34.359 to scale across multiple populations. 0:17:35.359 --> 0:17:37.840 We'll be right back after brief message from our partners 0:17:37.840 --> 0:17:50.600 at Intel. Welcome back to Technically Speaking an Intel Podcast. 0:17:51.000 --> 0:17:54.400 Let's pick up my conversation with Alex Flores and Peter 0:17:54.520 --> 0:18:02.000 Ship In season one of our pod, we talked a 0:18:02.040 --> 0:18:06.159 little bit about AI and privacy, and one of the 0:18:06.400 --> 0:18:10.560 I guess more contentious aspects is around patient medical history 0:18:10.600 --> 0:18:14.040 and their records. I like to get maybe Peter's thoughts 0:18:14.080 --> 0:18:18.399 first around the ability for AI to help centralize patient 0:18:18.440 --> 0:18:21.520 medical history and some of the dangers and some of 0:18:21.520 --> 0:18:24.960 the anxiety that people might have, you know, having their 0:18:25.000 --> 0:18:31.080 medical history cataloged and indexed and using AI and other algorithms. 0:18:31.600 --> 0:18:35.040 Yeah, it's always a tricky question, Graham. Yeah, that's why 0:18:35.080 --> 0:18:38.679 I asked it. And patient privacy here, but certainly I 0:18:38.680 --> 0:18:42.359 mean I think we recognize the importance of patient privacy 0:18:42.400 --> 0:18:44.760 and making sure that the patient still is in control 0:18:44.960 --> 0:18:48.119 of his or her data, especially healthcare data here. So 0:18:48.680 --> 0:18:52.000 from a Semen's Health in yourest perspective, as we develop 0:18:52.400 --> 0:18:56.720 AI algorithms and technologies that require all this data for us, 0:18:56.760 --> 0:19:00.719 it's important to establish that we focus on maintaining that 0:19:00.800 --> 0:19:03.920 patient privacy. And to that end, one of the big 0:19:03.960 --> 0:19:05.960 things that we do here at Seamans Health and HEARS 0:19:06.000 --> 0:19:09.080 is we've established what we call a big Data office, 0:19:09.320 --> 0:19:12.480 and what that big Data office is tasked with is 0:19:12.520 --> 0:19:17.280 actually to uphold the organization in terms of making sure 0:19:17.320 --> 0:19:20.560 that we respect that patient privacy tenant as it relates 0:19:20.600 --> 0:19:23.119 to patient data and data that we utilize to change 0:19:23.240 --> 0:19:27.240 these AI algorithms. So before we actually ingest any data 0:19:27.280 --> 0:19:31.480 into our organization for the purposes of developing artificial intelligence, 0:19:31.960 --> 0:19:34.320 all that data is actually quarantined, and what we do 0:19:34.440 --> 0:19:37.760 is we actually de identify all that data completely remove 0:19:37.800 --> 0:19:42.040 any PHI or PII associated with that data, even if 0:19:42.080 --> 0:19:45.080 the data was presented to us from either our clinical 0:19:45.119 --> 0:19:49.000 collaborators or other data sources. As being de identified, we 0:19:49.000 --> 0:19:52.119 actually go through the extra effort of de identifying it 0:19:52.200 --> 0:19:55.520 again before we actually utilize that. And then furthermore, we 0:19:55.560 --> 0:19:57.720 actually then make sure that the only people who have 0:19:57.800 --> 0:20:01.399 access to that data are folks who are actually developing 0:20:01.440 --> 0:20:04.439 the specific AI algorithms that they're looking to develop. So 0:20:04.920 --> 0:20:08.960 engineers within our organization have to declare what is their 0:20:09.119 --> 0:20:13.639 intention of utilizing the data for that AI algorithm development 0:20:13.680 --> 0:20:16.359 before they actually have access to the data. So we 0:20:16.400 --> 0:20:19.159 have a very stringent policy here as it relates to 0:20:19.359 --> 0:20:22.520 dealing with patient data. And again we don't ingest any 0:20:22.560 --> 0:20:25.480 of the data directly. We appreciate and honor kind of 0:20:25.520 --> 0:20:28.600 that relationship that the patient has with the provider in 0:20:28.680 --> 0:20:30.359 terms of what happens to their data. 0:20:31.040 --> 0:20:34.480 And then another example too is an INTEL One of 0:20:34.480 --> 0:20:37.960 the solutions that we created was around federated learning, and 0:20:38.240 --> 0:20:41.360 essentially it's really to kind of help address patient privacy 0:20:41.480 --> 0:20:46.560 specifically with data. So having the capability of moving the 0:20:46.720 --> 0:20:50.560 model to where the data is versus having the data 0:20:50.680 --> 0:20:53.440 move to where the model is, so really being able 0:20:53.480 --> 0:20:58.080 to facilitate that