Benjamin Kummer, MD, talks about telestroke and the future of teleneurology at a virtual session hosted by the American Medical Association (AMA) in collaboration with the American Academy of Neurology (AAN).
Chapters (Click to go to chapter start) Telestroke — definitions, history, and trends: The Basics components of a telestroke system: Why is telestroke such a good use case?: Telestroke care model: Telestroke evidence: Examples of asynchronous teleneurology and the future of teleneurology: Remove data sharing and patient monitoring: The Future: New Types of Data: Now: Interprofessional Consultations: The future new modes of communication/interactions: Now principal care management: The future: care coordination to support value-based care: The future: new ways of measuring quality and effectiveness: The future: emphasis on "techquity": The Future: artificial intelligence: thanks to the AMY and the A. N. For putting this terrific program together. I'm really honored to be here. Speaking alongside my distinguished colleagues, it's really very large shoes to fill to follow all that excellent material. Um Today I'll be talking mainly about tele stroke and the global 30,000 or maybe 10,000 ft view of the future of tele neurology. Um and I want to leave some time for questions so let's jump right into it. Next slide please. These are my disclosures. So talking about tele stroke 1st um next slide please. We're going to go over some brief definitions first. Um telescope is really the application of synchronous audio, video, telemedicine, two stroke neurology. Um And it's almost this sort of classic architectural tele neurology use case. Um It was first described in a Very very influential paper in stroke in 1999, Steve Levine and Marc Corman sort of wrote an opinion piece around it and shortly thereafter in the same year, also in stroke, a paper was published that compared NIH stroke scale assessments which is the physical exam assessment in stroke to video and a stroke scale assessments and basically found that both of those were essentially accurate and related if you will. So um since then we've we've seen a steady increase in tele stroke programs worldwide and in the United States given that physical examination over video and in person were essentially the same and as of 2019 at least a quarter of U. S. Hospitals have access to Tell a stroke programs and this comes from a recent paper from 2020 and that's probably an understatement given the fact that a lot of that was survey data from tele neurology company networks and Tell neurology and academic centers. Next slide please. So the basic components of the tele stroke system are very similar to the components of other systems. But as dr abuses mentioned, the setting in which stroke care typically occurs is not it's not typically at home. So there's always going to be a remote physician who's initiating the call for tele stroke too. Um Sorry the remote physician is going to be handling the call for for tele stroke from an on site physician or other provider who is typically located in an E. D. Or a mobile stroke unit and more rarely on the hospital floor. There also needs to be some kind of synchronous communication technology and infrastructure and system too. Work that technology. And finally a staffing model needs to be present as well. Next slide please. I want to just spend a little bit of time talking about why Tell a stroke is sort of a classic excellent use case in neurology and it really comes down to access to specialized stroke care which is a huge problem in our country and worldwide really. First acute ischemic stroke which is the most common cause of stroke arguably In this country affects about 700,000 patients every year. And as we all know strokes can be completely devastating and this disease imposes a huge social cost and burden to the public health system. Um The standard treatment for acute ischemic stroke which is known colloquially as T. P. A. It's generic name is altered place which is an I. V. Medication given as Ebola and then a drip over a specific period of time is really under utilized in the community either because of misconceptions around the evidence supporting T. P. A. Use. Um But also because there's a very very big shortage of stroke neurologists and stroke capable centers especially in rural areas of the United States. And you add to that the fact that T. P. A. Has a very very narrow therapeutic window you really have. You really should be giving T. P. A. If you're practicing by the guidelines at least between zero and 4.5 hours from stroke onset or last known well and the benefit of T. P. A. Tends to go down the further you get from the onset of the stroke and the greater the risk of hemorrhagic complications. And finally one of the reasons that makes it such a good use cases that you can really make the decision as to whether you want to give a patient T. P. A. And whether they're eligible or not really easily over audio video telehealth next slide please. So um tele stroke care models are basically divided into two general buckets. Um the hub and spoke model and the so called humblest model. Um the hub and spoke model refers to the analogy of a wheel if you will, where there's a center central hospital at the hub of the wheel that then is connected via telehealth relationships or audio connections to spoke hospitals that are on the periphery. You can see that in panel a right there on the top left, um and within the hub and spoke model there are really two different paradigms