ABSTRACT
Contact-tracing smartphone applications have been developed and used as a complement to manual contact tracing in the COVID-19 pandemic. The goal of these apps is to trace contacts between people and notify the mobile phone owners when one of their contacts tested positive. People who receive a notification should behave as exposed people, take a test, and possibly isolate themselves until they receive the result. Unfortunately, identifying contacts based on distance is technically a daunting task: apps can be configured conservatively (a very small number of people are notified, limiting the effectiveness of the app) or they may be more tolerant and produce a high number of notifications but also of false positives. We review the data available from Immuni, the Italian app, which provides detailed figures on the notifications sent and the positive users, and we show that Immuni was configured to generate a very large amount of notifications. We estimate the testing resources that the health system would have needed if the app was downloaded by 100%of the adult population, and every notified person would require a test. In such conditions, Immuni would have generated a number of tests orders of magnitude higher than what was available. We compare the performance of Immuni with the currently available literature on other apps and observe that contact-tracing apps had a limited impact on the second wave of the COVID-19 pandemic. As contact tracing exposes citizens to privacy risks, we discuss some ways to reshape the goal of the apps to achieve a better tradeoff between social benefit and risk. IEEE