The American Diabetes Association (ADA) and IBM Watson Health recently announced a long-term collaboration at the Association’s 76th Scientific Sessions. They hope to combine the cognitive computing power of Watson with the ADA’s vast expertise, including their repository of clinical and research data, in order to build a first-of-its-kind diabetes adviser for providers, patients, and caregivers. But that is only a part of what they hope to accomplish together. The organizations also announced a challenge to app developers to propose cognitive innovations that can take advantage of Watson Health and the ADA to transform how diabetes is prevented and managed. The challenge will be open for submissions this summer, and developers can visit watsonhealth.ibm.com/challengediabetes to learn more.
PM360 spoke to Kyu Rhee, MD, Chief Health Officer at Watson Health, about the collaboration, what they hope to accomplish together, what they will be looking for in the developer challenge, and how pharma and medical device companies can partner with Watson Health as well.
PM360: What do you imagine Watson will be able to do with all of the new data you are going to be able to get from the ADA as a result of this partnership?
Dr. Kyu Rhee: From a broad perspective, this partnership is about bringing our respective strengths together: The ADA has the science, the expertise, and the knowledge, but also, as you referenced, an extraordinary amount of data. This goes beyond the data you might traditionally associate with people with diabetes, as it also includes data on the Diabetes Information Center, the patient materials they have created, and even the ADA’s scientific journal. So, the idea is that the science and expertise of the ADA can be paired with the power of Watson to be able to fundamentally bring insights to that data that can be used by providers, patients, researchers, and entrepreneurs. We see so many opportunities and use cases to transform the way diabetes care and prevention can be delivered through this partnership.
I understand that Watson learns more as it gets more information, but how long do you anticipate it’s going to take for Watson to take this information and make it viable for providers, researchers, or patients to use? Will Watson need to be trained in diabetes before it can even take that next step?
Watson needs data and it needs expertise to train it. But I want to highlight the bigger picture—Watson has been in healthcare since the Watson Jeopardy Challenge. Right after that, healthcare became a priority area for us. So Watson, in some ways, went to medical school with the support of Cleveland Clinic. It learned oncology with Memorial Sloan Kettering. It’s learned clinical trials matching with Mayo Clinic. It’s learned the language of medicine, so Watson already knows how to read medical literature. Currently, it reads lots of oncology materials including 300 medical journals, 200 medical textbooks, and over 15 million pages of medical text. We have a lot of confidence that Watson understands medicine and while each condition is unique, we know that the amazing experts at the ADA and all of the science and guidelines that they crafted and created and evolved will be the perfect experts to train Watson in diabetes prevention and treatment.
But when do you anticipate that Watson’s recommendations or insights from the ADA partnership will become available for providers, researchers, patients, or caregivers to use?
Watson needs to learn from the data and experts at ADA and we plan to have it ready next year. To give you an example of how long it takes, we announced our partnership with Medtronic in April last year. In January, we announced the findings from the study involving 10,000 patients and how we discovered the ability to predict hypoglycemia two to four hours before the occurrence with up to 86% accuracy—which is very relevant to diabetes. And just recently we announced the app, which is designed to reveal a diabetic’s patterns and trends related insulin levels, continuous glucose monitoring, and nutritional data to help users make the connection between their behavior and their glucose levels. That is an example of how fast we can translate data to insights and a solution that people can use. We plan to move very quickly because, as the ADA has stated, diabetes is such an “invisible epidemic” and there is such an urgency and immediate need for cognitive solutions to help prevent and treat diabetes.
I understand IBM Research has been working in this arena for some time, can you talk about that?
This is an important aspect of the bigger picture view: Diabetes is something we’ve been doing for many, many years—even decades. IBM Research has been a pioneer in cognitive and in health/medical informatics and they have been doing extraordinary modeling with cognitive and the datasets that we have. For example, in the area of diabetes, IBM Research has published a number of studies that help to advance the understanding of diabetes through predictive modeling, care management, prevention and disease modeling, and in just the past year, IBM Research developed a first-of-a-kind predictive model to provide individualized ranking of diabetes risk factors for patients. In collaboration with Israeli Health Maintenance Organization Maccabi Healthcare services, IBM Research is using the latest advances in cognitive computing to build a predictive model that could help care providers design better personalized care management plans for the early detection of diabetic retinopathy, the leading cause of blindness for people with diabetes. By working with Maccabi’s 20 years of data on two million members or 25% of the Israeli population, we will use machine learning to find patterns in large amounts of high dimensional, retrospective clinical data, based on data from Maccabi’s completely anonymized diabetic patient database. Insights could help predict the need for eye examination, as well as the optimal intervals between eye examinations, personalized to the needs of each individual.
