Community health assessment using self-organizing maps and geographic information systems

Spatiotemporal evolution of COVID-19 in Portugal's Mainland with self-organizing maps

BACKGROUND

Self-Organizing Maps (SOM) are an unsupervised learning clustering and dimensionality reduction algorithm capable of mapping an initial complex high-dimensional data set into a low-dimensional domain, such as a two-dimensional grid of neurons. In the reduced space, the original complex patterns and their interactions can be better visualized, interpreted and understood.

METHODS

We use SOM to simultaneously couple the spatial and temporal domains of the COVID-19 evolution in the 278 municipalities of mainland Portugal during the first year of the pandemic. Temporal 14-days cumulative incidence time series along with socio-economic and demographic indicators per municipality were analyzed with SOM to identify regions of the country with similar behavior and infer the possible common origins of the incidence evolution.

RESULTS

The results show how neighbor municipalities tend to share a similar behavior of the disease, revealing the strong spatiotemporal relationship of the COVID-19 spreading beyond the administrative borders of each municipality. Additionally, we demonstrate how local socio-economic and demographic characteristics evolved as determinants of COVID-19 transmission, during the 1st wave school density per municipality was more relevant, where during 2nd wave jobs in the secondary sector and the deprivation score were more relevant.

CONCLUSIONS

The results show that SOM can be an effective tool to analysing the spatiotemporal behavior of COVID-19 and synthetize the history of the disease in mainland Portugal during the period in analysis. While SOM have been applied to diverse scientific fields, the application of SOM to study the spatiotemporal evolution of COVID-19 is still limited. This work illustrates how SOM can be used to describe the spatiotemporal behavior of epidemic events. While the example shown herein uses 14-days cumulative incidence curves, the same analysis can be performed using other relevant data such as mortality data, vaccination rates or even infection rates of other disease of infectious nature.

Using self-organising maps to predict and contain natural disasters and pandemics

The unfolding coronavirus (COVID-19) pandemic has highlighted the global need for robust predictive and containment tools and strategies. COVID-19 continues to cause widespread economic and social turmoil, and while the current focus is on both minimising the spread of the disease and deploying a range of vaccines to save lives, attention will soon turn to future proofing. In line with this, this paper proposes a prediction and containment model that could be used for pandemics and natural disasters. It combines selective lockdowns and protective cordons to rapidly contain the hazard while allowing minimally impacted local communities to conduct "business as usual" and/or offer support to highly impacted areas. A flexible, easy to use data analytics model, based on Self Organising Maps, is developed to facilitate easy decision making by governments and organisations. Comparative tests using publicly available data for Great Britain (GB) show that through the use of the proposed prediction and containment strategy, it is possible to reduce the peak infection rate, while keeping several regions (up to 25% of GB parliamentary constituencies) economically active within protective cordons.

Exploring the socio-economic and environmental components of infectious diseases using multivariate geovisualization: West Nile Virus

Background

This study postulates that underlying environmental conditions and a susceptible population's socio-economic status should be explored simultaneously to adequately understand a vector borne disease infection risk. Here we focus on West Nile Virus (WNV), a mosquito borne pathogen, as a case study for spatial data visualization of environmental characteristics of a vector's habitat alongside human demographic composition for understanding potential public health risks of infectious disease. Multiple efforts have attempted to predict WNV environmental risk, while others have documented factors related to human vulnerability to the disease. However, analytical modeling that combines the two is difficult due to the number of potential explanatory variables, varying spatial resolutions of available data, and differing research questions that drove the initial data collection. We propose that the use of geovisualization may provide a glimpse into the large number of potential variables influencing the disease and help distill them into a smaller number that might reveal hidden and unknown patterns. This geovisual look at the data might then guide development of analytical models that can combine environmental and socio-economic data.

Methods

Geovisualization was used to integrate an environmental model of the disease vector's habitat alongside human risk factors derived from socio-economic variables. County level WNV incidence rates from California, USA, were used to define a geographically constrained study area where environmental and socio-economic data were extracted from 1,133 census tracts. A previously developed mosquito habitat model that was significantly related to WNV infected dead birds was used to describe the environmental components of the study area. Self-organizing maps found 49 clusters, each of which contained census tracts that were more similar to each other in terms of WNV environmental and socio-economic data. Parallel coordinate plots permitted visualization of each cluster's data, uncovering patterns that allowed final census tract mapping exposing complex spatial patterns contained within the clusters.

Results

Our results suggest that simultaneously visualizing environmental and socio-economic data supports a fuller understanding of the underlying spatial processes for risks to vector-borne disease. Unexpected patterns were revealed in our study that would be useful for developing future multilevel analytical models. For example, when the cluster that contained census tracts with the highest median age was examined, it was determined that those census tracts only contained moderate mosquito habitat risk. Likewise, the cluster that contained census tracts with the highest mosquito habitat risk had populations with moderate median age. Finally, the cluster that contained census tracts with the highest WNV human incidence rates had unexpectedly low mosquito habitat risk.

A spatially explicit analysis of chronic diseases in small areas: a case study of diabetes in Santiago, Chile

Background

There is a strong spatial correlation between demographics and chronic diseases in urban areas. Thus, most of the public policies aimed at improving prevention plans and optimizing the allocation of resources in health networks should be designed specifically for the socioeconomic reality of the population. One way to tackle this challenge is by exploring within a small geographical area the spatial patterns that link the sociodemographic attributes that characterize a community, its risk of suffering chronic diseases, and the accessibility of health treatment. Due to the inherent complexity of cities, soft clustering methods are recommended to find fuzzy spatial patterns. Our main motivation is to provide health planners with valuable spatial information to support decision-making. For the case study, we chose to investigate diabetes in Santiago, Chile.

Methods

To deal with spatiality, we combine two statistical techniques: spatial microsimulation and a self-organizing map (SOM). Spatial microsimulation allows spatial disaggregation of health indicators data to a small area level. In turn, SOM, unlike classical clustering methods, incorporates a learning component through neural networks, which makes it more appropriate to model complex adaptive systems, such as cities. Thus, while spatial microsimulation generates the data for the analysis, the SOM method finds the relevant socio-economic clusters. We selected age, sex, income, prevalence of diabetes, distance to public health services, and type of health insurance as input variables. We used public surveys as input data.

Results

We found four significant spatial clusters representing 75 percent of the whole population in Santiago. Two clusters correspond to people with low educational levels, low income, high accessibility to public health services, and a high prevalence of diabetes. However, one presents a significantly higher level of diabetes than the other. The second pair of clusters is made up of people with high educational levels, high income, and low prevalence of diabetes. What differentiates both clusters is accessibility to health centers. The average distance to the health centers of one group almost doubles that of the other.

Conclusions

In this study, we combined two statistical techniques: spatial microsimulation and selforganising maps to explore the relationship between diabetes and socio-demographics in Santiago, Chile. The results have allowed us to corroborate the importance of the spatial factor in the analysis of chronic diseases as a way of suggesting differentiated solutions to spatially explicit problems. SOM turned out to be a good choice to deal with fuzzy health and socioeconomic data. The method explored and uncovered valuable spatial patterns for health decision-making. In turn, spatial microsimulation.

Decision Model for Predicting Social Vulnerability Using Artificial Intelligence

Social vulnerability, from a socio-environmental point of view, focuses on the identification of disadvantaged or vulnerable groups and the conditions and dynamics of the environments in which they live. To understand this issue, it is important to identify the factors that explain the difficulty of facing situations with a social disadvantage. Due to its complexity and multidimensionality, it is not always easy to point out the social groups and urban areas affected. This research aimed to assess the connection between certain dimensions of social vulnerability and its urban and dwelling context as a fundamental framework in which it occurs using a decision model useful for the planning of social and urban actions. For this purpose, a holistic approximation was carried out on the census and demographic data commonly used in this type of study, proposing the construction of (i) a knowledge model based on Artificial Neural Networks (Self-Organizing Map), with which a demographic profile is identified and characterized whose indicators point to a presence of social vulnerability, and (ii) a predictive model of such a profile based on rules from dwelling variables constructed by conditional inference trees. These models, in combination with Geographic Information Systems, make a decision model feasible for the prediction of social vulnerability based on housing information.

