Artificial intelligence-based assessment of built environment from Google Street View and coronary artery disease prevalence

BACKGROUND AND AIMS

Built environment plays an important role in the development of cardiovascular disease. Tools to evaluate the built environment using machine vision and informatic approaches have been limited. This study aimed to investigate the association between machine vision-based built environment and prevalence of cardiometabolic disease in US cities.

METHODS

This cross-sectional study used features extracted from Google Street View (GSV) images to measure the built environment and link them with prevalence of coronary heart disease (CHD). Convolutional neural networks, linear mixed-effects models, and activation maps were utilized to predict health outcomes and identify feature associations with CHD at the census tract level. The study obtained 0.53 million GSV images covering 789 census tracts in seven US cities (Cleveland, OH; Fremont, CA; Kansas City, MO; Detroit, MI; Bellevue, WA; Brownsville, TX; and Denver, CO).

RESULTS

Built environment features extracted from GSV using deep learning predicted 63% of the census tract variation in CHD prevalence. The addition of GSV features improved a model that only included census tract-level age, sex, race, income, and education or composite indices of social determinant of health. Activation maps from the features revealed a set of neighbourhood features represented by buildings and roads associated with CHD prevalence.

CONCLUSIONS

In this cross-sectional study, the prevalence of CHD was associated with built environment factors derived from GSV through deep learning analysis, independent of census tract demographics. Machine vision-enabled assessment of the built environment could potentially offer a more precise approach to identify at-risk neighbourhoods, thereby providing an efficient avenue to address and reduce cardiovascular health disparities in urban environments.

Data Insights for Sustainable Cities: Associations between Google Street View-Derived Urban Greenspace and Google Air View-Derived Pollution Levels

Unprecedented levels of urbanization have escalated urban environmental health issues, including increased air pollution in cities globally. Strategies for mitigating air pollution, including green urban planning, are essential for sustainable and healthy cities. State-of-the-art research investigating urban greenspace and pollution metrics has accelerated through the use of vast digital data sets and new analytical tools. In this study, we examined associations between Google Street View-derived urban greenspace levels and Google Air View-derived air quality, where both have been resolved in extremely high resolution, accuracy, and scale along the entire road network of Dublin City. Particulate matter of size fraction less than 2.5 μm (PM2.5), nitrogen dioxide, nitric oxide, carbon monoxide, and carbon dioxide were quantified using 5,030,143 Google Air View measurements, and greenspace was quantified using 403,409 Google Street View images. Significant (p < 0.001) negative associations between urban greenspace and pollution were observed. For example, an interquartile range increase in the Green View Index was associated with a 7.4% [95% confidence interval: -13.1%, -1.3%] decrease in NO2 at the point location spatial resolution. We provide insights into how large-scale digital data can be harnessed to elucidate urban environmental interactions that will have important planning and policy implications for sustainable future cities.

Retailer's Density and Single Stick Cigarette's Accessibility among School-Age Children in Indonesia

OBJECTIVE

The density of single-stick cigarette sales is related to the increase in tobacco epidemic-related diseases. This study aims to provide evidence of retailers' density and radius around the school location, accessibility of single-stick cigarette selling among school-age children, and retailers' response regarding the restriction policy options in urban areas in Indonesia.

METHODS

It is a cross-sectional study. The retailers' spatial density and the radius around schools in Daerah Khusus Ibukota (DKI) Jakarta Province were investigated using Google Maps and Google Street View (GSV). The coordinates of retailers and schools were geo-coded to Kernel Density Map. The accessibility of single-stick cigarettes among children and restriction policy options for cigarette selling were derived from random sampling using surveys of 64 retailers based on Google Data results.

RESULT

Virtually walking using google maps and GSV found 8,371 retailers in DKI Jakarta. There were ± 15 cigarette retailers every 1 km2, and an average of ± one cigarette retailer in every 1,000 residents. There were 456 (21.67%) retailers with a radius ≤ 100 meters around elementary schools, even an increase around junior high school locations of 167 (26.05%) retailers. The accessibility of cigarettes among children is easy because the price is relatively low, at Rp1,500/ $0.11 per stick. In addition, 58.1% of retailers allowed customers to buy on debt. Eleven percent of cigarette retailers intended to reduce the sale of cigarettes if the prohibition of single-stick cigarette sales were applied.

CONCLUSION

Cigarette retailers were very dense and single-stick cigarettes were still accessible to children in Indonesia. The implementation of the prohibition on single-stick cigarette sales should be added for future tobacco control in developing countries such as Indonesia.

Hyperlocal variation of nitrogen dioxide, black carbon, and ultrafine particles measured with Google Street View cars in Amsterdam and Copenhagen

Hyperlocal air quality maps are becoming increasingly common, as they provide useful insights into the spatial variation and sources of air pollutants. In this study, we produced several high-resolution concentration maps to assess the spatial differences of three traffic-related pollutants, Nitrogen dioxide (NO₂), Black Carbon (BC) and Ultrafine Particles (UFP), in Amsterdam, the Netherlands, and Copenhagen, Denmark. All maps were based on a mixed-effect model approach by using state-of-the-art mobile measurements conducted by Google Street View (GSV) cars, during October 2018 - March 2020, and Land-use Regression (LUR) models based on several land-use and traffic predictor variables. We then explored the concentration ratio between the different normalised pollutants to understand possible contributing sources to the observed hyperlocal variations. The maps developed in this work reflect, (i) expected elevated pollution concentrations along busy roads, and (ii) similar concentration patterns on specific road types, e.g., motorways, for both cities. In the ratio maps, we observed a clear pattern of elevated concentrations of UFP near the airport in both cities, compared to BC and NO₂. This is the first study to produce hyperlocal maps for BC and UFP using high-quality mobile measurements. These maps are important for policymakers and health-effect studies, trying to disentangle individual effects of key air pollutants of interest (e.g., UFP).

