Behavioral Nudging With Generative AI for Content Development in SMS Health Care Interventions: Case Study

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare
• Digital Health Interventions

Background:

• Brief message interventions show promise in healthcare but lack transparency and accessible data.
• Current researcher-driven content creation is resource-intensive and inefficient.

Purpose of the Study:

• To demonstrate generative artificial intelligence (AI) for efficient and transparent content creation in SMS health interventions.
• To provide the largest publicly available dataset of brief messages and source code for replication.

Main Methods:

• Utilized GPT-3.5 (OpenAI) to generate medication adherence messages for type 2 diabetes patients.
• Employed evidence-derived behavior change techniques and an attributed prompt adhering to content and SMS standards.

Main Results:

• Generated a dataset of 1150 brief messages, the largest for a single healthcare intervention.
• 89.91% of messages met character length requirements, and 80.7% met readability standards.
• AI-generated messages demonstrated diversity comparable to existing datasets.

Conclusions:

• Generative AI offers a novel, efficient approach to healthcare intervention content creation.
• Further research is necessary to evaluate the ethical, safety, and efficacy standards of AI-generated content.