Correspondence on Letter 1 regarding “Assessing the performance of ChatGPT in answering questions regarding cirrhosis and hepatocellular carcinoma”

Article information

Clin Mol Hepatol. 2023;29(3):821-822

Publication date (electronic) : 2023 May 31

doi : https://doi.org/10.3350/cmh.2023.0183

Yee Hui Yeo^,1, Jamil S. Samaan¹, Wee Han Ng²

¹Karsh Division of Gastroenterology and Hepatology, Department of Medicine, Cedars-Sinai Medical Center, Los Angeles, CA, USA

²Bristol Medical School, University of Bristol, Bristol, UK

Corresponding author : Yee Hui Yeo Karsh Division of Gastroenterology and Hepatology, Department of Medicine, Cedars-Sinai Medical Center, Los Angeles, CA 90048, USA Tel: +1-310-423-1971, Fax: +1-310-423-2356, E-mail: Yeehui.yeo@cshs.org

Editor: Seung Up Kim, Yonsei University College of Medicine, Korea

Received 2023 May 27; Accepted 2023 May 30.

Dear Editor,

We thank Dr. Ali for his insightful comments and for suggesting that a follow-up study on the potential advantages of the GPT-4 model would be of value [1]. We agreed with his observations about the ongoing advancements in artificial intelligence and its increasing utility in medical decision-making or education.

Cirrhosis and hepatocellular carcinoma (HCC), as highlighted by Dr. Ali, are conditions that requires active and thorough management to mitigate potential complications [2]. The nature of the disease requires patients to adhere to specific lifestyle changes and treatment protocols. Regular monitoring and timely intervention are essential in managing cirrhosis and HCC, as well as providing emotional support [3]. which often requires a comprehensive understanding of the disease by patients. In light of this, an artificial intelligence (AI) model like GPT-4 can play a pivotal role in patient education, rendering detailed and accurate information on basic knowledge and disease management.

Health literacy, indeed, holds paramount importance in improving patient outcomes [4]. A well-informed patient is more likely to adhere to recommended lifestyle modifications, medical appointments, and treatment plans, ultimately leading to better disease management. This is where ChatGPT can be instrumental [5,6]. As elucidated by Dr. Ali, GPT-4 can further provide more nuanced and easy-to-understand information, enabling patients to be proactive in their healthcare decisions and potentially reducing complications and readmissions.

Additionally, we would like to point out other features that GPT-4 may provide. We recently found that GPT-4’s multimodal capabilities can be valuable in addressing the needs of diverse patient populations [7]. For instance, it can be trained to understand and respond to queries in multiple languages, extending its potential to provide valuable information to a broader, more diverse patient population. By providing information in patients’ native languages, GPT-4 can help over-come language barriers that often contribute to health disparities. Our recent study showed that GPT-4 exhibits significantly higher accuracy in English and non-English cirrhosis-related questions. Specifically, an improvement was found in Mandarin and Korean responses, with a significantly smaller gap in the accuracy of the responses between them and English.

Lastly, we echo Dr. Ali’s sentiment regarding the verification of AI-generated information by a licensed healthcare provider. While the precision of responses from GPT-4 is impressive, it should be noted that these AI models supplement and do not replace the expert advice of physicians. Individual patient factors, regional guidelines, and evolving research are aspects that an AI, despite its learning capabilities, may not fully capture. Hence, while AI models like GPT-4 can serve as a reliable source of health information, they must be utilized with professional medical counsel.