ChatGPT can pass the medical licensing examination in mainland China

Wu Cheng-han is the attending physician in the Department of Neurosurgery at the Second Affiliated Hospital of Zhejiang University and also serves as a graduate tutor and the head of the international students’ class at Zhejiang University School of Medicine. Following tests by his research team, it was found that ChatGPT can pass the medical licensing examination in mainland China, correctly answering about 72% of the questions, scoring higher than over 80% of the medical students who took the test. The related achievements have been published in the journal “Digital Health”.

In an interview, Wu Cheng-han stated that the origin of this research was to understand the value of ChatGPT in clinical medicine. After proving that it could pass the medical licensing exam, the subsequent application basis was established.

Wu Cheng-han explained that the test is divided into a written test and a practical operation test. The former is a simple matter of having ChatGPT answer questions, while the latter involves having it simulate asking patients about their medical history, symptoms, etc., to test the degree of interaction with patients. The good performance of ChatGPT took him by surprise.

Wu Cheng-han’s team hopes that AI can act as a bridge between patients and doctors in the future. He said that when patients visit major hospitals, the waiting time is long and the actual consultation time is just a few minutes – even shorter with famous doctors – leading to poor patient experience. If ChatGPT could collect patient information in advance and generate a report for the doctor for reference, it could improve the doctor-patient relationship.

He believes that AI can provide doctors with hints based on case history, serving as an auxiliary tool to assist in differential diagnosis. In addition, AI has the potential to be an auxiliary tool in medical education. Wu Cheng-han is developing a conceptual model based on ChatGPT, turning AI into a patient to train and test medical students’ interaction with patients.

As for whether AI will replace doctors in the future, Wu Cheng-han frankly said, no matter how AI develops, human involvement is essential in the medical process, and it is the doctor who makes the final decision. He is not worried about AI replacing doctors, on the contrary, AI could make doctors’ work better and easier.

Furthermore, Wu Cheng-han mentioned that humans possess the trait of collective intelligence. As clinical work cannot make mistakes, the medical process is judged by resident doctors, attending doctors, and chief doctors, reducing the chance of errors after multiple stages, especially with the addition of ChatGPT. He emphasized that “medical care should never be a system where one person has the final say,” so collective human strategy is indispensable.

Wu Cheng-han is from Kaohsiung Taiwan. After graduating from the Department of Clinical Medicine at Peking University School of Medicine, he went on to the Clinical Medical Research Institute at Shanghai Jiao Tong University School of Medicine for his master’s degree, then went to the University of Leuven in Belgium for his PhD in Biomedical Science. After graduation, he continued his postdoctoral research work in the Department of Neurosurgery at Stanford University in the United States.

(Visited 11 times, 1 visits today)

Leave A Comment

發佈留言必須填寫的電子郵件地址不會公開。 必填欄位標示為 *