“AI predictions need a thoughtfully designed closed-loop to drive action”
In this episode, Mudit Garg discusses the evolution of Qventus and how they are applying AI to help hospitals and health systems in managing their operations with real-time predictions for improved care delivery.
With a focus on patient flow automation, Qventus helps hospitals and health systems in reducing the length of stay and other operational metrics. Based on early insights that machine learning models and prediction scores can be confusing to users in the absence of an accountability engine, the company has developed an interdisciplinary approach, augmenting AI with behavioral science principles to drive sustained improvements in healthcare operations.
Welcome to The Big Unlock podcast where we discuss data analytics and emerging technologies in healthcare. Here are some of the most innovative thinkers in healthcare information technology talk about the digital transformation of healthcare and how they are driving change in their organizations.
Paddy Padmanabhan: Hello again, everyone. This is Paddy and welcome back to my podcast. It’s my honor and privilege to have my special guest today, Mudit Garg, CEO and Founder of Qventus. Mudit, thank you so much for joining us and welcome to the show.
Mudit Garg: Thank you so much, Paddy. It’s an honor to be here. I appreciate the chance to be here. I’m looking forward to our conversation.
Paddy Padmanabhan: Wonderful. So, let’s get started. Tell us briefly about the company and its evolution and what is the marketplace need that Qventus is trying to address?
Mudit Garg: Yeah, it’s a great question. So, we provide a solution for hospitals and health systems to manage their operations in real-time. And specifically, where we focus is patient flow automation. So, reducing the length of stay, excess days in the inpatient environment, improving the throughput in the ED environment, things like that. Our platform has both artificial intelligence and behavioral sciences built together. And what it does is it empowers frontline managers so that they can identify, predict bottlenecks before they happen. But not just stop there but orchestrate solutions, drive accountability long term. And we layer on top of that a set of operational experts to bring those new capabilities to life in the hospital environments with process design and management practices alongside. So, that’s the market need where it comes from, as you know, is this immense pressure on hospitals and health systems across the country right now to drive a lot more efficiency. And this has been true for some time. You asked about the evolution of the company. And for me, a lot of this came from probably 10 to 12 years ago, doing process and performance improvement in health systems. And I don’t know if you probably felt the same way, but I definitely thought working in the hospital the first time that it was remarkable. The quality of the people; world-class equipment, world-class therapies, world-class clinicians are available at most hospitals that we work with. But on the other hand, as a patient, we really struggled to provide them with world-class care. And it’s really despite immense diving catches and super-heroic efforts from these clinicians. So, it’s really an odd dichotomy that those two things exist simultaneously. So, the market needs really, as we dug into it and as I dug into it early on, was how do you create operational reliability? How do you make everything else around the clinical care, reliable, repeatable and mature, so that world-class operations can exist to truly unlock the potential of people? So that’s kind of where we started from. The evolution of that has been very interesting. I was, you know, maybe biased a little bit to look at data. That’s one of its key ingredients to create that operational reliability from the beginning. But it was very clear in the beginning when we started focusing on AI and ML that the prediction was very important. The prediction of the bottleneck was very important, but not enough to drive that. That was the first phase of the company. We went from looking at the data and seeing that people were excited about dashboards but didn’t log in when they got busy. People decide what machine learning, but just putting a machine learning score and board didn’t do anything, that didn’t drive any action and what really needed was action. So, we built a really robust platform then that can take different parts of a machine learning distribution of a prediction and activate different decisions and action, over mobile, over boards, over email, over text, many, many different mechanisms. And that was really valuable in fact getting the insight into action. But as is the case with everything we saw that in the immense scale that exists in an operational environment like a hospital, things slowly started reverting back to the mean. So, then we built an accountability engine. And what this does is remarkable. If you’re in the world of manufacturing, there is something monitoring every machine so you can do preventive maintenance in the machine before it breaks down. Similarly, if you’re in a health system environment, there’s tons and tons of processes going on all the time. What sequence and preventative maintenance can you do? And that’s really behavioral. Right. You could have if you’re seeing this unit do a phenomenal job of planning, how can you make sure as a leader you can praise them? If this unit the person has changed or somehow the quality of the process is dropping, how can you actually make sure that there is coaching involved at the right time? Our statistical monitors are monitoring all these process metrics continuously and then searching opportunities for praises and opportunities for coaching for leaders so they can scale themselves. So those are the different phases of a company going from just information and dashboards to machine learning and prediction to action, to then the accountability engine coming through. And that was a core of the platform that we built in our evolution. I’d say the last probably evolutionary phase is recent in the last year or so. Whereas we went deeper and deeper, the market, we saw that to truly bring these capabilities to life; we needed to create a set of predefined best practices of how to use these capabilities where there’s predefined artificial intelligence models of software, but also operational processes and management practices, and then put together a team of world-class experts on clinical operations, doctors, nurses, AI experts to help bring all this to life. So those have been the evolution in trying to address that market need through the course of the company.
