BioModels Database logo

BioModels Database


Dalla Man et al., (2007). Meal simulation model of the glucose-insulin system.

January 2015, model of the month by Maciej Swat
Original model: BIOMD0000000379


A whole-body model of a glucose-insulin control system applied to normal and diseased subjects is presented here [1]. It is capable of describing the physiological process in normal humans as consequence of a standard mixed meal, which is much more difficult to handle mathematically than standard oral or intravenous glucose tests. The model has been applied to a rich data set of 204 normal subjects and 14 type 2 diabetic subjects. See Figure 1 for measured and estimated glucose and insulin concentrations and fluxes.

Figure 1

Figure 1Whole-body glucose-insulin control system schema visualising measured plasma concentration of glucose and insulin in relation to various glucose fluxes, such as rate of appearance (Ra), production (EGP), utilization (U), renal extraction (E), and insulin fluxes, such as secretion (S) and degradation (D). Figure taken from [1].


The model (Figure 2) aims to establish a quantitative relationship between measured concentrations and rates/fluxes, such as

  • glucose rate of appearance (Ra), production (EGP), utilization (U), renal extraction (E), and
  • insulin fluxes, such as secretion (S) and degradation (D)

Because of the complexity of the system it is impossible to build a realistic model based on experimentally measured plasma concentration for glucose and insulin only. This is due to the fact that there are many different, but equally satisfying, descriptions of these concentrations based on various representations of glucose rate of absorption (Ra), production (EGP) and utilisation (U).

Only after the inclusion of glucose fluxes a reasonable description is possible. The authors used so called forcing function strategy and developed number of parametric models for every process of interest. Following sub-models have been developed (most of them are developed by the same group in the past): glucose and insulin subsytem, endogenous glucose production, glucose rate of appearance, glucose utilisation and insulin secretion.

It is important to remark that the model has limitations in that it doesn't take into account regulatory hormones such a glucagon, epinephrine or growth hormone. Other factors that are not considered are free fatty acids and their interaction with glucose and insulin.


Figure 3 shows the results for normal and type 2 diabetic subjects. In both case, the predicted concentrations or fluxes for glucose and insulin with mean +/-1SD confidence limits show good agreement between the model and the experimental data. Moreover, based on the model the prediction of various control features is possible, e.g. insulin-dependent and independent components of glucose utilisation or hepatic insulin extraction. For the diabetic patients a number of expected results has been confirmed, e.g. slower gut absorption or lower dynamic and static beta-cell responsivity.

The presented model analysis, although applied to a large data set, doesn't consider population based methods such as non-linear mixed effect models, NLME, which would allow to account for inter-individual variability and covariate influence such as sex, body weight, age to name the most frequently used once.

Figure 2

Figure 2 Average values +/-SD for 204 healthy subjects for five measured variables plasma glucose and insulin concentrations, glucose rate of appearance, endogenous glucose production and glucose utilization. The last profile is the computationally estimated insulin secretion using deconvolution. Figure taken from [1].

Figure 3

Figure 3Model prediction versus measurements of plasma concentrations and fluxes in healthy volunteers and diabetes type 2 subjects. Figure taken from [1].

Bibliographic references

  1. Dalla Man et al. Meal simulation model of the glucose-insulin system. IEEE Trans Biomed Eng. 2007 Oct;54(10):1740-9.