A picture tells a thousand…exposures: Opportunities and challenges of deep learning image analyses in exposure science and environmental epidemiology

Population Neuroscience: Principles and Advances

In population neuroscience, three disciplines come together to advance our knowledge of factors that shape the human brain: neuroscience, genetics, and epidemiology (Paus, Human Brain Mapping 31:891-903, 2010). Here, I will come back to some of the background material reviewed in more detail in our previous book (Paus, Population Neuroscience, 2013), followed by a brief overview of current advances and challenges faced by this integrative approach.

Urban Air-Quality Estimation Using Visual Cues and a Deep Convolutional Neural Network in Bengaluru (Bangalore), India

Mobile monitoring provides robust measurements of air pollution. However, resource constraints often limit the number of measurements so that assessments cannot be obtained in all locations of interest. In response, surrogate measurement methodologies, such as videos and images, have been suggested. Previous studies of air pollution and images have used static images (e.g., satellite images or Google Street View images). The current study was designed to develop deep learning methodologies to infer on-road pollutant concentrations from videos acquired with dashboard cameras. Fifty hours of on-road measurements of four pollutants (black carbon, particle number concentration, PM2.5 mass concentration, carbon dioxide) in Bengaluru, India, were analyzed. The analysis of each video frame involved identifying objects and determining motion (by segmentation and optical flow). Based on these visual cues, a regression convolutional neural network (CNN) was used to deduce pollution concentrations. The findings showed that the CNN approach outperformed several other machine learning (ML) techniques and more conventional analyses (e.g., linear regression). The CO2 prediction model achieved a normalized root-mean-square error of 10-13.7% for the different train-validation division methods. The results here thus contribute to the literature by using video and the relative motion of on-screen objects rather than static images and by implementing a rapid-analysis approach enabling analysis of the video in real time. These methods can be applied to other mobile-monitoring campaigns since the only additional equipment they require is an inexpensive dashboard camera.

Beyond here and now: Evaluating pollution estimation across space and time from street view images with deep learning

Advances in computer vision, driven by deep learning, allows for the inference of environmental pollution and its potential sources from images. The spatial and temporal generalisability of image-based pollution models is crucial in their real-world application, but is currently understudied, particularly in low-income countries where infrastructure for measuring the complex patterns of pollution is limited and modelling may therefore provide the most utility. We employed convolutional neural networks (CNNs) for two complementary classification models, in both an end-to-end approach and as an interpretable feature extractor (object detection), to estimate spatially and temporally resolved fine particulate matter (PM2.5) and noise levels in Accra, Ghana. Data used for training the models were from a unique dataset of over 1.6 million images collected over 15 months at 145 representative locations across the city, paired with air and noise measurements. Both end-to-end CNN and object-based approaches surpassed null model benchmarks for predicting PM2.5 and noise at single locations, but performance deteriorated when applied to other locations. Model accuracy diminished when tested on images from locations unseen during training, but improved by sampling a greater number of locations during model training, even if the total quantity of data was reduced. The end-to-end models used characteristics of images associated with atmospheric visibility for predicting PM2.5, and specific objects such as vehicles and people for noise. The results demonstrate the potential and challenges of image-based, spatiotemporal air pollution and noise estimation, and that robust, environmental modelling with images requires integration with traditional sensor networks.

The leading global health challenges in the artificial intelligence era

Millions of people's health is at risk because of several factors and multiple overlapping crises, all of which hit the vulnerable the most. These challenges are dynamic and evolve in response to emerging health challenges and concerns, which need effective collaboration among countries working toward achieving Sustainable Development Goals (SDGs) and securing global health. Mental Health, the Impact of climate change, cardiovascular diseases (CVDs), diabetes, Infectious diseases, health system, and population aging are examples of challenges known to pose a vast burden worldwide. We are at a point known as the "digital revolution," characterized by the expansion of artificial intelligence (AI) and a fusion of technology types. AI has emerged as a powerful tool for addressing various health challenges, and the last ten years have been influential due to the rapid expansion in the production and accessibility of health-related data. The computational models and algorithms can understand complicated health and medical data to perform various functions and deep-learning strategies. This narrative mini-review summarizes the most current AI applications to address the leading global health challenges. Harnessing its capabilities can ultimately mitigate the Impact of these challenges and revolutionize the field. It has the ability to strengthen global health through personalized health care and improved preparedness and response to future challenges. However, ethical and legal concerns about individual or community privacy and autonomy must be addressed for effective implementation.

Predicting spatial variations in annual average outdoor ultrafine particle concentrations in Montreal and Toronto, Canada: Integrating land use regression and deep learning models

BACKGROUND

Concentrations of outdoor ultrafine particles (UFP; <0.1 µm) and black carbon (BC) can vary greatly within cities and long-term exposures to these pollutants have been associated with a variety of adverse health outcomes.

OBJECTIVE

This study integrated multiple approaches to develop new models to estimate within-city spatial variations in annual median (i.e. average) outdoor UFP and BC concentrations as well as mean UFP size in Canada's two largest cities, Montreal and Toronto.

METHODS

We conducted year-long mobile monitoring campaigns in each city that included evenings and weekends. We developed generalized additive models trained on land use parameters and deep Convolutional Neural Network (CNN) models trained on satellite-view images. Using predictions from these models, we developed final combined models.

RESULTS

In Toronto, the median observed UFP concentration, UFP size, and BC concentration values were 16,172pt/cm3, 33.7 nm, and 1225 ng/m3, respectively. In Montreal, the median observed UFP concentration, UFP size, and BC concentration values were 14,702pt/cm3, 29.7 nm, and 1060 ng/m3, respectively. For all pollutants in both cities, the proportion of spatial variation explained (i.e., R2) was slightly greater (1-2 percentage points) for the combined models than the generalized additive models and a greater (approximately 10 percentage points) than the deep CNN models. The Toronto combined model R2 values in the test set were 0.73, 0.55, and 0.61 for UFP concentrations, UFP size, and BC concentration, respectively. The Montreal combined model R2 values were 0.60, 0.49, and 0.60 for UFP concentration, UFP size, and BC concentration models respectively. For each pollutant, predictions from the combined, deep CNN, and generalized additive models were highly correlated with each other and differences between models were explored in sensitivity analyses.

CONCLUSION

Predictions from these models are available to support future epidemiological research examining long-term health impacts of outdoor UFPs and BC.

Environmental noise and health in low-middle-income-countries: A systematic review of epidemiological evidence

Evidence of the health impacts from environmental noise has largely been drawn from studies in high-income countries, which has then been used to inform development of noise guidelines. It is unclear whether findings in high-income countries can be readily translated into policy contexts in low-middle-income-countries (LMICs). We conducted this systematic review to summarise noise epidemiological studies in LMICs. We conducted a literature search of studies in Medline and Web of Science published during 2009-2021, supplemented with specialist journal hand searches. Screening, data extraction, assessment of risk of bias as well as overall quality and strength of evidence were conducted following established guidelines (e.g. Navigation Guide). 58 studies were identified, 53% of which were from India, China and Bulgaria. Most (92%) were cross-sectional studies. 53% of studies assessed noise exposure based on fixed-site measurements using sound level meters and 17% from propagation-based noise models. Mean noise exposure among all studies ranged from 48 to 120 dB (L_eq), with over half of the studies (52%) reporting the mean between 60 and 80 dB. The most studied health outcome was noise annoyance (43% of studies), followed by cardiovascular (17%) and mental health outcomes (17%). Studies generally reported a positive (i.e. adverse) relationship between noise exposure and annoyance. Some limited evidence based on only two studies showing that long-term noise exposure may be associated with higher prevalence of cardiovascular outcomes in adults. Findings on mental health outcomes were inconsistent across the studies. Overall, 4 studies (6%) had "probably low", 18 (31%) had "probably high" and 36 (62%) had "high" risk of bias. Quality of evidence was rated as 'low' for mental health outcomes and 'very low' for all other outcomes. Strength of evidence for each outcome was assessed as 'inadequate', highlighting high-quality epidemiological studies are urgently needed in LMICs to strengthen the evidence base.

