Susceptibility mapping of visceral leishmaniasis based on fuzzy modelling and group decision-making methods

Abstract

Abstract. Visceral leishmaniasis (VL) is a vector-borne disease, highly influenced by environmental factors, which is an increasing public health problem in Iran, especially in the north-western part of the country. A geographical information system was used to extract data and map environmental variables for all villages in the districts of Kalaybar and Ahar in the province of East Azerbaijan. An attempt to predict VL prevalence based on an analytical hierarchy process (AHP) module combined with ordered weighted averaging (OWA) with fuzzy quantifiers indicated that the south-eastern part of Ahar is particularly prone to high VL prevalence. With the main objective to locate the villages most at risk, the opinions of experts and specialists were generalised into a group decision-making process by means of fuzzy weighting methods and induced OWA. The prediction model was applied throughout the entire study area (even where the disease is prevalent and where data already exist). The predicted data were compared with registered VL incidence records in each area. The results suggest that linguistic fuzzy quantifiers, guided by an AHP-OWA model, are capable of predicting susceptive locations for VL prevalence with an accuracy exceeding 80%. The group decision-making process demonstrated that people in 15 villages live under particularly high risk for VL contagion, i.e. villages where the disease is highly prevalent. The findings of this study are relevant for the planning of effective control strategies for VL in northwest Iran.

Environmental modelling of visceral leishmaniasis by susceptibility-mapping using neural networks: a case study in north-western Iran

“…In order to determine an overall measure of the correlation between the predictive results created by our neural network analysis and those created by other Sheets et al (2010), Rajabi et al (2012) and Fu et al (2013). In this regard, pairwise comparisons were made between the predictive map resulting from the approach of this study (Fig.…”

“…In this regard, pairwise comparisons were made between the predictive map resulting from the approach of this study (Fig. 7) and predictive maps resulting from fuzzy modelling methods created by Rajabi et al (2012), which are almost based on the same map outcomes. The comparison shows that areas with a susceptibility value of >0.75 in maps shown by Rajabi et al (2012) coincide at least 89% with the high and very high susceptible areas according to the RBFLN approach in the northern part of the study area and 81% in the high and very high susceptible areas in the southern part.…”

“…sheepdogs, guard dogs and stray dogs) (Edrissian et al, 1988;Mirsamadi et al, 2003;Moshfe et al, 2008;Salahi-Moghaddam et al, 2010;Rajabi et al, 2012). The Meshkin-Shahr district in the Ardabil province is one of the most important endemic zones for VL in north-western Iran (SalahiMoghaddam et al, 2010) and the Iranian Ministry of Health (MoH) records indicate that the number of VL infections has noticeably increased in this and neighbouring districts during the last decade (Soleimanzadeh et al, 1993;Tamook et al, 2006).…”

“…For example, Bhunia et al (2011) studied the influence of the distribution of inland water bodies on transmission of VL and presence of its dominant vector in a GIS model. In a knowledge-driven approach, Rajabi et al (2012) used multi-criteria decision-making methods together with GIS to identify high-risk areas for VL outbreak in north-western Iran. The techniques applied can be divided into two groups: data-driven and knowledge-driven ones.…”

“…To specify the input data for the prediction model, the association between VL endemicity and various factors in the literature was explored. In this regard, seven items were deduced from literature to be the critical recognition criteria (CRC) for predictive mapping of VL risk-prone areas: temperature, precipitation, proximity to rivers, altitude, presence of health-centres, land cover and presence of nomads (Sudhakar et al, 2006;Bhunia et al, 2010;Salahi-Moghaddam et al, 2010;Rajabi et al, 2012;Tsegaw et al, 2013).…”