From Diagnostics to Data to Clinical Decision Making for Value Driven Patient Centered Care
Shahid Shah: Welcome back, HealthIMPACT audience. One of my favorite conversations, that I love to have about digital health is with pharmaceutical companies is with, people who have been in therapeutic world for a long time as opposed to a lot of startups or innovators that are getting into the space.
And so, with that, I'm really excited about this conversation with Sukhveer Roche, Sukhveer. Tell us a little bit about yourself and what you do at Roche.
Sukhveer Singh: Sure. Well, and then really excited to be here and looking forward to our conversation, Shahid So I've been you know, win Healthcare. Doing digital side of healthcare, you know, before it were called digital, right in the software and everything else for 25 plus years now in leading med tech and biotech companies.
Currently at Roche, I'm part of the organization called Roche Information Solutions, which is entity we created recently to focus on creating digitally enabled solutions for healthcare ecosystems, both in terms of patients and clinicians, and my primary focus. Neurodegenerative diseases, the dementia, Alzheimer's and cardiovascular metabolic you know, which of course, you know, those two areas are pretty interlinked with each other.
And then, you know, some other disease areas like acute scare setting related, for example, acute infections and population infections. So wide array of for areas, but really about creating digital solutions to help improve outcomes for clinicians and patients in this, Yeah, I
Shahid Shah: love it. In fact, when you look at what Roche does you know, people have probably heard of Roche for many years.
But you can actually buy stuff at a store. Like you can actually go and pick up a Roche product or service and you can buy pharmaceuticals and things like that. And so when you think about Roche the first thing that does not come to mind is digital solutions, information management, et cetera.
So what is it about therapeutic companies and diagnostic companies like Roche where you're thinking that there is some benefit to be on that digital side or the information management side that you somehow can't do without? You know,
Sukhveer Singh: this is a question we often get, you know, we often talk about it in terms of a right to play, but you know, as you mentioned that we have, you know, a proud 125 year history of being in healthcare and you know, we are a company which has very successful portfolio, both diagnostics and therapeutic solutions.
Right. But, you know, technology and software is not new to us. So, you know, I think one area we would like to differentiate ourselves from digital. Only topic is, I think I would rather rephrase it as we all realize that digitalization. Of healthcare ecosystems is playing a significant role and that ecosystems, you know, start with actual product development, drug development, or whether it's a diagnostic test development and then adoption in the market and impact on patients and clinical workflows.
So we have been doing that for decades now. Right? So if you look at a clinical trials in terms of analyzing data science in terms of our diagnos. Tests the labs and those results being part of the clinical workflows. So while software has been part of our dna, it was more inward focused, right?
Sort of, you know, making our labs more automated, making our sort of clinical trial workflows more automated, doing better analysis, but in the process, There are few things which we think in terms of our assets, which we have acquired, which could be leveraged to participate in creating solutions enabled by digital technologies for healthcare ecosystems, right?
So number one. So we are very intimate with clinical pathways, right? Given our participation, diagnostics and therapeutics, there's huge bench of talent and also the collaborations we have with the you know, health systems and other partners in the ecosystem, which we think that knowledge we can leverage and that knowledge is of course, you know, drives deeper understanding of how the healthcare ecosystems.
Number two also is a deep expertise in data science. You know, ai, machine learning, how to manage data, how to work with the sort of you know, low quality data, unstructured data, how to create better data out of that, and insights based on that. So that's the second area of it. And third, of course, you know, the global presence.
Intimacy and with healthcare ecosystems and how do we, how we can leverage that presence to add additional value in that space. So ultimately, I think I would say it's not just about creating digital health solutions. The way we look at it is that digital technologies. Can play a significant role in optimizing healthcare ecosystems and in fact ensuring greater value from, you know, whether it's therapeutic or diagnostic interventions in the outcomes.
You know, maybe we can go a little bit more into that, but it's really a focus is digital as an enabler. Of creating these deep clinical insights and making sure those insights are leveraged at right point in the workflow, whether it's for clinicians or for patients.
