Our guest this episode was Arjun Puri, CEO & Cofounder of Symbiotic AI. Arjun joined Reuben Hall to discuss the importance of leveraging healthcare information to improve decision-making for both clinicians and patients in cardiology.
“it is a complex and difficult decision to make when the patient is sitting on the cath lab table and you have to make that decision with the information that you have at the table” Arjun Puri on decision making in cardiology.
Find Moving Digital Health on Apple Podcasts and Spotify, and subscribe to the MindSea newsletter to be notified about future episodes.
Read Transcript:
Reuben (00:01)
Welcome to the MindSea podcast series, Moving Digital Health. Our guest today is Arjun Puri, CEO and co-founder of Symbiotic AI, a clinical decision support platform to improve heart care outcomes. Thanks for joining us today, Arjun.
Arjun Puri (00:17)
My pleasure. Thanks for having me.
Reuben (00:20)
Can you start by telling us a bit about your background?
Arjun Puri (00:31)
Yeah, happy to. So my background is in digital health innovation and clinical informatics. I started out my career really working on healthcare information and understanding how we could leverage that healthcare information to benefit clinical care and improve healthcare services for everybody in the world.
Reuben (00:44)
Cool. So what drew you to the digital health field?
Arjun Puri (00:49)
You know, my dad is an engineer. so I still remember stories of my dad coming home with old PCBs and circuits and saying, Hey, Arjun, go and work on these. And that only fueled my desire to learn more about information technology and how we use solutions and not just apps to solve problems. And that’s kind of where my journey began into digital health.
It was a perfect intersection of being able to have a positive impact in the world where the solutions that you create have the potential to impact lives, both patients, providers, and also the healthcare system that we all rely on every day. And that’s where my digital health journey began. fast forward to now, now I’ve had experiences with digital health innovation, with health systems commercializing and building.
digital health products in the medical imaging space and now in the clinical decision support space.
Reuben (01:50)
Cool. And so what was your first exposure into the healthcare field?
Arjun Puri (01:57)
Yeah, so after I finished my undergraduate degree, which was actually in sociology, so I was very interested in understanding, you know, the social aspects of how people operate. Initially, it was in the criminology field. Before, you know, getting into the healthcare space, I was really passionate about pursuing a career in the legal space. But as I started to learn more about clinical information, health informatics,
my attention really started to get piqued by how we use information to help make decisions. And after I finished my undergraduate degree, I actually ended up working with a organization in Ontario that evaluated the competency of healthcare professionals before they could practice in Ontario. So these organizations…
you know, primarily work with internationally educated graduates, whether they’re pharmacists, nurses, doctors, or specialist doctors, and want to ensure that they are ready to practice, you know, in the space that they’re entering into despite being trained elsewhere. And when I learned about, you know, the various challenges that physicians and other clinicians deal with on a daily basis and how similar they were, regardless of where the physician originated from,
that only further piqued my interest into understanding, how are we using the information that’s being collected by physicians? And that’s what led me to pursue a graduate degree in health informatics and clinical informatics. And that, you know, further fueled my interest and my intrigue. And, you know, I’m going to age myself now, but that was, you know, about 12 years ago or so now. So since then, the experiences that I’ve had and I’ve sought have all kind of
know, fueled that and added to that fire to get me to where I am today.
Reuben (03:59)
Interesting, yes. It is a field once you get into it. And your curiosity is peak, there’s infinite ways to take it. So now your latest venture, Symbiotic AI, maybe you tell us how that started.
Arjun Puri (04:17)
Yeah, you know, for a lot of innovators and entrepreneurs, their ventures start off with personal problems. And this was the case for me as well with symbiotic AI. Unfortunately, not too long ago, my mom had a heart attack and, you know, like others who are faced with that difficult challenge, you know, in the diagnosis of coronary artery disease, we tried our best to understand what this meant for my mother.
what it meant for our family, because your life changes. And then across the country, on the other side in the Western part of Canada, a colleague of mine, my co -founder, Dr. Jun Li, was doing research in coronary artery disease. And when we connected and shared some of the challenges that we were having on both fronts, whether that be about, you know, having access to the right information to make decisions or what sorts of
pieces of information would have been helpful for my mother and our family when we were making the decision about how to treat her condition. That’s what really gave rise to symbiotic AI. When we really decided to peer in to the domain of cardiology and have a better understanding of how decisions are being made currently and a realization that a lot of information can be left on the table.
while that decision is being made. And we may not be using the technological advances that we have access to, to their fullest potential. And that’s what gave birth to Symbiotic AI.
