Biomedical Innovation 101: Seminar 4 - Digital Health Technologies

Biomedical Innovation 101: Seminar 4 - Digital Health Technologies

Show Video

all right well good afternoon everyone um my name is John cervoss and I'm the director of commercialization education at fast forward medical Innovation uh also known as ffmi and I want to welcome you to today's webinar uh which is number four if you've been following along in our six part series with the Department of internal medicine here at the University of Michigan before we get things started I do have a few technical items I'd like to cover uh if you have a question today please feel free to type it into the Q a feature at the bottom of your screen uh our staff will be monitoring the Q a as well as the chat function during both presentations today and we'll be sure to link any link to any resources that are discussed today uh so that you can follow along I'll also note that the presentation is being recorded and we plan to send a a version of the recording along with slides and an opportunity to take an evaluation all post program via email later today today's webinar is titled digital Health Technologies software and mobile apps where we will be discussing the development pathway of digital Health Technologies and highlighting some unique considerations along the way as I mentioned this is a sponsored webinar series with the Department of internal medicine and today's webinars in particularly supported by U of M's Innovation Partnerships group as well I would suggest that you visit our website uh or in that post event email we can link to some of the remaining webinars including one on biomedical devices as well as additional resources and funding opportunities for those of you that are working on commercializing a project so with that I'm very pleased to introduce our two speakers today first you will hear from Drew Bennett who serves as the Director of software licensing and research Partnerships at the University of Michigan Innovation Partnerships group uh following Drew you'll hear from Michael Lanham who is an associate chief medical officer and assistant professor of both learning Health Sciences and OB GYN so thank you both for being here and I will turn things over to you drew great thanks John uh as John said my name is Drew Bennett I'm with the office of innovation Partnerships and lead the software practice there I'm happy to be doing this webinar today with Dr Lanham and our friends at ffmi always great to partner with them so I'm going to go through our a number of different things hopefully one of the takeaways at least from my side of the conversation is just uh to generate some enthusiasm for the variety of things that can be done raise some um thoughts around some of the foundational aspects of what really goes into digital Health opportunities and talk about some case examples there's lots of great stuff that's actually been done here at Michigan some people have varying levels of awareness around this I always like to highlight them because they're really really interesting um cool type of applications in digital health so it's not uh completely theoretical there's lots and lots of good things going on here so hopefully uh that's valuable to everybody and for John's comments certainly feel free to ask questions that's what we're here for this is intended to be as the title implies an introduction but certainly something to help people understand so just by way a backdrop to hopefully I understand who Innovation Partnerships is innovation Partnerships basically Works hand in hand with other partners across campus such as ffmi to help facilitate the transition of really high quality research that's done here at Michigan out into the marketplace through the translational research process our big uh primary responsibilities are working closely with pis around intellectual property production and then helping to facilitate commercialization through licensing and translational research development so these are some of our statistics everybody likes to thump their chests a little bit the good news for you is that you have a great support mechanism through Partnerships such as ffmi as well as our office we do a lot of stuff the University of Michigan is fortunate to be one of the handful of Institutions across the country that that does the volume of work that we do and that's a direct result of the Innovation and expertise at Michigan really that process in a nutshell is represented by this graphic at the front end of it we're trying to work with you to evaluate what we have determine what the potential protection strategies are what's the market for what uh you're doing from a work standpoint and then move it through that process of development and release into the marketplace so again all those translational activities lots of great programming such as this education wise lots of good funding mechanisms across campus campus to help you develop your product lots of mechanisms to reduce that risk gain feedback understand what the market needs and hopefully end up with a solution that really solves a problem in the marketplace so bottom line we're really here to help translate that activity From the Bench to the bedside and really have a big impact with you in the marketplace so from a foundational standpoint you know we've been uh working closely with ffmia and the medical school for a long long time I've been here going on 11 years now and if I really had to look at digital health and say what's the primary Foundation of it it really comes down to analytics it's basically those activities now that we're capable of doing based on consuming massive amounts of data and really starting to apply that in a clinical setting or an operational setting to change Healthcare if you walk away with any other takeaway from my presentation it's really to understand the premise from which this starts so it's really access to that data it's starting to analyze it interpret it and then apply it on a forward basis to the clinical setting again sometimes these things and we'll talk about some examples