國際醫療Business News

Doctors are now coming across more and more patients who are using AI and online resources to self-diagnose themselves based on their symptoms before coming to the hospital. This gave fourth-year resident student Riley Lyons the idea to assess the accuracy of AI chat boxes like ChatGPT to the evaluations of real doctors. At the end of the experiment, Lyons and his colleagues reported their findings to an online health science publisher called medRxiv. They found that ChatGPT did quite well in their assessments compared to human doctors who reviewed the same symptoms and actually performed significantly better than the symptom checker on WebMd. Although there have been issues in the past with ChatGPT making outright false statements, the Emory study claimed that “the most recent version of ChatGPT made zero ‘grossly inaccurate’ statements” (Leonard, 2023). Nieraj Jain, a professor at the Emory Eye Center, stated that the AI engine was “definitely an improvement over just putting something into a Google search bar and seeing what you find” (Leonard, 2023). 

Although an improvement to the standard Google search, there are still many questions and precautions when thinking about how to integrate artificial intelligence and other new technologies into the health care system. Many doctors seem to agree that AI-based medical tools should also undergo an approval process – similar to how the FDA approves drugs. However, it is unclear how exactly the FDA could regulate AIs such as ChatGPT in the same kind of process.

Med-PaLM is another AI chatbot and was specifically designed for medical use by Google computer scientists. According to a report published in Nature, Med-PaLM compared “favorably with answers given by clinicians” (Leonard, 2023). Some health care professionals even noted how much more empathetic the responses were from AI chatbots than from human doctors. Companies are exploring the idea of using these chatbots for services such as mental health therapy, one such company being Replika. 

However, critics express their concerns of the widespread use of AI especially in regards to privacy, bias, safety, transparency, liability, and minimal regulatory oversight. Mason Marks, professor and MD in health law, also brought up worries that “companies might want to push a particular product over another” which could lead to “the potential for exploitation of people and the commercialization of data” (Leonard, 2023). The company who developed ChatGPT, OpenAI, also urged caution by warning that “OpenAI's models are not fine-tuned to provide medical information… You should never use our models to provide diagnostic or treatment services for serious medical conditions.” (Leonard, 2023). However, in the meantime, Mayo Clinic has partnered with Google and HIA Technologies to integrate the Med-PaLM Chatbot and to provide interactive digital assistants for its website.

Source: https://www.npr.org/sections/health-shots/2023/09/16/1199924303/chatgpt-ai-medical-advice