Exploring the Role of a Large Language Model on Carpal Tunnel Syndrome Management: An Observation Study of ChatGPT

Ishith Seth, Yi Xie, Aaron Rodwell, Dylan Gracias, Gabriella Bulloch, David J. Hunter-Smith, Warren M. Rozen

Research output: Contribution to journalArticleResearchpeer-review

15 Citations (Scopus)

Abstract

Purpose: Recently, large language models, such as ChatGPT, have emerged as promising tools to facilitate scientific research and health care management. The present study aimed to explore the extent of knowledge possessed by ChatGPT concerning carpal tunnel syndrome (CTS), a compressive neuropathy that may lead to impaired hand function and that is frequently encountered in the field of hand surgery. Methods: Six questions pertaining to diagnosis and management of CTS were posed to ChatGPT. The responses were subsequently analyzed and evaluated based on their accuracy, coherence, and comprehensiveness. In addition, ChatGPT was requested to provide five high-level evidence references in support of its answers. A simulated doctor-patient consultation was also conducted to assess whether ChatGPT could offer safe medical advice. Results: ChatGPT supplied clinically relevant information regarding CTS, although at a relatively superficial level. In the context of doctor-patient interaction, ChatGPT suggested a diagnostic pathway that deviated from the widely accepted clinical consensus on CTS diagnosis. Nevertheless, it incorporated differential diagnoses and valuable management options for CTS. Although ChatGPT demonstrated the ability to retain and recall information from previous patient conversations, it infrequently produced pertinent references, many of which were either nonexistent or incorrect. Conclusions: ChatGPT displayed the capability to deliver validated medical information on CTS to nonmedical individuals. However, the generation of nonexistent and inaccurate references by ChatGPT presents a challenge to academic integrity. Clinical relevance: To increase their utility in medicine and academia, large language models must go through specialized reputable data set training and validation from experts. It is essential to note that at present, large language models cannot replace the expertise of health care professionals and may act as a supportive tool.

Original languageEnglish
Pages (from-to)1025-1033
Number of pages9
JournalJournal of Hand Surgery
Volume48
Issue number10
DOIs
Publication statusPublished - Oct 2023

Keywords

  • Artificial intelligence
  • carpal tunnel syndrome
  • chatbot
  • ChatGPT
  • CTS

Cite this