医学
疾病
比例(比率)
动作(物理)
重症监护医学
病理
量子力学
物理
作者
Fotis Dimitriadis,Stelina Alkagiet,Lamprini Tsigkriki,Panagiota Kleitsioti,George Sidiropoulos,Dimitris Efstratiou,Taisa Askalidi,Adam Tsaousidis,Michail Siarkos,Pinelopi Giannakopoulou,Angeliki-Despoina Mavrogianni,John Zarifis,George Koulaouzidis
标识
DOI:10.1177/00033197241238403
摘要
ChatGPT (Generative Pre-trained Transformer) is a large-scale language processing model, with possibilities for professional patient support in a patient-friendly way. The aim of the study was to examine the accuracy and reproducibility of ChatGPT in answering questions about knowledge and management of heart failure (HF). First, we recorded 47 most frequently asked questions by patients about HF. The answers of ChatGPT to these questions were independently assessed by two researchers. ChatGPT was able to render the definition of the disease in a very simple and explanatory way. It listed a number of the most important causes of HF and the most important risk factors for its occurrence. It provided correct answers about the most important diagnostic tests and why they are recommended. In addition, it answered health and dietary questions, such as the daily fluid and the alcohol intake. ChatGPT listed the most important classes of drugs in HF and their mechanism of action. It also answered with arguments to questions about patient's sex life, whether they could work, drive, or travel by plane. The performance of ChatGPT was described as very good as it was able to adequately answer all questions posed to it.
科研通智能强力驱动
Strongly Powered by AbleSci AI