S01E04 & S01E05
A conversation with Dr. John Hughes
Masha (Maryia) Samuel, Renée-Claude Bider, John B. Hughes, Meryem K. Talbo, Katherine Lan, Neeti Jain, Esther Kang, Predrag Jovanovic, Khiran Arumugam, Dylan Langburt, Susan Joanne Wang for the McGill Journal of Medicine1
Published online: May 27, 2023
The McGill Journal of Medicine (MJM) MedTalks podcast aims to share knowledge and advice with trainees in medicine and the health sciences through interviews with members of the medical community at McGill University and beyond on their careers, research, advocacy, and more. In this episode, Masha (Maryia) Samuel, MJM podcast team member and MSc student in Experimental Medicine interviews Dr. John Hughes, family physician and Assistant Professor at the McGill University Faculty of Medicine. In the first part of the interview, they discuss Dr. Hughes’ early training, his work on an Advanced Crew Medical System, and his involvement in space medicine. In the second part of the interview, they discuss the development of an electronic health record and Dr. Hughes' vision for the future of patient-doctor medical encounters. The episode is rounded off by Dr. Hughes’ advice for medical trainees and junior researchers. The show notes include a glossary of terms, links to publications, images, and videos referenced in the episode, and a transcript of Dr. Hughes and Masha Samuel’s conversation.
1:49 Dr. Hughes’ journey from engineering to medicine;
3:22 Why we desire to and how we justify sending humans to space: the intrinsic human desire to discover, the role of the space race, and how space research advances technology on Earth;
12:32 How the Advanced Crew Medical System, as envisioned by Dr. Hughes would work: information architecture and versatility, biomarker tracking, and predictive algorithms;
20:24 Biomedical sensors and how astronauts would use them;
25:15 The limits of the current telemedicine for astronauts compared to the proposed Advanced Crew Medical Systems and how predictive algorithms and biomarker monitoring are currently being used in medicine;
0:14 How the electronic health record intersects with Dr. Hughes’ work in space medicine;
9:17 Discussing the use of electronic paper and how it limits the power of the electronic health record;
13:16 Why the electronic health record has not progressed since the year 2000 and the importance of involving professional associations in decision-making;
28:49 Dr. Hughes hopes for the future of the electronic health record: a standardized system able to be queried and to diagnose;
30:00 The next steps in establishing an electronic health record, the role of the young scientist, and career advice from Dr. Hughes;
37:11 How to stay up to date with space medicine and helpful resources for those looking to learn more;
Advanced Crew Medical System: An autonomous and artificially intelligent system using biomarker tracking and predictive algorithms to provide medical support for the NASA Mars mission
Canada Health Infoway: An independent, not-for-profit, government-funded organization mandated to increase access to digital medical information. https://www.infoway-inforoute.ca
McMaster Oscar-EMR: A digital knowledge-sharing network established by Dr. David Chan and McMaster University’s Department of Family Medicine connecting McMaster University, Queen’s University, University of British Columbia, McGill University, and independent clinicians. https://fammed.mcmaster.ca/oscar-emr/
OECD countries: member countries of the Organization for Economic Co-operation and Development
Links and papers
- St. Mary’s Hospital, Advanced Crew Medical Systems Informational Video: https://www.youtube.com/watch?v=U7Csa-zBD7U
- Earthrise at Christmas image: https://www.nasa.gov/multimedia/imagegallery/image_feature_102.html
- Mass General Hospital Computer Lab DXplain Information: http://www.mghlcs.org/projects/dxplain
- Canada Space Agency Space Medicine Information page: https://www.asc-csa.gc.ca/eng/astronauts/space-medicine/
- CSA Report on Space Medicine: Canadian Space Agency (2021) Health Beyond - Report of the Advisory Council on Deep-Space Healthcare. © Her Majesty the Queen in Right of Canada, as represented by the Minister of Innovation, Science and Industry, 2021. ISBN: 978-0-660-39647-7
0:05 Masha Samuel (MS): The McGill Journal of Medicine Med Talks is a podcast series where members of the McGill Faculty of Medicine and Health Sciences are interviewed on topics related to career research, advocacy, and more. The aim of Med Talks is to open a space where faculty members can share information and advice for trainees in healthcare and Medical Sciences.
My name is Masha Samuel. I'm a master's student at McGill and in this two-part episode, I will be interviewing Doctor John Hughes, an Assistant Professor of Medicine and Health Sciences at McGill and the Quebec College of Physicians. Dr. Hughes studied Civil Engineering, Architecture, and Medicine at McGill. Today he runs a full-time primary care practice and has conducted research in advanced clinical decision support systems for Canada Health Infoway and the Canadian Space Agency.
In 2014, a team of investigators at Saint Mary's Research Center in Montreal, QC was awarded a contract by the Canadian Space Agency to develop the information systems architecture for an autonomous medicine application called the Advanced Crew Medical System, or ACMS, a completely autonomous and artificially intelligent system to provide medical support for the NASA Mars mission.
In today's episode, we will discuss why we send humans to space, its current limitations in medical assistance, and how an AI decision support system can allow for quicker medical support by constantly comparing an astronaut's physiological states. Here to talk about his experience as part of the research team that led this project is Dr. John Hughes. Well, thank you for joining us today. Dr. Hughes, it's a pleasure to have you.
