Comparing traditional and AI-enhanced strategies for developing patient decision aids: a multiple case study

Area of Science:

• Health Informatics
• Medical Decision Making
• Artificial Intelligence in Healthcare

Background:

• Patient decision aids (DAs) are crucial for informed healthcare choices.
• Traditional DA development is resource-intensive and can be time-consuming.
• The integration of artificial intelligence (AI) offers potential for optimizing DA content creation.

Purpose of the Study:

• To develop and evaluate AI-driven prompts for generating balanced, evidence-based patient DA content.
• To compare the efficacy of an AI-enhanced DA development strategy against a traditional human-led approach.

Main Methods:

• A comparative mixed-methods study analyzed eight DAs from the Ottawa Inventory.
• Two researchers independently extracted content and developed AI-enhanced content using refined prompts.
• Quantitative and qualitative analyses compared AI-enhanced and traditional DA development outputs.

Main Results:

• AI-enhanced strategies generated a significantly higher percentage of unique benefits/harms for option-focused DAs (66%) and unique options for outcome-focused DAs (47%).
• Evidence searches validated the AI-generated options, confirming their benefit and ruling out hallucinations.
• Qualitative analysis indicated that AI-enhanced content was generally richer, though optimal combinations were not suggested.

Conclusions:

• AI-powered prompt engineering can significantly improve the efficiency and quality of patient DA development.
• This study offers practical insights for leveraging AI in creating evidence-based patient decision support tools.
• Further research may explore AI's role in suggesting optimal combinations for DA content.