Spatial distribution and cluster analysis of dengue using self organizing maps in Andhra Pradesh, India, 2011-2013

BACKGROUND AND OBJECTIVES

Dengue is an emerging and re-emerging infectious disease, transmitted by mosquitoes. It is mostly prevalent in tropical and sub-tropical regions of the world, particularly, in Asia-Pacific region. To understand the epidemiology and spatial distribution of dengue, a retrospective surveillance study was conducted in the state of Andhra Pradesh, India during 2011-2013.

MATERIAL AND METHODS

District-wise disease endemicity levels were mapped through geographical information system (GIS) tools. Spatial statistical analysis such as Getis-Ord Gi* was performed to identify hot spots and cold spots of dengue disease. Similarly self organizing maps (SOM), a datamining tool was also applied to understand the endemicity patterns in study areas.

RESULTS

The analysis shows that districts of Warangal, Karimnagar, Khammam and Vizianagaram are reported as hot spot regions whereas Adilabad and Nizamabad reported as cold spots for dengue. The SOM classify 23 districts in 03 major (07 sub) clusters. These SOM clusters were projected in the geographical space and based on the disease/cases intensity the districts were characterized into low, medium and high endemic areas.

CONCLUSION

This visualization approach, SOM-GIS helps the public health officials to identify the disease endemic zones and take real time decisions for disease management.

An Overview of Literature Topics Related to Current Concepts, Methods, Tools, and Applications for Cumulative Risk Assessment (2007-2016)

Cumulative risk assessments (CRAs) address combined risks from exposures to multiple chemical and nonchemical stressors and may focus on vulnerable communities or populations. Significant contributions have been made to the development of concepts, methods, and applications for CRA over the past decade. Work in both human health and ecological cumulative risk has advanced in two different contexts. The first context is the effects of chemical mixtures that share common modes of action, or that cause common adverse outcomes. In this context two primary models are used for predicting mixture effects, dose addition or response addition. The second context is evaluating the combined effects of chemical and nonchemical (e.g., radiation, biological, nutritional, economic, psychological, habitat alteration, land-use change, global climate change, and natural disasters) stressors. CRA can be adapted to address risk in many contexts, and this adaptability is reflected in the range in disciplinary perspectives in the published literature. This article presents the results of a literature search and discusses a range of selected work with the intention to give a broad overview of relevant topics and provide a starting point for researchers interested in CRA applications.

Quantifying the Visual Impact of Classification Boundaries in Choropleth Maps

One critical visual task when using choropleth maps is to identify spatial clusters in the data. If spatial units have the same color and are in the same neighborhood, this region can be visually identified as a spatial cluster. However, the choice of classification method used to create the choropleth map determines the visual output. The critical map elements in the classification scheme are those that lie near the classification boundary as those elements could potentially belong to different classes with a slight adjustment of the classification boundary. Thus, these elements have the most potential to impact the visual features (i.e., spatial clusters) that occur in the choropleth map. We present a methodology to enable analysts and designers to identify spatial regions where the visual appearance may be the result of spurious data artifacts. The proposed methodology automatically detects the critical boundary cases that can impact the overall visual presentation of the choropleth map using a classification metric of cluster stability. The map elements that belong to a critical boundary case are then automatically assessed to quantify the visual impact of classification edge effects. Our results demonstrate the impact of boundary elements on the resulting visualization and suggest that special attention should be given to these elements during map design.

Self-organizing maps as an approach to exploring spatiotemporal diffusion patterns

Background

Self-organizing maps (SOMs) have now been applied for a number of years to identify patterns in large datasets; yet, their application in the spatiotemporal domain has been lagging. Here, we demonstrate how spatialtemporal disease diffusion patterns can be analysed using SOMs and Sammon’s projection.

Methods

SOMs were applied to identify synchrony between spatial locations, to group epidemic waves based on similarity of diffusion pattern and to construct sequence of maps of synoptic states. The Sammon’s projection was used to created diffusion trajectories from the SOM output. These methods were demonstrated with a dataset that reports Measles outbreaks that took place in Iceland in the period 1946–1970. The dataset reports the number of Measles cases per month in 50 medical districts.