Dr. John Hughes (JH): Well, thanks, Masha, for the invitation. Looking forward to it.
1:49 MS: Okay. So my first question for you is about your educational background. So your undergraduate degree is in civil engineering and architecture, and you later decided to get your MD. I just wanted to know, why did you decide to make that switch between the disciplines and where do you find that these two degrees might connect the most?
JH: Well, I guess I started off to be an engineer because that was kind of the family tradition. And what I learned in engineering was that it's essentially the process of applying scientific information to mathematical and geometric models to solve problems. While there, because the school of architecture is within the faculty of Engineering at McGill, they build on this principle, but they start to apply that to the human experience in that architecture is all about the built environment that we all inhabit or use in some way or another. So, and it was in architecture that I first got exposed to human psychology and the discipline of cybernetics, which is the human-machine interface. The human being, as you probably know, is the most complex entity in the universe and that we're aware of anyway. And my motivation to pursue this entity will probably become obvious when we get on to the next question, but it essentially has to do with pursuing the unknown.
3:22 MS: So sending humans to space is a huge undertaking that requires billions of dollars. Why do we want to send humans to space? And how do we justify the cost of autonomous space medicine when healthcare on Earth could probably use the same financial attention?
JH: Well, I think to begin with, the pursuit of exploration is hardwired into essentially every living creature, and, being living creatures, it's quite obvious in us, even though we don't need to as much as our ancestors did. But, in the upper-level vertebrates, this drive seems to be mediated by the dopamine motivational circuits and it is essentially one of the fundamental characteristics that allows for survival of the entity. Space, in the words of Gene Roddenberry during the introduction to Star Trek, The Next Generation is indeed the final frontier. And my time down at NASA made me– it was quite obvious that NASA is full of Trekkies who are just fascinated by this pursuit. On a side note, I had the great pleasure of working with William Shatner who played Captain Kirk in the original Star Trek series, and we did the narration for our video for the ACMS project together, which you can YouTube at Saint Mary's ACMS if you want to take a listen.
[…] One of the things that the space agency is very careful about is the transferability of the technologies that they fund development for to terrestrial uses. I think one of the best examples of that is the robotics development that was done many years ago when we were building the space station. As you know the Canadarm was the kind of Canada's contribution to that, and I'll explain a little bit about that. The way that NASA, ESO, which is the European agency, and the CSA Canada work together is that NASA is kind of the central orchestrator of things and in order […] for our astronauts to get rides upstairs, we have to contribute something to the program, and that relates to what I'll talk about a bit later about the Advanced Crew Medical system, the ACMS that I referred to. So, Canada looks for things that it can develop and contribute. But the CSA even with our project was very, very specific that there had to be really actionable usages terrestrially for whatever we developed. So, I think in that regard, they are quite responsible with their use of the money that they are allocated by parliament.
6:33 MS: Has that always been the case, the applicability to Earth or is that recent?
JH: Well, interesting. OK, you got to go back in your history and understand why we walked on the moon or why the Americans walked on the moon. That was the result of a race with the Soviet Union. The Soviet Union, following the end of the Second World War, was the recipient of some of the great rocket engine geniuses that fled Germany and really got the large boost vehicles off the ground literally before the Americans, which produced the nuclear threat that led to the Cold War, etc., etc. So back in the late 50s, early 60s, the Russians were out front in what we call […] the Space Race and Sputnik was the first human artifact to orbit the Earth. And this caused great consternation amongst the Western allies and led to John Kennedy's famous statements about going to the moon and led to the financing of that project which was, in the history of human enterprise, really quite spectacular. When I talk about this, the computer that put Neil Armstrong about 400 feet above the surface of the moon I think ran at 64 K, you know a couple of kilobytes of memory […].
Anyway, so it was originally a political Cold War-type enterprise and then once it became clear that the Americans had the lead and that the Soviet Union no longer had the financial resources to pursue it, that's when it started to translate into “what does this exploration activity produce for us?” I think it was clear from the original activities that being able to boost objects into space […] reliably and some of the first communication satellites that that that opened up transmission between various parts of the world and then moved into some of the terrestrial observation satellites. Canada was a big player in that one where we developed the Landsat series of satellites that started to help us analyze the Earth and understand the Earth from a – no pun intended – a global point of view. I was going to mention later on that, you know, when Bill Anders took that shot of Earthrise on Christmas Eve everybody was stunned. He was stunned. He didn't even have his camera loaded. And that's accredited, well, at least in part, with kickstarting the ecology movement here on Earth. Because what is extremely apparent – I hope everybody's seen that picture – but what is extremely apparent is that there's a vast 0-degree Kelvin environment that we live in on this little lifeboat called Earth and we seem bound to destroy it. And I think, confronting the constraints, stripping away all that Mother Earth gives us that we don't even appreciate, when we go up there in space, we don't have any of that. We have to think about it, figure out how to solve the problem, make it work, make it work reliably, and make sure that the people that are brave enough to do that come back alive.
