Innovative approach for online prediction of blood glucose profile in type 1 diabetes patients

Direct continuous time system identification of MISO transfer function models applied to type 1 diabetes

“…Responses are due to ingestion of a 45g carbohydrate breakfast (mixed meal) and insulin injection based on patients own decision. The high intra-patient variability is evident in the estimated parameter values for the 3 breakfasts on 3 different days (15). …”

“…Advanced control strategies such as model predictive control, which was intensively analyzed in this context [11] [12] require accurate simulation models or at least prediction models with horizons in the magnitude of the dominant system dynamics which is several hours. The modeling effort was devoted to both physiology based modeling [13] and pure data based modeling [14] [15].…”

Adaptive subspace-based prediction of T1DM glycemia

“…However, none of these works considered the dynamic interplay between previously injected insulin, meal intake and eventually exercise. Attempts at including in the prediction model insulin and carbohydrates can be found in [7], [8], [9], [10], [11] and [12]. A different approach is that of [13] and [14] focusing on insulin sensitivity prediction and consequently on the resulting blood glucose prediction by means of stochastic models for critically ill patients.…”

“…Nevertheless, a peculiarity of diabetic subjects blood glucose dynamics is that it heavily varies over time, often quickly and unexpectedly. As a consequence, a linear-time-invariant model may not be sufficient to produce accurate forecasts of future glycemia [12]. In addition, a sound and valid patientspecific dynamical model of the glucose metabolism is still unavailable, despite extensive research [15].…”

Bayesian combination of multiple plasma glucose predictors

“…2 the isometric cost of the chose cost function for different prediction errors at different BG values has been plotted together with the Clarke Grid Plot. The boundaries Three different predictors of different structure were used in this study; a state-space-based model (SS) (model M 1 in [26]), a recursive ARX model [13] and a kernel-based predictor [21]. The SS and ARX models furthermore utilized inputs generated by population parametrized sub models describing the flux and digestion of insulin and glucose following an insulin injection or meal intake [9].…”