An evaluation of different approaches which use Google Street View imagery to ground truth land degradation assessments

Member states of the United Nations Convention to Combat Desertification are required to report on the proportion of land that is degraded in their countries, a requirement that is also tied into the UN Sustainable Development Goals (SDGs). National land degradation assessments are often conducted with the use of remote sensing data which are not always ground truthed. Google Street View (GSV) provides high resolution, panoramic imagery across large parts of the world that has the potential to be used to ground truth land degradation assessments. We apply three different methodologies (visual interpretation of GSV images, GSV image classification and vegetation index extraction) to derive vegetation cover estimates from Google Street View imagery for the Hardeveld bioregion of the Succulent Karoo biome in South Africa. Visual estimates of cover best predict known habitat condition values (adjusted R² = 0.86), whilst estimates derived from an unsupervised classification of GSV images also predict habitat condition relatively well (adjusted R² = 0.52). These results show the potential for using GSV imagery, and other large collections of ground-level landscape photographs, as a rough ground-truthing tool, especially in instances where more traditional ground-truthing approaches are not possible.

Mixed-Effects Modeling Framework for Amsterdam and Copenhagen for Outdoor NO2 Concentrations Using Measurements Sampled with Google Street View Cars

High-resolution air quality (AQ) maps based on street-by-street measurements have become possible through large-scale mobile measurement campaigns. Such campaigns have produced data-only maps and have been used to produce empirical models [i.e., land use regression (LUR) models]. Assuming that all road segments are measured, we developed a mixed model framework that predicts concentrations by an LUR model, while allowing road segments to deviate from the LUR prediction based on between-segment variation as a random effect. We used Google Street View cars, equipped with high-quality AQ instruments, and measured the concentration of NO₂ on every street in Amsterdam (n = 46.664) and Copenhagen (n = 28.499) on average seven times over the course of 9 and 16 months, respectively. We compared the data-only mapping, LUR, and mixed model estimates with measurements from passive samplers (n = 82) and predictions from dispersion models in the same time window as mobile monitoring. In Amsterdam, mixed model estimates correlated r_s (Spearman correlation) = 0.85 with external measurements, whereas the data-only approach and LUR model estimates correlated r_s = 0.74 and 0.75, respectively. Mixed model estimates also correlated higher r_s = 0.65 with the deterministic model predictions compared to the data-only (r_s = 0.50) and LUR model (r_s = 0.61). In Copenhagen, mixed model estimates correlated r_s = 0.51 with external model predictions compared to r_s = 0.45 and r_s = 0.50 for data-only and LUR model, respectively. Correlation increased for 97 locations (r_s = 0.65) with more detailed traffic information. This means that the mixed model approach is able to combine the strength of data-only mapping (to show hyperlocal variation) and LUR models by shrinking uncertain concentrations toward the model output.

Measuring Perceived Psychological Stress in Urban Built Environments Using Google Street View and Deep Learning

An urban built environment is an important part of the daily lives of urban residents. Correspondingly, a poor design can lead to psychological stress, which can be harmful to their psychological and physical well-being. The relationship between the urban built environment and the perceived psychological stress of residents is a significant in many disciplines. Further research is needed to determine the stress level experienced by residents in the built environment on a large scale and identify the relationship between the visual components of the built environment and perceived psychological stress. Recent developments in big data and deep learning technology mean that the technical support required to measure the perceived psychological stress of residents has now become available. In this context, this study explored a method for a rapid and large-scale determination of the perceived psychological stress among urban residents through a deep learning approach. An empirical study was conducted in Gangnam District, Seoul, South Korea, and the SegNet deep learning algorithm was used to segment and classify the visual elements of street views. In addition, a human-machine adversarial model using random forest as a framework was employed to score the perception of the perceived psychological stress in the built environment. Consequently, we found a strong spatial autocorrelation in the perceived psychological stress in space, with more low-low clusters in the urban traffic arteries and riverine areas in Gangnam district and more high-high clusters in the commercial and residential areas. We also analyzed the street view images for three types of stress perception (i.e., low, medium and high) and obtained the percentage of each street view element combination under different stresses. Using multiple linear regression, we found that walls and buildings cause psychological stress, whereas sky, trees and roads relieve it. Our analytical study integrates street view big data with deep learning and proposes an innovative method for measuring the perceived psychological stress of residents in the built environment. The research methodology and results can be a reference for urban planning and design from a human centered perspective.

Training Computers to See the Built Environment Related to Physical Activity: Detection of Microscale Walkability Features Using Computer Vision

The study purpose was to train and validate a deep learning approach to detect microscale streetscape features related to pedestrian physical activity. This work innovates by combining computer vision techniques with Google Street View (GSV) images to overcome impediments to conducting audits (e.g., time, safety, and expert labor cost). The EfficientNETB5 architecture was used to build deep learning models for eight microscale features guided by the Microscale Audit of Pedestrian Streetscapes Mini tool: sidewalks, sidewalk buffers, curb cuts, zebra and line crosswalks, walk signals, bike symbols, and streetlights. We used a train−correct loop, whereby images were trained on a training dataset, evaluated using a separate validation dataset, and trained further until acceptable performance metrics were achieved. Further, we used trained models to audit participant (N = 512) neighborhoods in the WalkIT Arizona trial. Correlations were explored between microscale features and GIS-measured and participant-reported neighborhood macroscale walkability. Classifier precision, recall, and overall accuracy were all over >84%. Total microscale was associated with overall macroscale walkability (r = 0.30, p < 0.001). Positive associations were found between model-detected and self-reported sidewalks (r = 0.41, p < 0.001) and sidewalk buffers (r = 0.26, p < 0.001). The computer vision model results suggest an alternative to trained human raters, allowing for audits of hundreds or thousands of neighborhoods for population surveillance or hypothesis testing.