of treatment. One is called drip and ship and the other one is called drip and keep. I did not make these um descriptions up. These are actually legitimate descriptions in published literature as you can see in the upper right hand side there drip and ship just refers to treatment of patient with T. P. A. At a spoke hospital and then transfer to a hub hospital as you can see in panel A and dripping keep is just the patient gets TP at the spoke hospital and is kept at the spoke hospital for further evaluation work of placements whereas in drip and ship that occurs at the hub hospital. The contrast that you have the humblest model which is, it can sort of take two forms. One is either around a private practice group um and another is around a telemedicine company which is kind of functionally the same as a private practice organization. But if you think about it at a higher level, basically a humblest system is where the decision making is central but the transfers of patients is distributed process. So you can sort of see that in panel D to the right where there's a private practice group that is covering a bunch of spoke hospitals for decision making as to whether patients get treatment or whether they need to be transferred and then those transfers occur to other hub hospitals that are not necessarily related to the central private practice or in panel C the central telemedicine company. Next slide please. So dr busses did mention some of the evidence briefly in his presentation but there's a lot of very good quality data supporting the use of tele stroke and throughout The last two decades since its appearance and its publication and stroke in 1999. Um So as I mentioned before, neurological exams done over tele stroke are really accurate and basically the same in terms of their accuracy compared to in person examinations. Um And Telstra in a nutshell increases access to specialist stroke care and that includes treatment work up stroke units etcetera. So unsurprisingly in the literature telescope has been found to be associated with increases in thrum politic rates in the spoke hospitals or other types of arrangements. Um And because T. P. A. Improves mortality in patients with acute ischemic stroke and also results in improved functional outcomes at 90 days and further out uh in patients with ischemic stroke it's because your access is increased through tele stroke. It's unsurprising that telescope is really associated with improved mortality profiles and better and better effectiveness compared to in person care um in hospitals that do not have in person stroke coverage. So, you know, it's really essentially echoing what dr busses had mentioned. Tele stroke is the practice of stroke. So really increasing access and so the literature really reflects that. I think it's important to mention that most research on tele stroke does come from single center academic institutions, many of which are at the center of the hub and spoke model. So while those things are, there's very robust evidence and I didn't list all of it in the slide because it would probably be a whole topic of an hour long talk. If I did, all this information should be taken with a tiny little bit of a grain of salt. Huh, Next slide please. So I'm going to shift gears here and talk about asynchronous tele neurology in the future, where we're going with tele neurology. What I'm going to do in the next few minutes is I'm going to sort of talk about what's happening now in asynchronous tele neurology and various asynchronous services and try to come up with a prediction in the following slide around what I think is going to be happening in terms of general themes. Next slide please. So I think we're all witnessing a global shift of the health care system to going from synchronous care delivery systems to asynchronous care delivery systems. And we're seeing a shift that has been seen in banking and retail already. And health care is just a little bit behind that. And we're also seeing a convergence of telehealth and tele health systems onto mobile device platforms. Um essentially as you can see in this graph here we're moving from a conventional sort of centralized synchronous hospital based sort of live one on one. Um And audio visual interaction to a system that's more distributed, more convenient less centralized in a synchronous that doesn't involve live interactions one on one in that in the weather it sort of has been previously. This graph is actually from an excellent review of tele neurology from about five years ago in nature. Reviews neurology by Ray Dorsey. I highly recommended if anybody's interested in reviewing sort of thought pieces on tele neurology is really excellent. This graph actually chronicles the R. C. T. S. That have been done in telehealth according to the setting in which they have been conducted and in orange below you can see the year of the first RCT inception where they actually started and you can see just chronologically there's a shift from hospitals to clinics to home to the mobile phone. And it's very similar to what we've seen in finance or in retail where you have something happening, a transaction happening at the teller or at a store. Then you know moving to an A. T. M. Or a mall then web based transactions and finally converging onto smartphone transactions. So healthcare is really kind of going in the same direction where we're not going to be dealing with physicians as much face to face in a synchronous fashion. Next slide please. So as dr Morris mentioned, there's