thanks everybody for coming especially on a saturday uh it's nice to see a larger crowd than um i would have expected on saturday morning so i'm glad to be here i'm honored of course to be to have been uh uh named in connection with ruth sauber who was one of the folks that actually i worked with when i was a medical student here at brown almost 30 years ago 29 years ago i've got at least one of my classmates in the audience here and i think neither of us look like we went to medical school 29 years ago i want to talk to you today we got a lot of ground to cover i want to talk to you today about technology and health care my clinical training is as a pediatrician but i've mostly worked in global health and in technology in a very very non-traditional medical career and i'll actually be later at the medical school talking to medical students and maybe some of the parents about that non-traditional career but what i want to talk to you about today is more about the kind of changes that are actually not just going to happen but the changes that have happened uh in and around us and to us over the course the last 20 or 30 years and continue to change and alter our relationship to medicine and health care i do have one disclosure i should actually just popped up a moment ago uh i will be talking and showing things like the iphone and other products of apple and i i do have uh actually some apple stock which i in an impossibly lucky move purchased in the 1980s and and never got rid of uh that's about it for disclosures so i think any talk about uh the u.s healthcare system at all should start with this graph and this graph shows that even though the united states is by far the biggest spender per person on health care that's that's why we're over here at the nine thousand dollar mark um we are an extreme outlier among rich nations in terms of our lifespan uh and this is pre-pandemic and i don't know if you follow i don't know if you guys follow demography or this kind of stuff but basically uh the u.s lifespan has uh dropped again this past year that was from the pandemic but it's not just a pandemic related thing we have as you can see been separating from the rest of rich nations since about uh since the 80s effectively for lots of different reasons so we spend a lot of money and we don't seem to achieve the same results and just out of curiosity how many people here are physicians right so a fair number and how many folks are medical students or parents of medical students a couple couple of those too so we're talking about the system that we are in and or that we are about to launch a career in i'm going to go through this in three chapters uh the first one is barbarians at the gate and i'd like to talk to you about um something called disruption theory show of hands anybody know what disruption means okay a couple so i'm glad i'm going over the topic and i always start with this quote people don't want to buy a quarter inch drill they want a quarter inch hole this is from ted levitt who's passed away but was a famous economist and this tells me a couple things one it tells me that ted levitt was not really like a home handyman kind of guy because there's no really such thing as a quarter inch drill right you you have a drill and you buy a quarter inch bit for the drill but um so obviously he's kind of a nerdy maybe economist guy is how i pictured i love it but his point i always think at this point when i have physician colleagues who tell me things like well what patients really want is to have a good relationship with their doctor and i think is that really what patients want to have a good relationship with their doctor which is kind of like saying like what i really want is to have a good relationship with my auto mechanic right or would i rather just prefer not to have a relationship with my auto mechanic at all if that in fact was an option right let's not let's not confuse the method that we get the thing we want with the thing that we actually want now in terms of disruption uh everybody of course is familiar with the word disruption right obviously if someone pulled the fire alarm that would disrupt this lecture that is definitively not what we're talking about here so uh the second definition provided by the oxford dictionary is the one we're talking about here radical change to an existing industry or market from technological innovation and nowadays people are almost always talking about uh i.t disruption right digital disruption because that's the basically i t is the thing that's disrupting the most stuff nowadays but it doesn't necessarily have to be and so in fact i'm going to start even though i love that they specifically focus on no industry including health care being immune to digital disruption we're going to start a little bit further back talking about this so this is a classic example of disruption and it will complain what i'm talking about this is steel making so if we go back about 100 years this is a schematic sort of a map of the lackawanna steel plant in buffalo new york about 100 years ago and a hundred years ago this was the epitome the absolute peak the acme of steel making technology at the time they used something called a bessemer process they had an integrated plant that involved tens of thousands of people and hundreds of buildings and a very very complex operation again the height of the industry at that time this is something called a mini mill and this was this is technology that came out in the 1960s so whereas at the lackawanna plant they took steel that had been or rather they took ore that had been dug up out of the ground they were able to refine that ore process it in various different ways to make all sorts of products out of it the mini mill was different they just took scrap steel from junkyards and they put it into basically a big hot box and melted it down and they reused it that way and the thing about it is that compared to the integrated mill by almost every standard the mini mill was inferior right it was actually it was not nearly as good the integrated milk could produce any kind of steel from the most rough rebar right this stuff here that you see sticking out of concrete walls at construction sites right reinforcing bar that stuff is you know it's basically junk steel it does does the job but nobody's going to see it it's stuck inside a wall nobody really cares about the you know achieving perfect quality with that but the integrated mills were able to make everything all the way up to surgical steel you know fine refined steel steel for every possible thing and the only thing that many mills could do was to make that rebar but the one advantage that the mini mill had was they could do it for 20 cheaper and i want you to think throughout this lecture of what 20 cheaper does because what happened over the course of the next 50 years is this over the course of 50 years the mini mills ate the lunch of the integrated mills the mini mills took away all of the rebar business from the integrated mills and