Abstract
Objective Neighborhood characteristics can have profound effects on resident health. The aim of this study was to use an unsupervised learning approach to reduce the multi-dimensional assessment of a neighborhood using American Community Survey (ACS) data to simplify the assessment of neighborhood influence on health.
Method Multivariate quantitative characterization of the neighborhood was derived by performing a factor analysis on the 2011-2015 ACS data. The utility of the latent variables was examined by determining the association of these factors with poor mental health measures from the 500 Cities Project 2017 release.
Results A five-factor model provided the best fit for the data and the latent factors quantified the following characteristics of the census tract: (1) affluence, (2) proportion of singletons in neighborhood, (3) proportion of African-Americans in neighborhood, (4) proportion of seniors in neighborhood, and (5) proportion of noncitizens in neighborhood. African-Americans (R2 = 0.67) in neighborhood and Affluence (R2 = 0.83) were strongly associated with poor mental health.
Conclusions These findings indicate the importance of this factor model in future research focused on the relationship between neighborhood characteristics and resident health.
Disclosures and acknowledgments
None of the authors have conflicting equity ownership, profit-sharing agreements, royalties, patents. Mrs. Forthman and Drs. Yeh, Kuplicki, and Paulus report no competing interests.