Closing Healthcare Gaps with Technology w/ David Feinberg, MD
Dr David Feinberg was CEO of Ginger and UCLA health then LED Google Health before becoming CEO of Cerner for the past year he's been chairman of Oracle health so let's just say he's got a senior senior level perspective on healthc care provider systems and information technology today he joins us to discuss interoperability Ai and Health [Music] Equity welcome to car talk I'm America's home for incisive debate about Healthcare business and policy I'm David Williams president of Health business group and I'm John Drisco the president of Walgreens health have you checked out the care talk newsletter lately it's the source for trending healthcare industry news links to the latest podcast episodes and highlights and original blog posts that dive deeper into our podcast topics subscribe today at kod.com newsletter so Dr David what in your background as as a psychiatrist has prepared you to kind of lead Oracle Health well I think the training in Psychiatry we really think of the core training it's it's being able to listen and understand what's going on and then hopefully I'm actually a Child and Adolescent psychiatrist be able to tell a teenager who's using drugs how we're going to get out of this mess right so um I think some of those core people skills have been really really helpful and you know a lot of what you've been working on lately is is connecting the healthcare ecosystem with information technology why is that a valuable thing to pursue well I to me kind of the core of Health Care is people caring for people and it's built on trust and if you have a strong caring relationship where there's a trusted caregiver you can actually get folks to engage in their treatment they understand it and make sense to them they kind of want to get better I see technology as a component of that as really a tool to help in that core caregiving but all of us have gone to the dock and it seems like they don't really look at you when they're treating you because they're looking at a computer or they don't have the information or they explain things into you in a way that you don't understand and you just get bad Health outcomes it's like this doesn't make sense to me so for my whole career that's what I've been trying to work on and at Oracle the fun part is we got we kind of got dealt like this great hand and now we're going to play it where C part of it of course we have cloud and Erp supply chain human Capital Management uh claims processing clinical trials but really what that is is this disjointed Health ecosystem that all of us get caught up in whether you're on the provider side whether you're a payer whether you're a patient or a family or a community like make all that mess go away underneath and just present it to me if I'm a mom caring for someone at home or a professional caregiver caring for someone in a clinic in a way that's useful and allows me to do that core piece which is take care of and and but David as you and I know technology leaders have been saying for a generation meaningful use this EMR the the the connected Health System all the big Tech folks and the smaller uh Tech uh uh evangelists have been saying that technology is going to save healthc care for a generation what's different this time well there's a bunch of stuff that I think is actually different first of all we've now digitized the record so no one got treated during coid and they said oh we can't find your chart because Dr Smith took it home no pharmacist said I can't read this prescription because your doctor has bad handwriting now you might say why did it take us 40 years to get there but it was a slog to kind of digitize the record and now that it's digitized there's a few things that I think make a different this time one is actually you brought it up Co like when you think about it we gave everyone in the world a kind of Public Health 10one lesson like people now know what B the curve means vaccines people even know what antia means right and so we spent a lot what we relive through this experience we understood it Health Equity better and I mean we meaning the whole general population so that's one the records digitized the the things at a Breaking Point you know in my role now I have this opportunity to go all over the world and whatever Health System I talk to whatever Health System I talk to their number one issue is Staffing I me it it just got into this breaking so that's a potential opportunity right when you think about now can Tech come in and I think this time and I hope our organization has done this we're coming in and trying to be much more humble this isn't about here's an app you stupid people in healthcare you just use this everything will get better but rather here's an open platform that's going to be connected that's going to be built with normal kind of user interface because I don't think any of our stuff ever want was that actually makes it easier for you to take care of people so that that's why I'm hoping that it's better this time it's also the cost again that's another national International thing I hear we can't just keep going up it means we won't have schools or wage increases or jobs if the only thing in this case the United States does is produce healthare it's it's insane especially with our poor quality so I think there's a lot of factors and we can get into Ai and all that stuff but I think fundamentally it's a Breaking Point the record got digitized the tech is good and the populace says Hey I want to tell a health visit I had that a year ago it makes sense in this situation John is talking about uh you know some some jargon some industry terms like meaningful use which came out of a a federal requirement out of the high-tech act and more recently Federal government's been working on interoperability Frameworks are those important are they making an impact is that one of the is that going to move us forward hold us back I I mean to me those are mumbo jumbo words so you know a cner were the largest electronic health record company in the world uh much bigger than anyone else and when we're done with the VA install in the United States we'll also be the biggest in the United States right and what I say to the team is that means we have more of Grandma's blood sugar