Reuben (05:56)
And why do you think, like you said, some of that information is being left on the table and not contributing to the decisions? it just, is it they’re in disparate locations? Is there too much information to consolidate? What’s the challenge there?
Arjun Puri (06:19)
Yeah, you’re hitting on some of the factors there. And I think there’s multiple layers, right? And you start finding this out as you start to talk to more interventional cardiologists, community cardiologists and surgeons. And this isn’t something, you know, specific to the domain of cardiology either. The reality is that our healthcare systems are becoming more strapped for resources year after year. The population growth and the aging population certainly does put
strain on the healthcare system to be able to provide, you know, support. And we do collect more information than we ever have today. You know, Canada has spent billions of dollars into the adoption and uptake of electronic medical record systems. And we’ve been working on this for well over decades now. But the benefits are still, you know, being realized as we speak today. you know, clinicians
have a tough time being able to access disparate sources of information at the point of care when they’re making these complex and challenging decisions. And because the information is located in disparate sources and the analysis of that information is tough in a short period of time, it can be challenging for you to not leave information on the table and take every data point into account.
And I think that’s the benefit of, you know, some of the data science and machine learning foundations that we’ve been able to, you know, advance as a society. know, artificial intelligence was made for this. It was made for us to be able to analyze thousands of data points in a matter of seconds, to be able to provide, you know, risk forecast predictions of what might happen in the future and actionable insights. So, you know, there’s multiple layers to it. But, you know, definitely the strained healthcare system.
along with the spread information of sources are strong contributors to that challenge.
Reuben (08:17)
Okay, and so what stage is Symbiotic in right now? Knowing that it is still a fairly young company. Yeah, tell us about the stage and then where you are in the journey to.
Arjun Puri (08:36)
Yeah, excellent question, Ruben. So if the listeners are familiar with technology readiness levels, I can state it that way and then I’ll give a more broad strokes picture of where we’re at as well. So we are at technology readiness level five, which means that we validated components of our solution in simulated environments. In fact, we’ve actually also validated the accuracy and effectiveness
of our core models and our AI algorithms retrospectively. That’s what gave us the confidence that there’s a potential for impact to patient lives, clinician lives, and healthcare system sustainability. So we’re progressing quickly and passionately towards advancing our readiness levels, but that includes continuous evidence generation, iteration on our user experiences, because
It is a partnership. You know, we hear the term user centered design in product management and in product development all the time. But what it truly means is that you have to move at the pace as your end users do. And, you know, our end users are interventional cardiologists, cardiac surgeons. So moving with them hand in hand, we want to make sure that we are relentlessly focused on their workflow. So as to not interrupt.
that workflow and instead be supportive of that workflow. So we’re continuously iterating on our user experience. And I think, you know, that iteration doesn’t stop ever because you’re always trying to create a better product for your end user. But we are, you know, at the stage where we’re starting to have trials and pilots of Revise AI, which is the name of the clinical decision support system. It’s a play on the word revascularization because that’s the decision that we help.
cardiologists make, how to revascularize a patient who’s had a heart attack and has had a diagnosis of coronary artery disease. Because usually cardiologists will have three treatment options. It’s not always the case, but sometimes they might treat the patient with medical therapy, send them home with medicine because they may think the disease is stable and not advanced to a stage requiring invasive intervention. Sometimes they could
treat that with a percutaneous intervention or more commonly known as the placement of stents to ensure that collapsed and stenosed vessels are able to function as best possible. And then the third option is usually a cabbage surgery or a coronary artery bypass graft or bypass surgery, which indicative from our data can often be better for long -term outcomes for patients as opposed to stents.
But it is a complex and difficult decision to make when the patient is sitting on the cath lab table and you have to make that decision with the information that you have at the table.
Reuben (11:41)
And what does the feedback from cardiologists been so far?
