COVID-19: Analytic results for a modified SEIR model and comparison of different intervention strategies

TL;DR

This paper extends the SEIR model to include asymptomatic carriers for COVID-19, deriving analytic results and comparing intervention strategies like social distancing and testing-quarantining, with application to India’s data.

Contribution

It introduces an extended SEIR model with asymptomatic carriers, provides analytic solutions, and compares the effectiveness of intervention strategies in controlling COVID-19.

Findings

TQ is more efficient than SD at same R0 target.

Contact tracing and testing ratio are critical for TQ success.

Eigenvalue analysis explains disease progression and fits data.

Abstract

The Susceptible-Exposed-Infected-Recovered (SEIR) epidemiological model is one of the standard models of disease spreading. Here we analyse an extended SEIR model that accounts for asymptomatic carriers, believed to play an important role in COVID-19 transmission. For this model we derive a number of analytic results for important quantities such as the peak number of infections, the time taken to reach the peak and the size of the final affected population. We also propose an accurate way of specifying initial conditions for the numerics (from insufficient data) using the fact that the early time exponential growth is well-described by the dominant eigenvector of the linearized equations. Secondly we explore the effect of different intervention strategies such as social distancing (SD) and testing-quarantining (TQ). The two intervention strategies (SD and TQ) try to reduce the disease reproductive number, $R_0$, to a target value $R^{\rm target}_0 < 1$, but in distinct ways, which we implement in our model equations. We find that for the same $R^{\rm target}_0 < 1$, TQ is more efficient in controlling the pandemic than SD. However, for TQ to be effective, it has to be based on contact tracing and our study quantifies the required ratio of tests-per-day to the number of new cases-per-day. Our analysis shows that the largest eigenvalue of the linearised dynamics provides a simple understanding of the disease progression, both pre- and post- intervention, and explains observed data for many countries. We apply our results to the COVID data for India to obtain heuristic projections for the course of the pandemic, and note that the predictions strongly depend on the assumed fraction of asymptomatic carriers.