Lesson 19


1 System Overview

The human body carefully maintains fasting blood glucose concentrations within a range of 80–90 mg/dl of blood. This process is called glucose homeostasis. After a meal, blood glucose concentration increases. The increase in glucose is sensed by β\beta-cells in the pancreas (Figure 1) and they begin to secrete insulin in response. Insulin enters the blood where it is distributed throughout the body and signals cells to increase their glucose uptake from the blood. The liver, in particular, absorbs a significant amount of glucose compared to other tissues. The net result is that the blood glucose level falls back to its normal range. As the blood glucose level falls, the β\beta-cells sense less glucose and in response secrete less insulin. Figure 2 shows the effect of meals on blood glucose and blood insulin levels over a 24 hour period.

Figure 1: Overview of pancreas structure. Beta cells comprise the vast majority of cells within the pancreas. (After Moffet, Moffet, and Schauf 1993).
Figure 2: Blood (plasma) glucose and insulin levels over time. Note how insulin levels lag the glucose levels. (After Moffet, Moffet, and Schauf 1993).

2 System Model

Assume we want to model glucose homeostasis and insulin homeostasis as a single compartment model. Since we wish to model two different mass species, we will need two differential equations.

The first differential equation is a mass balance for glucose (GLU). There is one input for GLU. It can enter the system through absorption from the gastrointestinal tract, intravenous infusion, or supply from the liver. There are two outputs for GLU: (1) passive diffusion into the liver and tissue (controlled by GLU concentration) and (2) facilitated diffusion into the tissues (controlled by insulin (INS) concentration), which can be 20x larger than the passive diffusion rate. Note that “controlled by” means “multiplied by”. Assume the coefficient δ\delta is the rate constant for passive diffusion and the coefficient γ\gamma is the rate constant for facilitated diffusion.

The second differential equation is a mass balance for insulin (INS). There are two inputs for INS: (1) intravenous infusion (injection), and (2) production in the pancreas due to the beta-cells of the islets of Langerhans. Output is by inactivation (consumption) by blood insulinase activity. Assume the coefficient α\alpha is the rate constant for consumption and β\beta is the rate constant for secretion from the pancreas.

3 Student Exercises

3.1 Student Exercise 1

Write the mass balance equations for glucose and insulin. Use GLUGLU as the variable for glucose mass and INSINS as the variable for insulin mass.

dGLUdt=?anddINSdt=?\frac{dGLU}{dt} = \;? \quad \text{and} \quad \frac{dINS}{dt} = \;?

3.2 Student Exercise 2

Convert the equations from Exercise 1 into Laplace notation and enter them into MATLAB. Use solve() to find expressions for GLUGLU and INSINS. Use the function collect() to simplify the expressions once you have solved for them. Rearrange the equations to get something that looks like the equations shown below.

GLU(s)=somefractionṁinGLU(s)somefractionṁinINS(s)GLU(s) = \frac{\text{some}}{\text{fraction}} \dot{m}_{inGLU}(s) - \frac{\text{some}}{\text{fraction}} \dot{m}_{inINS}(s)

and

INS(s)=somefractionṁinGLU(s)+somefractionṁinINS(s)INS(s) = \frac{\text{some}}{\text{fraction}} \dot{m}_{inGLU}(s) + \frac{\text{some}}{\text{fraction}} \dot{m}_{inINS}(s)

3.3 Student Exercise 3

From visual inspection of the transfer functions from Exercise 2, what is the order of each system?

3.4 Student Exercise 4

Build a model containing the equations from Exercise 2 in Simulink. To get you started, consider visualizing the first equation like this:

Equation diagrammed to indicate inputs and outputs to facilitate entry into simulink.

Use a total simulation time of 24 (hrs). Use the variable solver ode15s. Enter the following values for the coefficients into your model:

α=0.916 hr1β=0.198 unit/hr/gmγ=3.23 gm/hr/unitδ=0.304 gm/hr/gm\begin{aligned} \alpha &= 0.916 \text{ hr}^{-1} \\ \beta &= 0.198 \text{ unit/hr/gm} \\ \gamma &= 3.23 \text{ gm/hr/unit} \\ \delta &= 0.304 \text{ gm/hr/gm} \end{aligned}

Use SCOPE blocks to plot GLUGLU and INSINS. Assume ṁinGLU\dot{m}_{inGLU} is a step input with a value of 100 gm/hr and ṁinINS=0\dot{m}_{inINS} = 0. What type of system does each response suggest? Why?

3.5 Student Exercise 5

Plot GLUGLU and INSINS if ṁinINS\dot{m}_{inINS} is an insulin step input with a value of 100 gm/hr and ṁinGLU=0\dot{m}_{inGLU} = 0. What type of system does each response suggest? Why?

3.6 Student Exercise 6

What is the %OS for Exercises 4 and 5?

3.7 Student Exercise 7

Assume the glucose transfer function is TGLU(s)=GLU(s)ṁinGLUT_{GLU}(s) = \frac{GLU(s)}{\dot{m}_{inGLU}} when ṁinINS=0\dot{m}_{inINS} = 0. Assume TINS(s)=INS(s)ṁinINST_{INS}(s) = \frac{INS(s)}{\dot{m}_{inINS}} when ṁinGLU=0\dot{m}_{inGLU} = 0. Are TGLU(s)T_{GLU}(s) and TINS(s)T_{INS}(s) stable? Prove your answer.

3.8 Student Exercise 8

Determine the steady-state error for TGLU(s)T_{GLU}(s) and TINS(s)T_{INS}(s). For TGLU(s)T_{GLU}(s) SSE, assume ṁinINS=0\dot{m}_{inINS} = 0 and ṁinGLU\dot{m}_{inGLU} is a step input of amplitude 100 gm/hr. For TINS(s)T_{INS}(s) SSE, assume ṁinGLU=0\dot{m}_{inGLU} = 0 and ṁinINS\dot{m}_{inINS} is a step input of amplitude 100 gm/hr.