Abstract
Continuous glucose monitoring (CGM) generates detailed temporal profiles of glucose dynamics, but its full potential for achieving glucose homeostasis and predicting long-term outcomes remains underutilized. Here we present GluFormer, a generative foundation model for CGM data trained with self-supervised learning on more than 10 million glucose measurements from 10,812 adults mainly without diabetes1,2. Using autoregressive prediction, the model learned representations that transferred across 19 external cohorts (n = 6,044) spanning 5 countries, 8 CGM devices and diverse pathophysiological states, including prediabetes, type 1 and type 2 diabetes, gestational diabetes and obesity. These representations provided consistent improvements over baseline blood glucose and HbA1c levels and other CGM-derived measures for forecasting glycaemic parameters3,4. In individuals with prediabetes, GluFormer stratified those likely to experience clinically significant increases in HbA1c over a 2-year period, outperforming baseline HbA1c and common CGM metrics. In a cohort of 580 adults with short-term CGM and a median follow-up of 11 years5, GluFormer identified individuals at elevated risk of diabetes and cardiovascular mortality more effectively than HbA1c. Specifically, 66% of incident diabetes cases and 69% of cardiovascular deaths occurred in the top risk quartile, compared with 7% and 0%, respectively, in the bottom quartile. In clinical trials, baseline CGM representations improved outcome prediction. A multimodal extension of the model that integrates dietary data generated plausible glucose trajectories and predicted individual glycaemic responses to food. Together, these findings indicate that GluFormer provides a generalizable framework for encoding glycaemic patterns and may inform precision medicine approaches for metabolic health.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$32.99 / 30 days
cancel any time
Subscribe to this journal
Receive 51 print issues and online access
$199.00 per year
only $3.90 per issue
Buy this article
- Purchase on SpringerLink
- Instant access to the full article PDF.
USD 39.95
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
Data availability
The data used in this paper are part of the HPP and are accessible to researchers from universities and other research institutions (https://humanphenotypeproject.org/data-access). Interested bona fide researchers should contact info@pheno.ai to obtain instructions for accessing the data. Deidentified participant data from the AEGIS study will be made available upon publication through the Runa Digital Repository (runa.sergas.gal). Access will require a signed data access agreement, and proposals should be directed to F.G.
Code availability
Implementation of GluFormer is available at GitHub (https://github.com/Guylu/GluFormer).
References
Shilo, S. et al. 10 K: a large-scale prospective longitudinal study in Israel. Eur. J. Epidemiol. 36, 1187–1194 (2021).
Reicher, L. et al. Deep phenotyping of health–disease continuum in the Human Phenotype Project. Nat. Med. 31, 3191–3203 (2025).
Nathan, D. M. et al. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N. Engl. J. Med. 329, 977–986 (1993).
King, P., Peacock, I. & Donnelly, R. The UK prospective diabetes study (UKPDS): clinical and therapeutic implications for type 2 diabetes. Br. J. Clin. Pharmacol. 48, 643–648 (1999).
Gude, F. et al. Glycemic variability and its association with demographics and lifestyles in a general adult population. J. Diabetes Sci. Technol. 11, 780–790 (2017).
Saab, K. et al. Capabilities of Gemini models in medicine. Preprint at https://doi.org/10.48550/arxiv.2404.18416 (2024).
Zhou, Y. et al. A foundation model for generalizable disease detection from retinal images. Nature 622, 156–163 (2023).
Lutsker, G., Rossman, H., Godiva, N. & Segal, E. COMPRER: a multimodal multi-objective pretraining framework for enhanced medical image representation. Preprint at https://doi.org/10.48550/arxiv.2403.09672 (2024).
Yuan, H. et al. Self-supervised learning for human activity recognition using 700,000 person-days of wearable data. npj Digit. Med. 7, 91 (2024).
Thapa, R. et al. SleepFM: multi-modal representation learning for sleep across brain activity, ECG and respiratory signals. Proc. Mach. Learn. Res. 235, 48019–48037 (2024).
Chen, R. J. et al. Towards a general-purpose foundation model for computational pathology. Nat. Med. 30, 850–862 (2024).
Krishnan, R., Rajpurkar, P. & Topol, E. J. Self-supervised learning in medicine and healthcare. Nat. Biomed. Eng. 6, 1346–1352 (2022).
GBD 2021 Diabetes Collaborators. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet 402, 203–234 (2023).
Rawshani, A. et al. Mortality and cardiovascular disease in type 1 and type 2 diabetes. N. Engl. J. Med. 376, 1407–1418 (2017).