you know the integrated mill said that's fine we don't make that much money on rebar that's like the low end stuff we're going to focus on the high end and they basically just laughed at the mini mills and you know they laughed until they almost all went out of business and this is um 2015 by now it's probably up to about maybe about 75 percent of steel in the united states is produced by the mini mills who by the way started out only making that rebar but then were able to make they got better they practiced they learned different methods and they were able to more cheaply make almost all the different kinds of steel right so first they ate the salad out of the integrated meals lunch then they ate the rest of the lunch and again this whole time the integrated mills saw what was happening they saw what was happening and they no integrated mill in the united states ever decided to to start using any mini mill process over the course of 50 years no one ever tried it they just said that's ridiculous we're much better look at the quality of stuff that we produce etc and they focused on their high-end customers uh and they ignored this threat for the entire existence until again the u.s steel industry is a tiny
shadow of his former self now just like healthcare is today us the u.s steel industry was one of the biggest employers in the country back when minnie mills came out and now they are not so this example is a famous example from a book called the innovator's dilemma and this came out in 1980 1997 pardon me by a guy down the road at harvard business school named clinton christensen who tragically passed away just a couple years ago because this is a guy who is really coming up with pretty amazing theories uh about how basically how technology causes change the things that christensen noted apply all to this steel example right of his examples of disruptive innovation one that the disruptor comes up with a product that's worse not better right we always think that oh someone's going to come up with technology that's way better and we'll all start using it but in fact many times giant companies or even industries are completely thrown at whack or destroyed by products that are actually worse but just better in one way and often the way that they're better the one way that they're better is that they're cheaper meaning that they're more accessible to more companies are accessible to more people this is especially important to think about when we're talking about the health care system which is very expensive right again classically the incumbents they know what's happening they see the challenge and they laugh at it because they say that's okay those guys do let them take the low end of the market let them do that low end stuff we're the real specialists the real folks who can do the fantastic work and unfortunately for them the disrupter gets takes the money that they make from biting that little bit of lunch off right from taking the low end they take the money they make and they get better and better and better we've seen this with the volkswagen when they introduced the volkswagen bug in the mid 20th century in the united states people said you know people used to drive in big detroit cars a lot of people couldn't afford to to buy those big detroit cars but detroit didn't care about those people who couldn't afford to buy them and people said nobody wants that car it's so small it's underpowered nobody wants that car now volkswagen is the most profitable and largest uh car manufacturer in the world so much for nobody wants those cars when netflix came out and i know there are people in the audience that don't probably remember that netflix was not always a streaming service they used to and i know this sounds crazy they used to actually mail you dvds it took about two or three days and people of course rightfully they said nobody wants to wait for nobody wants to to choose a movie and wait for two days to get it we want to go to a video store and spend two hours looking for it uh and mobile around the video store and of course uh netflix uh about two years after netflix came out uh blockbuster video which is the biggest if you don't know was the biggest video store chain filed for bankruptcy um and of course we're doing it right now with things like walmart health and other retail health right big hospital systems laugh at these inexpensive all prices listed on a menu type of operations now i'm originally from new york and i've ridden the subway many many times and so i've been able to observe another kind of interesting example and now we're shifting into more that the i.t right the digital transformation or digital disruption i'm not old enough to remember the paper tickets which they viewed with the debut of the subway in 1904 or 1905 and i think they went until the 50s i do remember the subway tokens subway tokens was a big innovation at the time right they were much more durable than paper tickets right you're much less likely to destroy them uh so that's fantastic and then the metro card which came out i think in the 80s and which still exists in the metro card meant that you didn't have to wait in line as many times to fill up your to get your subway rides right to get your tickets uh but all that stuff is now being uh supplanted by apple pay right being able to pay with your iphone which is just rolling out kind of new york city wide now or i think it's finally they finally related out to the whole city and for a while there they were testing in different stations the great thing about apple pay in this particular case is you never have to wait in line because your phone always has the money to pay for a ticket and you don't have to worry about losing your metro card fantastic the interesting thing though about the iphone which is the delivery system for apple pay is that it also sparked some other innovations in transportation like a couple years after the iphone came out uber and lyft came out and uh more recently we've had other innovations in transportation which i find that people either love or hate um which is these shared electric scooters which i'm gonna say i i love um and shared bicycles and all this other kind of stuff and you'll notice that um and this is association right we're not sure if this is causality but it is highly suggestive that um the iphone by introducing and enabling competitors to the subway system is actually decreasing the use of the subway system so you can have the same technology come out and that technology is both sustaining and disruptive and i these are terms of of christensen's from that book that i mentioned sustaining means that it improves the performance of an existing product right apple pay makes the subway better no question about it but apple pay also makes other things that make people not ride the subway and that may in fact mean that the subway eventually goes away that's disruption now the second chapter i want to talk about something called the consumerization of technology and you'll see how this