Results

Both stable and incidental synchronisation between medical districts were identified as well as two distinct groups of epidemic waves, a uniformly structured fast developing group and a multiform slow developing group. Diffusion trajectories for the fast developing group indicate a typical diffusion pattern from Reykjavik to the northern and eastern parts of the island. For the other group, diffusion trajectories are heterogeneous, deviating from the Reykjavik pattern.

Conclusions

This study demonstrates the applicability of SOMs (combined with Sammon’s Projection and GIS) in spatiotemporal diffusion analyses. It shows how to visualise diffusion patterns to identify (dis)similarity between individual waves and between individual waves and an overall time-series performing integrated analysis of synchrony and diffusion trajectories.

Algorithm to determine the outcome of patients with acute liver failure: a data-mining analysis using decision trees

BACKGROUND

We established algorithms to predict the prognosis of acute liver failure (ALF) patients through a data-mining analysis, in order to improve the indication criteria for liver transplantation.

METHODS

The subjects were 1,022 ALF patients seen between 1998 and 2007 and enrolled in a nationwide survey. Patients older than 65 years, and those who had undergone liver transplantation and received blood products before the onset of hepatic encephalopathy were excluded. Two data sets were used: patients seen between 1998 and 2003 (n=698), whose data were used for the formation of the algorithm, and those seen between 2004 and 2007 (n=324), whose data were used for the validation of the algorithm. Data on a total of 73 items, at the onset of encephalopathy and 5 days later, were collected from 371 of the 698 patients seen between 1998 and 2003, and their outcome was analyzed to establish decision trees. The obtained algorithm was validated using the data of 160 of the 324 patients seen between 2004 and 2007.

RESULTS

The outcome of the patients at the onset of encephalopathy was predicted through 5 items, and the patients were classified into 6 categories with mortality rates between 23% and89%. When the prognosis of the patients in the categories with mortality rates greater than 50% was predicted as "death", the accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the algorithm were 79, 78, 81, 83, and 75%, respectively. Similar high values were obtained when the algorithm was employed in the patients for validation. The outcome of the patients 5 days after the onset of encephalopathy was predicted through 7 items, and a similar high accuracy was found for both sets of patients.

CONCLUSIONS

Novel algorithms for predicting the outcome of ALF patients may be useful to determine the indication for liver transplantation.

Urban youths go 3000 miles: engaging and supporting young residents to conduct neighborhood asset mapping

In 2009, CARE (Community Alliance for Research and Engagement at Yale University) launched a multisectoral chronic disease prevention initiative that conducts baseline data collection, interventions, and follow-up data collection to measure change. Data collection includes asset mapping to assess environmental determinants of chronic disease risk factors in neighborhoods and around schools. CARE hired 7 local high school students to conduct asset mapping; they walked more than 3000 miles and collected 492 data points. Employing youths as community health workers to collect data greatly enriched the community research process and offered many advantages. We were able to efficiently and effectively conduct scientifically rigorous mapping while gaining entry into some of New Haven's most research-wary and skeptical neighborhoods.

Incorporating geospatial capacity within clinical data systems to address social determinants of health

Linking electronic health record (EHR) systems with community information systems (CIS) holds great promise for addressing inequities in social determinants of health (SDH). While EHRs are rich in location-specific data that allow us to uncover geographic inequities in health outcomes, CIS are rich in data that allow us to describe community-level characteristics relating to health. When meaningfully integrated, these data systems enable clinicians, researchers, and public health professionals to actively address the social etiologies of health disparities.This article describes a process for exploring SDH by geocoding and integrating EHR data with a comprehensive CIS covering a large metropolitan area. Because the systems were initially designed for different purposes and had different teams of experts involved in their development, integrating them presents challenges that require multidisciplinary expertise in informatics, geography, public health, and medicine. We identify these challenges and the means of addressing them and discuss the significance of the project as a model for similar projects.

Novel classification of acute liver failure through clustering using a self-organizing map: usefulness for prediction of the outcome

BACKGROUND

Patients with acute liver failure are classified according to the interval between the onset of hepatitis symptoms and the development of hepatic encephalopathy. We examined the validity of such classifications.