Shahid Shah: Yeah. That's fascinating. In fact, when you think about the legacy that you guys have, and I love that you called out the fact that you understand clinical pathways.
Cuz when you talk to a lot of newer innovators, When you say clinical pathways, they think of something completely different than what we really mean in healthcare, which is a lot of healthcare is already supposed to be following a set of paths with a set of best practices, et cetera. And these are well understood for many years.
But we would argue that some of those are not necessarily built and rooted in the amount of data. That we now have at our disposal in the real world data or the real world evidence side. Can you talk a little bit about how do you feel that going from the inward facing to the outward facing digital tooling can help you understand patients better?
and for one specific direction. And that is that a lot of us are trying to get into value-based care better and better, but there is no value-based care in one company, right? It's a value-based care is at the patient level, at the population level with companies working together. So what has Roche found so far with digital tools that might help it better understand patients and the pathways that they need to go through to assist in the value-based care?
Absolutely.
Sukhveer Singh: So I think the way. We could look at value based care is really, it's about taking systems thinking approach to optimizing the entire workflow. Right? You know, focus on value in healthcare has been there, but sometimes we have tried to implement it in specific silos. You know, you might recall there were efforts, couple of decades back to reduce imaging costs.
Now that's, you know, very well intentioned effort. But what are the consequences of reduce maybe longer term impacts of that? Right. So value based care really is how do we take a systems based approach to improve outcomes and efficiency of the healthcare ecosystems. Now, as you know, mentioned already the healthcare ecosystems are not a monolith.
They the, they run across institutional boundaries, departmental boundaries, and, you know, organizational boundaries. And so how do we, the way I would say, The challenge is about how do we connect these silos of excellence to create a system of excellence, right? And as you can see, that itself, the biggest limiting factor to that, of course there is about incentive design and everything else, which comes through the value based sort of program design, but the even more critical limiting factor there has been.
The ability to connect the dots, and that's where technology can play a significant role. Can we create a complete view of patients longitudinal journey? What happened to them, not only in a specific care episode, but across multiple years? Right? How do multiple factors not only say genetic factors or other factors, but how do, for example, social determinants of health play into the outcomes and all that?
So I. We have known all this, right? I think what we have now is technology's ability to connect these dots in real time and provide those insights. So I think that's where, so if you look at how we are approaching this problem, we really think it's a three stage problem. Sort of a pyramid of enablement number one.
Digital infrastructure to connect the dots to create rich longitudinal patient journeys so that we understand what's happening to patients in different care settings and not only in care settings. In fact, even more importantly, the non-traditional care settings like at home, like social determinants of health and other things, which all become part of patient journey Number two, how do.
Optimize the workflows to make sure these journeys are efficient. You know, for example, can patients show up on time for their care? Another example I would give you is, for example, let's say we have a beautiful algorithm for. Determining the sepsis progression. Right. In acute care setting, I was talking to one of the you know, the leaders at a large health system and his point was that your beautiful algorithm can still fail if these, the lab sample for the patient doesn't get picked up on time and the lab results are not done on time.
Right. So that could, the whole algorithm could fall apart. So before we can start introducing precision, Specific to patients, we have to ensure a standard of care, which applies to, all right, so only then we can sort of, you know, separate true deviations from the noise. So that's sort of the second layer.
And the third layer is when you have optimized health systems or workflows, that's when. It becomes even more impactful to apply clinical decision support, personalized healthcare intervention, and other aspects on top of those systems. Otherwise, those interventions will get lost in the noise. So I think these sort of, that's how we kind of think of it, this journey in a very three layered structured approach.
It's not that these will happen sequentially, it's just that, you know, this is almost a maturity model we have to look at before we jump into highly advanced clinical decision support. Very inefficient and disconnected workflow.
Shahid Shah: Yeah. You had me at, hello when you said longitudinal Health Record and as a longtime ehr vendor.
Cto. I've been a CTO for electronic health record solution vendors three times in my life, . And one of the things that we've al we always tried to say as an EHR vendor is, well, that longitudinal record. Is really not our problem. That's, you know, we are in your institution and we handle episodes of care within your institution.
