No abstract

Modelling infectious diseases is a complex and multi‐disciplinary problem that necessitates the combined use of multicriteria decision analysis (MCDA) and machine learning (ML) in a spatial framework. This research attempts to demonstrate the extensive applications of MCDA in the field of public health and to illustrate its utility with the combined use of spatial models and machine learning. The study investigates the risk factors for communicable diseases with a focus on vector‐borne infectious diseases, such as West Nile Virus (WNV), malaria, dengue, etc. It aims to quantify vector‐borne disease risk by examining the geographic contextual effects of socio‐economic, climatic, and environmental factors using the objective‐weighting technique adopted from MCDA and machine learning in a geographic information systems (GIS) framework. The authors attempted to minimize subjective bias from the decision space by utilizing an objective‐weighted technique to quantify the risk. The study adopted Shannon's entropy to derive weights for each factor and its classes. The derived weighted layers are fed to an artificial neural network to obtain a final map of risk susceptibility. This final risk map allows policymakers to examine vulnerable areas and identify the factors pivotal to the contribution of risk. Findings show the traffic volume as the most influential variable, and terrain slope as the least one in the disease spread for the study area. The risk appears to be concentrated and distributed along vegetation, wetlands, and around water bodies. The results produced by ensemble learning show great promise with more than 94% accuracy. The accuracy of the results was determined by the confusion matrix and the kappa index of agreement (KIA). The vector control programmes need to adapt to better manage the dynamic changes in patterns involving vector‐borne infectious diseases.

Purpose-The aim of this paper is to clarify the spatial multi-criteria workflow for stakeholders and decision makers, for which feedback rankings are vital to the success of the transportation planning. Design/methodology/approach-The experimental approach was designed to integrate in a novel fashion both analytical hierarchy process (AHP) and multi-criteria decision making (MCDM) within a geospatial information system (GIS) framework to deliver visual and objective tabular results useful to estimate environmental costs of the alignments generated. The method enables ranking, prioritization, selection, and refinement of preferred alternatives. The Interstate-269, the newly planned bypass of Memphis-TN, for which a recent environmental impact study (EIS) was completed, was selected as the experiment test-bed. Findings-The results indicate that the approach can automate the delivery of feasible alignments that closely approximate those generated by traditional approaches. Furthermore, via integration of local planning and ancillary spatial data, the method provided alignment results that avoided areas where local opposition was noted in the EIS. This enhanced method based on remote sensing and spatial information technologies delivers low or high-predicted environmental costs per feature criteria and cumulative predicted costs while preserving local values and plans. Practical implications-The method is highly transferable and limited solely by the availability of sources of geospatial data and coordination with stakeholders. The approach was implemented to derive results similar to traditional approaches with benefits in time, costs, and quality of solutions. Originality/value-A novel adaptation of MCDM and AHD within a spatial decision-making framework is presented. The paper suggests a clarification of multi-criteria workflow to design and select least-environmental-cost corridors. The case study application provides a starting point to develop practical tools that delivers environmental benefits through a collaborative process capturing stakeholder values and decision maker opinions.

A spatial interaction model to predict anthropogenically-initiated accidental and incendiary wildfire ignition probability is developed using fluid flow analogies for human movement patterns. Urban areas with large populations are identified as the sites of global influencing factors, and are modeled as the gravity term. The transportation corridors are identified as local influencing factors, and are modeled using fluid flow analogy as diffusion and convection terms. The model is implemented in Arc-GIS, and applied for the prediction of wildfire hazard distribution in southeastern Mississippi. The model shows 87 % correlation with historic data in the winter season, whereas the previously developed gravity model shows only 75 % correlation. The normalized error for convection-diffusion model predictions is about 5 % in the winter season, whereas the gravity model shows an error of 7 %. The proposed model is robust as it couples a multi-criteria behavioral pattern within a single dynamic equation to enhance predictive capability. At the same time, the proposed model is more costly than the gravity model as it requires evaluation of distance from intermodal transportation corridors, transportation corridor density, and traffic volume maps. Nonetheless, the model is developed in a modular fashion, such that either global or local terms can be neglected if required.