A ubiquitous method for street scale spatial data collection and analysis in challenging urban environments: mapping health risks using spatial video in Haiti

Using spatial video geonarratives to improve nursing care for people who use drugs and experience homelessness: A methodology for nurses

BACKGROUND

People who are insecurely housed and use drugs are disproportionately affected by drug poisonings. Nurses are uniquely positioned to utilize harm reduction strategies to address the needs of the whole person. Needle debris encompasses drug paraphernalia discarded in public spaces. Studying needle debris provides a strategic opportunity to identify where drugs are being used and target public health strategies accordingly.

AIM

Our aim in this article is to illustrate how spatial video geonarratives (SVG) combined GPS technology interviews, and videos of locations with needle debris, can elicit valuable data for nursing research.

METHODS

Using SVG required knowledge of how to collect data wearing cameras and practice sessions were necessary. A Miufly camera worn at waist height on a belt provided the stability to walk while interviewing stakeholders. We wore the cameras and conducted go-along interviews with outreach workers, while filming the built environment. Upon completion of data collection, both the interview and GPS information were analysed using Wordmapper software.

CONCLUSIONS

This methodology resulted in data presented uniquely in both a visual map and narrative. These data were richer than if a single modality had been used. These data highlighted specific contextual factors that were related to the location of needle debris, which created opportunities for nursing interventions to support people experiencing vulnerability.

The Use of Spatial Video to Map Dynamic and Challenging Environments: A Case Study of Cholera Risk in the Mujoga Relief Camp, D.R.C

In this paper, we provide an overview of how spatial video data collection enriched with contextual mapping can be used as a universal tool to investigate sub-neighborhood scale health risks, including cholera, in challenging environments. To illustrate the method’s flexibility, we consider the life cycle of the Mujoga relief camp set up after the Nyiragongo volcanic eruption in the Democratic Republic of Congo on 22 May 2021. More specifically we investigate how these methods have captured the deteriorating conditions in a camp which is also experiencing lab-confirmed cholera cases. Spatial video data are collected every month from June 2021 to March 2022. These coordinate-tagged images are used to make monthly camp maps, which are then returned to the field teams for added contextual insights. At the same time, a zoom-based geonarrative is used to discuss the camp’s changes, including the cessation of free water supplies and the visible deterioration of toilet facilities. The paper concludes by highlighting the next data science advances to be made with SV mapping, including machine learning to automatically identify and map risks, and how these are already being applied in Mujoga.

VisualCommunity: a platform for archiving and studying communities

VisualCommunity is a platform designed to support community or neighborhood scale research. The platform integrates mobile, AI, visualization techniques, along with tools to help domain researchers, practitioners, and students collecting and working with spatialized video and geo-narratives. These data, which provide granular spatialized imagery and associated context gained through expert commentary have previously provided value in understanding various community-scale challenges. This paper further enhances this work AI-based image processing and speech transcription tools available in VisualCommunity, allowing for the easy exploration of the acquired semantic and visual information about the area under investigation. In this paper we describe the specific advances through use case examples including COVID-19 related scenarios.

Delineating and Analyzing Locality-Level Determinants of Cholera, Haiti

Centre Department, Haiti, was the origin of a major cholera epidemic during 2010–2019. Although no fine-scale spatial delineation is officially available, we aimed to analyze determinants of cholera at the local level and identify priority localities in need of interventions. After estimating the likely boundaries of 1,730 localities by using Voronoi polygons, we mapped 5,322 suspected cholera cases reported during January 2015–September 2016 by locality alongside environmental and socioeconomic variables. A hierarchical clustering on principal components highlighted 2 classes with high cholera risk: localities close to rivers and unimproved water sources (standardized incidence ratio 1.71, 95% CI 1.02–2.87; p = 0.04) and urban localities with markets (standardized incidence ratio 1.69, 95% CI 1.25–2.29; p = 0.0006). Our analyses helped identify and characterize areas where efforts should be focused to reduce vulnerability to cholera and other waterborne diseases; these methods could be used in other contexts.

Exploring convolutional neural networks and spatial video for on-the-ground mapping in informal settlements

BACKGROUND

The health burden in developing world informal settlements often coincides with a lack of spatial data that could be used to guide intervention strategies. Spatial video (SV) has proven to be a useful tool to collect environmental and social data at a granular scale, though the effort required to turn these spatially encoded video frames into maps limits sustainability and scalability. In this paper we explore the use of convolution neural networks (CNN) to solve this problem by automatically identifying disease related environmental risks in a series of SV collected from Haiti. Our objective is to determine the potential of machine learning in health risk mapping for these environments by assessing the challenges faced in adequately training the required classification models.

