American clusters: using machine learning to understand health and health care disparities in the United States

Abstract

Health and health care access in the United States are plagued by high inequality. While machine learning (ML) is increasingly used in clinical settings to inform health care delivery decisions and predict health care utilization, using ML as a research tool to understand health care disparities in the United States and how these are connected to health outcomes, access to health care, and health system organization is less common. We utilized over 650 variables from 24 different databases aggregated by the Agency for Healthcare Research and Quality in their Social Determinants of Health (SDOH) database. We used k-means-a non-hierarchical ML clustering method-to cluster county-level data. Principal factor analysis created county-level index values for each SDOH domain and 2 health care domains: health care infrastructure and health care access. Logistic regression classification was used to identify the primary drivers of cluster classification. The most efficient cluster classification consists of 3 distinct clusters in the United States; the cluster having the highest life expectancy comprised only 10% of counties. The most efficient ML clusters do not identify the clusters with the widest health care disparities. ML clustering, using county-level data, shows that health care infrastructure and access are the primary drivers of cluster composition.

Keywords: American clusters; health disparities; machine learning; social determinants of health.

Figure 1.

Life expectancy, in years. Panel A shows the average life expectancy, 2020. Panel B shows the average life expectancy over time from 2015 to 2020, by year and cluster. Life expectancy is the mean value for all of the counties included in the dataset for all years (n = 1673, cluster 1; n = 310, cluster 2; n = 1161, cluster 3).

Figure 2.

Map of US counties for the machine learning 3-clusters analysis. Note that the clusters, although showing some tendencies for geographic concentration, are not defined by geography. For example, some counties in the upper peninsula of Michigan and the panhandle of Texas are both assigned to cluster 1 based on the underlying patterns of the 650 county variables, not geography. Maps are created using data for all counties in the dataset (n = 1673, cluster 1; n = 310, cluster 2; n = 1161, cluster 3).