Characterisation of urban environment and activity across space and time using street images and deep learning in Accra

The urban environment influences human health, safety and wellbeing. Cities in Africa are growing faster than other regions but have limited data to guide urban planning and policies. Our aim was to use smart sensing and analytics to characterise the spatial patterns and temporal dynamics of features of the urban environment relevant for health, liveability, safety and sustainability. We collected a novel dataset of 2.1 million time-lapsed day and night images at 145 representative locations throughout the Metropolis of Accra, Ghana. We manually labelled a subset of 1,250 images for 20 contextually relevant objects and used transfer learning with data augmentation to retrain a convolutional neural network to detect them in the remaining images. We identified 23.5 million instances of these objects including 9.66 million instances of persons (41% of all objects), followed by cars (4.19 million, 18%), umbrellas (3.00 million, 13%), and informally operated minibuses known as tro tros (2.94 million, 13%). People, large vehicles and market-related objects were most common in the commercial core and densely populated informal neighbourhoods, while refuse and animals were most observed in the peripheries. The daily variability of objects was smallest in densely populated settlements and largest in the commercial centre. Our novel data and methodology shows that smart sensing and analytics can inform planning and policy decisions for making cities more liveable, equitable, sustainable and healthy.

Digital health for climate change mitigation and response: a scoping review

OBJECTIVE

Climate change poses a major threat to the operation of global health systems, triggering large scale health events, and disrupting normal system operation. Digital health may have a role in the management of such challenges and in greenhouse gas emission reduction. This scoping review explores recent work on digital health responses and mitigation approaches to climate change.

MATERIALS AND METHODS

We searched Medline up to February 11, 2022, using terms for digital health and climate change. Included articles were categorized into 3 application domains (mitigation, infectious disease, or environmental health risk management), and 6 technical tasks (data sensing, monitoring, electronic data capture, modeling, decision support, and communication). The review was PRISMA-ScR compliant.

RESULTS

The 142 included publications reported a wide variety of research designs. Publication numbers have grown substantially in recent years, but few come from low- and middle-income countries. Digital health has the potential to reduce health system greenhouse gas emissions, for example by shifting to virtual services. It can assist in managing changing patterns of infectious diseases as well as environmental health events by timely detection, reducing exposure to risk factors, and facilitating the delivery of care to under-resourced areas.

DISCUSSION

While digital health has real potential to help in managing climate change, research remains preliminary with little real-world evaluation.

CONCLUSION

Significant acceleration in the quality and quantity of digital health climate change research is urgently needed, given the enormity of the global challenge.

Predicting walking-to-work using street-level imagery and deep learning in seven Canadian cities

New 'big data' streams such as street-level imagery are offering unprecedented possibilities for developing health-relevant data on the urban environment. Urban environmental features derived from street-level imagery have been used to assess pedestrian-friendly neighbourhood design and to predict active commuting, but few such studies have been conducted in Canada. Using 1.15 million Google Street View (GSV) images in seven Canadian cities, we applied image segmentation and object detection computer vision methods to extract data on persons, bicycles, buildings, sidewalks, open sky (without trees or buildings), and vegetation at postal codes. The associations between urban features and walk-to-work rates obtained from the Canadian Census were assessed. We also assessed how GSV-derived urban features perform in predicting walk-to-work rates relative to more widely used walkability measures. Results showed that features derived from street-level images are better able to predict the percent of people walking to work as their primary mode of transportation compared to data derived from traditional walkability metrics. Given the increasing coverage of street-level imagery around the world, there is considerable potential for machine learning and computer vision to help researchers study patterns of active transportation and other health-related behaviours and exposures.

Street-view greenspace exposure and objective sleep characteristics among children

Greenspace may benefit sleep by enhancing physical activity, reducing stress or air pollution exposure. Studies on greenspace and children's sleep are limited, and most use satellite-derived measures that do not capture ground-level exposures that may be important for sleep. We examined associations of street view imagery (SVI)-based greenspace with sleep in Project Viva, a Massachusetts pre-birth cohort. We used deep learning algorithms to derive novel metrics of greenspace (e.g., %trees, %grass) from SVI within 250m of participant residential addresses during 2007-2010 (mid-childhood, mean age 7.9 years) and 2012-2016 (early adolescence, 13.2y) (N = 533). In early adolescence, participants completed >5 days of wrist actigraphy. Sleep duration, efficiency, and time awake after sleep onset (WASO) were derived from actigraph data. We used linear regression to examine cross-sectional and prospective associations of mid-childhood and early adolescence greenspace exposure with early adolescence sleep, adjusting for confounders. We compared associations with satellite-based greenspace (Normalized Difference Vegetation Index, NDVI). In unadjusted models, mid-childhood SVI-based total greenspace and %trees (per interquartile range) were associated with longer sleep duration at early adolescence (9.4 min/day; 95%CI:3.2,15.7; 8.1; 95%CI:1.7,14.6 respectively). However, in fully adjusted models, only the association between %grass at mid-childhood and WASO was observed (4.1; 95%CI:0.2,7.9). No associations were observed between greenspace and sleep efficiency, nor in cross-sectional early adolescence models. The association between greenspace and sleep differed by racial and socioeconomic subgroups. For example, among Black participants, higher NDVI was associated with better sleep, in neighborhoods with low socio-economic status (SES), higher %grass was associated with worse sleep, and in neighborhoods with high SES, higher total greenspace and %grass were associated with better sleep time. SVI metrics may have the potential to identify specific features of greenspace that affect sleep.

Measuring and modelling perceptions of the built environment for epidemiological research using crowd-sourcing and image-based deep learning models

BACKGROUND

Perceptions of the built environment, such as nature quality, beauty, relaxation, and safety, may be key factors linking the built environment to human health. However, few studies have examined these types of perceptions due to the difficulty in quantifying them objectively in large populations.

OBJECTIVE

To measure and predict perceptions of the built environment from street-view images using crowd-sourced methods and deep learning models for application in epidemiologic studies.

METHODS

We used the Amazon Mechanical-Turk crowdsourcing platform where participants compared two street-view images and quantified perceptions of nature quality, beauty, relaxation, and safety. We optimized street-view image sampling methods to improve the quality and resulting perception data specific to participants enrolled in the Washington State Twin Registry (WSTR) health study. We used a transfer learning approach to train deep learning models by leveraging existing image perception data from the PlacePulse 2.0 dataset, which includes 1.1 million image comparisons, and refining based on new WSTR perception data. Resulting models were applied to WSTR addresses to estimate exposures and evaluate associations with traditional built environment measures.

RESULTS

We collected over 36,000 image comparisons and calculated perception measures for each image. Our final deep learning models explained 77.6% of nature quality, 68.1% of beauty, 72.0% of relaxation, and 64.7% of safety in pairwise image comparisons. Applying transfer learning with the new perception labels specific to the WSTR yielded an average improvement of 3.8% for model performance. Perception measures were weakly to moderately correlated with traditional built environment exposures for WSTR participant addresses; for example, nature quality and NDVI (r = 0.55), neighborhood area deprivation (r = -0.16), and walkability (r = -0.20), respectively.

SIGNIFICANCE

We were able to measure and model perceptions of the built environment optimized for a specific health study. Future applications will examine associations between these exposure measures and mental health in the WSTR.

IMPACT STATEMENT

Built environments influence health through complex pathways. Perceptions of nature quality, beauty, relaxation and safety may be particularly import for understanding these linkages, but few studies to-date have examined these perceptions objectively for large populations. For quantitative research, an exposure measure must be reproducible, accurate, and precise--here we work to develop such measures for perceptions of the urban environment. We created crowd-sourced and image-based deep learning methods that were able to measure and model these perceptions. Future applications will apply these models to examine associations with mental health in the Washington State Twin Registry.

