pith. sign in

arxiv: 2601.11505 · v2 · submitted 2026-01-16 · 💻 cs.LG · cs.AI· cs.SY· eess.SY· q-bio.QM

MetaboNet: The Largest Publicly Available Consolidated Dataset for Type 1 Diabetes Management

Miriam K. Wolff , Peter Calhoun , Eleonora Maria Aiello , Yao Qin , Sam F. Royston This is my paper

Pith reviewed 2026-05-16 13:17 UTC · model grok-4.3

classification 💻 cs.LG cs.AIcs.SYeess.SYq-bio.QM
keywords type 1 diabetesdataset consolidationCGMinsulin datapublic datasetmachine learningglycemic management
0
0 comments X

The pith

MetaboNet consolidates multiple type 1 diabetes datasets into one resource with 3135 subjects and 1228 patient-years of CGM and insulin data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The authors combine several existing public type 1 diabetes datasets that include both continuous glucose monitoring and insulin pump records. They convert the data into a single standardized format called MetaboNet while keeping auxiliary details such as carbohydrate intake and activity when available. The merged collection reaches 3135 subjects and 1228 patient-years, far larger than prior individual benchmark sets. A public subset is released for immediate download, and processing pipelines are supplied for the remaining data that require data-use agreements. The unified resource is intended to reduce preprocessing effort and support more comparable algorithm results across type 1 diabetes studies.

Core claim

By consolidating multiple publicly available T1D datasets that contain overlapping CGM and insulin pump dosing records into a unified format, the resulting MetaboNet dataset reaches a scale of 3135 subjects and 1228 patient-years while preserving auxiliary information such as carbohydrate intake and physical activity, with access provided through both direct public download and standardized pipelines for restricted components.

What carries the argument

The MetaboNet dataset formed by consolidating T1D datasets that supply both CGM and insulin records into one standardized structure.

If this is right

  • Algorithms can be developed and tested on a larger and more demographically varied patient population than any single prior dataset allows.
  • Standardized formatting removes the repeated preprocessing work that previously hindered cross-dataset comparisons.
  • Broad coverage of glycemic profiles supports claims of improved generalizability for new management tools.
  • Public availability of the main subset enables faster iteration by any researcher without application delays.
  • Processing pipelines for restricted data maintain inclusion while satisfying original data-use requirements.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Closed-loop insulin delivery systems could be trained and validated against this common large-scale resource before clinical trials.
  • Future observational studies may adopt the same data schema to allow automatic merging with MetaboNet.
  • Statistical power for detecting rare events such as nocturnal hypoglycemia increases with the combined patient-years.
  • Benchmarking platforms for machine-learning glucose predictors can now use a single reference dataset instead of multiple incompatible ones.

Load-bearing premise

Data collected under different protocols and devices can be merged into a single format without introducing meaningful biases or losing clinically important details.

What would settle it

An experiment showing that predictive models trained on the consolidated MetaboNet data produce measurably worse glycemic control outcomes than models trained on the original unmerged datasets.

Figures

Figures reproduced from arXiv: 2601.11505 by Eleonora Maria Aiello, Miriam K. Wolff, Peter Calhoun, Sam F. Royston, Yao Qin.

Figure 1
Figure 1. Figure 1: Overview of the scale of MetaboNet 2026, compared with the T1DEXI dataset after preprocessing according to the study’s inclusion criteria. The green portion represents publicly available data, orange indicates datasets governed by data use agreements (DUAs), and blue corresponds to the combined public and DUA￾protected datasets. Patient-years of CGM and insulin data are defined as periods during which cont… view at source ↗
Figure 2
Figure 2. Figure 2: Subject-level feature availability across the MetaboNet dataset. Each bar represents the number of subjects for which at least one non-missing value is available for the corresponding feature. This figure includes a subset of features, while the full list of available features is provided on the MetaboNet website [46]. offer opportunities to explore new hypotheses. The availability of physical activity ind… view at source ↗
Figure 3
Figure 3. Figure 3: Demographic distribution of the dataset. The top panels show the proportion of individuals by gender (left) and ethnicity (right), with the majority identifying as female and white, respectively, and a notable fraction in the “unknown” category for both attributes. The bottom panel displays the age and age of diagnosis distributions. participants span a wide range of body-mass index (BMI) values, further d… view at source ↗
Figure 4
Figure 4. Figure 4: Scatter plot showing the relationship between height (x-axis) and weight (y-axis). Each point represents one subject in the full dataset, with point colour indicating the individual’s BMI category. 3.2.1 Population Level Analyses Recent research highlights the need for investigating CGM-derived metrics [44, 16, 6], however, the sample size of the analyzed data is a limitation to the strength of the results… view at source ↗
Figure 5
Figure 5. Figure 5: Scatter plot showing the relationship between height (x-axis) and weight (y-axis). Each point represents one subject in the full dataset, with point colour indicating the individual’s BMI category. 4 How to Access MetaboNet The public part of the dataset can be accessed via https://metabo-net.org. The user must log in to access the public data directly. Data can be downloaded as either a single consolidate… view at source ↗
read the original abstract

