Transcripts For CSPAN2 Jag 20240706

And cardiac electrophysiology fellowships at mass general, and also earn a doctorate from oxford university, a master of sites in clinical investigation from m. I. T. Harvard, and a Research Fellowship at the frames home heart study. Dr. Singh is an a internationaly recognized scientist, educator and lecturer, and serves as an advisor to multiple medical device and ai companies. Joining him in conversation is jim morelli, a general assignment reporter for boston 25 news. Jim fregosi worked as a freelance correspondent for cnn, specializing in medical stories out of the boston area. Jim is a graduate of the Massachusetts College of pharmacy in boston, and he also holds a masters degree in Civil Engineering from tufts university. O jim is the author two nonfiction books, one of which, poison, how to handle the Hazardous Substances in your home, generated and appeared on the Oprah Winfrey show. Tonight, dr. Jag singh is presenting his new book, future care sensors, Artificial Intelligence, and the reinvention of medicine. In this book dr. Singh explores the ups will a virtual caracol the evolving role of sensors, and impact of Artificial Intelligence in medicine and healthcare through cutting edge assignments, big idea projections, and patient stories. Future care provides us an insight into how health care can become sensible, affordable, and practical, and what agoa needs to become a part of the solution. It is an important and timely contribution to the ongoing and increasingly urgent conversation about medicine, technology, and healthcare. Jane moran, chief information digital officer at mass general brigham says of this book, future care resonates with my experience leading digital and Information Technology at mass general brigham, and isnd quite honestly a template for all Digital Health strategy. Dr. Singh shows the virtual carv and Digital Health on not only the future of medicine, but can transform healthcare as we know it. He rightly highlights how digital capabilities can help move the practice of medicine from treatment to wellness and prevention. And most importantly, bring equity to healthcare. We are so pleased to host this event here at Harvard Book Store tonight. Please join me in welcoming dr. Jag singh and jim morelli. [applause]an thank you. Thank you, natasha. For really taking the time to be here at 7 p. M. On a work evening. And its incredible that youre taking the time. Thank you all. I want to start by thanking jim for being here and again, taking the time. He has a busy schedule. Hes managed to work me into it. I want to thank the howard big scott bookstore and natasha most of all, cspan for for this recording. So delighted to be here. So were going to do this as i thought. Ill just spend 7 to 10 minutes or so just giving an overview, probably what natasha has already said in a different way and do a short reading of one of the chapters just, just three or four minute reading to kind of give a flavor of what the stories within the book look like. And then well have a fireside chat without a but will light up the fire. Jim mock q a me and then well talk and then well open up to the audience for another 15, 20 minutes. So you guys ask questions. Feel free to even stop us in the middle of. Our dialog. Feel free to stop me the middle of what im going to be saying. So happy to be interrupted. It gives me time to think then so. But thank you for being here. So let me let me really start with a little introduction just to kind of i natasha already mentioned this, but i would really like to tell you who i am, where im from, because that gives you an idea of where im going or how i perceive the field evolving. So i grew up in india i, i practiced medicine for a few years after, training in india. Then i moved across, did my doctorate at oxford, and i ended up here in the us at mass have been at mass general for the last 25 years with training through there. So im a part of the woodwork out there. At least i like to think so until they excavate me out and the reason i bring that up is that ive in three different environments ive worked within the indian environment, which is a Resource Limited environment. I worked the uk, which is a resource constrained and ive worked in the which at mass general is as a well resourced environment and you know, all of them have their nuances and each of them have their advantages. But i can tell you that none of them work really well. But im going to focus our discussion today on primarily the us system and how we can advance the us system. This future care strategies that talking about with the intent of eventually providing Global Equity because i think thats the far reaching goal that all of us are aspiring towards. So, so with respect, you know, starting on, on the us Healthcare System, i think its important to recognize is that it is big but its also fat and sick and. I think its fat and sick because almost 20 of our gdp or 17 to be exact, or 4 trillion to be even more exact of our gdp Goes Towards Health care, which is larger than. The you know, the the i would say the economy of some of the most developed nations in europe. And despite we have issues in terms of inexplicable costs really high mortality. I mean, Life Expectancy is actually much lower than many of the other countries. And when we start looking at the quality of care metrics, were fairly indefensible. And i think much of that is because we are to some extent and i it pains me say that we are inept, we are ineffective, we are inequitable. To start off with. And moreover, system is very opaque. Our system is very opaque because its opaque in how we reimburse medicine. Its opaque in the way medicine is pract just because theres a lot of variance in practice across the country, across hospitals, across institutions, across the same region. And that in itself inflates the cost and i think thats been a big, big issue out there. So medicine right now. And the reason im excited is a state of transition. We are having this digital transition, right . And this digital metamorphosis will actually enable us to enhance the delivery of care and make it more equitable. And i think its possible now because we do have we do have unlimited connectivity now. We have massive Processing Power and we have ubiquitous data. And putting that together, we now have an opportunity of enhancing how we deliver. And with this Digital Strategies, the care is immutable. So that makes it very transparent and at the same time makes the cost of it also fairly transparent. And i think putting all these things together, there is hope in the future and the future of care as we look at it today. Now, the soul of this book is really in forecast, dying and predicting disease. And i think that is largely if i can leave you with one message today, that would be that the future of care will be partially virtual will be sensor aided will be powered by