Evaluating Time-Series Foundation Models and Multimodal Dietary Context for CGM Forecasting
Study finds zero-shot time-series foundation models underperform on CGM forecasting; fine-tuned Chronos-Bolt cuts RMSE up to 18.4% and dietary context adds signal.
The paper evaluates time-series foundation models for continuous glucose monitoring forecasting across eight public datasets covering Type 1 diabetes, Type 2 diabetes, and non-diabetes populations. Under a unified protocol, zero-shot foundation models did not consistently outperform baselines like Elastic Net and PatchTST, but lightweight fine-tuning did, with fine-tuned Chronos-Bolt reducing RMSE by 6.5%-18.4% in the T1D cohort and 8.6%-18.2% in the non-diabetes/T2D cohort. A residual-based fusion framework adding dietary context from CGMacros reduced overall RMSE by about 3% and postprandial RMSE by about 15% versus CGM-only baselines.
- Evaluation spans eight public CGM datasets including T1D and T2D cohorts
- Fine-tuned Chronos-Bolt cut RMSE 6.5-18.4% in the T1D cohort
- Dietary fusion reduced overall RMSE ~3% and postprandial RMSE ~15%
- Zero-shot foundation models lagged Elastic Net and PatchTST baselines
- Products
- CGMacros
- AI models
- Chronos-BoltPatchTST
From the original article
1 min read at arxiv.orgContinuous glucose monitoring (CGM) provides high-frequency measurements of glucose dynamics and enables short-term glucose forecasting for diabetes management. Although time-series foundation models have shown strong general forecasting ability, their effectiveness for CGM prediction and the added value of multimodal dietary context remain unclear.
By Bowen Zhang · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.