Machine learning review of hand surgery literature

Abstract

Background: Latent Dirichlet Allocation is an artificial intelligence model which processes text into topics, and has had broad application in medicine, political science, and engineering. As the orthopedic hand literature continues to grow, such technology may have value in efficiently conducting identifying trends and conducting systematic reviews. The purpose of this study is to demonstrate the use of Latent Dirichlet Allocation and machine learning to review literature and summarize the past 21 yr of hand surgery research. Methods: All original research articles published in the Journal of Hand Surgery (American), Journal of Hand Surgery (European), Hand, Journal of Bone and Joint Surgery (JBJS), Clinical Orthopaedics and Related Research (CORR), Journal of the American Academy of Orthopaedic Surgeons (JAAOS) and Plastic and Reconstructive Surgery (PRS) from 2000-2021 were analyzed using Latent Dirichlet Allocation, generating 50 topics which were then ranked by popularity and trended over the previous 21 yr. Results: Research article abstracts totaling 11,501 from 2000-2020 were extracted and analyzed to create 50 topics. Conclusion: This is the first study of its kind to utilize machine learning models for reviewing the hand surgery literature. Machine learning possesses the ability to rapidly process a large body of test and assess the current state of research and trends or research topics, which can aid clinicians and researchers in time-intensive tasks to provide clues that will promote areas of further study.