are very operational so they have to do with things that are more on the mechanical aspect of running the healthcare setting a number of these things are absolutely applied to clinical situations so they're diagnostic things they're prescriptive things and then there's a lot of things that are outside of the health system Telehealth not being the the least obvious of them that has changed substantially in the last five years certainly since the pandemic something that many people would have thought was on a slow roll is certainly been rapidly accelerated as a result of that so when we think about it let's really think about analytics and data as those foundational pieces to get a digital Health Solution off the ground so it's um as as has been said by lots and lots of people uh you know data is the new oil in many ways it is kind of the foundational gold for a lot of activities not the least of which certainly being in the healthcare setting so when we think about it you know the most obvious thing is the electronic health record what sits in Epic the EMR here at Michigan but there's tons of other things that are part of those foundational aspects that lead to a really significant change in what we're doing from a digital Health standpoint and not the least of which in that that category have to do with sensors the ability to essentially pick up signals that we were never able to see even as little as 10 years ago doesn't take a too many examples to appreciate how that was changed 10 years ago you couldn't have a sensor sitting on your wrist that was fundamentally able to detect you know cardiac anomalies you know now you can have an eye watch on your wrist it's going to tell you if it's picking up afib almost unheard of in a in the near past and of course on the back end of those things are a number of things that have to come with that the ability to generate that data is one thing the ability to compute against it and actually drive results uh has been assisted by things like massive compute capacity in the cloud and emerging tools like artificial intelligence and machine learning and deep learning we'll talk about that a little bit and then really the economics associated with that and the willingness to invest talk a little bit about kind of some some things here that are emerging areas in the healthcare space but EHR adoption if I were to highlight anything in that mix of things certainly is the foundational piece and the only thing to take away from the slide is really the adoption rate of the EHR through the various changes in legislation over the last several years has made a huge difference that the fact that we have this data over the longitudinal sweep of a patient's lifetime uh in pretty much every setting whether that's a clinical application in hospital setting or not of Hospital situation are really really massive massive changes so the EHR is something that's that's really been the fuel to a lot of the digital Health activity again on top of that just a massive massive amount of sensors whether it's things like blood glucose sensors that are now very common people running around with um afib detection or EKGs on their wrist everything motion sensors um csat sensors that people have at home that you can get for very very cheap uh inertial sensors of any kind they're all over the place so the the services Market is huge it's getting bigger and the artificial intelligence market so the ability to compute on top of all of this data whether it's gathered from the EHR or sensors starts to build that story of how digital Health Solutions come about and in the U.S it is particularly [Music] um important to remember that the amount of spending from a GDP standpoint that's associated with Health Care is just huge um the these numbers are a little bit dated at this point in time but you know 18 of GDP is a really really really huge number a three and a half trillion dollars uh tells you what that upside opportunity is and certainly when we look at that the reason digital Health in particular becomes probably a more unique and interesting opportunity is the fact that there's the uh possibility of changing via digital Health Solutions Telehealth remote sensing in hospital sensing uh that cost um ratio can be changed dramatically so certainly a lot of front-end investment but the opportunity to see some things change significantly it's really why there's a lot of excitement around this um just a couple of quick slides on AI and machine learning and deep learning you know most people have a general sense for this as far as the constellation of things there but all three of these all broadly in my estimation are the are now the foundation for most of the current exciting um digital Health Solutions they all have some aspect of it that um Ai and machine learning are looking at this Trove of data and fundamentally changing the way we're doing care so they're assistive in the sense of consuming a lot of data boiling that back to something that's usable from a clinical standpoint and then driving a change in care hopefully with respect to everything from Diagnostics to longitudinal tracking and treatment and and operationally as well and again if you look at these very specifically and examples where they're being applied we know them to be applied right now things such as ml being applied in logistic regression um and trying to evaluate skin lesions is this thing benign or is it like very very specific known case where again ml being applied to evaluating a lesion and determining the status of that um then as we move into the deep learning type of things lots of deep learning activity being applied in image analysis so detecting everything from a tumor to fracture detection to intravascular bleed characterization things of this nature so all these things previously were done by an expert who's looking at it and doing some type of assessment now it's assistive in the sense of it can be quantitative giving you a better detailed understanding of what these things are and then allowing you to understand things such as how treatment is affecting the