In recent years, the notion of artificial intelligence (AI) and its place in society has been on the tips of tongues around the globe. Film and television producers, long-haul truckers, customer service representatives — seemingly everyone, in nearly every industry, is busily navigating the integration of this asset into their daily lives.
At Belong Health, where humanistic whole-person care serves as our guiding star, we welcome a marriage of technological progress with grounded human experience — and we always place a heavy emphasis on interpersonal connection.
(That’s right: robot physicians need not apply.)
For us, automation and artificial intelligence are exciting only when they free up more time for human-to-human clinical engagement. And in late 2023, our technology, data management, and analytics teams figured out how to do just that.
Mac Davis, VP of Digital Product and Data at Belong, along with a small team of in-house engineers led by VP of Engineering Venki Ravi, and Data Engineer, Marc Zellick, worked on a recurring and time-consuming challenge. They aimed to make the navigation of an abundance of member information easier by building upon Belong's modern, near-real-time data foundation.
“Care management professionals were having to search across multiple systems to collect member data,” Davis said. “They were going into a dozen different systems, and sometimes, even within those systems, they had to go to multiple places to collect all the right material.”
For industry veterans, this is a familiar problem. Fortunately, their deep understanding of automated technology allowed the team to forge a smoother, quicker path for information retrieval and comprehension. Drawing upon traditional data engineering methods, as well as exploring the appropriate application of innovative A.I. tools (such as ChatGPT, Anthropic, and Bard), they developed a tool through which to corral and summarize key member information, exactly as it was needed.
“We’d realized we had specific needs where our staff had been spending a lot of effort on certain activities,” Davis said. “So we asked, ‘what [tools] can we build or launch to make our processes more efficient, ourselves?”
That efficiency was quickly quantifiable — and staggering. Two months since the tool’s launch, the time needed to prepare for and hold an interdisciplinary care management meeting had dropped by more than 68%. This meant more care managers were now able to streamline their prep and regain valuable facetime with an even greater number of complex-needs patients.
“A lot of that efficiency came from centralizing the data into a format that makes sense and making it securely available across our infrastructure,” Davis said, “but it also came from translating the data into things that are specifically meaningful to the person looking at it. Overall, our tool sought to reduce the cognitive stepping stones between raw data and action.”
Reducing Mental Clutter
One of the most significant benefits of Belong’s new tech tool is its ability to guide practitioners from an open question about social determinants of health (such as one may find on a Health Risk Assessment Tool [HRA]) to an actionable plan that addresses a member’s immediate needs.
Take, for example, the HRA question: Do you feel you’ll be able to pay rent this month?
A ‘no’ answer to this question means the respondent is ‘housing insecure.’ But in a traditional information gathering system, that conclusion might not immediately be recognized, as it could be lost amidst a flurry of other questions and contexts. Belong’s new tool, in such cases, eliminates the intermediate categorization or guesswork that could stand between a ‘no’ answer to such a question and a practical response to that member’s daily concern.
“We don’t want someone scanning an HRA that’s done somewhere else, logging that answer away and saying ‘This person is housing insecure as of this date,’” Davis said. “If some of that information is six months old, we can’t know if the person is still housing insecure. We need to be able process information a way that says, ‘this is current, this is valid, and we need to figure out how to connect this person with housing help.’”
In this and numerous other cases, the raw power of the Belong Health tech tools is its ability to ‘minimize mental processing’ of all the administrative steps that can delay connections between a medical practitioner and a person in need.
“Within our whole-person care approach is really a focus on maximizing the time care managers have available to physically work with members,” Davis said. “An algorithm isn’t going to be able to infer prioritization of what a person really needs, but it can help us get members on the phone and, person to person, focus those important conversations on what is at issue.”
The tech tool also enables Belong’s care management team to templatize their interdisciplinary meetings for each member, allowing experts to engage patients and primary care practitioners (PCP) more swiftly in the development of strategy.
“We've really been able to leverage cutting-edge AI tools to streamline care management,” says Natasha VanWright, Belong's SVP of Clinical Operations. “This has led to reduced administrative burdens, enhanced both the member experience and clinical engagement, all leading to better healthcare outcomes.”
Since late last year, Belong’s team rapidly iterated three versions of the tech tool and are excited for all the possibilities it may unlock.
Taz Alam, Belong’s Chief Technology Officer, shared, “In addition to the care management work, we’re working on a variety of other use cases to increase the efficiency for our teams or generate cost savings. Although a lot of the data sources we’re using are common, the way we synthesize that data is unique. We want to make sure that is front and center because that is truly a differentiator.”
Looking for ways to improve the productivity and efficiency for our teams and/or generate cost savings. We’re also interested in meaningfully differentiating our products and services, by adopting unique ways to synthesize and use the common data sources. We want to make sure that is front and center of our innovation.
Throughout such developments, the Belong team remains confident A.I. will serve as merely a capable ‘copilot’ to the great work Belong and its partners are already doing. Implementation of the Belong tech tools relies heavily on direct feedback from users, and on rapid investigation by data management and analytics teams when outliers emerge.
Perhaps most importantly, Davis says Belong’s ongoing development and refinement of these tools will only help stem the tide of regulatory work that risks slowing healthcare outcomes.
“A.I. isn’t going to get someone to stop smoking next week, but it can help us put our people in a position to make meaningful connections,” Davis said. “Whether you’re a doctor or a payer, there’s not going to be less work to do as regulation and administrative overhead grows. A.I. can help automate some of that, so that a doctor can ultimately spend more time with their patient.”