Assessing streetscape greenery with deep neural network using Google Street View

The importance of greenery in urban areas has traditionally been discussed from ecological and esthetic perspectives, as well as in public health and social science fields. The recent advancements in empirical studies were enabled by the combination of 'big data' of streetscapes and automated image recognition. However, the existing methods of automated image recognition for urban greenery have problems such as the confusion of green artificial objects and the excessive cost of model training. To ameliorate the drawbacks of existing methods, this study proposes to apply a patch-based semantic segmentation method for determining the green view index of certain urban areas by using Google Street View imagery and the 'chopped picture method'. We expect that our method will contribute to expanding the scope of studies on urban greenery in various fields.

Google Street View-Derived Neighborhood Characteristics in California Associated with Coronary Heart Disease, Hypertension, Diabetes

Characteristics of the neighborhood built environment influence health and health behavior. Google Street View (GSV) images may facilitate measures of the neighborhood environment that are meaningful, practical, and adaptable to any geographic boundary. We used GSV images and computer vision to characterize neighborhood environments (green streets, visible utility wires, and dilapidated buildings) and examined cross-sectional associations with chronic health outcomes among patients from the University of California, San Francisco Health system with outpatient visits from 2015 to 2017. Logistic regression models were adjusted for patient age, sex, marital status, race/ethnicity, insurance status, English as preferred language, assignment of a primary care provider, and neighborhood socioeconomic status of the census tract in which the patient resided. Among 214,163 patients residing in California, those living in communities in the highest tertile of green streets had 16–29% lower prevalence of coronary artery disease, hypertension, and diabetes compared to those living in communities in the lowest tertile. Conversely, a higher presence of visible utility wires overhead was associated with 10–26% more coronary artery disease and hypertension, and a higher presence of dilapidated buildings was associated with 12–20% greater prevalence of coronary artery disease, hypertension, and diabetes. GSV images and computer vision models can be used to understand contextual factors influencing patient health outcomes and inform structural and place-based interventions to promote population health.

"Biophilic Cities": Quantifying the Impact of Google Street View-Derived Greenspace Exposures on Socioeconomic Factors and Self-Reported Health

According to the biophilia hypothesis, humans have evolved to prefer natural environments that are essential to their thriving. With urbanization occurring at an unprecedented rate globally, urban greenspace has gained increased attention due to its environmental, health, and socioeconomic benefits. To unlock its full potential, an increased understanding of greenspace metrics is urgently required. In this first-of-a-kind study, we quantified street-level greenspace using 751 644 Google Street View images and computer vision methods for 125 274 locations in Ireland's major cities. We quantified population-weighted exposure to greenspace and investigated the impact of greenspace on health and socioeconomic determinants. To investigate the association between greenspace and self-reported health, a negative binomial regression analysis was applied. While controlling for other factors, an interquartile range increase in street-level greenspace was associated with a 2.78% increase in self-reported "good or very good" health [95% confidence interval: 2.25-3.31]. Additionally, we observed that populations in upper quartiles of greenspace exposure had higher levels of income and education than those in lower quartiles. This study provides groundbreaking insights into how urban greenspace can be quantified in unprecedented resolution, accuracy, and scale while also having important implications for urban planning and environmental health research and policy.

Evaluating Closures of Fresh Fruit and Vegetable Vendors During the COVID-19 Pandemic: Methodology and Preliminary Results Using Omnidirectional Street View Imagery

BACKGROUND

The COVID-19 pandemic has significantly disrupted the food retail environment. However, its impact on fresh fruit and vegetable vendors remains unclear; these are often smaller, more community centered, and may lack the financial infrastructure to withstand supply and demand changes induced by such crises.

OBJECTIVE

This study documents the methodology used to assess fresh fruit and vegetable vendor closures in New York City (NYC) following the start of the COVID-19 pandemic by using Google Street View, the new Apple Look Around database, and in-person checks.

METHODS

In total, 6 NYC neighborhoods (in Manhattan and Brooklyn) were selected for analysis; these included two socioeconomically advantaged neighborhoods (Upper East Side, Park Slope), two socioeconomically disadvantaged neighborhoods (East Harlem, Brownsville), and two Chinese ethnic neighborhoods (Chinatown, Sunset Park). For each neighborhood, Google Street View was used to virtually walk down each street and identify vendors (stores, storefronts, street vendors, or wholesalers) that were open and active in 2019 (ie, both produce and vendor personnel were present at a location). Past vendor surveillance (when available) was used to guide these virtual walks. Each identified vendor was geotagged as a Google Maps pinpoint that research assistants then physically visited. Using the "notes" feature of Google Maps as a data collection tool, notes were made on which of three categories best described each vendor: (1) open, (2) open with a more limited setup (eg, certain sections of the vendor unit that were open and active in 2019 were missing or closed during in-person checks), or (3) closed/absent.

RESULTS

Of the 135 open vendors identified in 2019 imagery data, 35% (n=47) were absent/closed and 10% (n=13) were open with more limited setups following the beginning of the COVID-19 pandemic. When comparing boroughs, 35% (28/80) of vendors in Manhattan were absent/closed, as were 35% (19/55) of vendors in Brooklyn. Although Google Street View was able to provide 2019 street view imagery data for most neighborhoods, Apple Look Around was required for 2019 imagery data for some areas of Park Slope. Past surveillance data helped to identify 3 additional established vendors in Chinatown that had been missed in street view imagery. The Google Maps "notes" feature was used by multiple research assistants simultaneously to rapidly collect observational data on mobile devices.

CONCLUSIONS

The methodology employed enabled the identification of closures in the fresh fruit and vegetable retail environment and can be used to assess closures in other contexts. The use of past baseline surveillance data to aid vendor identification was valuable for identifying vendors that may have been absent or visually obstructed in the street view imagery data. Data collection using Google Maps likewise has the potential to enhance the efficiency of fieldwork in future studies.