RESULTS

We show that SV can be a suitable source for automatically identifying and extracting health risk features using machine learning. While well-defined objects such as drains, buckets, tires and animals can be efficiently classified, more amorphous masses such as trash or standing water are difficult to classify. Our results further show that variations in the number of image frames selected, the image resolution, and combinations of these can be used to improve the overall model performance.

CONCLUSION

Machine learning in combination with spatial video can be used to automatically identify environmental risks associated with common health problems in informal settlements, though there are likely to be variations in the type of data needed for training based on location. Success based on the risk type being identified are also likely to vary geographically. However, we are confident in identifying a series of best practices for data collection, model training and performance in these settings. We also discuss the next step of testing these findings in other environments, and how adding in the simultaneously collected geographic data could be used to create an automatic health risk mapping tool.

Classifying crime places by neighborhood visual appearance and police geonarratives: a machine learning approach

The complex interrelationship between the built environment and social problems is often described but frequently lacks the data and analytical framework to explore the potential of such a relationship in different applications. We address this gap using a machine learning (ML) approach to study whether street-level built environment visuals can be used to classify locations with high-crime and lower-crime activities. For training the ML model, spatialized expert narratives are used to label different locations. Semantic categories (e.g., road, sky, greenery, etc.) are extracted from Google Street View (GSV) images of those locations through a deep learning image segmentation algorithm. From these, local visual representatives are generated and used to train the classification model. The model is applied to two cities in the U.S. to predict the locations as being linked to high crime. Results show our model can predict high- and lower-crime areas with high accuracies (above 98% and 95% in first and second test cities, accordingly).

Solid Wastes Provide Breeding Sites, Burrows, and Food for Biological Disease Vectors, and Urban Zoonotic Reservoirs: A Call to Action for Solutions-Based Research

Background: Infectious disease epidemiology and planetary health literature often cite solid waste and plastic pollution as risk factors for vector-borne diseases and urban zoonoses; however, no rigorous reviews of the risks to human health have been published since 1994. This paper aims to identify research gaps and outline potential solutions to interrupt the vicious cycle of solid wastes; disease vectors and reservoirs; infection and disease; and poverty.

Methods: We searched peer-reviewed publications from PubMed, Google Scholar, and Stanford Searchworks, and references from relevant articles using the search terms (“disease” OR “epidemiology”) AND (“plastic pollution,” “garbage,” and “trash,” “rubbish,” “refuse,” OR “solid waste”). Abstracts and reports from meetings were included only when they related directly to previously published work. Only articles published in English, Spanish, or Portuguese through 2018 were included, with a focus on post-1994, after the last comprehensive review was published. Cancer, diabetes, and food chain-specific articles were outside the scope and excluded. After completing the literature review, we further limited the literature to “urban zoonotic and biological vector-borne diseases” or to “zoonotic and biological vector-borne diseases of the urban environment.”

Results: Urban biological vector-borne diseases, especially Aedes-borne diseases, are associated with solid waste accumulation but vector preferences vary over season and region. Urban zoonosis, especially rodent and canine disease reservoirs, are associated with solid waste in urban settings, especially when garbage accumulates over time, creating burrowing sites and food for reservoirs. Although evidence suggests the link between plastic pollution/solid waste and human disease, measurements are not standardized, confounders are not rigorously controlled, and the quality of evidence varies. Here we propose a framework for solutions-based research in three areas: innovation, education, and policy.

Conclusions: Disease epidemics are increasing in scope and scale with urban populations growing, climate change providing newly suitable vector climates, and immunologically naïve populations becoming newly exposed. Sustainable solid waste management is crucial to prevention, specifically in urban environments that favor urban vectors such as Aedes species. We propose that next steps should include more robust epidemiological measurements and propose a framework for solutions-based research.

A Local View of Informal Urban Environments: a Mobile Phone-Based Neighborhood Audit of Street-Level Factors in a Brazilian Informal Community

