Spatial modelling, risk mapping, change detection, and outbreak trend analysis of coronavirus (COVID-19) in Iran (days between 19 February to 14 June 2020).

Abstract Objectives Coronavirus disease 2019 (COVID-19) represents a major pandemic threat that has spread to more than 212 countries and 2 international conveyance with more than 432,902 recorded deaths and 7,898,442 confirmed global worldwide so far (on June 14, 2020). It is crucial to investigate the spatial drivers to prevent and control the epidemic of COVID-19. Methods This is the first comprehensive study of COVID-19 in Iran and it undertakes spatial modeling, risk mapping, change detection, and outbreak trend analysis of the disease spread. Four main steps were taken: comparison of Iranian coronavirus data with the global trends; prediction of mortality trends using regression modelling; spatial modelling, risk mapping, and change detection using the random forest (RF) machine learning technique (MLT); and validation of the modelled risk map. Results The results show that from February 19 to June 14, 2020 the average growth rates (GR) of COVID-19 deaths and the total number of COVID-19 cases in Iran were 1.08 and 1.10, respectively. Based on World Health Organisation (WHO) data, Iran’s fatality (deaths/0.1 M pop) is 10.53. Other countries’ fatality rates were, for comparison, Belgium – 83.32, UK – 61.39, Spain – 58.04, Italy – 56.73, Sweden – 48.28, France – 45.04, USA – 35.52, Canada – 21.49, Brazil – 20.10, Peru – 19.70, Chile – 16.20, Mexico– 12.80, and Germany – 10.58. This fatality rate for China is 0.32 (deaths/0.1 M pop). The heatmap of the infected areas over time identified two critical time intervals for the COVID-19 outbreak in Iran. The provinces were classified in terms of disease and death rates into a large primary group and three provinces that had critical outbreaks that were separate from others. The heatmap of countries of the world show that China and Italy were distinguished from other countries in terms of nine viral infection-related parameters. The regression models for death cases showed an increasing trend but with some evidences of turning. A polynomial relationship was identified between coronavirus infection rate and province population density. In addition, a third-degree polynomial regression model for deaths showed an increasing trend recently, indicating that subsequent measures taken to cope with the outbreak have been insufficient and ineffective. The general trend of deaths in Iran is similar to the worlds, but it shows lower volatility. Change detection of COVID-19 risk maps with a random forest model for the period from March 11th to March 18th showed an increasing trend in COVID-19 in Iran’s provinces. It is worth noting that using the LASSO MLT to evaluate variables’ importance indicated that the most important variables were distance from bus stations, bakeries, hospitals, mosques, ATMs (automated teller machines), banks, and the minimum temperature of the coldest month. Conclusions We believe that the risk maps provided by this study is the primary, fundamental step for managing and controlling COVID-19 in Iran and its provinces.