So, there's a tremendous amount of insight. I mean, if you read about the work that's being done just from the point of view of nutrition for the Mars project, it's amazing. There'll be no animal protein on Mars except ours, us, but certainly […] we're not going to eat any animal protein on Mars. So, there's this whole research direction in vegetarian menus and nutrition and what can we grow, is it you know sufficient to meet our nutritional needs, etc., etc. So, I think that the space environment is a very, very constrained lab bench that does not allow you to take for granted that which we have here on Earth and […] from gravity to galactic cosmic radiation protection to our own biome. So […] we learn and the more we learn, I think probably that's our main modus operandi for what last 10,000 years.
12:26 MS: So, we know that long-term exposure to microgravity can cause deterioration of multiple body systems, hence the importance of medical monitoring. Can you tell us more about how the advanced crew medical system works? What kind of information do these biometric sensors, that each astronaut has, pick up information, biological information and how do they assist in providing healthcare to astronauts?
JH: So. then there's some interesting background behind that. I'll start with the story of the first engineer that […] I hired. He just graduated from Concordia University and I told him that […] I want to build an electronic medical record and he said “Okay Dr. Hughes […] first thing you got to do is tell me exactly what you want to do” and I said, “Well I can't because what I do today will probably be different tomorrow.” So, we started off with this premise that whatever we built or whatever we designed had to be expansible and changeable. So now we jump to [a] year or two after we finished the project for the space agency, we were contacted by some people down in Houston who told us this story about how NASA had been blindsided by some retinal problems that the astronauts were developing that they figured were related to weightlessness and none of the biometrics that they've been collecting, or any of the medical records they've been collecting, had given them a heads up as to the possibility of this happening. So, I was invited to go down and present at this conference […], it was actually an ontology conference to see […] what could be done to mitigate this issue of having things happen that we didn't expect to happen which kind of fit with what my philosophy of medical record development was; that we have to be able to cope with that which we don't know about yet.
So, in terms of what I call the information architecture behind the autonomous medical system that is the ACMS, the standards that we identified as allowing us to accommodate things that were unexpected had a lot to do with the way […] clinical entities were described. In other words, if you have an inconsistent way of describing A or B, then you never get a chance to aggregate A or B together and start to do analytics [on] the data. So, that […] was part of the reason for using information models in the data storage for the metrics that are collected. And the data storage, if you're using standardized information models, is totally agnostic to the hardware that's used, the database application that's used, and it's essentially future-proof. In other words, you get a data set that is usable today and usable years from today. The second standard that we identified as being critical to this ability was the ontalize– I'm going to invent the word here – the ontalization of domain knowledge. This is how the Gene Ontology is organized: the gene Ontology takes all of the genomic information that we have about all species and puts it together in a format and structure that allows you to easily move from species to species, from protein to protein, and figure out where the similarities are and where the differences are. So, it is with this kind of information technology that you can create data sets that are extremely useful in virtually any circumstance.
So, you then say – okay, […] we organized or selected a set of […] all the obvious vital signs we wanted to collect with our little wearable sensors[…]: pulse rate, temperature, O2 saturation, heart rate, electrocardiograph electroencephalography, and we moved on to things like […] eye movement velocities and even keystroke velocities because part of the project had to do with “is Bob or Jill fit to go outside the spacecraft and fix the antenna that's broken”. So, they wanted to know are people fit for [a] task, [which] was part of the requirement we had to satisfy. So that led to the notion [that] we take all of these biometrics that we can monitor now and we build a structure where if we find three more we could monitor tomorrow, we can add them in in a standardized way. And then, we're going to monitor the astronauts for three or four or six months prior to going up. And then, when they're up there, we're going to continue that monitoring and on a real-time basis compare those metrics to the metrics we'd already collected on earth to look for deviations. And this is what's very interesting about some of the AI applications is that if you have a data set that is interoperable, in other words, it is structured in the way I described where the […] apples are apples and oranges are oranges. Then you can start to create algorithms that crawl through the data sets looking for the digital signatures of what could be.
So, you have the, you have the data set, and then you have the outcomes. So, let's say, you know, Bob was having this set of parameters coming in, and then boom 9:00 o'clock that night he got a migraine headache. So, your software would be able to go through the antecedent data set and say “Okay, Was there a digital signature in here that would have predicted his migraine?” This is already being used in NICUs to predict septic shock in newborns, where they can take the data sets from – actually, they're literally video data sets of the flushing color – you can count the heartbeat of the baby from the video just by the flushing of the skin – and they analyze the frequency, the temperature, color temperature and our can successfully predict if a baby's going to go into septic shock. So, this is the kind of ability to predict decompensation, the assessment of competency for [a] task. And then the final part of the trilogy was to be able to diagnose something once it went wrong and figure out what to do with it.
20:20 MS: Interesting. I have a question for you about sort of how invasive are these sensors. What does this actually look like?