National Land Use Regression Model for NO2 Using Street View Imagery and Satellite Observations

Land use regression (LUR) models are widely applied to estimate intra-urban air pollution concentrations. National-scale LURs typically employ predictors from multiple curated geodatabases at neighborhood scales. In this study, we instead developed national NO₂ models relying on innovative street-level predictors extracted from Google Street View [GSV] imagery. Using machine learning (random forest), we developed two types of models: (1) GSV-only models, which use only GSV features, and (2) GSV + OMI models, which also include satellite observations of NO₂. Our results suggest that street view imagery alone may provide sufficient information to explain NO₂ variation. Satellite observations can improve model performance, but the contribution decreases as more images are available. Random 10-fold cross-validation R² of our best models were 0.88 (GSV-only) and 0.91 (GSV + OMI)─a performance that is comparable to traditional LUR approaches. Importantly, our models show that street-level features might have the potential to better capture intra-urban variation of NO₂ pollution than traditional LUR. Collectively, our findings indicate that street view image-based modeling has great potential for building large-scale air quality models under a unified framework. Toward that goal, we describe a cost-effective image sampling strategy for future studies based on a systematic evaluation of image availability and model performance.

Street View Imagery (SVI) in the Built Environment: A Theoretical and Systematic Review

Street view imagery (SVI) provides efficient access to data that can be used to research spatial quality at the human scale. The previous reviews have mainly focused on specific health findings and neighbourhood environments. There has not been a comprehensive review of this topic. In this paper, we systematically review the literature on the application of SVI in the built environment, following a formal innovation–decision framework. The main findings are as follows: (I) SVI remains an effective tool for automated research assessments. This offers a new research avenue to expand the built environment-measurement methods to include perceptions in addition to physical features. (II) Currently, SVI is functional and valuable for quantifying the built environment, spatial sentiment perception, and spatial semantic speculation. (III) The significant dilemmas concerning the adoption of this technology are related to image acquisition, the image quality, spatial and temporal distribution, and accuracy. (IV) This research provides a rapid assessment and provides researchers with guidance for the adoption and implementation of SVI. Data integration and management, proper image service provider selection, and spatial metrics measurements are the critical success factors. A notable trend is the application of SVI towards a focus on the perceptions of the built environment, which provides a more refined and effective way to depict urban forms in terms of physical and social spaces.

Integrating Multiscale Geospatial Environmental Data into Large Population Health Studies: Challenges and Opportunities

Quantifying the exposome is key to understanding how the environment impacts human health and disease. However, accurately, and cost-effectively quantifying exposure in large population health studies remains a major challenge. Geospatial technologies offer one mechanism to integrate high-dimensional environmental data into epidemiology studies, but can present several challenges. In June 2021, the National Institute of Environmental Health Sciences (NIEHS) held a workshop bringing together experts in exposure science, geospatial technologies, data science and population health to address the need for integrating multiscale geospatial environmental data into large population health studies. The primary objectives of the workshop were to highlight recent applications of geospatial technologies to examine the relationships between environmental exposures and health outcomes; identify research gaps and discuss future directions for exposure modeling, data integration and data analysis strategies; and facilitate communications and collaborations across geospatial and population health experts. This commentary provides a high-level overview of the scientific topics covered by the workshop and themes that emerged as areas for future work, including reducing measurement errors and uncertainty in exposure estimates, and improving data accessibility, data interoperability, and computational approaches for more effective multiscale and multi-source data integration, along with potential solutions.

What you see is what you breathe? Estimating air pollution spatial variation using street level imagery

High spatial resolution information on urban air pollution levels is unavailable in many areas globally, partially due to high input data needs of existing estimation approaches. Here we introduce a computer vision method to estimate annual means for air pollution levels from street level images. We used annual mean estimates of NO2 and PM2.5 concentrations from locally calibrated models as labels from London, New York, and Vancouver to allow for compilation of a sufficiently large dataset (~250k images for each city). Our experimental setup is designed to quantify intra and intercity transferability of image-based model estimates. Performances were high and comparable to traditional land-use regression (LUR) and dispersion models when training and testing on images from the same city (R2 values between 0.51 and 0.95 when validated on data from ground monitoring stations). Like LUR models, transferability of models between cities in different geographies is more difficult. Specifically, transferability between the three cities i.e., London, New York, and Vancouver, which have similar pollution source profiles were moderately successful (R2 values between zero and 0.67). Comparatively, performances when transferring models trained on these cities with very different source profiles i.e., Accra in Ghana and Hong Kong were lower (R2 between zero and 0.21) suggesting the need for local calibration with local calibration using additional measurement data from cities that share similar source profiles.

Quantifying Ecological Landscape Quality of Urban Street by Open Street View Images: A Case Study of Xiamen Island, China

With the unprecedented urbanization processes around the world, cities have become the main areas of political, cultural, and economic creation, but these regions have also caused environmental degradation and even affected public health. Ecological landscape is considered as an important way to mitigate the impact of environmental exposure on urban residents. Therefore, quantifying the quality of urban road landscape and exploring its spatial heterogeneity to obtain basic data on the urban environment and provide ideas for urban residents to improve the environment will be a meaningful preparation for further urban planning. In this study, we proposed a framework to achieve automatic quantifying urban street quality by integrating a mass of street view images based on deep learning and landscape ecology. We conducted a case study in Xiamen Island and mapped a series of spatial distribution for ecological indicators including PLAND, LPI, AI, DIVISION, FRAC_MN, LSI and SHDI. Additionally, we quantified street quality by the entropy weight method. Our results showed the streetscape quality of the roundabout in Xiamen was relatively lower, while the central urban area presented a belt-shaped area with excellent landscape quality. We suggested that managers could build vertical greening on some streets around the Xiamen Island to improve the street quality in order to provide greater well-being for urban residents. In this study, it was found that there were still large uncertainties in the mechanism of environmental impact on human beings. We proposed to strengthen the in-depth understanding of the mechanism of environmental impact on human beings in the process of interaction between environment and human beings, and continue to form general models to enhance the ability of insight into the urban ecosystem.

Urban greenspace and mental health in Chinese older adults: Associations across different greenspace measures and mediating effects of environmental perceptions

This study aimed to contrast the associations of street view-, land use- and satellite-derived greenspace measures with older adults' mental health and to examine the mediating effects of neighborhood environmental perceptions (i.e., noise, aesthetics and satisfaction with recreational opportunities) to explain potential heterogeneity in the associations. Data of 879 respondents aged 60 or older in Dalian, China were used, and multilevel regression models were conducted in Stata. Results indicated that the Normalized Difference Vegetation Index (NDVI), vegetation coverage, park coverage and streetscape grasses were positively correlated with older adults' mental health. The associations of exposure metrics measured by overhead view were stronger than those measured by the street view. Streetscape grasses had a stronger association with older adults' mental health than streetscape trees. Noise, aesthetics and satisfaction with recreational opportunities mediated these associations, but the strength of the mediating effects differed across the greenspace measures. Our findings confirm the necessity of multi-measures assessment for greenspace to examine associations with older adults' mental health in Chinese settings and can contribute to the realization of health benefits of urban greenspace.

Characterization of vapor intrusion sites with a deep learning-based data assimilation method

Appropriate characterization of site soils is essential for accurate risk assessment of soil vapor intrusion (VI). In this study, we develop a data assimilation method based on deep learning (i.e., ES_(DL)) to estimate the distribution of soil properties with limited measurements. Two hypothetical VI scenarios are employed to demonstrate site characterization using the ES_(DL) method, followed by validation with a laboratory sandbox experiment and then one practical site application. The results show that the ES_(DL) method can provide reasonable estimates of the effective diffusion coefficient distributions and corresponding emission rates (into the building) in all four cases. The spatial heterogeneity of site soils can be characterized by considerably enough measurements (i.e., 15 sampling points in the first hypothetical case); otherwise, layered characterization is recommended at the cost of neglecting horizontal heterogeneity of site soils. This new method provides an alternative to characterize VI sites with relatively fewer sampling efforts.