The Effects of Neighborhood Built Environment on Walking for Leisure and for Purpose Among Older People

BACKGROUND AND OBJECTIVES

Characteristics of a neighborhood's built environment affect the walking behavior of older people, yet studies typically rely on small nonrepresentative samples that use either subjective reports or aggregate indicators from administrative sources to represent neighborhood characteristics. Our analyses examine the usefulness of a novel method for observing neighborhoods-virtual observations-and assess the extent to which virtual-based observations predict walking among older adults.

RESEARCH DESIGN AND METHODS

Using Google Street View, we observed the neighborhoods of 2,224 older people and examined how characteristics of the neighborhood built environments are associated with the amount of time older people spend walking for leisure and purpose.

RESULTS

Multilevel model analyses revealed that sidewalk characteristics had significant associations with both walking for purpose and leisure. Land use, including the presence of multifamily dwellings, commercial businesses, and parking lots were positively associated with walking for purpose and single-family detached homes were negatively associated with walking for purpose, but none of these characteristics were associated with leisure walking. Gardens/flowers were associated with walking for leisure but not purpose. Garbage/litter was not associated with either type of walking behavior.

DISCUSSION AND IMPLICATIONS

Virtual observations are a useful method that provides meaningful information about neighborhoods. Findings demonstrate how neighborhood characteristics assessed virtually differentially impact walking for leisure and purpose among older adults and are interpreted within a social-ecological model.

Virtual audits of the urban streetscape: comparing the inter-rater reliability of GigaPan® to Google Street View

Background

Although previous research has highlighted the association between the built environment and individual health, methodological challenges in assessing the built environment remain. In particular, many researchers have demonstrated the high inter-rater reliability of assessing large or objective built environment features and the low inter-rater reliability of assessing small or subjective built environment features using Google Street View. New methods for auditing the built environment must be evaluated to understand if there are alternative tools through which researchers can assess all types of built environment features with high agreement. This paper investigates measures of inter-rater reliability of GigaPan®, a tool that assists with capturing high-definition panoramic images, relative to Google Street View.

Methods

Street segments (n = 614) in Pittsburgh, Pennsylvania in the United States were randomly selected to audit using GigaPan® and Google Street View. Each audit assessed features related to land use, traffic and safety, and public amenities. Inter-rater reliability statistics, including percent agreement, Cohen’s kappa, and the prevalence-adjusted bias-adjusted kappa (PABAK) were calculated for 106 street segments that were coded by two, different, human auditors.

Results

Most large-scale, objective features (e.g. bus stop presence or stop sign presence) demonstrated at least substantial inter-rater reliability for both methods, but significant differences emerged across finely detailed features (e.g. trash) and features at segment endpoints (e.g. sidewalk continuity). After adjusting for the effects of bias and prevalence, the inter-rater reliability estimates were consistently higher for almost all built environment features across GigaPan® and Google Street View.

Conclusion

GigaPan® is a reliable, alternative audit tool to Google Street View for studying the built environment. GigaPan® may be particularly well-suited for built environment projects with study settings in areas where Google Street View imagery is nonexistent or updated infrequently. The potential for enhanced, detailed imagery using GigaPan® will be most beneficial in studies in which current, time sensitive data are needed or microscale built environment features would be challenging to see in Google Street View. Furthermore, to better understand the effects of prevalence and bias in future reliability studies, researchers should consider using PABAK to supplement or expand upon Cohen’s kappa findings.

Assessing Google Street View Image Availability in Latin American Cities

Virtual audits using Google Street View are an increasingly popular method of assessing neighborhood environments for health and urban planning research. However, the validity of these studies may be threatened by issues of image availability, image age, and variance of image age, particularly in the Global South. This study identifies patterns of Street View image availability, image age, and image age variance across cities in Latin America and assesses relationships between these measures and measures of resident socioeconomic conditions. Image availability was assessed at 530,308 near-road points within the boundaries of 371 Latin American cities described by the SALURBAL (Salud Urbana en America Latina) project. At the subcity level, mixed-effect linear and logistic models were used to assess relationships between measures of socioeconomic conditions and image availability, average image age, and the standard deviation of image age. Street View imagery was available at 239,394 points (45.1%) of the total sampled, and rates of image availability varied widely between cities and countries. Subcity units with higher scores on measures of socioeconomic conditions had higher rates of image availability (OR = 1.11 per point increase of combined index, p < 0.001) and the imagery was newer on average (- 1.15 months per point increase of combined index, p < 0.001), but image capture date within these areas varied more (0.59-month increase in standard deviation of image age per point increase of combined index, p < 0.001). All three assessed threats to the validity of Street View virtual audit studies spatially covary with measures of socioeconomic conditions in Latin American cities. Researchers should be attentive to these issues when using Street View imagery.

Health and the built environment in United States cities: measuring associations using Google Street View-derived indicators of the built environment

BACKGROUND

The built environment is a structural determinant of health and has been shown to influence health expenditures, behaviors, and outcomes. Traditional methods of assessing built environment characteristics are time-consuming and difficult to combine or compare. Google Street View (GSV) images represent a large, publicly available data source that can be used to create indicators of characteristics of the physical environment with machine learning techniques. The aim of this study is to use GSV images to measure the association of built environment features with health-related behaviors and outcomes at the census tract level.

METHODS

We used computer vision techniques to derive built environment indicators from approximately 31 million GSV images at 7.8 million intersections. Associations between derived indicators and health-related behaviors and outcomes on the census-tract level were assessed using multivariate regression models, controlling for demographic factors and socioeconomic position. Statistical significance was assessed at the α = 0.05 level.