Street-level environment characteristics influence the health behaviors and safety of urban residents, and may particularly threaten health within informal communities. However, available data on how such characteristics vary within and among informal communities is limited. We sought to adapt street audit strategies designed to characterize the physical environment for use in a large informal community, Rio das Pedras (RdP) located in Rio de Janeiro, Brazil. A smartphone-based systematic observation protocol was used to gather street-level information for a high-density convenience sample of street segments (N = 630, estimated as 86% of all street segments in the community). We adapted items related to physical disorder and physical deterioration. Measures selected to illustrate the approach include the presence of the following: (1) low-hanging or tangled wires, (2) litter, (3) structural evidence of sinking, and (4) an unpleasant odor. Intercept-only spatial generalized additive models (GAM) were used to evaluate and visualize spatial variation within the RdP community. We also examined how our estimates and conclusions about spatial variation might have been affected by lower-density sampling from random subsets street observations. Random subsets were selected to determine the robustness of study results in scenarios with sparser street sampling. Selected characteristics were estimated to be present for between 18% (unpleasant odor) to 59% (low-hanging or tangled wires) of the street segments in RdP; estimates remain similar (± 6%) when relying on a random subset created to simulate lower-density spatial sampling. Spatial patterns of variation based on predicted probabilities across RdP differed by indicator. Structural sinking and low-hanging or tangled wires demonstrated relatively consistent spatial distribution patterns across full and random subset sample sizes. Smartphone-based systematic observations represent an efficient and potentially feasible approach to systematically studying neighborhood environments within informal communities. Future deployment of such tools will benefit from incorporating data collection across multiple time points to explore reliability and quantify neighborhood change. These tools can prove useful means to assess street-level exposures that can be modifiable health determinants across a wide range of informal urban settings. Findings can contribute to improved urban planning and provide useful information for identifying potential locations for neighborhood-scaled interventions that can improve living conditions for residents in Rio das Pedras.

Micro-Space Complexity and Context in the Space-Time Variation in Enteric Disease Risk for Three Informal Settlements of Port au Prince, Haiti

Diffusion of cholera and other diarrheal diseases in an informal settlement is a product of multiple behavioral, environmental and spatial risk factors. One of the most important components is the spatial interconnections among water points, drainage ditches, toilets and the intervening environment. This risk is also longitudinal and variable as water points fluctuate in relation to bacterial contamination. In this paper we consider part of this micro space complexity for three informal settlements in Port au Prince, Haiti. We expand on more typical epidemiological analysis of fecal coliforms at water points, drainage ditches and ocean sites by considering the importance of single point location fluctuation coupled with recording micro-space environmental conditions around each sample site. Results show that spatial variation in enteric disease risk occurs within neighborhoods, and that while certain trends are evident, the degree of individual site fluctuation should question the utility of both cross-sectional and more aggregate analysis. Various factors increase the counts of fecal coliform present, including the type of water point, how water was stored at that water point, and the proximity of the water point to local drainage. Some locations fluctuated considerably between being safe and unsafe on a monthly basis. Next steps to form a more comprehensive contextualized understanding of enteric disease risk in these environments should include the addition of behavioral factors and local insight.

The Use of Geonarratives to Add Context to Fine Scale Geospatial Research

There has been a move towards using mixed method approaches in geospatial research to gain context in understanding health related social patterns and processes. The central premise is that official data is often too reductionist and misses’ nuances that can help explain causality. One example is the geonarrative, a spatially relevant commentary or interview that can be mapped by content and/or location. While there have been several examples of geonarratives being used by researchers, there is no commonly available software that can easily transfer the associated text into spatial data. Having a standardized software platform is vital if these methods are to be used across different disciplines. This paper presents an overview of a solution, Wordmapper (WM), which is a standalone software developed to process geonarratives from a transcription and associated global positioning system (GPS) path. Apart from querying textual narrative data, Wordmapper facilitates qualitative coding which could be used to extract latent contextual information from the narratives. In order to improve interoperability, Wordmapper provides spatialized narrative data in formats, such as ESRI shape files, Keyhole Markup Language (KML), and Comma Separated Values (CSV). A case study based on five different spatial video geonarratives (SVG) collected to assess the human impacts following the 2011 Joplin, Missouri are used for illustration.

Spatial Video Health Risk Mapping in Informal Settlements: Correcting GPS Error

Informal settlements pose a continuing health concern. While spatial methodologies have proven to be valuable tools to support health interventions, several factors limit their widespread use in these challenging environments. One such technology, spatial video, has been used for fine-scale contextualized mapping. In this paper, we address one of the limitations of the technique: the global positioning system (GPS) coordinate error. More specifically, we show how spatial video coordinate streams can be corrected and synced back to the original video to facilitate risk mapping. Past spatial video collections for the Mathare informal settlement of Kenya are used as an illustration as these data had been previously discarded because of excessive GPS error. This paper will describe the bespoke software that makes these corrections possible, and then will go on to investigate patterns in the coordinate error.

