Addressing social determinants of health? Consider artificial intelligence and machine learning

By | January 30, 2019

“Social determinants of health” is one of the hot buzz-phrases in healthcare these days, and for good reason. SDOH refers to outside factors that may impact a patient’s health, such as employment status and access to education, and providers can improve efficiency and curb costs by addressing these factors.

Technology is often utilized to do so effectively — and lately that means artificial intelligence and machine learning. Automation, and technology that learns as it goes, is one way providers can make sense of the glut of SDOH data, and make informed decisions based on it.

These technologies aren’t new, exactly, but they’re still maturing. That doesn’t mean it’s too early to jump onboard. Embracing new approaches can be one way to ensure that SDOH are addressed in a timely and efficient manner.

PREDICTION VS. ESTIMATION

Dr. Greg Berg, vice president of analytics at AxisPoint Health, said AI and machine learning are constantly in the process of getting better and faster. They’re being used and applied more often in the healthcare space, to mostly encouraging results.

“What’s new is the emerging of two fields: computer science groups, where they focus on machine learning, and statistics groups from universities,” said Berg. “The blending and merging of these two groups is starting to produce some fruitful things.”

The way these groups apply AI highlights the strategic differences between prediction and estimation. Left to their own devices (so to speak), computers are essentially robotic research assistants which help researchers make decisions based on data. Pretty valuable stuff.

Statistics groups and economists make predictions, but also focus on estimation — for example, looking at information to prognosticate what clinical affect a particular treatment might have.

AI and machine learning’s potential to address SDOH lies at the intersection between those two concepts; making effective use of data means modeling out scenarios to predict outcomes.

“This blending is really at the forefront now — using machine learning to do what it does well, and then applying some of those techniques in estimation, such as treatment effects,” said Berg.

TACKLING SOCIAL DETERMINANTS

Philosophies and approaches may be evolving, but the core idea of social determinants has been around for a long time, Berg said.

“This isn’t something that’s new,” he said. “When you go to a doctor’s visit, they might ask you about family history. When I was in college, the physician knew I didn’t have insurance and gave me the antibiotic as a sample. So taking into account where people are has been going on for a long time.

“What’s changed is the quantity of data, coming in a more systematic way than it has in the past. This is where technology can help streamline some of the processes, so a physician wouldn’t have to ask all the questions they asked in the past. They have all of that information sent to them.”

That lends efficiency to the whole process, and underscores the scope of options that can be explored when considering SDOH. Looking beyond what can be treated strictly through medical means is a core component to success.

To emphasize the point, Berg offered the analogy of a school system. At its core, the school system’s mission is to educate students. Over the years, though, many factors have been added to that mission which facilitate education but aren’t directly related — such as offering free or reduced-price meals to students of low socioeconomic standing. Those social determinants of education, as it were, are an important factor in facilitating the primary goal of learning.

Such is the case in healthcare. Medicine, at its core, is supposed to help people, and factoring in SDOH broadens the scope of what providers can do to maximize clinical quality.

HARNESSING THE INFORMATION

Information is great. But what to do with it? A lot of info can be found in electronic health records, but using new technologies to harness it properly is still a work in progress. The shift to value-based care may be accelerating the push to address SDOH.

“As risk has shifted from a payer to a health plan to different payers, it seems to me that value-based care is really the catalyst for using social determinants of health,” Berg said. “If the risk is put on the ACOs or other health organizations, they have their own financial incentives to make their members healthier and not cost as much.”

Using the data in the best possible way entails a marriage between technology and process. Putting data into the EHR in a systematic way is helpful, but knowing what to do next with the data can trigger an actual clinical intervention — whether that takes into account the socioeconomic status of the patient, whether they live alone (which poses a greater health risk) or whatever other unknown factors may be affecting a person’s health.

Having that data already in the EHR might change physicians’ behavior, Berg said.

Renters, for instance, are correlated with more costs and emergency department visits than are homeowners. Those who have lived at their residences longer tend to represent fewer costs, perhaps owing to their stability. And access to a vehicle is important, as those without it are more likely to miss physician appointments, resulting in challenges to access their medical care.

“Social determinants are a factor in a person’s health,” said Berg. “Given that they’re a factor, can the data be readily available in a systematic way that can improve outcomes and reduce costs? That’s the goal.”

Twitter: @JELagasse

Email the writer: jeff.lagasse@himssmedia.com

News Feed