Google Street View Images as Predictors of Patient Health Outcomes, 2017-2019

Area of Science:

• Environmental health
• Urban planning
• Public health

Background:

• Neighborhood built environment data collection is resource-intensive.
• Google Street View (GSV) offers a scalable solution for neighborhood data.
• Previous GSV studies were ecological; this study links to individual health records.

Purpose of the Study:

• To construct neighborhood built environment indicators using GSV images.
• To evaluate associations between built environment and health outcomes.
• To control for individual-level predisposing factors in health outcome analysis.

Main Methods:

• Leveraged 1.4 million GSV images from Utah.
• Constructed neighborhood built environment indicators.
• Analyzed associations with 2017-2019 health outcomes for ~1/3 of Utah's population using electronic medical records.
• Controlled for age, race, ethnicity, religion, marital status, insurance, and area deprivation index.

Main Results:

• Living in areas with more green streets and non-single-family homes was associated with 10-27% lower rates of diabetes, uncontrolled diabetes, hypertension, and obesity.
• Visible utility wires were linked to 5-10% higher rates of these chronic conditions and substance use disorders.
• Non-single-family homes and green streets correlated with lower chronic disease prevalence.
• Visible utility wires and single-lane roads were associated with a higher burden of chronic conditions.

Conclusions:

• Neighborhood built environment characteristics derived from GSV are associated with individual health outcomes.
• Green streets and non-single-family housing may be protective factors against chronic diseases.
• Visible utility wires and single-lane roads may be risk factors for chronic conditions.
• Healthcare organizations can use residential environment data to understand patient health drivers.