Trend and prediction of COVID-19 outbreak in Iran: SEIR and ANFIS model

Abstract

Background: Mathematical and predictive modeling approaches can be used in COVID-19 crisis to forecast the trend of new cases for healthcare management purposes. Given the COVID-19 disease pandemic, the prediction of the epidemic trend of this disease is so important. Methods: We constructed an SEIR (Susceptible-Exposed-Infected-Recovered) model on the COVID-19 outbreak in Iran. We estimated model parameters by the data on notified cases in Iran in the time window 1/22/2020 - 20/7/2021. Global sensitivity analysis is performed to determine the correlation between epidemiological variables and SEIR model parameters and to assess SEIR model robustness against perturbation to parameters. We Combined Adaptive NeuroFuzzy Inference System (ANFIS) as a rigorous time series prediction approach with the SEIR model to predict the trend of COVID-19 new cases under two different scenarios including social distance and non-social distance. Results: The SEIR and ANFIS model predicted new cases of COVID-19 for the period February 7, 2021, till August 7, 2021. Model predictions in the non-social distancing scenario indicate that the corona epidemic in Iran may recur as an immortal oscillation and Iran may undergo a recurrence of the third peak. Conclusion: Combining parametrized SEIR model and ANFIS is effective in predicting the trend of COVID-19 new cases in Iran.