Progress in Type 1 Diabetes (T1D) algorithm development is limited by the fragmentation and lack of standardization across existing T1D management datasets. Current datasets differ substantially in structure and are time-consuming to access and process, which impedes data integration and reduces the comparability and generalizability of algorithmic developments. This work aims to establish a unified and accessible data resource for T1D algorithm development. Multiple publicly available T1D datasets were consolidated into a unified resource, termed the MetaboNet dataset. Inclusion required the availability of both continuous glucose monitoring (CGM) data and corresponding insulin pump dosing records. Additionally, auxiliary information such as reported carbohydrate intake and physical activity was retained when present. The MetaboNet dataset comprises 3135 subjects and 1228 patient-years of overlapping CGM and insulin data, making it substantially larger than existing standalone benchmark datasets. The resource is distributed as a fully public subset available for immediate download at https://metabo-net.org/ , and with a Data Use Agreement (DUA)-restricted subset accessible through their respective application processes. For the datasets in the latter subset, processing pipelines are provided to automatically convert the data into the standardized MetaboNet format. A consolidated public dataset for T1D research is presented, and the access pathways for both its unrestricted and DUA-governed components are described. The resulting dataset covers a broad range of glycemic profiles and demographics and thus can yield more generalizable algorithmic performance than individual datasets.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript presents MetaboNet, a consolidated dataset formed by aggregating multiple publicly available Type 1 Diabetes (T1D) management sources that contain overlapping continuous glucose monitoring (CGM) and insulin pump dosing records. Inclusion requires both CGM and insulin data, with auxiliary fields (carbohydrate intake, activity) retained when available. The resulting resource comprises 3135 subjects and 1228 patient-years, distributed as a fully public subset plus DUA-restricted components with provided conversion pipelines to a unified format. The central claim is that this unified, substantially larger dataset will support more generalizable T1D algorithm development than existing standalone benchmarks.

Significance. If the consolidation preserves data integrity across heterogeneous sources, the dataset would constitute a valuable, large-scale public resource for T1D research, exceeding the scale of current benchmarks and enabling broader algorithmic validation. The explicit access pathways and processing pipelines add immediate practical utility for the community.

major comments (2)
  1. [Abstract] Abstract: The headline size metrics (3135 subjects, 1228 patient-years) rest on successful unification, yet no quantitative validation is supplied for the consolidation pipeline, such as alignment success rates, handling of mismatched CGM sampling intervals (5-min vs 15-min), insulin unit standardization, or the fraction of records dropped due to missing overlap. This directly affects the claim that the dataset yields improved generalizability.
  2. [Abstract] Abstract and processing description: No post-consolidation metrics or bias checks (e.g., comparison of glycemic distributions or temporal correlations before/after unification) are reported, leaving open the possibility that protocol differences systematically alter the data and undermine the asserted advantage over standalone datasets.
minor comments (2)
  1. [Abstract] The manuscript should explicitly list the source datasets consolidated and their individual sizes to allow readers to assess the contribution of each.
  2. [Abstract] Verify that the provided download link (https://metabo-net.org/) remains functional and that the public subset description matches the final release.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and the recommendation of minor revision. We agree that additional quantitative details on the consolidation pipeline will strengthen the manuscript and will incorporate them in the revised version.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The headline size metrics (3135 subjects, 1228 patient-years) rest on successful unification, yet no quantitative validation is supplied for the consolidation pipeline, such as alignment success rates, handling of mismatched CGM sampling intervals (5-min vs 15-min), insulin unit standardization, or the fraction of records dropped due to missing overlap. This directly affects the claim that the dataset yields improved generalizability.