results foreign thing that comes into this mix is really what is going on with digital health and they're just a tremendous amount of activity in this I you know read this morning CVS at Oak Street Health which is a primary care solution provider for 10.6 billion dollars and that's just an add-on to a lot of activity that's going on out there I mean most people I think are familiar with Amazon three or four years ago bought a company called pillpack which is basically a online pharmacy solution Google purchasing uh Fitbit and a few other things Partnerships they've done with Beth Israel and other folks their care studio model that they've worked with Microsoft buying Nuance which is uh the voice technology company and then their big play in um basically cloud computing for healthcare so when we look at this there's lots of people and they're in very different sectors doing a lot of unique and different things certainly the ones on this slide are demonstrative of the different areas that people are operating in um and All Digital based Solutions so I wanted to spend a few minutes talking about um just some examples here that have come out of Michigan I'm going to go through these very quickly I apologize certainly if you want to know more about them I'm happy to talk one-on-one with folks and I'm happy to certainly spend some more time on it but just to give you a sense of the breadth of digital Health Solutions that have come out of Michigan just so you have some sense of what people are doing that's really what this next section is the uh this company general is a genomic literature search session it was started by a physician by the name of Dr Mark Keel came out of Michigan and fundamentally he was looking at genomic uh literature and saying you know the the challenge I have is not this is there's so much of this out there and I can't get to it all and fundamentally what genomina has built is a thing called Mastermind and it is a highly curated very specific search engine that's looking at scientific articles to say what's the state of the art currently to determine clinical care associated with a specific type of genetic variant or issue really really cool super interesting company has gone through several rounds of funding and this was developed at Michigan the original prototype was developed at Michigan another startup out of Michigan which is using sensor data as a company called fifthi and what they have is what's called an ews or an early warning system so they're fundamentally taking huge amounts of streaming data from various sensors that someone would have generally in critical care or even on the floor they're evaluating those things to determine if that person is going to have a hemodynamic decompensation type of event they'll call it hemodynamic instability so trying to see that before it's coming and then providing signaling to the care team to say hey what should we do here what are the opportunities to vein how can we help and make sure that this doesn't become a bigger issue obviously easier to help somebody in a preventative way than try to get them out of a very uh difficult decompensation event we have another startup it's going through a little bit of a change now but certainly is worth noting is automated scoring of colonoscopy videos one of the most common procedures done on an outpatient basis the challenging part is you know uh you you go through that you have a large video capture very difficult for any physician to be able to look at every one of those but what the solution does is in real time we'll evaluate that entire video post colonoscopy and then highlight areas that are particularly interesting that should be re-looked at and it does a couple of different things quantifies things like lesions or disturbances or ulcers this is super important with people with ulcerative colitis or Crohn's disease or IBS and it also geographically uh highlights where those things are at so if they come unfortunately folks who suffer from that um have multiple colonoscopies but allows the physician and Care team to be able to go back kind of subsequently want to determine if the treatment is resulted in any changes angioensite another image analysis tool is basically doing a 3D model from a 2d capture during a cath lab of activity to determine the total occlusion in a coronary artery and then as well do what's called ffr or frictional flow Reserve so essentially what's the flow through that particular set of arteries we're super interesting something that prior to the computational Firepower we have today and digital Health wasn't something that could be done but it's being done on the fly now and then I mentioned something earlier about really um operational type of activities we had a solution called maze analytics which was called explanation based auditing it's fundamentally determining if everyone is accessing a health record has an actual legitimate reason to do that in the example that the company would always use is you know if a celebrity were in a hospital is everybody who's looking at that medical record somebody who's actually on their car team our care team or some of those people just curious and basically the the solution that May's built allowed you to understand explanations automatically as to who was looking at the record and why and to help manage that and ensure that you know only people who had had or needed appropriate access were getting to it this company was actually acquired a couple of years ago so start up out of Michigan ran for several years and then was acquired another operational example example something that's been developed internally here basically loss prevention um unfortunately this is a real problem where you know occasionally drug cabinet access and um some of the the more challenging um Pharmaceuticals disappearing it's basically ability to try and manage what's been used what's been wasted what's been returned and I identify anything that's you know abnormal then a mobile example uh we have a mobile application that's been developed by a combination of engineering as well as a team in