When you start to look at connecting the dots and all of these other tasks these are things that are not in the DNA of most electronic health record vendors. Short of the big ones, the discerners, the epics of the world would make the case that they can do that. So, not to pick a fight between EHRs or not but think about.
What you guys are doing at Roche and what makes you well positioned in this case to be an accessory to already existing EHRs in order to do those three very important things that you talked about. And then we'll get to the value based care in a second. But what makes you well positioned would be a good
Sukhveer Singh: place to start.
Couple of things. It goes back to our experiences. So if you look at some of the clinical trial work we do in that space, we do create very rich longitudinal patient journeys in that regard, right? So that's one aspect of it, that connecting data from disparate sources and making meaningful insights out of that has something we they've been doing for a long time.
So that's one aspect. Secondly, if you look at a diagnostic portfolio, We touch every aspect of the patient journey. You know, diagnostics are not used just for diagnostics, you know, so if you look at diagnostics range from diagnostics for screening, diagnostics for actual disease, diagnostic, right?
Diagnostics for treatment, decision making, diagnostics for treatment, monitoring whether the treatment or the therapy is working or not, right? And then diagnosis for post follow up and sort of surveillance of patient's outcome, right? So the fact. Our current portfolio already touches patient journeys across different silos.
You know, so, you know, our labs are being used in at home care setting. Point of care setting. Acute care setting, right. So I think that's a very significant aspect of going back to an earlier point about we understand clinical pathways in disease pathways and that we can participate in them quite effectively and you know, build upon that, that found.
Shahid Shah: Yeah, so it seems
like low hanging fruit which is where a lot of smaller innovators so I've been part of a startup so many times. You probably have been as well, where we try to say, well, let's just look at some small low hanging fruit in this one very small area. What in the year 2022, which we're sitting entering 2023.
All the easy stuff is taken. Right. We have to do the hard stuff. So as you look at doing the hard stuff, what are some of the things that you've been working with at as far as your customers are concerned where they didn't realize how hard something was? Especially when you're dealing with longitudinal data across institutions mm-hmm. that still has to be patient first. Where have you jump jumped in and not to make it a case study example, but just some minor examples of. You can do some hard things because of your background as long as you guys have been doing this. Oh,
Sukhveer Singh: sure. You know, a couple of examples I'll get you. So for example, oncology, let's take, you know, it's one of the most complicated.
Diseases in terms of the workflows, right? One of the topics, the challenges we decided to take on was how do we allow more effective multidisciplinary decision making around treatment planning, right? So if you look at sort of more advanced organizations oncology centers, Every patient journey starts with the tumor board discussion where multiple specialists come together and they don't want to they want to talk about patient cases and decide what's the best path forward.
So, You wouldn't believe how difficult those meetings are to prepare for, because you're looking for information from pathology, from radiology from sort of the, you know, radiation oncology emo and, you know, all those aspects right at this. So it's not only just in bringing information together, now you're trying to figure out based on the latest sort of standards of care guidelines, so latest.
Trials and all, what is the best decision to be made and that, how do you document that decision? So look, it sounds very simple. It's like experts coming together, but we decided to take on that problem by creating. A tumor board you know, application. Now it sounds like a very simple application, but you won't believe the extent to which now we had to gather data from multiple sources.
Pathologies is in different system, radiology is in a different system. Clinical data is somewhere else, and patients own preferences, they are somewhere on the paper. Right? So bringing all this together, guidelines are. Digital guidelines, they are in some PDFs on some website, right? So how do you create all this information?
And this is one very small example, or one step in a patient journey and you know, so that's where, again, applying our data science expertise, our deep understanding of different sources of data, and of course the clinical pathways, we have been able to create a very effective tool and it shows significant operational.
Efficiencies of running such meetings and making those decisions. But also, you know, it has benefits for the patients of course, because now you're having a very comprehensive discussion around this topic. So this would be sort of an example where in fact all the strengths we mention are being brought together.
