Page 104 - Glucose Monitoring Devices

Applications of the SMBG measurement error models 103

therapies to understand how much errors in SMBG readings can affect the calcula-
tion of insulin doses, and thus the quality of glycemic control [37,42]. Other possible
applications include assessing the influence of SMBG accuracy and precision on
calibration algorithms for CGM sensors and algorithms that optimize SMBG-
based insulin dosing [48] or the frequency of SMBG testing [49].
An interesting application of the model of the BCN measurement error derived
by Vettoretti et al. [40], and presented in this chapter, is its incorporation in the T1D
patient decision simulator, a mathematical model of T1D patients making treatment
decisions based on glucose monitoring devices, which has been recently developed
by our research group at the University of Padova [50]. The T1D patient decision
simulator, schematized in Fig. 5.12, receives input meal data (I) and patient-
specific parameters describing both the patient’s physiology (P1) and therapy
(P2). The simulator output is the patient’s BG concentration (O), which is simulated
every minute, from which glycemic outcomes can be calculated. The T1D patient
decision simulator includes four main components:
A. the UVA/Padova T1D simulator that describes glucoseeinsulineglucagon
dynamics in T1D subjects [26,27];
B. a model of the device used for glucose monitoring that simulates SMBG [40]
and/or CGM measurements [31];
C. a model of the patient’s behavior in making treatment decisions, for example,
insulin dosing and hypoglycemia treatments;
D. a model of insulin delivery, for example, by insulin pump or multiple daily
injections.

FIGURE 5.12
Schematic representation of the T1D patient decision simulator.
From Vettoretti M, Facchinetti A, Sparacino G, Cobelli C. Type-1 diabetes patient decision simulator for in silico
testing safety and effectiveness of insulin treatments. IEEE Transactions on Biomedical Engineering 2018;
65(6):1281e1290; with permission.