ad-free
Channel / Source:
TEDx Talks
Published: 2016-10-12
Source: https://www.youtube.com/watch?v=Nj2YSLPn6OY
let me introduce you to Mrs Manny she came to the emergency room sections fifty two years old who came to the emergency room but the foot sore doctors investigated efforts or and she ended up staying there in the hospital for twenty two days here's what happened but you get to the emergency room for a foot sword the inspected her this on a real reason for medical
concern but they wanted to monitor in case what sort was infected so they put her in the general ward on day three she starts developing symptoms of what looks like a mile pneumonia the give her the usual treatment of antibiotics and all's good but then her condition starts to worsen on day six she develops what's called tachycardia that means in medical speak her heartbeat read them
has accelerated dramatically she then has trouble breathing on day seven she experiences septic shock that means her body is in crisis incidentally mortality in shock is one into not to this point the doctors get Vili concern and they transferred her to the intensive care unit I see use other units were the most critically ill patients get cared for they hear the giver every possible treatment to
stabilize her but her condition only worsens first a kidney start to feel then how long spew and on day twenty two she dies Mrs Mattie did receive the right set of treatments the problem is she received them only to rate what Mrs many experienced was an infection that turned into sepsis let me tell you a little bit about what substances sepsis occurs been infection releases chemicals
in your blood to tackle the infection so your body releases chemicals to fight the infection now this chemical can trigger and negative inflammatory response then this in Arben this information triggers this negative inflammatory response what it can then do is because a cascade of changes leading your organs to feel leading to death sepsis is the eleventh leading cause of death more than breast cancer and prostate
cancer combine turns out sepsis is preventable if treated early okay so then what's the catch doctors find it very hard to recognize sepsis in fact a Harvard study shows but ninety three leading academic experts but when they were given several cases of patients with and without sepsis they couldn't agree two years ago my nephew he was admitted to the best hospital in India and he died
of sepsis my family was devastated I am a machine learning experts and what I do is study ways in which we can use large messy datasets to enable intelligent decision making so not to question for me was could machine learning of hoped could machine learning of hope Mrs Manny and my nephew so this led to a massive effort but my colleagues at Hopkins to design would
be called the targeted real time only warning system or trews based on machine learning I'll give you a sneak peek into what to use this on how we're using it to tackle says let me take a step back and tell you a little bit about what machine learning is and what's E. I artificial intelligence is a field of study baby design a baby teach computers how
to learn Kate just like you teach your kids machine learning is one way of doing this by designing cold all programs the teach computers stuff over time by you by interacting with the environment all watching okay someone to show you a video off some robots learning how to walk I find it funny how it shudders so you're probably not our thinking this is hopeless well so
the question is how can you teach robots or machines how to work intuitively you can think of it as designing a game the goal of the game it's for the computer on the robot to learn how to walk for as long as possible without falling okay so do do this first you have to design right down the goal in a language the computer understands but this
bill use math so now you wandering wall how to eat right the goal off walking the dog following as long as possible in math well that's often hard for different tasks you can think of it as writing down a formula and what this formula does is it scores so indicates a walking into score every move the robot makes if the move it makes hopes the robot
walk it gets a high score if the written move that the robots baking makes the robot unstable it gets a low score and now the robot school is to experiment with the sequence of moves in order to be able to maximize its score so how does it know which moves to try right well there are two strategies for doing it first it expatriate arms by interacting
with the environment okay so here the robot will just make a guess it guesses it makes a move if the movie gets a high score that's positive feedback and the robot builds on it okay the second strategy is by watching other robots in other words the robot finds deter from past robots that are similar to this robot it watches but moves that baffle bar did but
it was in very similar positions and now it emulates a replicates those moves Kate so those are the tools strategies so now I'm going to show you a video offer robot learning how to walk using the strategy I just described okay so in the beginning it's gonna look hopeless but I promise you it gets better how and just to be clear this is not so this
is the skeleton off the robot and so this is not a human animator going there and just moving or animating this video this is really the robot the algorithm choosing which moves to made by moving the joints of the skeleton that you're seeing and you can see it's already getting better now suddenly the robots he would walk and run for a lot longer than it was
doing right so essentially the basic principle is as follows figure out a game that the computer can play you write it down using a language it understands and then be trained it to optimize the score right this is how we teach cars how to drive the puter is how to play the game of go and Alexa do understand say your preference of coconut water %HESITATION so
let's go back in our Keister the problem off sepsis so the goal here is to identify sepsis as quickly as possible right and put this truth learns by watching in other words using data from past patients this avoids the need for a truce drafter experiment on you patience right so to do that what are the pieces truth needs to do so one big change that has
happened in medicine that's interesting to note is in the past five years the introduction of electronic health records India chars every single measurement every single lap test that is ever done when you walk into the clinic or you're in the hospital gets collected choose analyzes the data from thousands of patients to identify subtle signs and symptoms that appear in patients with sepsis than those without but
