Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health...
Transcript of Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health...
![Page 1: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/1.jpg)
Urban GIS for Health Metrics
Presented at International Conference on Urban Health, March 5th, 2014
Dajun Dai
Department of Geosciences, Georgia State UniversityAtlanta, Georgia, United States
![Page 2: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/2.jpg)
People, Place, and Health• Location, location, location!!!
– Almost everything that happens, happens somewhere
– GIS keeps track not only events, activities, and things, but also where they happen
• Geography– Where (activity space & migration)– People affected by their environments (natural,
built, social, economic, etc)• Pubic health
– Not simply the absence of disease– State of physical, social, and emotional well-
being of residents
![Page 3: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/3.jpg)
3
What is Geographic Information Systems (GIS)?• Computer hardware & software for capturing, storing,
retrieving, analyzing, and output spatial data. – Maps: a vital role in the analysis (visualization) and display
components of GIS.
Digitizer
User
Remote Sensing
Digital Products
Data InputSubsystem
Data Storageand Retrieval
(DBMS)
Data Manipulationand Analysissubsystem
Reportingand Display
GIS Software
![Page 4: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/4.jpg)
Mapping Health• Spatio-temporal variation
– Geocoding for individual cases– Chropoleth map for aggregated cases – Dot density/size map
![Page 5: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/5.jpg)
Inset 1
5
Pedestrian Crashes in Metro Atlanta
![Page 6: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/6.jpg)
Built Envion’t• Cluster 1
– Statistically significant(P=0.01; 01-04)
• Sidewalks– discontinuous – only available on one side– Apartment complex
bisected by busy roads– Mixed with commercial
properties.
6
![Page 7: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/7.jpg)
Spatial Surveillance• Spatial clustering in GIS
– E.g., Spatial Clustering of HIV in Atlanta (Hixson et al, 2011)
• Identify core central areas of activities
![Page 8: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/8.jpg)
(Kimerling et al 2009)
![Page 9: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/9.jpg)
Space-Time Visualization
![Page 10: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/10.jpg)
Healthcare Access: Routing & Coverage
![Page 11: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/11.jpg)
Mapping spatial access to Mammography• Step 1: float a catchment on Mammography facilities
– For each facility: (1) search all population locations that are within the catchment; (2) inverse the population to obtain facility-population ratio (vj)
![Page 12: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/12.jpg)
Measure spatial access (con’t)• Step 2: float a catchment on population locations
– For each location: (1) search all facilities that are within the catchment; (2) weigh their facility-population ratios (vj) using the kernel function; and (3) summarize the weighted ratios
![Page 13: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/13.jpg)
Measure spatial access to health care (con’t)
• Integrating kernel function (Gaussian) to create weighted populations when computing the facility-population ratio
• Kernel bandwidth = catchment size
Facility
ZIP codecentroid
0
0.2
0.4
0.6
0.8
0 5 10
Distance
Weight
bandwidth
![Page 14: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/14.jpg)
Access to mammography facilities (d=10 min)Late stage breast cancer%
![Page 15: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/15.jpg)
Standard Linear Regression
• Multiple linear regression model:– Regression: Y=a+b1x1+b2x2+….bnxn+ε– Housing Price = Sq.ft. + Age + Median
Income + Dist_Marta + error
• Assumptions– Random errors have a mean of zero– Random errors have a constant variance
and are uncorrelated– Random errors have a normal distribution
• The assumptions may not be always satisfied in practice
Something the model can’t account for
![Page 16: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/16.jpg)
Spatial Regression Model
Why is spatial regression?• First law of geography: values of a
variable systemically related to geographic location
Price = Sq.ft. + Age + Median Income + Dist_Marta + error
Housing price is related to location nearby
Median income is related to location nearby
Housing price is related to median income nearby
Assumptions in standard regression may not be satisfied
![Page 17: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/17.jpg)
Spatial Regression Model
Price = W*Price + Sq.ft. + Age + Median Income + Dist_Marta + error
Spatial autocorrelation
Spatial Lag Model: Y=ρWY+aX+ε• Account for the spatial autocorrelation
![Page 18: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/18.jpg)
A case study using spatial lag model
Late-stage breast cancer and black residential segregation in City of Detroit and its 30-min buffer zone
![Page 19: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/19.jpg)
Standard Regression vs. Spatial Lag Model
St. Regression Model Spatial Lag ModelCoefficients t values Coefficients t values
Constant 0.261** 74.889 0.048** 3.843Black Segregation 0.113** 13.513 0.046** 7.184Spatial Lag 0.790** 17.344R2 0.544 0.817
**significance at the 0.001 level
![Page 20: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/20.jpg)
Mapping Childhood Drowning
• 2002-2008 childhood drowning cases (N=276)– Residential address– Demographic info– Drowning place (descriptive)– Source: Georgia Office of the Child
Advocate
![Page 21: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/21.jpg)
Spatial Smoothing Using Density
![Page 22: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/22.jpg)
Spatial Interpolation
• Spatial Interpolation– Use points with known values to
estimate values at other points• Sample points: points with
known values – The number and distribution
determine the accuracy of spatial interpolation
• Basic assumption– 1st Law of Geography (Everything
is related to everything else, but near things are more related than distant things)
![Page 23: Urban GIS for Health Metrics - WHO/OMS: Extranet Systems · 2018-12-13 · Urban GIS for Health Metrics Presented at International Conference on Urban Health, March 5th, 2014 Dajun](https://reader036.fdocuments.us/reader036/viewer/2022070708/5eb9114e40d21c36c77878bd/html5/thumbnails/23.jpg)
Challenges• Big Data
– Location• Privacy • Accuracy
– Latency– Migration