Application of Fuzzy Clustering and Discriminant Analysis to Identify Neighborhood Population Characteristics and Health Needs

• Activity Number: 257
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Ashely Pedigo* and William Seaver+ and Agricola Odoi
• Companies: University of Tennessee and University of Tennessee and University of Tennessee
• Keywords: Epidemiology ; Fuzzy clustering ; Discriminant analysis ; Health plannig ; Health policy ; Disparities in Care

Abstract

This study classified East Tennessee census tracts into peer neighborhoods (PNs) based on demographic, socioeconomic, and geographic factors so as to identify communities at high risk of stroke and heart attack. This information is useful for understanding neighborhood characteristics and health needs to support population health planning and policy. PNs were identified using fuzzy clustering, and nearest neighbor discriminant analysis was used to validate the PNs. Special attention was paid to misclassified observations, which were geographically located near the borders between two PNs. A closer examination of the misses and the PNs identified variables that differentiated the best. Other multivariate approaches did not yield solid common sense solutions. The identified PN classes will be useful in guiding health service provision, resource allocation and policy decisions.