62
4. Epidemics in the Marine System
OUTFLOWS:
INFECTION = CONTACT RATE*SUSCEPTIBLE *INFECTED
CONTACT_RATE = . 0 0 0 5
INFECTED_ONE = QELEM(INFECTED,l)
INFECTED_THREE = QELEM(INFECTED ,3)
I NFECTED_TWO = QELEM(INFECTED,2)
MORTALITY_RATE
. 1
TOTAL_POPULATION = IMMUNE + INFECTED + SUSCEPTIBLE
4.2. Sensitivity Analysis
4.2.1. Parameter Values and Initial Conditions
How sensitive are the model resu lts of the previous section to alternative
assumptions about initial conditions and parameter value s? To exp lore the
model's sensitivity, we cou ld vary-one at a time-the model's assumption,
run the model, and compare the results with previous runs. If we want to
choose a wide range of alternative parameter value s, then we need to run
the model man y times. Sensitivity analysis could become tedious were it
not for STELLA's ability to automate this process to some extent. For example, if we wish to expl ore how the severity and frequency of the disease
outbreak vary for thre e different contact rates, we can choose "Sensi Specs"
from the Run pull-d own menu and select CONTACT RATE as the parameter for which we wish to conduct the sensitivity analysis. Spe cify the number of run s as 3, then click on CONTACT RATE and specify a start and ending value for a set of "Incremental" changes in the cont act rate from run to
run . If you set the start value at .OOO5--Dur value from the previous model
run-and an ending value of .0010, then STELLA will choose an additional
value between these two to make up a total of three run s. For "Incremental" sen sitivity anal ysis, equally-spaced intermediate valuer s) will be chosen . Next, click on "Graph" and select INFECTED as the variable to be plotted for the three sensitivity runs .
Notice that when you want to run the model the menu now indicates "SRun," instead of "Run" in the pull-down menu. As you run the model, the
results for the three different assumptions are plotted in the same page of
the sensitivity graph. The results are shown in Figure 4.9.
Notice that for our choice of parameters the behavior of the disease is
relatively similar--outbreaks occur at roughly the same time and with approximately the same intensity, even though contact rates double from the
first to the third run . Will this result hold , in principle, for even higher contact rates? How sensitive are the model results to changes in initial conditions? How will the periodici ty and magnitude of disea se outbreaks change
4. Epidemics in the Marine System
OUTFLOWS:
INFECTION = CONTACT RATE*SUSCEPTIBLE *INFECTED
CONTACT_RATE = . 0 0 0 5
INFECTED_ONE = QELEM(INFECTED,l)
INFECTED_THREE = QELEM(INFECTED ,3)
I NFECTED_TWO = QELEM(INFECTED,2)
MORTALITY_RATE
. 1
TOTAL_POPULATION = IMMUNE + INFECTED + SUSCEPTIBLE
4.2. Sensitivity Analysis
4.2.1. Parameter Values and Initial Conditions
How sensitive are the model resu lts of the previous section to alternative
assumptions about initial conditions and parameter value s? To exp lore the
model's sensitivity, we cou ld vary-one at a time-the model's assumption,
run the model, and compare the results with previous runs. If we want to
choose a wide range of alternative parameter value s, then we need to run
the model man y times. Sensitivity analysis could become tedious were it
not for STELLA's ability to automate this process to some extent. For example, if we wish to expl ore how the severity and frequency of the disease
outbreak vary for thre e different contact rates, we can choose "Sensi Specs"
from the Run pull-d own menu and select CONTACT RATE as the parameter for which we wish to conduct the sensitivity analysis. Spe cify the number of run s as 3, then click on CONTACT RATE and specify a start and ending value for a set of "Incremental" changes in the cont act rate from run to
run . If you set the start value at .OOO5--Dur value from the previous model
run-and an ending value of .0010, then STELLA will choose an additional
value between these two to make up a total of three run s. For "Incremental" sen sitivity anal ysis, equally-spaced intermediate valuer s) will be chosen . Next, click on "Graph" and select INFECTED as the variable to be plotted for the three sensitivity runs .
Notice that when you want to run the model the menu now indicates "SRun," instead of "Run" in the pull-down menu. As you run the model, the
results for the three different assumptions are plotted in the same page of
the sensitivity graph. The results are shown in Figure 4.9.
Notice that for our choice of parameters the behavior of the disease is
relatively similar--outbreaks occur at roughly the same time and with approximately the same intensity, even though contact rates double from the
first to the third run . Will this result hold , in principle, for even higher contact rates? How sensitive are the model results to changes in initial conditions? How will the periodici ty and magnitude of disea se outbreaks change