Predictive Analysis for COVID-19 Spread in India by Adaptive Compartmental Model

The role of mathematical modelling in predicting spread of an epidemic is of vital importance. The purpose of present study is to develop and apply a computational tool for predicting evolution of different epidemiological variables for COVID-19 in India. We propose a dynamic SIRD (Susceptible-Infected-Recovered-Dead) and SEIRD (Susceptible-Exposed-Infected-Recovered-Dead) model for this purpose. In the dynamic model, time dependent infection rate is assumed for estimating evolution of different variables of the model. Parameter estimation of the model is the first step of the analysis which is performed by least square optimization of priori data. In the second step of the analysis, simulation is carried out by using evaluated parameters for prediction of the outbreak. The computational model has been validated against real data for COVID-19 outbreak in Italy. Time to reach peak, peak infected cases and total reported cases were compared with actual data and found to be in very good agreement. Next the model is applied for the case of India and various Indian states to predict different epidemiological parameters. Priori data was taken from the beginning of nation-wide lockdown on 24 March to 6 July. It was found that peak of the outbreak may reach in the month of August-September with maximum 4-5 lakhs active cases at peak. Total number of reported cases all over India would be in between three to five millions. State wise, Maharashtra, Tamilnadu and Delhi would be worst affected.