that's not alone what tools also needs to do is to figure out how to think about every signal in the context of every other signal let me give you an example that's a good example of creatinine so Clinton is a waste molecule okay and your kidneys filter it out but here's the catch so when your body is septic it affects your creek our kidneys it deteriorates
your kidneys ability to folk rock we hadn't %HESITATION creatinine level rises but there are many other things that can affect your kidneys ability to folks are planning for example if you have chronic kidney disease you're very likely to have high creatinine levels so now what tools has to do is to figure out is your creatinine high because of sepsis or because of chronic kidney disease or
the numerous other factors that need to high creatinine levels but that's not enough you need to do this for every single signal that exists in the electronic health record and to the stink about every signal in the context of every other signal to identify signs and symptoms that occur more often in patients with sepsis than those without let's return to Mrs Mackey research by Kumar and
colleagues have shown that for every hour treatment is delayed what type was about seventy percent the timing is critical we went and took Mrs magnesium and the ranch who's on it and here's what we found tools but have detected Mrs Mattie sepsis twelve hours before doctors car needed as my clinical colleagues would see that is the difference between life and death last year be short using
data from sixteen thousand patients that true was on average would have detected on most patients on average twenty four more than twenty four hours prior to the shop on set it's not twenty four hours in two thirds of these patients this abscess was detected prior to any organ dysfunction whatsoever and to put this result in context but sixty percent increase in performance with state of the
art so what was really doing is giving doctors in much longer window come in and intervene in order to prevent organ dysfunction and mortality this year be independently validated trews in data from Howard County General Hospital in Maryland and now we're working to do real time integration in order to make something like tools available to every doctor at Hopkins I'm also really excited because I have
to be published I papers several other health systems are now already implementing the published version of tools in order to be able to %HESITATION develop it in their own environment so I like like a few perhaps three salient characteristics but I think makes a strategy like who's very powerful okay first truest runs twenty four seven what it does is it gives doctors a second pair of
reliable ideas right to it's hard to scale up doctors it's easier I think much easier to scale of computers I want to use is really doing is allowing us to get expertise from the best doctors everywhere yes the third one which I think is very interesting in many cases like me seen sepsis we might not need new measurements the signs and symptoms were already in your
data and what tools is really doing is discovering the signs and symptoms we learn something that we couldn't see by I finally this but a lot of buzz about big data I don't make a little subtle point about a technical problem that I think to the solving that is very interesting it schools would be able to London learn much faster if it had a lot of
data on you or it could get more DR by experimenting on you but we don't want that right so what tools really has to do is leverage your limited data to figure out what's right for you right so in other words what tools really has to solve ism is a challenging small data problem in other words it has limited date on you it has to figure
out what is the right treatment for you and for that it has to let it leverages the vast amounts of data from other patients and figures out what information to borrow in order to make these assessments reliably and precisely so I also want to tell you a little bit about how this strategy is not unique the success so very broad leaf you think about it in
many diseases essentially where you have profile of symptoms and the response to treatment varies a great deal across individuals you can use a strategy like who's in order to target treatment so you're wondering like for example if you consider cancer diabetes multiple sclerosis our Parkinson's lupus so there are many such diseases on which a strategy like tools is amenable in fact in our own lab but
they are experts in rheumatic diseases autoimmune diseases in particular we're looking at how in scleroderma for instance we can use strategy similar trews two are avoid giving strong in municipal essence to patients who need them other colleagues the suicide William pallone Susan Murphy and their tea they're citing kids with ADHD and looking at how using similar data driven strategies they can identify when kids can be
L. benefit from behavioral therapy and we can avoid the need for giving them cycle stimulants altogether so the strategy is very very powerful so I was speaking about sepsis a let's go back to sepsis again so I said it was sepsis a Baroness month and %HESITATION the CDC has declared concepts is to be a medical emergency rightfully sort remember seven hunt fifty thousand people annually are
affected by sepsis a patient's family recently asked me what I take to bring this to a hospital near us I think that can be done in fact it can even be done within a year but we don't want to stop there we wanted to be possible to bring strategies like tools two hospitals everywhere and so the question is to do that what will it take right
so I think that three key things we need your help for one we need super smart engineers to be working and hope that we need your help in building and scaling up such technologies don't go to wallstreet healthcare needs you right we need policymakers create incentives open up electronic medical records as an expert at a leading health institution it's taken me more than a year because
the EMR so close in order to be able to figure out how to implement tools against the EMR it really should be easier than this pre pubescent healthcare system that's based on quality I could have got system is incentivized to optimize volume rather than quality right now you can choose rich restaurants to goto based on the quality of food should you be able to choose the
hospitals you goto based on quality of care part of the problem is that quality data at the moment is not very visible to consumers and we really need to make a bigger effort to make this quality of visible so that you can choose based on quality so to summarize sepsis is one preventable killer in many pressing medical problems like Nissan sepsis the answers for knowing who
