Current approaches to modeling of epidemic process of non-polio Enterovirus infections

Abstract

Aim. To study mathematical models for predicting the incidence of non-polio enterovirus infections (NPEVI) in the Kemerovo Region.Materials and Methods. Here we conducted a retrospective epidemiological study of NPEVI incidence in the Kemerovo region from 2006 to 2021 (n = 2152 cases). Epidemic process was studied using autocorrelation analysis, Fourier analysis, and neural networks using STATISTICA Automated Neural Networks (SANN) tool and StatTech v. 3.0.5.Results. The incidence rates of NPEVI were 9,39 per 100,000 population (2009), 15,78 per 100,000 population (2015) and 8,41 per 100,000 population (2019), exceeding the average median value (2006- 2021) by a factor of 2.4, 4.1, and 2.2, respectively. NPEVI incidence was largely determined by enteroviral meningitis. The majority of cases (89.94%) were registered in children. Notably, standard mathematical models failed to provide an objective analysis of the incidence trend. Autocorrelation analysis found the summer-autumn seasonality (August-October) by evaluating the ratio of actual data to 12-month rolling averages. Modeling of the epidemic process of NPEVI using neural networks highly likely predicted its incidence up to 52 months.Conclusion. The epidemic process of NPEVI in Kemerovo region has been characterized by a low intensity and summer-autumn seasonality. Neural networks are suggested as a promising tool to forecast the incidence of NPEVI.