Arjun Puri (11:48)
Yeah, that’s an excellent question. You know, a lot of times entrepreneurs go out there and they create things that they think people want. But they’re not, you know, certain but it all starts with a catalyst, right? Steve Jobs used to say people don’t know what they want until you give them what they want, right? So a lot of that is catalyzed with designs and prototypes that you think would be compelling after discovery of, you know, features and
information that would be important to your end users. So, you know, because we’ve had the privilege of partnering with heart catheterization labs here in Alberta, we’ve been able to work hand in hand with interventional cardiologists to learn about what their information needs are, where in their workflow would they like that information. And that has led to a strong positive response. But you know what, when we started this journey,
and didn’t have a face to put to the product and user experiences to put that product, it can be ethereal, right? And clinicians may not know how they feel about the product, but now we’re at the stage of, you know, where we can actually demonstrate those user experiences, demonstrate the value propositions that we’re able to offer with the capabilities of the product. You know, the focus shifts from awareness of the product to consideration about the product.
and making more intriguing questions come to the forefront, such as, I think this could be helpful here, or this would be particularly helpful in this patient’s case. And it’s that light bulb and aha moment when you understand how it fits into your workflow that really leads to that positive response. So we’ve been privileged to get that positive response. But like I said, the iteration never stops because you’re trying to create something that’s going to
continuously improve that workflow and offer additional benefits. So there’s always more work to do.
Reuben (13:49)
Yeah, there’s a wide spectrum of preconceptions about AI, right? Like people range from being super excited and thinking it’s going to solve all the problems to very skeptical and fear of, this is scary and bad. But like you said, once people start to see how it can practically fit into their day -to -day job and the tasks they need,
done, then it starts to actually settle in. this is what AI is for. This is how it’s useful. It’s not some scary thing, or it’s not magic, but there are specific places where it really helps. And it is exciting to find those applications.
Arjun Puri (14:38)
Absolutely.
Reuben (14:40)
So you mentioned Revise and how it works with clinicians. There’s a patient -facing side to this as well,
Arjun Puri (14:49)
There is, you know, anytime you’re getting a surgery as a patient, you need to provide consent to receive that procedure. Now there’s obviously exceptions to that, you know, a rule that, or that statement that I just made where, you know, you may not be capable of making that decision or providing that consent. But generally speaking, if you have stable angina and stable coronary artery disease,
It is a shared decision between the patient, sometimes the family caregivers, as well as the interventional cardiologist who’s sharing the treatment plan that’s been designed for that particular patient. But without the patient’s consent, things can’t go forward. And it’s important for patients to be able to understand their treatment options. And you can imagine when you are going through symptoms of a heart attack and you’ve presented
you know, in the heart catheterization lab and you’re getting that heart cath and you know, the clinician comes back and says, okay, we found something, you know, this level of blockage and disartery or multivessel disease, it can be pretty overwhelming, right? And because the wait times for surgeries and a repeat procedure are fairly long, know, patients can sometimes feel compelled to make that decision in a very short period of time.
So we’re still learning, to be honest, Ruben, we’re learning and we’re engaging our patients and their family caregivers to understand what would be helpful for you to make your decision and what information would you want to be presented with. You know, the level of information, the detail that we offer, as well as the, you know, I guess, jargon of the nature of the information that we offer.
to clinicians may not work and probably won’t work as well for patients and their caregivers. So we wanna be really intentional with what we develop for our patient partners and for our family caregiver partners. We think a lot of what is being shared with the clinician should be shared with the patient as well, but how they would like it delivered, how much time they need to be able to analyze that information to make that decision.
is still something that we’re learning a lot more about. But it’s, in our case at least, it starts with the clinician because the clinician is the individual who has to consume those insights to understand whether there’s value of those insights and those risk forecasts to be presented to the patient. If it’s a clear cut decision that this patient should absolutely be going forward with a PCI and having some stents placed,
then that discussion may be shorter and you may not need a information package for patients and family caregivers to have to go through. Although the conversation may be fueled and pursued and assisted by other pieces of information. But the intention is for this to have a clinician facing component as well as a patient facing component that helps that shared decision making be more accelerated and more informed so that patients can think about the long -term health
Reuben (17:49)
Mm
Arjun Puri (18:05)
consequences of one treatment option over the other, whether that be the risk of a future heart attack, the risk of death, the risk of affecting their health related quality of life, how many medications they might have to be on for treatment A versus treatment B. A lot of considerations have to be taken into account. And we’re still analyzing that data and waiting to see exactly what our patients want before we go out and build it.
So there’s a lot of consultation happening with our patient partners. So I’ll do a quick plug. If there’s any listeners who are, you know, experienced with coronary artery disease as, you know, family caregivers or patients themselves, we’d love to hear from you and learn more about what would have been helpful in your coronary artery disease journey or what could still be helpful to you in your coronary artery disease journey.
