Data driven healthcare: It's personal | Aaron Black | TEDxTysons



alright so I'm gonna do a little bit of a warm-up today a little bit of ad-hoc just because I need to do this so make this go so much better today if I do this so logistically I got it ok smile these are this group has to smile a lot today alright so I'm gonna start with the with actually a story there we go so about three months ago I'm driving on my way home from work and I heard that ping many of you were probably very familiar with what that is and it was a cell phone alert and kind of took me by surprise picked it up out of my console did the swipe and what it told me was I was on my way home and that the route was clear and I was amazed you're like that's kind of odd coming from a data guy but that's that's why because I I thought how did they do this and here's the reason was a couple reasons number one how did it know where my home was well I found out my wife had gone into my cell phone and put the contact information in right when I got the phone but even more curious was I very rarely dried to work I live very close to the Nova Health System so I probably driven to work 20 times in two years so that was kind of odd to me even more curious was every other day that I go to work I peddle my bike I have never once gotten that prompt and I Drive the same way on the same roads and hit that same stoplight so how did it know lastly it just prompted me on that day like what was so special about that day so made me think you know I'm also only one of probably million tens of millions of people who are beaming up geo locations at any one second and time so this is what the SAP was doing was it was filtering me out of tens of millions of rows every single second it knew that I was in a car and not a bike it knew that I was in the general direction of where I was at to my house it knew every it calculated my route and it knew based on everybody else's location on that road that the route was clear and it did this by the time I hit the first stoplight which is within seconds to me that's amazing and I immediately think how can we do something like this in health care so I have an example here another cellphone alert right so good for audience reaction so what this is is it's it's still a biological test right it still is a woman pregnant or not and really what we're doing is we're taking that information again from a healthcare guy like me and we're blending it with contextual data to make the response accurate but I immediately contrasted it against a second one okay so same test same response for the data guys it's the same thing but we got to worry about context it's not as simple as measuring biological outcomes right it's it's you've got to be able to contextualize this and that's where I believe the future of healthcare is is blending this data together so how does this all work now in ANOVA you know we had we do some tremendous things we have great a great team that does genomic testing so we do things that only ten or fifteen years ago could be imagined but if you end up in ANOVA typically that means you're sick I mean something's happened you have a symptom you're there for a reason the future of healthcare is being able to predict and prevent you from ever showing up at the hospital but that's a difficult thing to grasp and for a data guy like me I can give you an analogy so about two-and-a-half years ago I'm in Las Vegas for a conference over the weekend you know III there was a game that let's just say I was interested in and typically in Vegas what they do before the game starts they give you a prediction and that's what's shown up on the screen here but they have so much data and they can do this so quickly that they have something called in-game prediction which means as the game goes on they take the unique characteristics of that game and they start to change the prediction on you and you can interact with it so as touch downs happens if momentum changes field locations time left in the game they all would come into the in factor thus in each unique to that game sorry so as you see like my heart the game have been flowed quite a bit at the end because of just the unique characteristics of that game and if you go to any game that the NFL might have done or a college football game each one these prediction starts it's different its unique to that particular game not just one before you start here's what it is so what's the analogy to healthcare now we have a lot of data coming out as we have biological data genomics think DNA but there's much more molecular data that we're capturing electronic health record data environmental data we're about to censor you all up very very quickly and what that does is now we can look at what what I would could turn in life prediction which means we are going to start to build these predictive models behind who you are as a unique individual so you can see we might stop very very low and again this is an analogy but as you start throughout your life excuse me it starts pretty low but in your mid-20s let's say you get married you make up a little bit you have your first child now you're running after that child you want to go back you know go back and work out so maybe it goes down a little bit you get a promotion stress level goes up eventually you have now three kids now you need to you know you have tuition so now it still continues to go up finally empty nest in your 60s and you're like well you think it would go down but no now you have all this this extra time and all the stuff to do so it continues to go up and then you hit your 60s and something happens okay so this is this is where you typically make to the hospital right we can build all these prediction sharks but didn't stop you from going there so you get these you heard the terms precision medicine personalized medicine that's we're at right now right we're we're gonna we're gonna give you the price precise surgical procedures we have genomics that can tell you what drugs to use and and you you know you want it you want to live life better so just like that football game that stuff goes down but as beautiful as this chart is to a data guy like me and this isn't interactive right you're not going to go into your phone pick up a chart and print this out every day it just doesn't make any sense so we have to start doing back to my prompts is to start to alert you at certain times in your life as that trajectory goes up so we need to take advantage of cell phones and things that can can alert us like we did in the first couple slides so what that look like well I'm not a designer so bear with me here but we could start prompting you then this is the the where you're kind of going up on that original slope we can give you prompts they'll say you know what we we've measured this risk it's gone up 25% let's swipe it so that you can find some custom recommendations just for you but let's say you ignored it you swiped it too many alerts in life you know pyre and we turn it red explanation points swipe it now you can talk to your doctor directly now you can Skype with them you can maybe it's a virtual reality at some point time so that's really the futures building these predictive models but interacting with you in a way that's unique to you and I have one more prompt for this now this one there's two reasons to do this one two reasons number one there I have never seen us from mix-a-lot reference in a TED talk so check that one off but thank you okay but the contrast here again is that this is unique we have to build interactions that are unique to what is going to motivate you because we can build all the models in the background but if you don't react to them if you don't change the way you behave it makes no difference so this could be you know this is funny but it could be emotional it could be something that is personal to you and we have to build that in in this as well I've done this a couple times to people and typically what they tell me is this is it's too personal to put my health care on my phone right I just don't feel comfortable and I agree there's it's definitely a personal decision but I would argue that our technologies are becoming very personal and why I put this on here this is my kitchen table this is my wife and I cell phone case can you tell who's who right we are dressing our technology up now all right and our kids it's it's getting worse and I say worse or better depending on your perspective but this is the Barbie doll and the action figure of the 21st century right they're dressing up they're seeing us with this technology and like I want that too so I think this is the future you know I think interacting with us in the healthcare environment digital technology is the future so how do we do this you know just because this is the future you know is gonna be hard and you know building this ecosystem to data interacting with us in a very personal and precise way you know dynamically changing as we learn more about ourselves in the way we interact with that technology you know is that sound too good to be true and I can tell you firsthand that you know this is a monumental effort I joke with people and tell them a side effect of working with this data's boldness but I'm only half joking this data is is so much more complicated than driving directions or you know predicting the outcome of a football game and it's much riskier you know if you make a mistake with somebody on there driving directions they might lose 10 or 15 minutes on their commute if you do this in health care you can alter someone's life so it's also you know it's we have a lot of Technology in and I'm in the technology arena so you've heard buzzwords like cloud computing machine learning artificial intelligence the Internet of Things but those are just tools they're not magic as much as I like them to be they're not magic and you know we need to have cross-functional teams of doctors scientists nurses researchers paired with that technology and then teams like mine who can support those technologies that are coming at us very very quickly it's also going to need you the people who are giving this data now you might not know but you are in some respects you need to know how and where you're giving that data you need to evaluate why and if you truly want to give it if you do give it give it completely inaccurately makes my job easier and and lastly know what know what they're gonna use this for know your rights these are all very important so lastly if we're gonna build these data to build predictions out of those data you know we really need to think about it we need to validate it and that needs to be done over time because any missteps in that these predictions can have tragic consequences for us and we might not know them for years so taking a step back from the from the funny cell phone prompts and the alerts you know just doing this for one chronic condition will be a monumental effort and I do believe I mean we have a lot of talented people and a lot of really great information technology but to me there's an ingredient missing there's something that I think we need in order for us to do this effectively and coming from a data geek like me this might be a little surprising but I believe that's love and let me tell you why over the past six weeks well six weeks ago my mother called me and said that my dad was having heart problems I don't really remember the rest the phone conversation you know he went to the doctor he's had some procedures and over the past six weeks as I've been doing this presentation my wife's didn't you're looking at me because it made me really evaluate what I was building so a lot of these things you've already seen I had this done a while ago um but what it made me do is go back and realize you know what this data really means and you know as I was building this presentation I was going back it really I had direct interpretations to what's going on my dad's life so those in life predictions were very specific because these weren't contrived animations anymore you know these were his so I took the full circle kind of back and went through this presentation again and I have a different perspective and it's these bits and bytes are not just data elements you know these are parts of our lives each data element is a representative thing that happens in our lifetimes and if we get enough date we can we can actually find just about anything but is it the right thing you know we need to remember why we're doing this and it's not to build you know big technologies with sexy buzz words I can't even look at it until we really treat these data these technologies and predictions predictions with that same strong emotions that we show for these for our loved ones only then were they these technologies impact be realized all right so I have one more slide today and this one is this is actually an early Father's Day gift who might add he was able to make it here today first time I've seen him since his procedures and so this is one last slide for him as I I want to tell him I love him and happy Father's Day you