ties into disruption and this is something again you can tell by that subtitle that's been going on for a long time we'll start with this quote from one of my when i think is one of the most interesting people in the world mark andreessen uh he's the guy who invented the internet browser about 25 years ago and he is now a billionaire venture capitalist and i i thought it would be picking on him to put a photograph of him but he's he's got a bald pointed head and he looks just like a bond villain at this point so he sort of you know is a billionaire plotting the change of all these different industries but i don't want to focus on mark andreessen's appearance i want to focus on what he's talking about so if we look back over the last hundred or a little bit more than 100 years and we look at all the computer science that's happened there all the technology that's happened over this time we could see some trends that have occurred now of course one of the trend is that we no longer routinely use wood for example as a material to produce computers you can see that was very popular even up to the apple one which you can see up there at the top in the middle but the trend that i'm really talking about is the fact that for almost the first half of this period computers were things that were sold to institutions and i remember even when i was in college right computers were not things that people had they weren't things that small businesses had they were things that universities had or the dod had or corporations had right then around the 70s people started to or rather computer companies started to produce products that they were sold for dual use both to institutions and also to individuals right this is the era of the personal computer small small businesses started use computers and now we have fully pivoted into an era in which almost the entirety of computer technology is created not for industries or government or universities it is created for consumers right and people know this i think they have some idea that this is happening but it's really impossible to overstate what's going on here right this i don't know if you can see the line down here for the health care industry right which is the biggest under investor in technology and no one who's used any electronic health records in the hospital would be surprised to hear this right health almost doesn't even appear it almost doesn't even make the graph we are spending an enormous amount of money on consumer health that is to say consumers spend a huge amount of money on information technology far far more than the health care system spends on technology in fact far more than all the categories you see there combined including telecoms i mean telecoms is just i.t right and yet consumer technology is more than that in fact another way to put it and this is especially i think hits home for the medical professionals in the audience these are electronic health records systems there are rather companies that make them this is what the doctor uses to write down their notes and make orders etc electronically on the computer in the in the in the exam room these companies which people in medicine think of as guerrilla companies gigantic companies that dominate their industry companies like epic but the fact is that consumers spend more money on airpods than every hospital america has spent on any electronic health care system combined right consumers spend more money on airpods this is like airpods is not even like that doesn't appear on the top line of apples of apple's revenue statement right this is like an afterthought for apple and this is more money than is spent on technology and again we're not that surprised because all of us spend all of our time you know facing into our phones um i'm talking about all of you that's right i could see people looking at their phones right now i think they're watching the video version of this talk the streaming version of this talk which is very uh meta um but yeah we spend a lot of time on our phones we spend a lot of time on our technologies and it's produced uh depending on your opinion a virtuous or a vicious cycle whereby the more time we spend on it the more uh people want to create apps for us to look at which enables us makes us want to buy more phones which drives the cost of components down which makes more sales etc etc and one of the interesting things that's happened as if the the fact of all of us walking around carrying a super computer connected to the internet in our pockets wasn't enough is this this is inspiring another uh really revolution in software even beyond what we already have and it's because of the production of data as a result of what we're doing you know we talk about uh the data in electronic health record at the hospital but i can tell you for any of you if i wanted to be able to predict the the future of your health i would much rather be able to access the accelerometer data on your iphone to see how much you move and how often and how far right and how look the characteristics of your movement than i would to see the the occasional once a year visit that probably most of you have to hopefully most of you have to the doctor uh only once a year even for a chronic patient the information that we're generating just unconsciously all the time our phones our cars our tvs our refrigerators etc so much data in fact that if you look at all of the data that has been conceived since people were human right every single thing that was recorded whether it's art or its music or anything else everything that we have recorded that's considered to be about five exabytes or five billion uh gigabytes of data five exabytes and we produced that up until about 2007 when the iphone came out and now we produce about 100 exabytes every day every single day now um you know technology moves fast so there's a term that we used to use and when i say used to use it's about 10 years ago to describe this phenomenon and that was called big data now when i hear somebody ask me questions about big data i sort of feel like this person must not be really at the cutting edge of stuff because it's like oh that's so that's so that's so 2011.
