<div dir="ltr"><div>Hi Talitha, Hi John, Hi Olaf, Hi Steve, Hi Stephan, Hi Al, and greeting to all others who have not participated yet.</div><div><br></div><div>You have contributed thoughts to this discussion. I would like to point to two issues I see and add a third issue to the discussion</div><div><br></div><div>1. Dictionary like definition:</div><div>John and Olaf are trying to define the term like a dictionary. In this perspective, I myself would interpret population as: "A collection of X" where X can change as Madhav suggested before. It is probably a noun describing plurality of X.</div><div><br></div><div>2. Definition by Models included:</div><div>However, perhaps a more interesting discussion is what kind of models fall into our category. I recall Al and Steve trying to include Markov models, while Talitha and previously Stephan are excluding Markov Cohort models from falling within our definition of population modeling. This is to the best of my understating and I am surprised that Talitha suggested a strict definition since she some of her work is Markov model based.</div><div><br></div><div>What would you all feel about including Markov models within population modeling only if they stratify the cohort by age, gender, and/or other parameters. And non stratified models will define these as cohort models outside population modeling?</div><div><br></div><div><div>3. Modeling of individual uncertainty</div></div><div>This is another question. Would we consider description of uncertainty of parameters in an individual within our population modeling group? For example, does defining distribution of height of a person by mean, variance, and distribution function fall into our category? After all we are modeling a single entity from information derived from a population. I would say yes, especially if the information is extracted from a population that we can name. But what about correlation between anthropometric features such as defining the mean ratio between height and arm length as one number? Will one number be sufficient to call it population modeling or do we need a distribution? I would say if the number is associated to a population cohort then perhaps, and defiantly yes if we have several numbers associated with different population cohorts.</div><div><br></div><div>I am interested to learn more what others think of those questions. I hope others will join this discussion.</div><div><br></div><div> Jacob</div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Jan 13, 2015 at 3:25 PM, Talitha Feenstra <span dir="ltr"><<a href="mailto:talitha.feenstra@rivm.nl" target="_blank">talitha.feenstra@rivm.nl</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><font face="Default Sans Serif,Verdana,Arial,Helvetica,sans-serif"> <span>Dear all, <br><br>Following up on the definition discussion. <br><br>I would suggest that the structure of the model (state transition, or DES, or agent-based, or whatever) is not so important for the question whether or not a model is a population model. <br>More important seems that it reflects a "population". That is, rather than simulating a cohort of say 1000 people, it contains information on a population's epidemiology and demography. This may or may not include birth,and migration, and almost always includes mortality. This would then indeed exclude Markov cohort models, but more due to the cohort aspect than for their lack of modeling heterogeneity.<br><br>However the examples presented go very broad and also include populations of animals and even nonliving objects. Especially the latter gets confusing to me, how to distinguish between what I knew of as "agent-based modeling" which is a model structure, and "population modeling" ? <br><br>kind regards, Talitha Feenstra<br><br><br>TL Feenstra, PhD<br>health economist<br>RIVM, Centre for Nutrition, Prevention and Health Services Research<br>Po Box 1<br>3720 BA Bilthoven<br>*31 30 274 4387 (mon)<br>* 31 6 1186 87 60<br>UMCG, dept of epidemiology<br>PO Box 30.001<br>9700 RB Groningen<br>*31 50 361 5110 (tue/thu/fri)</span><br><br><font color="#990099">-----<a href="mailto:popmodwkgrpimag-news-bounces@simtk.org" target="_blank">popmodwkgrpimag-news-bounces@simtk.org</a> wrote: -----</font><div style="padding-left:5px"><div style="padding-right:0px;padding-left:5px;border-left-color:black;border-left-width:2px;border-left-style:solid">To: <a href="mailto:popmodwkgrpimag-news@simtk.org" target="_blank">popmodwkgrpimag-news@simtk.org</a><br>From: Stefan Scholz <u></u><br>Sent by: <a href="mailto:popmodwkgrpimag-news-bounces@simtk.org" target="_blank">popmodwkgrpimag-news-bounces@simtk.org</a><br>Date: 01/07/2015 10:58PM<br>Subject: Re: [Population Modeling] Population modeling definition<div><div class="h5"><br><br>
Hello all,<br>
<br>
I think the definition hits the broad subject really well. Maybe I