than any of el like let's get down to really what this means when you say words like interoperability or meaningful use we got to make sure that blood sugar is accurate that it's secure that it's getting to the right place that we understand it's more than blood sugar it could be about food in Grandma's Pantry right that's actually the business we're in and then we cloak it in these words around meaningful use and interoperability but hey we all have grandmas or we all had a grandma um and we don't want their thing messed up right that's what and and we want it to be coordinated and sometimes Grandma goes here for care and sometimes she goes there that's got to go around with Grandma so that's what to me this means when we say interoperability do we understand the social situation that a person's living in are we bringing in in some cases very important claims data do we understand the disperate parts of the medical record and then are we giving it to the caregiver in a way that is useful for them to kind of do their thing how when you when you say because there is a crisis of Staffing around the world and we've got a demographic boom and an aging population in in across the entire industrialized world and we're not making enough doctors and nurses fast enough how can technology historically technology has added tasks and added work and to your point gotten in the way of the relationship between doctors and nurses so arguably we've gone backwards why is it what's going to happen now that's going to allow us to move forward and how do how how do you how do you square that Circle given our our our our our really terrible history in terms of implementing Tech in healthare so so uh I think it was last month we were in Vegas for our Oracle Cloud world and I got to demo our new clinical digital assistant now this has been built and will'll be live in customers this year so this isn't make belief stuff but here's what happens I pull out my mobile phone I say hey Oracle tell me about my next patients read patient Julie Smith to me knock on the room put the phone down Julie had already agreed to uh ambient listening I go through my whole exam talk to her how's the pneumonia hey it looks like your uh uh hemoglobin a1c's have gone up hey Oracle show me hemoglobin a1c's I actually got prompted because Julie had food insecurity we just talked now as a psychiatrist I had a practice using the stethoscope but I did pretend and listen to Julie's lung and when I was done I said hey thanks Julie and the metformin got increased the hcc's got all of it got done and I never and we set it up this way looked at the computer right the computer was actually at my back the whole time I was looking at you so you can talk about gen Ai and all kinds of fancy stuff the technology allowed me to get back to being a doctor and when you talk about Staffing and people getting burned out if we're really able to get people back to the part that feels so good or when you talk about a nurse who spends 50% of the time at the terminal 30% of the time looking for supplies and 20% of the time nursing if I can eliminate that terminal time I I mean we didn't go from so many clicks to less Clicks in My Demo we went to zero clicks right if we go to zero clicks for the nurse goes in sees the IVs infiltrated ambient listing captures everything including the fided and documents the nurses note we've just double the amount of nurses that we have because they're now no longer connected to the computer right so from a staffing standpoint I think one thing that the technology can do is kind of free our people up to do what they want to do as opposed to you said John the technology has kind of pulled them away from what they want to do and that burnout keeps people saying hey I don't want to be a doc anymore I don't want to be a Nur anymore I don't want to go into this you know um I think it's our time and I don't think it's just us doing this I think the industry is doing this and um to me it's fantastic I mean we got to get this one right John let's take a second to thank our sponsor digital Diagnostics everyone should have access to ethical Dependable Health Care digital Diagnostics guiding mission is to make a positive impact on individual patients and the healthcare ecosystem through the integration of point of care diagnostic AI I providing access to highquality AI diagnosis is helping to transform Healthcare delivery focusing on the diagnostic process for the individual to ultimately transform patient outcomes across Healthcare check out digital diagnostics.com to learn more digital Diagnostics AI the right way so so Dr David there's a lot of conversations about AI there has been conversations about AI in healthcare since the 80s since I remember when I was in college and graduate school L morbin Minsky model and then there were the the Deep neural networks but it does feel like this time is different around generative AI around the massive compute available and frankly to your point the availab the the availability of structured relevant data how are you thinking about leveraging all of that access to data that you have at Cerner and Oracle the tools you have to analyze it at Oracle and the opportunities to kind of generate novel Solutions at you know at at at will yeah so I I'd say on one hand I'm like wildly excited and on the other hand I'm extremely nervous right and so let me kind of go over those so first of all I think a lot of people say oh we're going to do Ai and you know for things like computer vision it's definitely better than I think human performance in some cases human in the loop is always ideal but a lot of people say we're going to focus on the non-clinical things first like throughput in the e Ed before we get to you know improving diagnostic accuracy throughput in the Ed is a clinical intervention right because if you're doing CPR in the Ed because you couldn't get a bed upstairs like the emergency the emergency department yeah yes the emergency part so all everything in healthc care in my opinion is clinical the billing the coding it means does the person actually get their approval from the insurance company for the medication so I I get the talk about it separately but I think anything you're doing in healthcare fundamentally affects the patient so why am I