Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Archived » OHF » Using Stem to simulate dengue fever: two questions/problems
Using Stem to simulate dengue fever: two questions/problems [message #31820] Sun, 15 July 2007 22:17 Go to next message
Nicoletta De Maio is currently offline Nicoletta De Maio
Messages: 8
Registered: July 2009
Junior Member
Hi

I was told to post the following e-mail conversation to the newsgroup, so
there you go.

My original questions were:
----Quote----
Hello

I'm working in the bioinformatics group at NITD (Novartis Institute for
tropical Diseases) in Singapore and we're looking at using Stem to
simulate dengue fever epidemics. I'm the person assigned to evaluating
Stem to see if it may be useful to us.

I've collected as much of the documentation as I could find and it gave me
a pretty good understanding of the structure of a scenario, and of what's
available and what's not. But now I'm running into some practical problems.

First of all, I'd like to use the available geographic information for
Singapore rather than try and cobble something together myself. According
to the Models tab in the Designer View there is such a model (SGP (0,1,2)
Full infrastructure). But how do I get it into my project? I've created a
Scenario in the appropriate subfolder in the Project viewer and attached
the model to that. But I'm not able to copy it to the models folder within
my own project, at least not within the Designer view. (I haven't tried to
do it 'by hand', i.e. moving files around outside the running application,
by the way. I'm not sure that's wise...)

I need to do this because I'm planning to attach the Infector to the
scenario, but the Disease decorator to the model, or parts of it.

Which brings me to the next problem. I've used the wizard to create an
infector and dropped it in the Decorators folder of my project - but when
I try to do the same thing with a disease the dialog window will accept
all of my input before issuing an error message and do nothing when I
click 'finish'. The message name/description is simply the name of the
package 'org.eclipse.ohf.stem.ui'; there is no indication of what Stem
thinks happened. The stack trace is a bit long so here's the beginning of
it:

java.lang.reflect.InvocationTargetException
at
org.eclipse.jface.operation.ModalContext.runInCurrentThread( ModalContext.java:381)
at org.eclipse.jface.operation.ModalContext.run(ModalContext.ja va:313)
at org.eclipse.jface.wizard.WizardDialog.run(WizardDialog.java: 934)
at
org.eclipse.ohf.stem.ui.wizards.NewIdentifiableWizard.perfor mFinish(NewIdentifiableWizard.java:146)
at
org.eclipse.jface.wizard.WizardDialog.finishPressed(WizardDi alog.java:742)
at
org.eclipse.jface.wizard.WizardDialog.buttonPressed(WizardDi alog.java:373)
at org.eclipse.jface.dialogs.Dialog$2.widgetSelected(Dialog.jav a:616)
at
org.eclipse.swt.widgets.TypedListener.handleEvent(TypedListe ner.java:227)
at org.eclipse.swt.widgets.EventTable.sendEvent(EventTable.java :66)
at org.eclipse.swt.widgets.Widget.sendEvent(Widget.java:1101)
at org.eclipse.swt.widgets.Display.runDeferredEvents(Display.ja va:3320)
at org.eclipse.swt.widgets.Display.readAndDispatch(Display.java :2972)
at org.eclipse.jface.window.Window.runEventLoop(Window.java:820 )
at org.eclipse.jface.window.Window.open(Window.java:796)
at
org.eclipse.ohf.stem.ui.wizards.NewIdentifiableWizard$NewIde ntifiableWizardAction.run(NewIdentifiableWizard.java:1043)
at
org.eclipse.ohf.stem.ui.wizards.NewDiseaseWizard$NewAction.r un(NewDiseaseWizard.java:161)
at org.eclipse.ui.internal.PluginAction.runWithEvent(PluginActi on.java:256)
---<snip>---

(Still with me? Oh, good...:) )
I can't really make head or tail of it - do you have ideas what might be
going wrong? Oh, and I'm running Stem off the jarfile, not from within
eclipse.

So - the point of this e-mail. I'd appreciate any pointers or instructions
on how to solve either problem. I'm more than happy to provide further
information if necessary - just drop me a line. I'd really like to get
this off the ground (especially because so much geographical data is
already there) but it seems I have to get past the planning stage first!

By the way, feel free to forward all or part of this to someone else if
I've contacted the wrong developer. In that case - sorry to have bothered
you, it was not my intention to get on anyone's nerves. In any case:
thanks!

