An Exploration of Just-In-Time Adaptive Interventions for Eating Behaviors Using LLM-Based Messages

Eating behavior is one of the most critical components of weight loss and a vital contributor to weight loss management. Adults labeled as obese according to world-renown scientific organizations number more than 1.9 billion, with the United States classified as the OECD (Organization for Economic Co-operation and Development) country with the most obese adults among the world population. For the past several years, sensors have been instrumental in facilitating research and strides toward the goal of weight loss and management, and more recently, with eating behavior. Additionally, large language models have been explored as a tool within various areas including digital health. Some of the methods that LLMs have been useful for so far include public communication of health notices to identifying caloric intake. To mitigate obesity, how people eat and when they eat requires study to aid in reducing weight. To make these determinations, the utilization of various methods including sensors as well as LLMs is vital. In this work, LLMs will be utilized and reviewed for their strengths and weaknesses, and determine any future research to further enhance their usefulness toward the goal of further study of eating behavior to further mitigate obesity and general weight loss.

Grantor dc:publisher

University of Alabama Libraries

Year dc:date.issued

2025

Author dc:creator

• Moses, Eric Clifton

Advisor dc:contributor.advisor

• Crawford, Chris S

Contributors dc:contributor

• Anderson-Herzog, Monica
• Gray, Jeff
• Hong, Xiaoyan
• Sazonov, Edward

• Actionability
• Compliance
• Eating Behaviors
• JITAI
• Large Language Models
• Likeability

Statement dc:rights

• All rights reserved by the author unless otherwise indicated.

Language dc:language.iso

en_US, English

Dc Identifier Other

1162346

OAI identifier oai:identifier

oai:ir.ua.edu:123456789/17020