Convolutional Recurrent Neural Networks for Glucose Prediction

• Published in: IEEE journal of biomedical and health informatics Vol. 24; no. 2; pp. 603 - 613
• Main Authors: Li, Kezhi, Daniels, John, Liu, Chengyuan, Herrero, Pau, Georgiou, Pantelis
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.02.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Artificial neural networks; Blood glucose; Blood Glucose - metabolism; Calculators; Carbohydrates; Cases (containers); Computer simulation; continuous glucose monitor (CGM); Convolution; Data models; Datasets; Deep Learning; Diabetes; Diabetes mellitus; Diabetes mellitus (insulin dependent); Diabetes Mellitus, Type 1 - blood; Drug delivery systems; Glucose; Glucose monitoring; glucose prediction; Horizon; Humans; Insulin; Insulin - blood; Learning algorithms; long short term memory (LSTM); Machine learning; Medical research; Methyltestosterone; Neural networks; Neural Networks, Computer; Pancreas; Patients; Predictions; Recurrent neural networks; Root-mean-square errors; Smartphones; Sugar; Time lag; Time series analysis; Type 1 diabetes
• ISSN: 2168-2194; 2168-2208; 2168-2208
• DOI: 10.1109/JBHI.2019.2908488

Control of blood glucose is essential for diabetes management. Current digital therapeutic approaches for subjects with type 1 diabetes mellitus such as the artificial pancreas and insulin bolus calculators leverage machine learning techniques for predicting subcutaneous glucose for improved control. Deep learning has recently been applied in healthcare and medical research to achieve state-of-the-art results in a range of tasks including disease diagnosis, and patient state prediction among others. In this paper, we present a deep learning model that is capable of forecasting glucose levels with leading accuracy for simulated patient cases (root-mean-square error (RMSE) = 9.38 ± 0.71 [mg/dL] over a 30-min horizon, RMSE = 18.87 ± 2.25 [mg/dL] over a 60-min horizon) and real patient cases (RMSE = 21.07 ± 2.35 [mg/dL] for 30 min, RMSE = 33.27 ± 4.79% for 60 min). In addition, the model provides competitive performance in providing effective prediction horizon (PHeff) with minimal time lag both in a simulated patient dataset (PH eff = 29.0 ± 0.7 for 30 min and PHeff = 49.8 ± 2.9 for 60 min) and in a real patient dataset (PH eff = 19.3 ± 3.1 for 30 min and PH eff = 29.3 ± 9.4 for 60 min). This approach is evaluated on a dataset of ten simulated cases generated from the UVA/Padova simulator and a clinical dataset of ten real cases each containing glucose readings, insulin bolus, and meal (carbohydrate) data. Performance of the recurrent convolutional neural network is benchmarked against four algorithms. The proposed algorithm is implemented on an Android mobile phone, with an execution time of 6 ms on a phone compared to an execution time of 780 ms on a laptop.