Socio-environmental factors and diarrheal diseases in under five-year old children in the state of Tocantins, Brazil

BACKGROUND

Diarrhea is a waterborne disease that affects children, especially those under 5 years of age. The objective of this study was to identify the spatial patterns of distribution of diarrheal disease in under 5-year-old children in the State of Tocantins, Brazil, from 2008 to 2013.

METHODS

Geoprocessing tools were used to carry out an epidemiological study, to prepare thematic maps in the TerraView 4.2.2 software based on secondary data. General indicators of the disease, presence of spatial dependence through the Global Moran's Index (I) and the Spatial Association Index (LISA) were described.

RESULTS

There were 3,015 cases of under 5-year-old children hospitalized for diarrhea, with an average annual rate (AAR) of 4.10/1,000 inhabitants (inhab.). Among the main characteristics were: increasing rates in under 1-year-old children (6.16 to 9.66/1,000 inhabitants); children aged 1 to 4 full years (63%); males (55%); 8 deaths of under one-year-old children (75%); county of Araguaína (67%); incidence in the county of Nazaré (63.97/1,000 inhab.); prevalence and incidence in the Araguaína microregion (45%, AAR 9.38/1,000 inhab.). The presence of a cluster with spatial autocorrelation was found in the Araguaína microregion, which was statistically significant (I = 0.11, p-value < 0.03), with priority of intervention (Moran Map).

CONCLUSIONS

There was an increase in the number of hospitalizations for diarrhea in under 5-year-old children in the state of Tocantins. The spatial analysis identified clusters of priority areas for measures of maintenance and control of diarrheal diseases.

Community context and sub-neighborhood scale detail to explain dengue, chikungunya and Zika patterns in Cali, Colombia

Background

Cali, Colombia has experienced chikungunya and Zika outbreaks and hypoendemic dengue. Studies have explained Cali’s dengue patterns but lack the sub-neighborhood-scale detail investigated here.

Methods

Spatial-video geonarratives (SVG) with Ministry of Health officials and Community Health Workers were collected in hotspots, providing perspective on perceptions of why dengue, chikungunya and Zika hotspots exist, impediments to control, and social outcomes. Using spatial video and Google Street View, sub-neighborhood features possibly contributing to incidence were mapped to create risk surfaces, later compared with dengue, chikungunya and Zika case data.

Results

SVG captured insights in 24 neighborhoods. Trash and water risks in Calipso were mapped using SVG results. Perceived risk factors included proximity to standing water, canals, poverty, invasions, localized violence and military migration. These risks overlapped case density maps and identified areas that are suitable for transmission but are possibly underreporting to the surveillance system.

Conclusion

Resulting risk maps with local context could be leveraged to increase vector-control efficiency- targeting key areas of environmental risk.

Using Spatial Video to Analyze and Map the Water-Fetching Path in Challenging Environments: A Case Study of Dar es Salaam, Tanzania

Access to clean drinking water remains a significant health problem in the developing world. Traditional definitions of water access oversimplify the geographic context of water availability, the burden of water collection, and challenges faced along the path, mainly due to a lack of fine scale spatial data. This paper demonstrates how spatial video collected in three informal areas of Dar es Salaam, Tanzania, can be used to quantify aspects of the walk to water. These include impediments encountered along the path such as changes in elevation and proximity to traffic. All are mapped along with classic health-related environmental and social information, such as standing water, drains, and trash. The issue of GPS error was encountered due to the built environment that is typical of informal settlements. The spatial video allowed for the correction of the path to gain a more accurate estimate of time and distance for each walk. The resulting mapped health risks at this fine scale of detail reveal micro-geographies of concern. Spatial video is a useful tool for visualizing and analyzing the challenges of water collection. It also allows for data generated along the walk to become part of both a household and local area risk assessment.

Mapping to Support Fine Scale Epidemiological Cholera Investigations: A Case Study of Spatial Video in Haiti

The cartographic challenge in many developing world environments suffering a high disease burden is a lack of granular environmental covariates suitable for modeling disease outcomes. As a result, epidemiological questions, such as how disease diffuses at intra urban scales are extremely difficult to answer. This paper presents a novel geospatial methodology, spatial video, which can be used to collect and map environmental covariates, while also supporting field epidemiology. An example of epidemic cholera in a coastal town of Haiti is used to illustrate the potential of this new method. Water risks from a 2012 spatial video collection are used to guide a 2014 survey, which concurrently included the collection of water samples, two of which resulted in positive lab results “of interest” (bacteriophage specific for clinical cholera strains) to the current cholera situation. By overlaying sample sites on 2012 water risk maps, a further fifteen proposed water sample locations are suggested. These resulted in a third spatial video survey and an additional “of interest” positive water sample. A potential spatial connection between the “of interest” water samples is suggested. The paper concludes with how spatial video can be an integral part of future fine-scale epidemiological investigations for different pathogens.