Paddy Padmanabhan: Very interesting. There’s a lot to unpack from what you said. A couple of things that come to my mind right away that no matter how good a solution is unless you’re integrated into the clinical workflow of a hospital, it’s very hard to get users to adopt it and use it. And so, it seems like that was kind of one of the early insights you had, and you quickly went about addressing that. What I do like about the fact is this closing the loop, if you will, in terms of getting people to not just to use the platform, but holding themselves accountable in some ways through some kind of a feedback loop which tells them how they’re doing in order to avoid the reversal to the mean, which, as you know, is the bane of all management consulting. That’s right. Interesting. We’ll unpack some of that. Just tell us a little bit about from a growth perspective. I’m aware that you raised a significant amount of venture capital. Do you want to just quickly walk us through how much you’ve raised, who your major investors are?
Mudit Garg: Yeah, absolutely. I think from a growth standpoint, we’ve been fortunate in partnering with our health system partners to see growth in the work we do with them and concurrently, therefore, in the investment and that we can make in growing the company as well. We have raised to date about 45 million and we’ve been very, very blessed to have some very top tier investors from the valley and in the healthcare specifically as well. Some of the largest investors are Mayfield, Bessemer Ventures, Norwest Ventures. We also have YCombinator as a very early seed investor in the company and many other phenomenal seed investors in the company. And what’s been amazing is also some of the customers who work with us also felt compelled enough by the results they have seen to become strategic partners in investments as well.
Paddy Padmanabhan: But are you allowed to name any of your customers? You want to name one or two just for our listeners?
Mudit Garg: Yes. I mean, Dignity and New York-Presbyterian are two are the ones that have actually both been customers, but also investing in the company as well.
Paddy Padmanabhan: Dignity and New York-Presbyterian, is that what you said?
Mudit Garg: Yes.
Paddy Padmanabhan: OK. All right. Switching back to our topic at hand, which is your solution and your platform? How do you see yourself in the context of the ecosystem in which is the technology ecosystem in which you operate, namely electronic health record vendors, big tech firms that are building a lot of the capabilities that you’ve talked about in terms of advanced analytics, AI, machine learning-based decision making. And last but not least, other digital health startups who may be on to the same kind of ideas that you have. How do you place yourself in the context and in the milieu?
Mudit Garg: Yes. So, yes, you’re right. I mean, it’s a really exciting time to be in healthcare. That is just so much innovation and excitement across the ecosystem. I mean, to start with, EHRs are supercritical right there. We complement the investment health systems have made in the EHRs. Without that, we wouldn’t exist. If the data and the workflow itself wasn’t digitized, it’d be impossible to drive any kind of improvement. What our customers have found is that they’ve made substantial investment, in these areas of operational improvement over the years, both in terms of, you know, it could be in terms of process improvement, it could be a technology investment. If you look at the length of stay across the board, for example, that has really not budged much over the last 10-15 years, it’s been increasingly plateaued. So, when they look at looking to partner with us, what they’re looking for typically is the next step function improvement. And what really stands apart is, one, let’s take the inpatient, for example. In the inpatient environment we aren’t just helping you provide an understanding of [00:09:39] the workflow as isn, but our machine learning algorithms are, A) Identifying the problems, like this patient may need an MRI upfront several days in advance, then helping orchestrate the action and then helping manage throughout the ability. That’s super unique because one of the things that you do talk about the closed-loop part of it, just providing the AI/ machine learning is simply not good enough, in fact, and can sometimes be even more confusing to end users but closing that loop, providing the AI/ machine learning, helping create the action, and then helping create long term sustainability. That’s what is supercritical to our customers. And that’s where they see it most different. Of course, EHRs are pretty critical in view of the process improvement consulting teams that exist are pretty critical ingredient from a mindset standpoint, do all of that. But this infusion of AI and behavior science, not just AI, but the behavior science of, we think of like, how do I change behavior as a human? I need to have a cue, something prompting me to do something. I need to have the right thing to do. The easy thing to do. And I need to have some feedback on accountability. We have incorporated those vague principles in creating this organizational behavior change as well.