to help with that transparency of data 0:20:58.560 --> 0:21:01.680 so you can move that model get the benefits of 0:21:01.720 --> 0:21:05.320 being able to train that model on different data sets 0:21:05.359 --> 0:21:09.920 across various organizations and so forth, but still being able 0:21:09.960 --> 0:21:13.160 to respect the patient privacy. So that's an example of 0:21:13.240 --> 0:21:16.040 how we can work with Seemen's health and ears and 0:21:16.119 --> 0:21:18.080 the broader ecosystem in that space as well. 0:21:18.600 --> 0:21:21.960 Okay, and now thinking ahead in the future, I'm actually 0:21:21.960 --> 0:21:23.800 trying to figure out what that sort of time horizon 0:21:23.800 --> 0:21:26.439 I should give you, guys. But let's say once my 0:21:26.600 --> 0:21:29.480 kids have kids, so let's say twenty thirty years time, 0:21:30.520 --> 0:21:33.640 what do you think the hospital in doctor's office would 0:21:33.680 --> 0:21:37.960 look like in your minds using these sorts of technologies 0:21:37.960 --> 0:21:41.240 and obviously ones that are yet to come, you. 0:21:41.200 --> 0:21:43.560 Know, looking ahead in the crystal ball here, it's Seemens 0:21:43.600 --> 0:21:46.200 health in yours. Where we actually see the greatest potential 0:21:46.240 --> 0:21:49.919 for a technology like artificial intelligence is its ability to 0:21:50.000 --> 0:21:55.639 consume multiple pieces of patient clinical information, so really able 0:21:55.760 --> 0:21:58.439 to look at not just let's say, imaging data that 0:21:58.480 --> 0:22:00.840 comes from that X ray or that CT scan or 0:22:00.960 --> 0:22:05.160 MRI scan, but also looking at the patient's laboratory data, 0:22:05.240 --> 0:22:08.480 maybe their pathology data, maybe even their genomic data here, 0:22:08.800 --> 0:22:12.600 and then having AI actually find correlations in all that 0:22:12.720 --> 0:22:16.680 data to help the clinician make a more informed diagnosis 0:22:16.840 --> 0:22:20.960 or maybe a more personalized treatment for that patient. Now 0:22:21.000 --> 0:22:23.840 I can actually then go back to my broader patient 0:22:23.920 --> 0:22:28.000 population and look for other patients who might have similar 0:22:28.720 --> 0:22:32.720 imaging results or genomic results as my individual patient and 0:22:32.800 --> 0:22:36.639 apply that same treatment to that broader population with a 0:22:36.720 --> 0:22:39.480 higher level of a success. So here we're actually talking 0:22:39.480 --> 0:22:43.440 about true population health management. And then if you think 0:22:43.480 --> 0:22:46.200 about a gram like fast forward to those twenty thirty years, 0:22:46.680 --> 0:22:51.720 I could actually theoretically create a digital twin of that patient, 0:22:52.440 --> 0:22:55.199 which again is no simple task today but one that 0:22:55.240 --> 0:22:57.520 could happen in the future. But if you think about it, 0:22:57.600 --> 0:22:59.600 if I then had that digital twin of that patient, 0:22:59.680 --> 0:23:04.000 could actually start to now test certain therapies on that 0:23:04.040 --> 0:23:07.840 patient in this kind of virtual world here and figure 0:23:07.880 --> 0:23:11.879 out what's the optimal therapy for that patient on his 0:23:12.000 --> 0:23:14.800 or her digital twin, and then actually apply that to 0:23:14.840 --> 0:23:18.359 the patient with a greater level of success. And then finally, 0:23:18.400 --> 0:23:20.680 like if I can take that now digital twin, I 0:23:20.680 --> 0:23:23.359 could actually move all the way to the front of 0:23:23.400 --> 0:23:28.000 that patient's experience and really start focusing on preventative medicine. 0:23:28.160 --> 0:23:30.960 So rather than trying to figure out what's the optimal treatment, 0:23:31.440 --> 0:23:33.639 try to figure out what's the optimal way to prevent 0:23:33.800 --> 0:23:36.840 the patient from actually having to go into the healthcare 0:23:36.840 --> 0:23:38.040 system in the first place. 0:23:38.640 --> 0:23:42.240 Peter, you summarize that wonderfully. Two things I would add 0:23:42.320 --> 0:23:45.879 is one is also the integration of other data, so 0:23:46.000 --> 0:23:49.480 for example, maybe it's sleep data, maybe it's data from 0:23:49.560 --> 0:23:52.879 your wearable that you're tracking, or what you're eating, and 0:23:52.920 --> 0:23:57.439 so forth, to give you that really comprehensive view of 0:23:57.480 --> 0:23:59.640 your health, I think is what excites me the most 0:23:59.680 --> 0:24:03.320