Can "ChatGPT Prompt Formula" Support Patients with Breast Cancer in Information Needs Inquery?

Area of Science:

• Artificial Intelligence in Healthcare
• Clinical Decision Support
• Patient Education Technology

Background:

• Large language models like ChatGPT show potential for supporting health information needs.
• Effective communication is crucial for patient understanding and engagement in clinical settings.

Purpose of the Study:

• To evaluate the impact of different ChatGPT prompt engineering strategies on the quality of responses to patient inquiries.
• To identify optimal prompt components for generating informative and empathetic patient-facing medical information.

Main Methods:

• Five distinct ChatGPT prompt formulas were designed and tested using simulated patient inquiries.
• Prompt components included task, context, persona, exemplar, and format.
• Response quality was assessed based on informativeness, clarity, and empathy.

Main Results:

• ChatGPT responses provided valuable information, particularly for complex medical terminology.
• Prompts incorporating empathy and structured formats (e.g., question lists) yielded more beneficial patient-oriented answers.
• Advanced prompt engineering demonstrably increased the quality and utility of ChatGPT's output.

Conclusions:

• Sophisticated prompt engineering significantly enhances ChatGPT's capacity to address patient information needs in clinical contexts.
• Future patient education strategies can leverage these advanced prompt components for improved health literacy.
• Further empirical validation is recommended to solidify these findings.