Sun, Xingzhi et al., "GlucoAgent: Personalized Glucose Forecasting from Multimodal Personal Context", arXiv:2611.04126, CC BY 4.0
PreprintCatalog paper date: 2026Added to atlas: Nov 5, 2026Source
Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.
GlucoAgent is an LLM agent that forecasts glucose by writing and back-testing its own analysis code over a person's CGM, smartwatch, meal and profile history. On CGMacros it has the lowest four-horizon mean RMSE (22.5 vs Chronos-2 23.5 mg/dL), with the biggest advantage in post-meal windows.
Signals
Glucose, HR, meal logs/photos, profile
Form factor
wrist + CGM
Year
2026
Architecture
LLM agent (Claude Opus 5, high reasoning effort) that writes and back-tests code (meal retrieval, physiological and statistical models) over a person's multimodal context, then submits a forecast with reasons and evidence
Pretraining data
No pretraining; agentic tool use over each user's history
License
Paper arXiv CC BY 4.0; no code release found (the GitHub link in the paper is the CGM-LSM baseline)
LLM agent (Claude Opus 5, high reasoning effort) that writes and back-tests code (meal retrieval, physiological and statistical models) over a person's multimodal context, then submits a forecast with reasons and evidence
Continuous glucose monitors, smartwatches, meal logs, and health profiles form a rich multimodal personal context that can significantly enhance glucose forecasting. However, existing methods struggle to effectively utilize these signals to improve glucose prediction because the relevance of different modalities varies drastically across individuals and time. We present GlucoAgent, an LLM-based agentic system that addresses this challenge through dynamic, case-by-case analysis at inference time. For each forecast, GlucoAgent writes and runs code over an individual's multimodal context and returns predictions with a reason and supporting evidence. The generated programs are highly diverse, ranging from retrieving similar past meals to fitting and back-testing physiological and statistical models. On CGMacros, a multimodal glucose dataset, GlucoAgent outperforms time series foundation models on every forecasting horizon, achieving an average RMSE of 22.5 mg/dL compared to 23.5 mg/dL of Chronos-2, the leading multimodal time series foundation model. Furthermore, GlucoAgent is highly effective at leveraging multimodal data. The addition of multimodal data reduces its average RMSE by 1.4 mg/dL (24.0 mg/dL to 22.5 mg/dL), compared to a marginal 0.3 mg/dL (23.8 mg/dL to 23.5 mg/dL) improvement for Chronos-2. This advantage is most pronounced in challenging post-meal windows, where GlucoAgent uses multimodal context to reduce 60-minute forecast RMSE by 4.6 mg/dL (31.0 mg/dL to 26.5 mg/dL), far exceeding the 0.8 mg/dL (31.9 mg/dL to 31.1 mg/dL) reduction achieved by Chronos-2.
Verbatim fromarXiv· arXiv metadata (abstracts are released under CC0)
Key results
22.5
mean RMSE
On CGMacros it has the lowest four-horizon mean RMSE (22.5 vs Chronos-2 23.5 mg/dL), with the biggest advantage in post-meal windows.
How it works
Datasets linked in the hub; pretraining vs evaluation roles follow the paper.
Pretraining data
No pretraining; agentic tool use over each user's history
Architecture
LLM agent (Claude Opus 5, high reasoning effort) that writes and back-tests code (meal retrieval, physiological and statistical models) over a person's multimodal context, then submits a forecast with reasons and evidence
Cite the publication recorded for this model. Check it against the publisher before submitting.
Sun, X., Kim, H. J., Wang, Z., & Chen, H. (2026). GlucoAgent: Personalized Glucose Forecasting from Multimodal Personal Context (Meta Reality Labs; arXiv 2611.04126). arXiv 2611.04126v1. https://arxiv.org/abs/2611.04126
Sun, Xingzhi, Hyo Jin Kim, Zhiyuan Wang, and Hao Chen. 2026. "GlucoAgent: Personalized Glucose Forecasting from Multimodal Personal Context (Meta Reality Labs; arXiv 2611.04126)." arXiv 2611.04126v1. https://arxiv.org/abs/2611.04126.
Sun, X., Kim, H. J., Wang, Z. and Chen, H. (2026) 'GlucoAgent: Personalized Glucose Forecasting from Multimodal Personal Context (Meta Reality Labs; arXiv 2611.04126)', arXiv 2611.04126v1. Available at: https://arxiv.org/abs/2611.04126
Sun, Xingzhi, et al. "GlucoAgent: Personalized Glucose Forecasting from Multimodal Personal Context (Meta Reality Labs; arXiv 2611.04126)." arXiv 2611.04126v1, 2026. https://arxiv.org/abs/2611.04126.
X. Sun, H. J. Kim, Z. Wang, and H. Chen, "GlucoAgent: Personalized Glucose Forecasting from Multimodal Personal Context (Meta Reality Labs; arXiv 2611.04126)," arXiv 2611.04126v1, 2026, [Online]. Available: https://arxiv.org/abs/2611.04126.
@misc{sun2026glucoagent,
title = {{GlucoAgent: Personalized Glucose Forecasting from Multimodal Personal Context (Meta Reality Labs; arXiv 2611.04126)}},
author = {Sun, Xingzhi and Kim, Hyo Jin and Wang, Zhiyuan and Chen, Hao},
year = {2026},
howpublished = {arXiv 2611.04126v1},
eprint = {2611.04126},
archivePrefix = {arXiv}
}
History
Added to the ChatPPG research hub
Newsletter
Get new PPG datasets, papers and models
Get alerts when GlucoAgent is benchmarked on new datasets, plus the weekly digest.
New datasets, papers and models
Practical guides from the hub
Unsubscribe anytime
The waveform is an illustration of a PPG pulse, not a recording.