JH: […] What we designed is probably a bit of an antique now but we designed a set of biometric sensors that were adhesive and applied to the skin and they had an internal battery that had a life of about a week. So, what the protocol involved – this is kind of a sad thing for me – what the protocol involved was that they would have these somewhere between 7 and 9 sensors that they would apply to their body. They were micromechanical devices that our engineers developed. They were about the size of a dime and had a sticky back on them and they would just peel it off and stick them on various parts of their body and that was it. That's all they had to do. The rest was automated. Two points there, one point is that I think that with our computational technology and the power of our computing now, we should not be asking human beings to do the heavy lifting of, you know, routine activities that the computer could be doing for us. That's particularly relevant when I get on later to the electronic medical record. The second part of this – it's a little sad for me – was that when the contract – I guess you all know David St. Jacques, he was one of our residents at the time at Saint Mary's Hospital, and when he was scheduled to go up to the space station, the CSA put out the contract for a biometric sensor suite and some software to go with it and we actually won the contract to do that which was kind of made me very happy and but unfortunately Public Works Canada found a clause in the contract from our aerospace partner they didn't like and they canceled the whole thing, so it's a bittersweet thing for me.
22:44 MS: How does the current medical assistance provided in space compare to the autonomous one that you were trying to then implement?
JH: So the current model for taking care of the astronauts is telemedicine. […] Ping times, that's the time it takes to, you know when you check the speed on your computer, that's a ping time. So, in orbit-ping times, Lunar ping times are pretty good. They're short enough that you can have a normal conversation, but when you start to get out to the asteroid belt or to Mars, you're talking about delays that would be unacceptable in an urgent situation, and you're talking about a significant amount of time when Mars is actually in the shadow of the sun and there is no communication at all. So, theoretically, every crew would have a trained doctor on board, or at least a trained medical technician on board, and the assumption [at NASA] was that that person would [somehow] be not available. So, what would we do if that happened? Because they were going to be out there for three years. So that was the main driver behind the autonomous medicine concept for deep space travel. In Canada, it was Mark Garneau, who was Canada's first astronaut to fly, who really promoted this notion within the Canadian Space Agency as […] having potential for [a] Canadian contribution to this supporting of space exploration that I explained […] earlier that […] was kind of our ticket for getting some of our astronauts to be able to get on board with the projects.
MS: And you were talking about telemedicine, which I think is a great point. So, it's just far too slow–
JH: well and unavailable potentially. […] So, the final aspect to answer that question would be the biometrics suite that's collected. We were all quite surprised to find out how little or how few biometrics were collected on the astronauts and how intermittently they were collected. So the wearable sensor part of the project which you know fit right in with the development of the, I guess they call it the quantified self-movement with everybody wearing 2 Fitbits, […] it fit right in and really has a tremendous future in what we call medical hovering, which is patients or individuals with various disorders being able to wear biometric sensors that would again accumulate data sets in the way that I described earlier and allow us to look through them and predict what's going to happen. There's a certain amount of this already being done in England where they take limited data sets for people being discharged from hospital and predict who's going to need heavier levels of care at home to avoid readmission and that sort of thing.
26:24 MS: Thank you to Doctor John Hughes and our audience for joining us on the first part of this two-part episode on AI and space medicine. If you enjoyed this episode, tune back in for Part 2 where we'll explore the roots of this project and some of Doctor Hughes' aspirations for the future of medicine.
This podcast was edited and produced by the MJM Podcast team. Feel free to reach out to us on Twitter or Instagram at mcgilljmed or by email. We would love to have your feedback. See you next time.
0:06 MS: The McGill Journal of Medicine Med Talks is a podcast series where members of the McGill Faculty of Medicine and Health Sciences are interviewed on topics related to career research, advocacy, and more. The aim of Med Talks is to open a space where faculty members can share information and advice for trainees in healthcare and Medical Sciences. My name is Marsha Samuel. I'm a master’s student at McGill, and in this episode, we are continuing our discussion with Dr. John Hughes regarding his experience developing the Advanced Crew Medical System, or ACMS, a completely autonomous and artificially intelligent system to provide medical support for the NASA Mars mission. Today we will discuss the unexpected origins of this project, the difference between electronic medical records and electronic paper, and how the progression towards a universalized electronic medical record system can go a long way in the delivery of medicine on Earth and thus space. Welcome back, Dr. Hughes.
0:14 MS: Aside from space medicine, your research places emphasis on understanding electronic medical records, or EMR as you said, as well as health information systems. Where does this research meet advances in autonomous space medicine and why is EMR important for you to research?
JH: Right, Okay. So, that's a really great question because it was indeed the origin of our involvement with the space agency. I’m trying to remember, it's a while back now but […] in our teaching environment, we were being mandated to start to teach our residents how to use electronic medical records. And what was eminently apparent was that the electronic medical records that were available were really glorified [electronic] paper and really didn't live up to the potential that had been identified for [a] great many years. So, with the support of the foundation of Saint Mary's Hospital and the CEO Arvind Joshi who was at the hospital at that time, we received funding to start developing an electronic medical record that would meet the teaching mandates that we had [that] included not only teaching but research and obviously clinical care.