    Authors: We agree that explicit quantitative validation of the unification steps is needed to support the reported scale and generalizability claims. In the revised manuscript we will add a new subsection to the Methods that reports alignment success rates across source datasets, the resampling procedure used to standardize CGM intervals to 5 minutes, the mapping applied for insulin unit standardization, and the exact fraction of records excluded because of insufficient temporal overlap between CGM and insulin streams. These metrics will be presented both in aggregate and broken down by source to allow readers to assess the robustness of the consolidation. revision: yes

  2. Referee: [Abstract] Abstract and processing description: No post-consolidation metrics or bias checks (e.g., comparison of glycemic distributions or temporal correlations before/after unification) are reported, leaving open the possibility that protocol differences systematically alter the data and undermine the asserted advantage over standalone datasets.

    Authors: We acknowledge that post-consolidation bias diagnostics would further substantiate that the unified dataset preserves the statistical properties of the original sources. In the revision we will include a new Results subsection that compares key glycemic statistics (mean glucose, time-in-range, coefficient of variation) and selected temporal correlation measures before and after unification, stratified by source. Any observed shifts will be discussed with respect to known protocol differences, thereby addressing the concern that systematic alterations may have occurred. revision: yes

Circularity Check

0 steps flagged

No circularity: descriptive aggregation of existing datasets only

full rationale

The paper performs no derivations, predictions, or parameter fitting. It aggregates publicly available T1D datasets into a unified format, reports the resulting subject count and patient-years as direct sums after processing, and provides access pipelines. No equations, self-citations as load-bearing premises, uniqueness theorems, or ansatzes appear. The size figures are empirical tallies from source data, not outputs that reduce to the paper's own inputs by construction. The work is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that heterogeneous T1D datasets can be merged without critical loss of information or introduction of artifacts.

axioms (1)
  • domain assumption Data from different sources can be standardized without loss of critical information
    Invoked when defining inclusion criteria and conversion pipelines for CGM and insulin records.

pith-pipeline@v0.9.0 · 5598 in / 1148 out tokens · 80026 ms · 2026-05-16T13:17:12.531163+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. A unified data format for managing diabetes time-series data: DIAbetes eXchange (DIAX)

    cs.LG 2026-04 accept novelty 5.0

    DIAX is a standardized JSON format that unifies diabetes time-series data including CGM, insulin, and meal signals to enable better interoperability and reproducibility across datasets.

Reference graph

Works this paper leans on

57 extracted references · 57 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    Review of Automated Insulin Delivery Systems for Individuals with Type 1 Diabetes: Tailored Solutions for Subpopulations

    E. M. Aiello, S. Deshpande, B. Ozaslan, et al. “Review of Automated Insulin Delivery Systems for Individuals with Type 1 Diabetes: Tailored Solutions for Subpopulations”. In:Current Opinion in Biomedical Engineering19 (2021), p. 100312.DOI: 10.1016/j.cobme.2021. 100312

    work page doi:10.1016/j.cobme.2021 2021

  2. [2]

    Model Predictive Control (MPC) of an artificial pancreas with data-driven learning of multi-step-ahead blood glucose predictors

    E. M. Aiello, M. Jaloli, and M. Cescon. “Model Predictive Control (MPC) of an artificial pancreas with data-driven learning of multi-step-ahead blood glucose predictors”. In:Control Engineering Practice144 (2024), p. 105810.DOI: 10.1016/j.conengprac.2023.105810

    work page doi:10.1016/j.conengprac.2023.105810 2024

  3. [3]

    A hierarchical network model for the estimate of the energy expenditure in individuals with type 1 diabetes

    E. M. Aiello, C. Toffanin, M. C. Riddell, et al. “A hierarchical network model for the estimate of the energy expenditure in individuals with type 1 diabetes”. In:Engineering Applications of Artificial Intelligence159 (2025), p. 111758.DOI:10.1016/j.engappai.2025.111758

    work page doi:10.1016/j.engappai.2025.111758 2025

  4. [4]

    A Longitudinal Multimodal Dataset of Type 1 Diabetes

    A. Alsuhaymi, A. Bilal, D. G. García, et al. “A Longitudinal Multimodal Dataset of Type 1 Diabetes”. In:Scientific Data12.1 (2025), p. 1379.DOI:10.1038/s41597-025-05695-1

    work page doi:10.1038/s41597-025-05695-1 2025

  5. [5]

    Clinical Targets for Continuous Glucose Monitoring Data Interpretation: Recommendations From the International Consensus on Time in Range

