Harnessing Artificial Intelligence for Health Message Generation: The Folic Acid Message Engine

Area of Science:

• Artificial Intelligence
• Natural Language Processing
• Health Communication

Background:

• Social media campaigns are vital for public awareness but challenging to sustain due to the need for novel content.
• Generating consistent, relevant messages for communication campaigns is a resource-intensive bottleneck.

Purpose of the Study:

• To develop a pilot artificial intelligence (AI) system capable of generating numerous on-topic messages for awareness campaigns.
• To explore AI's potential in creating novel messages for public health initiatives, using folic acid as a case study.

Main Methods:

• Utilized the Generative Pretrained Transformer-2 (GPT-2) architecture, a machine learning model.
• Fine-tuned the GPT-2 model on a dataset of tweets related to #folicacid.
• Evaluated AI-generated messages against human-crafted messages through a web-based study with human raters.

Main Results:

• The AI message engine successfully generated hundreds of natural-sounding messages on the topic of folic acid.
• Human raters assessed the clarity and quality of AI-generated messages as equivalent to human-generated messages.
• Demonstrated the feasibility of using AI message engines for web-based awareness campaigns.

Conclusions:

• AI message engines can serve as a foundation for advanced AI-guided health communication systems.
• AI offers practical applications for social media campaigns and scientific potential for analyzing message characteristics.
• Future work should address ethical considerations for AI in health persuasion technologies.