Eclipse Community Forums: Spatiotemporal Epidemiological Modeler (STEM)

Home » Eclipse Projects » Spatiotemporal Epidemiological Modeler (STEM) » Modifying Disease Parameters by Location

Modifying Disease Parameters by Location [message #631285]

Wed, 06 October 2010 16:02

Eclipse User

I posted a request to tie location as well as time to the disease parameter modification process, but I'm wondering if anyone else has tried to do this and succeeded with some workarounds? The problem is something like closing schools in certain locations from days 10-17 of a run and simulating that by a reduced transmission rate. But could we apply the reduced transmission rate only to the locations containing the relevant school districts?

For example, I was thinking of assuming that the closure happens at the start time and infecting with multiple versions of the disease (different transmission rates) based on location? Maybe the timing aspect could still be included by piecing together multiple runs.

Or could this be addressed at all with the MultiPopulationSEIRDIseaseModel or the DemographicPopulationModel? Creating different populations based on location and then treating the populations differently wrt the triggers/modfiers etc?

Re: Modifying Disease Parameters by Location [message #631522 is a reply to message #631285]

Thu, 07 October 2010 16:11

Eclipse User

The most straightforward way of fixing this with the least amount of code changes would be adding a location property to disease models. We do this for population models, the reasoning was that the background birth/death rate could be very different in say Africa compared to Europe.

But in your example you would need to create one disease model for each location you want to reduce the transmission rate in. Not sure how many we are talking about. Perhaps if the location property was multi-valued we could reduce this.

Now that I think some more about this I'm not sure the code changes would be so small. For instance, the mixing code between neighboring regions doesn't handle multiple disease models right now, they assume a single disease model.

Here's another idea. We could write a custom disease model that uses some attribute of the node itself to determine the transmission rate. We have a few examples of this were the transmission rate depends on the lat/lon of the node. A label on the node could serve as a flag determining whether schools are closed or not. The trigger would just flip the flag, and the disease model would have two attributes for transmission rate depending upon the value of the flag.

Unfortunately, it's not really easy to add any custom labels to nodes right now, the planned graph editor will simplify this.

Another idea is to isolate the node where the school is located on by cutting of all edges (or rather setting the mixing to 0). Unfortunately right now mixing is a parameter that is part of the disease model and not the edge itself.

I'm open to more suggestions. I don't see an easy way to use the population initializers or the multi-population disease models/demographic population models to "fake" the thing you're trying to do.

Regards.
/ Stefan

Re: Modifying Disease Parameters by Location [message #631707 is a reply to message #631522]

Fri, 08 October 2010 10:07

Eclipse User

Thanks Stefan. I like the node idea. Seems like the graph editor is a desired feature - coming back to some of the stuff we wanted to do with transportation networks as well. Traci

Previous Topic:	STEM 1.1.0 Released!
Next Topic:	Macdonald Ross disease plugins available

Goto Forum:

-=] Back to Top [=-

Current Time: Sat Jun 14 07:56:16 EDT 2025

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter