6. [10 pts] Here, we assume that D1 uses a programmable insulin infusion pump, instead of the

first one that was only able to generate a constant infusion rate. This new pump is capable of

infusing insulin in rectangular pulses of different durations and insulin amounts at various

times during the 24-hour period. Imagine you are subject D1, program your insulin pump to

achieve the following goals:

a) Bring your mean blood glucose level over 24 hours as close as possible to the

corresponding level in Subject N;

b) Reduce the fluctuation in blood glucose level (as measured by the standard deviation) to

as low as possible over the 24 hours; and

c) Minimize the total amount of insulin (in mU) infused over 24 hours, since insulin is an

expensive drug.

Explore a variety of insulin infusion patterns. After some trial and error, try to learn from the

results what you think are the best strategies. Select the 3 infusion patterns that best satisfy

the above criteria. For each pattern that you design, plot the trajectories of: i) glucose

infusion, ii) insulin infusion, iii) blood glucose concentration and iv) blood insulin

concentration over 24 hours of simulation.

For each of these three selected cases, quantify the metrics that correspond to the 3 criteria

(a, b and c) described above. Be sure you specify the formulas employed to arrive at each of

the 3 criteria.

Case

N (from Q.4)

I

24-h mean glucose

conc. (mg/mL)

24-hr glucose conc.

fluctuations (mg/mL)

Total insulin infused

over 24 hours (mU)

Briefly justify why you chose these 3 patterns to be your "best 3", and what considerations

you used to arrive at each pattern.

Save the Simulink model, the best of the 3 cases, as glucose_D1_pump.slx.