Geographically weighted regression of land cover determinants of Plasmodium falciparum transmission in the Ashanti Region of Ghana

Managing vector-borne diseases in a geoAI-enabled society. Malaria as an example

More than 17 % of all infectious diseases are caused by vector-borne diseases resulting in more than 1 billion cases and over 1 million deaths each year. Of these malaria continues to be a global burden in over eighty countries. As societies become more digitalised, the availability of geospatially enabled health and disease information will become more abundant. With this, the ability to assess health and disease risks in real-time will become a reality. The purpose of this study was to examine how geographic information, geospatial technologies and spatial data science are being used to reduce the burden of vector-borne diseases such as malaria and explore the opportunities that lie ahead with GeoAI and other geospatial technology advancements. Malaria is a dynamic and complex system and as such a range of data and approaches are needed to tackle different parts of the malaria cycle at different local and global scales. Geospatial technologies provide an integrated framework vital for monitoring, analysing and managing vector-borne diseases. GeoAI and technological advancements are useful for enhancing real-time assessments, accelerating the decision making process and spatial targeting of interventions. Training is needed to enhance the use of geospatial information for the management of vector-borne diseases.

Integrating Spatial Modelling and Space-Time Pattern Mining Analytics for Vector Disease-Related Health Perspectives: A Case of Dengue Fever in Pakistan

The spatial–temporal assessment of vector diseases is imperative to design effective action plans and establish preventive strategies. Therefore, such assessments have potential public health planning-related implications. In this context, we here propose an integrated spatial disease evaluation (I-SpaDE) framework. The I-SpaDE integrates various techniques such as the Kernel Density Estimation, the Optimized Hot Spot Analysis, space–time assessment and prediction, and the Geographically Weighted Regression (GWR). It makes it possible to systematically assess the disease concentrations, patterns/trends, clustering, prediction dynamics, and spatially varying relationships between disease and different associated factors. To demonstrate the applicability and effectiveness of the I-SpaDE, we apply it in the second largest city of Pakistan, namely Lahore, using Dengue Fever (DF) during 2007–2016 as an example vector disease. The most significant clustering is evident during the years 2007–2008, 2010–2011, 2013, and 2016. Mostly, the clusters are found within the city’s central functional area. The prediction analysis shows an inclination of DF distribution from less to more urbanized areas. The results from the GWR show that among various socio-ecological factors, the temperature is the most significantly associated with the DF followed by vegetation and built-up area. While the results are important to understand the DF situation in the study area and have useful implications for public health planning, the proposed framework is flexible, replicable, and robust to be utilized in other similar regions, particularly in developing countries in the tropics and sub-tropics.

Spatial connectivity in mosquito-borne disease models: a systematic review of methods and assumptions

Spatial connectivity plays an important role in mosquito-borne disease transmission. Connectivity can arise for many reasons, including shared environments, vector ecology and human movement. This systematic review synthesizes the spatial methods used to model mosquito-borne diseases, their spatial connectivity assumptions and the data used to inform spatial model components. We identified 248 papers eligible for inclusion. Most used statistical models (84.2%), although mechanistic are increasingly used. We identified 17 spatial models which used one of four methods (spatial covariates, local regression, random effects/fields and movement matrices). Over 80% of studies assumed that connectivity was distance-based despite this approach ignoring distant connections and potentially oversimplifying the process of transmission. Studies were more likely to assume connectivity was driven by human movement if the disease was transmitted by an Aedes mosquito. Connectivity arising from human movement was more commonly assumed in studies using a mechanistic model, likely influenced by a lack of statistical models able to account for these connections. Although models have been increasing in complexity, it is important to select the most appropriate, parsimonious model available based on the research question, disease transmission process, the spatial scale and availability of data, and the way spatial connectivity is assumed to occur.

Geo-visual integration of health outcomes and risk factors using excess risk and conditioned choropleth maps: a case study of malaria incidence and sociodemographic determinants in Ghana

Background

Recently, exploratory spatial data analysis is for problem solving, hypothesis generation and knowledge construction. Unless geographically weighted regression, sophisticated spatial regression models best control spatial heterogeneity in outcomes and the associated risk factors but cannot visually display and identify areas of the significant associations. The under-utilised excess risk maps (ERMs) and conditioned choropleth maps (CCMs) are useful to address this issue and simplify epidemiological information to public health stakeholders without much statistical backgrounds. Using malaria and sociodemographic determinants in Ghana as case study, this paper applied ERM and CCM techniques for identification of areas at elevated risk of disease-risk factor co-location.