Artificial Intelligence or Predictive Analytics which, sustainable workflows in the hospitals that will translate into better outcomes. So i know its a long twisted sentence, but there are three components to it. Basically virtual care sensors, Artificial Intelligence and workflows, simple strategy. And lets break it down. Im going to break it down very quickly because i know there are time constraints out here to really get the discussion going. I think the most important thing is is the is is the sensor strategy. And i think all of us already are well immersed in the sensor strategies. Those are variable sensors. Theyre watching their earbuds, their necklaces, their caps. All of them provide Digital Information in that same brand is realme. There is the implantable sensors that i as implant devices, patients and these devices that pacemakers and defibrillators have sensors within them that provide us information regarding physical activity of lung fluid give idea of what their temperatures can and whether they are going to have an impending event in the near future. And some of these sensor strategies can also help us predict Heart Failure in patients well before it actually occurs. Right now, the important thing, sensors, is and this is really interesting understand is that every sense in our body, every analog experience we have touch, temperature, pain color. These are digital experiences, right . Because they are all sequenced in binary codes. Every cell in our body, electrical activity, every cell is connected to the other cell through electrical activity. And these binary codes are these analog perceptions, digital signals that go to our cpu or gpu or whatever you want to call it, our processing up here, which is really the interface of all these digital signals. And thats why i think sensor strategies allow us the opportunity of creating organ specific sensors that can create appropriate digital dashboards, that can allow us to look after patients in a very individualized way. Now, the next part is very quickly is, is you get all this data from all these sensors implantable and wearables. And, you know, for a spectrum of disease states. But how do you kind manage that data . And theres really nice quote that you can have. You can have data without information in which you cannot have information without data. So its so important to actually be able to use that data appropriately in managing our patients and giving the right kind of care to them. So the question is, what is that data . So i think the data we all recognize is the data that is there in our Electronic Medical, right. We all know all all our age, sex, demographic data, medicines all that stuff is there. We recognize that. But theres this whole invisible data now that we dont recognize that really is the bigger contributing to disease and that is social, cultural, Environmental Data from the wearables right now is still something that is dark to us because its not all integrated into Electronic Medical records right now. But there are strategies that are developing, can layer it. So will be approaches that we will be using a. I. To look at, not only the visible but also the data because we know machine eyes can things that we cant see. And just to give you an example now you can look at a simple electrocardiogram a 12 lead ekg and you predict which patients are going to develop atrial relation, which patients are going to life threatening arrhythmias, which might actually develop a stroke in the future from a single standard to a lead. There are algorithms now that are being developed that can help us predict events in the future. Now if you put together with many other diagnostic imaging modalities, you can only imagine that the perfection of actually being able to predict things in the future actually gets enhanced considerably more very quickly. I think its really important that alongside the the the use of sensors and ai there have to be in how we use these technologies to manage care change care and thats youre going to have many future models of and i look forward to chatting about them. But one of the standard approaches that we actually as cardiologists already in patients who have implanted is something an exception based care. Its a new concept but that is care. That is given only when the patient actually needs it. So if you have sensors that can actually continuously surveil patients, it is only when those parameters are out of whack that you say, okay, i to bring the patient in and see the patient so you dont have to see the patient. Six monthly 12 monthly, two yearly intervals when theyre not unwell but see them at the time that they actually need to be seen through the strategy of exception based care and i think thats a concept of medicine that is gradually and i think as we get more to sensor based approaches, better predictive strategies using Artificial Intelligence, we will be using some of these continuous approaches where we can provide timely care at the right time, the right person at the right location. Okay, this is beyond cardiology. This is every possible disease state, whether its diabetes, hypertension or chronic obstructive disease, has the opportunity of providing continuous tracking using sensors and approaches to actually provide individualized care. Now, beyond, there are these whole concepts that we can delve into and wired which is there in the book. Its called manage your own disease, because i think health care is nonsense, sustainable. It is nonsense attainable unless patients have some skin in the game, it is impossible to manage disease for patients. But if you allow patients and empower patients or allow patients to empower themselves to actually help manage their disease, it is the only way that medicine can actually become sustainable. And some of these sensor based strategies, a ai based algorithm that will allow individualized output patients can enable some that care. Im not going to get the details of digital twins but we can talk that thats the next future of how you can actually create virtual replicas of individuals out here. Some of this is already happening. We are already creating of individual organ states. So you decide whether a certain intervention is going to translate into a better outcome. But you can imagine digital twins which are virtual replicas in physiology. The anatomy biochemistry along molecular and genomic information that. You can actually simulate disease and, test the ability of an intervention to if it actually makes a difference. So really stuff out there and the last thing im going to touch on is the current existing care strategy of systems that we see where all the hospitals and their affiliate and their Surgical Care centers are all integrated into, a system largely to enhance cost efficiency and effectiveness of delivery of care. The next era is going to be networked this way. The social order of medicine is actually