Artificial Intelligence in Healthcare – It’s about Time | Casey Bennett | TEDxNashville



so people often ask me what I do or to explain what I do and my mom's ask me like eleven thousand thousand times but I like to tell them as a joke that I create artificial intelligence but really it's kind of artificially stupid and it's true are numerous examples from IBM's Watson on Jeopardy to walking robots that are foiled by pop-tarts the field of AI is replete with examples of AI gone wrong and you don't have to work in the field of AI very long to develop a deep appreciation and respect for that which is natural intelligence beauty of it the complexity really is a remarkable thing that we all carry around in our heads right a fundamental problem that we have is that the AI and analytics tools that we're building think in a different way than people do right there's a disconnect between the AI tools we're creating and the way that people approach the world a good example of this is driving a car alright something probably most people in this room have experience doing so imagine that you're in your car and you're driving down the road right and you come up to an intersection and you decide you want to make a right turn so you go through that mental decision right and then you go through the physical process turning the wheel and the car starts to turn then all of a sudden a pedestrian crosses the street right in front of you now you don't just keep doing what you're doing right now just ow right into them hopefully some of you guys laughed a little too loudly no you change what you're going to do so maybe you change your turning radius upon doing so you may find an entirely new obstacle in your way right maybe a tree or a fire hydrant or maybe a parked car and so you have to change what you're going to do again right there's a whole sequence of events so a sequence of decisions that you have to navigate in order to act intelligently in the world and based on that probability of that sequence you you may not you may choose not to turn it that particular intersection at all right so the not traffic or your prior knowledge is the number of pedestrians you may choose to take an entirely different route it's not about whether turning at that particular intersection is good or bad it's about sequences of decisions about understanding the way the world changes over time it's about understanding how the things we do now and the consequences thereof change what we can do in the future and nor the best assistants are AI told me to reflect that thought process it's not simply about option A versus option B or is treatment X better than treatment Y it's about sequences of decisions understanding the way the world changes over time and being able to adapt to that in real time just like you're driving a car just like Google Maps works many of the great advances in human history have largely been about taking what previously seemed to be complex and making it simple right so a good example of this keeping with our theme is your car right so the way your car works the way the engine works is based on something called the Lagrange equations right that's the math you see there up on the board now I guarantee that the auto mechanic who works on your car probably has no idea what any of that means but you can still work on your car why because we've taken the complexity and we've made it simple we've compartmentalize that we turn into components it can be plugged together and taken apart it can be used to build create fix and in the same vein in health care and healthcare AI and healthcare analytics we often have these models these algorithms right that maybe tell us something about the way certain clinical processes work or prognosis of various treatments and a lot of times we get lost in the elegance of the math we get lost in the art of solving the problem without ever really solving the problem but if we can take these models and parse out from the patterns that tell us about the way the world changes over time about how patients change over time and we can push those patterns into simple interfaces that let patients and providers see what to expect in the future what we find is that with real patients that we can both improve the quality of healthcare while simultaneously reducing the cost in other words we can make healthcare better and we can do it more cheaply at the same time all by understanding the way the world changes over time by building tools that reflect that not surprisingly in other fields people have come along these people and have taken what previously seemed to be complex and he made it simple all by pointing out the critical importance time and understanding those phenomena and also build robots so um it's not just about making better decisions it's also about having better information and newer technologies like socially assistive robots these cute furry cuddly robots are essentially walking sensor sweets that we can put into people's homes that come equipped with sensors for light and sound motion and they can collect that information in an interactive way in an engaging way provide portals in people's everyday health their activity levels their sleep patterns or diet what's actually going on with people on a day to day basis in real-time all without retrofitting people's homes with sensors or other smart home technologies all by giving people something they can play with because reality is health is not something that happens two or three times a year when you show up at the doctor's office health is something that happens every day is happening right now it's an accumulation of choices it's accumulation of lifestyle and simply having a snapshot of that a few times a year isn't good enough we need to be able to understand how things change over time how health changes over time how patients change over time in real time and be able to feed that information back to clinical decision makers so that they can do what they do better story of human civilization it's largely been about taking the things that we previously had to do physically or in our minds and pushing those into tools in our environment bring us up to focus on higher-level aspects of the task even the first act of chiseling something and stone was really an act of cognitive offload right I don't have to remember it I wrote it down right there and that rock and things like notepads and calculators and computers have all just been continuing chapters in that story the story mankind the story of human civilization some people fear AI if you're going to replace people or force us to do certain things and you hear Vinod Khosla talking about robot Dockers and Eric Topol talking about cyborgs clinicians the reality is AI is just another chapter in that story the story of mankind the story of civilization it's all about helping us do what we do better you know I remember as my grandparents got older and they got sicker as most of us are want to do one of my grandpa's had heart issues and another one had Alzheimer's and I remember it you know you take them to the doctor's office and the doctor would tell you to do this or that and then they'd send you on your way and you'd go home and you'd hope maybe pray but it would work she didn't really know maybe it worked for 60% of people but maybe it wouldn't work for you and it was scary being a patient in our healthcare system is often a very scary proposition for patients and loved ones like and the reality is that we know more about your car ride via uber or your table reservation via a table me are what you're most likely to buy on Amazon than we do about what the treatment you've been prescribed as actually going to work that's not good enough we can do better and we should do better for me AI is not simply about creating cars that drive themselves or robot doctors that diagnose autonomously it's about empowering people we can see more no more beyond the constraints with which we were born then we can do more evaluate our options know what to expect better understand what lies on the horizon it's not the things in life that are comfortable but those are challenges which the finest thank you you

13 comments

  1. Very interesting idea! – your auto mechanic doesn't need to know the basic physics behind an engine working. Provides hope that the AI revolution doesn't necessarily need to leave us behind if we're smart in the way we approach it

  2. Excellent! A paradigm shift is over due in terms of health care and how we solve problems in this time and age.

Leave a Reply

(*) Required, Your email will not be published