    Tadej Battelino, Thomas Danne, Richard M. Bergenstal, et al. “Clinical Targets for Continuous Glucose Monitoring Data Interpretation: Recommendations From the International Consensus on Time in Range”. In:Diabetes Care42.8 (2019), pp. 1593–1603.DOI: 10.2337/dci19- 0028

    work page doi:10.2337/dci19- 2019

  6. [6]

    A Comparison of Continuous Glucose Monitoring-Measured Time-in-Range 70-180 mg/dL Versus Time-in-Tight-Range 70-140 mg/dL

    R. W. Beck, D. Raghinaru, P. Calhoun, et al. “A Comparison of Continuous Glucose Monitoring-Measured Time-in-Range 70-180 mg/dL Versus Time-in-Tight-Range 70-140 mg/dL”. In:Diabetes Technology & Therapeutics26.3 (2024), pp. 151–155.DOI: 10.1089/ dia.2023.0380. 20 APREPRINT- JANUARY19, 2026

    work page arXiv 2024

  7. [7]

    Individualized Models for Glucose Prediction in Type 1 Diabetes: Comparing Black-Box Approaches to a Physiological White-Box One

    Giulia Cappon, Francesco Prendin, Andrea Facchinetti, et al. “Individualized Models for Glucose Prediction in Type 1 Diabetes: Comparing Black-Box Approaches to a Physiological White-Box One”. In:IEEE Transactions on Biomedical Engineering70.11 (2023), pp. 3105– 3115.DOI:10.1109/TBME.2023.3276193

    work page doi:10.1109/tbme.2023.3276193 2023

  8. [8]

    Event-Triggered Model Predictive Control For Embedded Artificial Pancreas Systems

    Ankush Chakrabarty, Styliani Zavitsanou, Francis J. Doyle, et al. “Event-Triggered Model Predictive Control For Embedded Artificial Pancreas Systems”. In:IEEE Transactions on Biomedical Engineering65.3 (2018), pp. 575–586.DOI:10.1109/TBME.2017.2707344

    work page doi:10.1109/tbme.2017.2707344 2018

  9. [9]

    Ten years after ImageNet: a 360° perspective on artificial intelligence

    S. Chawla, P. Nakov, A. Ali, et al. “Ten years after ImageNet: a 360° perspective on artificial intelligence”. In:Royal Society Open Science10.3 (2023), p. 221414.DOI: 10.1098/rsos. 221414

    work page doi:10.1098/rsos 2023

  10. [10]

    Design of a Real-Time Physical Activity Detection and Classification Framework for Individuals With Type 1 Diabetes

    S. Cho, E. M. Aiello, B. Ozaslan, et al. “Design of a Real-Time Physical Activity Detection and Classification Framework for Individuals With Type 1 Diabetes”. In:Journal of Diabetes Science and Technology18.5 (2024), pp. 1146–1156.DOI:10.1177/19322968231153896

    work page doi:10.1177/19322968231153896 2024

  11. [11]

    Evaluating clinical accuracy of systems for self-monitoring of blood glucose

    W. L. Clarke, D. Cox, L. A. Gonder-Frederick, et al. “Evaluating clinical accuracy of systems for self-monitoring of blood glucose”. In:Diabetes Care10.5 (1987), pp. 622–628.DOI: 10.2337/diacare.10.5.622

    work page doi:10.2337/diacare.10.5.622 1987

  12. [12]

    Glucose Variability Analysis in Two Large- Scale and Real-World Data Sets of Open-Source Automated Insulin Delivery Systems

    D. Cooper, B. Reinhold, A. Shahid, et al. “Glucose Variability Analysis in Two Large- Scale and Real-World Data Sets of Open-Source Automated Insulin Delivery Systems”. In:Journal of Diabetes Science and Technology19.3 (2025), pp. 649–657.DOI: 10.1177/ 19322968231198871

    work page 2025

  13. [13]

    Availability of Open Dynamic Glycemic Data in the Field of Diabetes Research: A Scoping Review

    L. L. Del Giudice, A. Piersanti, C. Göbl, et al. “Availability of Open Dynamic Glycemic Data in the Field of Diabetes Research: A Scoping Review”. In:Journal of Diabetes Science and Technology(2025), p. 19322968251316896.DOI:10.1177/19322968251316896

    work page doi:10.1177/19322968251316896 2025

  14. [14]