:-) Nicoletta De Maio

----Quote----

And here's Dan Ford's reply:
----Quote----
Nicoletta,
thanks for your note, you contacted the right person. I'm on
vacation right now with limited connectivity so I'll give a short reply
now and try to follow up with you next week when I return home. I also
you like to point you to the project newsgroup, it would be great for the
project to capture your questions and my answers for all to see. If you
want to post this response in the newsgroup that would be great.
http://www.eclipse.org/newsgroups/

1) with the current milestone release, 0.2.0M1, you cannot copy the
internal built-in models to user defined projects. I believe there is a
bug open on this issue, but a reference to will have to wait until next
week. This will be corrected, it just was one of the things we cut from
this release. You should, however, be able to drag it into a Model or
Scenario that you create with a wizard and it should work. The only
reason you would drag a copy to your own project would be to modify the
underlying model. The built-in version cannot be changed as other models
may refer to it and would not expect it to be different in any way.

There is a tutorial to be written that describes much of what you want to
accomplish, basically what you need to do is create a hierarchy of models
using the wizard and compose the hierarchy by dragging a child model on to
its parent. Try creating your lowest-level model and dragging a disease
model (one of the built in ones or one you create using the disease
wizard) into it. Then drag the built-in Singapore model and dropping it
there as well. Save that model. Create another model and in its editor
drag the built-in "Aggregating SEIR Disease Model" (decorator) to it.
This will aggregate values in the geographic hierarchy as well as
propagate population members along transportation links. Also, drag and
drop the first model you created to the new model's open editor to make it
a child. Save the second model. Create your Scenario using the Wizard,
and drag the second model you created to the Scenario in its editor and
drop it. Do the same with the Infector and the Sequencer you want to use.
Save it.

Double click on the new Scenario and if all is well, it will run.

2) The exception is a bug, could you open a bug report please. This will
help you follow its progress and help us track down and fix the problem.
Have a look at http://wiki.eclipse.org/Bug_Reporting_FAQ and if you have
any problems just send me a note.

3) I have some experience modeling Dengue fever. In the first version of
STEM (the eclipse version is the second), I implemented the model outlined
in:

Cummings DAT, Irizarry RA, Huang NE, Endy TP, Nisalak A, Ungchusak K,
Burke DS (2004) Travelling waves in the occurrence of dengue haemorrhagic
fever in Thailand. Nature 427 (6972): 344-347.

This model required a more sophisticated disease model than simple SEIR as
it segments the population by the sequence of serotypes that they have
been infected with. Unfortunately, that model is not yet implemented in
the new version of STEM.

Best Regards

Dan Ford
IBM Almaden
----Quote----

Hope this is useful to anyone besides myself.

:-) Nicoletta
Re: Using Stem to simulate dengue fever: two questions/problems [message #31858 is a reply to message #31820] Mon, 16 July 2007 05:17 Go to previous messageGo to next message
Nicoletta De Maio is currently offline Nicoletta De Maio
Messages: 8
Registered: July 2009
Junior Member
Hi Dan

First of all, thanks again for the help. Here's what came out of it:

1) Followed the instructions but when I run the scenario (which it lets me
do now) I get this bit of info (*not* a warning or an error) in the error
log:
-------------
Message: Aggregating SEIR Human didn't find a root node for the graph.
Missing containment relationship?
-------------
Important or not? (I don't think I've come across the term 'containment
relationship' in STEM before.) It doesn't stop the simulation as such but
I'm not sure how to tell if it is doing anything at all: Singapore is so
small that it is, for all practical purposes, impossible to find on the
map view if the rest of the world is not displayed and I have no point of
reference that tells me which part of the world I'm currently looking at.
Is there any way of centering the active i.e. selected area on the screen?
Or at least to display some sort of coordinate system? Or anything else
that would make finding a small island off the southern tip of Malaysia a
little easier? :)

2) Bug report has been submitted to Bugzilla. Wait and see...

3) See my e-mail. I've tracked down the paper you based the dengue
simulation on but I'm not sure I understand how you did it. I'm even more
interested in additional information now, as the paper does not list any
time-dependent differential equations of the sort that went into the SEIR
model.

:-) Nicoletta
Re: Using Stem to simulate dengue fever: two questions/problems [message #31893 is a reply to message #31858] Mon, 16 July 2007 11:17 Go to previous messageGo to next message
John Thomas is currently offline John Thomas
Messages: 14
Registered: July 2009
Junior Member
Nicoletta

Dan will be back from vacation on Wednesday and I am sure will be able
to help you make more progress..

In the meantime, I will try to answer one of of your questions.

The "missing containment relationship" is not important at this time.
Currently I believe that the US is the only area where we have
containment relationships fully specified. It refers to fully
specifying for each Administration area 2 which admin area 1 it belongs
to. and for each administration area 1 which administration area 0 it
belongs to. Currently the containment relationship is not being used
in the model so you can ignore the message.

One other comment. You might want to download the STEM source from the
Eclipse CVS site if you haven't already. That way if Dan does discover
some problems and fixes them, you will be able to use the fixed version.
The STEM wiki
http://wiki.eclipse.org/Installation_Guide
has the information that you will need to setup Eclipse and access the
STEM source code.

I hope this is helpful.

John


Nicoletta De Maio wrote:
> Hi Dan
>
> First of all, thanks again for the help. Here's what came out of it:
>
> 1) Followed the instructions but when I run the scenario (which it lets
> me do now) I get this bit of info (*not* a warning or an error) in the
> error log:
> -------------
> Message: Aggregating SEIR Human didn't find a root node for the graph.
> Missing containment relationship?
> -------------
> Important or not? (I don't think I've come across the term 'containment
> relationship' in STEM before.) It doesn't stop the simulation as such
> but I'm not sure how to tell if it is doing anything at all: Singapore
> is so small that it is, for all practical purposes, impossible to find
> on the map view if the rest of the world is not displayed and I have no
> point of reference that tells me which part of the world I'm currently
> looking at. Is there any way of centering the active i.e. selected area
> on the screen? Or at least to display some sort of coordinate system? Or
> anything else that would make finding a small island off the southern
> tip of Malaysia a little easier? :)
>
> 2) Bug report has been submitted to Bugzilla. Wait and see...
>
> 3) See my e-mail. I've tracked down the paper you based the dengue
> simulation on but I'm not sure I understand how you did it. I'm even
> more interested in additional information now, as the paper does not
> list any time-dependent differential equations of the sort that went
> into the SEIR model.
>
> :-) Nicoletta
>
Re: Using Stem to simulate dengue fever: two questions/problems [message #31928 is a reply to message #31893] Tue, 17 July 2007 05:29 Go to previous messageGo to next message
Nicoletta De Maio is currently offline Nicoletta De Maio
Messages: 8
Registered: July 2009
Junior Member
Hi

This is helpful in the sense that I don't have to worry about the message,
so thanks.

As for checking STEM out of the CVS repository, don't get me started - but
the problem is on my side (actually, our admin's...). I've got the source
code, though. I just hope I'll have access via CVS soon...

:-) Nicoletta
Re: Using Stem to simulate dengue fever: two questions/problems [message #32033 is a reply to message #31893] Wed, 18 July 2007 10:46 Go to previous messageGo to next message
Daniel Ford is currently offline Daniel Ford
Messages: 138
Registered: July 2009
Location: New York
Senior Member
Nicoletta,
the message you received was as John described. When the Aggregating
disease model walks the graph, it follows edges that represent geographic
"containment" relationships. The basic idea is that when a larger region
contains a number of smaller regions the state of the disease in the larger
region is the aggregate of the states of the smaller regions it contains.
The problem is that when it walked the graph it did not find a root node
representing the top-level of the heirarchy. I'll have to look at the data
set for Singapore, but I think John is right that we don't yet have the
contianment relationship specified completely.

Dan


"John Thomas" <jthomas119@gmail.com> wrote in message
news:469B8C11.6090904@gmail.com...
> Nicoletta
>
> Dan will be back from vacation on Wednesday and I am sure will be able to
> help you make more progress..
>
> In the meantime, I will try to answer one of of your questions.
>
> The "missing containment relationship" is not important at this time.
> Currently I believe that the US is the only area where we have containment
> relationships fully specified. It refers to fully specifying for each
> Administration area 2 which admin area 1 it belongs to. and for each
> administration area 1 which administration area 0 it belongs to.
> Currently the containment relationship is not being used in the model so
> you can ignore the message.
>
> One other comment. You might want to download the STEM source from the
> Eclipse CVS site if you haven't already. That way if Dan does discover
> some problems and fixes them, you will be able to use the fixed version.
> The STEM wiki
> http://wiki.eclipse.org/Installation_Guide
> has the information that you will need to setup Eclipse and access the
> STEM source code.
>
> I hope this is helpful.
>
> John
>
>
> Nicoletta De Maio wrote:
>> Hi Dan
>>
>> First of all, thanks again for the help. Here's what came out of it:
>>
>> 1) Followed the instructions but when I run the scenario (which it lets
>> me do now) I get this bit of info (*not* a warning or an error) in the
>> error log:
>> -------------
>> Message: Aggregating SEIR Human didn't find a root node for the graph.
>> Missing containment relationship?
>> -------------
>> Important or not? (I don't think I've come across the term 'containment
>> relationship' in STEM before.) It doesn't stop the simulation as such but
>> I'm not sure how to tell if it is doing anything at all: Singapore is so
>> small that it is, for all practical purposes, impossible to find on the
>> map view if the rest of the world is not displayed and I have no point of
>> reference that tells me which part of the world I'm currently looking at.
>> Is there any way of centering the active i.e. selected area on the
>> screen? Or at least to display some sort of coordinate system? Or
>> anything else that would make finding a small island off the southern tip
>> of Malaysia a little easier? :)
>>
>> 2) Bug report has been submitted to Bugzilla. Wait and see...
>>
>> 3) See my e-mail. I've tracked down the paper you based the dengue
>> simulation on but I'm not sure I understand how you did it. I'm even more
>> interested in additional information now, as the paper does not list any
>> time-dependent differential equations of the sort that went into the SEIR
>> model.
>>
>> :-) Nicoletta
>>
Re: Using Stem to simulate dengue fever [message #32174 is a reply to message #31858] Wed, 18 July 2007 11:40 Go to previous messageGo to next message
Daniel Ford is currently offline Daniel Ford
Messages: 138
Registered: July 2009
Location: New York
Senior Member
Nicoletta,
simulating Dengue Hemorrhagic Fever (DHF) is complicated because it is a
multiserotype disease that affects people differently depending upon their
personal serotype infection history. Basically, an initial infection with
any of the four different DHF serotypes will cause a similar "mild" reaction
in a person and give them immunity to that serotype. However, a second
infection by a different serotype will cause a much more severe, potentially
fatal, reaction and actually make the infected person more infectious to
others.

The problem with modeling such a disease is that you need to keep track of
the sequential serotype infection history of the population and to have
different infection and mortality rates depending upon the infection
history. This results in a combinatorial explosion in the number of states
that need to be represented and a much more complicated configuration and
implementation for the disease model. For example, in a simple SEIR model,
a population member can only be in one of four states (S, E, I, or R). In a
model for DHF, all population members start out in a single "S"
(Susceptible) state, but then they can transition to one of four "E" states
and then to the corresponding "I" state, of which there are four. From
there they will "recover" to the corresponding "R" state, of which there are
four. From there they can be exposed to any of the remaining three
serotypes and then transition to the appropriate "E" state that matches
their infection history, of which there are 12 (4x3) in total. This
continues to expand for all possible sequences of serotype infection until
the states eventually converge on a single "Recovered from everything"
state.

The first version of STEM contained a general implementation of this kind of
"combinatorial" model that could be configured for different numbers of
serotypes, etc. But, that code hasn't made it into the second version, yet.
We could look at migrating it if there is sufficient demand.

Dan

"Nicoletta De Maio" <ndemaio@student.ethz.ch> wrote in message
news:4e0c42a745bc749ec998029b1bf875b4$1@www.eclipse.org...
> Hi Dan
>
> First of all, thanks again for the help. Here's what came out of it:
>
> 1) Followed the instructions but when I run the scenario (which it lets me
> do now) I get this bit of info (*not* a warning or an error) in the error
> log:
> -------------
> Message: Aggregating SEIR Human didn't find a root node for the graph.
> Missing containment relationship?
> -------------
> Important or not? (I don't think I've come across the term 'containment
> relationship' in STEM before.) It doesn't stop the simulation as such but
> I'm not sure how to tell if it is doing anything at all: Singapore is so
> small that it is, for all practical purposes, impossible to find on the
> map view if the rest of the world is not displayed and I have no point of
> reference that tells me which part of the world I'm currently looking at.
> Is there any way of centering the active i.e. selected area on the screen?
> Or at least to display some sort of coordinate system? Or anything else
> that would make finding a small island off the southern tip of Malaysia a
> little easier? :)
>
> 2) Bug report has been submitted to Bugzilla. Wait and see...
>
> 3) See my e-mail. I've tracked down the paper you based the dengue
> simulation on but I'm not sure I understand how you did it. I'm even more
> interested in additional information now, as the paper does not list any
> time-dependent differential equations of the sort that went into the SEIR
> model.
>
> :-) Nicoletta
>
Re: Using Stem to simulate dengue fever [message #32454 is a reply to message #32174] Sun, 22 July 2007 21:41 Go to previous messageGo to next message
Nicoletta De Maio is currently offline Nicoletta De Maio
Messages: 8
Registered: July 2009
Junior Member
Hi

Thanks for the explanation. There are a number of things in the model that
I'd do differently, so I'm not sure if I shouldn't try and implement
something myself (if I ever get the thing to work the way I want it to...).