RESULTS

Single lane roads were associated with increased diabetes and obesity, while non-single-family home buildings were associated with decreased obesity, diabetes and inactivity. Street greenness was associated with decreased prevalence of physical and mental distress, as well as decreased binge drinking, but with increased obesity. Socioeconomic disadvantage was negatively associated with binge drinking prevalence and positively associated with all other health-related behaviors and outcomes.

CONCLUSIONS

Structural determinants of health such as the built environment can influence population health. Our study suggests that higher levels of urban development have mixed effects on health and adds further evidence that socioeconomic distress has adverse impacts on multiple physical and mental health outcomes.

Expanding Tools for Investigating Neighborhood Indicators of Drug Use and Violence: Validation of the NIfETy for Virtual Street Observation

A growing body of evidence suggests that characteristics of the neighborhood environment in urban areas significantly impact risk for drug use behavior and exposure to violent crime. Identifying areas of community need, prioritizing planning projects, and developing strategies for community improvement require inexpensive, easy to use, evidence-based tools to assess neighborhood disorder that can be used for a variety of research, urban planning, and community needs with an environmental justice frame. This study describes validation of the Neighborhood Inventory for Environmental Typology (NIfETy), a neighborhood environmental observational assessment tool designed to assess characteristics of the neighborhood environment related to violence, alcohol, and other drugs, for use with Google Street View (GSV). GSV data collection took place on a random sample of 350 blocks located throughout Baltimore City, Maryland, which had previously been assessed through in-person data collection. Inter-rater reliability metrics were strong for the majority of items (ICC ≥ 0.7), and items were highly correlated with in-person observations (r ≥ 0.6). Exploratory factor analysis and constrained factor analysis resulted in one, 14-item disorder scale with high internal consistency (alpha = 0.825) and acceptable fit indices (CFI = 0.982; RMSEA = 0.051). We further validated this disorder scale against locations of violent crimes, and we found that disorder score was significantly and positively associated with neighborhood crime (IRR = 1.221, 95% CI = (1.157, 1.288), p < 0.001). The NIfETy provides a valid, economical, and efficient tool for assessing modifiable neighborhood risk factors for drug use and violence prevention that can be employed for a variety of research, urban planning, and community needs.

Assessing the micro-scale environment using Google Street View: the Virtual Systematic Tool for Evaluating Pedestrian Streetscapes (Virtual-STEPS)

Background

Altering micro-scale features of neighborhood walkability (e.g., benches, sidewalks, and cues of social disorganization or crime) could be a relatively cost-effective method of creating environments that are conducive to active living. Traditionally, measuring the micro-scale environment has required researchers to perform observational audits. Technological advances have led to the development of virtual audits as alternatives to observational field audits with the enviable properties of cost-efficiency from elimination of travel time and increased safety for auditors. This study examined the reliability of the Virtual Systematic Tool for Evaluating Pedestrian Streetscapes (Virtual-STEPS), a Google Street View-based auditing tool specifically designed to remotely assess micro-scale characteristics of the built environment.

Methods

We created Virtual-STEPS, a tool with 40 items categorized into 6 domains (pedestrian infrastructure, traffic calming and streets, building characteristics, bicycling infrastructure, transit, and aesthetics). Items were selected based on their past abilities to predict active living and on their feasibility for a virtual auditing tool. Two raters performed virtual and field audits of street segments in Montreal neighborhoods stratified by the Walkscore that was used to determine the ‘walking-friendliness’ of a neighborhood. The reliability between virtual and field audits (n = 40), as well as inter-rater reliability (n = 60) were assessed using percent agreement, Cohen’s Kappa statistic, and the Intra-class Correlation Coefficient.

Results

Virtual audits and field audits (excluding travel time) took similar amounts of time to perform (9.8 versus 8.2 min). Percentage agreement between virtual and field audits, and for inter-rater agreement was 80% or more for the majority of items included in the Virtual-STEPS tool. There was high reliability between virtual and field audits with Kappa and ICC statistics indicating that 20 out of 40 (50.0%) items had almost perfect agreement and 13 (32.5%) items had substantial agreement. Inter-rater reliability was also high with 17 items (42.5%) with almost perfect agreement and 11 (27.5%) items with substantial agreement.

Conclusions

Virtual-STEPS is a reliable tool. Tools that measure the micro-scale environment are important because changing this environment could be a relatively cost-effective method of creating environments that are conducive to active living.

Electronic supplementary material

The online version of this article (10.1186/s12889-019-7460-3) contains supplementary material, which is available to authorized users.

Using Google Street View to examine associations between built environment characteristics and U.S. health outcomes

Neighborhood attributes have been shown to influence health, but advances in neighborhood research has been constrained by the lack of neighborhood data for many geographical areas and few neighborhood studies examine features of nonmetropolitan locations. We leveraged a massive source of Google Street View (GSV) images and computer vision to automatically characterize national neighborhood built environments. Using road network data and Google Street View API, from December 15, 2017-May 14, 2018 we retrieved over 16 million GSV images of street intersections across the United States. Computer vision was applied to label each image. We implemented regression models to estimate associations between built environments and county health outcomes, controlling for county-level demographics, economics, and population density. At the county level, greater presence of highways was related to lower chronic diseases and premature mortality. Areas characterized by street view images as 'rural' (having limited infrastructure) had higher obesity, diabetes, fair/poor self-rated health, premature mortality, physical distress, physical inactivity and teen birth rates but lower rates of excessive drinking. Analyses at the census tract level for 500 cities revealed similar adverse associations as was seen at the county level for neighborhood indicators of less urban development. Possible mechanisms include the greater abundance of services and facilities found in more developed areas with roads, enabling access to places and resources for promoting health. GSV images represents an underutilized resource for building national data on neighborhoods and examining the influence of built environments on community health outcomes across the United States.