Yeah,
Shahid Shah: that's a lovely application example because it also shows that until, unless you have all the personnel behind you that know this world, you couldn't even put together like a, if you have a company put together an app like that, they would make it an information management system rather than a.
Decision support system, and people often confuse those two things. Can you clarify a little bit what is an information collection system versus a actual clinical decision support system, which is what you would do in a tumor board? Yeah. You know,
Sukhveer Singh: I, I think that's a very important point, right? I mean, there's of course a role for information management systems, right?
But. What we should never underestimate is the sort of the workload and the stress under which the clinical or healthcare ecosystems are, or the clinicians are, or even patients are. So just throwing more information at them. And, you know, sort of help, you know, making them spend even more time to search for information is not going to be helpful.
I can, you know, the analogy you might have seen in last couple of years is the whole topic of remote patient monitoring. So if you talk to clinicians, they do understand the value of remote patient monitoring, but if it means now they get stream of data every day for thousands of their patients, that's not going to be helpful.
Right. So I think it's. Understanding of clinical context and converting that noise into actionable signal. That's where the real difference is because many of the information systems usually require the user to go look for actionable insight. Right? I think we have to flip that instead of that. We should be pushing actionable insight at point of decision making when the decision is being made, right?
So it's not about sending tons of alerts again, you know, we have, we know about that too, that there are too many alerts being sent and they get silenced. But it's really what is the intelligence of the system to understand what the real signal is. And what is the clinical context because you know, same signal might carry different relevance for, based on whether the patient is maybe being treated because of it's a chemo side effect, which is known.
So this doesn't require sort of, you know, five alarm fire versus it's another context. It might be highly relevant. Right. So I think that's where the difference you're pointing to. It comes between pure information system or sort of a knowledge reference system versus a clinical decision support system, which is based, which is contextualized.
For particular decision making and it's aware of those workflows. Yeah. What I
Shahid Shah: love about what how Roche is approaching this is that today we have a lot of systems where you can begin a task, like there's a to-do list manager and there's an ehr and this and that, but there are no systems optimized to help you finish a task and know that I've done and handing it over to the next one.
The other thing that we often don't have access to is the actual tool where you can make the decision documented and understand why you made it. . And so a lot of people will tell you that, oh, I've got this. Hey, as severe I've created this very cool dashboard, you'll be able to now give this to your doctors and the doctors will be able to do all of this.
And when I look at that, somebody would present that to me. I would. Giving me a new dashboard is a failure condition. It's not a success condition, right? Removing a dashboard and just sending a an email to the doctor saying the following four decisions have been made. Three nurses have already been connected of this fourth one, though needs for you to make the following decisions and connect to this for.
Right. That is that I think is a great way for you, that's what you're describing is the difference between the information system versus the clinical s and support system is that the better clinical decision support systems you have, the less the need for a information system review or a dashboard would be, is that how you would look at
Sukhveer Singh: it as well?
Yeah, absolutely. You know, that, that comes start and also, Human resource is sort of the most special resource in terms of clinician's time and everything else, right? So another role good or a well designed clinical decision support system can play is to focus human effort on the most complicated cases, right?
So how do you make sure that the interventions. Are aligned with the C severity of the signal, right? So that also becomes an important part. So for example, is that something side effect which a nurse practitioner could take care of? Or maybe this is actually a signal, you know, which is coming because of logistical issues with patients showing up for their appointment that actually might even not require clinical staff.
Right? So I think that's also a role we should Try to make sure clinical decision support is about optimizing human resources to focus on where their effort is most needed, like today. That can be a big challenge, right? So if every phone call is going to a nurse, say, with oncology side effects, so managing those side effects, it can be.
Not only, it's a lot of work, but also it can delay care for patients who really need help at the right time. Yeah. One, one
Shahid Shah: thing that I've noticed probably the last 15, 20 years is that we have venture capitalists who are initiating innovation and innovative companies. They're almost always telling their startups, stay away from the fda stay away from heavy regulation.
