Ep. 19: Alistair Erskine, MD, MBA, Chief Digital Health Officer of Mass General Brigham. Topic: Healthcare Innovation
Kathy: Welcome to Episode 19 of the Smarter Healthcare Podcast. Our guest today is Alistair Erskine, chief digital health officer at Mass General Brigham, a 14-hospital system in the Boston area.
Alistair is helping the hospital system to adopt new, innovative technologies to improve the patient experience and clinical care.
In this episode, we talk about how Mass General Brigham adopted new technologies to respond to the COVID-19 pandemic, and what the health system’s priorities are in the next several years.
I hope you enjoy our conversation.
Kathy: Alistair, welcome to our podcast today. Could you start by telling us a bit about yourself and your role at Mass General Brigham?
Alistair: Sure. I’m the Chief Digital Health Officer at Mass General Brigham. I’m an internist and a pediatrician physician as well. And I’m responsible for all things digital health across the organization, which include things like the electronic health record, what we do in terms of a digital front door for patients, the patient portal, our virtual care services and capabilities, that kind of thing.
Kathy: Great. Now the past year and a half was certainly disruptive for all of us, both personally and professionally. How did it change your technology priorities at Mass General Brigham?
Alistair: It certainly was an incredible unprecedented time. And we had the opportunity to learn a lot, from the things that were necessary to be done almost immediately. So, I think the good news is, we were on a chassis of a healthcare operating system, we use an electronic health record system, which worked really to our advantage, because when we needed to make changes that were almost immediate, we were able to promote those changes across the entire organization very quickly. The kinds of changes that were necessary initially were embedded into the system – different testing and diagnostic capabilities – and also extracting from the system the capacity of ICUs and of wards to be able to manage and take care of patients that were becoming sick, at a time when we really didn’t know a lot about COVID. And, of course, the immediate challenge was, with social distancing requirements, that we needed to virtualize a lot of things including ambulatory or outpatient visits. So we went from approximately 75 virtual visits a day to 12,500 virtual visits a day. So a fairly important increase. We had to actually swap out our video vendor just to be able to handle the volume of video calls that were coming through. And on the inpatient side because there was, if you remember, such a reduced availability of personal protective equipment, we had to find creative ways to still check on the patient without necessarily having to put on the masks and the gown to enter their room every single time. So we actually invested in a lot of iPad devices and built software to have an always-on virtual presence in the room. So that the nurse – he or she could check on the patient on a regular basis without the patient having to do anything, and so that consulting physicians, even if they were in a different part of the hospital or in fact at home or in a different hospital, they could still log in and see and take care of that patient virtually. So that virtualization became really important, as an aspect of care. I think that’s probably the thing that changed the most is not only the technology associated with video visits, an iPad and software you had to build on the fly, the process, in terms of getting people digitally upskilled to use that technology, and including the patients that needed to get digitally upskilled to be able to conduct these visits at home that may not have been used to it.
Kathy: And what do you think those important technology lessons are that we learned from the pandemic?
Alistair: So I think one of the most clear things we learned is the fact that we’re going to have to be a lot more nimble as a healthcare system to accommodate whatever comes at us, including a pandemic, or any future threats that come to our overall health as a system. I think the other piece that became important to consider is a public health infrastructure that is designed to support our state, and across the different states in the U.S., really has been dramatically underinvested in. In such a way that when it came time to rely on it, for example, understanding what the immunization registry information was, so every time a patient came in was getting vaccinated, why don’t we double-check before we invited them, did they already get vaccinated somewhere else? The volume that we needed to be able to check on patients every single day was not something that the immunization registry could handle as they got a barrage of requests from, as you can imagine, healthcare systems and hospitals and clinics all around the state. So it kind of gave us pause, and made us realize that that’s going to be something to consider in the future, in terms of how do we sustain and set up and prepare for another threat that comes along with a good public health infrastructure to be able to support that demand. And I think the other piece that was important is even though we have 7 million patients in our database and we had approximately 5 million patients that it was important for us to reach out to to invite to become, for example, vaccinated, some of the data that we had was not accurate about our patients. So it became very clear that we needed to have a better ongoing process to monitor phone numbers, addresses, e-mail addresses, and so forth, on an ongoing basis to make sure that we can always reach patients when we need to, not just when they’re trying to reach us, when we’re trying to reach them. And then the final thing is really important to bring up is what we didn’t want to do is exacerbate the digital divide that exists when, especially in COVID-prevalent areas, there were patients that weren’t endowed with a smartphone or technology or connectivity, yet we still wanted to reach out to that population. And so we had to engineer all kinds of new ways to reach out to a population that is harder to reach with traditional technology, means a patient portal, a smartphone, a text, and come up with new ways. And, in fact, we ended up using things like conversational AI bots that would call the patient and then interact with voice with a patient to be able to book an appointment, as an example. But these were things we weren’t doing before, not as mindful as we are now about making sure that when we offer a technology solution to one set of the population, we certainly consider alternative means to be able to reach all patients, based on, whether it’s language difficulties, technology connectivity difficulties, any race or ethnicity differences, that we want to make sure that we account for everyone when we try to reach out to them.