Associations between Body Mass Index and Urban "Green" Streetscape in Cleveland, Ohio, USA

Public health researchers are increasingly interested in assessing the impact of neighborhood environment on physical activities and chronic health issues among humans. Walkable streets and proximity to green space have long been believed to promote active lifestyles in cities, which contribute to positive health outcomes among residents. Traditionally, urban environmental metrics were calculated at the area level to describe the physical environment of neighborhoods. However, considering the fact that streets are the basic unit for human activities in cities, it is important to understand how the streetscape environment can influence human health conditions. In this study, we investigated the influence of street greenery and walkability on body mass index in Cleveland, Ohio, USA. Different from the area level and overhead view greenery metrics, we used the green view index calculated from the Google Street View to represent the amount of street greenery. The Walk Score was used to indicate the walkability of neighborhoods also at the street level. Statistical analysis results show that the Walk Score has a more significant association with decreased BMI for males than females and the street greenery has a more significant association with decreased BMI for females than males in Cleveland, Ohio. The results of this study would provide a reference for designing gender-specific healthy cities.

Using Deep Learning to Identify Utility Poles with Crossarms and Estimate Their Locations from Google Street View Images

Traditional methods of detecting and mapping utility poles are inefficient and costly because of the demand for visual interpretation with quality data sources or intense field inspection. The advent of deep learning for object detection provides an opportunity for detecting utility poles from side-view optical images. In this study, we proposed using a deep learning-based method for automatically mapping roadside utility poles with crossarms (UPCs) from Google Street View (GSV) images. The method combines the state-of-the-art DL object detection algorithm (i.e., the RetinaNet object detection algorithm) and a modified brute-force-based line-of-bearing (LOB, a LOB stands for the ray towards the location of the target [UPC at here] from the original location of the sensor [GSV mobile platform]) measurement method to estimate the locations of detected roadside UPCs from GSV. Experimental results indicate that: (1) both the average precision (AP) and the overall accuracy (OA) are around 0.78 when the intersection-over-union (IoU) threshold is greater than 0.3, based on the testing of 500 GSV images with a total number of 937 objects; and (2) around 2.6%, 47%, and 79% of estimated locations of utility poles are within 1 m, 5 m, and 10 m buffer zones, respectively, around the referenced locations of utility poles. In general, this study indicates that even in a complex background, most utility poles can be detected with the use of DL, and the LOB measurement method can estimate the locations of most UPCs.

Systematic review of the use of Google Street View in health research: Major themes, strengths, weaknesses and possibilities for future research

We systematically reviewed the current use of Google Street View (GSV) in health research and characterized major themes, strengths and weaknesses in order to highlight possibilities for future research. Of 54 qualifying studies, we found that most used GSV to assess the neighborhood built environment, followed by health policy compliance, study site selection, and disaster preparedness. Most studies were conducted in urban areas of North America, Europe, or New Zealand, with few studies from South America or Asia and none from Africa or rural areas. Health behaviors and outcomes of interest in these studies included injury, alcohol and tobacco use, physical activity and mental health. Major strengths of using GSV imagery included low cost, ease of use, and time saved. Identified weaknesses were image resolution and spatial and temporal availability, largely in developing regions of the world. Despite important limitations, GSV is a promising tool for automated environmental assessment for health research. Currently untapped areas of health research using GSV include identification of sources of air, soil or water pollution, park design and usage, amenity design and longitudinal research on neighborhood conditions.

Neighborhood Disorder and Obesity-Related Outcomes among Women in Chicago

Neighborhood psychosocial stressors like crime and physical disorder may influence obesity-related outcomes through chronic stress or through adverse effects on health behaviors. Google Street View imagery provides a low-cost, reliable method for auditing neighborhood physical disorder, but few studies have examined associations of Street View-derived physical disorder scores with health outcomes. We used Google Street View to audit measures of physical disorder for residential census blocks from 225 women aged 18⁻44 enrolled from 4 Chicago neighborhoods. Latent neighborhood physical disorder scores were estimated using an item response theory model and aggregated to the block group level. Block-group level physical disorder scores and rates of police-recorded crime and 311 calls for service requests were linked to participants based on home addresses. Associations were estimated for 6 obesity-related outcomes: body mass index, obesity, total moderate-to-vigorous physical activity, and weekly consumption of sugar-sweetened beverages, fast food, and snacks. Hierarchical regression models estimated cross-sectional associations adjusting for individual sociodemographics and neighborhood poverty. Higher neighborhood physical disorder was associated with greater odds of obesity (OR: 1.43, 95% CI: 1.01, 2.02). Living in a neighborhood with a higher crime rate was associated with an increase in weekly snack consumption of 3.06 (95% CI: 1.59, 4.54).

Smokefree signage at children's playgrounds: Field observations and comparison with Google Street View

Background

Although there is global growth in outdoor smokefree areas, little is known about the associated smokefree signage. We aimed to study smokefree signage at playgrounds and to compare field observations with images from Google Street View (GSV).

Methods

We randomly selected playgrounds in 21 contiguous local government areas in the lower North Island of New Zealand, all of which had smokefree playground policies. Field data were collected on smokefree signage along with dog control signage to allow for comparisons. The sensitivity and specificity of using GSV for data collection were calculated.

Results

Out of the 63 playgrounds studied, only 44% (95% CI: 33%–57%) had any smokefree signage within 10 m of the playground equipment. The mean number of such signs was 0.8 per playground (range: 0 to 6). Sign size varied greatly from 42 cm² up to 2880 cm²; but was typically fairly small (median = 600 cm²; ie, as per a 20 × 30 cm rectangle). Qualitatively the dog signs appeared to use clearer images and were less wordy than the smokefree signs.

Most playground equipment (82%), could be seen on GSV, but for these settings the sensitivity for identifying smokefree signs was poor at 16%. Yet specificity was reasonable at 96%.

