fb-pixel

SAN FRANCISCO — Talk to investors about why they like Omada Health, and they point you first to its technology. The startup, they rave, deploys sleek devices to monitor patients with chronic medical conditions; it collects millions of data points on their behavior; and it uses machine learning and other artificial intelligence techniques to automate health reminders and to flag individuals who need more intervention.

It’s a robust operation. It would also fall apart without the likes of DJ Moberly, a “health coach” who spends much of her day exchanging messages with Omada’s diabetes patients from her home in the Houston area — helping them set and stick to goals, encourage them to eat healthy, and offering them tips about controlling their blood sugar levels.

Advertisement



Digital health companies like Omada are now booming businesses. Although their models vary, they all monitor and educate people with chronic medical conditions such as diabetes, hypertension, obesity, and depression — and make money by charging their employers or health plans, with the promise of ultimately lowering medical costs. In just the past few months, leading companies in the sector have gone public, been acquired by Google, and raised tens of millions of dollars of venture capital funding.

Relatively underlooked, however, is just how reliant the sector is on human labor — namely, armies of health coaches like Moberly working from home offices hundreds or thousands of miles from corporate headquarters in Silicon Valley.

It’s a job that points to the limitations of the automation, AI algorithms, and sophisticated monitoring devices that have helped companies in the space achieve lofty valuations. Cutting-edge technology may impress investors and enterprise customers, but it’s the health coaches — many of them women, many of them moms with small children, and many of them living in the heartland — who power the operation.

Advertisement



“Despite the fact that I’m one of the biggest tech lovers out there, it’s hard to recreate human accountability,” said Omada CEO Sean Duffy.

That need for human contact is resonating with the growing number of employers being persuaded to pay health-tech companies to support their employees. Many of those employees have chronic conditions that are being poorly managed in the traditional health system, said Steven Wardell, a longtime health-tech analyst who recently opened his own consulting firm.

Those employers are “looking for something that shakes it up, because the current system’s not working,” Wardell said. “The most obvious, most powerful thing is the human coach.”

Health coaches bring wide expertise to the job. Some are registered dieticians, and others are certified diabetes educators.

Digital health companies are hiring these professionals away from jobs at hospitals, community health clinics, and — increasingly, as the sector grows — rival health-tech firms. Most of the applicants are women. While Omada wants more women to apply for its engineering jobs, it has the opposite challenge with health coaches: too few male applicants, Duffy said.

Some companies, including Omada and Livongo, another company focused on helping people manage their chronic conditions, hire their coaches as direct employees, just like their engineers and product managers. Other companies build their network of coaches differently. Fitbit’s health business, soon to be absorbed into Alphabet, relies on a mix of direct employees, contractors, and coaches who work for a third-party staffing service, sometimes in more traditional offices.

Advertisement



Leading companies in the sector count their health coaches in the multiple dozens, though none of them would disclose exactly how many are in their network, saying they consider that number to be a trade secret that gives them an advantage over the competition.

However many there are, they stay busy.

An Omada coach might work with roughly 300 to 400 participants at any given time, though that number can rise or fall significantly, depending on the participants’ conditions and their engagement levels, Duffy said. Coaches use digital dashboards built for coaches that track information about their patients.

At Fitbit, coaches who work with people with chronic conditions might have 300 people on their panel, while those who focus on wellness and prevention can work with more, said Amy McDonough, who leads Fitbit’s health business.

Livongo boasts a ratio of one coach for every several thousand of the patients that it refers to as members, according to Dr. Bimal Shah, the company’s chief medical officer. But that number reflects the many people who manage their conditions using only Livongo’s smart devices — which include a blood glucose meter, a blood pressure cuff, and a scale — and automated feedback. The number of people who regularly interact with a Livongo coach over time is smaller. On average, a Livongo coach might talk to eight or nine members in a given day, Ebeling estimated.

Coaches are matched to work with specific patients so that they can develop rapport over time, and those assignments are often based loosely on geography. Sometimes, it’s for practicality of being in the same or adjacent time zones, and other times, it’s for regulatory reasons: Illinois, for example, requires coaches to be physically based in the state to work with residents there, Shah said.

