Home PublicationsData Innovators 5 Q’s for Anthony Beverina, CSO of Socially Determined

5 Q’s for Anthony Beverina, CSO of Socially Determined

by Eline Chivot
Anthony Beverina CSO Socially Determined

The Center for Data Innovation spoke with Anthony Beverina, co-founder and chief strategy officer of Socially Determined, a healthcare analytics company based in Washington, DC that applies data science to measure how the social determinants of health impact a population. Beverina discussed Socially Determined’s analytic insights helps organizations that manage population health to take action that improves care in areas that may have traditionally been ignored.

Eline Chivot: What has led you to create Socially Determined? What are the challenges that you aimed to address?

Anthony Beverina: I am one of three founders at Socially Determined and I bring the perspective from outside the healthcare market—my other two co-founders are both medical doctors. My background is in national security, intelligence, and homeland security. I founded a company called Digital Sandbox that specialized in delivering advanced threat and risk analytics in some pretty important and mission-critical settings. Through building that company, which we sold in 2013, I learned a lot about how to understand people and places better and how to better model conditions “on the ground” to help organizations better predict outcomes and allocate precious resources.

For me, founding Socially Determined with my co-founders was a logical extension of my previous work—and an incredibly exciting challenge. Healthcare operates a multi-trillion dollar industry without knowing at least half the picture of those they serve. My co-founders and I felt there was a major gap in understanding of the social risks in communities and among individuals that was at the heart of healthcare disparities, rising costs, and poor outcomes. We felt that by providing “social risk intelligence” to organizations, we could help them implement strategic programs that would help people in need, while also helping their bottom line. For us it’s a win-win, and we are very excited to be working on such an important and potentially market-shifting solution. Imagine what is possible when these organizations spending so much money each year know the full story of their customers.

Chivot: Can you tell me more about how Socially Determined uses algorithms and machine learning models, and why is this technology particularly powerful and well-suited to help you deliver your vision?

Beverina: Being an analytics company, we got to work right away building a secure, scalable platform called SocialScape that mashes up clinical and claims data with social factors and uses algorithms and machine learning to uncover how and where social risk impacts utilization, cost, and disparities in health.

There are two things we must get right analytically to address this problem at scale. First, we must have a definitive way to express the nature and levels of social risk and need among people based on the pattern of their lives and the context of the communities where they live and work. Second, we must be able to quantify the impact those social risk factors have on things people care about like health, spending, and utilization patterns.

On the first issue of measurement, we created a set of risk metrics that fuse hundreds of data elements into scores for issues like financial strain, housing instability, and food insecurity. This involves an intense amount of research and data wrangling to ensure our scores are based on the right influencers and the best, most authoritative data. For this we formed data partnerships and have created entirely new algorithms that mashup combinations of information about the individual and also about the community where they live and its layout and resources.

To measure the impact that social risk has, we built an algorithmic framework in our SocialScape platform that uses machine learning algorithms to find patterns and make predictions across millions of people and arbitrarily large “market” footprints. Using these algorithms, we are able to combine a multitude of clinical, utilization, and social factors to zero in on groups of people and locations where social risk is creating disproportionate effects so that our customers can envision how they might use their levers to manage all these risks. It’s amazing to see how much innovation and purpose-driven action can be unleashed when you simply lay out the exact nature of these social challenges. Not to be too hyperbolic, but we believe the social risk factors we look at are at the heart of many of the health disparities and inequities we see in this country. It’s our obligation to uncover those and visualize those for our customers.

Chivot: As evidence cannot be found solely in an individual’s health record or insurance claims file, Socially Determined set out to improve health outcomes with data—not medical, clinical data, as one might expect, but with indicators and conditions that are called social determinants of health (SDOH) such as food insecurity, education, and access to transportation. Can you explain the extent to which these play a role in health outcomes?

Beverina: I would say it even more strongly than that. Not only can the evidence not be found solely in health records or insurance claims files, I would say that almost none of the evidence we are looking for is found in those records. We are talking about the link between issues like poor housing and asthma, living in a food desert and diabetes progression, and lack of transportation options and delayed cancer diagnoses. Organizations care a great deal about these things because often the social risk factors explain the unexplained variability they are seeing in their models and in their business. It’s the missing ingredient and the implications are measured in the billions across the industry and in even more powerful intangible ways like health and equity among our family, friends and neighbors. The evidence we need to understand these issues is all around us, but almost never in traditional healthcare sources.