First of all there's no biological evidence that the order in which a
patient gets the dengue subtypes is in any way significant. If anything,
the *number* of times a patient has had the disease may have an influence
on the severity of the next infection, but not which subtypes he's already
had, or when. So that will greatly reduce the number of possible states
(and probably remove a lot of potential headaches).

Second, the severity of the disease in a patient doesn't seem to be
related to the infectiousness of a particular strain. In fact, the two
things seem to be quite independent as far as I know. So it may make sense
to model the two things separately as well.

So, hypothetically and as a first test, how would I go about defining a
system with, say, two serotypes? Do I simply define two new diseases
Dengue-1 and Dengue-2, give them different parameters and attach them to
the model? (Oh. I guess I need two infectors as well, then. Right.) Since
hemorragic fever is comparatively rare, I'd ignore that for the moment,
the goal is just to get the hang of the system.

I can worry about including mosquitoes later...

:-) Nicoletta
Re: Using Stem to simulate dengue fever [message #32627 is a reply to message #32454] Mon, 23 July 2007 15:51 Go to previous message
Daniel Ford is currently offline Daniel Ford
Messages: 138
Registered: July 2009
Location: New York
Senior Member
Nicoletta,
perhaps I misspoke, the model I referenced didn't favor one sequence of
the four serotypes over another, but since exposure to one serotype gives
one immunity to that serotype
(http://www.stanford.edu/group/virus/flavi/2000/dengue.htm), it is necessary
to track which serotypes have affected the population and to partition the
population by their exposure sequence, not simply to track the raw number of
infections. Tracking this partitioning is also important to properly
incorporate antibody dependent enhancement (ADE)
http://en.wikipedia.org/wiki/Antibody_dependent_enhancement, and it's
potential impact in increased infectiousness
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=15 44250.

You could define two different diseases and label them "Dengue-1" and
"Dengue-2" and add them to a model. They will co-exist with no problem and,
"yes", you will need two infectors. The problem with this approach is that
the disease models are not written to incorporate state information from
other diseases so they will ignore each, meaning people who die from one
disease, could still be alive as far as the other is concerned.

Dan





Daniel Ford
IBM Almaden Research Center
San Jose, CA
"Nicoletta De Maio" <ndemaio@student.ethz.ch> wrote in message
news:bbce5f1a51f7a2abd3b8e2b0048eb3ad$1@www.eclipse.org...
> Hi
>
> Thanks for the explanation. There are a number of things in the model that
> I'd do differently, so I'm not sure if I shouldn't try and implement
> something myself (if I ever get the thing to work the way I want it
> to...).
>
> First of all there's no biological evidence that the order in which a
> patient gets the dengue subtypes is in any way significant. If anything,
> the *number* of times a patient has had the disease may have an influence
> on the severity of the next infection, but not which subtypes he's already
> had, or when. So that will greatly reduce the number of possible states
> (and probably remove a lot of potential headaches).
>
> Second, the severity of the disease in a patient doesn't seem to be
> related to the infectiousness of a particular strain. In fact, the two
> things seem to be quite independent as far as I know. So it may make sense
> to model the two things separately as well.
>
> So, hypothetically and as a first test, how would I go about defining a
> system with, say, two serotypes? Do I simply define two new diseases
> Dengue-1 and Dengue-2, give them different parameters and attach them to
> the model? (Oh. I guess I need two infectors as well, then. Right.) Since
> hemorragic fever is comparatively rare, I'd ignore that for the moment,
> the goal is just to get the hang of the system.
>
> I can worry about including mosquitoes later...
>
> :-) Nicoletta
>
Re: Using Stem to simulate dengue fever: two questions/problems [message #580037 is a reply to message #31820] Mon, 16 July 2007 05:17 Go to previous message
Nicoletta De Maio is currently offline Nicoletta De Maio
Messages: 8
Registered: July 2009
Junior Member
Hi Dan

First of all, thanks again for the help. Here's what came out of it:

1) Followed the instructions but when I run the scenario (which it lets me
do now) I get this bit of info (*not* a warning or an error) in the error
log:
-------------
Message: Aggregating SEIR Human didn't find a root node for the graph.
Missing containment relationship?
-------------
Important or not? (I don't think I've come across the term 'containment
relationship' in STEM before.) It doesn't stop the simulation as such but
I'm not sure how to tell if it is doing anything at all: Singapore is so
small that it is, for all practical purposes, impossible to find on the
map view if the rest of the world is not displayed and I have no point of
reference that tells me which part of the world I'm currently looking at.
Is there any way of centering the active i.e. selected area on the screen?
Or at least to display some sort of coordinate system? Or anything else
that would make finding a small island off the southern tip of Malaysia a
little easier? :)

2) Bug report has been submitted to Bugzilla. Wait and see...

