Mapping a Decade of Research Trends in Plastic Surgery: A Topic Modeling Analysis of PSTM Abstracts

Plastic Surgery The Meeting (PSTM) is a leading platform for presenting advancements in surgical techniques, research, and innovation. While research output has grown substantially over the past decade, systematic analyses of evolving academic themes in plastic surgery remain limited. We sought to determine which research themes have dominated PSTM abstracts over the past decade and how these themes have evolved over time.

We applied BERTopic, a transformer-based topic modeling algorithm, to the last decade of publicly available PSTM abstracts. Abstracts were extracted, preprocessed, and embedded using the all-mpnet-base-v2 model. Dimensionality reduction (UMAP), clustering (HDBSCAN), and topic representation (TF-IDF) with maximal marginal relevance) were used to generate and refine topics.

Analysis of 4,867 PSTM abstracts (2014-2023) identified 11 coherent research topics and one outlier group. The most prevalent themes were "Breast Reconstruction & Implants" (26.1%) and "Cleft Lip, Palate & Craniofacial Surgery" (17.4%). Several areas demonstrated substantial growth, most notably "Transgender & Gender-Affirming Surgery" (36.3% CAGR) and "Carpal Tunnel & Wrist Surgery" (31.2% CAGR). A heat map reveals stable long-standing domains alongside rapidly expanding emerging fields, while the intertopic network highlights semantic links between clinically related areas, including the positioning of gender-affirming surgery adjacent to major reconstructive topics.

The BERTopic model effectively identified core areas of focus, emerging areas of interest, and relationships between research topics. This approach offers a data-driven framework for understanding evolving academic trends and can inform conference programming, funding priorities, and curriculum development as the field continues to advance.