Publication

Abstract : Epidemics present significant global health challenges, requiring proactive, data-driven approaches for effective management and mitigation. Existing methodologies for epidemic forecasting often encounter limitations such as reliance on static datasets, inadequate integration of multivariate factors, and a lack of real-time visualization capabilities. These constraints hinder timely decision-making and resource allocation in response to emerging outbreaks. This research introduces a comprehensive framework for real-time epidemic prediction and mapping, integrating machine learning, time-series modelling, and geospatial visualization tools. The methodology combines Random Forest and Gradient Boosting models for predictive accuracy, ARIMA and Vector Autoregression (VAR) for time-series trend analysis, and Folium for interactive geospatial visualization. These components are applied to analyse the spread of 15 diseases across Indian states, leveraging epidemiological, environmental, and demographic data to identify high-risk regions and forecast outbreak dynamics. Our framework addresses the limitations of traditional approaches by incorporating multivariate analysis, enabling real-time visualizations, and achieving high prediction accuracy. The results demonstrate improved forecasts for diseases like Dengue Fever, Hepatitis B, and Leptospirosis, facilitating targeted public health interventions. This work advances the state-of-the-art in epidemic modelling by providing actionable insights, improving response times, and enhancing resource optimization for epidemic control.