Mapping Inequalities in the Physical, Built and Social Environment in Population-Based Studies of Brain Health

This mini-tutorial describes how combining aggregate-level data about the physical, built and social environment can facilitate our understanding of factors shaping the human brain and, in turn, brain health. It provides entry-level information about methods and approaches one can use to uncover how inequalities in the local environment lead to health inequalities in general, and those in brain health in particular. This background knowledge should be helpful to those who are interested in using neuroimaging to investigate how environmental factors shape inter-individual variations in the human brain.

Abnormal Respiratory Sounds Classification Using Deep CNN Through Artificial Noise Addition

Respiratory sound (RS) attributes and their analyses structure a fundamental piece of pneumonic pathology, and it gives symptomatic data regarding a patient's lung. A couple of decades back, doctors depended on their hearing to distinguish symptomatic signs in lung audios by utilizing the typical stethoscope, which is usually considered a cheap and secure method for examining the patients. Lung disease is the third most ordinary cause of death worldwide, so; it is essential to classify the RS abnormality accurately to overcome the death rate. In this research, we have applied Fourier analysis for the visual inspection of abnormal respiratory sounds. Spectrum analysis was done through Artificial Noise Addition (ANA) in conjunction with different deep convolutional neural networks (CNN) to classify the seven abnormal respiratory sounds—both continuous (CAS) and discontinuous (DAS). The proposed framework contains an adaptive mechanism of adding a similar type of noise to unhealthy respiratory sounds. ANA makes sound features enough reach to be identified more accurately than the respiratory sounds without ANA. The obtained results using the proposed framework are superior to previous techniques since we simultaneously considered the seven different abnormal respiratory sound classes.

Predicting Perceptions of the Built Environment using GIS, Satellite and Street View Image Approaches

BACKGROUND

High quality built environments are important for human health and wellbeing. Numerous studies have characterized built environment physical features and environmental exposures, but few have examined urban perceptions at geographic scales needed for population-based research. The degree to which urban perceptions are associated with different environmental features, and traditional environmental exposures such as air pollution or urban green space, is largely unknown.

OBJECTIVE

To determine built environment factors associated with safety, lively and beauty perceptions across 56 cities.

METHODS

We examined perceptions collected in the open source Place Pulse 2.0 dataset, which assigned safety, lively and beauty scores to street view images based on crowd-sourced labelling. We derived built environment measures for the locations of these images (110,000 locations across 56 global cities) using GIS and remote sensing datasets as well as street view imagery features (e.g. trees, cars) using deep learning image segmentation. Linear regression models were developed using Lasso penalized variable selection to predict perceptions based on visible (street level images) and GIS/remote sensing built environment variables.

RESULTS

Population density, impervious surface area, major roads, traffic air pollution, tree cover and Normalized Difference Vegetation Index (NDVI) showed statistically significant differences between high and low safety, lively, and beauty perception locations. Visible street level features explained approximately 18% of the variation in safety, lively, and beauty perceptions, compared to 3-10% explained by GIS/remote sensing. Large differences in prediction were seen when modelling between city (R2 67-81%) versus within city (R2 11-13%) perceptions. Important predictor variables included visible accessibility features (e.g. streetlights, benches) and roads for safety, visible plants and buildings for lively, and visible green space and NDVI for beauty.

CONCLUSION

Substantial within and between city differences in built environment perceptions exist, which visible street level features and GIS/remote sensing variables only partly explain. This offers a new research avenue to expand built environment measurement methods to include perceptions in addition to physical features.

National Empirical Models of Air Pollution Using Microscale Measures of the Urban Environment

National-scale empirical models of air pollution (e.g., Land Use Regression) rely on predictor variables (e.g., population density, land cover) at different geographic scales. These models typically lack microscale variables (e.g., street level), which may improve prediction with fine-spatial gradients. We developed microscale variables of the urban environment including Point of Interest (POI) data, Google Street View (GSV) imagery, and satellite-based measures of urban form. We developed United States national models for six criteria pollutants (NO₂, PM_2.5, O₃, CO, PM₁₀, SO₂) using various modeling approaches: Stepwise Regression + kriging (SW-K), Partial Least Squares + kriging (PLS-K), and Machine Learning + kriging (ML-K). We compared predictor variables (e.g., traditional vs microscale) and emerging modeling approaches (ML-K) to well-established approaches (i.e., traditional variables in a PLS-K or SW-K framework). We found that combined predictor variables (traditional + microscale) in the ML-K models outperformed the well-established approaches (10-fold spatial cross-validation (CV) R² increased 0.02-0.42 [average: 0.19] among six criteria pollutants). Comparing all model types using microscale variables to models with traditional variables, the performance is similar (average difference of 10-fold spatial CV R² = 0.05) suggesting microscale variables are a suitable substitute for traditional variables. ML-K and microscale variables show promise for improving national empirical models.

Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study

What is the relationship between mortality and satellite images as elucidated through the use of Convolutional Neural Networks? Background: Following a century of increase, life expectancy in the United States has stagnated and begun to decline in recent decades. Using satellite images and street view images, prior work has demonstrated associations of the built environment with income, education, access to care, and health factors such as obesity. However, assessment of learned image feature relationships with variation in crude mortality rate across the United States has been lacking. Objective: We sought to investigate if county-level mortality rates in the U.S. could be predicted from satellite images. Methods: Satellite images of neighborhoods surrounding schools were extracted with the Google Static Maps application programming interface for 430 counties representing ~68.9% of the US population. A convolutional neural network was trained using crude mortality rates for each county in 2015 to predict mortality. Learned image features were interpreted using Shapley Additive Feature Explanations, clustered, and compared to mortality and its associated covariate predictors. Results: Predicted mortality from satellite images in a held-out test set of counties was strongly correlated to the true crude mortality rate (Pearson r = 0.72). Direct prediction of mortality using a deep learning model across a cross-section of 430 U.S. counties identified key features in the environment (e.g., sidewalks, driveways, and hiking trails) associated with lower mortality. Learned image features were clustered, and we identified 10 clusters that were associated with education, income, geographical region, race, and age. Conclusions: The application of deep learning techniques to remotely-sensed features of the built environment can serve as a useful predictor of mortality in the United States. Although we identified features that were largely associated with demographic information, future modeling approaches that directly identify image features associated with health-related outcomes have the potential to inform targeted public health interventions.

Predicting Within-City Spatial Variations in Outdoor Ultrafine Particle and Black Carbon Concentrations in Bucaramanga, Colombia: A Hybrid Approach Using Open-Source Geographic Data and Digital Images

Outdoor ultrafine particles (UFP, <0.1 μm) and black carbon (BC) vary greatly within cities and may have adverse impacts on human health. In this study, we used a hybrid approach to develop new models to estimate within-city spatial variations in outdoor UFP and BC concentrations across Bucaramanga, Colombia. We conducted a mobile monitoring campaign over 20 days in 2019. Regression models were trained on land use data and combined with predictions from convolutional neural networks (CNN) trained to predict UFP and BC concentrations using satellite and street-level images. The combined UFP model (R² = 0.54) outperformed the CNN (R² = 0.47) and land use regression (LUR) models (R² = 0.47) on their own. Similarly, the combined BC model also outperformed the CNN and LUR BC models (R² = 0.51 vs 0.43 and 0.45, respectively). Spatial variations in model performance were more stable for the CNN and combined models compared to the LUR models, suggesting that the combined approach may be less likely to contribute to differential exposure measurement error in epidemiological studies. In general, our findings demonstrated that satellite and street-level images can be combined with a traditional LUR modeling approach to improve predictions of within-city spatial variations in outdoor UFP and BC concentrations.

