Geospatial Examination in General Wellbeing Spatial Bunch Recognition M.J. School, Jalgaon India September 22-26, 2008 G.


98 views
Uploaded on:
Description
Geospatial Examination in General Wellbeing Spatial Bunch Recognition M.J. School, Jalgaon India September 22-26, 2008 Glen D. Johnson New York State Branch of Wellbeing and The College at Albany School of General Wellbeing Bureau of Natural Wellbeing Sciences Affirmation:
Transcripts
Slide 1

Geospatial Analysis in Public Health Spatial Cluster Detection M.J. School, Jalgaon India September 22-26, 2008 Glen D. Johnson New York State Department of Health and The University at Albany School of Public Health Department of Environmental Health Sciences

Slide 2

Acknowledgment: Some of the accompanying design on bunch discovery are compliments of Tom Talbot, MSPH of the New York State Department of Health - co-shows “GIS in Public Health” with Glen Johnson and Frank Boscoe at the University at Albany, S.U.N.Y.

Slide 3

Cluster various comparable things gathered firmly together Webster’s Dictionary Concentrations of wellbeing occasions in space and/or time Public Health Definition

Slide 4

Clustering of wellbeing results may be brought on by various group level factors… Occupation blend Demographic blend (i.e. Race, Age, Sex) Socioeconomic status Cultural/Behavioral Environmental Exposure (dependably an unavoidable issue) Time and/or Space (catches unexplained variables that co-fluctuate with the result)

Slide 5

Cluster location affected by scaling and zoning impacts: … as must be considered for every single spatial measurement and mapping/representation - the Modifiable Area Unit Problem (MAUP)

Slide 9

Different size of observational units: Coarser total

Slide 10

Different zonation: Grid shift

Slide 13

Cluster Questions Does an illness bunch in space? Does a sickness group in both time and space? Where is the probably bunch?

Slide 14

More Cluster Questions At what geographic or populace scale do groups show up? Are instances of sickness bunched in territories of high introduction? - or all the more for the most part, “Can the group be clarified as being connected with an option that is other than chance?”

Slide 15

Nearest Neighbor Analysis Cuzick & Edwards Method Count the quantity of cases whose closest neighbors are cases and not controls. At the point when cases are grouped the closest neighbor to a case will have a tendency to be another case, and the test measurement will be expansive.

Slide 16

Nearest Neighbor Analyses

Slide 17

Advantages Accounts for the geographic variety in populace thickness Accounts for confounders through sensible choice of controls Can recognize grouping with numerous little bunches

Slide 18

Disadvantages Must have spatial areas of cases & controls Doesn’t show area of the groups

Slide 19

Knox Method test for space-time communication When space-time connection is available cases close in space will be close in time, the test measurement will be expansive. Test measurement: The quantity of sets of cases that are close in both time and space. P quality is computed through irregular reenactments of the time estimation of the cases Need to characterize discriminating space and time separations. i.e. characterize what is close?

Slide 20

Advantages Do not need to guide controls Determines if there is a space-time communication. Can identify space-time bunching notwithstanding when the general illness rate has continued as before after some time

Slide 21

Disadvantage Computationally tedious with countless. Does not focus regions or time times of where groups happen.

Slide 22

Spatial Scan Statistic Martin Kulldorff http://www.satscan.org/Determines areas with hoisted rates that are factually noteworthy. Alter for various testing of the numerous conceivable areas and zone sizes of bunches. Theory testing taking into account Monte Carlo recreations of the invalid, totally irregular, spatial dispersion

Slide 23

Following is a case of how the sweep measurement calculation portrays all conceivable roundabout bunches, in light of enumeration squares in the city of Albany …

Slide 28

A probability proportion is then processed for each roundabout window, where every window speaks to a potential spatial group. For instance, accepting a Poisson circulation of checks, the probability proportion is corresponding to … for watched cases c and expected cases E[c] inside the inquiry window, and C aggregate watched cases all through the area, including inside of the pursuit window.

Slide 29

The circle with the greatest probability proportion is then distinguished as the in all likelihood group, and all others are rank-requested beneath the most extreme. An invalid dissemination of greatest probability proportions is acquired by rehashing the examination on a randomized variant of the information, getting the maximum. probability proportion, and rehashing this activity for, say, 999 times. A p - quality is gotten for every circle by contrasting it’s probability proportion with the recreated invalid appropriation. In this way, for a probability proportion whose rank is R inside of the mimicked invalid qualities, then the p - esteem = R/(# reproductions +1).

Slide 30

Example, low conception weight

Slide 32

Note that E[c] = n*C/N for populace n in the circle and aggregate number of cases and Population = C and N separately or for covariate class i (a “indirect standardization”) or E[c] may even be anticipated from a relapse model.

Slide 33

Recent headways in the spatial output measurement went for conquering the limitation of the fairly subjective state of roundabout groups Patil GP, Taillie C. Upper level set output measurement for recognizing self-assertively formed hotspots. Environ Ecol Stat 2004;183-197. Duczmal L, Assuncao RM . A recreated strengthening procedure for the recognition of subjectively molded spatial groups. Comp Stat Data Anal 2004; 45:269-286. Tango T, Takahashi K . An adaptably formed spatial output measurement for distinguishing bunches. Int J Health Geographics 2005; 4:11.

Slide 34

Regression Analysis Control for known danger components before dissecting for spatial bunching Analyze for unexplained groups. Follow-up in regions with vast relapse residuals with conventional case-control or partner studies Obtain extra hazard component information to represent the substantial residuals.

Slide 35

Example, Child Lead Poisoning

Slide 36

.:

Recommended
View more...