    , author Dong, W

    J. Deng, W. Dong, R. Socher, et al. “ImageNet: A Large-Scale Hierarchical Image Database”. In:Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2009.DOI:10.1109/CVPR.2009.5206848

    work page doi:10.1109/cvpr.2009.5206848 2009

  15. [15]

    Variation in Hypoglycemia Risk During Real-World Physical Activity in Adults with Type 1 Diabetes: Insights from the Type 1 Diabetes Exercise Initiative

    M. Dhaliwal, K. Tang, E. M. Aiello, et al. “Variation in Hypoglycemia Risk During Real-World Physical Activity in Adults with Type 1 Diabetes: Insights from the Type 1 Diabetes Exercise Initiative”. In:Diabetes Technology & Therapeutics(2025), p. 15209156251400209.DOI: 10.1177/15209156251400209

    work page doi:10.1177/15209156251400209 2025

  16. [16]

    Is It Time to Move Beyond TIR to TITR? Real-World Data from Over 20,000 Users of Continuous Glucose Monitoring in Patients with Type 1 and Type 2 Diabetes

    T. C. Dunn, R. A. Ajjan, R. M. Bergenstal, et al. “Is It Time to Move Beyond TIR to TITR? Real-World Data from Over 20,000 Users of Continuous Glucose Monitoring in Patients with Type 1 and Type 2 Diabetes”. In:Diabetes Technology & Therapeutics26.3 (2024), pp. 203–210.DOI:10.1089/dia.2023.0565

    work page doi:10.1089/dia.2023.0565 2024

  17. [17]

    Offline reinforcement learning for safer blood glucose control in people with type 1 diabetes

    H. Emerson, M. Guy, and R. McConville. “Offline reinforcement learning for safer blood glucose control in people with type 1 diabetes”. In:Journal of Biomedical Informatics142 (2023), p. 104376.DOI:10.1016/j.jbi.2023.104376

    work page doi:10.1016/j.jbi.2023.104376 2023

  18. [18]

    RSSDI consensus recommendations on insulin therapy in the management of diabetes

    Expert panel (extended), R. Chawla, B. M. Makkar, et al. “RSSDI consensus recommendations on insulin therapy in the management of diabetes”. In:International Journal of Diabetes in Developing Countries39.S2 (2019), pp. 43–92.DOI:10.1007/s13410-019-00783-6

    work page doi:10.1007/s13410-019-00783-6 2019

  19. [19]

    J. Fu, M. Norouzi, O. Nachum, et al.Benchmarks for Deep Off-Policy Evaluation. 2021.DOI: 10.48550/arXiv.2103.16596. arXiv:2103.16596

    work page doi:10.48550/arxiv.2103.16596 2021

  20. [20]

    Comparing Machine Learning Techniques for Blood Glucose Forecasting Using Free-living and Patient Generated Data

    H. Hameed and S. Kleinberg. “Comparing Machine Learning Techniques for Blood Glucose Forecasting Using Free-living and Patient Generated Data”. In:Proceedings of Machine Learning Research. V ol. 126. 2020, pp. 871–894

    work page 2020

  21. [21]

    Behavior,

    Trevor Hastie, Robert Tibshirani, and Jerome Friedman.The Elements of Statistical Learning. Springer Series in Statistics. New York, NY: Springer, 2009.DOI: 10.1007/978-0-387- 84858-7

    work page doi:10.1007/978-0-387- 2009

  22. [22]

    jaeb.org/datasets/diabetes(visited on 10/26/2025)

    JAEB Center for Health Research.Diabetes Datasets and Documents.URL: https://public. jaeb.org/datasets/diabetes(visited on 10/26/2025)

    work page 2025

  23. [23]

    Control-Relevant Models for Glucose Control Using A Priori Patient Characteristics

    K. van Heusden, E. Dassau, H. C. Zisser, et al. “Control-Relevant Models for Glucose Control Using A Priori Patient Characteristics”. In:IEEE Transactions on Biomedical Engineering 59.7 (2012), pp. 1839–1849.DOI:10.1109/TBME.2011.2176939

    work page doi:10.1109/tbme.2011.2176939 2012

  24. [24]

    HUPA-UCM diabetes dataset

    J. I. Hidalgo, J. Alvarado, M. Botella, et al. “HUPA-UCM diabetes dataset”. In:Data in Brief 55 (2024), p. 110559.DOI:10.1016/j.dib.2024.110559. 21 APREPRINT- JANUARY19, 2026

    work page doi:10.1016/j.dib.2024.110559 2024

  25. [25]