Moser, E. G., Crew, L. B. & Garg, S. K. Role of continuous glucose monitoring in diabetes management. Av. Diabetol. 26, 73–78 (2010).
Battelino, T. et al. Continuous glucose monitoring and metrics for clinical trials: an international consensus statement. Lancet Diabetes Endocrinol. 11, 42–57 (2023).
Kieu, A., King, J., Govender, R. D. & Östlundh, L. The benefits of utilizing continuous glucose monitoring of diabetes mellitus in primary care: a systematic review. J. Diabetes Sci. Technol. 17, 762–774 (2023).
Holzer, R., Bloch, W. & Brinkmann, C. Continuous glucose monitoring in healthy adults-possible applications in health care, wellness, and sports. Sensors 22, 2030 (2022).
Zahedani, A. D. et al. Digital health application integrating wearable data and behavioral patterns improves metabolic health. npj Digit. Med. 6, 216 (2023).
Shilo, S. et al. Continuous glucose monitoring and intrapersonal variability in fasting glucose. Nat. Med. 30, 1424–1431 (2024).
U.S. Food & Drug Administration. FDA clears first over-the-counter continuous glucose monitor. FDA https://www.fda.gov/news-events/press-announcements/fda-clears-first-over-counter-continuous-glucose-monitor (2024).
McInnes, L., Healy, J. & Melville, J. UMAP: uniform manifold approximation and projection for dimension reduction. Preprint at https://doi.org/10.48550/arxiv.1802.03426 (2018).
Bergenstal, R. M. et al. Glucose management indicator (GMI): a new term for estimating A1C from continuous glucose monitoring. Diabetes Care 41, 2275–2280 (2018).
Broll, S. et al. Interpreting blood GLUcose data with R package iglu. PLoS ONE 16, e0248560 (2021).
Rodbard, D. New and improved methods to characterize glycemic variability using continuous glucose monitoring. Diabetes Technol. Ther. 11, 551–565 (2009).
Wang, J. et al. Self-improving generative foundation model for synthetic medical image generation and clinical applications. Nat. Med. 31, 609–617 (2025).
Keshet, A. et al. CGMap: characterizing continuous glucose monitor data in thousands of non-diabetic individuals. Cell Metab. 35, 758–769 (2023).
Vickers, A. J., Van Calster, B. & Steyerberg, E. W. Net benefit approaches to the evaluation of prediction models, molecular markers, and diagnostic tests. BMJ 352, i6 (2016).
Van Calster, B. et al. Performance evaluation of predictive AI models to support medical decisions: overview and guidance. Preprint at https://doi.org/10.48550/arxiv.2412.10288 (2024).
Htet, T. D. et al. Rationale and design of a randomised controlled trial testing the effect of personalised diet in individuals with pre-diabetes or type 2 diabetes mellitus treated with metformin. BMJ Open 10, e037859 (2020).
Rein, M. S. et al. BREAst Cancer Personalised NuTrition (BREACPNT): dietary intervention in breast cancer survivors treated with endocrine therapy—a protocol for a randomised clinical trial. BMJ Open 12, e062498 (2022).
Ben-Yacov, O. et al. Personalized postprandial glucose response-targeting diet versus Mediterranean diet for glycemic control in prediabetes. Diabetes Care 44, 1980–1991 (2021).
The Diabetes Prevention Program Research Group. The Diabetes Prevention Program. Design and methods for a clinical trial in the prevention of type 2 diabetes. Diabetes Care 22, 623–634 (1999).
International Expert Committee. International Expert Committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care 32, 1327–1334 (2009).
Cersosimo, E., Solis-Herrera, C., Trautmann, M. E., Malloy, J. & Triplitt, C. L. Assessment of pancreatic β-cell function: review of methods and clinical applications. Curr. Diabetes Rev. 10, 2–42 (2014).
Abdul-Ghani, M. A. et al. The relationship between fasting hyperglycemia and insulin secretion in subjects with normal or impaired glucose tolerance. Am. J. Physiol. Endocrinol. Metab. 295, E401–E406 (2008).
Ansari, A. F. et al. Chronos: learning the language of time series. Transact. Mach. Learn. Res. https://openreview.net/forum?id=gerNCVqqtR (2024).
Rabanser, S., Januschowski, T., Flunkert, V., Salinas, D. & Gasthaus, J. The effectiveness of discretization in forecasting: an empirical study on neural time series models. Preprint at https://doi.org/10.48550/arxiv.2005.10111 (2020).
van den Oord, A. et al. WaveNet: a generative model for raw audio. In Proc. 9th ISCA Speech Synthesis Workshop 125 (2016).
van den Oord, A., Kalchbrenner, N. & Kavukcuoglu, K. Pixel recurrent neural networks. In Proc. 33rd International Conference on Machine Learning (eds Balcan, M. F. & Weinberger, K. Q.) 1747–1756 (2016).