nobody uses that term anymore back in 2011 mckinsey did a report on this big data on this unexpected unanticipated generation of all this enormous quantities of data and basically the important thing about what they said was oh big data is data that's really hard for us to do anything with right it's just so big we can't analyze it we can't use any of our tools right this is the thing the the most important defining characteristic of big data according to mckinsey in 2011 was you couldn't use it you couldn't do anything with it and yet not very much after that back down in 2017 that's only six years later the economist comes around and says well now even more valuable than oil the world's most valuable resource isn't oil it's data so the question and in fact some people i think really went out on a limb really went out on women said data is the new bacon which you know i'm i'm ambivalent about but i want to i want to put it out there and the question is what happened between 2011 and 2017 to change data from being well it's this amazing large volume thing that we can't do anything with to being wow this is the most important thing and it's the new bacon and what happened in between 2011 and 2017 is more software people came up with software or rather were able to use some software that had been around for a while software called ai software a lot of this stuff a lot of the concepts between for artificial intelligence software were developed back in the earlier the 80s and the 90s but at that time we didn't have fast enough processors to do to deal with it and we didn't have the data which i'm going to explain in a moment the data that we need to make it work right we didn't have smartphones we didn't have smart devices constantly churning out all this data to use so i should say first of all when we talk about data about ai about artificial intelligence sometimes people talk about artificial intelligence they're talking about general a.i that is you know conscious robots or conscious computers right terminator and all that kind of stuff obviously we've seen them in movies and tv that is clearly not what we're talking about here maybe that will happen maybe it won't i i doubt it will be in my lifetime in any case something that's right here right now is something called narrow ai and this is just another piece of software that automates some tasks that it used to be only people could do a task like image recognition right so to give you an example the military had been working even in the 70s and certainly in the 80s on things called expert systems where the idea was you're going to write a very specific if this then that series of algorithms to tell you basically to tell the tank uh to basically stay on the path now this is pretty easy with a highway right but if you're you know in some area where everything looks like red dirt right think about trying to write down a set of instructions for someone who's coming behind you to explain to them how you know that this is the path and that's not the path i mean it's hard for us to even write it in just plain old english much less to write a computer program and it was so hard in fact that it basically failed so the accuracy as you can see is 40 to 60 percent uh we know that 46 60 basically means like flipping a coin that's not what you hope to get when you spend 100 billion dollars on on expert systems but that's what we got what we have now is a completely new method this is you know people use a lot of different terms talking about specific kinds of artificial intelligence like machine learning etc i'm just going to refer to all this as ai for the purpose of this talk but happy to talk to people about more more in depth after the talk during the q a so with ai the idea is you have to input a lot of example data so this is a road this is not a road this is a path this is not a path here's a hundred thousand examples of what roads look like here's a hundred thousand examples of things that are not roads and once you do that um it's able to learn it is basically able to look at a new picture of a road or not road and tell you which one it is now first of all where are we going to get lots of example data i know you guys know the answer to this question right and very fast processors where are we going to get this well from the phone meaning we're going to get it from the consumer side where we're not going to get it we're not going to get it from the health side because that's that tiny little graph right that's a tiny little graph where we're not producing much data right so we have data for health but we have way more data for consume for the consumer side and that's where most of the advancements been happening and as a result of this advancement i mean enormous steps forward have been made enabling us to tackle some of the most difficult problems ever in computer science even the most complicated things and of course i know everyone knows what i'm going to show here because this is the thing that we've wrestled with for for for 50 years which is you know this is basically the the olympics of image recognition how can you tell the difference between a chihuahua and a blueberry muffin and and i mean let's be honest i mean some of those images i'm i'm not sure and i'm standing right next to the i'm standing right next to the image you know uh you can imagine what those expert systems i mean how could you write an algorithm to explain well if it's got three dots no no scratch that if it's tan colored with black no scratch that right it's very difficult to do and yet with this new system in which we basically just feed lots of sample images in we find that this has better than 95 accuracy and growing all the time uh basically better than human accuracy right the system will know and i don't know if you realize this but this is you know this is rolled out to all of us on on all of our phones android iphone um if you go to your photos app you can type in tree and it'll show you all the pictures that have trees in it or baby and i'll show you the pictures with babies in it you can even type pine tree or you know blank and or chihuahua dog and it'll show you all the pictures that have that without any labeling at all because of the ai that is running behind the scenes on your device or in some cases in the in the cloud connected to your device the thing about this is that the more data you can put into this box into this algorithm the better the result is and that means ai is getting better faster in some sectors and these are the sectors that have the most data and again we know what sectors those are and are not so we are surrounded as consumers we're surrounded with uh amazing amazing information technology um in almost everything that we do commercially right if you get netflix recommendations of movies if you have a bmw that tells you when you're veering out of your lane if you are on spotify and it recommends songs to you everything from recommendations stock picks sales things all this stuff is again it's not the future it's what's happening right now i was at a panel discussion with someone a month or two ago and someone said well you know we're not really sure what kind of technology will come down the road for health care and i said yeah we're sure it's whatever retail is doing right now is what healthcare will be doing 10 years from now right because healthcare is always behind the times when it comes to information technology now a book came out a couple years ago which i thought phrased this kind of well it helped me to solidify my own thoughts about this the book was called prediction machines and they said really boiling it down that computing right the the advent of computing was a revolution in cheap math right the ability to just do mathematical equations very quickly process them and that ai is a revolution in what they called