To me, those solving a healthcare problem and staying away from regulation seems like a dichotomy. So given obviously you guys are not afraid of the fda, you're not afraid of making devices where you have to depend on them for safety, can you talk a little bit about is it possible to make true clinical decision support where you can pull humans at some level?
You're not trying to replace humans everywhere, but trying to remove some humans where they're not necess. To move the the work over to where the humans are absolutely necessary. That requires a level of either a class one or class two medical device, or some higher level AI ml, which is going to have to go through some regulatory bodies.
So is Roche afraid of any of those? One is a good place to start. And then how much could you actually drew in true clinical support without FDA regulation or regulated systems in.
Sukhveer Singh: Well, you know, look you know, regulations exist for a reason, right? And I think they have played an important role in creating safe and effective sort of solutions in the healthcare market.
And, you know, there's fda, there are regulatory bodies are across the globe and we work very closely with them. And, you know, so, so I think regardless of regulation, You know, we from our data science expertise and expertise in this area understand all the challenges with creating clinical algorithms and their deployment.
You know, whether it's about, you know, data shapes, whether it's you know, making sure those the right, they're trained on the right data models, that they can be applied from one population to the other, right? So there is aspect of this. Which we really need to watch out for. Right. And there have been countless examples of where algorithms have un lifted up to their expectations.
And I think while, you know, there would be an element of learning or improving the algorithm in the field, but what are the tools for monitoring that shift? What is the tool of manta monitoring the quality of the elbow and all that? So I, you know, It's not an easy task. It requires commitment of resources of, you know, sort of, and we are being very diligent about it.
Right. Is there a role for data driven algorithms, machine learning and healthcare? Absolutely. Right. Just the sheer amount of data and knowledge we have, there is no way we can just process it at the human pace. Right? So, so absolutely there's a rule, but we kind of always think of these tools as elevate.
Human capability to make decisions, right? Not necessarily replacing it, right? So that's one part. And second is how can we sort of, you know, how can we leverage these tools to increase the gold standard bar, right? So as you can imagine in any, he, any system, including healthcare ecosystem, decisions are being made with different quality.
So can we learn from the best and make that collective experience? You know, and distill it into these tools and make it accessible to people who might not have the, you know, sort of the access to the latest knowledge. You know, how can we make every physician, every nurse operate at sort of the, with highest capability in terms of knowledge and access to these tools.
So that's, you know, we will constantly looking of, Bought it in that terms rather than sort of, you know, looking at that these tools will replace sort of actual clinical decision making, right? So we just want it to be, for example, look at a tool. Which instead of a physician having to go to a website and, you know, browse through few hundred pages of guidelines to figure out can we contextualize and digitize those guidelines and provide the two or three potential next steps in a decision journey or a patient journey.
That, that makes them more efficient. Ultimately they are making the clinical decisions. So the tools have a role to play. But again, I wanna emphasize that it's about elevating our ability to make those decisions rather than replacing it. Yeah, I love that.
Shahid Shah: In fact that last piece, which is to say that the clinical decision support, it's in the name right decision support tool, and really what one lesson to learn from here that you've taught us.
That big difference between an information system that is just gathering and presenting information rather than a tool that is actually helping you make a decision by giving you supporting materials the latest research on it, collecting the information together and presenting it one way, and that the tumor board's job is not to make a file bigger and bigger, but in fact it's to make the.
Small as possible so that when you present it to that group that is making these decisions, they can make them really quickly. That's fantastic. And in fact then, you know, we have a few more minutes left in our in the rest of this conversation. If you can talk a little bit about where do people today get the idea of value based care quote wrong in your opinion, in the sense that they're not thinking about value based care in, in, in a particular way that Roche is thinking about it.
And a couple of things to me come to mind is you can't just say, assuming that the, all the incentivization is there, which I think the government's trying to do its job regulations will come in. Telemedicine. Is there remote patient monitoring? Is there, what is it about value-based care that we're still missing that you think somebody the size and scope and hef and resources of Roche could help bridge the gaps?
