Decentralized contact tracing protocols and a risk analysis approach to pandemic control

Abstract

Non-pharmaceutical interventions (NPIs) can protect against pandemic pathogens, but they depend on behaviour change, and so can impose costs on quality of life and civil liberties. With careful system design and risk analysis these tradeoffs can be improved, enabling more effective disease control at a lower cost. In this thesis, I propose a method for decentralized digital contact tracing that is fast, scalable, and cannot be used for mass surveillance. I show how targeted quarantine substitutes for broad social distancing and use this relationship to estimate the optimal quarantine risk threshold – finding that it strongly depends on disease prevalence. Using the joint distribution for infectiousness and test sensitivity, quarantine duration and test timing can be chosen to minimize the dura- tion of quarantine without increasing expected transmissions. Decentralized digital contact notification apps were used by close to 100 million people during the COVID-19 pandemic and prevented a significant number of transmissions despite challenges with system ro- bustness. Decentralized digital contact tracing combined with adaptive risk analysis can efficiently suppress infectious disease in the idealized (high participation) case, however more work is needed to design solutions that are both robust and socially acceptable.