Abstract
<title>Abstract</title> <p>Background Prehypertension is an important risk factor for hypertension and related cardiovascular diseases. Physical activity (PA) is an effective non-drug approach for managing blood pressure and preventing the progression of prehypertension; however, maintaining long-term PA habits is often challenging. Combining behavior change theories with wearable devices and large language model (LLM)-based chatbots may offer more adaptive and personalized support by integrating user profiles, real-time activity data, and conversational feedback to address individual barriers to PA. However, little is known about the effectiveness of these interventions. Objective This study aims to assess the effectiveness of HabitBot, an LLM-integrated PA behavior change chatbot, in promoting PA habits among adults with prehypertension. It also explores how PA habits are formed and evaluates the intervention’s usability and acceptance. Methods This is a 12-month, two-arm parallel, pragmatic randomized controlled trial (RCT) with an embedded 3-month intensive longitudinal study. A total of 118 participants will be recruited from two companies and social media platforms in Beijing, China. Participants will be randomly assigned to one of two groups: the intervention group, which will receive a wearable blood pressure monitoring smartwatch paired with HabitBot, a chatbot powered by an LLM designed to support PA habit formation through personalized PA prescriptions, health advice, and motivational feedback; or the control group, which will receive the same smartwatch, a standard PA prescription, and access to a health app offering general education on blood pressure management. The primary outcome is daily step count, measured via the smartwatch. Secondary outcomes include physical activity levels, blood pressure, body composition, and psychological factors relevant to behavior change theories. Data will be collected at baseline, 3, 6, and 12 months. Results This study will provide insights into the effectiveness of HabitBot in promoting PA habits and improving health among prehypertensive adults. It will also explore the mechanisms underlying PA habit formation through detailed longitudinal data analysis. Conclusion The results of this study could help shape future LLM-integrated interventions aimed at long-term PA promotion and prehypertension management. By examining both reflective and reflexive pathways of habit formation, this study advances digital health strategies for reducing cardiovascular risk. Trial Registration Chinese Clinical Trials Registry, ChiCTR2400085073, Registered May 30, 2024</p>