Psychiatry that fundamentally is passively evaluating the voice of bipolar patients to determine if they may be having a change in their care and and essentially it's attempting to determine beforehand whether somebody's going into a manic or depressive episode really cool solution so a bunch of different examples there just intended to give you a couple of key things to think about there's some things from the digital Health standpoint that I would point out are certainly key adoption challenges and things we want to be aware of not the least of which is the FDA is super interested in digital Health Solutions doesn't mean they regulate all of them but they certainly continue to look at things like mobile apps software's medical device the use of artificial intelligence alerting early man Early Learning Systems image analysis and that becomes a big part of it part of the education through ffmi and some of the services that we support here in Innovation Partnerships is to get an early read on whether what you're building or what you're doing research on requires uh regulatory oversight we certainly can help with that privacy huge part of Health Care certainly there's legislative activity around that but there's privacy in the sense of uh technical restrictions and requirements uh that are required to ensure that the infrastructure that you're using meets those needs and then the last two I would mention here are really education these things are all different a lot of them are new they are profoundly different for some of the patients that we're asking to use these things everything from Telehealth to the the mobile application that I just mentioned to sensors and other things that we're asking people to to use so the education on that for for patients is huge and it's a central part to most digital Health Solutions as well as where does it fit in the workflow how do we get clinical adoption how do we ensure that these things are actually being used as used as we intend and then there's uh a lot of interesting kind of uh factors with respect to intellectual property and Innovation management we live in a world where people want to try to get a uh kind of secure position from an intellectual property standpoint usually in the form of patents and on the other end of the spectrum we have a lot of things that are kind of opposed to that in the form of open science and really open access and open business models where we want to make things freely accessible to people so creating that appropriate balance there is something that's part of the discussion that we try to get engaged with early on to understand what that is there's no right answer there's just an answer that's right for your situation um and it's something that we enjoy working through that process with folks then just to summarize that a lot of that stuff that I just talked about is really the big things that people are talking about now as far as adoption challenges have to do with data rights and privacy who owns that data who owns the patient's data how do you manage that in a way that works both from a privacy standpoint but also helps Advance science and care now there's certainly legislative things around data such as HIPAA gdpr and HIPAA High trust legislation things of this nature are are very topical I don't see them changing uh at any time in the near future and then when it comes to all of the AI machine learning and Associated Technologies there's a lot of discussion about ensuring that what we are actually doing with those models um is representative there's lots of concerns about bias based on how the model has been trained and on what data it's been trained on there's lots of concerns about being overfit or overly sensitive to a specific population or a particular situation and then of course going back to the regulatory side validation is a big part of the discussion so is this working does it continue to work how does it change over time and with that I'm going to flip uh the Reigns over to Dr Lanham and he can drive us the rest way through we're happy to like I say ask answer questions and uh take it from there thanks dear so for my piece of this I'm going to give a little bit of background on who I am and how I got to what I'm doing today and then tell you guys two stories of two different pieces of software um that I've been involved in the development of and then with my associate chief medical information officer hat on just do a little tiny little primer on fire web resources um to answer a question that came in before the talk about kind of what they are and where they might be useful and just as Drew was talking about early on about Healthcare analytics it is one of the ways the legislation is helping or has helped improve interoperability and getting data out of the HR so just wanted to talk about that a little bit so next slide please um so I have been in Ann Arbor since medical school um I did my first resident or my residency in OB Joanne here and one of my first forays into the EHR space was as the house Officers Association president being the representative for the house officers and many of the initial my chart implementation conversations and then as a fellow in reproductive endocrinology and fertility continued to kind of be the my chart representative for my division and do some other representation there and continue to be interested in the electronic health record but also this is where I first noted the need for a particular Improvement in a clinical workflow I've always been very interested in workflows and making things more efficient but also patient friendly and so that's where the idea for on track which is one of the softwares that I'll talk about in a second came from and I'll explain more about that in a moment I started as faculty here in 2014 and as part of my initial starting package had times set aside to be what's called a physician Builder which is an epic term for getting training to be able to do specific configuration within epic I slowly learned and then got the security to actually do custom programming and epic as well and then was