Health technology brings care plans alive

by Andrew Oram

This was originally published on O’Reilly Media’s Strata blog, February 6, 2014.

Challenges and hackathons are meant to surprise you. If the winner is a known leader in the field with lists of familiar credentials festooning the team’s resumes, there was no point to starting the challenge in the first place.

Pharmaceutical company Merck and the Heritage Provider Network, the largest physician-led health network in the US, were looking for something new when they launched their challenge on diabetes and health disease. These conditions are virtual epidemics, world-wide.

The incidence of diabetes in New York’s rich Upper West Side is only a few percentage points, whereas it is five times as great in Harlem a few blocks away, according to Mark Wagar, President of Heritage Medical Systems (a subsidiary of HPN). But the important statistic is that the incidence of diabetes is growing in both neighborhoods.

Faced with chronic diseases like these, doctors are great at drawing up care plans. The patients—not so good at following them. Only half of the people with diabetes or health disease [AO: I missed the details—is it these particular conditions or all chronic conditions] follow through on their care plans. Something new has got to be tried fast. Basic medical tasks such as taking prescribed medication and coming to follow-up visits are part of the problem, but exercise, healthy eating, and stress reduction all enter the care plan as well.

Five judges followed a grueling scheduling, evaluating 90 contestants against a list of criteria. After a live demo day in late January, the judges chose two finalists: Wellframe and Sense Health. Each uses deceptively simple technologies and techniques to address the care plan issue.

Starting this month, the two semi-finalists will work with Desert Oasis Healthcare, California, a member of HPN, to refine their platforms. The winner will be announced in May.

Wellframe makes the clinician-patient connection a matter of lifestyle

Dealing with a chronic condition such as diabetes, heart failure, or mental illness is not a series of discrete treatments but a whole lifestyle. Although clinicians have tried to educate people about the necessary changes they need to make in all the little things they do each day, they are frustrated by the traditional model of seeing patients once every few months—or at best three times a week for a few months, as hospitals ask patients to do when recovering from a heart attack.

Wellframe has a twenty-first century solution to this problem that’s technologically straightforward but includes a number of subtleties. They have just gone commercial. I talked to the leader of the Wellframe team, Harvard epidemiologist Jacob Sattelmair, about how Wellframe developed experimentally and what it offers the patient and her doctor.

Wellframe presents the patient with their care plan on their mobile device. It runs on both phones and tablets powered by iOS or Android. But unlike the care plan traditionally handed to you on a piece of paper when you’re discharged from a hospital, this plan reminds you to take your meds, asks how you’re feeling, and lets you communicate with a nurse or other clinician back at your treatment center. The app lets you self-report back to the clinician each time you take medication or perform another task.

Sattelmair compares a traditional care plan to a set of driving directions, and Wellframe to a GPS. It breaks down the complex plan into small steps and presents them to the patient at the moment she is expected to do them. Also like a GPS, Wellframe can adjust the plan dynamically. Sattelmair mentions, as an example, that one nurse found out her patient was attending a big family gathering every Wednesday where he was presented with temptingly heart-unfriendly foods. She instructed his Wellframe app to message him every Wednesday afternoon with reminders about what he could healthily eat.

As the company enhances Wellframe, it will be able to adjust more and more flexibly as the clinician notices changes in the patient’s behavior, like greater or lesser adherence to taking medication. It will help the patient and clinician transition from an intense early intervention during the early phase of treatment to a more relaxed, lifestyle type of collaboration.

Data between patient and clinic is transmitted in an encrypted form, and the whole platform is HIPAA-compliant.

On the clinician’s side, a dashboard lets the nurse see what the patient is doing in real-time. Not only does this permit feedback and encouragement on a day-to-day basis, but the data collected gives the clinicians a much fuller picture of the patient when she does return for a follow-up visit. Instead of wasting time asking the patient what she has been doing for the past few months (questions she may or may not answer accurately), the doctor can take all that data in hand and launch immediately into a deeper, more meaningful discussion.

