Research Report.

Executive summary

Dengue incidence has risen significantly in Nepal in recent years, posing an escalating public health challenge. Understanding the spatial distribution of cases and the environmental drivers that shape transmission is essential for guiding effective control measures and targeted intervention strategies. In this study, we utilized publicly available environmental and socioeconomic geospatial covariates sourced from multiple geoportals, alongside monthly dengue case data extracted from situation reports published by the Epidemiology and Disease Control Division (EDCD), Government of Nepal.
We first characterized the spatiotemporal dynamics of the outbreak and mapped spatial patterns using Global Moran’s I and Local Indicators of Spatial Association (LISA) to identify statistically significant clustering. Subsequently, we applied a Random Forest (RF) machine learning model to predict case distribution and assess associations between dengue incidence and geographic covariates.
The results revealed heterogeneous spatial distribution of dengue cases across the study years, with significant clusters emerging in different regions of the country. Monthly patterns showed sporadic and spatially unstructured distribution until June, followed by a sharp rise from July, peaking in September, and declining thereafter—producing statistically significant monthly clusters. The RF model indicated that districtlevel spatial variation was primarily driven by urban related environmental factors, including road density, nighttime lights (NTL), and the Human Footprint Index, along with precipitation. In contrast, temperature played a comparatively minor and variable role across outbreak years.
These findings provide important insights for designing spatially targeted dengue control and intervention strategies in Nepal.