Social determinants of health (SDOH) influence between 30-55% of all health outcomes. Understanding them at the population health and individual level is an integral part of impacting quality of life and controlling healthcare costs. That’s why Community Risk Data is a cornerstone of our SocialScape® platform, powering our risk analytics with deep insights into those social determinants of health. Our tools help define geographic and population-level risk, equipping clients with the data they need to design impactful health strategies.
Accuracy is critical, so we regularly refresh our data to ensure up-to-date, reliable metrics. With the current shift in the availability of public datasets, a stable source of truth is more important than ever. That’s why we’re proud to announce our latest Community Risk refresh—ensuring our insights remain robust, even as public data availability fluctuates.
To dive deeper, we spoke with our VP of Data Science, Dan McMichael, about the refresh process and what it means for our clients, including how updates to models focused on food, transportation, and social connectedness help address key drivers of health outcomes and costs.
What sets Socially Determined’s approach to data and insights apart from competitors?
Our team has taken a comprehensive approach to understanding social risk – specifically as it impacts health outcomes – at the community and individual level. There are several factors that set us apart, from our approach to building models, to the data we use to produce results, and finally how we deliver those results to our customers.
We begin with a model-first approach, combining research with expert opinion to consider what factors influence risk in a community. Take our Food Landscape model as an example. When we want to describe what drives potential risk, we think about the availability of healthy food choices in a location and what ability people may even have to make good choices about food. By taking this model-first approach, we work out how we can describe risk before we focus on how we will get the data to build out the models.
From there, we focus on the data and how to present the results. Unlike the SVI and ADI, which are commonly used resources that assess social risk, we go beyond the American Community Survey (the Census) in our community models, instead incorporating 17 unique public and commercial data sources, using over 500 data points.
Additionally, in our Food and Transportation models, we use points-of-interest data to describe how resources can be utilized to increase healthier outcomes. In our individual models we use three commercial sources and supplement our transportation model with publicly available data regarding public transportation resources.
Finally, we present our risk metrics at a far more granular level than our competitors. We produce community data packages at a ZCTA (ZIP Code Tabulation Areas), census tract, and county level. This gives our customers a first step towards understanding their areas of interest. We also present our community data at 200 to 400-meter diameter hexagons across the entire United States. This granularity at scale gives powerful insights that help our customers focus on highest-need areas. We run these community models and produce data deliverables and individual risk scores every six months on data sources refreshed every three months, ensuring up-to-date insights. Further, we produce a data package with risk scores and metrics that allow our customers the freedom to integrate that data into their own analytic environments to be analyzed and combined with their own data assets.
What advantages does POI (point-of-interest) data provide over relying soley on public datasets?
POI data is a big factor in what makes us unique in the industry and it’s an exciting aspect of our models.
Our POI data partner gives us the power and flexibility to categorize our data that gives us next-level insights. For example, in our food models we divide the data into healthy and unhealthy food offerings, leveraging those categories to better describe an area. Then, using research that suggests the conditions in place that make a person choose to go to a grocery store versus a fast-food restaurant, we can build a model that considers factors present in the real world.
Similarly, our transportation models leverage the same POI data, but here we categorize healthcare workers, like PCPs, high-volume specialists, and mental health specialists. We use this data to build a map of available resources to a person or a community to then start to understand the challenges to access. Finally, we use POI data to help describe conditions that lead to social isolation and incorporate those factors in our community Social Connectedness model.
We incorporate those results into our models and then pass those insights directly to our clients through our data packages. Our goal is to go beyond guessing what is available and drill in to how the real conditions in an area can support or prevent healthy outcomes.
What steps has your team taken to ensure that our data products can withstand changes in data sources and/or industry standards of data reporting?
The recent news around restrictions to government data sources is certainly troubling. However, we have worked hard to ensure that we do not rely on any one source to build our models. By incorporating commercial data, especially the point location data we use for food, healthcare, and social support services, we have a built-in buffer against some of the changes occurring in the public data space. We have also explored ways to apply our individual-level data to describe social risk at the community level and feel confident that, if needed, we can absorb drastic changes to the data we currently use. We communicate with our commercial data partners on a regular basis to ensure that the data we use to provide individual insights remains steady. Here, too, we have a buffer by using two different sources for individual-level data.
The recent mandate to remove certain public web pages and datasets highlights the critical need for high-quality, independent SDOH data. What factors impact your decision-making when evaluating new and existing data sources?
Even with the recent disruptions, our approach to evaluating data sources remains unchanged. By focusing on our models first, we can shift what data can be used to describe the factors on the ground. We rely on sources with trusted backgrounds and proven track records that come at the resolutions that provide the best coverage and detail possible. We also consider who else is using the data and for what purpose to help us understand how others interpret the usefulness. Finally, we run analyses on the data to be sure that what we are receiving is relevant and helps us tell the story we are trying to tell.
All of this makes the vetting process easier. We know if the data satisfies the need in the model, if it comes at a resolution that will help bring granularity to results, if it covers the communities the customer serves, and how recent it is. Still, our process is intensive. The data science team spends weeks pulling data in and evaluating its applicability to the models at every stage of the process. All along the way, we’re ensuring that from factor, to feature, to influencer, each stage produces meaningful metrics and results.
Get Started Today with Our Refreshed Community Risk Data
Regular data refreshes like this ensure accuracy, an accuracy that delivers results for any program incorporating SDOH data, from cost reduction to health equity and clinical research at a time where public data access can feel limited. For existing customers, our refreshed Community Risk data is live now in SocialScape. Not a customer yet? Reach out to our sales team to learn how you can get access to these industry leading insights!