A study reveals that an AI chatbot can produce more comprehensible, thorough and precise informed consent documents than surgeons.
ChatGPT-3.5 is an AI-powered language model developed by OpenAI that can generate human-like text based on context and past conversations and is designed to engage in conversations, answer questions, and help with various tasks.
The study compared its capabilities with those of surgeons.
The results have significant implications for improving the quality of informed consent for patients undergoing standard surgical procedures.
The research was carried out by a team at the University of California San Francisco (UCSF) and led by Dr Genevieve Melton-Meaux, MD, PhD, director of the Centre for Learning Health System Sciences (CLHSS).
It aimed to determine whether large language model-based chatbots could effectively communicate information about surgical procedures to patients, thereby improving the informed consent process.
Although the principles of informed consent are crucial to best practice, relevant documents are often too complex and challenging for patients to comprehend, which risks misunderstandings and mistrust.
The study compared the readability, accuracy and completeness of informed consent documents generated by ChatGPT-3.5 with those created by surgeons for six commonly performed surgical procedures:
- Colectomy
- Coronary artery bypass graft
- Laparoscopic cholecystectomy
- Inguinal hernia repair
- Knee arthroplasty
- Spinal fusion.
The goal was to evaluate whether chatbots could produce descriptions of surgical risks and benefits and the alternatives superior to those written by surgeons.
While the chatbot and surgeon-written documents were more complex than ideal for patient understanding, the chatbot’s descriptions were generally more straightforward.
It’s worth noting that surgeons would still need to review the automatically generated documents for accuracy because chatbots are not infallible.
The study demonstrated that large language model-based chatbots could offer a way for surgeons to save time and provide effective written communication around informed consent information for common surgical procedures.
By utilising these chatbots, physicians can enhance the informed consent process while easing the documentation burden, benefiting patients and healthcare providers.
The study’s findings suggest that large language models (LLM)--based chatbots are a promising tool for generating informed consent forms, reducing errors, and providing clear, understandable information to patients.
This development can revolutionise how patients interact with informed consent documents, allowing them to make more informed decisions about their medical procedures.
Further research may be needed to evaluate the practical application of LLM-based chatbots in real-world healthcare settings. Additionally, patient perspectives and opinions on the adequacy of LLM-generated informed consent documentation could be considered in future studies.
These advanced technologies may be crucial in transforming the healthcare industry, improving patient care, and ensuring the informed consent process is as clear and informative as possible.
As large language model-based chatbots continue to evolve and improve, their impact on healthcare documentation and communication is expected only to become more pronounced.
Integrating LLMs into electronic health records while adhering to privacy regulations could create more personalised and informative informed consent documents for patients, enhancing transparency and trust in the healthcare system.