going to change. And this is going to change because of the expectations of patients. This is going to change because. Our patients are going to are going to expect that they get care wherever. They are whenever they want, for whatever they want, wherever that location may be. And thats the future of where medical care will eventually go. I know that seems like utopia at this point in time, but there is a drift towards net network ness already. So im going to start by saying that, you know, we can talk about all these phenomenal and i think its cool stuff, but these are all to the human touch and the human bond. They cannot replace the empathy and the human touch that we encounter as clinicians every day. And you as clinicians or patients to on a daily basis, i think medicine always will will will have that human there. And id like to close off by this really nice saying from Francis Peabody who says that the secret of care lies in actually caring for the patient. And i think that can only be possible if the human touch is preserved while you use all these sensor based strategies and ai based outages to provide care. So with that, im going to conclude my my sermon, but im going to actually go on to do a short reading of the book so i can we can then sit down, just have a chat so that okay, thank you for bearing with youll have to continue, to hear me for another three, three and a half minutes. So i picked up, you know, i almost dropped the book down, said which, which page opens up. So i picked up one of the ones on creating the air culture this is chapter 11. And this is a little, you know, so every, every chapter in the book i write about 20, 25 or so, all of them start with patient stories and their intertwined with how those patient stories translate into how outcomes and Clinical Care needs to evolve. Right . So it kind of provides that. So theyre peppered with stories all over. So this one starts with this quote from, charles lindbergh, who says is like a landscape. You live in the midst of it, but can describe it only from the Vantage Point of distance. Okay. I met victoria for the first and last time in the of 2019. She was a 73 year old. Strikingly majestic black woman who had air of confidence that made it clear me that she was the matriarch and the boss of her family. She was here to see me with, her 24 year old granddaughter, alice, her chaperon for the day. I had been asked by Oncology Team to see victoria for increased shortness of breath and swollen, supposedly early manifestations of a weak heart. But she clearly had bigger problems that day. Victoria was wearing a long sleeved light yellow tshirt, imprinted with several faces, intrigued by her t shirt and in an attempt to break the ice, i asked her about the faces was wearing. I was taken aback when she introduced of the seven faces as our family members, three brothers, two sisters and two daughters, and began recounting how each of them had died of either of cancer or Heart Disease. She carried them around with her on this personal journey said that it gave her faith strength and continually reminded her to be grateful for every time. She might have. She might still have. She also told me that she was not afraid of going to the other as she knew they would be waiting for her. She pointed out that alice was accompanying her today, had herself survived leukemia at the tender age of seven and was studying to be nurse victoria now had recurrent Breast Cancer. Shed been diagnosed 25 years ago received chemotherapy. And since then she had had two surgeries with a left breast mastectomy and reconstructive surgery. Her weak heart now most likely a result of the chemotherapy, had received. She what we call a classic case of the triple whammy. First, cancer, now cancer again, and then Heart Failure. She was told that she had a recurrence in her right breast with evidence metastases to the lungs and signs of fluid building up around her lungs and her heart. She was here on borrowed time. She was very direct in her line of questioning. She wanted to know whether another of chemotherapy with a weaker heart would hasten her death and make her even more uncomfortable. I was not surprised. At the end of the clinic visit victoria. It clear that she was not interested chemotherapy or in any form heroics to help her weak heart, which is what i was seeing her for. She died three months later. There were so many things that went beyond my simple understanding of clinical medicine, why did she have a recurrence after so many years, could have been predicted . Could an earlier intervention detection have saved victoria from the discovery of metastases all over her body . Was there a way we could predict, which patients would develop Heart Failure from chemotherapy . Are there simple baseline predictors of, chemotherapy induced side effects . Are there downstream events . Granddaughter might have to face that our clinical clinical insight cannot see. Why did all her family members have such poor outcomes. Is there an interaction, race, genetics and cancer was overlooked or not detected in a timely fashion. Can i help us here . Can i help . See below the tip of the iceberg to where the vast body of the unknown that constitutes the true of evidence lies. How does one gets ones head around this notion of beginning to use . I lets try to break this down. Im going to stop here and then i break down. Thank you very much. I got that. Yeah. All right, jim. Okay, so this is. Yeah, yes. Okay. All right. All right. So i cover and mayhem five days a week, and so i never take anything for granted. Air is a term everyone has heard. Can you succinctly define air tell me what its capabilities are now and what they could be . And you may have done this here, but there was no there. No, no. It totally. Can you guys hear me . Everybody hear me right. So, you know, i, you know, when i when i think of air, i still think of the air in air as as additional intelligence or augmented intelligence or assistive intelligence more than artificial. But the Artificial Intelligence is real and its happening. I will talk about that in the next few. But when you look at a. I. , if you really want to subcategories it, i categorize as conventionally ai and within conventionally i categorize it as narrow ai. And generally i. And then we have generally by which i think everybody is hearing about with gpt and all the Large Language Models out there. So i think from the narrow a. I. As an electrophysiologist implants, devices that can treat patients with sudden cardiac death. So i just want to make that clear that the icds or the defibrillators we implant in patients work off a single cord that can do a job better than human intelligence that is pick an arrhythmia and shock arrhythmia. Thats a form of a. I. That is a single code that already exists and has been there for 20, 25 years and is getting continually refined. The conventional a. I. That talk about