excited why am I so nervous when I was uh at Google with actually Cerner in the NHS we tried to see if we could use AI to improve the diagnosis of acute kidney injury in hospitalized patients and so the beginning was an app no AI but a nice app told the rapid response team this patient's creatinin going up hey come and see the patient just good user interface no AI it went from a 4our time to diagnosis to 14 minutes there was a 30% decrease in cardiac arrest and wow decrease in the cost of care like mind-blowing then we said okay let's do AI so instead of just looking at creatin the measure of kidney function sure we trained the model on 600,000 variables per patient times 70,000 patients so great training model we did this with the VA and instead of four hours and we've got it down to 14 minutes it now goes to negative 48 hours two days before any signs or symptoms this patient is no sign or symptom that there's acute kidney injury the computer with 90% accuracy says this patient's going to end up on dialysis like it's mind-blowing stuff so you go that's why I'm super excited like this is like a new field of anticipatory medicine here's why and you found that probabilistically that was accurate it wasn't just it wasn't just it wasn't just a retrospect that hey we're verifying what we already know no no no we we trained the model and then gave a new patients and they hit it 90% of the time time right 48 hours before normally would take four hours if you did go ux you got it to 14 minutes okay but we trained it on the VA well that's 93% male guess what the thing doesn't work as well on females so as we train these models that actually improve performance let's make sure that when we use them they come with a warning it's been trained in this situation it probably works pretty well we're not sure in this situation and please don't use it in this situation right because we don't want to make gaps in Care Health Equity um issues that we know are built unfortunately into our health system we don't want a hard wire anymore just keep making them worse so I am excited because I think it can really augment human caring and at the same time I'm nervous be think because I think it could actually make some of the bad stuff we do even worse and and could be used in the wrong situations so on the AI it sounds like there's you know pluses and minuses from your standpoint or things you're worried about there are a number of folks in the AI Community itself who've taken it a lot further not just to say it may reinforce bad things that are happening but may unleash all sorts of forces that we can't predict and lead to the uh the end of end of the human race are you have do you have that concern as well and is it a completely separate one from what you were just describing about you know reinforcing existing biases so um I'm spending a lot of time think about this one and and I think about it specifically around healthare um if you look at our current health care System let's just talk the US where we're getting close to 20% of GDP and um black women are dying at you know two to three times the rate of white women in childbirth even if they're not poor like it's amazingly how it's amazing how much we spend and how bad our outcomes are so I'm of the belief and I'm talking Healthcare now I'm not talking military or anything like that I think for healthcare these Technologies can really really up our game now let's be super thoughtful about them but I and we may need to contain them in certain ways and I'm certainly we need to regulate them but we have an opportunity the whole point of digitizing the record was not just fixing legibility and making sure that the doctor had the chart in the ER this is Magic we can put on top so let's make sure we capture that in the right way um I think it's a huge opportunity to really make care more Equitable more cost effective more culturally sensitive for us to really become a learning organization to learn from this you know hey Doc we've seen 10,000 cases just like this and no one has ever prescribed that medicine are you sure you want to do it I mean there are so many errors in health care because humans are human that we have the ability to augment that with really really powerful tools again that are used appropriately that were thought about that we continue to about so so I'm bullish on it it's it's a really interesting thought David I mean I I sort of share your concern about some of the feral models that the large language models that have been put out without any regulation or controls but to Leverage The Power of compute to actually ring fence stupidity and error and ra and and racism I mean because it's really the the math on on on maternal mortality for women of color across all income levels it's it's it's a crime and to think about the the computer as your as as a tool to protect you don't actually have to get into the area of unknown novelty uh you can actually work it work to use it almost as as a warning system for what's for what we already know is right and wrong that that's a really interesting frame yeah I I I I mean it gives us a huge opportunity to in a sense close those gaps yeah when I asked you about interoperability earlier you you pushed toward you know Grandma's uh sugars and which I take is sort of part of a theme sounds like a song did I hear it wrong John as part of a theme about uh you know putting patients and families at the at the center of care does does AI help to do that are these the sort of things that we can that we can do better with AI or is there a completely you know non-technological approach that's needed for that yeah yeah I think it can certainly help families I mean when you think about it if you're looking on your phone at your Labs through some portal people don't really want to see their Labs what they want to see is their bad cholesterol or when is my gy appointment right so AI can help us translate medical ease into what what am I what do you think I'm searching for what am I searching for and I don't have to think your way you know hey this thing came back from the pathologist that I have something in my prostate is that good or bad well that's where Ai and especially gen AI can I think really help in that translation so from a consumer face standpoint it's really really