    Artificial intelligence and machine learning for improving glycemic control in diabetes: best practices, pitfalls and opportunities

    Peter G. Jacobs, Pau Herrero, Andrea Facchinetti, et al. “Artificial intelligence and machine learning for improving glycemic control in diabetes: best practices, pitfalls and opportunities”. In:IEEE Reviews in Biomedical Engineering(2023). Early Access.DOI: 10.1109/RBME. 2023.3331297

    work page doi:10.1109/rbme 2023

  26. [26]

    n.d.URL: https://public

    JAEB Center for Health Research.A Multi-Center Study of the Control-IQ Closed Loop Control System in Children with Type 1 Diabetes (DCLP5). n.d.URL: https://public. jaeb.org/dataset/535(visited on 01/12/2026)

    work page 2026

  27. [27]

    n.d.URL:https://public.jaeb.org/dataset/546(visited on 01/12/2026)

    JAEB Center for Health Research.A Randomized Trial Comparing Continuous Glucose Monitoring With and Without Routine Blood Glucose Monitoring in Adults with Type 1 Diabetes. n.d.URL:https://public.jaeb.org/dataset/546(visited on 01/12/2026)

    work page 2026

  28. [28]

    n.d.URL: https://public.jaeb

    JAEB Center for Health Research.An Observational Study of Individuals with Type 1 Diabetes Using the Loop System for Automated Insulin Delivery. n.d.URL: https://public.jaeb. org/dataset/560(visited on 01/12/2026)

    work page 2026

  29. [29]

    PID + Fuzzy Logic) in Individuals with Type 1 Diabetes

    JAEB Center for Health Research.FLAIR — Fuzzy Logic Automated Insulin Regulation: A Crossover Study Comparing Two Automated Insulin Delivery System Algorithms (PID vs. PID + Fuzzy Logic) in Individuals with Type 1 Diabetes. n.d.URL: https://public.jaeb.org/ dataset/566(visited on 01/12/2026)

    work page 2026

  30. [30]

    n.d.URL: https://public.jaeb.org/ dataset/581(visited on 01/12/2026)

    JAEB Center for Health Research.Pilot Study 3 of Outpatient Control-to-Range: Safety and Efficacy with Day-and-Night In-Home Use (CTR3). n.d.URL: https://public.jaeb.org/ dataset/581(visited on 01/12/2026)

    work page 2026

  31. [31]

    n.d.URL: https://public.jaeb.org/ dataset/579(visited on 01/12/2026)

    JAEB Center for Health Research.The Insulin-Only Bionic Pancreas Pivotal Trial: Testing the iLet in Adults and Children with Type 1 Diabetes. n.d.URL: https://public.jaeb.org/ dataset/579(visited on 01/12/2026)

    work page 2026

  32. [32]

    n.d.URL: https://public.jaeb.org/dataset/573 (visited on 01/12/2026)

    JAEB Center for Health Research.The International Diabetes Closed Loop (iDCL) Trial: Clinical Acceptance of the Artificial Pancreas - A Pivotal Study of t:Slim X2 with Control-IQ Technology (DCLP3). n.d.URL: https://public.jaeb.org/dataset/573 (visited on 01/12/2026)

    work page 2026

  33. [33]

    n.d.URL: https://public.jaeb

    JAEB Center for Health Research.The Pediatric Artificial Pancreas (PEDAP) Trial of Control- IQ Technology in Young Children in Type 1 Diabetes. n.d.URL: https://public.jaeb. org/dataset/599(visited on 01/12/2026)

    work page 2026

  34. [34]

    2025.DOI: 10

    Sam Gordon James.BrisT1D-Open Dataset. 2025.DOI: 10 . 5523 / bris . 33z5jc8fa6tob21ptrugzqog08

    work page 2025

  35. [35]

    Khamesian, A

    S. Khamesian, A. Arefeen, B. M. Thompson, et al.AZT1D: A Real-World Dataset for Type 1 Diabetes. Version 1. 2025.DOI:10.17632/gk9m674wcx.1

    work page doi:10.17632/gk9m674wcx.1 2025

  36. [36]

    Neural-Net Artificial Pancreas: A Random- ized Crossover Trial of a First-in-Class Automated Insulin Delivery Algorithm