Using machine learning to examine street green space types at a high spatial resolution: Application in Los Angeles County on socioeconomic disparities in exposure

BACKGROUND

Compared to commonly-used green space indicators from downward-facing satellite imagery, street view-based green space may capture different types of green space and represent how environments are perceived and experienced by people on the ground, which is important to elucidate the underlying mechanisms linking green space and health.

OBJECTIVES

This study aimed to evaluate machine learning models that can classify the type of vegetation (i.e., tree, low-lying vegetation, grass) from street view images; and to investigate the associations between street green space and socioeconomic (SES) factors, in Los Angeles County, California.

METHODS

SES variables were obtained from the CalEnviroScreen3.0 dataset. Microsoft Bing Maps images in conjunction with deep learning were used to measure total and types of street view green space, which were compared to normalized difference vegetation index (NDVI) as commonly-used satellite-based green space measure. Generalized linear mixed model was used to examine associations between green space and census tract SES, adjusting for population density and rural/urban status.

RESULTS

The accuracy of the deep learning model was high with 92.5% mean intersection over union. NDVI were moderately correlated with total street view-based green space and tree, and weakly correlated with low-lying vegetation and grass. Total and three types of green space showed significant negative associations with neighborhood SES. The percentage of total green space decreased by 2.62 [95% confidence interval (CI): -3.02, -2.21, p < 0.001] with each interquartile range increase in CalEnviroScreen3.0 score. Disadvantaged communities contained approximately 5% less average street green space than other communities.

CONCLUSION

Street view imagery coupled with deep learning approach can accurately and efficiently measure eye-level street green space and distinguish vegetation types. In Los Angeles County, disadvantaged communities had substantively less street green space. Governments and urban planners need to consider the type and visibility of street green space from pedestrian's perspective.

Using machine learning to examine street green space types at a high spatial resolution: Application in Los Angeles County on socioeconomic disparities in exposure

BACKGROUND

Compared to commonly-used green space indicators from downward-facing satellite imagery, street view-based green space may capture different types of green space and represent how environments are perceived and experienced by people on the ground, which is important to elucidate the underlying mechanisms linking green space and health.

OBJECTIVES

This study aimed to evaluate machine learning models that can classify the type of vegetation (i.e., tree, low-lying vegetation, grass) from street view images; and to investigate the associations between street green space and socioeconomic (SES) factors, in Los Angeles County, California.

METHODS

SES variables were obtained from the CalEnviroScreen3.0 dataset. Microsoft Bing Maps images in conjunction with deep learning were used to measure total and types of street view green space, which were compared to normalized difference vegetation index (NDVI) as commonly-used satellite-based green space measure. Generalized linear mixed model was used to examine associations between green space and census tract SES, adjusting for population density and rural/urban status.

RESULTS

The accuracy of the deep learning model was high with 92.5% mean intersection over union. NDVI were moderately correlated with total street view-based green space and tree, and weakly correlated with low-lying vegetation and grass. Total and three types of green space showed significant negative associations with neighborhood SES. The percentage of total green space decreased by 2.62 [95% confidence interval (CI): -3.02, -2.21, p < 0.001] with each interquartile range increase in CalEnviroScreen3.0 score. Disadvantaged communities contained approximately 5% less average street green space than other communities.

CONCLUSION

Street view imagery coupled with deep learning approach can accurately and efficiently measure eye-level street green space and distinguish vegetation types. In Los Angeles County, disadvantaged communities had substantively less street green space. Governments and urban planners need to consider the type and visibility of street green space from pedestrian's perspective.

Backyard benefits? A cross-sectional study of yard size and greenness and children's physical activity and outdoor play

BACKGROUND

The home environment is the most important location in young children's lives, yet few studies have examined the relationship between the outdoor home environment and child physical activity levels, and even fewer have used objectively measured exposures and outcomes. This study examined relationships between objectively assessed home yard size and greenness, and child physical activity and outdoor play.

METHODS

Data were drawn from the HealthNuts study, a longitudinal study of 5276 children in Melbourne, Australia. We used cross-sectional data from a sample at Wave 3 (2013-2016) when participants were aged 6 years (n = 1648). A sub-sample of 391 children had valid accelerometer data collected from Tri-axial GENEActive accelerometers worn on their non-dominant wrist for 8 consecutive days. Yard area and greenness were calculated using geographic information systems. Objective outcome measures were minutes/day in sedentary, light, and moderate-vigorous physical activity (weekday and weekend separately). Parent-reported outcome measures were minutes/day playing outdoors (weekend and weekday combined). Multi-level regression models (adjusted for child's sex, mother's age at the birth of child, neighbourhood socioeconomic index, maternal education, and maternal ethnicity) estimated effects of yard size and greenness on physical activity.

RESULTS

Data were available on outdoor play for 1648 children and usable accelerometer data for 391. Associations between yard size/greenness and components of physical activity were minimal. For example, during weekdays, yard size was not associated with daily minutes in sedentary behaviour (β: 2.4, 95% CI: - 6.2, 11.0), light physical activity (β: 1.4, 95% CI: - 5.7, 8.5) or MVPA (β: -2.4, 95% CI: - 6.5, 1.7), with similar patterns at weekends. There was no relationship between median annual yard greenness and physical activity or play.

CONCLUSION

In our study of young children residing in higher socio-economic areas of Melbourne yard characteristics did not appear to have a major impact on children's physical activity. Larger studies with greater variation in yard characteristics and identification of activity location are needed to better understand the importance of home outdoor spaces and guide sustainable city planning.

Statistical Emulation of Winter Ambient Fine Particulate Matter Concentrations From Emission Changes in China

Air pollution exposure remains a leading public health problem in China. The use of chemical transport models to quantify the impacts of various emission changes on air quality is limited by their large computational demands. Machine learning models can emulate chemical transport models to provide computationally efficient predictions of outputs based on statistical associations with inputs. We developed novel emulators relating emission changes in five key anthropogenic sectors (residential, industry, land transport, agriculture, and power generation) to winter ambient fine particulate matter (PM_2.5) concentrations across China. The emulators were optimized based on Gaussian process regressors with Matern kernels. The emulators predicted 99.9% of the variance in PM_2.5 concentrations for a given input configuration of emission changes. PM_2.5 concentrations are primarily sensitive to residential (51%-94% of first-order sensitivity index), industrial (7%-31%), and agricultural emissions (0%-24%). Sensitivities of PM_2.5 concentrations to land transport and power generation emissions are all under 5%, except in South West China where land transport emissions contributed 13%. The largest reduction in winter PM_2.5 exposure for changes in the five emission sectors is by 68%-81%, down to 15.3-25.9 μg m^-3, remaining above the World Health Organization annual guideline of 10 μg m^-3. The greatest reductions in PM_2.5 exposure are driven by reducing residential and industrial emissions, emphasizing the importance of emission reductions in these key sectors. We show that the annual National Air Quality Target of 35 μg m^-3 is unlikely to be achieved during winter without strong emission reductions from the residential and industrial sectors.