Breaking Out of Surveillance Silos: Integrative Geospatial Data Collection for Child Injury Risk and Active School Transport

The preponderance of active school transport (AST) and child injury research has occurred independently, yet they are inherently related. This is particularly true in urban areas where the environmental context of AST may pose risks to safety. However, it can be difficult to make these connections due to the often segregated nature in which these veins of research operate. Spatial video presents a geospatial approach for simultaneous data collection related to both issues. This article reports on a multi-sector pilot project among researchers, a children's hospital, and a police department, using spatial video to map child AST behaviors; a geographic information system (GIS) is used to analyze these data in the environmental context of child pedestrian injury and community violence.

Spatial video geonarratives and health: case studies in post-disaster recovery, crime, mosquito control and tuberculosis in the homeless

Background

A call has recently been made by the public health and medical communities to understand the neighborhood context of a patient’s life in order to improve education and treatment. To do this, methods are required that can collect “contextual” characteristics while complementing the spatial analysis of more traditional data. This also needs to happen within a standardized, transferable, easy-to-implement framework.

Methods

The Spatial Video Geonarrative (SVG) is an environmentally-cued narrative where place is used to stimulate discussion about fine-scale geographic characteristics of an area and the context of their occurrence. It is a simple yet powerful approach to enable collection and spatial analysis of expert and resident health-related perceptions and experiences of places. Participants comment about where they live or work while guiding a driver through the area. Four GPS-enabled cameras are attached to the vehicle to capture the places that are observed and discussed by the participant. Audio recording of this narrative is linked to the video via time stamp. A program (G-Code) is then used to geotag each word as a point in a geographic information system (GIS). Querying and density analysis can then be performed on the narrative text to identify spatial patterns within one narrative or across multiple narratives. This approach is illustrated using case studies on post-disaster psychopathology, crime, mosquito control, and TB in homeless populations.

Results

SVG can be used to map individual, group, or contested group context for an environment. The method can also gather data for cohorts where traditional spatial data are absent. In addition, SVG provides a means to spatially capture, map and archive institutional knowledge.

Conclusions

SVG GIS output can be used to advance theory by being used as input into qualitative and/or spatial analyses. SVG can also be used to gain near-real time insight therefore supporting applied interventions. Advances over existing geonarrative approaches include the simultaneous collection of video data to visually support any commentary, and the ease-of-application making it a transferable method across different environments and skillsets.

Water-related infrastructure in a region of post-earthquake Haiti: high levels of fecal contamination and need for ongoing monitoring

We inventoried non-surface water sources in the Leogane and Gressier region of Haiti (approximately 270 km²) in 2012 and 2013 and screened water from 345 sites for fecal coliforms and Vibrio cholerae. An international organization/non-governmental organization responsible for construction could be identified for only 56% of water points evaluated. Sixteen percent of water points were non-functional at any given time; 37% had evidence of fecal contamination, with spatial clustering of contaminated sites. Among improved water sources (76% of sites), 24.6% had fecal coliforms versus 80.9% in unimproved sources. Fecal contamination levels increased significantly from 36% to 51% immediately after the passage of Tropical Storm Sandy in October of 2012, with a return to 34% contamination in March of 2013. Long-term sustainability of potable water delivery at a regional scale requires ongoing assessment of water quality, functionality, and development of community-based management schemes supported by a national plan for the management of potable water.

A system for household enumeration and re-identification in densely populated slums to facilitate community research, education, and advocacy

BACKGROUND

We devised and implemented an innovative Location-Based Household Coding System (LBHCS) appropriate to a densely populated informal settlement in Mumbai, India.

METHODS AND FINDINGS

LBHCS codes were designed to double as unique household identifiers and as walking directions; when an entire community is enumerated, LBHCS codes can be used to identify the number of households located per road (or lane) segment. LBHCS was used in community-wide biometric, mental health, diarrheal disease, and water poverty studies. It also facilitated targeted health interventions by a research team of youth from Mumbai, including intensive door-to-door education of residents, targeted follow-up meetings, and a full census. In addition, LBHCS permitted rapid and low-cost preparation of GIS mapping of all households in the slum, and spatial summation and spatial analysis of survey data.

CONCLUSION

LBHCS was an effective, easy-to-use, affordable approach to household enumeration and re-identification in a densely populated informal settlement where alternative satellite imagery and GPS technologies could not be used.