Assessing climatic drivers and multi-scale forecasting of Japanese encephalitis in endemic regions of eastern India

Abstract

Abstract Japanese Encephalitis (JE) remains a seasonal public health issue in eastern India, but the factors related to climate and how we predict them differ by state. We looked at monthly JE case counts from five states where the disease is common, using correlation analysis, Generalised Additive Models (GAMs), Seasonal Autoregressive Integrated Moving Average (SARIMA) and a wavelet-Seasonal Autoregressive Integrated Moving Average with Exogenous Variables (SARIMAX) framework. The main climatic factors affecting JE were minimum temperature(Tmin), rainfall, and humidity. Warm nights and monsoonal moisture were particularly influential, while maximum temperature(Tmax) had lesser or unpredictable effects. The GAM results indicated significant humidity effects that matched monsoon conditions, with a clear Tmin threshold around 18–22 $$^\circ $$ C and a specific response to rainfall. Although basic SARIMA worked well in Uttar Pradesh, our forecasting tests showed that using wavelet decomposition and including state-specific external factors improved accuracy in other areas. The Wavelet-SARIMAX model produced the smallest errors in Assam, Bihar, Odisha, and West Bengal. In summary, we achieved more reliable short-term JE forecasts by combining key weather factors with specific decomposition methods, which supports better early warning systems in affected regions.