We seek to understand the communities where people live and what the resources and access look like in those communities. We seek to understand the patterns of peoples’ lives and issues like financial options, health literacy, and social connectedness. This can only be achieved by harvesting a great deal of data from a widely disparate set of sources. We harvest consumer and alternative risk data from commercial vendors, business data on communities from commercial sources, and an enormous amount of open data from federal, state, and local sources. People ask me the “what data do you use?” question all the time and the best way to describe it is that we are a model-first company. We know what kind of risk factors we are looking for, so we hunt and process any data source that we think provides good evidence for these models. There is no way around a massive amount of data prospecting, wrangling, and cleaning in this business. Organizations that have tried to model on a single source of consumer data, or a bit of screening data from clinical settings have not been successful. Like a risk analytic problem, you get out of it what you put into it.

Here’s an exercise to illustrate this point: Let’s look at the impact of food insecurity on one’s health. Food insecurity is a complex issue and there is no single dataset available to establish a person’s risk for food insecurity. The evidence cannot be found solely in an individual’s health record or insurance claims file. Our customers are trying to find the answers to important questions, such as: What is the link between food insecurity and diabetes progression? How does food insecurity impact maternal health and birth outcomes? What positive effects do fruit and vegetable prescription programs have on spending and health outcomes among program participants?

By following our model first and then evaluating the right datasets, we can answer these questions. Our models consider several aspects in this. First, the use of commercial business data and open data about the person’s neighborhood as evidence of access to healthy or unhealthy food. Second, harvesting unstructured data from clinical notes as evidence of what providers assess about a patient’s eating habits, susceptibility to hunger, or food literacy. Third, screening the individual for direct evidence of hunger or their inability to afford food. Fourth, using commercial patterns of life data on the individual as evidence of their ability to reliably afford food for their family.

Chivot: Socially Determined works with public and private organizations such as healthcare systems, life sciences companies, and insurance companies. How can they use your analytics platform, and what insights do you deliver to them?

Beverina: Each client has different goals and strategies, but our support essentially comes down to the same basic idea: We show them where and how social risk is impacting those they serve and the business metrics that matter to them and then help them prioritize for action and measure results. It’s full-cycle analytics as we want to make sure that they can measure the results of their actions. Getting that right starts with accurately characterizing their population and market and we do that by onboarding their data (some combination of member roster, clinical records, claims data) and create a “risk baseline” so they can see exactly (for the first time ever) the contours of social risk exposure in their business. From there it’s all about quantifying the impact and helping them think through their tactical and strategic actions. Sometimes that may be a tactical program like community-based organization referrals (referring a hungry person to a food bank), and often it is very strategic (produce prescription program, financial counseling, building a grocery store in a food desert). We will provide our expertise and support throughout implementation and provide program evaluation frameworks.

Invariably in these programs, we are seeing both an improvement to the lives of the participants and improvement in business metrics like costs and utilization patterns. We believe that if we are rigorous in our approach, we will build programs with our customers that are sustainable and scalable because they are not only the right thing to do but they also make business sense. For me, that is the power of domain-specific analytics to make a real difference.

Chivot: The COVID-19 pandemic has had a disproportionate impact on underprivileged populations. How does or can your approach/model play a role in overcoming such a challenge?

Beverina: Unfortunately, I don’t believe that we are going to like what we see when we analyze how differentially this pandemic has ravaged our population. You can read the press and see plenty written on what is happening in communities of color. Sadly for us at Socially Determined, we are not surprised by this and we got into this business to see if we could start to bend that particular curve. We feel it is important to dig into more than just outcome data and start really analyzing the conditions on the ground with respect to access, transportation, health literacy, and other factors to get a better understanding of the underlying social risk factors, just like we look at underlying health risk factors. We have been called on by our customers to work on this from a couple different perspectives—and there will likely be more going forward. First, we have looked across large populations and identified people that we deemed most at-risk for poor outcomes should they get the virus. For one customer, we looked at underlying health conditions and a host of social factors that enabled the organization to provide emergency outreach to support these vulnerable people. This is a great example of how our insights can enable an organization to proactively apply limited resources to direct those resources to the area of greatest need. Another large customer has asked us to look at confirmed cases to uncover patterns related to social determinants of health. We’re still in the middle of this analysis, but it will be the first study we have heard of that really has had the right data to fully understand how social risk factors play a role in who gets the virus and what happens once they do. It’s important work, and given how long this pandemic may linger, hopefully a meaningful contribution to the nation and world.

You may also like

Show Buttons
Hide Buttons