LeadingAge Magazine · July-August 2018 • Volume 08 • Number 04

Cognitive Computing and the Social Determinants of Care

July 18, 2018 | by Gene Mitchell

A conversation with Judy Murphy, R.N., chief nursing officer at IBM Global Healthcare, about the use of data to improve health care.

What is cognitive computing, and why will it be important to health care in general, and aging services in particular, in years to come?

At this year’s Long Term & Post-Acute Care (LTPAC) Health IT Summit, held in June in Washington, DC, that question was discussed by one of the keynote speakers: Judy Murphy, R.N., chief nursing officer at IBM Global Healthcare.

Judy Murphy, R.N.
Judy Murphy, R.N.

Before moving to IBM, Murphy was chief nursing officer and deputy national coordinator for programs and policy at the Office of the National Coordinator for Health IT (ONC) in Washington, DC. Her time at ONC came on the heels of more than 25 years of health informatics experience at Aurora Health Care in Wisconsin, an integrated delivery network with 15 hospitals, 120 ambulatory centers, and over 30,000 employees, where she was vice president-EHR applications.

LeadingAge interviewed Murphy after the LTPAC conference, to build on her presentation there.

LeadingAge: You describe today’s health care environment as a movement from a “transactional, fee-for-service system” toward value-based care. Can you explain why that transition is so important?

Judy Murphy: The transition is important for both health and health care reasons. The financial model we have right now is unsustainable; our costs keep going up and up. We have to be thinking about both health promotion and preventive care: preventing people from having chronic diseases or at least discovering it as early as we can, preventing them from falling and getting hip fractures, and doing all the other things that must be done to have a healthier population. So the focus shifts from “treating the people who you see that have conditions,” to “treating the people that you don’t see to prevent them from needing care.”

That’s why it’s so important for people to be more engaged in their own care, because they’re not consistently attached to a health care organization throughout their lives.

LeadingAge: You talked about the idea that we look at health care as something outside of us, and that you believe it should be more inside of us—a part of our regular lives.

Judy Murphy: That’s actually an idea I evolved during a talk earlier this year. I was thinking, “What was so different about the time before now?” The big difference was that we’ve grown up thinking about health care as something outside of ourselves. Our doctor worries about our blood pressure, so “It’s his problem.” We need to see our own health as part of our everyday lives, whether that’s nutrition, or exercise, or taking care of a chronic condition, or monitoring for the potential onset of a chronic condition, or various preventive measures. All of those things are in our control as individuals. It’s important to see them not as our doctor’s problems but as our own problems.

LeadingAge: How do you tie that concept to older adults and the services they need?

Judy Murphy: They are our best example, right? They really grew up thinking that the doctor knew it all, and that the doctor would take care of me, and this plays itself out over and over. For example, take my mother. She’s 92 and began having inadequate pain relief, and immediately wanted to go see the doctor. I told her, “The doctor isn’t going to be able to tell you anything we don’t already know. This is chronic pain from your arthritis and diabetic neuropathy. We need to think about how you can change your activity or how you’re taking your pain meds in order to get pain relief. This isn’t something magic that the doctor is going to be able to take care of.”

You can imagine having someone older go to an ER, where the likelihood of a person’s medication regime being altered is high, where the likelihood of something untoward happening is high, and yet that’s the first thing older adults think about. So a culture change is exactly what we need here. There is a time and place for care from a doctor, but it is not an end-all for everything that goes on in our lives.

LeadingAge: Where is the responsibility for getting the message across? Does it fall on health care providers only, or someone else?

Judy Murphy: It falls on all of us. That’s part of our dilemma now, because all of us are doing it—which, essentially, means that in many cases we get duplicative messages and therapies. We see examples where both the provider and the insurer send people to the individual’s home to check on them post-hospitalization, and Medicare does it now routinely as well.

So that’s an example of us doing it right, but we’re not organized enough in our clinically integrated networks, or ACOs, or partnerships to not duplicate services a little. But in many cases, these home visits are probably preventing a whole lot of untoward effects from occurring, so it’s probably still worth it.

LeadingAge: You discussed “digital reinvention” as developing in 3 phases, the last of which includes population health management, cognitive computing, personalization and precision medicine. I was struck by a remark you made to the effect that we need to do things differently, and that does not mean just automating all the old things we used to do. Can you offer an example or two of that?

Judy Murphy: Electronic health records (EHRs) are probably the best example. When we implemented them, we pretty much just automated the same processes that we had. The first step was just getting the EHR into place, but the second step was actually harder, because that’s where we wanted to use the EHR to transform health care and do things differently. Many refer to that as the “EHR optimization” phase.

A good example to demonstrate this reinvention is Uber. Nobody sat around and said, “We’ll build this to change transportation.” It was only possible to create Uber because of the technology; the technology allowed a new way for people to be transported from point A to B. As it applies to health care, the biggest examples I can think of are the eventual ways we will personalize medication prescribing. Today, whatever your condition is, we’re pretty much using evidence-based medicine to determine what therapy to use for an individual. Once we have genomics and a profile for each individual, and have done the research to learn which medications work on which types of profiles, or with which genetic markers, we can use that to determine how best to treat individuals as compared to using the typical evidence-based approach, which was looking at the overall population and pretty much treating everyone the same. Creating profiles for individuals that allow us to be more specific about what drugs will or won’t work for individuals is like turning the industry on its head. You’re starting with the genome rather than starting with the medication.

