Ventana software support forum

Hi folks,
I'm trying to implement a model for HIV/AIDS with a group of students, can anyone provide some example of models for this topic?

Many thanks in advance

G

Hi,

There is one very similar to the HIV/AIDS model you are trying to implement, used by the Vensim PLE Online Tutorial at
http://www.shodor.org/tutorials/vensim/pre.php, too simple as it is built.

Perhaps you may be familar with this model, but I'd like to describe it again to warm up brains of netizens interested in your problem if necessary. As an example of system dynamics applied by the online tutorial, this model assumes that:
On a Monday 100 students return to school, one of whom has the flu. As time passes, the sick student unwittingly infects his classmates, who in turn infect others. This could be generalized as follows:
[align=center]Over time, susceptible people get sick and become infected people.[/align]

To begin building the model, create two box variables: one for Susceptible People and one for Infected People. These will allow you to keep track of how many healthy and sick people there are.
......

Carefully following the instructions stated in this online tutorial, I conducted the model one step after another without any skip or omission. Unfortunately, this failed to lead me to success in modeling this elementary SD problem, so as to result in "ERROR: No units specified for - Get sick" when "Checking Units" is conducted. According to this online tutorial, the rate Get Sick is defined as follows:

1 Assume that every susceptibleperson interacts with every infected person once per day. This can be expressed
mathematically as Susceptible People * Infected People. Not every interaction results in an infection though, so multiply this total number of interactions by the infection fraction to get
the number of people who get sick.

2 Following the PROCEDURE: with the Equations button still selected, click on Get Sick. Notice that the other variables in the system are listed in the box on the right of the input window. Because you connected these variables to this rate, they must be used in the formula you build. You can add them to your formula by clicking once on them. Enter the following formula:
[align=center]Susceptible People * Infected People * Infection Fraction[/align]
Leaving the units blank for now, click OK to finish.

Please pay attention on the said statement "Leaving the units blank for now, click OK to finish". This is why "ERROR: No units specified for - Get sick" when "Checking Units" has to be caused.

To solve this error problem, I tried such all the units available in the model as 1/(people*day), Day, Donth, People, and Dmn1, but failed to correct the error until now. The system still reports errors again and again.

Can anyony kindly offer me a hand? Thanks in advance.

Encl.: Systems Dynamics Using Vensim Personal Learning Edition (PLE) Quick Start Tutorial

[Edited on 11-7-2010 by CATS_Whisker]

Can you post the model that has the error unit?
JJ

GREAT! SO FAST ECHO! Here comes the model with said error unit. 100s of thanks for your kind concern.

[Edited on 11-7-2010 by CATS_Whisker]

Hi

joined the corrected model.

The infection fraction has unit 1/day and the infected people influence the get sick rate, not directly as in your model but influencing the infection fraction, eventually by the ratio infected by susceptibles via a lookup table.
Regards.
JJ

JJ,

Thanks for your kind gift of updated model. It works soundly okey. But you may failed to correct the model I built following instructions of the tutorial provided by vensim.com and downloadable at vensim.com/freedownload.html.

My question on the example model used by the tutorial is why a clear statement is made that "Leaving the units blank for now, click OK to finish", which encounters barrier (or ERROR: No units specified for - Get sick) when "Checking Units" is conducted.

Your comment that the infection fraction has unit 1/day is not applicable for my case, because replacing 1/(people*day) (or unit of infection fraction used in my model) with 1/day will still ends in identical barrier which goes "ERROR: No units specified for - Get sick".

I will be grateful if you are kind enough to correct this error in the un-updated model. Thanks again for your generosity to give the updated model to netizen friends as a gift.

Regards

Cats Whisker

Hi

One first remark: whatever the context, if you enter a variable with a blank unit, it will always report an error.
Maybe there is an error in the tutorial. I will try to find the tutorial you are talking about.
Second remark: Joined two models: the first is with the fraction with a unit equal to 1/day with no link from infected people and a second model that is your first model with a unit for the get sick. One can consider two different meanings for the fraction. One is that it is the probability of a susceptible to be infected in one day and a second that it is the probability of a susceptible person to be infected if there is only one infected person. This formulation is not wrong mathematically speaking, but it does not make sense.
So the logical formulation of the fraction unit is 1/day.
There is of course a link between the infected people and the get sick rate but one must formulate it otherwise, like the model I first posted.
The majority of modellers do not understand the nature and utility of dimensions. The dimension of a variable is a characteristic of that variable that is independant from the equations. Coyle in his book recommends to set the dimension of a variable (or constant or level) at the qualitative diagram step, when there are not yet any equations written, and then afterwards to verify if there is no unit errors. So the unit checking is not a unit check but an equations check, provided that the dimensions are correctedly formulated. One can too consider that it is a cross checking of units and equations that verify the coherence of both.
Regards.
JJ

[Edited on 8-11-2010 by LAUJJL]

Hi, thanks for your soonest reply and here come the VensimTut.pdf attach for your convenience. I will make a perusal of your new model tomorrow. - CW

I checked the tutorial that you posted. There is noting to say about the leave the unit blank. It does not say explicitly to use people/day as unit.
it specifies that fraction as 1/(people*day) for instructional reasons, even if one thinks a bit it has no sense.
The value of this fraction will be sensibly different if the total population varies. One infected person will have a greater effect if the total population is 100 or 100000! But I suppose the objective of the tutorial is to learn the principles of modeling without insisting on something that could be considered as too subtle for that level of learning.
Regards.
JJ

Hi

There is even a better and simple formulation.
The infection fraction is the probability to be infected if one suscecptible meets one infected, and to calculate the probability of one susceptible to be infected one must multiply the infection rate by the probability to meet one infected which is proportionnal to the ratio infected by the total population. One then has a correct unit formulation while keeping the proportionnality of the get sick with the number of infected.
Joined the model.
Regards.
JJ

Hi,

Thx for your feedback.

There is an EXPLICIT STATEMENT to "Leav(ing) the units blank for now, click OK to finish" on page 5 of the tutorial I posted. "Ctrl+F" instruction may help you find out this statement in the tutorial.

I agree on your comment that the model used by the tutorial specifies infection fraction as 1/(people*day) for instructional reasons.

It is a clever to add one more variable of total population in this simple model, so as not only to solve the unit error problem but also to make sense in case that total population becomes variable.

Thank you again for your susceptible_infected_people_5.mdl.

Regards
CW