感染症と予防医学ジャーナル
オープンアクセス

概要

A Prediction Model for COVID-19 Prevalence Based on Demographic and Healthcare Parameters in Iran

Parimah Emadi Safavi, Karim Rahimian, Alireza Doustmohammadi, Mahla Safari Dastjerdei, Ahmadreza Rasouli, Javad Zahiri

Background: Coronavirus Disease 2019 (COVID-19) pandemic has become the greatest threat to global health in only a matter of months. Iran struggling with COVID-19 coincidence with Nowruz vacations has led to horrendous consequences for both people and the public health workforce. Modeling approaches have been proved to be highly advantageous in taking appropriate actions in the early stages of the pandemic. To this date, no study has been conducted to model the disease to investigate the prevalence of COVID-19 cases, especially after travel restrictions in Iran. We investigated contributing factors of early-stage coronavirus spread via generating a model to predict daily confirmed cases in Iran.

Methods: In this study, we collected publicly available data of confirmed cases in 24 provinces from April 4, 2020, to May 2, 2020, with a list of explanatory factors. we exploited the opportunities that Artificial neural networks offer to investigate contributing factors of early-stage coronavirus spread via generating a model to predict daily confirmed cases in Iran. The factors were checked separately for any linear associations and to train and validate a multilayer perceptron network.

Results: The accuracy of the models was evaluated; the R2 scores were 0.842 for population distribution, 0.822 for health index, and 0.864 for the population in the provinces. No significant linear associations were seen between collected factors and COVID-19 incidence in provinces. Also, the relative distance from the disease epicenter had no relation to the disease incidence in provinces.

Conclusion: Our results show the significant impact of population, population density, and health infrastructure quality on the cross-province spread of COVID-19 in the time of travel restrictions when the vacation ended. Accordingly, this information can be implicated in assessing the risk of epidemics and future policy-making in this area.