Method

We computed and smoothed mean district-specific malaria incidences for the period 2010 to 2014 as a function of sociodemographic determinants. The spatial distribution of malaria was investigated through global and local spatial autocorrelations, and the association with sociodemographic risk factors evaluated with bivariate correlations. ERMs and CCMs were produced for the statistically significant risk factors.

Results

The incidence of malaria increased over time with cluster locations detected, predominantly at the northern parts but later few spread to the middle parts of the country. Our results suggested that with respect to sociodemographic determinants, district variations in malaria rates might be explained by inequalities in seven sociodemographics, including an unexpected significant negative association with non-religious affiliation. The sociodemographics had positive spatial autocorrelations, exhibited statistically significant interactions and the strongest was observed in urbanisation-basic education correlation (p< 0.01, r = +0.969). The ERMs and CCMs specifically identified locations with lower or higher than expected rates with respect to particular risk factor(s) where improving risk factor(s) such as employment-to-population ratio in rural areas, basic education could have cascade effects to reduce the expected malaria incidence in endemic areas.

Conclusion

Ghana remains malaria hyperendemic region with district-level spatial heterogeneity. Significant association between malaria and sociodemographics was detected and the ERMs and CCMs geo-visually pinpointed locations of these significant associations. To complement sophisticated spatial regression models, the easily interpretable ERMs and CCMs could be used to specify where disease-risk factor associations are significant, simplifying complex spatial epidemiological information for efficient public health administration.

Modeling the spatially varying risk factors of dengue fever in Jhapa district, Nepal, using the semi-parametric geographically weighted regression model

Dengue fever is expanding rapidly in many tropical and subtropical countries since the last few decades. However, due to limited research, little is known about the spatial patterns and associated risk factors on a local scale particularly in the newly emerged areas. In this study, we explored spatial patterns and evaluated associated potential environmental and socioeconomic risk factors in the distribution of dengue fever incidence in Jhapa district, Nepal. Global and local Moran's I were used to assess global and local clustering patterns of the disease. The ordinary least square (OLS), geographically weighted regression (GWR), and semi-parametric geographically weighted regression (s-GWR) models were compared to describe spatial relationship of potential environmental and socioeconomic risk factors with dengue incidence. Our result revealed heterogeneous and highly clustered distribution of dengue incidence in Jhapa district during the study period. The s-GWR model best explained the spatial association of potential risk factors with dengue incidence and was used to produce the predictive map. The statistical relationship between dengue incidence and proportion of urban area, proximity to road, and population density varied significantly among the wards while the associations of land surface temperature (LST) and normalized difference vegetation index (NDVI) remained constant spatially showing importance of mixed geographical modeling approach (s-GWR) in the spatial distribution of dengue fever. This finding could be used in the formulation and execution of evidence-based dengue control and management program to allocate scare resources locally.

Applications of Space Technologies to Global Health: Scoping Review

Background

Space technology has an impact on many domains of activity on earth, including in the field of global health. With the recent adoption of the United Nations’ Sustainable Development Goals that highlight the need for strengthening partnerships in different domains, it is useful to better characterize the relationship between space technology and global health.

Objective

The aim of this study was to identify the applications of space technologies to global health, the key stakeholders in the field, as well as gaps and challenges.

Methods

We used a scoping review methodology, including a literature review and the involvement of stakeholders, via a brief self-administered, open-response questionnaire. A distinct search on several search engines was conducted for each of the four key technological domains that were previously identified by the UN Office for Outer Space Affairs’ Expert Group on Space and Global Health (Domain A: remote sensing; Domain B: global navigation satellite systems; Domain C: satellite communication; and Domain D: human space flight). Themes in which space technologies are of benefit to global health were extracted. Key stakeholders, as well as gaps, challenges, and perspectives were identified.

Results

A total of 222 sources were included for Domain A, 82 sources for Domain B, 144 sources for Domain C, and 31 sources for Domain D. A total of 3 questionnaires out of 16 sent were answered. Global navigation satellite systems and geographic information systems are used for the study and forecasting of communicable and noncommunicable diseases; satellite communication and global navigation satellite systems for disaster response; satellite communication for telemedicine and tele-education; and global navigation satellite systems for autonomy improvement, access to health care, as well as for safe and efficient transportation. Various health research and technologies developed for inhabited space flights have been adapted for terrestrial use.

Conclusions

Although numerous examples of space technology applications to global health exist, improved awareness, training, and collaboration of the research community is needed.

