A compartmental Mathematical model of COVID-19 intervention scenarios for Mumbai

Abstract

AbstractA new mathematical method with an outstanding potential to predict the incidence of COVID-19 diseases has been proposed. The model proposed is an improvement to the SEIR model. In order to improve the basic understanding of disease spread and outcomes, four compartments included presymptomatic, asymptomatic, quarantine hospitalized and hospitalized. We have studied COVID-19 cases in the city of Mumbai. We first gather clinical details and fit it on death cases using the Lavenberg-Marquardt model to approximate the various parameters. The model uses logistic regression to calculate the basic reproduction number over time and the case fatality rate based on the age-category scenario of the city of Mumbai. Two types of case fatality rate are calculated by the model: one is CFR daily, and the other is total CFR. The total case fatality rate is 4.2, which is almost the same as the actual scenario. The proposed model predicts the approximate time when the disease is at its worst and the approximate time when death cases barely arise and determines how many hospital beds in the peak days of infection would be expected. The proposed model outperforms the classic ARX, SARIMAX and the ARIMA model. And It also outperforms the deep learning models LSTM and Seq2Seq model. To validate results, RMSE, MAPE and R squared matrices are used and are represented using Taylor diagrams graphically.