A meta-learning approach to personalized blood glucose prediction in type 1 diabetes. (June 2023)

• Record Type: Journal Article
• Authors: Langarica, Saúl; Rodriguez-Fernandez, Maria; Núñez, Felipe; Doyle, Francis J.
• Is Part Of: Control engineering practice. Volume 135(2023)
• Publication Date: 2023-06
• Subjects: Deep learning -- Meta learning -- Type 1 diabetes -- Artificial pancreas; Automatic control -- Periodicals; 629.89
• DOI: 10.1016/j.conengprac.2023.105498
• Languages: English
• ISSNs: 0967-0661

Abstract

Abstract: Accurate blood glucose prediction is a critical element in modern artificial pancreas systems. Recently, many deep learning-based models have been proposed for glucose prediction, showing encouraging results in population modeling. However, due to the large amount of data required for training deep learning-based models, few studies have successfully addressed personalized modeling, which is critical to ensure safe policies in a closed-loop scheme given the high inter-patient variability. To address this concern, we propose a meta-learning-based technique for accurate personalized modeling that requires minimal data volume to personalize from its population version, needs few training iterations, and has a low risk of over-fitting. Results using the UVA/Padova simulator show that the proposed technique generalizes better and outperforms other approaches in standard and task-specific metrics, particularly for longer prediction horizons and higher degrees of distributional shifts. Highlights: Glucose systems exhibit wide inter-and intra-patient variability. Population data-driven methods are not safe for control. Proposed method uses Meta-Learning to personalize to any number of patients. Our method surpass other techniques in several metrics. Our method surpass transfer-learning for personalize models.