Development and Analysis of AI Generated Motivational Health Intervention Content Using Prompt Engineering Architecture
Open Access
Author:
Pye, Nolan
Area of Honors:
Information Sciences and Technology
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Saeed M Abdullah, Thesis Supervisor Carleen Maitland, Thesis Honors Advisor
Keywords:
Generative AI Prompt Engineering Artificial Intelligence Natural Language Processing
Abstract:
Over 70% of adults in the United States are insufficiently active. While effective interventions
often require personalization, current approaches to motivational messaging mostly rely on manual
generation of intervention messages, which is not scalable. To address this pitfall, this thesis
investigates the generation and effectiveness of AI-generated motivational health messages through
the lens of GAIMPLAY (Generative AI to Motivate PhysicaL ActivitY), a novel, theory-informed
prompt engineering methodology designed for the tailored creation of motivational content. This
research utilizes natural language processing (NLP) tools for a quantitative analysis of sentiment,
readability, and text comparisons, assessing the effectiveness of prompt chaining in generating
tailored motivational content. Additionally, the thesis evaluates the utility of quantitative NLP
metrics in the analysis of AI-generated health messaging, aiming to bridge the gap between the
need for personalization in health interventions and the scalability challenges of manual message
generation.