Tuesday, June 26, 2012

Part 1. Detection of disease local clustering (Bernoulli Model)

Project example: Obesity cluster detection in Atlanta Metro Region (Project 1)
Data: Coordinate of each subject, BMI of each subject  and Obesity status (yes or no)
Software: R, ArcGIS/QGIS, and SatScan
Step 1: Descriptive statistic in R
  - check total number of subject, completeness of data, and descriptive statistics of BMI and obesity.
  - check outliers of BMI and validate if possible.
Step 2: Plot clean data into ArcGIS/QGIS
 - check the projection datum
Step 3. Prepare data set for cluster analysis in SatScan
   - Separate case & control data. (Code in R)
   - Check the variable needed for cluster analysis such as, Unique ID, number of cases for each unique location, coordinate variable name in "lat" and "long".