Who Posts on Instagram? Using Natural Language Processing to Assess the Relationship Between Training Background and Content of 700,000 Posts

A grant from the One-University Open Access Fund at the University of Kansas was used to defray the author's publication fees in this Open Access journal. The Open Access Fund, administered by librarians from the KU, KU Law, and KUMC libraries, is made possible by contributions from the offices of KU Provost, KU Vice Chancellor for Research & Graduate Studies, and KUMC Vice Chancellor for Research. For more information about the Open Access Fund, please see http://library.kumc.edu/authors-fund.xml. ; Background Instagram (Menlo Park, CA) is a major platform for the dissemination of plastic surgery (PS) information, but the training background of users is difficult to ascertain. Objectives We sought to better characterize the source and content of PS-related posts on Instagram. Methods Metadata from publicly available Instagram posts containing PS relevant hashtags was collected from December 2018 to August 2020 using Node.js (Node.js Foundation, San Francisco, CA). The data was characterized by account type, and post topics were analyzed using a custom dictionary of PS procedures applied with natural language processing. All data analyses were performed with R (The R Foundation, Vienna, Austria). Results Board-certified plastic surgeons account for 38% of posts on Instagram, followed by organizations (31%), nonplastics-trained physicians (19%), facial plastics (5%), oculoplastics (1%), and nonphysician providers (5%). Oculoplastics had the highest engagement rate with their posts (3.7 ± 5.1), whereas plastic surgeons had the lowest (2.7 ± 4.2). Breast aesthetics was the predominant topic posted by plastic surgeons (42%, P < .001), and board certification phrases distinguished their posts from other account types (23%, P < .001). Nonphysician posts focused on nonsurgical aesthetics like Botox and fillers (80%). However, nonplastics-trained physicians and organizations significantly contributed to procedural subcategories in a similar distribution to plastic surgeons. Conclusions Board-certified plastic ...