To me, Wellframe as an exemplar of technology that enhances the doctor/patient relationship rather than (like all too much technology today) getting in the way. Sattelmair calls it "high touch." I see it working on three levels: simple, timely reminders, higher-level communications between patient and clinician, and changing care plans in response to real-life needs.

But for clinicians, Wellframe does even more. It can let clinicians know the effects of their interventions over time (Is the patient exercising properly? Is he reducing his stress?) and let them focus in on patients that need more or different interventions.

Wellframe also contains educational programs that a doctor can deliver to patients when appropriate. The app can follow up on the program by asking questions that reveal how well the patient understood the material. Depression screening is also available, because depression is a critical factor in many physical illnesses, with strong impacts on the patient’s ability to recover.

Wellframe also does large-scale data analysis, like many apps. It uses de-identified patient data to detect trends that may be useful in population health management (trust an epidemiologist to think that up).

Having learned a lot through past experience about the challenges inherent to engaging patients with technology, Sattelmair came to this project very alert to the need for feedback from patients and clinicians. Numerous mock-ups and prototypes of Wellframe were tested. It was deployed experimentally in three areas:

Sattelmair says, "It’s no longer OK to treat patients in an episodic way." Wellframe is currently available in English and Spanish, with plans to deploy it in other languages as well.

Sense Health puts algorithms to work maintaining a care manager/patient relationship

We all know the patients who need the most medical care (besides the declining elderly, a category all their own): patients with multiple illnesses, who have developed problems such as obesity and high blood pressure over many, many years, and who often don’t seem to harbor as much motivation to get better as their health providers or public health advocates do. Patients like the hot spotters famously treated by Jeffrey Brenner and reported in an article by Atul Gawande.

Sense Health is handling these patients through one of the most rudimentary communication technologies we have: text messaging. They partner with care managers, who typically deal with 50 to 100 Medicaid recipients at a time and can’t afford to check in with each one every day. So care managers enter data about each patient into Sense Health—a quick task listing the major medical conditions, the patient’s level of education and motivation, and forth—and use it to manage an ongoing, consistent relationship.

The sophistication comes with choosing the right texts and arranging them in an effective sequence that motivates and encourages the patient without alienating him. Sense Health combines the patient-specific information with an algorithm that assesses the patient’s knowledge and readiness to take the reins for his own health. This algorithm is similar to the patient-activation-measure developed by Insignia Health. Sense Health chooses a unique sequence of messages in four categories:

Founder Stan Berkow told me the system contains more than one thousand messages of these types, written by health and wellness professionals. Care managers are welcome to customize them. A care manager may insert the patient’s name into messages, or add some personal note that they know the patient responds to positively. Sense Health is working on addressing cultural and language differences.

The system must also recognize when the patient reports a condition such as pain that requires quick intervention, and alert the care manager. Sense Health monitors the patients’ responses to text messages—or lack of response—and uses AI techniques based on brain research to mine data over time in a bid to improve its algorithms. AI is useful to Sense Health in two areas: improving the support plans to adapt the message streams to different types of people, and determining when the health provider should be brought back into the conversation.

The developers have found the patients response to be gratifying. Even though only one-quarter of the messages contain direct questions, patients in a clinical trial responded to 75% of the messages. Sense Health summarizes their status and sends an email message to the care manager each morning. Patients are warned that they can’t expect an immediate response to a message, even though it is casual and looks like a message sent by a person.

SMS text messaging is not secure, but is the best technology to ensure that most Medicaid recipients can receive the communications and the medium most familiar to a large range of people. Because text messages cannot be encrypted, Sense Health ensures that messages sent out contain no protected health information. It even checks messages sent by care managers and warns them if the messages contain key words such as disease names. The back end is HIPAA-compliant.

Author’s home page
Other articles in chronological order
Index to other articles