Epidemiological parameter sensitivity in Covid-19 dynamics and estimation

Abstract: Covid-19 is one of the most dreaded pandemics/epidemics in the world threatening the human population. The dynamics of this pandemic is quite complicated and prediction of pandemic states often fails. In this work, we study and correlate the SIR epidemiological model with the ongoing pandemic and found that pandemic dynamics and states are quite sensitively dependent on model parameters. The analysis of the exact parametric solution of the deterministic SIR model shows that the fixed points ($g^*$) depend on the SIR parameters, where, if $g^*\langle\frac{\gamma}{\beta x_0}$ then $g^*$ is stable and the pandemic can be controlled, whereas, if $g^*\rangle\frac{\gamma}{\beta x_0}$ then $g^*$ is unstable indicating active pandemic state, and $g^*=0$ corresponds to an endemic state. The dynamics show asymptotic stability bifurcation. The analytical solution of the stochastic SIR model allows to predict endemic time $t_E\propto\frac{1}{\gamma}$ and mean infected population becomes constant for small values of $\gamma$. The noise in the system even can modulate pandemic dynamics. Hence, we estimated the values of these parameters during lockdown periods using the ABC SSA algorithm and found strong dependence on control strategy, namely, lockdown and follows power-law behaviors. The numerical results show strong agreement with the real data and have the positive effect of social distancing or lockdown in the pandemic control.