Reuben (18:58)
Well, that’s excellent. I love the user -centered approach you’re taking to really bring the patients and clinicians into the process to really, really understand their needs. Now, maybe you could talk about how the AI insights are surfaced in the tool. Do you qualify or explain how symbiotic
came to that recommendation or insight and what the different factors are. I know there’s some problems with like black box AI that you get this insight, but a clinician has no idea why you’re getting that insight. then it can be mistrusted and kind of thrown out. like, that doesn’t make any sense to me. I’m just gonna go with my own knowledge and training here.
Arjun Puri (19:51)
Yeah, that’s an excellent question, Ruben. And I think you and I could probably talk about this for hours. But I’ll try to keep this succinct and say that you’re right. You’re right.
that needs to be built with AI solutions is certainly a challenge. And it’s not just a challenge in medicine, it’s challenge everywhere. And it’s appropriate for all of us to ask this question of what is the logic model and how is an algorithm or a solution preparing the information that it is? And a lot of the advancement that we’ve had in artificial intelligence solutions as of late,
sometimes can’t explain where that came from. Whether it’s the weighting of features, whether it’s the type of model that’s being developed, it all has an effect. So it’s up to companies like Symbiotic AI to put their best foot forward, to make sure that their systems are explainable, that their insights, origins have traceable steps. And that’s something that we’ve taken into account. So we are not a black
AI and trust with clinicians is very, very important. So to that end, what we do is we provide a variety of different features that explain how our algorithms and our solutions arrive at the insights that they do, whether they be specific patient characteristics that led to an increase of a risk in a particular event, whether that be monitoring our models over time.
as they’re ingesting new data and understanding how their accuracy, how their specificity and sensitivity has changed with new data or old data in there. These are all different aspects that are presented both to clinicians. And then there’s another aspect to Rubin, which is IT teams at enterprise institutions and hospitals, hard calf labs are going to be tasked with the implementation and monitoring of products that have artificial intelligence components built in.
There’s also a component in Revised AI that’s intended for IT administrators, chief information officers, chief technology officers, so that they have the confidence to understand that not only are our systems providing value to clinicians and providing that traceable logic model, just like the clinicians would have, okay, I’m analyzing this factor for this patient. I’ve read this randomized control trial. This is the case with a particular patient and this is how I’m arriving to my…
you know, a conclusion for the best treatment approach, they can trace that back in our system as well. And then for IT administrators, they can overall monitor the health and also have compliance, which is, you know, a big thing for artificial intelligence systems right now. You know, we operate in the software as a medical device space. So there can be, you know, although minimal risk to patients, they are real, which means that we must, and we do.
Reuben (22:49)
Mm
Arjun Puri (22:57)
ensure that we have quality management systems and risk management systems in place to ensure that what we are building has traceability as well. So for those requirements from the standard side, it’s essential that companies and clinicians alike are aware of what those requirements are, whether it’s maintenance of those systems or how to responsibly train models as they’re
heading into adoption of some of these solutions because they might not be mandatory requirements from regulatory organizations today. But my prediction would be that that’s not going to be the case for very long and that regulatory organizations will want to see responsibility from software as a medical device companies or medical device companies that leverage artificial intelligence to follow those standards to prove that they’re able to do this responsibly and ethically.
And that’s another place where symbiotic AI is differentiated in that the data that we have received access to to build our models is not just scraped off the internet without permissions or licensing. We have securely, ethically accessed data with appropriate licensing terms so that we can get to this stage. And that continues to be built into our terms and conditions and business agreements.
Reuben (24:20)
Mm
Arjun Puri (24:26)
You know, we do this responsibly moving forward as well.
Reuben (24:30)
Well, I love all of that and the approach you’re taking to really responsibly build this product. Obviously, it takes a lot of different skills and knowledge to make something like this come to life. Maybe you could talk a little bit about the team, of different roles and kind of how you’ve built up the team as you’ve built the product.
Arjun Puri (24:58)
Yeah, you you’re only as good as your team, right? And that’s where, you know, successful CEOs really prove their worth, right? How can they attract the right teammates who are driven by, you know, similar motivations and can support the unified mission of the company and the vision of the company and what you’re trying to do? And fortunately for us, we’ve been privileged that we’ve been able to find these teammates and work with these teammates, some from day one.
And we continue to add more teammates who are passionate about solving similar problems that we do. You know, we are trying to democratize precision heart health for as many people as we can around the globe. And that is what’s really kind of the unifying force between our team. So right now we’re small and mighty. We have my co -founder and chief technology officer who has a deep expertise in data science and machine learning. Dr. June Lee, who’s also a professor at the University of Calgary.