a lot of activity around remote patient monitoring and collection of data by sensors and photographs and other types of devices and wearables and sending those data streams to responsible providers that then make decisions or assessments on that data. Um and I think it's important to mention that a lot of those systems are not well integrated into the HRS. Um and that's something that's definitely a gap that needs to be further explored. And I'm contrasting remote data sharing and patient monitoring here in the sense that patient monitoring really there's an expectation a frequent data monitoring and decision making that may be more relevant in the blood pressure management program where there's much more fluctuation um and remote data sharing where the data is being transmitted but there's no expectation that the data is being monitored in real time or quasi real time. And so sort of to address this gap of integration. We've actually piloted something at Sinai where we are using the apple health kit and google fit architecture and I'll talk about that in a second to collect. Step counts in patients with multiple sclerosis and step counts have been shown in the literature to be associated with disability and multiple sclerosis. Um And the beauty of the system is that it actually um leverages the patient's own device uh and essentially connects epic through a flow sheet and my chart too. A patient's individual device and to their their their apple health google fit app which then can connect to other wearables um that you may have or that already exists in the sensors of the phone itself. So it starts with the physician order that gets sent to my chart. The patient then sinks their device to my chart. And then once that happens there's passive data flowing into the HR. So we're getting for about 20 patients over the last year. We're collecting daily step counts in M. S. Patients that they don't even think about. It just uses the standard step counts that they already have and patients and patients get that information reviewed by the M. S. Physician at the time of their visit. So it's not real time monitoring but it's more remote data sharing. Next slide please. This is just our cumulative enrollments. Um so we're up to 20 patients at this point since April of 2021. Next slide please. I think towards the future we're gonna be talking about and seeing many new types of data being integrated and incorporated into our decision making into our armamentarium of patient assessments and I think one of the main categories that we're going to be seeing this in is an imaging. I think we're going to be seeing a lot more transmission and analysis of video data namely of physical examination and neurology um and other phenomenology such as paroxysmal events like Estonia or seizures. Um We're also going to be seeing a sort of higher level integration of complex time locked physiologic data, streams of wearables. Right now there's a lot of integration or just single streams of data from single wearable single sensors. But I think we're going to be seeing a lot of computational lee complex integration of all of those different wearable signals to phenotype, different motor disorders for example or standardized assessments and disease and that's actually that's already starting to happen in Parkinson disease. We're also going to be seeing the digitization of social social network activity, sociability and displacement and and what somebody's digital life space actually is and what it means for their condition. This can be particularly relevant for patients that have cognitive disorders or visual spatial disorders. And we're going to be seeing a lot of integration of more treatment data namely around symptom diaries, responses to various therapies and adverse events. Next slide please there was some mention in a number of my esteemed colleagues presentations on inter professional consultations which are a form of asynchronous tele neurology And we've actually launched any concert program in neurology in October of 2019 which is all epic integrated. Um Similar to the ask a doc program at Geisinger um where the epic order basically is initiated by the requesting physician that contains a clinical question that's pictured in the lower left hand corner here. Um It has a lot of clinical history, very very detailed clinical history for a variety of different symptoms. This once the order is signed routes to an in basket pool which is pictured right here on the lower The lower right hand side. Um that's staffed by five different neurologists for each of the day of the week. Um and the consultant will then feel the question complete any consult note and epic and then drop a bill. The exact code is 99451 For inter professional consultation and the note will actually route back to the individual requester. We've been pretty successful in this program. We've done about 430 consultations since we started. Most of these referrals and consults really come from Medicaid internal medicine clinics in the Sinai system. Next slide please. This is actually graph I wanted to show so out of all of the inter professional consultations that we've done in neurology. Um The majority have actually resulted in an electronic recommendation and the majority of those electronic recommendations basically avoided a referral for an in person visit in our neurology clinics which are chronically overbooked and the wait times for which are extremely long. So I think this is a big win in the sense that we're reducing