Danne, T. et al. International consensus on use of continuous glucose monitoring. Diabetes Care 40, 1631–1640 (2017).
Cefalu, W. T. et al. A global initiative to deliver precision health in diabetes. Nat. Med. 30, 1819–1822 (2024).
Ahlqvist, E., Prasad, R. B. & Groop, L. Subtypes of type 2 diabetes determined from clinical parameters. Diabetes 69, 2086–2093 (2020).
Xiong, Z. et al. How generalizable are foundation models when applied to different demographic groups and settings? NEJM AI https://doi.org/10.1056/AIcs2400497 (2024).
Vaswani, A. et al. Attention is all you need. In Adv. Neural Information Processing Systems (eds Guyon, I. et al.) 30, 5998–6008 (2017).
Loshchilov, I. & Hutter, F. Decoupled weight decay regularization. In Proc. 7th International Conference on Learning Representations https://openreview.net/forum?id=Bkg6RiCqY7 (2019).
Chen, T., Kornblith, S., Norouzi, M. & Hinton, G. A simple framework for contrastive learning of visual representations. In Proc. 37th International Conference on Machine Learning (eds Daumé, H. D. III & Singh, A.) 119, 1597–1607 (2020).
He, K., Fan, H., Wu, Y., Xie, S. & Girshick, R. Momentum contrast for unsupervised visual representation learning. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition 9729–9738 (2020).
Radford, A. et al. Learning transferable visual models from natural language supervision. In Proc. 38th International Conference on Machine Learning (eds. Meila, M. & Zhang, T.) 139, 8748–8763 (2021).
Devlin, J., Chang, M.-W., Lee, K. & Toutanova, K. BERT: pre-training of deep bidirectional transformers for language understanding. In Proc. 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (eds. Burstein, J. et al.) Vol. 1 (Long and Short Papers), 4171–4186 (2019).
Yuan, H. et al. Self-supervised learning of accelerometer data provides new insights for sleep and its association with mortality. npj Digit. Med. 7, 86 (2024).
Acknowledgements
We thank all members of the Segal Laboratory, the Pheno.AI data science and NVIDIA Tel Aviv Research groups, and E. Barkan and A. Shocher for discussions. J.M. was supported by Novo Nordisk Foundation grant NNF23SA0084103, an EFSD/Novo Nordisk Foundation Future Leaders Award (no. 0094134) and the European Union (HORIZON-EIC-2023-PATHFINDERCHALLENGES-01-101161509). Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or European Innovation Council and SMEs Executive Agency (EISMEA). Neither the European Union nor the granting authority can be held responsible for them.
Author information
Authors and Affiliations
Contributions
G.L. conceived the project, designed and conducted all analyses, interpreted the results and wrote the manuscript. G.S. developed protocols, interpreted the results and wrote the manuscript. A.G. designed pipelines and created preprocessing scripts. S.S. interpreted the results and wrote the manuscript. J.R.G. and D.S.-B. acquired the PREDICT cohort data. R.D. interpreted the results and wrote the manuscript. F.G. wrote the manuscript. J.M. interpreted the results and wrote the manuscript. S.M. guided computational analyses. E.M. guided computational analyses and managed code-running infrastructure. E.P.X. interpreted the results and wrote the manuscript. G.C. interpreted the results and wrote the manuscript, and directed the project. H.R. conceived and directed the project and analyses, designed the analyses, interpreted the results and wrote the manuscript. E.S. conceived and supervised the project and analyses, designed the analyses, interpreted the results and wrote the manuscript.
Corresponding authors
Ethics declarations
Competing interests
G.S. and H.R. are employees in Pheno.AI, a biomedical data science company from Tel-Aviv, Israel. E.S. is a paid consultant of Pheno.AI. G.L.’s work was done during an internship at NVIDIA Research. The other authors declare no competing interests.
Peer review
Peer review information
Nature thanks Jessilyn Dunn and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Supplementary Information (download DOCX )
Supplementary Figs. 1–36 and Supplementary Tables 1–7.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Lutsker, G., Sapir, G., Shilo, S. et al. A foundation model for continuous glucose monitoring data. Nature 650, 978–986 (2026). https://doi.org/10.1038/s41586-025-09925-9
Received:
Accepted:
Published:
Version of record:
Issue date:
DOI: https://doi.org/10.1038/s41586-025-09925-9