out here and i think there are several ways to actually slice this. And i could go on for hours on this. It could be either in the diagnosis of disease, the prediction of disease, the treatment of disease at this point time, one of the things that i mentioned in my talk was that you can use ekg to predict events 25 years down the line. Thats something the machine can see in an ekg that the human eyes cannot. We can only diagnose whats happening right now, but were not able to predict whats going to happen. And some of that deep Neural Network that can look at multiple of ekgs and come up with these algorithms predict is a form of conventional ai. There are many other forms of conventional a. I. And i think some of them are now already being used in helping for example, let me just pick up since we were talking about cancer, if you ask what are the uses of ai and cancer, it could be and is being used now for early diagnosis is right. The second is its being potentially used for Precision Medicine where it uses genomic and molecular data to provide more specific directed targeted therapy individuals. It can used for Drug Discovery, which is already is being used for Drug Discovery for many neuro chemotherapeutic. And besides that imaging, im circling back to imaging, looking at scans now, you can predict which patients may be at risk for certain cancers. Well before cancer actually occurs. And some of that work is even happening. You know, cancers like pancreatic cancer are necessarily know are fatal. But if you can predict them now, i think the potential of changing the course really is around corner. So are we talking about you mentioned sensors. Are we talking about implants in . Every patient of some sort. And and which organs . How would you choose the organs going to implant to give you information on . Right. Absolutely. So i think when we talk about sensors i think its important to categorize them as implantable and wearables implantable are people who already getting a device or a certain condition, whether its Heart Failure or whether its know you have a pacemaker or need. There are sensors in those devices that already exist at the other implant sensors that we do implant right now, our large need to pick up heart rhythm in individuals and theyre selectively in the chest area and thats a kind of an organ specific sensor to the heart where it measures heart rate arrhythmias at the same can measure fluid around the lungs. It can measure physical activity, it can measure respiratory rate and. Now, some of the newer sensors can even measure the tidal volume that is what your respiratory rate is and whether in Heart Failure or not. So those implantable sensors. But interesting thing is that there are a slew of variable sensors now. You can use the watch, for example the watch. The simple watch can give you your ekg, it can give you a heart rate. It can give you your physical activity, can you your oxygen saturation. There are watches now that can measure the electrolyte changes in your sweat. They can measure autonomic tone they can measure hydration status. So so i think as we and then there are for example so thats thats the watches but the same watch can even measure gait stability or instability of your monitoring a patient with parkinsons if youre monitoring a patient with neuropathy there are other that can be variable now that can provide information. One of the commonest one that i think many people know about is is diabetes. There implantable or semi implantable monitors that you can stick to the skin, little electrode they can measure your glucose levels on a continuous basis and can provide you the opportunity of managing your glucose levels continuously. Its not every morning or every afternoon when youre finger yourself, but continuously and now im just going to couple this with a part. There are ai based strategies that link these cgm that is these continuous glucose monitors with implantable insulin pumps work in concert that the the glucose levels informs the pump as to how much insulin it actually needs to give. So you have ai based approaches that are working in concert with sensor based approaches. So so of the the long winded answer or the short answer is that there are individualized variables for individual conditions that eventually become a part of practice. And then we will have Remote Monitoring platforms where all that information will be integrated, where you can see that patient in the form of a digital dashboard beyond just a single disease, but their whole entire body structure and organ problems. So theres a very moving story in the book a sad story about your friend maya. Is that her name . She was in their thirties and developed out of the blue pancreatic. No, no history or anything like that. And died within four months. And i think you mentioned in there a sensor would be great, but you got to know what youre sensing for, right. And there are conditions for which we dont know. Right. Right. So unfortunately for pancreatic cancer, there are no clear evidence sensor strategies of yet because it comes on unheralded and people dont know when they get it. But there are now approaches using imaging that can actually look. So what they did was they looked at patients with who had scanned images of their pancreas and who developed pancreas, pancreatic cancer at a later stage. And then looking normal scan images now, they can predict which patients actually went on to develop not the human eye but the machine. I can predict that. Its very interesting. The machine i can look inside your retina and while looking for eye diseases, it can look at your inside and predict whether you are going to develop sudden cardiac death or develop coronary disease or develop other atherosclerotic Heart Disease issues just by looking inside the retina. And thats the machine again. So so youre absolutely, absolutely right that that we dont have sensors for every disease state. But i think thats where medicine is moving towards i think all the care we deliver even its whatever virtual form of care we deliver has to be objectified and that objective i think that if that maybe award is going to be through sensor strategies, i would think you would want to know whos most susceptible to the diseases, otherwise youd be scanning everybody and looking for everyone to get it right. Absolutely. So. So i think there are Machine Learning approaches that we can currently use of our Electronic Medical records and even off data and sensor based data that you can risk stratify patients. So so there are so what are the things we did with sensors . This is 2009. I think we were amongst the first to actually come up with a risk score for using sensors from within the devices and. This is a paper i published a long time ago which looked at heart, which looked at physical, which looked at autonomy tone and created risk score from these three measures from an implantable