Now, if you look at software development and the cost of software development, thinking that […] a minimal EMR has anywhere from three to 10 million lines of code, and each line of code, depending on how well it's documented, can cost anywhere from $3 to $10, you're talking about a lot of money and you need a substantive market to be able to justify the business model for a company to get involved in that. What we realized was that the market for something that would accommodate the requirements of teaching, research, and care was just too small in Canada to be able to expect a company to come along and say “Okay, we're going to develop that”. The guys out of McMaster had already developed Oscar as kind of a starting point, but it really didn't satisfy what I saw as being the potential for a platform that would incorporate some of the advanced computational capabilities that we're now calling AI.
So we started down this road and decided we were looking at the [...] Gartner Corporation is one of the big consulting firms in this area and they had produced this kind of road map for electronic records that showed that okay you go from step 1 to step 2 to step 3 to Step 4 to Step 5. And we looked at this and we said, “Oh God, okay, why don't we do what they did in Africa with telephones. We'll just skip the wires and the switches and we'll go straight to cell phones”. So, we decided we were going to go straight to Level 5. Level 5 of Gartner was essentially the autonomous medicine application. It was an advanced application that was capable of offering both […] diagnostic and clinical advice at the point of care. And so that’s [what] we started working on that and that's when the CSA came by and said “Hey you guys you want to bid on this contract”. So, we bid on the contract and got it. Our intention was not to do space research, our intention was to develop advanced clinical support applications in large part because – I'm showing my age here – we've been in a healthcare crisis since the 1960s and you know we've got another one hitting the front pages and the media right now and the federal government and the ministers are all, you know, gnashing their teeth and saying you know we need 27 billion and the feds are saying well you're only going to get 4 billion. But in all honesty, the current predicament we're in started seriously back in ‘96 when at the behest of the IMF Canada closed 20% of its acute care beds […] and bought out all the older doctors and older nurses and cut down the size of the medical school classes in an attempt to save money, and we're still paying the price for all of that today.
JH: Yeah. Then okay. So we got a crisis. We look at the data, look at the data, look at the percentage of GDP that healthcare consumes for all the OECD countries and all [we] see is this endless ascent. Our society cannot afford to continue to provide healthcare in the way we have traditionally done it, and this was identified by the federal government in 1999 in the report of the advisory committee on the Health Infoway, which was the founding document for Canada Health Infoway, which allocated billions of dollars to [provide], quote-unquote, “an interoperable electronic record for every Canadian”. And the reason that that was going to be done was because Canadians, quote-unquote “in 1999, after the closures in 96, were becoming worried that the healthcare they would need won't be there”. Now, I have a minimum of two to three people a day in my office [expressing] that sentiment to me, so this is not new, and it's extremely important that we have to start to look at new models for delivering healthcare. I'm currently involved in a nurse practitioner model development where we're going to use advanced decision support software to support the practice of both nurse LPNs and nurse practitioners.
We have to look to our technologies to provide greater and greater efficiency and effectiveness in our system. Right now, we have an electronic medical record environment in Canada that according to the Canadian Medical Protective Association, which is our, how shall I say, group insurance for doctors, is the number one cause of burnout amongst doctors. It doesn't sound like a very good solution when that's your number one cause of burnout. There's so much more that could be done. This [is part] of the reason I was willing to talk to you; I think we have to get the message out there to people to say, “Look, you know, we can't afford to do what we keep doing […]”. It's unsustainable. We need to look at new ways of doing it and we need to look at how our new technologies can help us do that in a very pragmatic way.
9:17 MS: So we'll chat about Health Infoway in sort of the next question. But I wanted to just go back to something you said electronic paper is the term that you used. Can you sort of tell the listener […] why is that useless for us? What would be more useful? Maybe we had chatted about – what was the term – health informatics, the language of […] health informatics. Why is the electronic paper useless for us in what you're trying to do with the autonomous space model?
JH: Well, I don't know of anybody who can index, well, I guess you can parse a PDF but you cannot compute upon a PDF. So, if all you're doing is storing PDFs. That's not really of much value in terms of being able to write a search. You know what we need to be able to do is write a short couple of lines of code that answers the question. We did a study for Canada Health Infoway [where] they wanted to know our Canadian electron– Actually before I go down [that road], let's just clarify something about electronic medical records. The original term was “computer-based patient record”. The term “electronic medical record” has become meaningless. It is used by people as a placeholder for a wish that they have that they wish this thing would come along and solve the problems. It has not and will not in its current format. Let's be absolutely clear about that. The computer-based patient record as foreseen in the 1980s and 1990s, was a record in which the data was stored in a way that you could search through it. You could write a couple of lines of code and find this, that, or the other thing.
So [Infoway] I asked us “OK, you guys, do a study. We want to see if the electronic records out there in Canada are capable of addressing population health management”. So, we had a set of questions like “Give us a list of all your patients over 65 who have not had a pneumococcal vaccination” and […] we had study sites from coast to coast […]. There were a couple of electronic records that did very well, but they […] happened to be in clinics where the owner or the main guy or girl –I don't know whether it was one or the other– was a computer geek and was actually keeping a parallel database to the supplier of the electronic record so that they could write their own queries on their database because they weren't allowed to touch the database of the electronic record provider. And there were other sites where they were virtually unable to answer the question. They had to submit a proposal to the record provider and the record provider had to figure out how much they were going to charge them to do it and then write the code and blah blah blah. And we studied one paper clinic and some of the worst electronic record clinics were not much better than the paper clinic. So, the thing – actually we're going to touch on this a bit when […] we're going to talk a bit about the record of the future. So there's– I’m [going to] paint you the future from hell and then the future from my fantasy.
