Accuracy of ChatGPT‐Generated Information on Head and Neck and Oromaxillofacial Surgery: A Multicenter Collaborative Analysis

Abstract

ObjectiveTo investigate the accuracy of Chat‐Based Generative Pre‐trained Transformer (ChatGPT) in answering questions and solving clinical scenarios of head and neck surgery.Study DesignObservational and valuative study.SettingEighteen surgeons from 14 Italian head and neck surgery units.MethodsA total of 144 clinical questions encompassing different subspecialities of head and neck surgery and 15 comprehensive clinical scenarios were developed. Questions and scenarios were inputted into ChatGPT4, and the resulting answers were evaluated by the researchers using accuracy (range 1‐6), completeness (range 1‐3), and references' quality Likert scales.ResultsThe overall median score of open‐ended questions was 6 (interquartile range[IQR]: 5‐6) for accuracy and 3 (IQR: 2‐3) for completeness. Overall, the reviewers rated the answer as entirely or nearly entirely correct in 87.2% of cases and as comprehensive and covering all aspects of the question in 73% of cases. The artificial intelligence (AI) model achieved a correct response in 84.7% of the closed‐ended questions (11 wrong answers). As for the clinical scenarios, ChatGPT provided a fully or nearly fully correct diagnosis in 81.7% of cases. The proposed diagnostic or therapeutic procedure was judged to be complete in 56.7% of cases. The overall quality of the bibliographic references was poor, and sources were nonexistent in 46.4% of the cases.ConclusionThe results generally demonstrate a good level of accuracy in the AI's answers. The AI's ability to resolve complex clinical scenarios is promising, but it still falls short of being considered a reliable support for the decision‐making process of specialists in head‐neck surgery.

Systematic Review of Large Language Models for Patient Care: Current Applications and Challenges

“…The authors were primarily affiliated with institutions in the United States (n=47 of 122 different countries identified per publication, 38.5%), followed by Germany (n=11/122, 9%), Turkey (n=7/122, 5.7%), the United Kingdom (n=6/122, 4.9%), China/Australia/Italy (n=5/122, 4.1%, respectively), and 24 (n=36/122, 29.5%) other countries. Most studies examined one or more applications based on the GPT-3.5 architecture (n=66 of 124 different LLMs examined per study, 53.2%) 13,26–29,31–34,36–40,42–49,52–54,56–61,63,65–67,71,72,74,75,77,78,81–89,91,92,94,95,97–100,102–104,106–109,111 , followed by GPT-4 (n=33/124, 26.6%) 13,25,27,29,30,34–36,41,43,50,51,54,55,58,61,64,68–70,74,76,79–81,83,87,89,90,93,96,98,99,101,105 , Bard (n=10/124, 8.1%; now known as Gemini) 33,48,49,55,73,74,80,87,94,99 , Bing Chat (n=7/124, 5.7%; now Microsoft Copilot) 49,51,55,73,94,99,110 , and other applications based on Bidirectional Encoder Representations from Transformers (BERT; n=4/124, 3...…”

“…Most studies examined one or more applications based on the GPT-3.5 architecture (n=66 of 124 different LLMs examined per study, 53.2%) 13,26–29,31–34,36–40,42–49,52–54,56–61,63,65–67,71,72,74,75,77,78,81–89,91,92,94,95,97–100,102–104,106–109,111 , followed by GPT-4 (n=33/124, 26.6%) 13,25,27,29,30,34–36,41,43,50,51,54,55,58,61,64,68–70,74,76,79–81,83,87,89,90,93,96,98,99,101,105 , Bard (n=10/124, 8.1%; now known as Gemini) 33,48,49,55,73,74,80,87,94,99 , Bing Chat (n=7/124, 5.7%; now Microsoft Copilot) 49,51,55,73,94,99,110 , and other applications based on Bidirectional Encoder Representations from Transformers (BERT; n=4/124, 3.2%) 13,83,84 , Large Language Model Meta-AI (LLaMA; n=3/124, 2.4%) 55 , or Claude by Anthropic (n=1/124, 0.8%) 55 . The majority of applications were p...…”

“…Most reports evaluated LLMs in English (n=88/89, 98.9%) 13,24–103,105–111 , followed by Arabic (n=2/84, 2.3%) 32,104 , Mandarin (n=2/84, 2.3%) 36,75 , and Korean or Spanish (n=1/89, 1.1%, respectively) 75 . The top-five specialties studied were ophthalmology (n=10/89, 11.2%) 37,40,48,51,65,74,97,98,100,101 , gastro-enterology (n=9/89, 10.1%) 25,32,34,36,39,61,62,72,96 , head and neck surgery/otolaryngology (n=8/89, 9%) 35,42,56,64,66,76,78,79 , and radiology 59,70,88–90,110 or plastic surgery 45,47,49,102,107,108 (n=6/89, 6.7%, respectively). A schematic illustration of the identified concepts of LLM applications in patient care is shown in Figure 2.…”

“…In terms of design limitations, many authors noted the limitation that LLMs are not optimized for medical use (n=46/89, 51.7%) 13,26,28,34,35,37–39,46,49,50,54–59,61,62,65,66,68,70,71,79–81,83–85,88,91,93–98,100–107,109 , including implicit knowledge/lack of clinical context (n=13/89, 14.6%) 28,39,46,66,71,79,81,83–85,98,103 , limitations in clinical reasoning (n=7/89, 7.9%) 55,84,95,102–105 , limitations in medical image processing/production (n=5/89, 5.6%) 37,55,91,106,107 , and misunderstanding of medical information and terms by the model (n=7/89, 7.9%) 28,38,39,59,62,65,97 . In addition, data-related limitations were identified, including limited access to data on the internet (n=22/89, 24.7%) 38,39,41,43,54–57,59,60,64,76,79,82–84,88,91,94,96,104,109 , the undisclosed origin of training data (n=36/89, 40.5%) 25,26,29,30,32,34,36,37,40,46,…”

Validation of the QAMAI tool to assess the quality of health information provided by AI

“…These tools have primarily been designed for manual, human-centric assessments and are not compatible with AI-generated outputs. To date, no validated tool exists to accurately assess the health information provided by ChatGPT, and the few clinical studies published on this topic have used non-validated instruments [9,10,17,[24][25][26][27].…”