Micro-spatial distribution of malaria cases and control strategies at ward level in Gwanda district, Matabeleland South, Zimbabwe

Background

Although there has been a decline in the number of malaria cases in Zimbabwe since 2010, the disease remains the biggest public health threat in the country. Gwanda district, located in Matabeleland South Province of Zimbabwe has progressed to the malaria pre-elimination phase. The aim of this study was to determine the spatial distribution of malaria incidence at ward level for improving the planning and implementation of malaria elimination in the district.

Methods

The Poisson purely spatial model was used to detect malaria clusters and their properties, including relative risk and significance levels at ward level. The geographically weighted Poisson regression (GWPR) model was used to explore the potential role and significance of environmental variables [rainfall, minimum and maximum temperature, altitude, Enhanced Vegetation Index (EVI), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), rural/urban] and malaria control strategies [indoor residual spraying (IRS) and long-lasting insecticide-treated nets (LLINs)] on the spatial patterns of malaria incidence at ward level.

Results

Two significant clusters (p < 0.05) of malaria cases were identified: (1) ward 24 south of Gwanda district and (2) ward 9 in the urban municipality, with relative risks of 5.583 and 4.316, respectively. The semiparametric-GWPR model with both local and global variables had higher performance based on AICc (70.882) compared to global regression (74.390) and GWPR which assumed that all variables varied locally (73.364). The semiparametric-GWPR captured the spatially non-stationary relationship between malaria cases and minimum temperature, NDVI, NDWI, and altitude at the ward level. The influence of LLINs, IRS and rural or urban did not vary and remained in the model as global terms. NDWI (positive coefficients) and NDVI (range from negative to positive coefficients) showed significant association with malaria cases in some of the wards. The IRS had a protection effect on malaria incidence as expected.

Conclusions

Malaria incidence is heterogeneous even in low-transmission zones including those in pre-elimination phase. The relationship between malaria cases and NDWI, NDVI, altitude, and minimum temperature may vary at local level. The results of this study can be used in planning and implementation of malaria control strategies at district and ward levels.

Remotely Sensed Environmental Conditions and Malaria Mortality in Three Malaria Endemic Regions in Western Kenya

Background

Malaria is an important cause of morbidity and mortality in malaria endemic countries. The malaria mosquito vectors depend on environmental conditions, such as temperature and rainfall, for reproduction and survival. To investigate the potential for weather driven early warning systems to prevent disease occurrence, the disease relationship to weather conditions need to be carefully investigated. Where meteorological observations are scarce, satellite derived products provide new opportunities to study the disease patterns depending on remotely sensed variables. In this study, we explored the lagged association of Normalized Difference Vegetation Index (NVDI), day Land Surface Temperature (LST) and precipitation on malaria mortality in three areas in Western Kenya.

Methodology and Findings

The lagged effect of each environmental variable on weekly malaria mortality was modeled using a Distributed Lag Non Linear Modeling approach. For each variable we constructed a natural spline basis with 3 degrees of freedom for both the lag dimension and the variable. Lag periods up to 12 weeks were considered. The effect of day LST varied between the areas with longer lags. In all the three areas, malaria mortality was associated with precipitation. The risk increased with increasing weekly total precipitation above 20 mm and peaking at 80 mm. The NDVI threshold for increased mortality risk was between 0.3 and 0.4 at shorter lags.

Conclusion

This study identified lag patterns and association of remote- sensing environmental factors and malaria mortality in three malaria endemic regions in Western Kenya. Our results show that rainfall has the most consistent predictive pattern to malaria transmission in the endemic study area. Results highlight a potential for development of locally based early warning forecasts that could potentially reduce the disease burden by enabling timely control actions.

Seasonal and Geographic Variation of Pediatric Malaria in Burundi: 2011 to 2012

We analyzed hospitalization records from 2011 to 2012 to examine the spatial patterns of pediatric malaria in Burundi. Malaria case data for those below the age of five years were categorized according to the four principal seasons of Burundi, which are two rainy seasons (February to May; September to November) and two dry seasons (June to August; December to January). The Getis-Ord Gi* statistic was used to examine seasonal spatial patterns of pediatric malaria, whereas geographically weighted regression (GWR) were used to examine the potential role of environmental variables on the spatial patterns of cases. There were a total of 19,890 pediatric malaria cases reported during the study period. The incidence among males was higher than that among females; and it was higher in rural districts. The seasonal incidence peaks occurred in the northern half of the country during the wet season while during the dry season, incidence was higher in southern Burundi. Elevation played a greater role in explaining variance in the prevalence of pediatric malaria during seasonal peaks than rainfall. The counterintuitive finding in northern Burundi confirms previous findings and suggests other factors (e.g., land cover/land use) facilitate the persistence of the mosquito population in the highlands of Africa.