Advertisement



Moberly, the Omada health coach in Houston, said she loves her job in part because it lets her “help and engage with a larger number of people than what I ever could within a human-interaction in-person clinical space.” She recently chimed into a group discussion about healthy eating, suggesting healthier ways to cook okra to participants who were only familiar with frying the vegetable. Try cooking it with a bit of oil and sea salt on a nonstick skillet, she suggested, or grill it, or combine it with tomatoes.

For coaches, there’s also the matter of convenience.

Consider Myra Ebeling, who used to spend 90 minutes to three hours in the car each day, commuting to and from the clinics where she worked with diabetes patients in the San Diego area. She loved the work, but with a baby and a toddler at home, all that driving was getting exhausting.

Four years ago, Ebeling took a job at Livongo, which offered her the opportunity to work from home, coaching diabetes patients all over the country. The role came with advancement opportunities that brought better pay, along with equity in the Silicon Valley company, which went public over the summer in one of this year’s most closely watched initial public offerings.

Advertisement



At first, Ebeling worried about how to connect with her patients, since she’d no longer be sitting a few feet away from them in an office. But she quickly took to the medium. “There are no preconceptions,” she said. “People can really open up over the phone.”

For companies monitoring chronic diseases, the health-coaches model can come with challenges.

“You’re talking about having a potentially large and rapidly growing and virtual workforce,” said Andrew Matzin, a partner at the consulting firm Health Advances. “In any industry, if that’s the situation you’re in, you may run into challenges of sustaining high-quality levels, sustaining the training experience, keeping retention high as you scale, avoiding the temptation to lower the bar in recruiting in order to meet aggressive hiring goals.”

Health coaches aren’t practicing medicine, and what they do happens outside of the traditional health system. While coaches do have a considerable amount of data on patients’ vitals, behaviors, and their medications, their systems generally aren’t linked up with patients’ electronic health records maintained by their primary care provider, creating the potential for inconsistencies in care, Wardell said.

As companies in the space develop ever-more sophisticated algorithms, they’re making decisions about what to automate — and what needs a human touch from coaches.

At Omada, the company ran a test to see if patients were more likely to heed reminders to weigh in on a smart scale if the suggestion came typed out by a real human coach — or via automated stock language. There wasn’t a difference, so the company decided to delegate such alerts to the bots.

Omada also uses the AI technique known as natural language processing to tell coaches drafting a message to a participant whether they’ve used similar wording in previous messages, so that they can avoid sounding too robotic.

At Livongo, the company uses machine-learning algorithms to provide patients with tips and insights at moments when they’re actively engaging with the service, such as when they’re testing their blood glucose levels or weighing in.

“Did you know that 17 grapes equals one carb choice?” reads one message on the screen of a blood glucose meter. “Did you know that you can have fun while exercising? Plan active outings with family and friends,” says another.

“It just makes it more ubiquitous, more personalized, more scalable when those messages are delivered at the opportune time digitally — as opposed to forcing a human interaction when somebody’s quickly just trying to get through their day,” Shah said.

Human interactions, by contrast, are reserved for when Livongo’s members really need human support.

Take William McLeod, a 66-year-old retiree in New Jersey. He was diagnosed with type 2 diabetes two decades ago. It was not until 2017 that he first worked with a health coach, when his health plan, Horizon Blue Cross Blue Shield of New Jersey, signed up as a Livongo customer.

Just about every month since, McLeod and his coach have held 30-minute check-in phone calls. She’s helped him set goals for weight loss and develop strategies for what to do when his blood glucose readings come in too high or too low.

McLeod said his coach has become a friend. And since he started working with her, his A1C level, a measure of glucose in the blood, has dropped steadily.

“I’m right where I’d like to be,” he said.


Rebecca Robbins
can be reached at rebecca.robbins @statnews.com. Follow her on Twitter @RebeccaDRobbins. Follow Stat on Twitter @statnews.