Paddy Padmanabhan: Yeah, fascinating. So, is it fair to say you mentioned length of stay a few times? So, is it fair to say that, that is one specific problem that you’re focused on and have been able to demonstrate results, and by extension, is that kind of the main use case for your platform?
Mudit Garg: So, our platform is fairly extensible. Patient flow ends up often being the first-place customers start because as one of the things for health systems among many, many things they can do. Length of stay and patient flow is something that no matter where in the spectrum of fee-for-service to the value-for-service organization you might be, It’s one of those rare problems where the incentive of the customer, the patient, the incentive of the hospital, of CMS or the payer are all aligned. No one wants the patients to have to stay an excess amount of time in the hospital. The patient doesn’t want that, payer doesn’t want that, the hospital doesn’t want that. So that is an area where we’ve seen a tremendous amount of pool as a result from the market. And increasingly as hospitals and health systems look at Medicare break even. How do we break even on a Medicare patient long term? Length of stay is such a massive part of that problem that we’ve seen a strong amount of pull there. But our platform extends beyond that to throughput in the E.D. and the operating room. We have worked in the outpatient access space. We are working on system operations similarly as well. But for most health systems inpatient as a state has been a place that they have had a strong interest and often a place to start with us. We have seen, as you asked the question, pretty significant improvements there as well. Statistically significant reductions in length of stay between 0.2 to 0.7 days, which is phenomenal. If you’re in a capacity constraints institution and being able to serve more patients and if you’re not in a capacity constraint, information in terms of being able to reduce the cost to serve patients.
Paddy Padmanabhan: So, it is a business case, fairly straightforward. Because it is a single number that you can track, which is length of stay. And if you reduce the length of stay by a factor of 0.2 or to 0.7, as you mentioned, the results kind of automatically speak for themselves and they are visible. Is it a fairly straightforward business case?
Mudit Garg: The business case is straightforward. The problem is complex. Yes, I think that would be a fair assessment. The business case, I mean length of stay is a top initiative for many, many health systems in the impetus to drive down unnecessary cost. And so therefore, that already exists. Any excess their patient spends is at least a thousand dollars of excess costsin the hospital. Along with the propensity to have a hospital-acquired condition or infection or other pieces and the lack of satisfaction that comes from it. So not even counting for those other downstream effects just a core excess cost is significant by itself.
Paddy Padmanabhan: Yeah. Who is your target audience for something like this? Who do you normally start conversations with? Is it a CIO, Chief Digital Officer, Chief Medical Officer?
Mudit Garg: That’s a good question. I mean, the operator is the Chief Operating Officer, the CNO, CMO. Those are ones who are already strategically often focusing on this problem, like the problem of length of stay, throughput flow. We help remove the cognitive boredom from the front-line teams that help to ease some burnout. So those are the folks that are often probably most directly eating and seeing the problem and looking to solve it. The CIO is a very critical stakeholder in the discussions right because we are complementing the EMR. They may have other tech investments. We want to make sure we have a good understanding of the data transfer and data lakes and all of those things. So, they are a critical component. And then the last piece of it, which is given the compelling financial return, the CFO are often important stakeholders as well. It can be between 10 to hundred billion dollars of annual financial benefit for assistance to the CFO and to be critical to that conversation as well.
Paddy Padmanabhan: So, if you look back at the past several years that you’ve built this business and built the platform and gone through their evolution. Can you talk to what have been the most significant challenges that you’ve faced in really validating your solution against a known problem?
Mudit Garg: So, most significant challenges in validating or are those two questions validating the solution and the challenges we face?
