Epidemic models with digital and manual contact tracing

Dongni Zhang,Tom Britton

2023-11-09

Abstract:We consider the Markovian SIR epidemic model, but where individuals either recover naturally or are diagnosed, the latter implying isolation and subject to being contact traced. We focus on the novel digital contact tracing based on individuals using a tracing app and consider both the situation when this is the only form of contact tracing, and when combined with manual contact tracing. It is proven that the epidemic process converges to a limiting process as the population size $n$ grows large. The limiting process is not a branching process (as is common for epidemic processes) because infectees may be contact traced "from" their infectors, thus creating dependencies. But by instead treating to-be-traced components as macro-individuals allows a multi-type branching process interpretation, thus giving an expression for the reproduction number $R$. This reproduction number (for the digital-only case) is then converted to an individual reproduction number $R^{(ind)}$ being easier to interpret. It is proven that, contrary to $R$, $R^{(ind)}$ decays monotonically with $\pi$ (the fraction of app-users), and the two reproduction numbers are proven to have the same threshold at 1. The digital (only) contact tracing is then compared with earlier results for manual (only) contact tracing where a fraction $p$ of all contacts are traced. It is proven that the critical fraction app-users $\pi_c$ for which digital contact tracing is reduced to $R=1$, and the critical fraction manually contact traced $p_c$ to reach $R=1$, satisfy $\pi_c>p_c$.

Probability,Populations and Evolution