#43 - Dosis Co-Founder and CEO Shiv Chhabra on Using AI to Transform Chronic Disease Management

Sep 22 · 12:50 min

In this Episode:

In this podcast Jeff Terry is joined by Shiv Chhabra, the CEO and founder of Dosis, which is a company that helps get chronic disease patients the right dosage of medicine. Learn more about the AI technology here.

Jeff Terry:
Hello, and welcome. I'm Jeff Terry, delighted to be joined today by Shiv Chhabra, who's the CEO and founder of Dosis, which is a company that helps get chronic disease patients the right dosage of medicine, which is obviously more difficult than it sounds. So, Shiv, welcome, hi.

Shivrat Chhabra:
Thank you so much, Jeff. Thanks for having me.

Jeff Terry:
My pleasure. Please, tell us about Dosis. What is it? What problems does it help solve? Why did you found it?

Shivrat Chhabra:
Absolutely. Dosis is an artificial intelligence powered personalized dosing platform, that's designed to change the way that chronic diseases are managed. It does that by helping clinicians, that is doctors and nurses, dose medications more efficiently, and in many cases more effectively than they're currently able to do.

Our vision, and really the reason we were founded, is to become the standard of care for chronic disease management over the next decade. The way we're approaching that is by taking a targeted approach, so starting with a specific class of drugs and a specific disease, to show that precision dosing is possible at scale, and it can really deliver value if operationalized effectively, and that it really does have the potential to transform chronic disease care. So far, we've executed on that. Since we were founded in 2017, we've grown to be widely used in dialysis clinics nationwide.

Jeff Terry:
Brilliant. Dialysis is obviously a super important topic, touches a lot of people, so using that as an example, how does your technology help get the right dose? And why is that important specifically in a dialysis setting?

Shivrat Chhabra:
Absolutely. Our first tool, our flagship tool, is called Strategic Anemia Advisor, or SAA, and that's focused on chronic anemia, which is mostly prevalent amongst kidney disease patients, so over 90% of patients who are on dialysis also have chronic anemia. Their anemia is managed by a class of drugs called erythropoietin-stimulating agents, or ESAs, which are absolutely critical for those patients, to prevent blood transfusions, but are also associated with increased instances of adverse events, like heart attacks and strokes. Because of those concerns, ESAs are supposed to be given in the lowest doses necessary to prevent blood transfusions.

That's where SAA comes in. SAA evaluates a patient's historical ESA dosing, and the response to the drugs they were given, finds that patient's place on the spectrum of dose response, and recommends future dosing to keep their anemia as well-managed as possible, while exposing them to less drug in the process. We've found that providers that use SAA can reduce their patient exposure, or their patient's ESA exposure, by an average 25%, while maintaining or improving their anemia outcomes.

Jeff Terry:
So it's titrating the dosage in a very sophisticated way. I want to unpack the technology of how you do that, but before we do, what are some other diseases or medicines where this is relevant, or is in your pipeline to be relevant in the future?

Shivrat Chhabra:
In our immediate pipeline, we have an iron submodule, that was actually just released last week, that will work in tandem with SAA to comprehensively manage anemia. So SAA is what helps with titration of ESAs. Intravenous iron is often given with ESAs. It's one of the building blocks to manufacture red blood cells. That was just released last week, so our customers will now be able to simultaneously dose ESAs and iron.


Next in our pipeline is a module that personalizes the dosing of mineral and bone disorder, which is also a condition that is experienced by kidney disease patients. That is treated with three drugs in tandem: vitamin D, calcimimetics, and phosphate binders. Our module would actually be, we think the first module on the market to do this kind of AI-powered smart dosing for three drugs in tandem, to manage this fairly complicated condition.

Jeff Terry:
Smart dosing, by the way. That's a very crisp and clear phrase, as what this space really is. I love that. So those are all great, by the way. Congratulations. Kidney disease, relevant. Kind of what's the next chronic disease or two where you think you might apply this in the future?

Shivrat Chhabra:
That's a great question. There's a whole host of chronic diseases that this can be applied to in the future. The platform is actually very extensible. And the way that we would look at expansion outside of kidney disease in the future is really around the economic models, and the economic shift that we're seeing in healthcare.

One thing that I'm sure that you and your audience are really familiar with, and really in touch with, is this industry-wide shift that we're seeing from volume-based care to value-based care, in the healthcare space. That's really what our tech, and tech like ours...

Jeff Terry:
Enable, yeah.

Shivrat Chhabra:
Yeah, yeah. So, tech like Dosis is really needed, because it does what I just mentioned. It maximizes quality outcome while reducing cost, and quality over cost is how value is defined in healthcare, and tools like ours are uniquely poised to deliver that value successfully. Given this large-scale shift we're seeing, and as we're seeing healthcare providers struggle to adapt to these new payment systems, they're facing a lot of pressure to reduce the total cost of care without sacrificing the quality outcomes, and clinical decision support tools, and especially the AI-powered CDS tools, like ours, empower them to do exactly that, to compete most effectively in that payment method.