    B. P. Kovatchev, A. C. Frasquet, E. C. Pryor, et al. “Neural-Net Artificial Pancreas: A Random- ized Crossover Trial of a First-in-Class Automated Insulin Delivery Algorithm”. In:Diabetes Technology & Therapeutics(2024).DOI:10.1089/dia.2023.0469

    work page doi:10.1089/dia.2023.0469 2024

  37. [37]

    Limitations in Achieving Glycemic Targets From CGM Data and Persistence of Severe Hypoglycemia in Adults With Type 1 Diabetes Regardless of Insulin Delivery Method

    L. M. Laffel, J. L. Sherr, J. Liu, et al. “Limitations in Achieving Glycemic Targets From CGM Data and Persistence of Severe Hypoglycemia in Adults With Type 1 Diabetes Regardless of Insulin Delivery Method”. In:Diabetes Care48.2 (2025), pp. 273–278.DOI: 10.2337/dc24- 1474

    work page doi:10.2337/dc24- 2025

  38. [38]

    Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems

    S. Levine, A. Kumar, G. Tucker, et al.Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems. 2020.DOI: 10.48550/arXiv.2005.01643 . arXiv: 2005.01643

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2005.01643 2020

  39. [39]

    GluNet: A Deep Learning Framework for Accurate Glucose Forecasting

    K. Li, C. Liu, T. Zhu, et al. “GluNet: A Deep Learning Framework for Accurate Glucose Forecasting”. In:IEEE Journal of Biomedical and Health Informatics24.2 (2020), pp. 414– 423.DOI:10.1109/JBHI.2019.2931842

    work page doi:10.1109/jbhi.2019.2931842 2020

  40. [40]

    Artificial intelligence for diabetes management – a review

    S. Maheshwari, A. Kalia, J. Tewari, et al. “Artificial intelligence for diabetes management – a review”. In:JDMDC12.1 (2025), pp. 24–32.DOI:10.15406/jdmdc.2025.12.00292

    work page doi:10.15406/jdmdc.2025.12.00292 2025

  41. [41]

    The OhioT1DM Dataset for Blood Glucose Level Prediction: Update 2020

    C. Marling and R. Bunescu. “The OhioT1DM Dataset for Blood Glucose Level Prediction: Update 2020”. In:CEUR Workshop Proceedings. V ol. 2675. 2020, pp. 71–74

    work page 2020

  42. [42]

    doi: 10.1093/jamia/ ocy068

    A. Neinstein, J. Wong, H. Look, et al. “A case study in open source innovation: developing the Tidepool Platform for interoperability in type 1 diabetes management”. In:Journal of the American Medical Informatics Association23.2 (2016), pp. 324–332.DOI: 10.1093/jamia/ ocv104. 22 APREPRINT- JANUARY19, 2026

    work page doi:10.1093/jamia/ 2016

  43. [43]

    Nudge BG.Babelbetes.URL: https://nudgebg.github.io/babelbetes/ (visited on 10/26/2025)

    work page 2025

  44. [44]

    An Exploratory Analysis of Continuous Glu- cose Monitoring Metrics in Relation to Prediabetes in Youths with Obesity

    C. Piona, E. M. Aiello, V . Mancioppi, et al. “An Exploratory Analysis of Continuous Glu- cose Monitoring Metrics in Relation to Prediabetes in Youths with Obesity”. In:Diabetes Technology & Therapeutics(2025).DOI:10.1177/15209156251407959

    work page doi:10.1177/15209156251407959 2025

  45. [45]

    Diatrend: A dataset from advanced diabetes technol- ogy to enable development of novel analytic solu- tions,

    T. Prioleau, A. Bartolome, R. Comi, et al. “DiaTrend: A dataset from advanced diabetes technology to enable development of novel analytic solutions”. In:Scientific Data10.1 (2023), p. 556.DOI:10.1038/s41597-023-02469-5

    work page doi:10.1038/s41597-023-02469-5 2023

  46. [46]

    Accessed 2025-11-07

    Replica Health.MetaboNet. Accessed 2025-11-07. n.d.URL: https://metabo-net.org/ data-dictionary

    work page 2025

  47. [47]

    Accessed 2025-10-26

    Replica Health.MetaboNet Processor. Accessed 2025-10-26. n.d.URL: https://github. com/replicahealth/metabonet_processor

    work page 2025

  48. [48]