improving access to neurological services without worsening the patient experience if you will and a number of other important outcomes. Next slide please. I think we're going to be seeing many different modes of interacting between patients and providers and communicating between both parties as we move towards the future. I think the technology of chatbots is going to become much more sophisticated than what it is now, namely because the artificial intelligence algorithms and natural language processing algorithms that are developing and will be developed are going to be much more sophisticated in handling and parsing text and illness scripts and triaging such that patients will really have 24 7 availability, two asynchronous care if they have a question about their medication or they have a symptom that they don't really know what to do with. And I think there's also going to be the leverage of sort of more accessible traditional technologies like text and smS to do digital outreach and surveillance and vulnerable populations that may be on the other side of the digital divide, which I'll speak about in a few slides, um sort of in parallel. I think digital patient journeys are starting already starting to be developed and will continue to become much more important around symptoms. I think patients are going to stop going through their physician networks to kind of get referrals and I think they're going to start with technology to find a referral and intelligent way and to be connected to resources. So a few examples of that are apps that are so called digital front doors. Um Health systems are going to be developing their own apps which are basically health system, all things health system on an application on your mobile phone. And patients will start to search for their symptoms in these apps and then be connected to resources providers, different tools to manage the disease as opposed to going through the traditional channels. Again, this is entirely asynchronous, It does not have to take place during an audio visual or face to face encounter with the physician. So again, just kind of going back to that slide with the Ray Dorsey figure, we're really going towards a more sort of distributed model um and we're also going to be seeing similarly in the private sector doctor on demand services for specific disease verticals like Rohan Hims, roman hymns have established themselves really in the male urology and sort of pattern baldness vertical but there very shortly going to be branching out into additional verticals um so that patients can be connected to providers immediately based on symptoms and requests. Next slide, please, I want to briefly talk about PCM since a number of my colleagues mentioned this um just to reiterate PCM is a service that essentially leverages the entire health team which can comprise providers a P. P. S. P. T. S. S. LPS to what have you um for the care, coordination and management of one chronic condition for at least three months per team. And this is a time based service. So you have to accrue a certain number of minutes every Months before you can before you can build a service. But um we've actually built a PCM module and epic and are piloting this at Sinai with a number of different neurological conditions and we've received an $80,000 office of well being resilience grant to build this module. And this is from scratch because epic does not actually have anything like this out of the box. This allows the assignment of care plans and goals. Care targets, task tracking. Um It's very complex, very cool and very proud of it personally. Um And on the right hand side here you can see the enrollment form because as dr and dr abuses mentioned consent and sort of patient involvement in decision making is key in in rolling a patient the program. And then in the bottom you see a smart form where you can track your time for the services that you may render. Um So we've launched this in A. L. S. Q. Four of last year and we just started doing this in M. S. Um and if you can just scroll through the next animation there's a approval um graph here where you can see we're up to about 23 patients in A. L. S. Um and six patients, 16 members that have been trained. So let's go to the next slide please. And shifting towards the future to echo something that doctors everybody said. Principal care management obviously requires care planning, that's part and parcel of the PCM service. There's goal setting around those and embedded in those care plans. But I think an important thing to mention is that PCM doesn't really specify how the provider of the care team interacts with the patient. It can be a synchronous, can be synchronous, it can be on the phone, it can be a digitally mm service. It can be a variety of different things. Um and that includes data review of a synchronous wearable information like we're doing in step counts mm. S. Um That may not be billable by the the traditional sort of by the book cPT coding guidelines for remote patient monitoring because it's not enough d a approved device and so on. You can actually count those services towards PCM time spent and towards PCM building. So I think to echo doctors at this point, what we're gonna be seeing in the future is that PCM and these care coordination services are really going to be blazing a path towards value based care and value based payment models which have been talked about for a very long time and really the minority of a lot of services um in the United States but will probably