MetaboNet-Bench
An open, extensible benchmark for multimodal glucose forecasting in type 1 diabetes that standardizes evaluation across models leveraging continuous glucose monitoring, insulin dosing and carbohydrate-intake signals, addressing the field's lack of fair, reproducible comparison frameworks for glycemic-control algorithms.
Composite
53.0
Experimental validation
Retrospective
Stages
Clinical Development (cross-phase)Post-market / RWE
Modalities
small molecule
Task types
regression
Size
modalities_signals: 3
splits: {'train': 0, 'val': 0, 'test': 0}
note: Multimodal glucose/insulin/carbohydrate time-series; open-source evaluation framework benchmarking several recently published forecasting models plus a custom multimodal time-series baseline. Exact patient/record counts to be confirmed from the released dataset.
splits: {'train': 0, 'val': 0, 'test': 0}
note: Multimodal glucose/insulin/carbohydrate time-series; open-source evaluation framework benchmarking several recently published forecasting models plus a custom multimodal time-series baseline. Exact patient/record counts to be confirmed from the released dataset.
License
Other — open-source evaluation framework (license to be confirmed from release)
First release
2026-06-17
Last updated
2026-06-17
Official site
Leaderboard
→ leaderboard
Dataset
→ dataset
Code / GitHub
→ repository
HuggingFace
→ HF
Paper
MetaboNet-Bench: A Multi-modal Benchmark for Glucose Forecasting in Type 1 Diabetes · Nathaniel Jeffries, Miriam Wolff, Sam Royston, Elizabeth Healey, Caleb Mayer, David Klonoff, Michael Snyder, Tao Wang · 2026 · paper · doi:N/A — arXiv preprint 2606.18640 · 0 citations
Flags
none
Experts
—
Groups
—
Hosted by
—
Related benchmarks
Rubric (7-criterion)
rigor
4
coverage
3
maintenance
2
adoption
1
quality
4
accessibility
4
industry_relevance
3
Notes
Fills a genuine gap: standardized, multimodal (glucose+insulin+carbs) evaluation for T1D glucose forecasting where prior work was CGM-only and incomparable (rigor 4, quality 4). Open-source extensible framework benchmarking multiple published models (accessibility 4). Senior authorship (Snyder, Klonoff) lends credibility. New (Jun 2026), no citations/leaderboard yet (maintenance 2, adoption 1); narrow to one indication and a forecasting task (coverage 3). Clinically relevant device/biomarker domain but not yet pharma-program-validated (industry 3).