Maniaci A, Lazzeroni M, Cozzi A, Fraccaroli F, Gaffuri M, Chiesa-Estomba C, Capaccio P. Can chatbots enhance the management of pediatric sialadenitis in clinical practice?
Eur Arch Otorhinolaryngol 2024:10.1007/s00405-024-08798-4. [PMID:
38955859 DOI:
10.1007/s00405-024-08798-4]
[Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/06/2024] [Accepted: 06/17/2024] [Indexed: 07/04/2024]
Abstract
OBJECTIVE
The purpose of this study was to assess how well ChatGPT, an AI-powered chatbot, performed in helping to manage pediatric sialadenitis and identify when sialendoscopy was necessary.
METHODS
49 clinical cases of pediatric sialadenitis were retrospectively reviewed. ChatGPT was given patient data, and it offered differential diagnoses, proposed further tests, and suggested treatments. The decisions made by the treating otolaryngologists were contrasted with the answers provided by ChatGPT. Analysis was done on ChatGPT response consistency and interrater reliability.
RESULTS
ChatGPT showed 78.57% accuracy in primary diagnosis, and 17.35% of cases were considered likely. On the other hand, otolaryngologists recommended fewer further examinations than ChatGPT (111 vs. 60, p < 0.001). For additional exams, poor agreement was found between ChatGPT and otolaryngologists. Only 28.57% of cases received a pertinent and essential treatment plan via ChatGPT, indicating that the platform's treatment recommendations were frequently lacking. For treatment ratings, judges' interrater reliability was greatest (Kendall's tau = 0.824, p < 0.001). For the most part, ChatGPT's response constancy was high.
CONCLUSIONS
Although ChatGPT has the potential to correctly diagnose pediatric sialadenitis, there are a number of noteworthy limitations with regard to its ability to suggest further testing and treatment regimens. Before widespread clinical use, more research and confirmation are required. To guarantee that chatbots are utilized properly and effectively to supplement human expertise rather than to replace it, a critical viewpoint is required.
Collapse