Gluformer: Transformer-Based Personalized Glucose Forecasting with Uncertainty Quantification

TL;DR

Gluformer introduces a Transformer-based model for personalized blood glucose forecasting that effectively quantifies uncertainty, improving prediction accuracy especially for complex, multi-modal trajectories, facilitating clinical adoption.

Contribution

The paper presents a novel Transformer-based approach modeling glucose trajectories as infinite mixture distributions to incorporate uncertainty quantification.

Findings

Outperforms existing methods in accuracy on benchmark datasets.

Provides reliable uncertainty estimates for glucose predictions.

Demonstrates effectiveness on synthetic and real data.

Abstract

Deep learning models achieve state-of-the art results in predicting blood glucose trajectories, with a wide range of architectures being proposed. However, the adaptation of such models in clinical practice is slow, largely due to the lack of uncertainty quantification of provided predictions. In this work, we propose to model the future glucose trajectory conditioned on the past as an infinite mixture of basis distributions (i.e., Gaussian, Laplace, etc.). This change allows us to learn the uncertainty and predict more accurately in the cases when the trajectory has a heterogeneous or multi-modal distribution. To estimate the parameters of the predictive distribution, we utilize the Transformer architecture. We empirically demonstrate the superiority of our method over existing state-of-the-art techniques both in terms of accuracy and uncertainty on the synthetic and benchmark glucose data sets.