13:16 MS: Okay, well, maybe to tangent off of that, in 2000 we had talked about how Canada Health Infoway promised each Canadian an interoperable electronic medical record and so that was sort of the breakthrough more or less in 2000 with EMR, but [we] want more and we know that we need more than just the electronic paper. Has EMR progressed since the year 2000? And if not, sort of what's the biggest thing holding us back from progressing?
JH: The short answer there is no, it has not. Actually, the record has not progressed since 1969. 1969 was when the first efforts […] hit the deck and were massively constrained by limited computational power and limited database software. But, the principles were established of what was needed. In the interim, obviously, the computing horsepower and the database architectures developed and the standards actually that are necessary developed. And one of the key lessons I learned was taught to me by the Chair of Medical Informatics at the University of Buffalo after a very exciting meeting I attended in Brazil. [We] had this idea to create an online sandbox for students who could take all these Lego parts and put them together and create their own electronic record and all kinds of stuff. And [we] walked out of the meeting and I said
“Gee, that was really cool meeting,”
and he said, “Yeah, it's very cool, but it'll never work.”
“Oh, okay here, you seem to be quite into it while you were in there.”
And he said, “No, the problem is, is that it's the CFO who decides which application is purchased”.
That's what we're seeing in Quebec and Canada right now; it's actually the Treasury Board in the government of Quebec that […] has the final word on what applications are developed and used and which ones are not. So, the people that write the checks don't understand, and the people that understand don't write the checks. Now it's understandable from my point of view because you're talking about investments of billions of dollars. And the guy who's going to sign that check, you know, doesn't want to get hung out to drive because he made a mistake. But I think that the only way to address that, and we tried to do this number of years back, is to get the professional associations involved, the ones who are responsible for setting the standards of education and care for healthcare in Canada and say, “Okay, we have people within our associations that understand this and here's the recommendation, and this is the reason why”. But to date, we've been unable to do that despite trying. And you know, this is a bit sad, but if I were to sit here and tell you the true data about our morbidity and mortality related to medical errors. People would shut down, they don't want to hear it because it is terrible and it's like, “Oh no, no, no, don't tell me that we're going to, we're going to give you a bit $4 billion and you're going to make it all work.” But nobody ever sits down and figures out how to make it work. That's my beef with why we still have electric paper. We could do better, but we haven't.
17:27 MS: If we do look into the future, in about 10 to 20 years in the future, maybe 30, where do you see EMR progressing? Are you hopeful? Are you pessimistic? Are you realistic about it? What do you think?
JH: Well, I'm hopeful because I'm talking to you and I will I take my – how shall I say – my inspiration from a guy named Larry Weed. […] Every medical student in the world is taught what Larry Weed developed, which is the SOAP Note, the subjective-objective assessment and plan. He developed that at the University of Vermont –he was a family doc by the way– in the 1960s as being a requirement for the problem-oriented medical information system which was one of the first electronic records to function in a clinical setting. He ran an entire floor in the University of Vermont Health Center in Burlington on an IBM360 mainframe. Google an IBM360 mainframe, it's about the size of a large bathroom and it has less horsepower than my iPhone but it was a good proof of concept that it could work and could provide the requirements he had specified and those requirements still aren’t being met today by our current electronic records. So, if we go forward, I'm an optimist because of him because. He worked until into his 80s trying to promote this notion. He was actually one of the consultants on our project. I had been a fan of his for years and years and he was kind enough to give us some insights into what are called “problem knowledge couplers”, which are part of the advanced decision support system.
So if I take off my optimist hat, I say “Well, It's been a half-century that we've been using the same level of software function. It'll be another half century and we'll still be using the same level of software function unless we do something about it”. Now if I put on my optimist hat and I say, “Oh, OK, let's say that we do get the professional associations to buy into this need, and let's say we do get the people that sign the checks to have the courage to start to at least fund the proof of concepts and the beta sites for this kind of application”. Then, you have a scenario that is very different. You have a scenario where, you know, people are now coming into my office saying, “well I, you know, I have this pain in my left eye. And I went on ‘Doctor Google’ and he told me this, that, and the other thing” I [say] “Okay, that's interesting,” but we would actually have […] the clinical encounter start long before the patient even saw the doctor. Because the patient would be able to go into a patient computer interview application and start to document all of the symptoms that they were having and document all of their past medical issues that may or may not be existent in the medical record that, then, could all be subjected to decision support analysis – and I'll go into that in a second.
So, then the patient comes in to see the doctor and the doctor sits down and has this entire report generated in front of them as the starting point to the clinical encounter. The clinical encounter would have the – how shall I say that – the clinical encounter interface that the doctor would have with the computer would be actually a customized interface. And then, this is what is done in the domain called Human Factors Engineering where you design the interface with the electronic system, the computer system, specifically for the human user, in the context they're in, this is what's done in the airline industry. [...] When a pilot looks at their computer screen, you know when they're coming in in the fog, a heavy crosswind. They are only being fed the information they need to know to be able to do what they need to do at that point in time and all of the information they need to know at that point in time. So, the interface with the doctor would be structured like that in accordance with his preferences. For example, I'm a very graphic person. I like, […] charts and that kind of stuff. Other people are very narrative kind of people. So, I'd have my interface set up so that all the information would be graphed for me and, you know, timelines and yada, yada, yada.