Jeff Terry:
In that value equation and cost equation, clearly there's reducing the amount of medicine that's used while getting the right outcome, right? And finding that point. And know, but is it also the case that using this technology reduces the skill level of the physician required, so there's less specialist involvement required to find that? Is that true at all?

Shivrat Chhabra:
It is true in certain ways. For example, currently, we're receiving a lot of inbound interest from endocrinology practices that manage chronic kidney disease patients. Chronic kidney disease is one of the conditions that is kind of on the forefront of the shift to value-based care. Previously, a lot of CKD patients were reimbursed in a volume-based system, and often, advanced stage CKD patients, CKD stage 3 and above, do experience anemia as well, so you have a lot of these practices that, in the past, have been kind of referring these patients outward, but now are seeing that actually, they are going to be held responsible for the cost incurred even by third-party providers that help them manage their anemia, so they would like to in-house a lot of that management, and they think that they can do it more effectively.

But in the past, if they were to try to do that, they would need to have at least a full-time manager, probably multiple full-time managers, just to manage anemia, whereas tools like ours, and specifically our tool, can help them manage a much higher volume of patients, with a relatively fewer number of anemia managers required. So it saves them a lot of time.

Jeff Terry:
Absolutely. And I've heard that. That's the real definition of scale, right? More benefit with less labor required. Can we talk a bit about the tech? My first question is, is SAA a 510(k) medical device?

Shivrat Chhabra:
The FDA distinguishes between clinical decision support devices and clinical decision support non-devices. SAA meets all the criteria of a clinical decision support non-device, and one of the most fundamental of those criteria is transparency. A user should be able to follow what the inputs are, have an independent basis of judging the outputs and evaluating them, and also, should be able to follow the logical flow of what's happening within the algorithm. Within our UI, our users are able to do all of those things.


We do have a FDA 510(k) cleared version of the same algorithmic engine, and that's meant to be an embeddable backend solution, that we just receive data from providers and we send them back recommendations. That version of the tool is a CDS device, and that has 510(k) clearance.

Jeff Terry:
That's brilliant. Very clear distinction, and I think a good distinction that the FDA's made there. Cool to know that you guys can do it either way. Can you say a bit more about the AI? Things on my mind are, I guess, two specific things. One, what type of data is it ingesting? Two, does the algorithm learn? Is it sort of a machine learning type thing? Then, three, does the UI live inside another system, like an electronic medical record? Sort of how does the tech work and present to the user?

Shivrat Chhabra:
Sure, so our UI is a standalone, cloud-based system, so they can access it using any web browser. The way that the process actually works... The data that's taken in is very simple. It's historical hemoglobins and historical ESA doses. We take up to three months of data, but we don't need all three months of data in order to issue recommendations. We can really start with just one data point, and take up to those three months in order to kind of make the recommendation as optimal as possible, and as informed as possible. But essentially, from that input data, we first calculate the patient's place on the spectrum of dose response, from hyper-responder, or someone who is very responsive to the medication, to hypo-responder, or someone who is really not responsive to the medication at all.

That's where the AI component comes in, is finding that patient's specific place on that spectrum and building out a personalized dose response profile for that patient. After that point, we project out, based on the available range of medications available, we project out what the hemoglobin trajectory for that patient is going to be, based on their place on the spectrum of dose response, and we... That's the basis of our optimization, so based on the expected trajectory, we are able to select the optimal dose that gets that patient as close to target range as possible over the next four-month period.

Jeff Terry:
That's brilliant. Well, congratulations to you both on that success, and yeah, thank you very much for joining the podcast today.

Shivrat Chhabra:
Absolutely.

Jeff Terry:
Perfect. And with that, I'll close the podcast.

As a rapidly growing health system serving a wide geographical and largely rural area, Deaconess needed a way to ...
Serving a large geographical area across four states, Deaconess Health System is evolving their telehealth programs to ...
As an experienced charge nurse for a fast-paced Medicine unit with complex, high-acuity patients, Vicky Martin and her ...
We must sustain our caregivers if we’re going to achieve our other sustainability goals. See how leading health systems ...
Seth Sternberg and his team at Honor are committed to changing the aging experience. For them, that starts with caring ...
Matthew Mawby is a nurse and co-founder of StaffHealth, and he's working to take the hassle out of nurse staffing. Jeff ...
As the Regional Director of Capacity and Throughput for the Providence Health System in Oregon, Helene Anderson has ...
Leveraging technology already in place at Oregon Science & Health University, Apprise Health Insights partnered ...

Subscribe to Podcast Updates: