Web-based GIS for spatial pattern detection: application to malaria incidence in Vietnam

Web-based information dissemination

The fast development of Internet and Web-based services helps health practitioners and communities to effectively and efficiently share health information and disseminate risks of diseases (Highfield et al. 2011), even though the services were limited in simple display and query tasks (Longley 2005). Recently, burgeoning efforts have been made to develop more active and dynamic systems and to make Web-based GIS more interactive for end users, such as statistical packages in health data processing and health information dissemination. Technically, geo-technology has enabled the health data sharing and analysis, but socially, it results in a major challenge in protecting patient geo-privacy while allowing analyses of spatial data (Hampton and Fitch 2010). The larger number of data sets become public, the more health information is exposed to communities. Recently, the opening up of health data leads to the development of filter techniques to mask out locations of patient’s location information to preserve confidence. Although enormous efforts have been made, the instilment of sensitivity to geo-privacy and disclosure risk have been unsuccessful (Kounadi and Leitner 2014), particularly in the era of internet.

There is significant demand to exploit current geospatial information infrastructures in health applications (Granell et al. 2014). Researches have focused on the area of Web-based health information and more researchers (academic and non-academic) are using the web for data sharing, research and planning purposes (Lee 2010). Gao (2009) also emphasized the importance of comprehensive representation of heath information to users. Highfield et al. (2011) worked on interactive web-based mapping that provides health data easily accessible for a wide variety of research audiences. The authors urged that the system need to be as interactive as possible, allowing users to dynamically define their research parameters rather than being forced to accept pre-set queries. Boulos (2011) pointed out the application of open map services as Google, Openstreetmap to collect data from communities (crowd-sourcing) and discussed the benefits to get local communities involved in health surveillance. Delmelle and Zhu (2014) recently mentioned the state-of-the art web-based applications in epidemiology and produced an analytical module to map the outbreaks of dengue fever in Colombia, by using accelerated Kernel Density Estimation. Almost all mentioned researches assessed the available Web-based package for health information dissemination, but there are restrictions on spatial analytical capabilities (Delmelle and Zhu 2014) and lack of web-based pattern mapping tool, fostering multi-level participation to support identification of diseases clustering. Even though many web-based health applications dynamically generate view-only maps or interactive maps (Gao 2009) and allow users to study spatio-temporal distributions and evolution of patients (Hammond and Barzyk 2008). However, web-based health pattern detection tools were limited to some specialized researches, using specialized tools in particular a Web-based system (Mills 2008).

This paper focuses on creating web-based spatial processing tool that can be used to measure spatial patterns of the incidence over health data in Vietnam (in this case malaria samples were used). We develop three tools, namely nearest neighbour analysis, K-function, spatial autocorrelation to test the significances of clustering of point distribution. All points are treated in the distribution as if they are all the same they do not distinguish points by their attributes in the first two tools. The third uses aggregated data and takes into consideration both location and locational characteristics. These pattern detection tools can provide preliminary overview on how diseases clustering magnitude be found and will be useful for health practitioners.

Malaria prevalence and pattern

The occurrence of malaria is closely related to environmental, climatic (Odongo-Aginya et al. 2005) and socio-economic conditions (Kienberger and Hagenlocher 2014). Malaria is still a major health problem in the world and Vietnam in particular. Though its prevalence has been reduced, there is great risk of diseases re-emergence to susceptible communities which live in the remote regions, especially those living in border areas between Vietnam and Laos and Cambodia (Hoang 2014) and in highland regions of Vietnam. Most of researches on malaria in Vietnam have focused on malaria epidemiology and achieved many results in prevention of malaria. But the disease is not adequately addressed due to lack of research on disease management, monitoring, detection and early treatment of malaria patients among local communities. A mechanism to share health information throughout communities is an important route to reduce national-wide public health issues (Hoang 2014). This research used malaria distribution as example data for spatial pattern detection tools.

Web-based spatial analysis tool

Spatial-analytical tools have increasingly been used with epidemiologic data to address the complex spatial nature of exposure-disease relationships. Management of the data is performed using geographic information systems (GIS) and database management systems, and is of relevance throughout the various phases of spatial data analysis. Many applications and GIS platforms demonstrate the application of spatial analysis in epidemiologic research, introducing analytic approaches which integrate GIS and standard statistical software tools to enhance epidemiologic assessments. They provide tools in visualization, exploration and modelling (Abdullahi et al. 2010; Dominkovics et al. 2011; Gao 2009). They cover approaches for identifying spatial clustering of disease, spatial interpolation and imputation methods, geographically weighted and land use regression, and basic spatial statistics useful for studying the spatial distribution of exposure and disease. Even though many Web-based health applications dynamically generate view-only maps or interactive maps, background processing and metadata are often missing (Gao et al. 2009).

In the literature review, several projects have focused on the use of representational state transfer (REST) architectural style and have proved it as an effective and efficient style in Web applications (Arakawa and Kido 2010; Dominkovics et al. 2011; Flemons and Guralnick 2007; Highfield et al. 2011; Kvilekval 2010). In general, clients interact directly with the exposed resources through a uniform interface, materialized basically through the combination of HTTP (hypertext transfer protocol), URI (uniform resource identifier) and standard formats such as HTML (hypertext markup language) and XML (extensible markup language).

System design

Web-based system provides fundamental tools for visualization, query and for spatial analysis in some cases. Users can interactively define study area and test whether observed distribution and known pattern are statistically different. It also provides an Information communication page. The communication component enables sharing of local/scientific knowledge and guidance of decision makers. The information system will be getting more and more enriched with analysed data by scientists, information verification and feedbacks from the users.

Figure 2 shows conceptual epidemiology data analysis capabilities in GIS, in which the authors propose an open-source framework to deal with pattern visualization and pattern analysis. The criteria for selection of systems embrace many aspects, such as Free and Open source software (FOSS), leverage open standards, hierarchical layering, spatial functionality, query layout, documentation and loading time. Among the two most popular open-source mapping servers UMN Mapserver and Geoserver, Geoserver was selected because of its extension that serves Web Processing Service. From the client side, there are many available frameworks that are sophisticated enough to compare with each other. This research employed GXP library, Openlayers, Extjs and GeoExt. It offers n-tier architecture consisting of: data layer, visualization and processing layer, and presentation layers.

Density visualization

Based on test data, this tool performs a density measurement and visualization that provides a preliminary density visualization of diseases. Results depend on zoom level to show different magnitude of clustering. The processing service performs a heat map web processing service (WPS) service to compute a spatial density map to detect high concentration area of malaria cases. Malaria reported cases were stored in PostGIS and served by Geoserver. When the user makes a change to zoom level, request is sent back to server to recalculate the spatial map with changes in values according to the zoom level (Fig. 4).

Nearest neighbour analysis

The test is based on comparing the observed average distances between nearest neighbouring points and those of a known pattern (Wong 2005) and is used to estimate the spatial proximity among events. User can draw a box around cluster of points to define boundary of study area, then select level of significance. The tool will calculate R scale, standard error, Z score through calling PostGIS function and compare it to significant interval to check whether the observed distribution is significant different from random pattern. The example showed that Z score is between (−1.96SE– >1.96SE) at 5 % significant level that means observed and random pattern are not statistically different (Fig. 5).

With all measured parameters, this tool will calculate D statistic value and D_α for K–S test, by calling PostGIS function. The example showed that D statistics is greater than D_α = 5 % that means two distributions are significantly different in a statistical sense (Fig. 5a). As in the figure, we defined an area in Lam Dong province and the result was considered to be a non-random pattern at the α = 0.05 level.

K-function

One of the limitations of the nearest neighbour distance method is that it uses only the nearest distance, as such it only considers only the shortest scales of variation. The K-function provides an estimate of spatial clustering over a wider range of scales. This higher-order analysis based on all the distances between events in the study area and assumes isotropy over the region. The tool starts with drawing a box as a bounding box for malaria point selection, and continues selecting a distance or spatial lag. For each point in the selected area, a buffer with radius equal to spatial lag is drawn and sum of numbers of points within each buffer is calculated.

Spatial autocorrelation

Neither the nearest neighbour, nor the K-function take point’s attributes into consideration. Different geographic locations rarely have identical characteristics. Because, this is aggregated data, in which one location might contain information on malaria incidences of an area (in this case, areas can be at commune or district level). The number of cases in an area can be used as weight of points in spatial analysis. Spatial autocorrelation is used in considering the effects of both distances between points and their attributes. With spatial autocorrelation, Moran’s I index is used to measure the proximity of location and similarity of the locations. In this study we develop a PostGIS tool to measure Moran’s I, in which data can be either continuous or count data.

Expected E(I) is measured as E(I) = −1/(number of points −1). Cluster pattern, Random pattern and dispersed patterns are detected when calculated Moran’s I value is higher than E(I), equal to E(i) or small than E(I) respectively. In this example, with the calculated score was higher than expected E(I) of −0.185, the malaria points showed a slightly statistically significant departure from a random pattern. This was demonstrated by high score of 0.0213 [a bit higher than random E(I)].

The selection of proper geographic scale for clustering analysis affects the analysis results. Objects may be detected as clustered, dispersed or random depending on the actual scale (in this case zoom scale) we define before proceeding the analysis. In fact, data was geocoded using reported addresses. However, there were missing in documenting correct addresses, then they were aggregated into communes or even into districts. The effect is known as Modifiable aerial unit problem that can radically affect the analysis results.

Indeed, the examples describe three spatial detection tools that compare the observed distribution to random pattern of points, or more specifically pattern of diseases. These functionalities are powerful testing tool to convey message to and to support decision making. By zooming in and out, in combination with clustering tool, users define the boundary of study area and understand changes of analytical results that vary with zoom scale. Nearest neighbour can be used for the global test of clustering, it measures distances between points and compare these distances to known pattern. It has been extended (in comparison to Quadrat statistics) to accommodate second, third, and higher order neighbourhood definitions. The outcome results of this analysis vary depending on spatial scale, because some look clustered in small scale but seem dispersed at large scale. If boundary box size is too small, each polygon may contain a couple of points. On the other hand, if it is too large, each polygon contains many points. This measure does not take into account that different points may be different in how points are represented (due to their differences in characteristics).

The second analysis takes into consideration distances among points and results in plot, describing areas where patterns are clustered or dispersed. All spatial processing of three analyses are done in PostGIS and measured statistics are sent back to client. The third one measures the proximity of location in considering effects of locational characteristics. However, this test considers the population at risk is evenly distributed within the study area and correlation in all directions are to be the same (Moran 1950). This is a drawback of the Moran’s I test. Through the Web based pattern detection tools, end users can interactively define (draw) area for measurement. The tools are three straightforward measurement of clustering and diversion that provide preliminary assessment to disease distribution. More spatial tests should be further incorporated.