Piloting a Generative Artificial Intelligence Chatbot in a Mobile Weight Loss Program

12-week behavioral weight loss intervention that includes daily self-monitoring (tracking) of dietary intake, physical activity, and weight; personalized daily goals for dietary intake and physical activity; weekly behavioral lessons in the study app; weekly tailored feedback on goal progress and weight change; and 1-2 intervention messages per day that are tailored based on current diet, activity, and/or weight progress. Participants receive a Fitbit activity tracker and scale and track their activity and weight in the Fitbit app. Participants use a simplified approach to self-monitoring their dietary intake by tracking 'red' (high calorie) foods in the study's food log smartphone app. This data is used to inform the daily intervention messages and weekly tailored feedback.

The AGILE chatbot operates via a trained generative artificial intelligence large language model and offers additional support to participants for making changes in their dietary and physical activity behaviors. The chatbot is available for conversation only after a message has been delivered in the AGILE app. Participants can choose to start a conversation or not, and the ability to start a conversation expires at midnight. The chatbot is able to provide evidence-based cognitive behavioral support to participants including, but not limited to: 1) strategies and tips for dietary self-monitoring, physical activity tracking (wearing the Fitbit), and self-weighing; 2) overcoming common barriers to dietary and physical activity goal achievement; 3) general support for remaining engaged in behavior change efforts during the course of the program.