For years, Len Testa has been using mathematical models to help families answer one of their most challenging questions: how to optimize summer vacation by selecting the best theme parks with the shortest lines, the best restaurants, and the most cost effective hotels? His website, Touring Plans, provides all the decision making your family of four will hopefully need to make the most of your next trip to Disney World. Today, however, Len has his sights set on a bigger challenge: clinical decision making for diabetes management. Len’s new interest is driven by his belief that, “…we’re pushing too much complex decision-making on to healthcare providers…We shouldn’t be surprised when we see sub-optimal outcomes.”
Working with diabetes specialist Dr. Bradley Eilerman and utilizing the power of Vericred’s databases, Len has developed GlucosePATH. GlucosePATH is an application that takes in a wide array of data points about each patient and determines the class of medication the clinician should probably prescribe. The tool goes beyond evaluating the vitals and historical clinical data that drive most other diabetes management tools by incorporating financial factors such as the patient’s health insurance plan and budget for a co-pay. This is where Vericred comes into play providing normalized, structured information on health insurance plan design and formulary data that is used by GlucosePATH’s decision making tools.
Today, GlucosePATH has successfully navigated the FDA and is now in its first clinical trial with initial results coming soon. In the meantime, Medgadget had a chance to connect with all three key players involved in the creation of GlucosePATH to tell us a little about their involvement in the future of diabetes management.
First, we had a chance to hear from the driving force behind Touring Plans himself who has turned his penchant for improving decision making to healthcare.
Michael Batista, Medgadget: What led you to making the leap from tourism to healthcare with the development of GlucosePATH? Are there parallels between your previous work at Touring Plans and GlucosePATH?
Len Testa: Brad really got things started by emailing me out of the blue back in 2015. He had read an interview where I described the math behind avoiding long lines at theme parks. He asked if it was possible to do the same sort of thing, to choose medicines for treating type 2 diabetes.
It turns out that we were already solving a similar problem – trying to find the least expensive set of admission tickets to Walt Disney World for your vacation. It’s an example of the “knapsack problem” in Operations Research.
Because the problems are so similar, it only took a couple of weekends of work to adapt the tickets code to make the prototype of GlucosePATH. After that, it was a process of refining the decisions and the data, which is another way of saying we needed to get Brad’s decade of clinical experience into code. It helped a lot that Brad had programming experience, too, and had thought about this problem for a long time.
The most interesting thing to me as a computer scientist is that the problem of choosing a medication regimen is much easier – from a computational perspective – than minimizing waits in a theme park. There are around 6 million different combinations of diabetes medications to choose from in a typical office visit, but there are 2,432,902,008,176,640,000 different ways to visit 20 rides in a theme park. So the theme park problem is about 400 billion times bigger in terms of things that have to be considered. I’m moderately surprised that more medical decision-making isn’t automated, given how relatively straightforward it is to just evaluate every possible option.
Medgadget: How does GlucosePATH differ from other mobile health tools for diabetes patients? (i.e. WellDoc, MySugr)
Len: I think the two big differences are that we incorporate the patient’s specific insurance plan, monthly budget, and coupons into our decisions; and we evaluate every possible combination of therapies for that patient’s medical condition.
The ability to link in a specific insurance plan through Vericred is critical, because it allows the physician and patient to share in the decision-making around treatment options. We hear from pharmacists all the time who see patients walk up to the cash register, see the price of their four diabetes medicines, and then have to decide which two of the four they can afford. That’s because the price of the medicine wasn’t discussed during the patient’s appointment. We can change that.
Our approach seems to be unique, which is mostly very good news. We did have some difficulty when submitting our 510(k) to the FDA. They ask you to identify a ‘predicate device’ – something that they’ve seen before that you can point to and say ‘this thing we want you to approve is like this other thing that you’ve already approved’.
There aren’t many examples of decision-support software in the FDA’s predicate device list. We identified what we thought was a reasonable example for our 510(k). In our face-to-face meeting in Washington, the response we got back was along the lines of “we’ve never seen anything like this.” But the FDA was absolutely fantastic to work with. The feedback they gave was specific and actionable, and made the software so much better. We couldn’t have got through the process without their help.
Medgadget: What anecdotal or clinical evidence can you share about the impact of GlucosePATH on patients using the tool to manage their diabetes?
Len: The first big “proof of concept” we did with the software was a 2016 retrospective analysis of 200 type 2 diabetes patients. It compared their latest HbA1c levels and physicians’ medications recommendations to what GlucosePATH would have done. That indicated the software’s recommendations would have lowered the patients’ HbA1c by 0.8, and save each patient around $750 per year.
Based on that we were able to start a clinical trial of GlucosePATH at Saint Elizabeth Healthcare in Cincinnati in October 2016. We’re measuring the HbA1c change for those patients 90 days after changing their medications to what GlucosePATH recommends. We should be able to report interim results for that very soon.
Medgadget: What are the current plans to commercialize and grow the use of GlucosePATH?