I think we’ll learn a lot with our analysis of the huge amounts of data we’re collecting through our EHRs, and it’s also important to get at some data outside the EHR, such as the social determinants of health. What really makes an impact? That gets us back to cognitive computing that can inform us: How does diabetes develop, and can we stop that? How does high blood pressure develop and how can we stop it? And maybe more importantly, we can tailor the incentives to individuals that will cause them to use preventive measures. What motivates you to check your BP every day? What motivates you to have a colonoscopy, because we’re not all the same? For me, that’s another example of turning things on their [head], because we start with the individual and work back to the treatment, rather than using the traditional treatment.

LeadingAge: My 2 favorite slides from your presentation (see figures 1 and 2 below), concerned “What makes us healthy vs. what we spend on being healthy,” and the risk stratifications and relative costs of health care for different segments of the American population. It struck me that LeadingAge members—who work with older adults every day because in most cases they are housing them as well—might have a better grasp of the social determinants of care than most other people in the health care field. What do you think?

What Makes Healthy
Figure 1.

Judy Murphy: Yes, I think that’s true. As an example, my mother moved into a community less than a year ago. She was living alone in an apartment and had lots of different issues, and was depressed, and was taking a lot of pain meds. We finally talked her into moving into a community. For the last 5 years, I had been telling her there were 3 safety issues: her medications, her eating, and her depression/lack of social interaction. We solved the eating issue with Meals on Wheels, and solved the medication safety because my sister sets up her meds each week. But we never really got to the social interactions part, and the depression part.

The bottom line is, we moved her into a community and to see the change in her was unbelievable. She loves playing cards; now she plays every day. She’s got lots of friends, and she takes one meal a day and spends time in the common areas where people go and play games and such. She stopped taking her pain meds within a month or 2, and hasn’t taken a dose since. It’s absolutely amazing to see the difference not only in her medication, but in her perception about the pain she was having, and how that was interacting with the depression she was in.

Not everyone will have the same outcomes my mom did, but that relationship of the behavioral to the physical is so strong, that these communities are just doing wonders. Again, it’s that they pay attention to all aspects of residents’ lives, the social-behavioral as well as the physical. We can’t undermine the importance of that. It goes right back to population health management, right back to this idea that it’s about the way we live, not just the specific care we get.

LeadingAge: Before we talk about cognitive computing, can you tell us what do you do at IBM?

Judy Murphy: I am a non-billable resource. I can be a consultant during the selling phase, during the implementation phase, and during the post-implementation phase, for any of the products and services that we sell. Because of my depth and breadth of expertise in health IT, I help bridge the gap between the solutions we have and the clients we serve.

I also work internally with the people at IBM. A lot of them do not have deep health care experience or expertise. So I spend time developing programs and doing training for the people doing the selling and implementation of our health care solutions.

Pop. Health Mgmt.
Figure 2.


LeadingAge: In your speech you talked about genomic and exogenous data that is needed to really understand human health, and to allow cognitive computing to work. Can you briefly define those terms, and explain how we will get to the point of collecting those types of data?

Judy Murphy: This will be a journey and we won’t get there overnight. As I talk with individuals around the country, they are pulling from different kinds of sources: weather data, data from social media and data from government databases. For example, there might be demographic information from certain ZIP codes. In other countries, things like access to water is really important. Access to any kind of fresh vegetables is important, and that shows up here: Do you have access to that or is the only place in your ZIP code a 7-11 convenience store? So people don’t totally understand what’s going to make a difference—I don’t totally understand either—but we have to start pulling that data in, and using cognitive computing to help us learn what does make the difference. If you go 2 ZIP codes away, and your life expectancy goes down by 3 years, we need to understand why. We need to understand the social determinants of health and the exogenous data that will show a causal relationship between people in one ZIP code and another.

Certainly, census data gives us a sense of what the cost of living is in the area, and the average salary, and who’s unemployed. All that data can be consumed, and we can begin to look at some of the correlations and use them to try to determine causation. I’m talking about using cognitive computing to understand better, from a population health management standpoint, what exogenous data and social determinants make a difference. This is an important role for cognitive computing on our big data, that will start to open up some insights on population health.

LeadingAge: I’m going to quote you to you: One of your slides, “Care Management: The Future,” says, “Everyone has a cognitive care plan with clinical decision support and analytics that help determine the most effective mix of high-touch, low-touch and remote interventions.” Would you care to speculate on what part of the population will first reach that point?

Judy Murphy: If you go back to that slide that you liked, on the “Population Health Management Model,” you can see that the engagement strategies listed, from the healthy/low-risk group at 40-60% of the population, who are low- or very low-touch, to the 2-3% of the population with active disease who are high-touch. This is the balance when we talk about population health: Where do we invest our money and how do we take our strategies to scale?

The idea is to shift a greater percentage to the low-touch and healthy low-risk segment, where we’ve motivated them enough that all they need are little pings of reminders about things, and don’t need the heavy hand of care coordination. But the people who are in the chronic disease category, creeping into active disease, need high-touch and possibly even daily care coordination. So cognitive computing will help us identify not just which people fall into each category, but the proper mix of our dollars in health care to spend on each segment of the population, so we’re not overspending on the healthy and underspending on the person that might get high blood pressure.

Gene Mitchell is editor of LeadingAge magazine.