device by the categorization, we could predict which patients were going to die. One year, 25 of the ones with the risk score four died within one year, and we could predict which patients were going to do just fine so risk categorization is going to be important, but not through implantable sensors necessarily. You can even do them. Looking at the conventional medical records, if there is enough data out there, you can create risk stratification algorithms to help you decide which patients need to focus in on more. So you mentioned patients having so you mentioned patients having skin and game and how important that is. Thats important now but we have a population where it doesnt look like theres toooo much skn and again. We have people, obesity office and contribute a lot of disease, lack of exercise come poor diet, et cetera, et cetera. What do you mean different by that . Or do you mean anything to . Absolutely. When i say skin in the game its really important for patients, for example, issues example of diabetes, right . You have a patient who is diabetic and doesnt look after him or herself, will develop complications of diabetes, whether its peripheral disease, neuropathies or Heart Disease. Their impact on Health Care System because of not looking after themselves is significant, right . Whereas on the same page about somebody who looks after themselves and they need to get some sort of credit for being engaged in the own healthcare. And some of this is already happening where Insurance Companies already incentivizing patients to look afterem themselves by adjustingiu their premiums, if the wearing a wah and recording their physical activity on a regular basis and providing that information, that means they arere adopting healty lifestyles, and the Insurance Companies now are figuring out that if they adopt healthier lifestyles, the likelihood of them getting admitted for untoward events is much lower so theyre going to save money for the Insurance Companies. As result oftr which the incentivization strategies are evolving. Some of those evolving strategies are where patients will automatically have more skin and again. Now having said that, that every single course incentivization. Patients themselves dont feel empowered enough in the way medicine is practiced right now to know how to look after themselves. I think it isus incumbent upon s asce clinicians to be a part of that process to help patients understand, to get more empowered, and look after themselves. Because thatsdu really what we can can reduce the burden on Healthcare System on the clinicians and patients more engaged and involved in their care. So my wife isle a nurse at Newton Wellesley hospital. She works in postpartum and every night when she comes home i hear more stories about how they are short staffed, they dont have nurses to care for the patients. One nurse has six patients that night. That obviously is a continuing issue. How can, or can ai alleviate that . Out of me by replacing people necessarily, but by perhaps, i dont know, is there a way . Lets use the same example that you picked up of within the hospital floors. How will i felt that particular situation . Y two ways immediately come to mind. One is, for example, i work on an arrhythmia floor, and arrhythmia floors have monitors that have alarms are constantly going off. Theres alarm fatigue for our nursing staff and for the clinicians out there. And a part of that is because those alarms are erroneous. Oftentimes artifacts are just patients believe their arms, can set up the wires and have alarms fatigue. Ai strategy is becoming into play to reduce set alarm fatigue and help really pick out the true alarms out there, reduce the false positivity of those alarms and enhance the positive Predictive Value of those particular signals. So thats one. The second is actually, i forget which one ital is, trinity Healthcare System is already usingir within nursing they are using virtual nursing at the same time alongside life nursing. So that is on the floor you have a certain number of nurses but you also have virtual nurses on the same floor i can connect immediately with patients in the rooms through videobased strategies, but also provide advice at the same time while the other nurses are able to do their other work. So there are going to be shift in how work distributions going to occur. So when we talk about the use of ai, i think there are going to be approaches in how we repurpose, redeployed and redefine some of the job structures of how we do, not just nursing, even operate as clinicians, with an healthcare landscape. Much of that is evolving at a think its going to transform theen way we are actual practice medicine. Do you think with ai taking kuwait role we will still need more healthcare workers . Absolutely. I thinky youre going to continually aspects i think whats going to happen is, is that the jobs will be redefined by the use of ai but i think i actually going to enhance our ability to spend time with our patients. And i can give an example. One of the things that really drags us down h in clinic when e are seeing for patients and have a couple of patients powder, they will say jag doesnt make eye contact and hes on the computer. When theyre seeing me. I try to avoid that but that happens 90 of the time. Most clinicians are looking at the computer and not at their patients. And thats because they are trying to get the note to done because they do want to do it at 9 p. M. When they go home. After seeing 20, 25 patients. There are already large language model strategies, or chatgpt equivalents that will allow us to generate those synthetic notes during a visit would you dont, its what we call keyboard liberation. Two you are getting rid of the keyboards, you are and chaining herself from the mundane activities of prior authorizations and building and prescriptions which will allow, so ai has potential actually to enhance the physician and Patient Experience or a clinician and Patient Experience overall. Think the jobs will go way at all. I just think they will be redefined. Some of the middle managere jos may go and the administrator jobs make it impacted by that but i think but the clinician jobs i think would just be redefined what they are not going to get significantly impacted. It makes me think a pharmacist for example. Pharmacist, lets fill prescriptions but now the routine is more they handle the clinical aspects of it versus a mechanical. Absolutely. I think the pharmacists now are getting a chance to work at the top of their license. Correct. Which you didnt have for so many years and thats because of ai. And i think we cant survive without talking to ourdr pharmacists not because it tells weut drug interactions that dont remove off the top of our head and a part of our clinical rounds right now which they couldnt in the past because i they had to do all the work in the past but now there are some forms of