Enhancing infectious disease modelling through a Kalman-based (HCRD-R) epidemiological approach. Application to COVID-19 data during the vaccination period

Abstract

The COVID-19 pandemic represented one of the most pressing global health crises of the 21st century, imposing substantial challenges upon national healthcare systems. Efforts to model this disease have become paramount in devising informed strategies for its containment. However, prevailing mathematical methodologies predominantly rely on deterministic models, which often oversimplify the complex dynamics of epidemic spread as they cannot quantify the uncertainty that accompanies this physical phenomenon. Notably, there is a dearth of models that specifically address the nuanced dynamics of hospitalized and intensive care unit (ICU) admitted cases, crucial for healthcare system planning. This paper, introduces a novel HCRD-R model, tailored to focus exclusively on hospitalized, ICU admitted, recovered and deceased cases in combination with a Kalman filter for predictive analysis within hospital settings. This focused approach aims to optimize resources while ensuring reliability in modeling and efficiently predicting critical epidemiological scenarios. The reliability of the model is examined on daily recording of COVID-19 in Italy, over a time interval of 106 days which takes place after the onset of the vaccination period. Several comparisons with deterministic and stochastic alternatives are also explored, validating the efficiency of the proposed epidemiological model. Finally, through our analysis we underscore the enhanced efficacy of stochastic epidemiological models over their deterministic counterparts, rendering them suitable for assessing the severity of existing and new epidemic outbreaks.