cheap prediction right um to me this language is a little bit awkward because what they're saying is you show it a photo and it will predict whether that's likely to be a chihuahua dog or a blueberry muffin again that doesn't really to me sound like plain english so i prefer to say it's a revolution in cheap expertise because most of the time that's what we mean by expertise right um you know if you were living in downton abbey days and you happen to be like lord grantham or something you could probably hire a human to you know curate your library and sort of tell you what other books you might want to read but nowadays we just have to go to amazon and amazon does the same thing for us right you could have had a book expert doing that for you now we have that expertise available to us all the time through this ai revolution that's going on what about for health how many people saw either this nature article which came out in 2017 or maybe more likely the new yorker article that came out in 2017. so not too many in the room right this uh article uh maybe it's obvious this article prompted that article ai versus md this very you know aggressively named uh named article but basically what they found and they have confirmed this again and again and again and again is that they are able to use this image recognition to do some of the stuff that previously it took trained doctors to do certainly not everything but some things and uh uh siddhartha mukherjee the guy who wrote the emperor of maladies which if you haven't read it is a phenomenal book about about cancer just an amazing book he's a great writer he wrote this article basically talking about what are the implications of when medical expertise can be can be automated or scalable and since that time advances have only grown and some of these of course everybody knows i think pretty much everybody knows that your apple watch can tell if you have certain heart arrhythmias like afib and now apple announced that they think it will be able to actually diagnose other arrhythmias that they they didn't think it could before but there's other things happening as well we'll touch on a couple of them so this is idrx right so idrx is a system that takes a picture of your eye and looks at the picture and says whether you have something called diabetic retinopathy diabetic retinopathy is something that you get when you have diabetes and the back of your eyes the retina starts to have problems and previously this could that diagnosis could only be made by an ophthalmologist right a very highly paid very highly skilled physician this machine doesn't make a suggestion it doesn't say i think that this is diabetic retinopathy and show it to the ophthalmologist to then confirm this makes this is cleared by the fda to make the diagnosis without a doctor and what the company is now positioning this as something that they will sell as a business decision sell to primary care and places where they're dealing with a lot of diabetics so the primary care doctor can just you know they'll go sit down at that machine okay and then they look at the readout but the primary doctor isn't actually doing anything and you can easily imagine that they might put a machine like this at some point when the price drops right next to the blood pressure cuff in the in the cvs drug store right oh i'm just worried whether my my my diabetic retinopathy is advancing i better sit down in that machine while i'm waiting for my prescription and find out again something that used to take you know years and years and years of education to be able to do another of my favorites is walbot i was talking about this uh yesterday with some friends robot is a chat an automated chat system originally it was on facebook chat i think it's now on a couple different platforms and it does something called cognitive behavioral therapy or cbt this is a kind of very um algorithmic therapy that you do for people who often for people who are depressed and it's a way of making them sort of change the pattern of their thoughts so um one psychologist who's now the head of this company thought this is so cookie cutter you know this type of therapy like what the patient says and how we respond we should probably be able to automate this and you know it's still early days but as you can see even within the field even in within the mental health field they're saying sentences like it's possible to contemplate a time in the future when therapists have gone the way of bank tellers now i will say that they've only done a couple studies of of robot in which they found that it was as effective as a human therapist in reducing the symptoms of depression and it's actually quite interesting also when you look at the surveys they do with the patients when they ask them questions like do you like robot yes do you think that robot respects you yes and you're like it's almost like a uh sort of what they what they call it in literature willing suspension of disbelief like you know that that robot is not a conscious creature but you you know when they say do you think robot res the robot respects you yeah it feels like the robot respects me i like that robot um and there's plenty of other things as well and what's interesting to me is where they are and who they're positioned to be sold to so wobot is a direct to consumer thing obviously apple watch is also a direct to consumer thing um the diabetic retinopathy as i mentioned is being sold to primary care providers but i think it's just a matter of time before it winds up being a direct to consumer thing as well as things like identifying you know do i have melanoma what's what's this thing on my on my hand like should i be worried about this maybe should i go to see a dermatologist about it kind of that first line of diagnosis um and then there's other stuff that's i think even more interesting radiology was one of the first fields they started applying this right this uh image recognition to and um that they've made a lot of the progress in um but some stuff has happened along the way like res app so rezap is uh this is a group of australian researchers docs and phds out of australia they started a company and what they are doing is they said you know as a doc if i think a patient might have pneumonia what do i do i listen to them breathing i use a stethoscope and i listen to the breathing and i know because i've listened to many examples of people breathing if it sounds like it's pneumonia and not and certainly if it sounds like it's not normal and they said we ought to be able to automate that because this it's not just image recognition that ai does it's pattern recognition right it's identifying whether something is like these examples or it's like those examples and so they have a smartphone app which they're now in the process of testing and the smartphone app if you think you're you know if your kid has a fever and they're coughing they breathe in front of the smartphone app and it says do they have croup is it lower respirator respiratory