Previously submitted to: Journal of Medical Internet Research (no longer under consideration since Sep 18, 2025)
Date Submitted: Apr 3, 2025
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Leveraging large language model (LLM)-powered agents when addressing problems in mental health and chronic conditions: a scoping review
ABSTRACT
Background:
Recent advancements in large language models (LLMs) present promising opportunities for improving the management of chronic health conditions and mental health challenges. However, their integration into healthcare remains complex, especially with rapid developments like ChatGPT since 2022. These developments highlight the need for collaboration between healthcare professionals (HCPs) and LLM-powered agents to ensure safe and effective adoption.
Objective:
This scoping review aims to assess and synthesize how LLM agents are integrated into healthcare to support chronic disease and mental health management while guiding informed adoption.
Methods:
We conducted a scoping review following Arksey and O'Malley's framework. We searched PubMed/MEDLINE, Embase, ACM Digital Library, and Web of Science for English-language articles published between 1st Jan 2020 to 17th April 2024 that examined the use of LLM-powered agents as assistive tools in managing chronic conditions and mental health. After screening for inclusion, we performed data extraction and summarized the findings.
Results:
We total included 14 studies. Among those six used pretrained models without any modification, while eight studies fine-tuned or augmented the models to improve performance. LLM-powered agents made contributions across various medical decision-making scenarios by assisting with health data interpretation, automating clinical assessment and monitoring, and providing clinical decision references. Three key aspects of LLM evaluation emerged: model performance, comparative evaluation, and human evaluation.
Conclusions:
Responsible integration of LLM-powered agents requires caution due to early-stage challenges like data limitations and AI hallucinations. Sustainable adoption will benefit from practical insights into user interactions and real-world implementation. Future studies should establish performance benchmarks, focus on user experience, improve LLM functionality, and set protocols for long-term evaluation, addressing stability and reliability to ensure lasting benefits in patient care.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.