Tom Lawry is the National Director for Artificial Intelligence, Health and Life Sciences at Microsoft.
“The future is already here. It’s just not evenly distributed.” – William Gibson, Futurist.
With so many types of data available today to help monitor and manage health, here’s a question: Which is a better predictor of health status: your genetic code or your zip code?
Where you live affects how you live. It impacts whether you have access to healthy food, places to exercise or health services when needed. Your “living location” also affects your personal and family’s economic prosperity based on the availability of jobs, unemployment rates, education and training opportunities. These “social” factors shape and determine health and longevity across your lifespan.
Social determinants of health (SDOH) matter when it comes to addressing how we improve the health status of individuals, communities and nations. SDOH are conditions where people live, learn and work that affect a wide range of health and quality-of life-risks.
Here’s an example from a MedCity News article: “Two 60-year-old women live 10 miles apart in the Washington DC area. They’ve both been prescribed beta-blockers for high blood pressure, both have family histories of Type 2 diabetes, and have missed their last few annual check-ups. What should their care plans look like? Should they be different?”
“Clinically, they’re spitting images of each other. However, one piece of data — their zip code — can dramatically tilt the equation. Turns out, they face radically different life expectancies (63 versus 96 years), just based on the difference in their geographic locations. This 33-year life expectancy gap can be chalked up to differences in income level, education level, and access to grocery stores with fresh food.”
Traditional health systems have historically used data to understand the physiologic aspects of a health or medical condition. Such data is important in making diagnoses and managing health, but this only shows part of the picture. Social and environmental factors are much more indicative of a patient’s health outcome than once thought. One study suggests that 60% of a patient’s healthcare outcome is driven by their behavior and social and economic factors, 10% by their clinical care, and 30% by their genetics.
The Color Of Coronavirus
According to an article from The Conversation, “Neighborhoods with large black populations tend to have lower life expectancies than communities that are majority white, Hispanic or Asian. Such racial differences reflect the places in which different races live, not the individual characteristics of people themselves.”
Covid-19 gave voice to these issues. For example, at the beginning of the pandemic, Black Americans were twice as likely to die from Covid-19 even though they were a smaller percentage of the overall population. In taking a closer look, two things became self-evident: First, higher death rates related to where people lived. Second, the “twice as likely to die” was a statistical average. Underneath this average was the true story. In reality, if you were Black and living in Washington D.C., you were six times more likely to die of Covid at that time. Living in Michigan meant that you were four times more likely to die of Covid.
The Conversation article also notes that “Research shows that black communities are less likely to have access to resources that promote health, like grocery stores with fresh foods, places to exercise and quality health care facilities. This is true even in middle-class neighborhoods.”
Artificial Intelligence (AI) As A Turning Point
AI gives us the ability to better understand and proactively address social determinants impacting health.
To factor SDOH into health planning, health organizations must first be able to identify consumers facing adverse SDOH. Once identified, such factors can be incorporated into personal health management and population health strategies.
AI can help automate the identification of people whose health is likely impacted by their living situation. Opportunities include adding intelligent features to EMRs and proactive assessments of patient populations. Such activities help to identify and triage at-risk populations and enable organizations to build intelligent workflows for referrals and follow-up.
One study found that AI accurately predicted inpatient and emergency department utilization using only publicly available SDOH data such as gender, age, race and address.
AI promises to make it more practical to incorporate SDOH into care management and population health strategies. AI can identify consumers whose health issues are related to SDOH and then help clinicians with targeted interventions to help them better manage their health while maximizing the use of resources.
Understanding and incorporating social determinants of health in health planning is at the heart of moving toward healthier citizens and communities.