Kathy: As you said, you had to do a lot that you hadn’t done before. Do you think after the pandemic is over it will be as easy to innovate in healthcare?
Alistair: I think it will be – yes, I think it will be as easy. Maybe the other part of the question is will it be easier? In the sense that we had to act very, very quickly and change processes on the fly, sometimes multiple times per day. Information was flowing through in such a manner that sometimes across the system, what would come out of the CDC or what would come out of state requirements, the mandates would change frequently. So I think that puts an organization, especially of the size of Mass General Brigham, on its toes in terms of how do you respond that quickly? What was a silver lining of the pandemic is because we were in incident command structure, in fact we were in incident command for over a year, so we had – Paul Biddinger, and Ron Walls, and Peter Markell, led the process of putting us into an incident command, to be able to respond quickly to the needs. Our decision-making ability was dramatically accelerated. Because there wasn’t time to pull a whole committee together and get agreement on every different component we needed to be done, it was much more of a fast decision-making process. That actually ends up helping when it comes to what to innovate on, because you don’t have to go to a number of people to sort that out and figure it out, you can go to a smaller set of executive leaders to understand what’s the most important thing to innovate. That ends up being a really important part of innovation is innovate on the right thing. The right problem to solve. Innovation that’s unmoored or not couched in the problem that is in front of you tends to be more investigatory, opportunistic as opposed to innovation that’s actually anchored into something that is a problem that’s right in front of you. The other piece is we learned and came to appreciate the importance of iteration across the innovation spectrum, in terms of we didn’t wait for anything to be perfect, we had to put things out there as they came and then iterate and make it better as we learned more, as the information was slowly flowing through. And so, I think having that skill and that capability across the team that deals with technology and that has come to expect that things need to occur faster and more iteratively, as opposed to what’s referred to as a waterfall way of being able to pull a project and pull something together. I think we got more comfortable with learning as we went along, and making core adjustments, than we had before. Of course, there were – Mass General Brigham is chock full of clever, innovative, brilliant minds, but being able to tug all those in the right and similar direction so that they were actually able to scale very quickly, I think that muscle that we built around how to innovate as we needed to will perpetuate into the future.
Kathy: Now a lot of times when we start to talk about innovation in healthcare there are some larger tech companies that are starting to get into this like Google and Apple. What do you think of them trying to enter the healthcare market?
Alistair: So I think it is critical that they do enter the healthcare market. And the reason I say that is because I don’t think that if we’re really going to transform healthcare we’re going to do it on our own as a healthcare system. First of all, there are an enormous number of players across the entire supply, value chain, but large tech companies bring very important expertise and capabilities to a healthcare system. We’re not a software development shop. We certainly do develop some software, but we’re really in the business of improving the health of a patient and being the academic medical center of the future that puts the patient at the center of everything it does. We need partners in technology that all they do is engineer and software develop and build in technology devices. So I think that partnership becomes really important. Now will we always be in a position where it’s not a competition and a collaboration? Well, most likely. And I think of organizations like Amazon with Amazon Care, and its desire to take on some of the virtual primary care business. Same thing with CVS Health. CVS Health is also, with health hubs around – even around Boston – is trying to offer a service to their customers that overlap in some of the services that we develop. So I also see that not only can large technology companies bring their expertise in technology and machine learning and software development, but they can also bring their market impact in terms of consumerism and setting expectations for consumers, which ultimately are patients, and make sure that we then respond with the right approach in terms of the products and the services that we offer as Mass General Brigham to make sure that they are convenient, and they’re close, and they’re inexpensive for patients. So I think it will push us, so we need partners, we need some competition to push us to do better, and I think we do that with these large tech companies.
Kathy: Now let’s look a little bit to the future. As you look forward to the next year or so, where will Mass General Brigham be focusing its tech resources?