able to start a consulting company that I still have where I can do custom programming and build for other epic sites and during that five-year period also began to get funding with Cardiology to do a custom epic build at Michigan medicine to support a doec dashboard which is the second project that I'll talk to you about in 20 middle of 2019 I transitioned away from seeing patients directly and um now I'm also boarded in clinical informatics which I forgot to list all the work I do now is informatics related and I've flexed my time from being very part-time to less part-time to full-time over the last four years but I'm currently the associate cmio that covers provider Builders so physician Builders as well as apps that have that same build access that I described custom code and workflows so anytime we can't get epic to do something right out of the box folks can often come to me and we can figure out a way to do it with a little snippet of custom code here and there and then I'll sum the acmio who's meant to understand the web resources and buyer resources and how Integrations with other pieces of software may use those to improve either our provider experience or our patient experience or patient safety I also happen to be the site lead for a new project called nhsn collab which is helping the CDC and nhsn create a new piece of software called nhsn link to reduce the need to do manual uploads of some of the documentation and Reporting that's required and instead use these same bioweb resources I've been working with them for a couple years now which has helped extend my knowledge of what fire can and cannot do uh next slide please so Story number one um as a fellow and an infertility Clinic you're responsible for overseeing the in vitro fertilization cycles for many different patients and you often also had to do the nurse's work when they were on vacation and so one of the things you learned about earlier creating these calendars an example which is on the screen unfortunately with fertility Cycles as things get started they immediately change for what you think the plan is going to be with the doses that the patient is going to take as far as their medications go and sometimes the ultrasounds and blood draws that you do actually mean that everything even gets pushed out a week or two weeks and so what I saw happening was the nurses would spend all this time on this expel spreadsheet and then have to change it immediately or you can see here the patient might have been given a printed copy and then they would go through and scribble out all the dates and change the dates themselves and then even a few days in it was wrong and so what I saw was what I saw what I thought was a waste of time from the nurse's perspective and then from the patient's perspective on updating this thing manually over and over and so I wanted to create a solution that would be some sort of dynamic calendar program potentially on an iPad that would help keep this work from having to be done over and over so that's what I started with next slide please um but early on I also went and talked to Drew an office attack transfer as well as some other advisors there and learned about this value proposition canvas and understanding the real problem and trying to avoid having a solution that didn't have a problem with it and so took this model and then went and found 14 patients and partners and actually talked to them about their Cycles next slide please foreign this is just a little snapshot of um this was 11 patients and three partners and I didn't have really any training on this but it turned out to be really beneficial to kind of learn it on the Fly and sat with them I had asked them to give me an hour and many of them wanted to talk for two or more about their cycle and what worked well what didn't and I took that prior framework and worked on determining what were the jobs pains and gains that the patients had to get to recycle and then even asked them to help me kind of prioritize features of this future application and in retrospect the best thing about doing all of this is there was no programming that happened before any of it so it was very very cheap to figure out what the end user might actually care about and ultimately I found out things that I hadn't thought about at all when trying to think about this this Dynamic calendar application that the real problem that from patient's perspective is that they wanted to stay to know that they were doing everything right as you can imagine folks who are doing an in vitro fertilization cycle usually have infertility so already stress going into this process and it is also very common for patients to make medication errors so to either run out of these meds because they're expensive and hard to keep on hand at home or they are often from a specialty pharmacy or these are people who haven't had chronic diseases and haven't had it to give themselves medication who suddenly are doing daily medication injections so about 15 of our Cycles have some Med air and even in the setting where clinically you might be able to say probably didn't make a difference for Cycles where you really only have at most a 50 success rate the patients who don't have success it's hard to convince them that if there was a problem that it didn't have any implication and so there's a lot of stress that comes along with this too next slide please so several things happened relatively soon after that initial customer Discovery there was a hackathon which I'd never been part of before but somebody told me about it so I went and it was in Detroit and was meant to be this International hackathon because there were people from Canada as well and um met Christina was a nurse in our clinic and I took her to the hackathon and then met Alex who's a developer from Canada and there was a lot of synergy there and we won several prizes at that hackathon for um like the application most likely to become a real thing um and had the opportunity to do a boot camp through spark in Ann Arbor next slide um but when I thought this was just going to be a dynamic calendar I'd