Spatiotemporal modelling and composite indices for pandemic preparedness: A novel geospatial-temporal approach (In revision)

Effective pandemic preparedness requires robust analytical frameworks that capture the spatiotemporal evolution of infectious diseases. We propose a novel geographically and temporally weighted random forest (GTWRF) model designed to enhance predictive accuracy in epidemic modelling by integrating spatial and temporal dependencies within an adaptive weighting framework. GTWRF incorporates one novel and two existing spatiotemporal distance (STD) functions into Gaussian distance-decay weight function, dynamically assigning higher weights to observations closer in space and time. This model was applied to COVID-19 incidence across US counties, addressing key limitations of traditional geographically weighted random forest (GWRF) models. By leveraging local epidemiological, demographic (race-ethnic), socioeconomic (adversity), and environmental composite indicators, GTWRF captures dynamic regional vulnerabilities during three critical pandemic waves. Performance evaluations demonstrate that the novel STD-based GTWRF model significantly outperforms standard GWRF, reducing out-of-bag mean absolute error and root mean square error. Findings indicate that epidemiological indicators consistently drive COVID-19 incidence, while demographic (race-ethnic) factors dominate early (first wave) outbreaks, environmental factors peak in influence during the second, and socioeconomic adversity becomes more prominent in the third wave. Regional vulnerabilities, particularly in the South and West, persisted, with multiple indicators contributing significantly to COVID-19 incidence. The superior predictive capability of GTWRF in modelling spatiotemporal disease spread underscores its potential for real-time outbreak monitoring, resource allocation and targeted intervention strategies. This study provides valuable insights for public health policymakers seeking to enhance pandemic preparedness strategies and mitigate structural vulnerabilities in future outbreaks.