3) See my e-mail. I've tracked down the paper you based the dengue
simulation on but I'm not sure I understand how you did it. I'm even more
interested in additional information now, as the paper does not list any
time-dependent differential equations of the sort that went into the SEIR
model.

:-) Nicoletta
Re: Using Stem to simulate dengue fever: two questions/problems [message #580049 is a reply to message #31858] Mon, 16 July 2007 11:17 Go to previous message
John Thomas is currently offline John Thomas
Messages: 14
Registered: July 2009
Junior Member
Nicoletta

Dan will be back from vacation on Wednesday and I am sure will be able
to help you make more progress..

In the meantime, I will try to answer one of of your questions.

The "missing containment relationship" is not important at this time.
Currently I believe that the US is the only area where we have
containment relationships fully specified. It refers to fully
specifying for each Administration area 2 which admin area 1 it belongs
to. and for each administration area 1 which administration area 0 it
belongs to. Currently the containment relationship is not being used
in the model so you can ignore the message.

One other comment. You might want to download the STEM source from the
Eclipse CVS site if you haven't already. That way if Dan does discover
some problems and fixes them, you will be able to use the fixed version.
The STEM wiki
http://wiki.eclipse.org/Installation_Guide
has the information that you will need to setup Eclipse and access the
STEM source code.

I hope this is helpful.

John


Nicoletta De Maio wrote:
> Hi Dan
>
> First of all, thanks again for the help. Here's what came out of it:
>
> 1) Followed the instructions but when I run the scenario (which it lets
> me do now) I get this bit of info (*not* a warning or an error) in the
> error log:
> -------------
> Message: Aggregating SEIR Human didn't find a root node for the graph.
> Missing containment relationship?
> -------------
> Important or not? (I don't think I've come across the term 'containment
> relationship' in STEM before.) It doesn't stop the simulation as such
> but I'm not sure how to tell if it is doing anything at all: Singapore
> is so small that it is, for all practical purposes, impossible to find
> on the map view if the rest of the world is not displayed and I have no
> point of reference that tells me which part of the world I'm currently
> looking at. Is there any way of centering the active i.e. selected area
> on the screen? Or at least to display some sort of coordinate system? Or
> anything else that would make finding a small island off the southern
> tip of Malaysia a little easier? :)
>
> 2) Bug report has been submitted to Bugzilla. Wait and see...
>
> 3) See my e-mail. I've tracked down the paper you based the dengue
> simulation on but I'm not sure I understand how you did it. I'm even
> more interested in additional information now, as the paper does not
> list any time-dependent differential equations of the sort that went
> into the SEIR model.
>
> :-) Nicoletta
>
Re: Using Stem to simulate dengue fever: two questions/problems [message #580059 is a reply to message #31893] Tue, 17 July 2007 05:29 Go to previous message
Nicoletta De Maio is currently offline Nicoletta De Maio
Messages: 8
Registered: July 2009
Junior Member
Hi

This is helpful in the sense that I don't have to worry about the message,
so thanks.

As for checking STEM out of the CVS repository, don't get me started - but
the problem is on my side (actually, our admin's...). I've got the source
code, though. I just hope I'll have access via CVS soon...

:-) Nicoletta
Re: Using Stem to simulate dengue fever: two questions/problems [message #580137 is a reply to message #31893] Wed, 18 July 2007 10:46 Go to previous message
Daniel Ford is currently offline Daniel Ford
Messages: 138
Registered: July 2009
Location: New York
Senior Member
Nicoletta,
the message you received was as John described. When the Aggregating
disease model walks the graph, it follows edges that represent geographic
"containment" relationships. The basic idea is that when a larger region
contains a number of smaller regions the state of the disease in the larger
region is the aggregate of the states of the smaller regions it contains.
The problem is that when it walked the graph it did not find a root node
representing the top-level of the heirarchy. I'll have to look at the data
set for Singapore, but I think John is right that we don't yet have the
contianment relationship specified completely.

