Non-seasonal ARIMA modeling of stroke incidence

Abstract

This study employs the ARIMA Box-Jenkins methodology to model monthly stroke incidence data from January 2011 to December 2021. A non-seasonal ARIMA (p,d,q) model is fitted, with ARIMA (1,1,1) identified as the optimal model based on its lowest BIC (=781.5278) and AIC (=772.9022) values compared to alternative models. The Ljung-Box statistic (19.931, df = 22, p = 0.5885 > 0.05) affirms the model's suitability. Projections derived from the fitted model indicate a discernible trend towards increasing stroke incidence. Given the anticipated socioeconomic implications, particularly regarding employment loss, income reduction, and social network disruption among stroke survivors, urgent action from governmental and health authorities is imperative to address the projected challenges effectively.