Sukhveer Singh: But you know, I think as with any topic, so there are great examples of value-based care being implemented and then sort of there's the long tail, you know, along that, where the variation. So I think we are part, you know, it's not that, you know, we have sort of as an institution, all the answers. I think as a collective participant, stakeholders in healthcare ecosystems, we are all learning about value-based care and how to implement it, right?
So I think it's really about. How big is the system we are trying to optimize, right? So, you know, are we trying to optimize an episode of care? Are we trying to optimize a three year journey of the patient, or are we trying to optimize long-term journey of the patient? So, I think we would all agree that the current cost, health cost escalation is not sustainable.
Right. So as a society, I think if you look at those curves, I don't think anybody would say that. This looks great. Right. This is, that's right. So the question is, one part of it is, of course it's an important goal, which is how do we make each part of. Sort of cost more efficient. You know, this is about lowering the cost overall in every part of it, but that won't be sufficient.
Right? And by that I mean, for example, if a patient does get hospitalized, there's only this much you can do in terms of making that process efficient. You have to monitor for certain things. You have to do those things. What's going to be important is how do we often people call it shift the cost curve to the left, right?
So, That focus in value based care is just emerging, right? So, which is about let's identify what are the populations at risk for certain diseases, right? Let's then do better screening and diagnosis of those populations, right? So, and then, you know, similarly, how do we. Make sure that if you look at it, this journey is a progression from a low risk to high risk to actual diseases, right?
Even if, so for example, let's take diabetes journey, right? It starts with diabetes, chronic kidney disease to nsta, renal disease. You know, so it's a, I think our goal is first of all, slow down or stop progression from one stage to the other, and that's what is going to impact the curve of total cost curve.
So I think question. Are we effectively setting up sort of, if it's not about creating a monolith, but inter stakeholder collaborations, information exchange incentive design to enable this long term view of patient journeys and making sure we are optimizing for this system. Right. And which is very difficult to do sometimes because, for example, for a commercial, The length of a patient's association with a commercial plan is maybe about 18 months, right?
But if you're talking of a cost curve optimization, which will range say 18 years, 30 years, say for a diabetes patient, how do we bring that about? So I don't think it's about that we know something which the system doesn't know, but I think as a collective stakeholders in the system, that's where the real sort of, opportunity lies.
Otherwise, we might end. What I would call doing lot of things, which would amount to squeezing the balloon. Right? So you just push the air to some other part of the balloon, but the total cost curve is not being impacted. Yeah, it's
Shahid Shah: a, it's a great way of putting it. So in the last 15, 20 seconds that we have you have a lot of our audience members at the Health Impact side are coming from the care delivery side.
They make decisions on helping or choosing partners like Roche. What would be a simple ask for them? What kind of help would you want from them to work with Roan to solve some of these kind of problems? A quick call to action would be
Sukhveer Singh: helpful. Absolutely. And I think it's a, I really are I think us would be we have to do this together, right?
So it's not like, you know, I would say Steve Jobs that we can bring the new version of iPhone and surprise everybody, right? So if we are looking to. And create new algorithms to test them in real world and all that. We need real world settings into which to act and to work closely with committed clinical systems where we are working together to bring these innovations.
So I think it's in invitation is really about. You know, we would love to hear from institutions who would like to work with us as we go on this journey. It's a long journey, but it will require that collaboration, co-creation, and commitment to be able to do this.
Shahid Shah: That's fantastic. And in fact, my last piece of advice would be, Just call ups , the Roche group, to explain and show a demo of just a tumor board on how to a tumor board app that Sevi is talking about.
I would love to have everybody take a look at that and understand what they can do, because that would be a great conversation starter about if you can actually write apps that use our modern information systems to help come to an agreement in a tumor board about next. And actions for our patient.
Whoa. That's a big deal. And definitely needed in the market. With that, VIR, thank you so much for joining us here with the health Impact team, and we look forward to hearing from you in the future.
Sukhveer Singh: Thank you so much. It was a great conversation.
Shahid Shah: Appreciate it.