applied for some figs function or sorry figs funding and had come up with a spreadsheet based solution and then off of the hackathon and creating the team and doing more customer Discovery we were able to get two more sets of funding with a couple different Milestones the Kickstart funding through Amtrak allowed us to help to use memo Innovations to help us with some rapid low Fidelity prototyping so basically creating some designs on paper and they also had the benefit of going being able to go into people's homes and actually watch them do their medication administration without their a physician being nearby which was very beneficial for getting some more detail on where people were struggling and then we got additional funding with them track ran a randomized control trial at Washington St Louis um Kansas and here and looked at the number of medication errors and also some Stress and Anxiety scores over time in patients who are doing Cycles either with um just tracking things on their own versus using our application um during the RCT a company called new bundle reached out and we're interested in licensing the technology so that started in 2019 and then so that was licensed by new bundle through the end of 2020 or so when they were then unable to raise more funding so that license has reverted back to University of Michigan um but that is in a nutshell the story from the beginning of a concept um having an idea for a piece of software that ultimately we never built a dynamic calendar application it became much more what the patients wanted was a medication tracker and being able to check off when they had done things I didn't talk about a particular feature but we also had an escalation feature so that if patients hadn't checked off doing their medications it could escalate up to the nurses who were doing the study so that we could reach out to them and say hey what's going on we also had transparency onto their medication inventory so that if they were about to run out of medications we knew as long as they knew and that was totally novel because we don't didn't have that functionality before and don't have that functionality within epic today next slide um so Lessons Learned I've said a couple times we were successful in avoiding programming a solution that actually wasn't what end users wanted and learned we're grateful to be able to Pivot away from the initial idea if it's not what they wanted one other lesson learned that I've talked about with folks before is that because of the need to design the product to be able to do the RCT as a milestone of the funding it did drive us to create a much more functional product out of the gate than I would have necessarily if we had been kind of a startup outside the University where basically we had people not paying for this and I would have been interested to have run the experiment on what would be the smallest thing people could pay for um but we we had the opportunity to create more than that um we also had a version that was patient facing that was beautiful and then the site that was Clinic facing that was very much just what we needed to have to be able to have people put in their doses for the day and that was very intentional because if the patient didn't check off that they had done their work or didn't enter their inventory that's where all the data was coming from and so it allowed us to realize we needed to focus on that interface first and make sure that it was useful [Music] um and oh with the the last piece is just about the licensing with new bundle um even though they weren't able to raise funding and that was ultimately where that license agreement disintegrated there were also some pieces of the culture of the company and the way they did their development and their design that I happen to have opinions on but given it was my first time doing it was very excited about the license happening I would just ask some different questions next time around um because I would have behooved me to know a little bit more sooner rather than later so that was just my own learning from that particular license agreement next slide please okay so totally flipping gears as an OB GYN I don't know that I had ever even had a patient especially an infertility patient on a doe act prior to hearing about them in around 2017 or 18. but given the Specialties from which the attendees today are coming from at least in what I was sent I think you guys are much more likely to know about these than I was at the time but director anticoagulants introduced in the early 20 teens as an alternative to warfriend for reducing the risk of things like stroke with aphid or recurrent blood clot after VTE there are many million users in the U.S

and the high utility of it but also the easy uh the ease with which it can be incorrectly dosed due to changes in renal function age weight or interacting medications has led to it being the number one class for cause of adverse drug events since 2018. Michigan anticoagulation quality improvement initiative or Mackie is one of the cqis in the state of Michigan funded by Blue Cross Blue Shield and they had the goal of having something in place in some of their hospitals to be able to avoid this problem of a needle in the haystack of about 15 percent of patients who are on the wrong dose and thus at an increased risk of bleeding or clotting there at Michigan medicine alone there are 12 000 patients that are on this medication and the outpatient side so it is impossible to conceive of the pharmacist or nurse going through all of those patients individually and looking for um dosing issues there was a manual process where a small subset of patients who are on them were being manually abstracted into another database and they had a way for nurses to identify dosing errors at that time and then reach out to those folks and then the VA had implemented a population Health dashboard that would calculate expected dose or look for drug drug interactions or other issues with the dosing and So based on that fact Mackie via Jeff Barnes reached out to me with what I knew about epic to say can