Spatially variable risk factors for malaria in a geographically heterogeneous landscape, western Kenya: an explorative study

Background

Large reductions in malaria transmission and mortality have been achieved over the last decade, and this has mainly been attributed to the scale-up of long-lasting insecticidal bed nets and indoor residual spraying with insecticides. Despite these gains considerable residual, spatially heterogeneous, transmission remains. To reduce transmission in these foci, researchers need to consider the local demographical, environmental and social context, and design an appropriate set of interventions. Exploring spatially variable risk factors for malaria can give insight into which human and environmental characteristics play important roles in sustaining malaria transmission.

Methods

On Rusinga Island, western Kenya, malaria infection was tested by rapid diagnostic tests during two cross-sectional surveys conducted 3 months apart in 3632 individuals from 790 households. For all households demographic data were collected by means of questionnaires. Environmental variables were derived using Quickbird satellite images. Analyses were performed on 81 project clusters constructed by a traveling salesman algorithm, each containing 50–51 households. A standard linear regression model was fitted containing multiple variables to determine how much of the spatial variation in malaria prevalence could be explained by the demographic and environmental data. Subsequently, a geographically-weighted regression (GWR) was performed assuming non-stationarity of risk factors. Special attention was taken to investigate the effect of residual spatial autocorrelation and local multicollinearity.

Results

Combining the data from both surveys, overall malaria prevalence was 24 %. Scan statistics revealed two clusters which had significantly elevated numbers of malaria cases compared to the background prevalence across the rest of the study area. A multivariable linear model including environmental and household factors revealed that higher socioeconomic status, outdoor occupation and population density were associated with increased malaria risk. The local GWR model improved the model fit considerably and the relationship of malaria with risk factors was found to vary spatially over the island; in different areas of the island socio-economic status, outdoor occupation and population density were found to be positively or negatively associated with malaria prevalence.

Discussion

Identification of risk factors for malaria that vary geographically can provide insight into the local epidemiology of malaria. Examining spatially variable relationships can be a helpful tool in exploring which set of targeted interventions could locally be implemented. Supplementary malaria control may be directed at areas, which are identified as at risk. For instance, areas with many people that work outdoors at night may need more focus in terms of vector control.

Trial registration: Trialregister.nl NTR3496—SolarMal, registered on 20 June 2012

Application of geographically-weighted regression analysis to assess risk factors for malaria hotspots in Keur Soce health and demographic surveillance site

Background

In Senegal, considerable efforts have been made to reduce malaria morbidity and mortality during the last decade. This resulted in a marked decrease of malaria cases. With the decline of malaria cases, transmission has become sparse in most Senegalese health districts. This study investigated malaria hotspots in Keur Soce sites by using geographically-weighted regression. Because of the occurrence of hotspots, spatial modelling of malaria cases could have a considerable effect in disease surveillance.

Methods

This study explored and analysed the spatial relationships between malaria occurrence and socio-economic and environmental factors in small communities in Keur Soce, Senegal, using 6 months passive surveillance. Geographically-weighted regression was used to explore the spatial variability of relationships between malaria incidence or persistence and the selected socio-economic, and human predictors. A model comparison of between ordinary least square and geographically-weighted regression was also explored. Vector dataset (spatial) of the study area by village levels and statistical data (non-spatial) on malaria confirmed cases, socio-economic status (bed net use), population data (size of the household) and environmental factors (temperature, rain fall) were used in this exploratory analysis. ArcMap 10.2 and Stata 11 were used to perform malaria hotspots analysis.

Results

From Jun to December, a total of 408 confirmed malaria cases were notified. The explanatory variables-household size, housing materials, sleeping rooms, sheep and distance to breeding site returned significant t values of −0.25, 2.3, 4.39, 1.25 and 2.36, respectively. The OLS global model revealed that it explained about 70 % (adjusted R² = 0.70) of the variation in malaria occurrence with AIC = 756.23. The geographically-weighted regression of malaria hotspots resulted in coefficient intercept ranging from 1.89 to 6.22 with a median of 3.5. Large positive values are distributed mainly in the southeast of the district where hotspots are more accurate while low values are mainly found in the centre and in the north.

Conclusion

Geographically-weighted regression and OLS showed important risks factors of malaria hotspots in Keur Soce. The outputs of such models can be a useful tool to understand occurrence of malaria hotspots in Senegal. An understanding of geographical variation and determination of the core areas of the disease may provide an explanation regarding possible proximal and distal contributors to malaria elimination in Senegal.