Blood glucose forecasting in patients with insulin dependent diabetes mellitus with the Universal Process Modeling Algorithm
暂无分享,去创建一个
Acute and chronic complications in patients with IDDM result from insufficiently controlled blood glucose. In order to assist insulin therapy we applied the Universal Process Modeling Algorithm (UPMA) to forecast blood glucose in patients with IDDM. Two diabetic patients (one stable and one unstable) documented blood glucose, therapy, physical activity and diets with an electronic diary (handyDOC™) over 12 months each. That data of the preceeding month were used to predict blood glucose changes in the consecutive month. Predictions of the short-term blood glucose fluctuations were significantly correlated with the observed blood glucose changes, correlation coefficients ranged from 0.45 to 0.73 (p<0.01) in either patient. Extremes of blood glucose changes (i.e. hypo- and hyperglycemias) were identified correctly in the majority (>70%) of the events. In contrast, neither of two diabetologists was able to produce a reliable prediction of future blood glucose changes (0.05 < r < 0.12; p>0.l).
[1] Zlatko Trajanoski,et al. Einsatz von Neuronalen Netzwerken zur Blutzuckervorhersage bei Typ l Diabetes , 1991 .
[2] G. Wießpeiner,et al. System zur Dokumentation und Optimierung der Stoffwechseleinstellung bei Typ-I-Diabetes mellitus , 1990 .
[3] M. Egger,et al. Increasing Incidence of Hypoglycemic Coma in Children With IDDM , 1991, Diabetes Care.
[4] J. Skyler,et al. Algorithms for Adjustment of Insulin Dosage by Patients Who Monitor Blood Glucose , 1981, Diabetes Care.