So, then you would take the information from the patient, take the history, do the exam, and be able to enter it using the minimum number of keystrokes, and where you needed narrative, you could put in [the] narrative. Where you needed very structured data, you put in very structured data. And while this is accumulating, in the background is running the diagnostic software. And […] if you want to go online and check out something called DXplain, D-X-p-l-a-i-n, it was the clinical decision support software developed at the Mass General Hospital computer lab started by Octo Barnett back in the 60s. It's the most powerful application of its type in the world. As you start entering data into it – I mean it's problematic because you have to manually enter data, my system of course would have the data entered into it automatically as it appeared – but as you enter data, it’s like you put in male: 50,000 diagnoses. 40 years old: 30,000 diagnoses. Caucasian: 25,000 diagnoses. Headache: 10,000 diagnoses. So, as you enter findings from the clinical encounter or the machine is obviously doing this for you in advance, the software is coming up with its synopsis of the medical knowledge base as correlated to the findings. And this is what we developed back in the 70s […] called the Problem Knowledge Coupler where you had your knowledge base and you had your findings from the clinical encounter and it's a relatively simple process of putting them together if you can automate the process.
[…] This is where the lethality algorithm in our project came in. So, we had this list of what we call a differential diagnosis to explain the presenting problem of the patient and with our lethality algorithm, we said “Okay, well, possibility #4 only has a 4% probability of being the right diagnosis, but if we don't do something about it, the guy's going to be dead in 10 minutes”. So that kind of overrode the fact that while it's most likely just a cold and you know we give him some nasal spray and he'll be fine. So there, you get that. Then, that kind of secondary refinement to the diagnosis where the lethality of the potential diagnosis is taken into account. This is the same sort of operating system that battleships used to decide what to shoot at. You know, when there [are] 10,000 incoming threats to a battleship, they have software that decides which one to shoot at. So, it's the same sort of idea. Again, this is just an example of how we took a lot of, how shall we say, knowledge and examples from other domains and sought to apply them to healthcare. There [are] publications on that for those who are interested.
So then we decide, OK, and then the same type of algorithms have gone through all the therapeutics. Now this is where it gets really interesting if you have a large data set that's interoperable. So now I've decided, “OK, Bob is diabetic and these are all his characteristics, his age, his genotype, his phenotype, and everything else. And now I want to know what to give him”. So now I say, “OK. Find me the best therapy.” Well, it's already gone and done that because it knows that I'm going to ask the question. So, it crawls through the database of a couple 100,000 people who meet his criteria: male, 40, blah, blah, blah, blah, blah and looks at what was done and what the outcomes [are] and said “okay, you know, 80% of the people with these characteristics that had this done for them ended up doing this”. Well, I mean, this is stunning. If you could imagine that and this has been dreamed about for over a decade. I first read about that sort of stuff over a decade ago, and it's all very doable with what we know how to do now. We don't need to invent anything new to do this.
That's the – yeah, okay let me just say something backtracked into the space agency. We talked about, […] working for them, [and] one of the really interesting things was what's called a technology readiness level. Now there's a scale from zero to– I think it goes up to seven or eight, where a technology readiness level of 0 is “, that's a really neat idea that I dreamt about last night”, to an 8 where “okay we can buy 12 off the shelf tomorrow and ship them out right away”. And we had to function within a readiness level somewhere between 5 and 6. So we didn't have to be able to say “Well you can go out and buy it tomorrow,” but we had to be able to say that what we are proposing to do is already done over here and we're going to bring it from over there and put it in here and make it work. So that was a very – how shall I say – good lesson to learn that if you can reuse existing technology and fit it into the application you have, you really can advance quite rapidly.
28:49 MS: […] Whenever you talk about this, I always remember that image that you flashed in the very first lecture I saw you do. And it's that image of these little people pulling a wagon with square wheels.
JH: I love that.
MS: And someone suggesting “Hey do you want to do a circular one, a round one” And they're like “No, we don't have time to change wheels, we’re doing our work here.”
JH: […] That's […] in healthcare, that's the brush fire type situation we're in. You know one of the problems is getting clinicians into a meeting. I mean they've got the beepers […] going off and the people are crashing in the emergency room and you want me to go to a meeting to talk about you know, so it's a very, very real problem and an understandable problem. But regardless, it should not prevent us from chasing that dream that I have where I can give my patient the best treatment possible relatively easily.
30:00 MS: If we are talking about the future, I want to ask you what are some of the next steps in this project, in sort of the EMR project? And more specifically, let's talk about maybe the young scientists. The coming generations of scientists, what can they do to advocate for the development of EMR and therefore the advanced crew medical system?