liberation from some of those tasks. Obviously in the future we are whenever different generation of people seeking healthcare as they get older. In your book you have this kind of depressing thought. Theres a new generation of healthcare consumers who have no time for complex personal interactions. They would rather avoid human contact, some younger consumers are also letting go of their primary care physicians for the sake of privacy. I cannot getng the demographic youre talking about here but i wonder, to discard your pcp, wouldnt it be better to have a relationship with your doctor early versus trying to establish one later on, especially given we talk about preventative medicine . For sure pics i think thats all influx right now. I mean if you look at many of the large academic centers, our primary care centers are actually i, would say divorcing themselves from the hospitals. There are more thirdparty vendors that are now becoming part of larger organizations inviting primary care patients require. But the thing is with the Younger Generation they want care now, ay, or they want yesterday. They would rather see their doctor or any doctor just to get the advice or comical or something. I dont need to see their same primary care physician for advice and wait three months to see that particular doctor or clinician or Nurse Practitioner or physician assistant. So i think the expectations are changing and i think that is where i i feel that theseso continuous surveillance strategies come the sensorbased approaches with automated ai algorithm approaches along with healthcare navigator will help, reduce the burden ofid both ends providing instantaneous care. I have an example in the book which is actually kind of a funny story where i get page on a saturday morning at 6 55 or at 6 59 a. M. And im so irritable one of picking up the phone and say oh, my gosh its 7 a. M. On a weekend, im getting paged. And then it hit my patients love the voice and i melt there immediately and she had been waiting all night long to ask that question. She missed her blood thinnerr e night before and shekn did know whether she needed to double up her dose orhe not take the next dose, take theni dose in the morning. But can you imagine the angst and anxiety it had on her mind for seven hours from when she missed her dose and what she should do the dishes worried about it. And it wasay not my day on calli just had not signed by pager at an obviously a f question i need to be answered, i felt that it was a really good thing that it did. I provided care. Care that i provided by relieving her anxiety. She felt she got valuebased care. But the insurance systems, the hospital, seeing that kind of transaction is fluff and is really meaningless in the books of medicine. Because you cant charge for . No, its not appreciator or charge or incentivize or recognized. Now, you should say medicine is alter mystics or you should just be happy that the patient recognize you for your time you spent, and thats fine that truly is what inspires us. That i think that doesnt inspire every clinician. And i think its important to recognize that that question could have been answered by a simple chat bot or aibased strategy at night immediately within 15ro minutes of her havig the problem, as she would not have to suffer through the eight hours of anxiety and then paid me on a saturday morning at 60 to 90. When you get a page at 6 05 09 a. M. You know someone is waiting for the clock to turn 7 a. M. Because they are trying desperately to get in touch with you at a reasonable hour. So answering your question, i think there will be aibased approaches that will change the way with actual operate on oa daily basis. These are things we need to factor in because, and the Younger Generation of patients, and even clinicians, have a different approach to life. And i think many Patients Want care wherever they are at any point in time, and theyre happy to see another clinician if they can get that opinion. And they dont necessarily need to see their same primary care clinician. Ifat you have continuous surveillance strategies, since abased data that are i would say input into the emr, you can have ahe strategy or anybody can pick up the care out there because you have a longterm narrative because right now the primary care physician narrative that you have is broken up into 12 minutes of a discussion every six months to a year, and you are expected to know their entire life and that he or my to get sensorbased approaches that provide a continuous narrative, supplemented by that individual contact, i think you will have a much better representation of who the patient is, where they are and where they are going. It sounds like you think the personal interaction is still going to be very important. Absolutely, absolutely. I wanted to ask you about covid. If you dont know, dr. Singh was one of the first patient in massachusetts who develop covid and yet quite a time with it and wound up in the icu in fact. I wonder if the time when you were very ill and isolate acuity committed by ipad, dated influence yourr writing of the booklets were you sitting around thinking, watching things and seeing how they been better perhaps . For sure, for sure. That was a tough time obviously because in march of 2020 there was a fair amount of uncertainty, you know, whether you going to make it or not take the wrong left turn because we were seeing all these images from spain and france were people were out on the streets dying. But the care i got actually was phenomenal. I was fortunate to be one of the best hospitals in the country at mass general, and the care i got was from the nursing staff who were there all the time and the physicians not necessarily so. And i think it changed my perception of care considerably. It gave me a different Vantage Point where i was now look at life not justt from one side of the bed but even from other side of the bed. And i think that certainly, had a Significant Impact on the tone of the book because i realized that you can talk about the best of technology, but if it doesnt have humanistic appeal, its kind of meaningless. And i think what gave meaning to my care when i was down with covid was the nursing care. And i think that human touch really was impacted my writing tone. And i think i modified my tone a few times when i was writing this book since this was my first nonacademic book. And i think it was very much influenced by my covid experience. Finally leading internet, you dedicated the book to nurses. Can you tell us that is true, that is too. So i did dedicate the book to nurses, and to think my wife was thrilled. I think, you know, i think nursing is undervalued. Very often in the care we deliver. The clinician comes in, spent five minutes, maybe ten minutes and the remainingng 99. 