Alistair: We’ve made some great investment in the electronic health record. I think again, like I was saying before, that really serves as a kind of health care operating system. The next place we’re going is we’ve made some important investments and in partnership with Microsoft relative to the Azure Cloud sets of services. So a data ecosystem that is being built to be able to take advantage of all of this data that people are dutifully entering on their keyboards every single day. Both patients and providers and everybody else that helps manage a patient through a healthcare episode. All that data, that rich - the exhaust of all this digital activity, has a huge amount of value and can be reorganized in a different way for analytical purposes that then can provide insight both operationally at the point of care in terms of dashboards and situational awareness, but also in terms of insight even in the form of big data streams whether it’s a physiologic monitor that is informing what’s going on with a patient’s blood pressure, whether it’s information that comes from their home, in terms of devices and sensors. All creating a digital twin of that patient that we can then run a various number of AI and machine learning algorithms on without affecting the patient directly but giving us insight into what could be the next best treatment for that patient. So I think we’re clearly interested in organizing our data in a way that we can use it operationally and for advancing the insights for a patient. I think the other really important area that we are also invested in is our consumer relationship management approach. So we have, of course, the electronic health record holds information about the patient, knowing what’s going on with their clinical case, and in some cases on the back end in terms of how to bill to the various different insurance companies and so forth. But we don’t necessarily use that system to collect all the information about the patient that we want to - in terms of their preference, in terms of how they want to be communicated with – is it e-mail, is it through the portal, is it through snail mail, is it through a text? And we don’t usually keep information on what their experience was over the past, for example, 12 months, to know where are we not doing a good job in terms of the patient’s convenience and the patient’s experience? And how do we – where do we surface that information in our system? So when you call Mass General Brigham, we want to make it so that we already know who you are right when we pick up the phone. So you don’t have to explain your name, you don’t have to explain which doctor you normally see, which clinic you normally go to – that information is at our fingertips somewhere in the system, we want to surface it to the right agent that’s answering the phone. And similarly, when taking care of you as you’re navigating seamlessly across the entire system, independent of which part of the system you’re navigating, that experience should be the same. And we want to store how you prefer to navigate through that system, so that you don’t end up being the human interface across disparate processes and technology, but instead we have a mechanism to track that. So that’s a lot of what consumer relationship management tools do, very well-utilized in other industries, not so much in healthcare, but I think that we’re making a bet that that is actually an important missing transactional system in the healthcare platform ecosystem, and we want to learn by trying to implement and then seeing how much value that actually adds to the patient’s experience.
Kathy: You briefly mentioned AI and machine learning before. Where do you think that market is headed in the next several years?
Alistair: So in the next several years I think there will be a lot of excitement about AI and machine learning, and deep machine learning. And I think that there definitely is – it’s important to continue to learn about AI and machine learning. And I think the challenge is going to be, a lot of things that were previously referred to as more traditional data analytics is now referred to as – everything is AI. It’s almost like somebody found the software box and put ‘New and Improved AI’ on it. And I think the challenge is there are brand new problems that we’re going to have to wrestle with. So number one is that AI is dependent on the data that feeds it. And that data in some cases can be very biased, either in the way that it’s inputted, or even in the way that the population is organized within one area of the market. You move AI from Boston to Los Angeles, you may not get those algorithms working as well, because of the bias that’s inherent in our population that’s here. The other piece is that the AI can be interpretable or it can be non-interpretable. So when it’s interpretable I can tell the reason why something actually is suggested by the model. When it’s not interpretable and it’s using spears and mathematical models and vectorization to be able to figure out what the ultimate answer is, it becomes more difficult to kind of go back and try to figure out why did the AI suggest option one versus option two? So I think that’s another piece about it. And then the third thing is just because the model was working today doesn’t mean the model is going to work well tomorrow. It may need to be re-trained. It’s going to have to learn. We’re going to have to constantly go back through our governance model and figure out how to support it, how to remove it when it’s time to remove it, and so that whole life cycle of AI from concept to building the model in the first place to then implementing that model and then retraining and governing that model I think will be things that we’re wrestling with. And we’re not alone wrestling with that. I think the industry is wrestling with that, and the FDA is also trying to understand what role should it play in terms of either policing or managing or trying to regulate what occurs with AI. Is what has traditionally been called decision support in terms of alerting somebody of a drug interaction, how much does that then convert over to AI and is considered a complete different family of not clinical decision support but clinical data science. And so I think – it’s a spectrum, it’s not a black and white type of scenario. And so I think we’ll have to also try to understand how to accommodate this new advent of AI as it applies in clinical care. And the other piece is clinical care is not a place where you can just sort of casually experiment. These are life and death situations. So we also want to make sure that they – first, do no harm, that’s kind of the motto that we live by as clinicians. We have to make sure that’s the case. And there are clever ways we can ensure that it’s not doing anything untowards, but just having to stay on top of it, making sure we can maintain it, making sure that we have it documented so the person that invents it and goes off to another organization, that we know how to continue to maintain it. Those are the kinds of things that we have to worry about.