tract disease is it pneumonia whoa yeah whoa whoa i mean now as a pediatrician i could tell you that you know this is something that we do all the time in the clinic but if they can automate this this will not be something that we do all the time in the clinic and probably more interestingly you know there's all this talk about um about whether or not we're going to need radiologists because ai will be able to read the x-rays but this doesn't end run around the whole thing right this basically says well we don't need to take the x-ray in the first place because we found a better way a less invasive way a less harmful way to diagnose pneumonia in the first place and this i think is one of the most interesting things about ai which is that it enables us to look at this these scads of data and they reveal things about us and our health and our world that before we never really thought there was a connection even in a situation like rezap where the connection was pretty obvious we've known for a long time that your breath sounds different when you have pneumonia than otherwise we just never didn't realize we could actually automate it now it's not all about ai there's lots of other apps that are changing health care that don't involve ai at all for example um and i love this middle app which for about a year i mispronounced in public presentations as nerks if you know this company it's pronounced new rx so um and interestingly uh happily the people in the audience knew maybe about the same as i did and nobody ever caught me making that mistake but it's it's new rx and what they do is they said wow you know what are the things that are most annoying to women in terms of women's health one of them is getting a birth control prescription or a refill a birth control prescription i got to make an appointment i got to go down i gotta talk to the doctor i gotta take time off of work i gotta get somebody to watch the kids all this kind of stuff and they said why don't we just have someone answer those questions that the doctor asked them answer them on the phone in an app oh and you know what else we'll do we're gonna make it asynchronous because there's no reason you have to have like like teledoc on the left is a video thing you're talking to the doc in real time said you don't have to talk to the doc in real time to answer these questions and have them determine whether or not you shouldn't take birth control or what kind you should take i should be able to just i mean basically send them an email and that's effectively what the app does you answer the questions and you'll get a text message or an email within the next you know few hours because a doc someplace has reviewed that stuff and then your prescription is winging its way and will arrive at your house the next day or the day afterwards right i mean think about this again it's not ai this is not amazing advanced technology beyond the amazing advanced technology of a smartphone but it completely changes the way that we might interact with the health care system now this is one of those things where i think i know a lot of ob docs were like well you know that's fine you know we don't we don't make a lot of money writing writing prescriptions for for birth control like this is not something we're worried about giving up but that's what the steel company said too right right we don't care about that because that's the low end we don't make that much money doing that and that's assuming that you think these companies are going to stand still right they're not ai not yet okay but you know i read an interesting article a couple years ago i know people caught this i think it mostly went under the radar fda might green light drug prescribing apps for chronic ailments wait what drug prescribing apps yeah that's right because if we can if we can automate certain expertise even at this point low-level expertise of physicians like should you get a should you get a the birth control or not well why don't we just instead of having a doc review it asynchronously why don't we just automate the whole damn process there's a lot of people in this country who are not on for example anti-cholesterol meds that should be on anti-cholesterol meds because of problems with access and cost and insurance and everything else why don't we just automate the process and indeed the fda is thinking about these things where the puck is going this is from uh this is from that famous well steve jobs by way or wayne gretzky by way of steve jobs quotes i think people probably recognize the quote from wayne gretzky the great hockey player who said i like to skate to where the puck is going not to where the puck has been and steve jobs used that same quote when he when they released the iphone talking about how they were trying to leapfrog over the aspirations of other phone makers at that time another quote to start with i really have to say i hate these robot doctor illustrations but it's almost you have to include it people should stop training radiologists now it's just completely obvious that within five years deep learning is going to do better than radiologists so i was lucky enough to actually be there when jeff at the conference where jeff hinton said this and created an enormous hubbub especially six years ago jeff hinton is a guy who he's this uh he's a computer scientist from this illustrious illustrious british family like you know he's the kind of person who's like great grandfather was a celebrated entomologist you know sorry edmund entomologist insect studier right you know this this kind of a family you know i'm sure his great grandfather knew charles darwin but he is a huge pioneer works at google of course um in ai and he's the guy who came up with one of the techniques that's used that really advance the field of image recognition like by a huge amount like they solved one problem and suddenly everything was 100 times better and he made the statement that people you know mocked it just you know hated this obviously radiologists you know mocked it and said this is ridiculous people often say that he they often think that he said that radiologists will be gone five years from now but that isn't what he said he said basically what he's saying is that i think we have enough radiologists now in 2016 trained to last us the next let's say 20 to 25 years until we won't need them anymore right there's no point training any radiologists now and when i give this talk or something like it to radiologists it's very interesting because i can always tell by the color of their hair how gray their hair is how they're going to react to this if their hair looks like mine they're kind of like oh that's you know that's kind of theoretically interesting but my ira is fine and you know i think i'm gonna be okay and it's but it's the the resident the resident radiologists that are like what does this mean for my career this obviously makes us think in general about can machines actually replace doctors and you read all sorts of things saying yes they can