And that’s really how the company was initially kind of spun off and our foundations were in academic research. It came from an evidence -based foundation. So that way we could have some confidence in the fact that what we were creating is evidence -based, is going to lead to heavy impacts for the healthcare field. And then we added to that team by bringing on
talented and motivated teammates in the user experience design and development space, in the product management space. Because we’re building healthcare products and healthcare informatics products, information flow is very important. So we also have interface developers on our team who understand what it takes for information to flow from one information system to another information system.
how that information is passed back and forth securely and privately, because when you are dealing with personal health information, that is of utmost importance. And we have an advisory board that has interventional cardiologists with practice experience, both in Canada and the USA, which really adds a tremendous amount of value because it’s like having a resident end user who’s available.
for you to bounce ideas off of, for you to understand whether what you’re creating is truly of value. These components of the team have been massively helpful to help us accelerate our journey and our growth. We continue to be interested in bringing on additional individuals, of course, software developers. And because we’re working in the AI space, we’re also cognizant that the company has a, the team has a commitment and a responsibility
to AI ethics in the medical field. So we’re also inviting medical ethicists and data science ethicists to work with us and collaborate with us in that space. We also work with health economists as well as human factor specialists because what we’re creating has to be IEC compliant from a use case specifications perspective. unless you can quantify the benefits
of what your solution is able to provide through appropriate health economics forecasts. You’re not really able to accurately claim that this is what the potential of the product is with respect to a return on investment. So these are some of the critical components of the team at this stage and we’re privileged to have passionate and motivated people working with us.
Reuben (28:35)
Right.
next month. Well, it sounds like you have a great team around you and you’re solving an important problem. It might be too early to talk about like, you know, success stories for patient outcomes, but in your mind, like, you know, what’s a future ideal success story of how, you know, Revise AI has impacted patient outcomes?
Arjun Puri (29:08)
Yeah, so this is where you have to put on that future vision hat and look into the future and think about what would it be like if it was an ideal state and you were helping millions if not billions of people around the world. The reality is that coronary artery disease is a leading killer of individuals worldwide and it tends to be a strain on healthcare systems across the world. So for us, success is
you know, avoiding repeat procedures for patients who don’t need that repeat procedure. And the first decision is the appropriate decision for us, you know, avoiding future heart attacks or strokes is success for us clinicians spending less time behind the computer, documenting everything that they’ve, you know, received in terms of insights and automating that is important. And that’s success criteria. And then, you know, making the healthcare system and the
delivery of heart care more sustainable because you’re avoiding repeat procedures, because you’re being mindful about physician time is also a success metric. know, often in the healthcare space, you’ll refer to or hear about a framework called a quadruple aim or the triple aim. And, you know, that’s really what we’ve focused on. How do we improve the health of our populations? How do we improve the provider?
and patient experience and how do we make care more sustainable? And if we can have some of these impacts and these metrics, you know, with the revised data that we’re forecasting, we’re confident that we’ll get there. That’s what success would look like for us. And then there’s, you know, always opportunities to apply what you’ve learned and done in the cardiology space and discipline into other disciplines of medicine, right? So I think the, if we can,
transcend and transfer our learnings and the foundations that we’re building in the cardiology space to other disease spaces later on, you maybe long into the future, that would give us an opportunity to have an impact on more than one discipline in medicine and, you know, kind of grow our product offering at the same time. And, you know, at the same time, we want to make sure that our clinicians are happy, our patients are happy and living better lives at the end of the day, and the team is happy.
right, because the team’s got to be motivated and willing to come to work every day to build what we build. And a lot of that comes from one -on -one engagement with patient partners and physician partners. So we’re always seeking those opportunities that help propel that.
Reuben (31:48)
Well, I can certainly get on board with that vision Arjun. I appreciate that you’re putting all your energy and passion into this product. And thank you for joining me on the podcast today.
Arjun Puri (32:01)
It’s a pleasure. Thank you for having me. Fantastic questions. And I hope we were able to share information with your listeners that they found intriguing. If you want to learn more, visit our website, symbioticai.ca, or feel free to get in touch with me.
Reuben (32:17)
Thank you and thanks to everyone for listening to Moving Digital Health podcast. If you enjoy the conversation, please go to movingdigitalhealth.com to subscribe to the MindSea newsletter and be notified about future episodes.
