[Population Modeling] Population modeling definition

[H. Stephen Leff]

Re: Definition
This may be too late for the paper, but here is my 2 cents.

I think it is important to distinguish between system of care population models and epidemiological or etiological population models (as I understand them from discussions on this list).
Epidemiological or etiological population models (and I may not have this right, since I do not do this type of modeling) seem to focus on the causes and distributions of illnesses or other things in populations.  It seems to me that these models usually assume a system of care – something like treatment as usual, although this is not always made explicit, although I think models are more useful if the nature of the care available is made explicit.

Systems of care population models focus on distributions of persons in systems of care, usually categorized according to health or service states (more on that in a moment) and how persons move or transition through theses states as a function of the type of service they receive and their characteristics or the course of their disorder.
Re: Jacob’s comment below:

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.

I’m not sure I understand this, but I would say Markov models commonly consider differences between individual entities, if what is meant by “differences” is differences in health states.  Moreover, different Markov models can be constructed for different populations distinguished by things like diagnoses, then their results can be combined to model more complex populations.

One more point: Although systems of care often use as health states locations in particular services (e.g., inpatient care), I think the most useful systems of care models use health states based on epidemiological categories (to muddy the waters a bit) such as “acutely disturbed,” “mildly disturbed,” and so on.  When we use service locations we confound health states with the services available in systems.  So in mental health systems higher functioning persons can be in hospitals because community residential care is not available, not because their heath states warrant inpatient care.

Anyway, the discussions on the list have been very thought provoking.  Thanks, everyone.
Virtually,
Steve Leff

From: popmodwkgrpimag-news-bounces at simtk.org [mailto:popmodwkgrpimag-news-bounces at simtk.org] On Behalf Of Jacob Barhak
Sent: Wednesday, January 07, 2015 4:18 PM
To: popmodwkgrpimag-news at simtk.org
Subject: [Population Modeling] Population modeling definition


Hello All,

There is a growing demand to define the term population modeling. This need is visible in some recent personal communications.

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.

So here are key quotes:

Sergey Nuzhdin USC:

"we are doing lots of population / quantitative modeling, and i would love to get engaged in human community;

but so far, our efforts have been in Drosophila and plants"

John Rice - Society for Simulation in Heathcare (SSiH)  government relations vice chair:

"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."

Sergey replying:
" you are right, John, precisely what we are doing (among other things);

how to predict collective properties from individual ones;

there is plenty done about it when the subjects are homogeneous, and very little when heterogeneous;"

Madhav Marathe - Virginia Tech:
"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."

It seems that the above discussion matches the ad hoc definition we reached at the IMAG meeting at the NIH:

"Modeling a collection of entities with different levels of heterogeneity"

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.

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.

I would appreciate your thoughts and will try to incorporate those in the paper.

          Jacob
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://simtk.org/pipermail/popmodwkgrpimag-news/attachments/20150108/3d5e0862/attachment-0001.html