Zhu, T; Yao, X; Li, K; Herrero, P; Georgiou, P; (2020) Blood glucose prediction for type 1 diabetes using generative adversarial networks. In: Bach, K and Bunescu, R and Marling, C and Wiratunga, N, (eds.) Proceedings of the 5th International Workshop on Knowledge Discovery in Healthcare Data co-located with 24th European Conference on Artificial Intelligence (ECAI 2020). (pp. pp. 90-94). : Santiago de Compostela, Spain.

Preview

Text paper15.pdf - Published Version Download (1MB) | Preview

Abstract

Maintaining blood glucose in a target range is essential for people living with Type 1 diabetes in order to avoid excessive periods in hypoglycemia and hyperglycemia which can result in severe complications. Accurate blood glucose prediction can reduce this risk and enhance early interventions to improve diabetes management. However, due to the complex nature of glucose metabolism and the various lifestyle related factors which can disrupt this, diabetes management still remains challenging. In this work we propose a novel deep learning model to predict future BG levels based on the historical continuous glucose monitoring measurements, meal ingestion, and insulin delivery. We adopt a modified architecture of the generative adversarial network that comprises of a generator and a discriminator. The generator computes the BG predictions by a recurrent neural network with gated recurrent units, and the auxiliary discriminator employs a one-dimensional convolutional neural network to distinguish between the predictive and real BG values. Two modules are trained in an adversarial process with a combination of loss. The experiments were conducted using the OhioT1DM dataset that contains the data of six T1D contributors over 40 days. The proposed algorithm achieves an average root mean square error (RMSE) of 18.34 ± 0.17 mg/dL with a mean absolute error (MAE) of 13.37 ± 0.18 mg/dL for the 30-minute prediction horizon (PH) and an average RMSE of 32.31 ± 0.46 mg/dL with a MAE of 24.20 ± 0.42 for the 60-minute PH. The results are compared for clinical relevance using the Clarke error grid which confirms the promising performance of the proposed model.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item