we develop something like that in Epic and so um with the Mackie funding we were able to create in Epic the technology happens to be a registry that then filters up to a reporting workbench report that filters up to a radar dashboard but allows Pharmacists and nurses to identify the patients who have potential issues or have known issue based on their clinical information in the chart because Mackie covers multiple hospitals in the state of Michigan one of the goals was to be able to kind of create a build playbook for other sites to implement that we were able to go one step further with epic's help and use functionality they have called turbocharger to actually just move the configuration which took several hundred hours of build time at Michigan and translated it into about eight hours of active analyst time plus about 10 hours of testing at three other hospitals so this is now also live at Beaumont um or what it was Beaumont what was Spectrum and what still I think is Henry Ford and so they're over 6 out 60 000 patients managed annually last year um for looking for these stoic issues so it's really been a boon for finding this needle in a haystack next slide please so that was the beginning and that is the upper left quadrant of this little two by two table where you imagine um hospitals that are on Epic and managing their outpatient sites um we also uh By Word of Mouth other hospitals in the country have found out that we have this functionality and it's not something that epic has at Baseline and so there are several sites and conversations with Ott about a license for the initial configuration and some number of years of support um for updates that need to happen to maintain that at the same time we've gone to Blue Cross Blue Shield and um gotten funding to create a non-epic specific version of this that would use fire resources and I'll talk more about what that means in a moment that would allow us to either plug into an epic version that doesn't want to have to to continue to touch the configuration in the Epic version if things change in the package inserts then right now there is configuration that has to happen in Michigan medicine and then I send the exchanges to the other sites and say this is what you need to change and in the fire enabled version we would be able to make that process a little more seamless from an epic analyst's perspective so we have that funding we have the design complete and or starting the development now and intend to implement at least in one site by the end of this year through the anticoagulation form in FDA we've also gotten funding to create a similar thing that's epic specific on the inpatient side it will use a slightly different technology and that it'll be patient lists rather than reporting workbench but going to fill that gap for the 70 to 70 to 100 patients that are on dox at any one time inpatient rather than pharmacists having to look for those problems daily and then our applying for looking for funding for a fire enable the HR version on the inpatient side right now as well so next slide so with my asml hat on briefly what are fireweb resources fire stands for fast Healthcare interoperability resources it means both the specification for how the data is meant to be structured as well as the technology of the web resources where an outside instance of software can make a call to a patient's EHR or to provider's EHR and return information about a particular patient or about appointments or about several different pieces of data many ehrs have implemented it and there are a couple different things to know about the versions that are relevant for trying to write software that hits those web resources yourself the dstu stands for draft um and then the r is the final is the most recent version the specifications change over time and are not necessarily Backward Compatible but the other thing you might read about are the uscdi versions so uscdi stands for United States core data for interoperability they are defined by onc and they Define what the different web resources should expose from the electronic health record which is meant to then support the ability for one developer to write a piece of software that works with epic via fire and then also with Cerner and then also with Athena Health with having to do a lot of rework there is a link between implementing those versions so it's uh behooves epic to implement those different versions of uscdi so that their EHR certified and then it behooves Michigan medicine or other hospitals to have a certified EHR because of CMS reimbursement however we have learned I've learned through the nhsn project and then with the other projects that we're working on that everything you would want is not necessarily there in those different versions and so as you begin to do a project you're going to want to take a look at what the versions offer and then there are different actually public comment periods um that anyone could comment as they're putting out these newest new uscdi versions so hopefully over the next five to ten years they'll become more useful than they are now but they're a place to start next slide then finally just some other terminology you might read smart bonfire smart stands for what's on the screen it is essentially if you think about if you go to a website and they ask you if you want to log in with your Facebook or your Apple ID and you use your Facebook password to log into something else it's the same concept but instead you're using your in our world my chart password to log into a different product so that that uses the oauth 2 standard for Authentication and it also defines some standards for retrieving the patient data and some bulk data but at its core it's really about the authorization authentication it's not smart like AI smart or any other way you might Define smart it just means how you log into it CDs hooks are a way for providing essentially alerts to different electronic health records with the same set of data or the same logic um but the way the vendor gets to control how they're implemented and so it is they haven't really