Paddy Padmanabhan: The length of stay problem is a well-known problem; it’s been a problem for long. So, when you went about trying to sell the idea of it. What were the most significant challenges you had to overcome in the process?
Mudit Garg: That makes sense. So, I mean, look, it is a big, big problem, right? Like when a patient, for example, is being taken care of in an inpatient environment, there are just so many things involved. There are physicians taking care of them, diagnostics taking care of them, there are procedures happening. They are going for imaging, a pharmacist helping them with medication reconciliation, so many things beingdone. It is a fairly complex problem. The first thing that we had done, and we needed to do was just to make sure we understood the complexity and size of the problem we’re going after. It is a hard problem. The next challenge is healthcare data, as you know, is often hard and messy because unlike advertising, where all the data is machine-generated, much of the healthcare data is not machine-generated. It’s human-generated, so by nature it’s messy. So, for us, one of the very core needs early on was to build a pretty significant platform capability to do real-time and automated data quality checks so we can pick up when things are off and not looking right. So, they don’t affect all the downstream applications significantly. So that was one of the big challenges, just the quality of the data, the availability of the data. How do you make sure, for example, building a machine learning model in an academic environment would control data where you can take out the outliers, where you have historical data and it has been cleaned. That is way different than running it in real-time. So, first that was a challenge to solve. The second piece is also just recognizing how much change people are going through already in healthcare on the front lines, how little time they have. And to some extent, how much fatigue or change fatigue may set in situations like that. And therefore, not just taking, hey, this is a cool prediction and putting it up on the board and expecting something to come out of it, but really very thoughtfully designing the closed loop that we were talking about. And I think that was an important challenge to recognize and to work towards because otherwise, it is easier in some ways to solve the mathematical problem that is to be solved. But forget the true problem of trying to drive change in the environment. Those I would say are probably two of the challenges along the path that we had to face, that shaped as I said in the beginning of the conversation our evolution from just predicting and prescribing actions, to actually building the accountability and getting the platform and then to actually creating these prescriptive, proven methodology combinations of tech and process that we now deployed to a team on the ground.
Paddy Padmanabhan: That’s fascinating. We hear a lot about the struggles of digital health startups, death by pilot, long sales cycles. So, on and so forth. Healthcare organizations want the innovation, they need to innovate. There is alignment around the problem to be solved to your point around something like the length of stay. But in reality, executing on an innovation program is incredibly hard. As you pointed out, should healthcare organizations be doing based on your experience so far to accelerate the adoption of digital innovation with all these constraints? You don’t have time, we have fifty other things, but we also want innovation, right?
Mudit Garg: Yeah. I mean, it is hard, right? It is really hard. I understand where it comes from because the business and the care, they are providing already is complex. There’s a lot of change in the market, so that takes a good amount of the bandwidth away from the day to day already. And so then in a way, it becomes harder. But I think what I’ve seen very effective organizations do is, one really thinks about what the no brainer moves are. We may not know what the latest CMS guidelines are going to be almost certain things. We may not know how the regulatory environment might shape. What are the no brainer things that we need to do as health systems no matter what? So for a lot of them, like reliability and cost comes to the top. Okay, so that’s that then. I think oftentimes in saying of what are the one or two or three partners we can pick and go deep with them? And that doesn’t mean you have to start with the big bang right away. It just means that, , you engage with them deeply. And I think that is supercritical. From an innovation standpoint, that it’s very hard to have a spray and pray kind of an approach where you have your hands in a lot of things or you’re just assessing the market, but actually doing it because I think for the organization to see some wins and to see some action, that’s supercritical. Honestly, even if it doesn’t result in wins, see that something was talked about, done and learnt from is critical. And the last thing I’m saying from an innovation standpoint that’s critical is that finding the operational alignment is important. The innovation cannot be devoid of what the people are feeling day to day. That operational alignment must exist. And I think in doing these three things, what we’ve found is like, for example, for us, we worked with early customers, went deep on their specific problems, created these best practices, AI models and all that stuff. So, when customers, then innovating, appreciated it. There are parts of what we do where they don’t need to reinvent the wheel and they are parts of innovation where we are learning with them and just sort of appreciating. We’re not trying to reinvent the wheel entirely. But, taking what’s already there and then actually finding unique ways of improving the innovation as well is something that I find to be effective. It is a hard space. So, the sales cycles for enterprise, in general, are hard, not just healthcare, they are long. But I think if we can do these things, then when you align with someone, you find the right partners and you make sure that it actually drives through, impacts the business and starts fueling an appetite for more and more innovation over time.