become more and more important as we go forward. Next slide please. We're also going to be seeing a lot more high quality evidence on the value proposition in tele neurology in the covid pandemic context and sort of after the covid pandemic or in the covid pandemic that we're obviously moving through. Um And we'll also be seeing a lot of a lot more alignment of professional societies like the A. N. With the evaluation of tele neurological care and care quality um with formal quality frameworks like that developed by the N. QF. As well as the framework that you all have developed at the A. M. A. Um and an increasing shift towards digital patient reported outcome measures to measure effectiveness of tele neurological care. Next slide please. I think I would be remiss if I didn't mention the buzzword that is tech witty tele neurology I think is also going to increasingly attempt to address the digital divide and the issues of equity, diversity and inclusion that we experienced during the massive shift to digital medicine and digital neurology by that extension during the covid pandemic um I don't think this is something that health systems and practitioners can really ignore anymore. This is a very, very hot topic right now and I think what we're gonna be seeing in tele neurology specifically is efforts towards increasing accessibility to very well established technologies like texting for vulnerable populations as well as decreasing barriers to newer technologies that may require greater digital literacy or greater access to broadband or other technologies such as video synchronous video platforms and other wearable devices. Next slide please. And finally I want to sort of close with the future um state of tele neurology with respect to artificial intelligence which is going to be increasingly present in our world not just in neurology but really everywhere in medicine I want to break down different uh Conceptual Categories of Artificial Intelligence. There are three of them namely assistive ai analytics ai and autonomous Ai assisted Ai is really stuff that you've probably seen in your practice or even in your in your daily routines. It's really the pattern recognition and automation of simple processes. One really really good example of that which is still kind of in its experimental phase is detection of seizures automatically from an E. G. Waveform. It's not making decisions, it's not acting independently as a human being but it's recognizing a pattern and maybe it alerts you of it. For example analytic ai is a little bit more complex in the sense that it sort of enhances or adds to human decisions. A good example of that could be a chat bot that detects an illness script of a headache. Um a specific type of headache and suggests multiple choice of diagnosis or hones in on the diagnosis of migraine or trigeminal neuralgia and then notifies the physician and suggest something but does not make the decision that this is trigeminal neuralgia and have to start carBAMazepine for example. And then finally the sort of more aspirational further off and almost slightly scary. Ai to be perfectly honest is the autonomous AI which is designed to act independently of humans and really make decisions on its own. I think a good example of this which we're going to be tending towards in the future is a smart home or smart exam room that has multiple cameras, voice recognition technology that sort of leverages the same technology that we use for amazon um and and Alexa and Siri um in motion sensors. I think that's going to be increasingly prevalent in cognitive disorders, movement disorders. For example a smart home could identify that the patient's fallen and automatically activate the M. S. Automatically make some referrals automatically notify the care team and sort of act independently to a certain extent. So that's still a ways off. But it's something that's coming on the horizon and we're going to have to talk about it to increasing degrees next slide please. I think it's very important to mention that Ai in the future. I don't visualize this and this may be a topic for debate but my opinion is that AI is not going to replace providers. I think it's really gonna enable providers to practice at the top of their license. It's going to be able to automate a lot of the wrote sort of less complicated stuff that physicians have to do day to day and really allow them to focus on the real nuances of their specialties or taking care of their patients clinically. Um I think it's patients will still need human touch that's inevitable. And so in that sense Ai is not here to replace providers. I think it's really here as a support of technology and I think we're going to be seeing an increasing amount of consideration being devoted to equity and ethics in Ai model design. I know that there's a at Sinai for example, there is a committee on artificial intelligence, Equity that was recently formed to regulate the way that we put together our models and integrate them into the health system to make sure that this is equitable for everyone. I will say though that as we move towards more asynchronous distributed services with increasing amounts of touch points between in person visits. Um And I do I would argue that the frequency of in person interactions may decrease because touchpoints outside of these synchronous visits will increase. But again, I do want to make the point that patients will still need that human touch and they still will Ai will not be replacing providers in my opinion