going to I think help make healthc care more understandable right because right now it's particularly confusing I mean just think about any of us who get an eov I'm an MD I got an MBA I've run Health Systems I get eobs from my own health system and I have no I say to my wife what do you think this is yeah so talk about someone who has less health literacy than me we have created a system that is totally incomprehensible well that's actually what AI is pretty good at like take all this stuff and make it make sense for me in my language and uh I I think that's a big potential I think eobs are everybody's favorite you know what I like about it is it says okay this is not a bill I said okay I got that now what else is maybe look can work out like what all other things that isn't and then I'll figure out what it is from that but they just say it's not a bill but it that nor is it nor is it terribly predictive of what that bill is going to like it's or descriptive of what it is you know one of the more interesting applications I I I heard about recently uh Dr David is that the AI was more effective at writing sympathetic apology letters to people around basic mistakes that hospitals and doctors have made than the hospitals and doctors you know complaints department were that there is a there's Nuance here around automating and simplifying but I I thought it was remarkable that it was easier for the computer to be heart centered an apology than it was for a department head it I think we are an K we are in a weird weird territory the the computer's probably a little less defensive than the person exactly hey sorry that we we sent you the wrong result on your buyouts yeah so you know we had coid we had uh you know 2019 come conveniently 101 years after the 1918 flu outbreak these come every 100 years or do you expect a little less than 100 years or a little bit more and if a new pandemic were to hit in 2024 will the system be ready well let me be a little edgy I I'd say we we have another pandemic going on if you really look at at illnesses that are driven by uh lack of access to good food unsafe housing poor education difficulty with Transportation um wage stagnation all of those things Drive multiple times things like diabetes hypertension high cholesterol kidney disease and if we want to decrease the health care cost in America by half overnight we just need to address those things right and so what we've done is create a system to try to fix those things once they got bad as opposed to going upstream and when you play this out we're not even good at fixing those things but if we were to really to go upstream and treat what I'm describing as this current pandemic of you know food born illnesses and and lack of opportunities in communities and poor access to education um we know those are drivers of help we just know that loneliness kills people and um living in neighborhoods where people are shooting one another leads to stress and medical and mental conditions like this is not new stuff so I as opposed to worrying about what's coming in a hundred years why aren't we talking today about these issues addressing these issues not only are they nice from a hey I'll feel better in a human standpoint they would actually solve the Financial issue of the Health Care system and I think a big way of getting that is to say how how can we stop talking about moving at a feif for service to a feif for value and actually move to really a value based system because now all of a sudden those things matter I don't want you going to the ER because you got chest pain and it's from a panic attack I want to treat your panic attack where in the current system no one really cares you know David I think you and I probably agree that when people ask what's the one thing you could do to fix Healthcare I said fix the farm bill and stop subsidizing the things that are killing us and and and and start start start supporting the things that actually would provide healthy people and ensure that everybody is is knowing as food and secure but I want to go a slightly different direction we'll hit the farm bill later David don't worry okay good thanks which which is which is how do you C can the machines with this epidemic of loneliness with your unique perspective as a mandarin of one of the Great Tech compan companes while also having the background of of of a psychiatrist can the machines help us with this epidemic of loneliness and if they can how well I think they're going to be really good and you kind of talked about that about the the apology note in really understanding emotions and seeing hey this is how I this is how I want my robot to talk with me and maybe you want to have a different kind of relationship with your robot or your AI companion um so it feels a little creepy to me yeah totally there's still something about a person caring for you um and but I think it could augment and and so I mean I I sometimes give an answer on you know is AI gonna replace doctors you're talking more about companionship but but I say AI is not going to replace Radiologists but Radiologists that don't Embrace AI will be replaced so I I would give a similar answer around this loneliness issue for example I think that computers and Tech um in some ways can make you even more lonely like you can get kind of lost in your social media and you never look up um if used appropriately we got to figure this out I don't know that any of us have the answer I think can help um address some of these things but I still think from a human standpoint we're we're people that need to be with other people and we saw that during Co like just being isolated make we just saw dramatic increases in mental health so maybe the tech can help bring us together and can augment how we're together and can help me be with you in a better way but I still think it's I still think we're fundamentally in a people business well that's a good note to end on that's it for yet another episode of care talk Our Guest today has been Dr David Feinberg he's chairman of Oracle health I'm David Willams president of Health business group and I'm John Driscoll the president of Walgreens health if you like what heard or you didn't please subscribe on your favorite service and thank you David for joining [Music] us
2023-10-20 07:06