9 of the time byly the bedside is the nue who actually delivers that care and actually looks after that patient. And i think that was very selfevident to me when i became a patient, a kind of gave me that Vantage Point and i realized how much we undervalued the care we give. Any think its really important to recognize, and i recognize that at that point in time for sure, is that medicine is a team sport. You know, nurses are really aninvaluable team members who cn play any position in that team sport. They cannot just lead hospitals. They can be at the front line looking after you when you have covid, and they can play almost any position in that team sport. It truly phenomenal to see them in that action but it seemed to recognize their undervalued. But i will say this and its not me, someone said this, that the culture of an institution and the happiness of an institution is measured best at the bedside by the nurse. Very good. Thank you, dr. Singh. You have cleared up a lot of things for me. Thank you much. Do we want to open it up to the audience . Yes. So thank you for the talk. I am very familiar with the technology of insulin pumps and sensors at how much this is in the cloud and patients can login and additions can log in remotely. Family member contract and call the daughter in a different state if they have low sugar in the middle ofnt the night. So we got to that extent that patients and families can control it remotely. Do you see, and this is wonderful. Do you see any ramifications of ai regarding data hacking and data privacy . Lets say somebody come to c a three people can hijack two planes, a terrible tragedy. One person can hijack 10 million pumps. Half a million pumps and deliver too much insulin. Current ai check . Could ai have layers of interdependence between independence of an autonomy of a person and ai layers . Would you see the problem in the future . I think thats a really good question. I think cybersecurity, hacking of algorithms are a big problem, and it something that we need to be aware of. The potential certainly does exist. Im going to kind of sidestep after and ill come back to your question. Ai is a form of an organic light they can actually take over organic life, thats the human, right . And much of that is he blames it all saying that if the humanca brain can be engineered, its packable, too. And so what is a pump . A pump can be hacked is pretty straightforward. I think thats importance of the organizations that are actually felt pumps is to create. The best way to fight a isbn offense and and fight with ai. And there are ai strategies that most of theseav Device Companies have in play that prevents people from getting access to those pump related data. There are layers out there that prevent that, and at least come for example, in implantable devices that we use you can still pack a a pacemaker or defibrillator project to come within, less than ten ten centimeters of that device to actually be able to hack it. So that our guardrails. There are barriers to hack ability anything with respect insulin pumps alsoin its something that the industry is very aware of. Its just like, you know, automatic cars. You are kind of always worried about these cars, whether someone is going to take control of your car that using the movies, youre going to lose control of them completely. So i think your question is valid. I think its a legit concern, and folks are making sure that they have the appropriate layers of added to prevent ai from some of these algorithms are also self learning algorithms, not the insulin pump once but again thats a problem that needs algorithms start teaching themselves can sometimes get out of control or the potential of them getting out of control is also their. Let me follow up. If that does happen, if the Artificial Intelligence makes a mistake, when ami doctor makes a mistake and we know who to sue, who do we submit ai makes the mistake . The company . Who is liable . Its a very hot ethical debate right now, and much of it in the realm of robotic surgery, for example. You have robotic surgery or automatic algorithms that allow you to do certain surgical maneuvers, and you have a disaster, is at the robot, is that theom physician, is at thei company or is it the company that actually created the hardware . Right now its the physician, and the organization because they are the final word because its manual override. But there are discussions out thereus as to, because there are many nuances to this and there are some that are completely automated algorithms and the armys are automated but have manual override. At this point in time theres nothing that is completely automated in medicine. We try our best to avoid that unless its something with a single code, but using ai as a blanket strategy is still has a lot of manual override issues. Engagement. Patient privacy concern, you abort in three different countries. One of the things that would be really important for Ai Development is very large data sets that truly make, make the algorithms reliable. The converse of protecting data is making it easily shareable. Do you have any idea what is currently being done to implement International Rules for data sharing . What are the barriers that are being reduced to help speed up the algorithmic finetuning . Sure. No, no, absolutely. So i think again this is a work in progress but i can truly some of the International Work ive done. So i work with data sets from france, india and the u. S. In a combined data set where we had the only identifiable information we had was age and gender. We have nothing else that we knew about the patient besides their ekg records. Some of the work i did was actually off a patch monitor off the first 24 hours predicting which patients were at risk for sudden death inn the next 30 days. C. Really really really good predictive ability for, sudden cardiac death in the short time period but that was based off a collaborative dataset from countries the problem is that thats a a single covariate, right . As a date, what youre looking for is to have large datasets that large data sets that really are meaningful with the invisible data that we talked about, with environmental cultural and molecular data. Some are the fed rated approaches where the institutions actually dont share their data, they keep them in their own institutions. They create the algorithm, they share the algorithm and allow the algorithm to be tweaked at each of the other institutions so you maintain possession of your data. Thats one of the strategies and obviously others that kind of bring many data bases together. But the issue out there again is especially in the world of generative ai that you can identify or deidentify, you know, data sets and thats a risk that people are still dealing with and there are the safe ai forums that are underway, actually specifically addressing the question that youre asking. So at this point in timed fed rated approach is one of the ways and the other is limiting the amount of covariants that you share with other