Large language models in obesity: a systematic review

Area of Science:

• Artificial Intelligence in Healthcare
• Obesity Management Technologies
• Clinical Decision Support Systems

Background:

• Generative AI and large language models (LLMs) present novel opportunities for addressing complex obesity challenges.
• The rapid evolution of AI necessitates a comprehensive understanding of its current role in obesity care.

Purpose of the Study:

• To systematically review and synthesize research on large language model (LLM) applications in obesity management.
• To critically evaluate the performance, limitations, and future research directions of LLMs in this field.

Main Methods:

• Systematic literature search of MEDLINE, OVID, and SCOPUS databases up to August 2025.
• Inclusion of studies on LLM applications in medical and surgical obesity care.
• Narrative synthesis of findings and risk of bias assessment using ROBINS-I.

Main Results:

• Thirty-three studies were included, exploring LLM use in personalized nutrition, education, medical therapy, and weight loss strategies.
• LLMs demonstrated promising accuracy and utility in some areas, but significant variability and limitations were observed.
• Identified limitations include inconsistent recommendations, inaccuracies, challenges with complex cases, and potential bias.

Conclusions:

• Generative AI and LLMs hold significant potential for improving obesity management, including personalized interventions and clinical decision support.
• Current LLM technology has critical limitations in accuracy, consistency, and handling nuanced clinical situations, requiring ongoing clinician oversight.
• Further research is essential to enhance model training, validate real-world performance, and address ethical concerns for clinical implementation.