no they can't you can never replace the human touch uh many times at this point i can say safely that many times in talking about this topic i have had someone who is in the question and answer period hopefully not today who's clearly a little angry or agitated who will say something like and you can tell they're trying to control you know like not to appear like a complete lunatic they try to control themselves and say you can never replace the judgment of a position with a machine and on that topic i want to talk to you about a certain machine that we we're all we've all been living with for a long time um so imagine if i told you there was a machine and by the way bonus points for everybody except for rider windham for guessing who the little boy and the dog are if you know this reference this is something from my childhood but imagine if there was a machine and i told you that that machine has already eliminated 60 000 full-time medical practitioner jobs sixty thousand which is about six percent of the total of practicing physicians in active physicians in the united states what if i told you that was the case you'd say well you know if you weren't a doctor uh you'd say well that's that's some machine that's pretty impressive but in fact it's not really a machine it's actually the fda what am i talking about i'm talking about over-the-counter drugs now i've read a million articles about whether machines or ai is going to replace doctors and i've never met a single one about whether over-the-counter drugs are going to replace doctors and yet the fact is that since 1976 with about 124 different active ingredients that have switched from requires a prescription to doesn't require a prescription booze what's now booz allen made a report back in 2012 almost 10 years ago saying if they we didn't have these over-the-counter drugs we'd need an extra roughly 60 000 physicians to make the appointments to see these people to give them their currently over-the-counter drugs and you think well that's nuts why would you need a prescription for advil well i don't know why you needed a prescription for advil but you did and now you don't any longer and again we're not talking a trivial amount of full-time equivalents right we're not talking a trivial amount of jobs we're talking six percent of the total and that's the fda without trying right that's just from over-the-counter drugs but it's not just over-the-counter drugs because this also fits into this narration or this narrative i'm telling you about in terms of the consumerization of technology and the data because we've changed over the course of the last 30 years and how we've changed is we have computers now and we have the internet and we have google now we are different consumers than we were 30 years ago i mean i remember when i was a kid if you wanted to know something or look it up you had to go to the library and maybe even ask the librarian to like show you the dewey decimal system in the in the wooden card files right that's that's how old i am right but the point is now we have all these technologies and all these technologies i can tell you as a physician make me a better physician but they also make the consumer a little bit of a better physician right they'll make the consumer every consumer a little bit better able to care for themselves you can think of it this way if we look at a graph of capabilities and cost you'd say well okay in terms of you know healthcare diagnostic and curative capabilities a physician obviously has more capabilities than the average consumer the average doc and certainly the average nurse practitioner does as well but of course they cost more it costs a lot of money to to educate a physician what happens when we add you know a smartphone with google to each of these different persons well they all get smarter right they all get more capable it's a little more expensive because a smartphone is not free right and your internet service is not free but it's you know it's pretty negligible when you compare it to the cost of medical school and the fact is that since we have these devices now we have this access to the internet all the time it has advanced our capabilities where we can do things now that the fda thought we couldn't do before like treat ourselves for our own seasonal allergies right now this this has changed so much that it seems completely absurd like that business about netflix mailing people dvds right you're like that can't be true grandpa but it is true the fact is that 20 years ago if you have you had a stuffy nose and itchy eyes in the spring you had to contact a scheduler speak with that person then speak with a human at the front desk maybe speak with the parking attendant in the parking garage and take time off from work get shown by a nurse into an examining room get seen by a physician in the examining room maybe even have to see an even more highly paid allergist so you could then get another piece of paper and you could carry it to another human in a drugstore so they could give you your claritin right i mean think of like some little part of all those people's jobs was getting you your claritin and now of course we don't have to do that because our capabilities have expanded by the technology that's available to us not just for seasonalities but for lots of other things things that again it just seems crazy to me like you have athlete's foot you used to have to go to see a doctor about it and this adds up to about 60 000 full-time medical practitioner physicians that we don't need think about that when people tell you that we have a physician shortage or we're facing a physician shortage we might be facing a care shortage within with an uh an aging population but i don't think we're facing a physician shortage per se what happens to this when we add this amazingly powerful new tool of ai well it's predictable all of a sudden everybody gets a lot more capable the primary care doctor gets a lot more capable the nurse practitioner gets a lot more capable and you get a lot more capable there's all sorts of things that you go to the doctor for now that you won't go to the doctor for in the future you won't need to because your phone or your watch or your your glasses or your implant are going to enable you to know these things that you can't know right now and that's going to give us the capability to treat all sorts of other stuff which now may seem unthinkable like treating our heart disease or our pneumonia or our diabetes or well depression i guess wobot has already taken care of that but being able to treat all of these things and again remember this fda green lighting apps that prescribe drugs so in conclusion i'd like to tell you about one last example of disruption classic example health making so what we see on the left is the salem va medical center this is a integrated mill of health where you produce all different sorts of health products you have all kinds of specialists thousands of people work there in hundreds of different buildings and