organizations. Not the data, not furthering the whole objective. I have a followup question. Of course. And then ill let other people have a chance. Sure, please. Another challenge, i feel, in pushing ai to routine use in medicine is actually having physicians validate data and of course, weve discussed how taxed the Health Care System is, especially in terms of resources and time. One of the challenges that ive run up against is having physicians being willing to sit down and comb through lots of data and validate, which is difficult work and a time consuming work. What do you think could be incentive or do you think in the future will be dedicated like a subdiscipline especially for overseeing that within radiology or how do we make that something that is more done, i guess. Absolutely. So just for the others, you know, the question that is largely being asked is, how do you really truly have the ground trust that is identified and validated by a. 0 institution by a certain event and there are organizations specifically that are specifically, you know, they have data scientists who are advocating and creating data sets, obviously, data now is in a common thing and dont want to share it. If theyre ahead in their data preferably not share it at all. Thats in the work force. I can tell you as a clinician, i dont have the time to even remotely care i hate to say this and im saying this on national tv to really ensure the complete veracity of everything i put in the note. I will have a nice summary statement, but i dont go into the details absolutely making sure is this research worthy data that im entering into the Electronic Health record and thats a problem. And thats a problem and i think thats where generative ai will have a role so youre not going to need data scientists to help annotate that, youll have one to separate them, cureate them and do the work that clinicians dont have the time to do. Yes. I was just wondering if you could expand upon a little more about the racial barriers that exist in health care and how ai can break those down. Absolutely. So i think covid gave us a big microphone to the social determinants of health and to correct some of the wrongdoings of systemic racism that existed for centuries. I think what Digital Health does, it gives us the unique opportunity of looking at this proactively in a perspective fashion. So we can create Digital Strategies that can enchance equity and allow dissemination of care that does not have the undercurrents of racism. The good thing about Digital Health, you can actually code an algorithm to not be racist, but its very difficult to change the mindset of human beings, so you can actually, i think, make significant progress using Digital Strategies in a perspective way, not subsequently going in and filling in the gaps that we do right now, but prospectively ensuring that those issues are being addressed. Okay. Go ahead. Please, go ahead. Absolutely. Hi. Sorry. So i was just thinking about you were saying that ais currently also being used to predict diseases like 10 years down the line and im thinking as a patient, if i knew that i have, and im at a high risk of having a disease 10 years down the line it will make me more paranoid instead of trying to do things to like maybe trying too much to prevent that and then when im worried then i can get sicker and i can even like keep calling the doctor saying, is this wrong or that wrong . What an amazing question. What an amazing question, that was a really elegant question. You know, theres this whole concept of survivor and provider. Survivor is when you have a disease and you survived it and now youre dealing with it, youre dealing with the consequence of having had that disease. When youve not had the disease, but when you know youre at risk with that disease how do you deal with that whole evolving landscape of medicine. You know, its better to know that you have a disease than not know that youre going to have a disease in the future. And i think what will help people understand is how do you create the right probabilities around that. The potential for having a disease, 10 or 95 . If it 95 , lets do something to prevent it. So, i think it arms you with the opportunity of actually intervening in a way that can prevent disease so there are two sides to the story and i think the whole component with the Breast Cancer gene or at braca gene, and family live with the fear of it and so people take proactive action or wait for it to show up. It gives you the opportunity to make the decision. Seeking help and how to use that data rather than get paranoid about it is important. Its going to be a whole evolving platform of care as we deliver it. Last question, all right. So, a lot of current diagnostic techniques, like geognomic testing and the data selection to which theyre based how do you ensure that ai algorithms accurately treat illnesses and access to health care and skewed and how do you make sure access to ai service is equitable. Thats a long question. Do we have another hour out here . I would say this, for example, watson. Watson was the computer, super computer strategy that i think that Sloan Kettering created with ibm to create a strategy to predict and treat cancer and didnt work and it crashed and burned. The reason it crashed and burned was there were gaps in the genomic data in gaps in the clinical data, and how you would approach patients with cancer, and approach with late cancer, but lack of uniformity, but how to approach patients with early cancer and the Decision Making and engagement of the clinician and the patient is important. I take your point. I think theres a lot of ground to cover. Im not going to break your question into all the five parts that you had, but ill say that theres lots of ground to cover in terms of equity, in terms of getting the right data sets, in terms of making the data sets generalizable without bias because i think its so important that if youre using an algorithm off a particular data set for a particular patient, if that patient was not represented in the data sets that was creating that algorithm, you know, its not going to work. And there are so many instances that i could go into to exemplify that point, but i know were running out of time, but i would conclude by saying that we have a lot of ground to cover and were moving in the right direction. [applause] thank you very much. Weekends on cspan2 are an intellectual feast. Every saturday American History tv documents americas story and on sunday, book tv brings you the latest in nonfiction books and authorities. Funding from cspan2 comes from these Television Companies and more, including sparklight. The greatest town on earth is the place you call home, and right now were facing our greatest challenge. Thats why sparklight is working around the clock to keep you protected. Were doing our part so you can do yours. Sparklight along with these television cpa