taken off particularly well at least in Epic specific sites and I think partly because the way epic has implemented what they look like when you get them triggered but it would be a way for you to have one piece of data spread to different ehrs if you wanted to but I'm happy to talk offline about what they really mean and how they would be implemented if that were something you'd be interested in doing next slide I'll try to get through this quickly so we've got some time quote for questions um very specifically the uscdi versions um started with what I think of as significant lack of specificity so you would imagine if it says something like medications you might get stuff like the medication order the prescription or even on the inpatient side when the patient received the medications and in its first version medication just meant defining penicillin for example had nothing to do with patient context and even the most recent versions of the medication resources that are implemented in Epic don't provide things like the patient got the med in the hospital and so it makes it difficult to do some of the nhsn CDC work we're trying to do in figuring out things like hypoglycemia due to medication administration and it all essentially comes down to the fact that vendors don't have to actually implement it in a particular way and then individual customers may also have workflow that puts data somewhere different and then isn't exposed in the fire web resources so the implementation of this type of software you have to be patient with both the vendor and the customer and yourself when you're trying to figure out what data you can get from where next slide and then finally lessons learning on the second piece of things uh being able to support these multiple sites through Mackie has been beneficial for learning the different ways that the Epic customers Implement things and we are also within the Michigan medicine ecosystem beginning to be able to build support workflows or problems that come up such that when this ultimately gets spun off or license we'll already have some of that worked out um it's also been nice to learn that folks want the outpatient dashboard that we've created and get a feel for what some of that might be worth to people will be nice here and um I can go and go to the next slide and then I'll just give a pitch to talking to Drew and his staff early especially about learning to do customer Discovery and avoiding doing programming if you're not 100 sure that that it's the solution like you know what the problem is going out and seeing what people are doing then being humble enough to identify the difference between what you are assuming and what you actually know getting low Fidelity designs like stuff on paper in front of people to quote use in order to get feedback and then failing early and pivoting your design early rather than putting time into development if you don't know what you're going to design next slide I'm happy to take any questions or feel free to email me after the talk as well here's a question Michael and I think uh you're probably equipped to answer this one could you please elaborate on your statement avoid trying to create Solutions looking for a problem yeah so um it can if I go back to my example with on track where I saw the problem as the fact that people were remaking this calendar and spending time to do that when I went and talked to the end users of it I realized that they didn't see the writing over the dates as a problem they actually kind of liked to the writing over the dates in that particular situation because it meant that they knew what was going on but they still had real problems they had problems keeping track of their medication realizing they'd run out at 9 00 PM when you can't go get a new medication so I in going and talking to the people who are the real users of the the workflow I was able to identify those problems Drew I might ask you if you don't mind can you although it can you think of any examples in your career of something similar um where people have had this really exciting technology better than trying to push it back into a to get it out there without a real problem yeah I I mean I I I think there's numerous examples that are you know out there in the workplace that are are just in real life where people things have been built that they could not seem to get off the ground I mean the I I only laugh about this one because some people will be familiar with it Apple developed a product called a Newton which fundamentally was an iPad but it was about 25 years too early and people just weren't ready for it right so and I I think that fundamentally goes to the message you're trying to communicate which is customer Discovery is huge right your ideas are are probably um well-founded but it's you know always amazing what you learn when you talk to people who are potentially the end users going back to the the calendaring function and what's the real problem right so uh talk to people get that feedback is is certainly the message yeah John not sure if there's anything else no I was just just going to plug you know the FF my fast-paced course for that concept around you know speaking to different stakeholders and conducting custom Discovery and understanding problems before developing uh elaborate Solutions so I'll I'll put a link to that in the chat here and give you a chance to to venture over to that website if you have an interest um but happy to answer any questions related to that course uh that might come up um and we are at about one o'clock so unless there are additional questions let me check real quick nope um I'll thank you both again for your your time this afternoon um we really do appreciate it uh both great presentations I'll remind the audience we will be sending a copy of the uh recording as well as as the slides and then again an opportunity for you to fill out an evaluation uh in an email shortly after we get off the call here today so thank you everyone for for joining us uh have a great afternoon and the rest of your day

2023-02-15 02:40

Show Video

Other news