Conclusions

The presence and quality of smokefree signage was poor in this sample of children’s playgrounds in this developed country setting. There appears to be value in comparing smokefree signage with other types of signage (eg, dog control signage). Google Street View was not a sensitive tool for studying such signage.

Electronic supplementary material

The online version of this article (10.1186/s12971-017-0143-x) contains supplementary material, which is available to authorized users.

Street Audits to Measure Neighborhood Disorder: Virtual or In-Person?

Neighborhood conditions may influence a broad range of health indicators, including obesity, injury, and psychopathology. In particular, neighborhood physical disorder-a measure of urban deterioration-is thought to encourage crime and high-risk behaviors, leading to poor mental and physical health. In studies to assess neighborhood physical disorder, investigators typically rely on time-consuming and expensive in-person systematic neighborhood audits. We compared 2 audit-based measures of neighborhood physical disorder in the city of Detroit, Michigan: One used Google Street View imagery from 2009 and the other used an in-person survey conducted in 2008. Each measure used spatial interpolation to estimate disorder at unobserved locations. In total, the virtual audit required approximately 3% of the time required by the in-person audit. However, the final physical disorder measures were significantly positively correlated at census block centroids (r = 0.52), identified the same regions as highly disordered, and displayed comparable leave-one-out cross-validation accuracy. The measures resulted in very similar convergent validity characteristics (correlation coefficients within 0.03 of each other). The virtual audit-based physical disorder measure could substitute for the in-person one with little to no loss of precision. Virtual audits appear to be a viable and much less expensive alternative to in-person audits for assessing neighborhood conditions.

Invited Commentary: Observing Neighborhood Physical Disorder in an Age of Technological Innovation

Researchers across several disciplines have argued that the characteristics of neighborhood environments can affect a variety of individual- and neighborhood-level outcomes. Physical disorder is one feature of neighborhoods that scholars have argued is important, but data that capture physical disorder have been limited because of the time and resources required for in-person audits. The advent of Google Street View, which provides publicly available street-level imagery with nearly complete coverage of the United States, opens new possibilities for researchers. In this issue of the Journal, Mooney et al. (Am J Epidemiol. 2017;186(3):265-273) compare in-person and virtual audits in Detroit, Michigan, and demonstrate that virtual audits offer key advantages to measuring neighborhood physical disorder over in-person audits, including substantial reductions in time and resources with little to no loss of measurement precision. In this invited commentary, I welcome the use of virtual audits for advancing the study of neighborhoods and outline areas in which they can advance understanding of neighborhood effects. I also describe areas of caution in their implementation and outline how new innovations can advance the use of virtual audits for furthering understanding of neighborhood environments.

​Parcel feature data derived from Google Street View images for urban land use classification in Brooklyn, New York Cityfor urban land use classification in Brooklyn, New York Cityretain-->

Google Street View (GSV) was used for urban land use classification, together with airborne light detection and ranging (LiDAR) data and high resolution orthoimagery, by a parcel-based method. In this data article, we present the input raw GSV images, intermediate products of GSV images, and final urban land use classification data that are related to our research article "Parcel-based urban land use classification in megacity using airborne LiDAR, high resolution orthoimagery, and Google Street View" (Zhang et al., 2017) [1]. More detail about other used data and our findings can be found in Zhang et al. (2017) [1].

Validation of a Google Street View-Based Neighborhood Disorder Observational Scale

Recently, there has been a growing interest in developing new tools to measure neighborhood features using the benefits of emerging technologies. This study aimed to assess the psychometric properties of a neighborhood disorder observational scale using Google Street View (GSV). Two groups of raters conducted virtual audits of neighborhood disorder on all census block groups (N = 92) in a district of the city of Valencia (Spain). Four different analyses were conducted to validate the instrument. First, inter-rater reliability was assessed through intraclass correlation coefficients, indicating moderated levels of agreement among raters. Second, confirmatory factor analyses were performed to test the latent structure of the scale. A bifactor solution was proposed, comprising a general factor (general neighborhood disorder) and two specific factors (physical disorder and physical decay). Third, the virtual audit scores were assessed with the physical audit scores, showing a positive relationship between both audit methods. In addition, correlations between the factor scores and socioeconomic and criminality indicators were assessed. Finally, we analyzed the spatial autocorrelation of the scale factors, and two fully Bayesian spatial regression models were run to study the influence of these factors on drug-related police interventions and interventions with young offenders. All these indicators showed an association with the general neighborhood disorder. Taking together, results suggest that the GSV-based neighborhood disorder scale is a reliable, concise, and valid instrument to assess neighborhood disorder using new technologies.

Optimizing Scoring and Sampling Methods for Assessing Built Neighborhood Environment Quality in Residential Areas

Optimization of existing measurement tools is necessary to explore links between aspects of the neighborhood built environment and health behaviors or outcomes. We evaluate a scoring method for virtual neighborhood audits utilizing the Active Neighborhood Checklist (the Checklist), a neighborhood audit measure, and assess street segment representativeness in low-income neighborhoods. Eighty-two home neighborhoods of Washington, D.C. Cardiovascular Health/Needs Assessment (NCT01927783) participants were audited using Google Street View imagery and the Checklist (five sections with 89 total questions). Twelve street segments per home address were assessed for (1) Land-Use Type; (2) Public Transportation Availability; (3) Street Characteristics; (4) Environment Quality and (5) Sidewalks/Walking/Biking features. Checklist items were scored 0–2 points/question. A combinations algorithm was developed to assess street segments’ representativeness. Spearman correlations were calculated between built environment quality scores and Walk Score^®, a validated neighborhood walkability measure. Street segment quality scores ranged 10–47 (Mean = 29.4 ± 6.9) and overall neighborhood quality scores, 172–475 (Mean = 352.3 ± 63.6). Walk scores^® ranged 0–91 (Mean = 46.7 ± 26.3). Street segment combinations’ correlation coefficients ranged 0.75–1.0. Significant positive correlations were found between overall neighborhood quality scores, four of the five Checklist subsection scores, and Walk Scores^® (r = 0.62, p < 0.001). This scoring method adequately captures neighborhood features in low-income, residential areas and may aid in delineating impact of specific built environment features on health behaviors and outcomes.

