Artificial intelligence can play a key role in diabetes management programs by enabling more personalized health "nudges" for members, which drives higher engagement and better health outcomes, according to new data released by Teladoc Health.
The telehealth giant presented the results of two studies at the American Diabetes Association’s 84th Scientific Sessions, which began Friday in Orlando, Florida.
Both studies evaluated the role of Teladoc's predictive modeling capabilities to help members with type 2 diabetes control their blood sugar.
One study looked at the impact of personalized health nudges on clinical outcomes. Teladoc examined the effectiveness of the AI-powered health nudges, in the form of notifications sent to mobile or cellular-connected devices, to help improve members' self-monitoring of type 2 diabetes, particularly for those individuals that were previously identified as being at-risk for uncontrolled outcomes in the future.
Data from the study conducted over nine months shows a clear connection between program engagement and improved clinical outcomes, with a 3X increase in engagement leading to an additional 0.4 reduction in A1c (8.2 to 7.8) for members targeted with personalized health nudges, according to Teladoc.
As the excitement about artificial intelligence in healthcare continues to build, Teladoc, along with most healthcare organizations and health tech companies, is exploring how AI and machine learning can help to deliver better care, said Tejaswi Kompala, M.D., an endocrinologist and head of cardiometabolic clinical strategy at Teladoc.
As part of its virtual care services, Teladoc offers chronic condition management programs for diabetes management, hypertension and diabetes prevention and weight management. The company currently has 1.1 million members in its chronic care management programs, and that's up 9% from a year ago, Teladoc executives said during the company's first-quarter earnings call in April.
As of the end of 2023, diabetes management comprised approximately half of Teladoc's chronic care program enrollment, chief financial officer Mala Murthy said during the company's fourth-quarter earnings call in February.
"With 1 million members who are using our programs, there has to be technology and digital to deliver care to populations at scale," Kompala said in an exclusive interview about the new data. "Of course, so much of our program continues to focus on the human touch with member interactions with our health coaches. We're never trying to replace the relationship someone has with their existing healthcare team—and yet, we know that digital interventions like ours can help to augment that care."
"What's exciting about machine learning models is that it's just helping us be smarter at how we should engage with members in our program and turn those touchpoints into being more successful ones that are likely to lead to ongoing engagement," she said.
The results come on the heels of previous research that demonstrated Teladoc Health’s ability to proactively identify a person at risk for uncontrolled outcomes more than a year in advance using AI.
The ability of AI to forecast this is crucial for the early detection and management of diabetes; it allows for more timely, personalized interventions to avoid complications and improve outcomes, according to Teladoc executives. Improved clinical outcomes also help employers and health plans better control costs related to chronic conditions.
“Teladoc Health has a long-standing history of successfully using data to improve health outcomes for our members, and new applications of AI are helping us accelerate our impact,” Sal Shafiq, chief data and analytics officer at Teladoc Health, said in a statement.
Teladoc has been in the telehealth market for 20 years, historically offering virtual urgent care and behavioral health solutions. The company moved deeper into digital chronic condition management with its blockbuster $18.5 billion acquisition of Livongo in October 2020. That deal added Livongo's data science-driven chronic disease management solutions and remote patient monitoring services.
With the acquisition, Teladoc aimed to combine Livongo's expertise in chronic disease management with its telehealth capabilities and offer those solutions all in one place. At the time of the acquisition, then-CEO Jason Gorevic said the combination of Livongo and Teladoc would enable "personalized, technology-enabled longitudinal care" that spans primary care to chronic condition management.
Teladoc Health’s diabetes management program provides members with education and tools to self-manage their diabetes through mobile technology. The program offers participants a cellular-enabled, 2-way messaging device that measures blood glucose (BG) levels and delivers personalized insights into their glycemic management as well as real-time support from diabetes response specialists and access to certified diabetes care and education specialists. Members in the program also have access to a web-based app and mobile phone app to monitor blood glucose levels while also providing real-time support and access to an asynchronous chat with coaches.
Teladoc has been able to leverage "billions" of data points from its members to build predictive models to help identify individuals at risk of uncontrolled diabetes and then provide more timely intervention, Kompala noted.
"We're looking at factors like demographics, we're looking at how they engage with our program, both with the use of our app and with the use of our various devices. And, then, of course, we're looking at some of the device data itself. How are my blood glucose levels trending over time? With that combination of information, along with some other information related to the type of medications that a person is using, we were able to develop these models," she said.
In another study presented at the ADA's conference, Teladoc data showed that diabetes members who received personalized next-best actions in their weekly email were 50% more likely to engage with a health coach.
Previous research indicates the use of dedicated coaching can improve diabetes control. In this latest study, Teladoc evaluated how tailored email content can be used to increase 1:1 coaching in digital programs for type 2 diabetes.
These newsletters used predictive models to suggest the next best action, whether coaching or digital activities, for managing their chronic condition based on members' engagement with Teladoc Health’s services. Members who received personalized next-best actions were more likely to engage in coaching services compared to those who received the standard newsletter, the study found.
In March, the Peterson Health Technology Institute unveiled an analysis that poured cold water on the effectiveness of widely used digital diabetes management solutions, stirring up discussion about how best to evaluate the growing market of digital health tools.
The blistering report concluded that diabetes monitoring apps "do not deliver meaningful clinical benefits, and result in increased healthcare spending." For its analysis, PHTI evaluated eight widely used digital health tools that support people with Type 2 diabetes from DarioHealth, Glooko, Omada, Perry Health, Teladoc (Livongo), Verily (Onduo), Vida Health and Virta Health.
Health technology vendors were quick to push back on the report. A Teladoc Health spokesperson said at the time that the company's data show its diabetes management program improves weight, blood pressure and medication adherence. "For uncontrolled members in our diabetes management programs (starting at A1c>=9), our program delivers meaningful results and sustains those health improvements over time. 2.9% A1C reduction at three months sustained over five years," the spokesperson said via email.
As an endocrinologist and a physician in digital health, Kompala is excited about the potential for tech, including digital health tools combined with AI, machine learning and analytics, to improve healthcare delivery. Technology can provide more touchpoints for patients to help augment episodic care and fill in any gaps between doctor visits, she noted.
"Twenty years ago, diabetes care was all just face-to-face interactions between the patient and provider. But, we know that in chronic disease management, it's so much about what's happening day in and day out between those visits. This machine learning-based model is to help with that—what's happening between the visits and making productive use of the time," she said.
AI and machine learning also can make chronic condition management programs more personalized to each individual, she added. "Chronic disease management is all about sustained engagement. The ability to keep someone engaged over time is so critical to be able to continue with longer-term control," Kompala said.
Teladoc's use of AI to personalize content in members' weekly newsletters is one example of a "high touch, high tech" approach, Kompala noted. "That drove increased engagement with that newsletter over time and that also translated into increasing engagement with the human touch, with our coaches. I think that's a good example of collaborative intelligence; we're using our coaches, but we're using AI to help drive more of that," she said.
Teladoc plans to continue exploring the use of AI to expand the reach and efficiency of its coaching services and its providers and to double down on personalized education and content.
"We're continuing to think about 'How do we give people really personalized guidance that will be meaningful to them?'," she said. "I'm excited to see where the future heads. This is just the beginning, our first steps into understanding and seeing how artificial intelligence and machine learning can really be employed effectively within the context of a digital health program like ours. I think there's a lot of exciting opportunity ahead."