MetaboNet-Bench: A Multi-modal Benchmark for Glucose Forecasting in Type 1 Diabetes

2026-06-16

AI Summary

This study addresses the lack of a standardized multimodal evaluation benchmark in existing Type 1 diabetes glucose prediction research, which hinders fair model comparison. To bridge this gap, the authors propose MetaboNet-Bench—an open-source, extensible benchmark that, for the first time, integrates continuous glucose monitoring (CGM), insulin dosing, and carbohydrate intake data within a unified preprocessing, training, and evaluation pipeline. The framework systematically evaluates various state-of-the-art time series models, revealing the performance gains conferred by multimodal data—particularly for complex architectures—and incorporates clinically relevant metrics to expose critical limitations in current approaches. By establishing a consistent and reproducible foundation, MetaboNet-Bench provides both a standardized platform and clear guidance for future research in glucose prediction.

Abstract

Glucose forecasting algorithms are an important aspect of glycemic control management in type 1 diabetes. So far, the research community has developed numerous algorithms and models for forecasting. However, it is well-recognized that the lack of standardized model performance evaluation benchmarks makes fair comparison difficult and hinders further innovation, and thus benchmark standardization is in urgent need. Furthermore, many published glucose forecasting algorithms are limited to CGM data alone, ignoring other multimodal signals such as insulin dosing and carbohydrate intake. Here, we introduce MetaboNet-Bench, a benchmark for multimodal glucose forecasting for patients with type 1 diabetes that provides an extensible open-source evaluation framework for comparison of glucose forecasting algorithms that leverage glucose, insulin, and carbohydrate data. We then demonstrate its utility by benchmarking several recently published glucose forecasting models and a custom multimodal time-series model, representing different model architectures. The results show that the benefit of adding data modalities is conditioned on the complexity of the model and that incorporating more clinical metrics helps identify meaningful gaps to fill for future research.

Problem

• glucose forecasting
• type 1 diabetes
• multimodal benchmark
• standardized evaluation
• glycemic control

Innovation

• multimodal benchmark
• glucose forecasting
• type 1 diabetes
• model evaluation framework
• clinical time-series