Alcohol in urban streetscapes: a comparison of the use of Google Street View and on-street observation

Background

Alcohol-related harm is a major global health issue, and controls on alcohol marketing are one intervention utilized by governments. This study investigated the use of Google Street View (GSV) as a novel research method for collecting alcohol-related data in the urban environment.

Methods

The efficacy of GSV and on-street observation by observer teams was compared by surveying 400 m stretches of 12 streets in Wellington, the capital city of New Zealand. Data on alcohol sale, alcohol-related advertising, health promotion materials, regulatory information and visible alcohol consumption were collected.

Results

A total of 403 retailers with evidence of alcohol sales and 1161 items of alcohol-related communication were identified in on-street observation. Of the latter, 1028 items (89 %) were for alcohol marketing and 133 (11 %) were for alcohol-related health promotion and alcohol regulation. GSV was found to be a less sensitive tool than on-street observation with only 50 % of the alcohol venues identified and 52 % of the venue-associated brand marketing identified. A high degree of inter-observer reliability was generally found between pairs of observers e.g., for the detection of alcohol retail venues the intra-class correlation coefficient (ICC) was 0.93 (95 % CI: 0.78 to 0.98) for on-street observation and 0.85 (95 % CI: 0.49 to 0.96) for using GSV.

Conclusions

GSV does not seem suitable for the comprehensive study of the influences on alcohol consumption in the urban streetscape. However, it may still have value for large, static objects in the environment and be more time efficient than traditional on-street observation measures, especially when used to collect data across a wide geographical area. Furthermore, GSV might become a more useful research tool in settings with better image quality (such as more ‘footpath views’) and with more regularly updated GSV imagery.

Residential environments, alcohol advertising, and initiation and continuation of alcohol consumption among adolescents in urban Taiwan: A prospective multilevel study

Background

Research indicates that place characteristics and the media environment are important contextual determinants of underage drinking behaviors in Western countries, but it is unknown whether these exposures influence adolescent alcohol consumption outside Western contexts, including in Asia׳s emerging global alcohol markets. Guided by the social ecological framework, we prospectively investigated the influences of place characteristics and alcohol advertising on initiation and continuation of alcohol consumption among adolescents in Taipei, Taiwan.

Methods

Data on individual-level characteristics, including alcohol use behaviors and perceived exposure to alcohol advertising, were obtained from two waves of a longitudinal school-based study through a stratified probability sampling method in 2010 (Grade 7/Grade 8, aged 13-14 years old) and 2011-2012 (Grade 9, aged 15 years old) from 1795 adolescents residing in 22 of 41 districts in Taipei. Data on district-level characteristics were drawn from administrative sources and Google Street View virtual audit to describe districts where adolescents lived at baseline. Hierarchical generalized linear models tested hypotheses about the associations of place characteristics and perceived alcohol advertising with underage drinking, with stratification by baseline lifetime alcohol consumption.

Results

Among alcohol-naïve adolescents, lower district-level economic disadvantage, a higher proportion of betel nut kiosks (a relatively unregulated alcohol source) compared to off-premises alcohol outlets, and exposure to television-based alcohol advertising predicted increased likelihood of alcohol initiation at one-year follow-up. Among alcohol-experienced adolescents, greater spatial access to off-premises alcohol outlets, and lower access to metro rapid transportation (MRT) and to temples were found to predict a subsequent increased likelihood of continued alcohol use. Parental drinking moderated the relationship between district-level violent crime and initiation of alcohol consumption.

Conclusions

These findings suggest that local social economic status, alcohol access, and institutional resource and individual media exposure affect underage drinking behaviors in Taiwan. We discuss potential public health implications for place-based interventions. Future research on place, media, and adolescent alcohol consumption in Asian contexts is warranted.

Matching study areas using Google Street View: A new application for an emerging technology

Google Street View (GSV) can be used as an effective tool to conduct virtual neighborhood audits. We expand on this research by exploring the utility of a GSV-based neighborhood audit to measure and match target and comparison study areas. We developed a GSV-based inventory to measure characteristics of retail alcohol stores and their surrounding neighborhoods. We assessed its reliability and assessed the utility of GSV-based audits for matching target and comparison study areas. We found that GSV-based neighborhood audits can be a useful, reliable, and cost-effective tool for matching target and comparison study areas when archival data are insufficient and primary data collection is prohibitive. We suggest that researchers focus on characteristics that are easily visible on GSV and are relatively stable over time when creating future GSV-based measuring and matching tools. Dividing the study area into small segments may also provide more accurate measurements and more precise matching.

Development and deployment of the Computer Assisted Neighborhood Visual Assessment System (CANVAS) to measure health-related neighborhood conditions

Public health research has shown that neighborhood conditions are associated with health behaviors and outcomes. Systematic neighborhood audits have helped researchers measure neighborhood conditions that they deem theoretically relevant but not available in existing administrative data. Systematic audits, however, are expensive to conduct and rarely comparable across geographic regions. We describe the development of an online application, the Computer Assisted Neighborhood Visual Assessment System (CANVAS), that uses Google Street View to conduct virtual audits of neighborhood environments. We use this system to assess the inter-rater reliability of 187 items related to walkability and physical disorder on a national sample of 150 street segments in the United States. We find that many items are reliably measured across auditors using CANVAS and that agreement between auditors appears to be uncorrelated with neighborhood demographic characteristics. Based on our results we conclude that Google Street View and CANVAS offer opportunities to develop greater comparability across neighborhood audit studies.