Network analysis of the 2016-2021 interventional radiology residency match: algorithmic identification of dynamic communities and influencers

ABSTRACT

Purpose: Interventional radiology (IR) residency grew from 7 programs and 15 positions in 2016 to 87 programs and 164 positions in 2021. Analysis of prior years’ match data can reveal the most well-connected schools and programs, communities, influencers that bridge communities, and the network’s evolution over time. Applicants to IR and residency program stakeholders will gain valuable insights from analysis of the IR training network. Materials and Methods: 618 of 726 (85%) IR matches between 2016-2021 were compiled from publicly available sources NRMP data, department websites, and medical school match lists. Network plots were generated using Gephi 0.9.2. Medical schools and residency programs (nodes) and matches (edges) were assessed for number of connections, communities, and Eigenvector centrality (relative influence based on how well-connected a node is and how well-connected are its connections). Results: Preliminary analysis of the manually sourced public data reveals that medical schools graduating the greatest number of students matched to IR are Rosalind Franklin (15), Georgetown (14), U South Florida (14), and Northwestern (13), while residency programs matching the greatest number of students into IR are Emory (20), Rush (20), Vanderbilt (18), and U Pennsylvania (18). Overall, the most well-connected nodes in the network of IR matches are Rush (29), U Pennsylvania (28), Georgetown (26), Vanderbilt (23), and Yale (23). The most influential node in the 5-year network is U Pennsylvania with an Eigenvector centrality of 1.0, followed by Duke (0.70), Stanford (0.49), Emory (0.45), and Mount Sinai (0.43). A force-directed network algorithm illustrates that even as the number of matched students invariably grows, the number of communities of schools and programs dynamically expands and contracts 6 in 2016, 16 in 2017, 11 in 2018, 9 in 2019, 8 in 2020, and 8 in 2021. Further analyses are ongoing of year-over-year community stability and geographic trends, as well as impact of COVID and virtual interviews on the network. Interactive images are available. Conclusion: IR is among the most competitive specialties with challenges for applicants and programs alike—an issue exacerbated by the relative ease of virtual interviews and resultant increase in applications. Students may use network analysis to focus on programs within their own or other desirable communities, or to identify an influential program for an away rotation to bridge communities. Programs may use network analysis to guide outreach and inform perceptions of a program’s relative standing within and amongst communities or the broader IR training network.