10 comments

  1. This was one of the better Tedx Talks I have seen in a while. Implementing data analytics and machine learning into healthcare is what makes healthcare and the field of health informatics so intriguing. While there may be some folks who disagree with adding new technologies into healthcare, I would beg to differ. The use of electronic medical/health records have been evolving and while there is the likelihood of those technologies failing us, it is important to note that computers and other applications can store reams of data. This makes it faster and easier to track down a person's medical history/health rather than searching for folders upon folders of information. Overall these tools not only help in the field of healthcare but also in public health.

  2. I did not see the date of this talk, but, by the time I watched this video this technology was already very much a part of my experience.

    I am already subject to data driven, interactive, computer monitored, smart phone communicated, messaging – not unlike what the speaker is describing. It is not just inadequate, it pretty much fails to accurately assess what is actually going on medically. Consequently, it hijacks my day on a regular basis because I have to contact my health care providers and my medical insurance and durable medical equipment providers, etc., to correct the records, and untangle the confusing mess.

    This kind of system is only as good as the data that is (or is not) entered by humans (or, inadequately collected by unintelligent machines). Instead of making life simpler, it is putting people out of work, over complicating, and making an effen mess of things, and making life worse. It is also a way for health insurance companies to avoid costs that will reduce their profits.

  3. Wonderful examples with real-life implications! Thank you for sharing this information and how it is affecting everyone even now.

Leave a Reply

(*) Required, Your email will not be published