Your SlideShare is downloading. ×
0
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
3A_1_spatially clustered associations
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

3A_1_spatially clustered associations

233

Published on

Session 3A, Presentation 1

Session 3A, Presentation 1

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
233
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide
  • Data mashups
  • Efficient geoprocessing service vs pertinent statistical method /pertinent model
  • Cluster detection /cluster association
  • Cluster detection /cluster association
  • Transcript

    • 1. Spatially Clustered Associations in Health GIS “mash-ups” Didier Leibovici 1 , Lucy Bastin 2 , Suchith Anand 1 , Jerry Swan 1 , Gobe Hobona 1 and Mike Jackson 1 1 Centre for Geospatial Sciences, University of Nottingham, UK 2 School of Engineering & Applied Science, Aston University, UK Tel. (+44(0)115 846 8408) Fax (+44(0)115 951 5249) [email_address] Spatial Statistics, Multiway Data, Marked Point Process, Spatial Pattern, Spatial Interaction, Multi-Scale Analysis Co-Occurrences,
    • 2. an historical perspective
      • data acquisition / storage / exchange
      • conflation / data analysis / geocomputation
      • for general public / for practitioners / for researchers
    • 3. advanced conflation/analysis
    • 4. Cluster detection vs Clustered associations
      • Detection of clusters
        • Epidemiology Ecology Health Geography ...
      • GAM Openshaw et al. (1987)
      • Besag-Newell (areals) Besag & Newell (1991) DCluster
      • SaTScan Kulldorf & Nagarwalla (1995) www.SaTScan.org
      • spatclus
      • cases / observed, expected,
      • non cases / background population, population at risk
    • 5. Cluster detection vs Clustered associations
      • Clustered associations
        • Epidemiology Ecology Health Geography ...
      • “ multiple” cases / observed, (expected)
      • risk factors / attributes, areal attributes
      • functions/package & web service
      • ...
      detecting clusters of multivariate associations between attributes of one or more populations localised spatially
    • 6. a simple example star dot square 3 spatial patterns from 3 point processes focus is on: “ associations” , local profiles ??? ??? ??? e.g. contagion factors
    • 7. spatial associations
      • talking about labels
        • presence of all labels, uniformity,
        • dominance of one label, particular profiles
      • talking about variables
      • independence, correlation
      • co-occurrences counts
      • uniformity / non-uniformity of spatial profiles | entropy
      • independence, lack of independence of variables | chi2
      in a vicinity
    • 8.
      • chi2
      • entropy
      spatial associations / co-occurrences counts global & local The Co-o ccurrences at a chosen order , of attributes from the same or different processes , build multinomial distributions at the root of spatial organisation and interactions of processes according to: the collocation d istance, and the order of collocation . see Leibovici et al. (2008) (2009) CAkOO and SOOk methods for global / overall “ ”
    • 9.
      • chi2 / entropy local statistics
              • at each neighbourhood
              • a) “condition of sufficiency” for
              • N bp points (any points)
              • N bc cases (list of cases labels)
              • b) co-occurrences counts
              • - at “ d ” chosen, “ all ” ( d =d N ), “ opti mal” (best d )
              • c) statistical map, chi2 and entropy to
              • d) choice of growing
      spatial associations / co-occurrences counts in a vicinity x S V x S V x S V x S x S ScankOO package kOO (to be finalised with CAkOO, SOOk, selSOOk) d x “ ” d N x x S x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x V x S x x
    • 10. > library(spatstat) > source(“kOO.R”) ={“star”,“dot”,”square”} dot square star 47 46 50 1/HSu HSu Min Q1 Median Q3 Max 0.51 0.85 0.92 0.97 0.99 1/HsSu HsSu Min Q1 Median Q3 Max 0.13 0.59 0.86 0.94 1
    • 11. chi2 minimum ... “ “ independence” ”
    • 12. - + 1450 428 -1 - 3 - 5 + 1 +3 + 5 571 478 401 55 138 235 Epidemiological study Infectious disease dataset somewhere in UK 626 616 636 1000 878
    • 13. more -1 and less +5 9.2% profile with same +3 and +5
    • 14. relatively more +3 and less +5 reds higher odd - for 1 higher odd + for 3 than blacks
    • 15.  
    • 16. SOOk analysis (global)
    • 17. /21
    • 18. 150 subjects 0 Sept04-Feb05 3 5 - 59 7 + 5 4 / Mar05-Aug05 3 5 - 50 8 + 12 5 <45 45-75 >75 3 5
    • 19. conclusions and further work
      • scankOO (local view)
      • + statistics from local co-occurrences
      • + neighourhood control (points, cases)
      • + map of local spatial entropies / spatial chi2
      • + more tests needed on d ( opti) and Nbp Nbc (grows)
      • + monte- carlo, random fields p-values for local maximum (or minimum)
      • other exploratory analysis based on co-occurrences (global view)
      • + SOOk, SelSOOk and CAkOO
      • + scale analysis
      • + needs correction edge effects
      • integration in an OGC Web Processing Service (using R)
      • + R wp server, Rapache, parallel processing
      package kOO (to be finalised with CAkOO, SOOk, selSOOk)

    ×