Dan


"John Thomas" <jthomas119@gmail.com> wrote in message
news:469B8C11.6090904@gmail.com...
> Nicoletta
>
> Dan will be back from vacation on Wednesday and I am sure will be able to
> help you make more progress..
>
> In the meantime, I will try to answer one of of your questions.
>
> The "missing containment relationship" is not important at this time.
> Currently I believe that the US is the only area where we have containment
> relationships fully specified. It refers to fully specifying for each
> Administration area 2 which admin area 1 it belongs to. and for each
> administration area 1 which administration area 0 it belongs to.
> Currently the containment relationship is not being used in the model so
> you can ignore the message.
>
> One other comment. You might want to download the STEM source from the
> Eclipse CVS site if you haven't already. That way if Dan does discover
> some problems and fixes them, you will be able to use the fixed version.
> The STEM wiki
> http://wiki.eclipse.org/Installation_Guide
> has the information that you will need to setup Eclipse and access the
> STEM source code.
>
> I hope this is helpful.
>
> John
>
>
> Nicoletta De Maio wrote:
>> Hi Dan
>>
>> First of all, thanks again for the help. Here's what came out of it:
>>
>> 1) Followed the instructions but when I run the scenario (which it lets
>> me do now) I get this bit of info (*not* a warning or an error) in the
>> error log:
>> -------------
>> Message: Aggregating SEIR Human didn't find a root node for the graph.
>> Missing containment relationship?
>> -------------
>> Important or not? (I don't think I've come across the term 'containment
>> relationship' in STEM before.) It doesn't stop the simulation as such but
>> I'm not sure how to tell if it is doing anything at all: Singapore is so
>> small that it is, for all practical purposes, impossible to find on the
>> map view if the rest of the world is not displayed and I have no point of
>> reference that tells me which part of the world I'm currently looking at.
>> Is there any way of centering the active i.e. selected area on the
>> screen? Or at least to display some sort of coordinate system? Or
>> anything else that would make finding a small island off the southern tip
>> of Malaysia a little easier? :)
>>
>> 2) Bug report has been submitted to Bugzilla. Wait and see...
>>
>> 3) See my e-mail. I've tracked down the paper you based the dengue
>> simulation on but I'm not sure I understand how you did it. I'm even more
>> interested in additional information now, as the paper does not list any
>> time-dependent differential equations of the sort that went into the SEIR
>> model.
>>
>> :-) Nicoletta
>>
Re: Using Stem to simulate dengue fever [message #580227 is a reply to message #31858] Wed, 18 July 2007 11:40 Go to previous message
Daniel Ford is currently offline Daniel Ford
Messages: 138
Registered: July 2009
Location: New York
Senior Member
Nicoletta,
simulating Dengue Hemorrhagic Fever (DHF) is complicated because it is a
multiserotype disease that affects people differently depending upon their
personal serotype infection history. Basically, an initial infection with
any of the four different DHF serotypes will cause a similar "mild" reaction
in a person and give them immunity to that serotype. However, a second
infection by a different serotype will cause a much more severe, potentially
fatal, reaction and actually make the infected person more infectious to
others.

The problem with modeling such a disease is that you need to keep track of
the sequential serotype infection history of the population and to have
different infection and mortality rates depending upon the infection
history. This results in a combinatorial explosion in the number of states
that need to be represented and a much more complicated configuration and
implementation for the disease model. For example, in a simple SEIR model,
a population member can only be in one of four states (S, E, I, or R). In a
model for DHF, all population members start out in a single "S"
(Susceptible) state, but then they can transition to one of four "E" states
and then to the corresponding "I" state, of which there are four. From
there they will "recover" to the corresponding "R" state, of which there are
four. From there they can be exposed to any of the remaining three
serotypes and then transition to the appropriate "E" state that matches
their infection history, of which there are 12 (4x3) in total. This
continues to expand for all possible sequences of serotype infection until
the states eventually converge on a single "Recovered from everything"
state.

The first version of STEM contained a general implementation of this kind of
"combinatorial" model that could be configured for different numbers of
serotypes, etc. But, that code hasn't made it into the second version, yet.
We could look at migrating it if there is sufficient demand.

Dan

"Nicoletta De Maio" <ndemaio@student.ethz.ch> wrote in message
news:4e0c42a745bc749ec998029b1bf875b4$1@www.eclipse.org...
> Hi Dan
>
> First of all, thanks again for the help. Here's what came out of it:
>
> 1) Followed the instructions but when I run the scenario (which it lets me
> do now) I get this bit of info (*not* a warning or an error) in the error
> log:
> -------------
> Message: Aggregating SEIR Human didn't find a root node for the graph.
> Missing containment relationship?
> -------------
> Important or not? (I don't think I've come across the term 'containment
> relationship' in STEM before.) It doesn't stop the simulation as such but
> I'm not sure how to tell if it is doing anything at all: Singapore is so
> small that it is, for all practical purposes, impossible to find on the
> map view if the rest of the world is not displayed and I have no point of
> reference that tells me which part of the world I'm currently looking at.
> Is there any way of centering the active i.e. selected area on the screen?
> Or at least to display some sort of coordinate system? Or anything else
> that would make finding a small island off the southern tip of Malaysia a
> little easier? :)
>
> 2) Bug report has been submitted to Bugzilla. Wait and see...
>
> 3) See my e-mail. I've tracked down the paper you based the dengue
> simulation on but I'm not sure I understand how you did it. I'm even more
> interested in additional information now, as the paper does not list any
> time-dependent differential equations of the sort that went into the SEIR
> model.
>
> :-) Nicoletta
>
Re: Using Stem to simulate dengue fever [message #580405 is a reply to message #32174] Sun, 22 July 2007 21:41 Go to previous message
Nicoletta De Maio is currently offline Nicoletta De Maio
Messages: 8
Registered: July 2009
Junior Member
Hi