    M. C. Riddell, Z. Li, R. L. Gal, et al. “Examining the Acute Glycemic Effects of Different Types of Structured Exercise Sessions in Type 1 Diabetes in a Real-World Setting: The Type 1 Diabetes and Exercise Initiative (T1DEXI)”. In:Diabetes Care46.4 (2023), p. 704.DOI: 10.2337/dc22-1721

    work page doi:10.2337/dc22-1721 2023

  49. [49]

    Large-Scale Data Analysis for Glucose Variability Outcomes with Open-Source Automated Insulin Delivery Systems

    A. Shahid and D. M. Lewis. “Large-Scale Data Analysis for Glucose Variability Outcomes with Open-Source Automated Insulin Delivery Systems”. In:Nutrients14.9 (2022), p. 1906. DOI:10.3390/nu14091906

    work page doi:10.3390/nu14091906 2022

  50. [50]

    Hypoglycemia Early Alarm Systems Based On Multivariable Models

    K. Turksoy, E. S. Bayrak, L. Quinn, et al. “Hypoglycemia Early Alarm Systems Based On Multivariable Models”. In:Industrial & Engineering Chemistry Research52.35 (2013), pp. 12329–12336.DOI:10.1021/ie3034015

    work page doi:10.1021/ie3034015 2013

  51. [51]

    Hierarchical Amortized GAN for 3D High Resolution Medical Image Synthesis,

    K. Turksoy, S. Samadi, J. Feng, et al. “Meal Detection in Patients With Type 1 Diabetes: A New Module for the Multivariable Adaptive Artificial Pancreas Control System”. In:IEEE Journal of Biomedical and Health Informatics20.1 (2016), pp. 47–54.DOI: 10.1109/JBHI. 2015.2446413

    work page doi:10.1109/jbhi 2016

  52. [52]

    A perspective on harmonizing diabetes management datasets

    M. K. Wolff, S. Royston, A. L. Fougner, et al. “A perspective on harmonizing diabetes management datasets”. In:Data in Brief59 (2025), p. 111399.DOI: 10.1016/j.dib.2025. 111399

    work page doi:10.1016/j.dib.2025 2025

  53. [53]

    GluPredKit: A Python Package for Blood Glucose Prediction and Evaluation

    M. K. Wolff, S. Royston, and R. V olden. “GluPredKit: A Python Package for Blood Glucose Prediction and Evaluation”. In:Journal of Open Source Software9.101 (2024), p. 6904.DOI: 10.21105/joss.06904

    work page doi:10.21105/joss.06904 2024

  54. [54]

    Blood Glucose Prediction Algorithms Require Clinically Relevant Performance Criteria Beyond Accuracy

    M. K. Wolff, H. G. Schaathun, S. Gros, et al. “Blood Glucose Prediction Algorithms Require Clinically Relevant Performance Criteria Beyond Accuracy”. In:Diabetes Technology & Therapeutics27.10 (2025), pp. 858–870.DOI:10.1089/dia.2025.0074

    work page doi:10.1089/dia.2025.0074 2025

  55. [55]

    Schalk, D.J

    J. Xie and Q. Wang. “Benchmarking Machine Learning Algorithms on Blood Glucose Pre- diction for Type I Diabetes in Comparison With Classical Time-Series Models”. In:IEEE Transactions on Biomedical Engineering67.11 (2020), pp. 3101–3124.DOI: 10.1109/TBME. 2020.2975959

    work page doi:10.1109/tbme 2020

  56. [56]

    Chinese diabetes datasets for data-driven machine learning

    Q. Zhao, J. Zhu, X. Shen, et al. “Chinese diabetes datasets for data-driven machine learning”. In:Scientific Data10.1 (2023), p. 35.DOI:10.1038/s41597-023-01940-7

    work page doi:10.1038/s41597-023-01940-7 2023

  57. [57]

    Offline Deep Reinforcement Learning and Off-Policy Evalua- tion for Personalized Basal Insulin Control in Type 1 Diabetes

    T. Zhu, K. Li, and P. Georgiou. “Offline Deep Reinforcement Learning and Off-Policy Evalua- tion for Personalized Basal Insulin Control in Type 1 Diabetes”. In:IEEE Journal of Biomedical and Health Informatics27.10 (2023), pp. 5087–5098.DOI:10.1109/JBHI.2023.3303367. 23

    work page doi:10.1109/jbhi.2023.3303367 2023