PURPOSE OF REVIEW: Innovations in information technology, initiatives by local governments to share administrative data, and growing inventories of data available from commercial data aggregators have immensely expanded the information available to describe neighborhood environments, supporting an approach to research we call Urban Health Informatics. This review evaluates the application of machine learning to this new wealth of data for studies of the effects of neighborhood environments on health. RECENT FINDINGS: Prominent machine learning applications in this field include automated image analysis of archived imagery such as Google Street View images, variable selection methods to identify neighborhood environment factors that predict health outcomes from large pools of exposure variables, and spatial interpolation methods to estimate neighborhood conditions across large geographic areas. SUMMARY: In each domain, we highlight successes and cautions in the application of machine learning, particularly highlighting legal issues in applying machine learning approaches to Google's geo-spatial data.
In-person audits to collect data on neighborhood characteristics offer opportunities to study the mechanisms that link neighborhood conditions to unequal outcomes for individuals and communities, but the expense and logistical difficulties associated with conducting neighborhood audits have limited their use. The images collected by Google Street View provide a promising alternative for researchers to measure neighborhood environments across cities and to examine how neighborhood conditions vary across a wider geographic scope. We describe the benefits of using "virtual" neighborhood audits and discuss the practicalities of collecting data from virtual audits. We provide an example of individual- and neighborhood-level inequality in the distribution of disorder for older adults across four cities: New York, San Jose, Philadelphia, and Detroit. Despite the promise of virtual audits, they also introduce perils that must be addressed as research progresses; we introduce and discuss those perils here.
<p><span style="font-family: Times New Roman;"><strong><span style="font-size: medium;">Objectives:</span></strong><span style="font-size: medium;"> African American children are at higher risk of obesity than White children and African American women are more likely to undergo caesarean-section (CS) delivery than White women.</span><span style="font-size: medium;"> </span><span style="font-size: medium;">CS is associated with childhood obesity, however, little is known whether this relationship varies by race.</span><span style="font-size: medium;">We examined if the association of CS with obesity at age 2 years varied by race.</span><span style="font-size: medium;"> </span></span></p><p><span style="font-family: Times New Roman;"><strong><span style="font-size: medium;">Design: </span></strong><span style="font-size: medium;">Longitudinal birth cohort.</span><strong></strong></span></p><p><span style="font-family: Times New Roman;"><strong><span style="font-size: medium;">Setting:</span></strong><span style="font-size: medium;"> Birth cohort conducted in a health care system in metropolitan Detroit, Michigan with follow-up at age 2 years.</span></span></p><p><span style="font-family: Times New Roman;"><strong><span style="font-size: medium;">Participants:</span></strong><span style="font-size: medium;"> 639 birth cohort participants; 367 children (57.4%) were born to African American mothers and 230 (36.0%) children were born via CS.</span></span></p><p><span style="font-family: Times New Roman;"><strong><span style="font-size: medium;">Main Outcome Measure: </span></strong><span style="font-size: medium;">Obesity defined as body mass index </span><strong></strong><span style="font-size: medium;">≥95</span><sup><span style="font-size: small;">th</span></sup><span style="font-size: medium;"> percentile at age 2 years.</span></span></p><p><span style="font-family: Times New Roman;"><strong><span style="font-size: medium;">Results:</span></strong><span style="font-size: medium;"> Slightly more children of African American (n=37; 10.1%) than non-African American mothers (n=18; 6.6%) were obese (</span><span style="font-size: medium;">P</span><span style="font-size: medium;">=.12). There was evidence of effect modification between race and delivery mode with obesity at age 2 years (interaction<em> </em></span><span style="font-size: medium;">P</span><span style="font-size: medium;">=.020).</span><span style="font-size: medium;"> </span><span style="font-size: medium;">In children of African-American mothers, CS compared to vaginal birth was associated with a significantly higher odds of obesity (aOR=2.35 (95% CI: 1.16, 4.77), </span><em><span style="font-size: medium;">P</span></em><span style="font-size: medium;">=.017).</span><span style="font-size: medium;"> </span><span style="font-size: medium;">In contrast, delivery mode was not associated with obesity at age 2 years in children of non-African-American mothers (aOR=.47 (95% CI: .13, 1.71), </span><span style="font-size: medium;">P</span><span style="font-size: medium;">=.25).</span><span style="font-size: medium;"> </span></span></p><p><span style="font-family: Times New Roman;"><strong><span style="font-size: medium;">Conclusions:</span></strong><span style="font-size: medium;"> There is evidence for a race-specific effect of CS on obesity at age 2 years; potential underlying mechanisms may be racial differences in the developing gut microbiome or in epigenetic programming.</span><span style="font-size: medium;"> </span><span style="font-size: medium;">Future research is needed to determine if this racial difference persists into later childhood. <em>Ethn Dis.</em> 2016;26(1):61-68; doi:10.18865/ed.26.1.61<br /></span></span></p><p> </p>
