Table 1.
Associated features in the three sets of clustering analyses.
Fig 1.
Clustering results displayed on a map of WA: Set 1 surgical care procedure subgroup labels and other features clustering.
Note: The background of the map illustrates the division of zip codes (which was also used as the division of social indices calculation) within the state of Washington (WA). Each symbol on the map represents a hospital, where the geographic location is indicated by the symbol’s placement. The color of the symbol represents the cluster to which the hospital belongs, and the shape denotes the designated trauma level.
Fig 2.
Clustering results displayed on a 2-dimensional space: (a) Set 1 surgical specialty procedure subgroup labels and other features clustering, (b) Set 2 surgical care PCG distribution and other features clustering, (c) Set 3–1 surgical care volume clustering, (d) Set 3–2 surgical care distribution clustering. Note: Each symbol represents a hospital, with the distance indicating the relative distances based on all the clustering features. Hospitals with high-dimensional data for all features are visualized using the t-SNE method, which assigns each data point a location in a two-dimensional space, mapping similar data points closely together. Other features include sex, age, admission type, transfer status, insurance payer type, ISS, injury mechanism, and social indices.
Table 2.
Key features contributed to the TCs/non-TCs clusters from Set 1 surgical care procedure subgroup labels and other features clustering.
Table 3.
Key features contributed to the TCs/non-TCs clusters from Set 2 surgical care PCG distribution and other features clustering.