Uncertainty-aware Blood Glucose Prediction from Continuous Glucose Monitoring Data

In this work, we investigate uncertainty-aware neural network models for blood glucose prediction and adverse glycemic event identification in Type 1 diabetes. We consider three families of sequence models based on LSTM, GRU, and Transformer architectures, with uncertainty quantification enabled by either Monte Carlo dropout or through evidential output layers compatible with Deep Evidential Regression. Using the HUPA-UCM diabetes dataset for validation, we find that Transformer-based models equipped with evidential output heads provide the most effective uncertainty-aware framework, achieving consistently higher predictive accuracies and better-calibrated uncertainty estimates whose magnitudes significantly correlate with prediction errors. We further evaluate the clinical risk of each model using the recently proposed Diabetes Technology Society error grid, with risk categories defined by international expert consensus. Our results demonstrate the value of integrating principled uncertainty quantification into real-time machine-learning-based blood glucose prediction systems.

• Publication year

  2026

• Venue

  Unknown venue

• Publication date

  2026-03-05

• Fields of study

  Medicine, Physics, Computer Science

• Identifiers
  arXiv 2603.04955
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Data-Efficiency with Comparable Accuracy: Personalized LSTM Neural Network Training for Blood Glucose Prediction in Type 1 Diabetes Management

  2025influential reference
• Intermittently scanned continuous glucose monitoring provides no benefit over structured self-monitoring of blood glucose in type 2 diabetes not on prandial insulin, in the context of diabetes self-management education: GLucose monitoring programme SingaporE (GLiMPSE).

  2024cited by this paper
• HUPA-UCM diabetes dataset

  2024influential reference
• The Diabetes Technology Society Error Grid and Trend Accuracy Matrix for Glucose Monitors

  2024cited by this paper
• Deep Learning-Based Glucose Prediction Models: A Guide for Practitioners and a Curated Dataset for Improved Diabetes Management

  2024cited by this paper
• Deep evidential learning for radiotherapy dose prediction

  2024cited by this paper
• An LSTM-Based Neural Network Wearable System for Blood Glucose Prediction in People With Diabetes

  2023cited by this paper
• 6. Glycemic Goals and Hypoglycemia: Standards of Care in Diabetes-2024.

  2023cited by this paper
• Machine Learning Models for Blood Glucose Level Prediction in Patients With Diabetes Mellitus: Systematic Review and Network Meta-Analysis

  2023influential reference
• Continuous glucose monitoring and metrics for clinical trials: an international consensus statement.

  2022influential reference
• Personalized Blood Glucose Prediction for Type 1 Diabetes Using Evidential Deep Learning and Meta-Learning

  2022influential reference
• The OhioT1DM Dataset for Blood Glucose Level Prediction: Update 2020

  2020cited by this paper
• Assessing Reliability and Challenges of Uncertainty Estimations for Medical Image Segmentation

  2019cited by this paper
• Deep Evidential Regression

  2019influential reference
• Beyond HbA1c: Comparing Glycemic Variability and Glycemic Indices in Predicting Hypoglycemia in Type 1 and Type 2 Diabetes

  2018cited by this paper
• Evidential Deep Learning to Quantify Classification Uncertainty

  2018cited by this paper
• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?

  2017cited by this paper
• International Consensus on Use of Continuous Glucose Monitoring

  2017influential reference
• The changing face of diabetes complications.

  2016cited by this paper
• Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

  2015cited by this paper
• Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

  2014cited by this paper
• Strictly Proper Scoring Rules, Prediction, and Estimation

  2007cited by this paper
• Elements of Information Theory

  2005cited by this paper
• Long Short-Term Memory

  1997cited by this paper
• On Information and Sufficiency

  1997cited by this paper
• Bayesian Interpolation

  1992cited by this paper
• Evaluating Clinical Accuracy of Systems for Self-Monitoring of Blood Glucose

  1987influential reference
• VERIFICATION OF FORECASTS EXPRESSED IN TERMS OF PROBABILITY

  1950cited by this paper

• No citing papers are available for this paper.