Statistical Emulation of Winter Ambient Fine Particulate Matter Concentrations From Emission Changes in China

Air pollution exposure remains a leading public health problem in China. The use of chemical transport models to quantify the impacts of various emission changes on air quality is limited by their large computational demands. Machine learning models can emulate chemical transport models to provide computationally efficient predictions of outputs based on statistical associations with inputs. We developed novel emulators relating emission changes in five key anthropogenic sectors (residential, industry, land transport, agriculture, and power generation) to winter ambient fine particulate matter (PM2.5) concentrations across China. The emulators were optimized based on Gaussian process regressors with Matern kernels. The emulators predicted 99.9% of the variance in PM2.5 concentrations for a given input configuration of emission changes. PM2.5 concentrations are primarily sensitive to residential (51%-94% of first-order sensitivity index), industrial (7%-31%), and agricultural emissions (0%-24%). Sensitivities of PM2.5 concentrations to land transport and power generation emissions are all under 5%, except in South West China where land transport emissions contributed 13%. The largest reduction in winter PM2.5 exposure for changes in the five emission sectors is by 68%-81%, down to 15.3-25.9 μg m-3, remaining above the World Health Organization annual guideline of 10 μg m-3. The greatest reductions in PM2.5 exposure are driven by reducing residential and industrial emissions, emphasizing the importance of emission reductions in these key sectors. We show that the annual National Air Quality Target of 35 μg m-3 is unlikely to be achieved during winter without strong emission reductions from the residential and industrial sectors.

Combining citizen science and deep learning for large-scale estimation of outdoor nitrogen dioxide concentrations

Reliable estimates of outdoor air pollution concentrations are needed to support global actions to improve public health. We developed a new approach to estimating annual average outdoor nitrogen dioxide (NO₂) concentrations using approximately 20,000 ground-level measurements in Flanders, Belgium combined with aerial images and deep neural networks. Our final model explained 79% of the spatial variability in NO₂ (root mean square error of 10-fold cross-validation = 3.58 μg/m³) using only images as model inputs. This novel approach offers an alternative means of estimating large-scale spatial variations in ambient air quality and may be particularly useful for regions of the world without detailed emissions data or land use information typically used to estimate outdoor air pollution concentrations.

Multimodal deep learning from satellite and street-level imagery for measuring income, overcrowding, and environmental deprivation in urban areas

Data collected at large scale and low cost (e.g. satellite and street level imagery) have the potential to substantially improve resolution, spatial coverage, and temporal frequency of measurement of urban inequalities. Multiple types of data from different sources are often available for a given geographic area. Yet, most studies utilize a single type of input data when making measurements due to methodological difficulties in their joint use. We propose two deep learning-based methods for jointly utilizing satellite and street level imagery for measuring urban inequalities. We use London as a case study for three selected outputs, each measured in decile classes: income, overcrowding, and environmental deprivation. We compare the performances of our proposed multimodal models to corresponding unimodal ones using mean absolute error (MAE). First, satellite tiles are appended to street level imagery to enhance predictions at locations where street images are available leading to improvements in accuracy by 20, 10, and 9% in units of decile classes for income, overcrowding, and living environment. The second approach, novel to the best of our knowledge, uses a U-Net architecture to make predictions for all grid cells in a city at high spatial resolution (e.g. for 3 m × 3 m pixels in London in our experiments). It can utilize city wide availability of satellite images as well as more sparse information from street-level images where they are available leading to improvements in accuracy by 6, 10, and 11%. We also show examples of prediction maps from both approaches to visually highlight performance differences.

Using Street View Imagery to Predict Street-Level Particulate Air Pollution

Land-use regression (LUR) models are frequently applied to estimate spatial patterns of air pollution. Traditional LUR often relies on fixed-site measurements and GIS-derived variables with limited spatial resolution. We present an approach that leverages Google Street View (GSV) imagery to predict street-level particulate air pollution (i.e., black carbon [BC] and particle number [PN] concentrations). We developed empirical models based on mobile monitoring data and features extracted from ∼52 500 GSV images using a deep learning model. We tested theory- and data-driven feature selection methods as well as models using images within varying buffer sizes (50-2000 m). Compared to LUR models with traditional variables, our models achieved similar model performance using the street-level predictors while also identifying additional potential hotspots. Adjusted R² (10-fold CV R²) with integrated feature selection was 0.57-0.64 (0.50-0.57) and 0.65-0.73 (0.61-0.66) for BC and PN models, respectively. Models using only features near the measurement locations (i.e., GSV images within 250 m) explained ∼50% of air pollution variability, indicating PN and BC are strongly affected by the street-level built environment. Our results suggest that GSV imagery, processed with computer vision techniques, is a promising data source to develop LUR models with high spatial resolution and consistent predictor variables across administrative boundaries.

Informatics Approaches for Recognition, Management, and Prevention of Occupational Respiratory Disease

Computer and information systems can improve occupational respiratory disease prevention and surveillance by providing efficient resources for patients, workers, clinicians, and public health practitioners. Advances include interlinking electronic health records, autocoding surveillance data, clinical decision support systems, and social media applications for acquiring and disseminating information. Obstacles to advances include inflexible hierarchical coding schemes, inadequate occupational health electronic health record systems, and inadequate public focus on occupational respiratory disease. Potentially transformative approaches include machine learning, natural language processing, and improved ontologies.

Centralizing environmental datasets to support (inter)national chronic disease research: Values, challenges, and recommendations

Whereas environmental data are increasingly available, it is often not clear how or if datasets are available for health research. Exposure metrics are typically developed for specific research initiatives using disparate exposure assessment methods and no mechanisms are put in place for centralizing, archiving, or distributing environmental datasets. In parallel, potentially vast amounts of environmental data are emerging due to new technologies such as high resolution imagery and machine learning.

Machine Learning Within Studies of Early-Life Environmental Exposures and Child Health: Review of the Current Literature and Discussion of Next Steps

PURPOSE OF REVIEW

The goal of this article is to review the use of machine learning (ML) within studies of environmental exposures and children's health, identify common themes across studies, and provide recommendations to advance their use in research and practice.

RECENT FINDINGS

We identified 42 articles reporting upon the use of ML within studies of environmental exposures and children's health between 2017 and 2019. The common themes among the articles were analysis of mixture data, exposure prediction, disease prediction and forecasting, analysis of complex data, and causal inference. With the increasing complexity of environmental health data, we anticipate greater use of ML to address the challenges that cannot be handled by traditional analytics. In order for these methods to beneficially impact public health, the ML techniques we use need to be appropriate for our study questions, rigorously evaluated and reported in a way that can be critically assessed by the scientific community.

Predicting outdoor ultrafine particle number concentrations, particle size, and noise using street-level images and audio data

Outdoor ultrafine particles (UFPs) (<0.1 µm) may have an important impact on public health but exposure assessment remains a challenge in epidemiological studies. We developed a novel method of estimating spatiotemporal variations in outdoor UFP number concentrations and particle diameters using street-level images and audio data in Montreal, Canada. As a secondary aim, we also developed models for noise. Convolutional neural networks were first trained to predict 10-second average UFP/noise parameters using a large database of images and audio spectrogram data paired with measurements collected between April 2019 and February 2020. Final multivariable linear regression and generalized additive models were developed to predict 5-minute average UFP/noise parameters including covariates from deep learning models based on image and audio data along with outdoor temperature and wind speed. The best performing final models had mean cross-validation R² values of 0.677 and 0.523 for UFP number concentrations and 0.825 and 0.735 for UFP size using two different test sets. Audio predictions from deep learning models were stronger predictors of spatiotemporal variations in UFP parameters than predictions based on street-level images; this was not explained only by noise levels captured in the audio signal. All final noise models had R² values above 0.90. Collectively, our findings suggest that street-level images and audio data can be used to estimate spatiotemporal variations in outdoor UFPs and noise. This approach may be useful in developing exposure models over broad spatial scales and such models can be regularly updated to expand generalizability as more measurements become available.

Healthy built environment: Spatial patterns and relationships of multiple exposures and deprivation in Toronto, Montreal and Vancouver

BACKGROUND

Various aspects of the urban environment and neighbourhood socio-economic status interact with each other to affect health. Few studies to date have quantitatively assessed intersections of multiple urban environmental factors and their distribution across levels of deprivation.

OBJECTIVES

To explore the spatial patterns of urban environmental exposures within three large Canadian cities, assess how exposures are distributed across socio-economic deprivation gradients, and identify clusters of favourable or unfavourable environmental characteristics.