METHODS

The subjects were 1,022 patients enrolled in a nationwide survey in Japan. The intervals between the onset of the hepatitis symptoms and the development of encephalopathy were 10 days or less in 472 patients (group-A), between 11 and 56 days in 468 patients (group-B), and longer than 56 days in 82 patients (group-C). Data on a total of 104 items collected from the patients were subjected to clustering using a self-organizing map.

RESULTS

The patients were classified into three clusters. The first cluster consisted of 411 patients (group-A: 57%, group-B: 39%, group-C: 4%). Their incidence of complications was low; 34% underwent liver transplantation (LT), and their survival rate was 90%, while 94% of those treated without transplant were rescued. The second cluster consisted of 320 patients (21, 65, and 14% groups A, B, and C, respectively), who showed a high incidence of complications; the survival rate was 7% in the patients treated conservatively without LT. Sixteen percent underwent LT and survival rate of these patients was 52%. There was a third cluster, of 291 patients (59, 34, and 7% groups A, B, and C, respectively). Without LT, 81% of the patients died. Seven percent were treated by LT and their survival rate was 60%.

CONCLUSIONS

Clustering revealed that patients with acute liver failure could be classified into three clusters independent of the interval between the onset of disease symptoms and the development of encephalopathy. This technique may be useful, since the outcomes of the patients differed markedly among the clusters.

An eight-year snapshot of geospatial cancer research (2002-2009): clinico-epidemiological and methodological findings and trends

Geographic information systems (GIS) offer a very rich toolbox of methods and technologies, and powerful research tools that extend far beyond the mere production of maps, making it possible to cross-link and study the complex interaction of disease data and factors originating from a wide range of disparate sources. Despite their potential indispensable role in cancer prevention and control programmes, GIS are underrepresented in specialised oncology literature. The latter has provided an impetus for the current review. The review provides an eight-year snapshot of geospatial cancer research in peer-reviewed literature (2002-2009), presenting the clinico-epidemiological and methodological findings and trends in the covered corpus (93 papers). The authors concluded that understanding the relationship between location and cancer/cancer care services can play a crucial role in disease control and prevention, and in better service planning, and appropriate resource utilisation. Nevertheless, there are still barriers that hinder the wide-scale adoption of GIS and related technologies in everyday oncology practice.

Application of geographic information systems and asset mapping to facilitate identification of colorectal cancer screening resources

OBJECTIVE

We sought to identify and map the geographic distribution of available colorectal cancer screening resources; following identification of this priority within a needs assessment of a local community-academic collaborative to reduce cancer health disparities in medically underserved communities.

METHODS

We used geographic information systems (GIS) and asset mapping tools to visually depict resources in the context of geography and a population of interest. We illustrate two examples, offer step-by-step directions for mapping, and discuss the challenges, lessons learned, and future directions for research and practice.

RESULTS

Our positive asset driven, community-based approach illustrated the distribution of existing colonoscopy screening facilities and locations of populations and organizations who might use these resources. A need for additional affordable and accessible colonoscopy resources was identified.

CONCLUSION

These transdisciplinary community mapping efforts highlight the benefit of innovative community-academic partnerships for addressing cancer health disparities by bolstering infrastructure and community capacity-building for increased access to colonoscopies.

A small-area study of environmental risk assessment of outdoor falls

Falls in public places are an issue of great health concern especially for the elderly. Falls among the elderly is also a major health burden in many countries. This study describes a spatial approach to assess environmental causes of outdoor falls using a small urban community in Hong Kong as an example. The method involves collecting data on fall occurrences and mapping their geographic positions to examine circumstances and environmental evidence that contribute to falls. High risk locations or hot spots of falls are identified on the bases of spatial proximity and concentration of falls within a threshold distance by means of kernel smoothing and standard deviational ellipses. This method of geographic aggregation of individual fall incidents for a small-area study yields hot spots of manageable sizes. The spatial clustering approach is effective in two ways. Firstly, it allows visualisation and isolation of fall hot spots to draw focus. Secondly and especially under conditions of resource decline, policy makers are able to target specific locations to examine the underlying causal mechanisms and strategise effective response and preventive measures based on the types of environmental risk factors identified.