they can basically do anything that you need to get done about your health that was introduced about 100 years ago and on the right we see something called the mini mill or retail clinic and that was introduced in the 60s i don't know if this is starting to sound familiar of course at the integrated mill they can do anything related to health and the minnie mill is much more limited but the mini mill can do it for about 20 cheaper at least and of course there's another disrupter that's in the picture here and that is us just regular old people right we can do all sorts of other things as i've been illustrating and we're going to be able to do even more things as the future progresses as our as we advance in this and again important to keep in mind we keep thinking i i find people keep thinking it's the health care system that's going to be making most of these advances but that's not where the data is folks that's not where the advanced computing is that's not where the data is most of the most of the advances aren't going to be coming from the healthcare system maybe from the medical research system right maybe from apple and google and other other tech companies but very unlikely to be coming from the with the sad record of of information technology and health care very unlikely to be coming from health care and i'm not going to say this is my prediction for the future because it's not it's this is a description of what's just happened right this is what's happening around us this is what we've done over the course of our lifetimes we gradually get are able to do more and more retail clinics and urgent care centers can gradually do more and more and healthcare will do less and less and if you're a medical student now i think it's really important for you to think about this to think about where do you want to position yourself in terms of your career i think that the medical students who are studying right now are going to be the last generation of medical students who could confidently think that the most likely outcome for them is clinical practice that they're going to be seeing patients that's what they're going to do because i think it's going to be it's going to be nurse practitioners it's going to be physicians assistants it's going to be apps on your phone it's going to be retail clinics it's going to be all those different things but i think it's unlikely to be people who study for 16 years in order to be able to do the especially those low-level simple things now does this mean this is going to eliminate physicians no of course not i think there'll always be other things that you'll need that depth of experience and that depth of education to be able to do and i i think thankfully these are among the most interesting things but there's absolutely no doubt that this is exactly what's going on and the most recent piece of data i was able to find uh actually quite quite recently from last year to some in support of exactly what i'm talking about is this one from gangly i think at harvard this was published last year to make the declining use of primary care among commercially insured adults in the united states 2008 to 2016 an overall decline of 24 of people going to the basically going to the doctor going to the primary care doctor why well i mean it could be because the primary care doctor is expensive people don't want to pay the co-pay or maybe they're not insured lots of financial reasons you might not want to do that but what they noted in the discussion was that this was especially true for younger adults who are increasingly comfortable with online self-care right they just go to the internet they just go to an app uh they don't want to go in for the kind of things that they now know they can just search the internet to find the answer for how they can cure themselves right exactly what we've been talking about so again this is not the future this is the this is the past and the future now talked about a lot of stuff over the course of 48 minutes uh i put some slides here just to remind you about all the things that we that we've shown there again we talked about the just examples of what disruption is we talked about the consumerization of technology and how now almost all the great tech and almost all the great data to power ai is in the consumer side um we talked about software changing the world just how this has completely changed how people look at their health especially younger people and how that's already along with actions by the fda with regulation etc this is already changing the way that people basically get healthy when they're sick or stay healthy when they're already healthy i would be happy to take questions uh we have again plenty of time for questions and i know we have a microphone which i'm going to carry around to different folks so they can ask them owen thank you for your attention i think it's on yeah so um an adjacent question on healthtech given sort of the change drivers and the role of data why do you think google has failed to do this famously recently and then 10 years ago as well yeah so the question is given all this stuff and i guess intrinsic to the question is and google's such a player in technology obviously why has google failed so much in this and honestly i think that um i think part of is our expectations about google google does one thing really well and that is search advertisements that's where 99 of google's money comes from google has all sorts of moonshots that they announced they're going to do this they're going to have a contact lens that detects diabetes they're going to do this and that honestly i feel like that's all marketing i mean they they they announce all these things they never make money on any of them they don't they don't even make money on android which is a great product right i mean that's a very successful and great product but they give it away right so i i just think that google um they haven't yet demonstrated that they can actually do something other than do really effective search ads just my opinion uh yeah in the brown shirt oh sorry i guess probably too many brown shirts i think that kind of gets into his question as well most of the deep learning tech is at google tensorflow facebook pytarch and you know the big tech essentially right and i think one of the reasons why they've not had tremendous success in health and i think there are structural reasons in health domain for adopting big tech because of privacy concerns i mean are we really going to trust our data to facebook yeah right you know are we really going to trust our data to google given given everything that has happened in the last two weeks so i think the problem is these people have the most sophisticated tech yeah from a deep learning algorithms and all of that but you know they have a huge credibility and trust issue yeah right so how do you see those two things being bridged sure so that's that's a great question and people are talking a lot about privacy and about these concerns but i w
2022-03-08