METHODS

We indexed nationally standardized estimates of active living friendliness (i.e. "walkability"), NO₂ air pollution, and greenness to 6-digit postal codes within the cities of Toronto, Montreal and Vancouver. We compared the distribution of within-city exposure tertiles across quintiles of material deprivation. Tertiles of each exposure were then overlaid with each other in order to identify potentially favorable (high walkability, low NO₂, high greenness) and unfavorable (low walkability, high NO₂, and low greenness) environments.

RESULTS

In all three cities, high walkability was more common in least deprived areas and less prevalent in highly deprived areas. We also generally saw a greater prevalence of postal codes with high vegetation indices and low NO₂ in areas with low deprivation, and a lower greenness prevalence and higher NO₂ concentrations in highly deprived areas, suggesting environmental inequity is occurring. Our study showed that relatively few postal codes were simultaneously characterized by desirable or undesirable walkability, NO₂and greenness tertiles.

DISCUSSION

Spatial analyses of multiple standardized urban environmental factors such as the ones presented in this manuscript can help refine municipal investments and policy priorities. This study illustrates a methodology to prioritize areas for interventions that increase active living and exposure to urban vegetation, as well as lower air pollution. Our results also highlight the importance of considering the intersections between the built environment and socio-economic status in city planning and urban public health decision-making.

Assessment of kitchen emissions using a backpropagation neural network model based on urinary hydroxy polycyclic aromatic hydrocarbons

Kitchen emissions are mixed indoor air pollutants with adverse health effects, but the large-scale assessment is limited by costly equipment and survey methods. This study aimed to discuss the application of backpropagation (BP) neural network models in the assessment of kitchen emissions based on the exposure marker. A total of 3686 participants were recruited for the kitchen survey, and their sleep quality was measured by the Pittsburgh sleep quality index (PSQI). After excluding the confounders, 365 participants were selected to assess their urinary hydroxy polycyclic aromatic hydrocarbons (OH-PAHs) concentrations by ultra-high-performance liquid chromatography/tandem mass spectrometry. Two BP neural network models were then set up using the survey and detection data from the 365 participants and used to predict the total urinary OH-PAHs concentrations of all participants. The total urinary OH-PAHs and 1-hydroxy-naphthalene (1-OHNap) concentrations were significantly higher among the 365 participants with poor sleep quality (global PSQI score > 5; P < 0.05). Results from internal and external validation showed that our model has high credibility (model 2). Further, the participants with higher predicted total urinary OH-PAHs concentrations were associated with the global PSQI score of >5 (odds ratio (OR) = 1.284, 95% confidence interval (CI) = 1.082-1.525 for participants with predicted total urinary OH-PAHs concentrations of over 1.897 μg/mmol creatinine in model 1, and OR = 1.467, 95% CI = 1.240-1.735 for participants with predicted total urinary OH-PAHs concentrations of over 2.253 μg/mmol creatinine in model 2) after adjusting for the confounders. Findings suggest that the BP neural network model is suitable for assessing kitchen emissions, and the urinary OH-PAHs concentrations can be taken as the model outlay.

High-resolution spatiotemporal measurement of air and environmental noise pollution in Sub-Saharan African cities: Pathways to Equitable Health Cities Study protocol for Accra, Ghana

INTRODUCTION

Air and noise pollution are emerging environmental health hazards in African cities, with potentially complex spatial and temporal patterns. Limited local data are a barrier to the formulation and evaluation of policies to reduce air and noise pollution.

METHODS AND ANALYSIS

We designed a year-long measurement campaign to characterise air and noise pollution and their sources at high-resolution within the Greater Accra Metropolitan Area (GAMA), Ghana. Our design uses a combination of fixed (year-long, n=10) and rotating (week-long, n =~130) sites, selected to represent a range of land uses and source influences (eg, background, road traffic, commercial, industrial and residential areas, and various neighbourhood socioeconomic classes). We will collect data on fine particulate matter (PM_2.5), nitrogen oxides (NO_x), weather variables, sound (noise level and audio) along with street-level time-lapse images. We deploy low-cost, low-power, lightweight monitoring devices that are robust, socially unobtrusive, and able to function in Sub-Saharan African (SSA) climate. We will use state-of-the-art methods, including spatial statistics, deep/machine learning, and processed-based emissions modelling, to capture highly resolved temporal and spatial variations in pollution levels across the GAMA and to identify their potential sources. This protocol can serve as a prototype for other SSA cities.

ETHICS AND DISSEMINATION

This environmental study was deemed exempt from full ethics review at Imperial College London and the University of Massachusetts Amherst; it was approved by the University of Ghana Ethics Committee (ECH 149/18-19). This protocol is designed to be implementable in SSA cities to map environmental pollution to inform urban planning decisions to reduce health harming exposures to air and noise pollution. It will be disseminated through local stakeholder engagement (public and private sectors), peer-reviewed publications, contribution to policy documents, media, and conference presentations.

Derivation of Time-Activity Data Using Wearable Cameras and Measures of Personal Inhalation Exposure among Workers at an Informal Electronic-Waste Recovery Site in Ghana

OBJECTIVES

Approximately 2 billion workers globally are employed in informal settings, which are characterized by substantial risk from hazardous exposures and varying job tasks and schedules. Existing methods for identifying occupational hazards must be adapted for unregulated and challenging work environments. We designed and applied a method for objectively deriving time-activity patterns from wearable camera data and matched images with continuous measurements of personal inhalation exposure to size-specific particulate matter (PM) among workers at an informal electronic-waste (e-waste) recovery site.

METHODS

One hundred and forty-two workers at the Agbogbloshie e-waste site in Accra, Ghana, wore sampling backpacks equipped with wearable cameras and real-time particle monitors during a total of 171 shifts. Self-reported recall of time-activity (30-min resolution) was collected during the end of shift interviews. Images (N = 35,588) and simultaneously measured PM2.5 were collected each minute and processed to identify activities established through worker interviews, observation, and existing literature. Descriptive statistics were generated for activity types, frequencies, and associated PM2.5 exposures. A kappa statistic measured agreement between self-reported and image-based time-activity data.

RESULTS

Based on image-based time-activity patterns, workers primarily dismantled, sorted/loaded, burned, and transported e-waste materials for metal recovery with high variability in activity duration. Image-based and self-reported time-activity data had poor agreement (kappa = 0.17). Most measured exposures (90%) exceeded the World Health Organization (WHO) 24-h ambient PM2.5 target of 25 µg m-3. The average on-site PM2.5 was 81 µg m-3 (SD: 94). PM2.5 levels were highest during burning, sorting/loading and dismantling (203, 89, 83 µg m-3, respectively). PM2.5 exposure during long periods of non-work-related activities also exceeded the WHO standard in 88% of measured data.

CONCLUSIONS

In complex, informal work environments, wearable cameras can improve occupational exposure assessments and, in conjunction with monitoring equipment, identify activities associated with high exposures to workplace hazards by providing high-resolution time-activity data.