Len: Our goal this year was to address the two big questions that people have around the software: “Does this work in the real world?” and “What does the FDA think about all of this?” I think those are answered now. The FDA determined that we were covered under parts of the 21st Century Cures Act, and thus not subject to FDA review. We’re confident in the science and technology behind these decisions.
The systems and insurers we’re talking to now are interested in answering questions around population health and analytics. For example, we identified $1 million in annual healthcare savings for one company by eliminating ineffective medications, better adherence to the formulary the company negotiated, and judicious use of manufacturer coupons.
We’re also talking to insurers who are interested in seeing if lower HbA1c levels lead to lower long-term healthcare costs by reducing expensive events like emergency room visits and complications. There’s lots of research that seems to indicate it will, but quantifying the software’s role in that for a specific population would be great.
We’ve also done some interesting one-off projects, like trying to determine the optimal price of a specific, branded drug in a company’s formulary, given the performance characteristics of the drug and the characteristics and price of every other drug in the formulary. We did that for a pharmaceutical company but it’s the kind of question that every insurer and employer should be asking.
Next, we heard from Dr. Bradley Eilerman, a diabetes specialist who played a key role in the vision and clinical development of GlucosePATH.
Dr. Bradley Eilerman
Medgadget: How did you become involved in the ideation and development of GlucosePATH?
Dr. Bradley Eilerman: Developing decision support for Type 2 diabetes management has been a dream of mine for some time. I realized that sophisticated medical management could be quantified into an algorithm when I was giving lectures to other physicians about the endocrinology approach towards medical therapy. The addition of a cost layer completed the idea. All physicians are faced with hundreds of formularies with different configurations and copays. Ignorance in cost knowledge leads to nonadherence and frustrated patients. I wanted to be able to offer affordable regimens that did not compromise my medical values. I knew that Len was brilliant from his work in travel. I was fortunate that he could help my early ideas become a reality. It has been the most satisfying work in my career so far.
Medgadget: How does GlucosePATH improve upon the current standard by which specialists help patients manage diabetes? In what situations do you find the technology most valuable?
Dr. Eilerman: I realized the power of GlucosePATH early in our testing phase. I would run the software on patients that I saw throughout the day. I noticed that the GlucosePATH results were logical but sometimes a little different than my own. When I looked at the choices in more detail, I came to the realization that I was favoring certain medicines because of familiarity more than logic and not moving aggressively enough. Now that the software is much more sophisticated, I use it on three main levels. I use it on patients with challenging medical needs (i.e. renal insufficiency, fear of needles, etc.). I use it in challenging economic situations, largely in Medicare scenarios. Lastly, I use GlucosePATH for population analysis to find where new therapies or cost changes affect prescribing trends.
Medgadget: What impact do you see digital health technology having in patient care? Is this technology already in play or are we still in the process of realizing what is possible in digital health?
Dr. Eilerman: I think we are just beginning to scratch the surface of where digital tools can help in clinical decision making. Part of the disappointment with technology to this point is the fact that we have largely reproduced a paper world in a digital form. This improves access to information, but does not provide the tools to analyze the information. The myriad of calculations inherent in analysis of information is where our digital world will shine. We are excited to be part of the early process of leveraging this power to make good choices.
Finally, we heard from Michael Levin, CEO of Vericred, the big data platform powering tools like GlucosePATH.
Medgadget: How does Vericred power GlucosePATH?
Michael Levin: Vericred’s role with GlucosePATH was, and is, the same as it is with other insurtech and digital health applications. Vericred is the data layer on which these applications are built.
Vericred did not have any direct role in the development of GlucosePATH. However, we deliver complete, normalized and structured data, in this case health insurance plan design and formulary data, through a modern API so that Len and his team could focus on building the algorithms and the user experience that make GlucosePATH what it is, instead of building the data.
Medgadget: We’re seeing a shift towards more data transparency in healthcare. What opportunities does this trend create in the market?
Levin: More data transparency, and data liquidity, are critical to delivering on the “Triple Aim” of healthcare: improving the patient experience of care (including quality and satisfaction); improving the health of populations; reducing the per capita cost of health care.
GlucosePATH is a perfect embodiment of advancing the triple aim. It improves the patient experience by using algorithms to recommend a drug regimen that reduces HbA1c, minimizes side effects, and is affordable to the patient. It holds the opportunity to “move the needle” on population health given that diabetes is such a broad-based problem, and it has been proven to reduce costs. Without access to the underlying data many of these GlucosePATH benefits could not be realized.
Medgadget: Would developing GlucosePATH not have been possible 5 or 10 years ago just in terms of the availability of the data that Vericred is using to power this technology?
Levin: While the algorithms could have been developed without the data, it is the data that fuels the algorithm, and indeed the app. In the absence of Vericred’s centralized plan and formulary data, it is likely that the applicability of GlucosePATH, meaning the patients it can serve, would have been severely limited. For example, it may have been limited to an employer, or a carrier’s members. Instead, the benefits of GlucosePATH can be enjoyed nationwide across a large swath of the population.
Link: GlucosePATH…