Thanks for the explanation. There are a number of things in the model that
I'd do differently, so I'm not sure if I shouldn't try and implement
something myself (if I ever get the thing to work the way I want it to...).

First of all there's no biological evidence that the order in which a
patient gets the dengue subtypes is in any way significant. If anything,
the *number* of times a patient has had the disease may have an influence
on the severity of the next infection, but not which subtypes he's already
had, or when. So that will greatly reduce the number of possible states
(and probably remove a lot of potential headaches).

Second, the severity of the disease in a patient doesn't seem to be
related to the infectiousness of a particular strain. In fact, the two
things seem to be quite independent as far as I know. So it may make sense
to model the two things separately as well.

So, hypothetically and as a first test, how would I go about defining a
system with, say, two serotypes? Do I simply define two new diseases
Dengue-1 and Dengue-2, give them different parameters and attach them to
the model? (Oh. I guess I need two infectors as well, then. Right.) Since
hemorragic fever is comparatively rare, I'd ignore that for the moment,
the goal is just to get the hang of the system.

I can worry about including mosquitoes later...

:-) Nicoletta
Re: Using Stem to simulate dengue fever [message #580593 is a reply to message #32454] Mon, 23 July 2007 15:51 Go to previous message
Daniel Ford is currently offline Daniel Ford
Messages: 138
Registered: July 2009
Location: New York
Senior Member
Nicoletta,
perhaps I misspoke, the model I referenced didn't favor one sequence of
the four serotypes over another, but since exposure to one serotype gives
one immunity to that serotype
(http://www.stanford.edu/group/virus/flavi/2000/dengue.htm), it is necessary
to track which serotypes have affected the population and to partition the
population by their exposure sequence, not simply to track the raw number of
infections. Tracking this partitioning is also important to properly
incorporate antibody dependent enhancement (ADE)
http://en.wikipedia.org/wiki/Antibody_dependent_enhancement, and it's
potential impact in increased infectiousness
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=15 44250

You could define two different diseases and label them "Dengue-1" and
"Dengue-2" and add them to a model. They will co-exist with no problem and,
"yes", you will need two infectors. The problem with this approach is that
the disease models are not written to incorporate state information from
other diseases so they will ignore each, meaning people who die from one
disease, could still be alive as far as the other is concerned.

Dan





Daniel Ford
IBM Almaden Research Center
San Jose, CA
"Nicoletta De Maio" <ndemaio@student.ethz.ch> wrote in message
news:bbce5f1a51f7a2abd3b8e2b0048eb3ad$1@www.eclipse.org...
> Hi
>
> Thanks for the explanation. There are a number of things in the model that
> I'd do differently, so I'm not sure if I shouldn't try and implement
> something myself (if I ever get the thing to work the way I want it
> to...).
>
> First of all there's no biological evidence that the order in which a
> patient gets the dengue subtypes is in any way significant. If anything,
> the *number* of times a patient has had the disease may have an influence
> on the severity of the next infection, but not which subtypes he's already
> had, or when. So that will greatly reduce the number of possible states
> (and probably remove a lot of potential headaches).
>
> Second, the severity of the disease in a patient doesn't seem to be
> related to the infectiousness of a particular strain. In fact, the two
> things seem to be quite independent as far as I know. So it may make sense
> to model the two things separately as well.
>
> So, hypothetically and as a first test, how would I go about defining a
> system with, say, two serotypes? Do I simply define two new diseases
> Dengue-1 and Dengue-2, give them different parameters and attach them to
> the model? (Oh. I guess I need two infectors as well, then. Right.) Since
> hemorragic fever is comparatively rare, I'd ignore that for the moment,
> the goal is just to get the hang of the system.
>
> I can worry about including mosquitoes later...
>
> :-) Nicoletta
>
Previous Topic:STEM: problem building a project
Next Topic:STEM: problem building a project
Goto Forum:
  


Current Time: Tue Jul 29 00:53:16 EDT 2014

Powered by FUDForum. Page generated in 0.02558 seconds