A systematic review of empirical and simulation studies evaluating the health impact of transportation interventions

Urban transportation is an important determinant of health and environmental outcomes, and therefore essential to achieving the United Nation's Sustainable Development Goals. To better understand the health impacts of transportation initiatives, we conducted a systematic review of longitudinal health evaluations involving: a) bus rapid transit (BRT); b) bicycle lanes; c) Open Streets programs; and d) aerial trams/cable cars. We also synthesized systems-based simulation studies of the health-related consequences of walking, bicycling, aerial tram, bus and BRT use. Two reviewers screened 3302 unique titles and abstracts identified through a systematic search of MEDLINE (Ovid), Scopus, TRID and LILACS databases. We included 39 studies: 29 longitudinal evaluations and 10 simulation studies. Five studies focused on low- and middle-income contexts. Of the 29 evaluation studies, 19 focused on single component bicycle lane interventions; the rest evaluated multi-component interventions involving: bicycle lanes (n = 5), aerial trams (n = 1), and combined bicycle lane/BRT systems (n = 4). Bicycle lanes and BRT systems appeared effective at increasing bicycle and BRT mode share, active transport duration, and number of trips using these modes. Of the 10 simulation studies, there were 9 agent-based models and one system dynamics model. Five studies focused on bus/BRT expansions and incentives, three on interventions for active travel, and the rest investigated combinations of public transport and active travel policies. Synergistic effects were observed when multiple policies were implemented, with several studies showing that sizable interventions are required to significantly shift travel mode choices. Our review indicates that bicycle lanes and BRT systems represent promising initiatives for promoting population health. There is also evidence to suggest that synergistic effects might be achieved through the combined implementation of multiple transportation policies. However, more rigorous evaluation and simulation studies focusing on low- and middle-income countries, aerial trams and Open Streets programs, and a more diverse set of health and health equity outcomes is required.

Translational data analytics in exposure science and environmental health: a citizen science approach with high school students

BACKGROUND

Translational data analytics aims to apply data analytics principles and techniques to bring about broader societal or human impact. Translational data analytics for environmental health is an emerging discipline and the objective of this study is to describe a real-world example of this emerging discipline.

METHODS

We implemented a citizen-science project at a local high school. Multiple cohorts of citizen scientists, who were students, fabricated and deployed low-cost air quality sensors. A cloud-computing solution provided real-time air quality data for risk screening purposes, data analytics and curricular activities.

RESULTS

The citizen-science project engaged with 14 high school students over a four-year period that is continuing to this day. The project led to the development of a website that displayed sensor-based measurements in local neighborhoods and a GitHub-like repository for open source code and instructions. Preliminary results showed a reasonable comparison between sensor-based and EPA land-based federal reference monitor data for CO and NOx.

CONCLUSIONS

Initial sensor-based data collection efforts showed reasonable agreement with land-based federal reference monitors but more work needs to be done to validate these results. Lessons learned were: 1) the need for sustained funding because citizen science-based project timelines are a function of community needs/capacity and building interdisciplinary rapport in academic settings and 2) the need for a dedicated staff to manage academic-community relationships.

Emerging Opportunities Provided by Technology to Advance Research in Child Health Globally

Current approaches to longitudinal assessment of children's developmental and psychological well-being, as mandated in the United Nations Sustainable Development Goals, are expensive and time consuming. Substantive understanding of global progress toward these goals will require a suite of new robust, cost-effective research tools designed to assess key developmental processes in diverse settings. While first steps have been taken toward this end through efforts such as the National Institutes of Health's Toolbox, experience-near approaches including naturalistic observation have remained too costly and time consuming to scale to the population level. This perspective presents 4 emerging technologies with high potential for advancing the field of child health and development research, namely (1) affective computing, (2) ubiquitous computing, (3) eye tracking, and (4) machine learning. By drawing attention of scientists, policy makers, investors/funders, and the media to the applications and potential risks of these emerging opportunities, we hope to inspire a fresh wave of innovation and new solutions to the global challenges faced by children and their families.

Bringing Big Data to Bear in Environmental Public Health: Challenges and Recommendations

Understanding the role that the environment plays in influencing public health often involves collecting and studying large, complex data sets. There have been a number of private and public efforts to gather sufficient information and confront significant unknowns in the field of environmental public health, yet there is a persistent and largely unmet need for findable, accessible, interoperable, and reusable (FAIR) data. Even when data are readily available, the ability to create, analyze, and draw conclusions from these data using emerging computational tools, such as augmented and artificial inteligence (AI) and machine learning, requires technical skills not currently implemented on a programmatic level across research hubs and academic institutions. We argue that collaborative efforts in data curation and storage, scientific computing, and training are of paramount importance to empower researchers within environmental sciences and the broader public health community to apply AI approaches and fully realize their potential. Leaders in the field were asked to prioritize challenges in incorporating big data in environmental public health research: inconsistent implementation of FAIR principles in data collection and sharing, a lack of skilled data scientists and appropriate cyber-infrastructures, and limited understanding of possibilities and communication of benefits were among those identified. These issues are discussed, and actionable recommendations are provided.

Residential greenness, air pollution and psychological well-being among urban residents in Guangzhou, China

China's rapid urbanization has led to an increasing level of exposure to air pollution and a decreasing level of exposure to vegetation among urban populations. Both trends may pose threats to psychological well-being. Previous studies on the interrelationships among greenness, air pollution and psychological well-being rely on exposure measures from remote sensing data, which may fail to accurately capture how people perceive vegetation on the ground. To address this research gap, this study aimed to explore relationships among neighbourhood greenness, air pollution exposure and psychological well-being, using survey data on 1029 adults residing in 35 neighbourhoods in Guangzhou, China. We used the Normalized Difference Vegetation Index (NDVI) and streetscape greenery (SVG) to assess greenery exposure at the neighbourhood level, and we distinguished between trees (SVG-tree) and grasses (SVG-grass) when generating streetscape greenery exposure metrics. We used two objective (PM_2.5 and NO₂ concentrations) measures and one subjective (perceived air pollution) measure to quantify air pollution exposure. We quantified psychological well-being using the World Health Organization Well-Being Index (WHO-5). Results from multilevel structural equation models (SEM) showed that, for parallel mediation models, while the association between SVG-grass and psychological well-being was completely mediated by perceived air pollution and NO₂, the relationship between SVG-tree and psychological well-being was completely mediated by ambient PM_2.5, NO₂ and perceived air pollution. None of three air pollution indicators mediated the association between psychological well-being and NDVI. For serial mediation models, measures of air pollution did not mediate the relationship between NDVI and psychological well-being. While the linkage between SVG-grass and psychological well-being scores was partially mediated by NO₂-perceived air pollution, SVG-tree was partially mediated by both ambient PM_2.5-perceived air pollution and NO₂-perceived air pollution. Our results suggest that street trees may be more related to lower air pollution levels and better mental health than grasses are.

Imaging methods used in the assessment of environmental disease networks: a brief review for clinicians

Background

Across the globe, diseases secondary to environmental exposures have been described, and it was also found that existing diseases have been modified by exposure to environmental chemicals or an environmental factor that has been found in their pathogenesis. The Institute of Medicine has shared a permanent concern related to the nations environmental health capacity since 1988.

Main body

Contemporary imaging methods in the last 15 years started reporting alterations in different human systems such as the central nervous system, cardiovascular system and pulmonary system among others; evidence suggests the existence of a human environmental disease network. The primary anatomic regions, affected by environmental diseases, recently assessed with imaging methods include Brain (lead exposure, cerebral stroke, pesticide neurotoxicity), uses MRI, DTI, carotid ultrasonography and MRS; Lungs (smoke inhalation, organophosphates poisoning) are mainly assessed with radiography; Gastrointestinal system (chronic inflammatory bowel disease), recent studies have reported the use of aortic ultrasound; Heart (myocardial infarction), its link to environmental diseased has been proved with carotid ultrasound; and Arteries (artery hypertension), the impairment of aortic mechanical properties has been revealed with the use of aortic and brachial ultrasound.

Conclusions

Environmental epidemiology has revealed that several organs and systems in the human body are targets of air pollutants. Current imaging methods that can assess the deleterious effects of pollutants includes a whole spectrum: radiography, US, CT and MRI. Future studies will help to reveal additional links among environmental disease networks.

Approximate and Situated Causality in Deep Learning

Causality is the most important topic in the history of western science, and since the beginning of the statistical paradigm, its meaning has been reconceptualized many times. Causality entered into the realm of multi-causal and statistical scenarios some centuries ago. Despite widespread critics, today deep learning and machine learning advances are not weakening causality but are creating a new way of finding correlations between indirect factors. This process makes it possible for us to talk about approximate causality, as well as about a situated causality.