Paddy Padmanabhan: One of my guests on one of my earlier podcasts mentioned that she benefited greatly from having these sponsorship and support of some early believers and risk-takers in their client environment. And let’s face it, there is a lot of risk involved in innovation. Healthcare is a margin constrained environment. There’s not a lot of like to have. You also had that experience, you know, the early believers who make a huge difference who you co-innovate with and, you know, somehow make it happen? Has that been your experience as well?
Mudit Garg: Yeah. I mean, if you look back all the way back in the beginning, now we have a ton of outcomes and we have customers and stuff. So, it’s a little bit of a different risk profile today. If I go back three-four years ago, those same questions existed. And I think Mercy was a great example of an institution where we had operational champions, the head of their ED for example, the presence of a disintegration across the board. Folks who just jumped in, worked hard with us, understood it. But you have to have those early champions who are willing to connect the business pain to the problem to be solved, to the solution that’s coming through and help sort of get everyone excited and fired up on that. And that’s critical, especially even more so in the earlier days. Of course, as you go deeper in the market people evaluation and sort of reasons to buy change and become more and more business focused. But early on, that is absolutely critical.
Paddy Padmanabhan: Fascinating. So, guess we have coming up to the end of our time here. I have just one last question for you. You’re a classic Silicon Valley digital health startup. So, tell us what are the upsides and downsides of being in the valley?
Mudit Garg: Yeah, that’s a good question. Not surprisingly, everything sort of has this upside and downside. I mean, what I love about being in the valley is you have access to world-class talent; technical talent, and business talent. People who been at the vanguard of the AI and behavior science and innovative technology and most importantly, who have scaled businesses and transformation of industries and other industries before. And that is amazing and very, very critical to have that ecosystem around you and to have that quality of talent, it seems to be brought to bear for the problem you’re trying to solve. I mean, the flip side of that is, of course, it’s the value become extremely both costly and sometimes in that they hard to scale and just as so much of innovation comes out of here and I think over time now, a lot of the talent is actually going to other places in the country, which is phenomenal. And that’s allowing us to create much, much more of a remote enabled culture as well, where folks can still have the core of the same ethos that came out of here but actually scattered across the country and beyond. So that’s sort of the upside and downside of being here. The upside of this, that is like so the quality of talent and the experience of folks before the downside is dropped, perhaps the cost is sort of a distraction value that exists by being in the valley.
Paddy Padmanabhan: Well, not to mention the traffic on one-on-one.
Mudit Garg: Ya we’re gonna one-on-one is not a very friendly yes. Yes, you’re absolutely right.
Paddy Padmanabhan: Mudit, it’s been such a pleasure talking to you. And thank you so much for sharing your deep insights from all the work that you’ve done. And congratulations on the progress so far. All the very best to you and your team. And we’ll be watching.
Mudit Garg: Yeah. Thank you so much, Paddy. It’s always a great conversation with you. I am excited to continue the conversation. Thank you for having me.
About our guest
Mudit Garg is the CEO and Co-Founder of Qventus. Qventus is an AI and behavioral science-based system that integrates with EHRs and automates patient flow. In this role, Mudit works closely with leading health systems including Dignity, Emory, Fairview, Mercy, NewYork-Presbyterian, and Stanford. Together with Qventus, these organizations have been able to transform their operations, reducing length of stay by 0.3 to 0.8 days, eliminating thousands of excess days, decreasing ED LWBS by 50%, and more – ultimately resulting in higher margins, decreased staff burnout, and a better patient experience.
Prior to Qventus, Mudit co-founded multiple technology companies including Vdopia and Hive. He also spent time in McKinsey & Company’s healthcare practice helping large providers with organizational transformation and performance improvement. Mudit has been recognized for leadership as one of the Silicon Valley Business Journal 40 Under 40. He is a Stanford-StartX mentor. He earned his Master’s in Business Administration and Electrical Engineering from Stanford University and a Bachelors from the Indian Institute of Technology.
About the host