The other announcement you recently made was your contribution to the HelpAround app. What does that involve?
HelpAround is an innovative digital health startup that is building Watson APIs into the company’s diabetes support community to optimize their ability to analyze patients’ questions and requests in real-time. The HelpAround mobile platform provides quality-of-life support to people living with diabetes by matching them with a resource that could help. Watson will enable HelpAround to analyze every help request in real-time, assess its sentiment and tone, and identify frustrations, dissatisfaction, and expressions of urgency.
What I love about HelpAround goes back to my time as a physician when I was practicing daily and how I witnessed the power and impact of social interactions in improving diabetes care and outcomes as delivered through the group visit. Physicians like me would go to the group visit because the system was set up this way that I had to show up to do a quick exam with the folks. Then I would be there for the next hour and a half with a group of maybe 12 or 15 people with diabetes. And they were talking to each other. That was the best way in which people with diabetes learned about their conditions—from people like them. As a medical director and physician who looked at the diabetes registries we had, I could see that the patients who went to the group actually had better health outcomes than the people who didn’t.
My point here is HelpAround is a great demonstration of what I would call social support—peers who support each other. It is a like having a global virtual group visit wherever you are whenever you need it. And cognitive computing is now embedded into their platform.
That seems like a good segue to the developer challenge you just announced. Ultimately, will you just be choosing one of the developers or are you planning on helping anyone who presents an idea that could be useful?
The details are yet to be shared and determined. Right now thousands of these health and wellness apps are out there. Many of them purport to try to help address the risk factors of diabetes or help manage your diabetes. I would suggest that not a lot of science is behind some of these apps. So imagine this opportunity to be able to connect to the science and expertise from the ADA and to the power of Watson and cognitive computing—two pieces that the winner or winners would get access to. While the details have still not been announced, I know I will be a judge, and personally I like Shark Tank, so I would like to call this Smart Tank. We are looking for entrepreneurs who are doing extraordinary things that are transformational and then we will do our best to help accelerate those ideas and scale and spread them to the people who need them.
You are also working on a diabetes adviser yourselves for providers, patients, and caregivers. Will Watson be able to make personalized recommendations based on different factors for each individual patient?
Here is what I would say. One of the essential factors that determine health outcomes is definitely healthcare factors. However, we need to recognize that the science tells us there are other key factors that impact health outcomes: Social factors, environmental factors, behavioral factors, demographic factors, and more. So to me, personalization leverages all of these factors to promote and nudge the best, unique advice to the person with diabetes. In addition, I would call this diabetes adviser, which is built on a cognitive system like Watson, dynamic. It continues to understand, reason, and learn. So the personalization you might get today, based on new knowledge and insights, may be different than what you would get tomorrow.
You mentioned that you have been in healthcare for a while. Besides diabetes, what other areas are you currently focused on and what plans do you have for the future?
Back in February, we announced the partnership with the American Heart Association to address the prevention of heart disease and stroke. Then in April, it was the American Cancer Society to create an adviser for people with cancer and their caregivers. We believe it is essential that we partner with professional groups and societies that represent both the providers and the patients who they serve. Our partnerships with the AHA, ACS, and ADA represent this commitment. Our goal is to be a catalyst to transform health and healthcare and, similar to the diverse, transdisciplinary healthcare team that delivers care one patient at a time, this will require a diverse, trans-sectoral team that delivers care one population at a time. So we are excited about the partners we have and there will be more to come because it does require an ongoing team effort and a broad ecosystem of partners.
Most of our readers work at pharmaceutical companies and med device companies. You already mentioned your partnership with some of them, such as Medtronic. So for companies that haven’t already partnered with you yet, what do they need to know if they want to work with Watson Health?
We are very proud of our collaborations with Medtronic and Novo Nordisk for diabetes care and prevention. For companies that we haven’t partnered with, I would ask them to consider what data they have and have they gotten the best insights they can from that data. At IBM, we view data as a natural resource, but it has to be cultivated and there is an opportunity to bring visibility or insights to key health conditions through that data. This partnership with the American Diabetes Association is addressing what the ADA President referenced as an “invisible” disease and epidemic. So we are excited to partner with the ADA to bring “visibility” to this invisible condition and urgent epidemic. We know that we can help the other partners who are interested in transforming diabetes care and prevention. If you have that common mission, we are already interested. If you have an interest in bringing insights out of data through the power of cognitive computing and Watson, we are very interested in partnering with you.
