Generative AI/LLMs for Plain Language Medical Information for Patients, Caregivers and General Public: Opportunities,...

Area of Science:

• Artificial Intelligence in Healthcare
• Medical Informatics
• Health Communication

Background:

• Generative artificial intelligence (gAI) and large language models (LLMs) are increasingly used by the public for medical information.
• These AI tools offer potential benefits for patient self-care, health literacy, and clinician engagement.
• However, inaccurate or unreliable information from AI can lead to significant negative health outcomes.

Purpose of the Study:

• To review published findings on the opportunities and risks of using gAI/LLMs for plain language medical information.
• To identify key themes related to accuracy, readability, and potential harms.
• To explore ethical considerations and recommendations for responsible AI implementation in healthcare.

Main Methods:

• Systematic review of 44 articles published between January 2023 and July 2024.
• Analysis focused on opportunities, risks, accuracy, readability, and patient involvement.
• Synthesis of findings regarding common risks and ethical concerns.

Main Results:

• Studies focused on readability and accuracy, with AI responses generally accurate, readable, and detailed.
• Common risks include oversimplification, over-generalization, reduced accuracy for complex queries, and lack of source transparency.
• Few studies involved actual patients, highlighting a gap in real-world user experience data.

Conclusions:

• AI tools show promise for medical information but require enhanced transparency, governance, and monitoring.
• Addressing ethical concerns like health equity, inclusiveness, and data privacy is paramount.
• Education for non-specialist users on optimal AI tool utilization is essential for safe and effective adoption.