Kathy: Now if we look a little bit further ahead, are there any tech advancements that may be in the early stages right now that you think hold a lot of promise long-term?
Alistair: Yeah, so, clearly there is an important aspect of moving from a fee-for-service model to a value-based care model, and that requires redirecting the way that we look at patients in general. So in a fee-for-service world, for example, it’s a lot about volume and episodes of care. In other words, somebody comes in, they have a transaction with the healthcare system, and that’s how we kind of repeat that process, almost like a vertical view of the world. When we move over to more on the value-based care side of the world, it reorients things more horizontally. Where we’re taking care of populations of patients, we’re identifying cohorts of patients that need more additional types of care. So there’s going to be a lot of investment. Mass General Brigham is committed to reducing our total medical expense, in terms of how efficient we are in taking care of patients and populations of patients. And so that invites the idea of remote patient monitoring for patients that are willing to consider that. It could be a scale for a patient that has heart failure. Or a glucometer for a patient that has diabetes. Or a blood pressure cuff that’s connected to a cellular network for a patient with hypertension. That data will come in, and then along with what’s available in the clinical record, and ultimately will be available in the CRM, in the consumer relation management tool, pull that together to be able to figure out what’s the next best thing to do for that patient, and then when are they scheduled to come in, and how can we bundle things together to make it more convenient? How can we get care out of the hospital, out of the clinic, into the home when it’s appropriate? Even concepts like acute care, like home hospital type of environments, where the patient needs to be admitted, comes to the emergency department, and instead of being admitted to the hospital, gets admitted to home with all the services they need to be able to do home hospital care. We actually do that today. We look to increase the amount of home-based care, and technology to support home-based care, for patients that are acutely ill or for patients that just need clinic management. So I think that’s one important area and future direction that we’re going. I think another one that we should consider, especially after pulling together a data ecosystem, is we sit on a trove of data that can inform what to do with a patient when there is no evidence, or when the expertise is not there, about our patient. So not every patient comes in having read the textbook of medicine, and presents in that way. And not every patient do we have randomized controlled trials that help us understand what’s the next best thing to do. And in those situations, we need to get better at figuring out a way to tap into that data, with the right privacy parameters in place, and the right security parameters in place, and the right set of people with the right training to be able to access that, and get that information to clinicians at the front line to be able to tackle what to do – what’s the best thing to do for the patient that’s right in front of them. And I’ll tell you, there’s one other piece that hasn’t really come to fruition yet, which is there’s a new kind of caregiver that is making themselves more obvious as time goes on. Who is a doctor that you consult for – does this AI or machine learning tool apply to my patient? What is the data that’s in the system that is going to help me manage this patient? What is the right digital health tool that could send this patient home with, just like a prescription for a drug, that’s a prescription for a digital health setup for them. That clinician at the bedside doesn’t exist yet, but you can imagine, just like you’d consult an ID doctor, or a cardiologist, you may want to also consult a digital health doctor that is able to come at the bedside and provide that kind of insight and support the same way they would consult for anything else. So I think those are other directions that we’re going to explore, because all of this technology has the potential to just fragment the clinical workflow and fragment the patient experience. We need people that are going to be able to tie this together, stitch the different technologies and try to de-fragment this with the right set of platforms, and also provide consultation even at the bedside for these kinds of emerging technologies.
Kathy: That all sounds really interesting. Thank you, Alistair, this was a great conversation.
Alistair: Thank you very much.
Kathy: Thank you for joining me for this episode of the Smarter Healthcare Podcast.
If you’d like to learn more about Alistair, you can follow him on Twitter @transformatics.
You can follow me on Twitter @ksucich or @smarthcpodcast. Feel free to get in touch with comments or guest suggestions.
To listen to more episodes, visit our website at www.smarthcpodcast.com or find us on your favorite podcast app. I’d appreciate if you would subscribe, rate, and review.
Thanks for listening!
