Using Machine Learning to Increase Patient Engagement | Bob Gold | Gold Group Enterprises

GoMoHealth_logo_website_header.png

Podcast:


Full Conversation:


“Patient Engagement” is one of the biggest themes, and highest goals, of today’s healthcare industry.

If we can “engage patients” we can reduce costs by helping them be more adherent and avoid negative outcomes and readmissions. If we “engage patients” we can make the healthcare experience a better one for them and their support network. If we “engage patients…” we can help them avoid becoming patients at all through surveillance and prevention.

It’s all good and necessary. And if it was easy, we’d already have it nailed down.

The problem is that “patient engagement” is a complex process that includes reaching people in the first place, getting their attention, and then - and probably most difficult - getting them to take action and likely make a sustained change in behavior. Again, if this was easy...

Bob Gold loves the problem of behavior change. Gold, CEO of Gold Group Enterprises, is a behavioral technologist who focuses on applying his work to the healthcare industry. One of his subsidiary companies, GoMo Health, is located in Asbury Park, New Jersey. Gold says, “we're about 40 people, and we're working on helping change healthcare patient engagement and care coordination. Not only in the U.S. but we're deployed around the world. For example, we're doing a prenatal program for women in Kenya, and we're about to start some programs in other countries in Africa and Central Asia."

GoMo Health’s primary product is a concierge care system built on a proprietary system called BehaviorRx. The system “interacts with people in their daily lives and asks them questions and responds to them.” Gold says it’s like Siri or Alexa “but it’s very specific to your healthcare and your condition, your lifestyle, your behaviors, your ability to cognitively understand, and it’s able to engage you. So it’s sort of like ‘nurse meets Ritz Carlton concierge’” and helps you make good decisions throughout the day.


The cost to the system

Although Gold talks about a number of different principles that underlie Concierge Care, the fundamental one is “resiliency.” Which roughly means adherence and sustained behavior change. Gold notes two common examples of situations where there’s low resiliency: taking medications (“Seventy percent of people in the United States are not adherent to their medication”) and going to the gym (we don’t need to quote him here, do we? You know…)

It matters because lack of adherence is a massive burden on the healthcare system. Gold pegged non-adherence at 70%, a 2014 review put it at 25-50%. That same review quoted an estimate that 20-30% of healthcare dollars are wasted - which includes but is not limited to medication non-adherence. In 2012, a review in Annals of Internal Medicine stated that non-adherence is estimated to cause approximately 125 000 deaths, at least 10% of hospitalizations, and a substantial increase in morbidity and mortality. Nonadherence has been estimated to cost the U.S. health care system between $100 billion and $289 billion annually.” (Emphasis added)

So just looking at the single issue, it’s clear that we have a massive problem. Gold notes that we’ve got maybe 20 years worth of Medicare and Medicaid money left. Burning what little we have on things that don’t work isn’t a viable option.


It's hard to beat human wiring

Why is non-adherence such a problem? Human psychology, of course. Historically, medical staff will provide a relatively standardized set of discharge teachings to, say, a post-surgical patient. “Take these meds this often, shower but don’t take baths, call us if you feel this.” Sometimes it works, often it doesn’t.

For example, we have a family member who works on a cardiovascular surgery unit. One of her responsibilities is discharge teaching, which includes that line about not sitting in baths or hot tubs. A few years ago, she told the story of a patient who returned to the hospital not long after open-heart surgery with a badly infected incision (we’ll spare you the details). It didn’t take long to figure out that the patient had enjoyed a warm bath. “Why did you take a bath?” “I thought it would feel good. I thought it would help.”

Clearly there was a disconnect. And Gold can explain exactly what it was:

"Your brain goes through a process when you're trying to absorb information. One of the reasons why healthcare has such problems in transitions of care […] from a patient engagement standpoint, or even clinical engagement, is a lot of information never makes it into long term memory. So the patient doesn't forget. They never remembered.
"So for example if you come out of the hospital or skilled nursing facility on something complex there's typically a discharge protocol where the nurse, social worker, or relevant clinician sits with you and maybe a caregiver, and they review your discharge procedures and what they expect you to do.
"Typically at the time you’re so overwhelmed, stress blocks some of the neurotransmitters from storing things. So the clinician may feel you’re listening and you're trying to listen but it really never makes it into memory.
"And then later when you don't do something the team gets frustrated say hey why didn't they do that, I clearly told them! And the reason is you can't do it because it’s not in your memory.”

So memory is one thing. Another is simply communicating and interacting with people in a way that matches the way they’re wired. Consider two patients leaving the hospital with the same condition and needing to follow the same post-discharge care plan:

“If you self-assess on a scale of one to ten…I just got out of the hospital for an operation for congestive heart failure and I self assess a one or two, where ten means I can follow a care plan, and somebody else assesses an eight, it's ridiculous to communicate to those two people in the same way, with the same voice, the same frequency, ask them the same questions. Because one person is telling you they want to help every step of the way and the other person is saying I think I got it.”

Put very simply, we’re all different, so the way our care needs to be delivered should be, too.

As always around here, we want to be careful not to sound too critical of the professionals doing incredible work to help patients. This is a design problem, not a people problem. Care teams may be able to adjust their interaction with patients to some extent (presenting things differently to a patient who is cognitively impaired, for example), but creating a detailed, customized plan for each patient simply isn’t feasible.


Beating the system with personalization & technology

This is where Gold, and many others, are making a difference. Key to behavior change in general is personalization. In an earlier podcast David Vivero of Amino called it “meeting people where they are.” And no one reading this is any stranger to the push towards personalization in healthcare.

So Gold and the team at GoMo Health (as well as so many others in different areas of healthcare) are using behavioral science and burgeoning new technologies like machine learning to offer the personalization that we need but don’t have the resources to do manually. With the machine learning component, BehaviorRx is able to ask questions and adapt over time, fitting into the individual’s lifestyle and learning style. Traditional protocols “typically don’t operate how people behave cognitively and how they live their day.

“So [it’s] personalization, and then it's dealing not just with their healthcare in the engagement protocol, but their life related to it.”

Darren Hay of Care Angel discussed this last year with us. It’s using powerful technology on the back end to simplify and personalize things on the front end:

“You don't need a smartphone or tablet and you just basically need the voice and that allows you to interact and give data and give feedback on how you're doing - how the patient is doing - to really allow you to start taking action on their healthcare in real time.”

And maybe that’s the most exciting thing, that we are now at a point technologically where all of this is possible. Sure, our system is collapsing under its own weight, but recent advances have put us in a position to make significant and rapid improvements. One patient with a positive outcome thanks to a personalized plan and intervention is good, for sure - probably enough to make it worth it. But we have the chance to do way better than that. We can change - fix - the entire system by working with instead of against human nature.