JH: So I think the first thing to do is to drop the term EMR. I think, as I said earlier, I think it obfuscates a bit the real situation. I think what we have is a medical record and that medical record can live in any number of containers. What's important is the record and its content and structure. There has to be a recognition of what doctors [and] clinicians do. […] I have to include nursing and […] it’s to a lesser extent applicable for physios and resp. techs and others because they actually do touch people and do things to them. But most doctors other than surgeons in the operating room don't realize what they're doing. Now that's a strange statement, right? But if you think about that scenario I just ran through my dream scenario of [a] patient comes in, I collect all the history, the physical exam, all the findings, I stuff it in my brain and correlate it to my meager knowledge of the medical literature, which is you know, in this day and age, it's impossible to know it all. And then I make a decision and I write an order to do something. So, what is the overarching thing that each one of those steps represents? Each one of those steps is part is doing something. And that “something” has a commonality between all those steps, and the commonality is that I'm dealing with information. So, I'm collecting information, I'm storing information, I'm processing information, I'm retrieving information. And that is the definition of informatics, the science of information. Now, that science has a set of formalisms associated with it that are extremely well-developed and are the basis behind most of what I've been talking about here. But it is not understood or broadly understood as being important in clinical medicine.
So, for the young researcher going out there, if you're interested in this domain, lesson number one would be to understand what informatics is and that everything – except maybe, well, even cutting out the gallbladder has some informatics involved in it – everything we do in in healthcare involves informatics because we're processing information, we're dealing with information. So then the second thing I'd say to young […] students, people who are interested, young clinicians even. Make sure that you find your passion first. You got it. If you're not out there following your passion, take a trip to Spain or something and think about it okay because it's– to be able to go through the years and retain your interest, you've got to be doing something that you're really interested in. […] You take as long as you need to find that and if you find the wrong one, don't be afraid to change.
In the world of research, there [are] many ways to do it and you're essentially expressing some of that [...] – So, I guess [it] was the second question I answered that – intrinsic need we have to go out there and discover things, you're satisfying that need within yourself and you can do it in different ways. You don't need to go down the mainstream academic researcher who's essentially you know living hand to mouth with [money] from CIHR grants. You know if you like doing that, great, that's cool but make sure you like doing that […]. The other end of the spectrum, I think it's more like what I did, which was I'm very ADD, so I change my interests a lot. But I found that the informatics interest was, yes, persisted over decades for me, so, I know it's my passion. But you can – you know, some of the greatest researchers in history […] did it in their spare time. You don't have to be a professional researcher to do research and enjoy it and fulfill yourself. So, I for years, funded my own research; I funded my meetings, I hired engineers. I was fortunate enough to have a profession that allowed me to do that and it was only later on that the people came to us and said, “Oh Gee, would you do this for us? And here we'll give you money to do it.” Okay, that's great.
So, there's a number of different roads and if you end up being a clinician, a lot of clinicians– the classic trilogy in in academic clinicians is […] you have your clinical care, you have your research interest and you have your teaching. So that's the trilogy that you work from and you balance them in accordance with what the demands are for your particular circumstance. […] If you're the only plastic surgeon for 500 miles in a particular area of the province, then there's going to be a lot of constraints on your ability to […] go into the wet lab and fool around with how […] to do new sutures or something. So, you kind of customize it to what your lifestyle [and] work style allows you to do, but as long as you're following your passion, I think you'll be happy. And if you're not happy in life, well then take another trip to Spain.
37:11 MS: How can those interested in the progression of space medicine stay interested? Basically, how do we know what's going on? Where do we find that information? How do we stay up to date? How do we get involved in space med?
JH: So you've got to familiarize yourself with the last two CSA publications related to space medicine which – I'm trying to remember the title – the last one was called Health Beyond. So, if you Google “CSA Health Beyond” you'll see the document, the last one they published, I think it was the end of the end of 2021. That's the last one I saw, there might be a new one since then. You'll see a number of names associated with it: Robert Thirsk, David David Saint-Jacques, [Dave] Williams, you know, the Canadian astronauts. Then you do a similar– then you troll the CSA website, try and get on their listservs for announcements about funding opportunities. Troll the space medicine literature, find out who's researching what [and] where. There [are] a couple of places out West where there [are] some guys that are more dedicated to it than I am that are running projects all the time. [The] same thing is true in the States. You do the same thing with the NASA website. There's a ton of information out there related to what's going on with healthcare and space, and you look at the publications of who's doing what and see something interesting. You send them an email and say, “Hey, I read your stuff, I'm interested, you know, willing to talk” and you'll get a lot of dead ends, but you [have to] go out there and look for it.
39:05 MS: Thank you so much for joining us today, Dr. Hughes, it's been a pleasure to hear your advice and about your research. So thank you.
JH: Well, thank you for the invitation and I have greatly appreciated your insightful questions. They've allowed me to reflect on many years of accumulating various types of experiences in this domain.
39:35 MS: Thank you to Doctor John Hughes in our audience for joining us on another episode of the McGill Journal of Medicine Med Talk series. This podcast was edited and produced by the MJM Podcast team. Feel free to reach out to us on Twitter or Instagram at mcgilljmed or by email. We would love to have your feedback and don't forget to join us on our next episode.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.