Help save little old ladies with statistics!

[bogamol]

The current model of dynamic deployment in EMS (ambulance transport) systems is called System Status Management and it works by doing hourly demand analysis to determine how many units are required to meet the demand ay various times of the day. Most services expect an ambulance to be on a call about 33% of the time they are in service in order to assure fastest possible reaponse. We then have set places that we sit in order to minimize distance to travel based on how many ambulances are available. eg if there is only one ambulance available it is posted in the most central point of the service area. If two then split the area into 2 zones and put them both in the center of their reapective zones.

It works kind of like zone defense in soccer or football.

I think we can do better by being able to predict, based on historical data where calls are most likely to happen. eg there are more accidents during rush hour so an ambulance should be poated closer to high call volume freeways during that time. More people wake up with chest pain early in the morning so ambulances should be posted near neoghborhoods where the residents are demographically older.

The question is I can use Morisitas index to determine whether calls are significantly clumped together but if the R value is low it would be difficult to determine the actual center of the clumps or whether it would be beneficial to alter the way we do things. Any ideas on how to analyze calls and call locations as they cluster themselves on a map?
In a highly skilled system it would work like man coverage.

[Old School]

With the baby boomers retiring at ever greater rates, all those women going through menopause and hot flashes; is it any wonder the planet is getting warmer?
_________________
I am not young enough to know everything.
- Oscar Wilde

[notageek]

[BoneKracker]

Reading this article would be a good place to start:

Operations Research Models for the Deployment of Emergency Service Vehicles

Also: see here.
_________________
Oldthinkers unbellyfeel INGSOC.
-- Headline of a document on Winston Smith's terminal in his cubicle at the Ministry of Truth, seen briefly in the background in one scene of the movie rendition of Nineteen Eighty-Four.

[richk449]

BKs review article is probably a good place to start. Personally though, my immediate thought was clustering algorithms, since it sounds like you are really trying to position response teams near the expected problem clusters.

There is plenty on the web regarding clustering algorithms for emergency response, although much of it is for large scale disasters, rather than ordinary ones. Regardless, here are a few articles I found, some with emphasis on clustering:

http://dvd1.ecs.umass.edu/wireless/resource_alloc/resource%20allocation%20kondaveti.pdf
http://users.cis.fiu.edu/~taoli/pub/disaster-survey.pdf
http://dns5405.dizinc.com/PDF/Connections/v26/2005_I-2-5.pdf
http://www1bpt.bridgeport.edu/~jelee/courses/CS556_SU08/paper/5.pdf
http://www.gdmc.nl/publications/2008/SDI_Emergency_Response_Netherlands.pdf

These are also interesting, although they are bit different, since they are highly focused on using real-time data to predict current crises:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.165.9143&rep=rep1&type=pdf
http://nd.edu/~icensa/papers/ICCS2007aOPTIMIZE.pdf

Disclaimer: I didn't read any of the articles I posted.

[bogamol]

I found some relevant search terms that I can use to research my methods and I also got some good information about what levels of granularity I should be looking at.

Thanks, BK and rich. Your links were very helpful.

[BoneKracker]

You are welcome. You might get some pro bono help on this from a university or from some local consultants who work in the operations research or simulation / modeling field (promise them public recognition / free publicity, and your higher ups saying good things about them to their higher ups).
_________________
Oldthinkers unbellyfeel INGSOC.
-- Headline of a document on Winston Smith's terminal in his cubicle at the Ministry of Truth, seen briefly in the background in one scene of the movie rendition of Nineteen Eighty-Four.