OpenAI's latest upgraded large language model (LLM) ChatGPT-4.0 passed a clinical neurology exam with an 85% accuracy rate in a conceptual validation study. The authors of the research report believe that after some fine-tuning, LLM will have "significant applications" in clinical neurology.

A group of researchers from Heidelberg University Hospital and the German Cancer Research Center in Heidelberg announced the experimental results on December 7. The test was conducted on May 31 using two LLMs, namely ChatGPT-3.5 and its later version ChatGPT-4.0.

The researchers used the question banks of the American Psychiatric and Neurological Association and a small portion of questions from the European Neurological Association.

The old version of ChatGPT had a score rate of 66.8%, answering 1306 out of 1956 questions correctly, while the latest version ChatGPT-4.0 had a score rate of 85%, answering 1662 questions correctly. The average human score is 73.8%. ChatGPT-4.0 outperformed human users in behavioral, cognitive, and psychological-related questions and effectively "passed" the neurological exam, as in educational institutions, 70% of correct answers are usually considered a passing score.

However, in tasks requiring "higher-order thinking," the performance of both models is relatively weaker compared to tasks that only require "lower-order thinking."

According to the research group conducting the experiment, these results suggest the use of LLMs in clinical neurology after some modifications:

The researchers pointed out that there are still some reservations. While the prospects of applying LLMs to documents and decision support systems are clear, neurologists should be cautious in their actual use because they are still imperfect in higher-order cognitive tasks. Dr. Varun Venkataramani, one of the authors of the study, said in an interview with Cointelegraph:

"Artificial intelligence has already played a role in some significant tasks in the healthcare field, such as finding cancer treatments for AstraZeneca or combating the phenomenon of overprescribing antibiotics in Hong Kong."

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。