am seeing too much through the eyes of a health economist, but I
would draw the line of population models right after (cohort) markov
models. In my opinion mostly all of those cohort models ignore
patient heterogeneity and focus more on the disease rather than the
population. On the other hand I would say that "Individual Sampling
Models" and DES as presented by Al Chrosny and Jon Karnon do
incorporate patient heterogeneity and should be presented by the
term "population model". So maybe we can draw the line between
cohort and individual Markov models.<br>
<br>
Best regards,<br>
Stefan<br>
<br>
<br>
<br>
<div>On 07.01.2015 22:17, Jacob Barhak
wrote:<br>
</div>
<blockquote type="cite">
<p dir="ltr">Hello All,</p>
<p dir="ltr">There is a growing demand to define the term
population modeling. This need is visible in some recent
personal communications.</p>
<p dir="ltr">This post will try to bring some quotes from
different people who try to define it during conversation. I
intend to use some of this in the paper, so I am bringing it to
the list first to get opinions of more people.</p>
<p dir="ltr">So here are key quotes:</p>
<p dir="ltr">Sergey Nuzhdin USC:</p>
<p dir="ltr">"we are doing lots of population / quantitative
modeling, and i would love to get engaged in human community;</p>
<p dir="ltr">but so far, our efforts have been in Drosophila and
plants"</p>
<p dir="ltr">John Rice - Society for Simulation in Heathcare
(SSiH) government relations vice chair:</p>
<p dir="ltr">"Sergey, perfect! I want to learn about PopMod for
non human sciences and engineer applications. If we had a
population of trees in a forrest, in stead of a generalized
collective description. Could we predict the course of a forrest
fire better? Only beginning to think about that, then here you
are, modeling a populations of plants."</p>
<p dir="ltr">Sergey replying: <br>
" you are right, John, precisely what we are doing (among other
things);</p>
<p dir="ltr">how to predict collective properties from individual
ones;</p>
<p dir="ltr">there is plenty done about it when the subjects are
homogeneous, and very little when heterogeneous;"</p>
<p dir="ltr">Madhav Marathe - Virginia Tech: <br>
"The population modeling group is intended to be broad. Ofcourse
making it way too broad might make it harder for a cohesive
conversation but we can see how it plays out. I am calling this
population of things; it appears things is the best word I can
use to keep the intended generality. We have studied cells,
wireless devices, people and animals in the past for instance."</p>
<p dir="ltr">It seems that the above discussion matches the ad hoc
definition we reached at the IMAG meeting at the NIH:</p>
<p dir="ltr">"Modeling a collection of entities with different
levels of heterogeneity"</p>
<p dir="ltr">We reached this definition quickly and it was a broad
consensus. However, others on the list may want to discuss this
definition and offer alternatives.</p>
<p dir="ltr">For example Markov models address populations mostly
as time series and seldom consider differences between
individual entities, yet I would still include Markov models
under the umbrella of population modeling. It is a difficult fit
to the above definition unless hetrogeneity among states is
considered within time.</p>
<p dir="ltr">I would appreciate your thoughts and will try to
incorporate those in the paper.</p>
<p dir="ltr"> Jacob</p>
<br>
<fieldset></fieldset>
<br>
<div><font face="Courier New,Courier,monospace">_______________________________________________<br>PopModWkGrpIMAG-news mailing list<br><a href="mailto:PopModWkGrpIMAG-news@simtk.org" target="_blank">PopModWkGrpIMAG-news@simtk.org</a><br><a href="https://simtk.org/mailman/listinfo/popmodwkgrpimag-news" target="_blank">https://simtk.org/mailman/listinfo/popmodwkgrpimag-news</a><br></font></div>
</blockquote>
<br>
<div><font face="Courier New,Courier,monospace">_______________________________________________<br>PopModWkGrpIMAG-news mailing list<br><a href="mailto:PopModWkGrpIMAG-news@simtk.org" target="_blank">PopModWkGrpIMAG-news@simtk.org</a><br><a href="https://simtk.org/mailman/listinfo/popmodwkgrpimag-news" target="_blank">https://simtk.org/mailman/listinfo/popmodwkgrpimag-news</a><br></font></div><u></u></div></div></div></div><div><br><font face="Verdana" style="font-size:8pt"><a href="http://www.rivm.nl/Proclaimer" target="_blank"><br>Proclaimer RIVM http://www.rivm.nl/Proclaimer</a></font></div></font><br>_______________________________________________<br>
PopModWkGrpIMAG-news mailing list<br>
<a href="mailto:PopModWkGrpIMAG-news@simtk.org">PopModWkGrpIMAG-news@simtk.org</a><br>
<a href="https://simtk.org/mailman/listinfo/popmodwkgrpimag-news" target="_blank">https://simtk.org/mailman/listinfo/popmodwkgrpimag-news</a><br>
<br></blockquote></div><br></div>