On the Optimization of Automatic Exposure Detection in Contact Tracing

Automatic exposure detection and notification is a decentralized system that notifies individuals of potential exposure to viruses such as COVID-19. The protocol is based on proximity estimation through smart phones using Bluetooth Low Energy (BLE) technology. It operates using guidelines that specify distance and time thresholds to determine the occurrence of contacts with infected individuals. In the United States, 2 m and 15 min thresholds are used to measure the duration of exposures due to such contacts. In this paper, using two metrics for the system performance, i.e., the number of unnecessary quarantines and undetected infections, we provide a mathematical framework for the optimization of the distance and time thresholds. We show the relative gain in the performance of the automatic exposure detection when the optimal time and distance thresholds are used. The framework is also extended to demonstrate the advantage of using different thresholds for individuals with various risk tolerance levels. Other factors such as the severity of the outbreak in the community, transmissibility of the virus as well as the surrounding environment (e.g., indoor vs. outdoor) can also be considered in the optimization framework. This study is done through an agent-based simulation platform which emulates automatic exposure notification with BLE-based proximity detection. The platform allows for comprehensive analysis with several user-controlled parameters that can specify different scenarios. The ultimate objective of this study is to enhance the effectiveness of automatic exposure notification in contact tracing systems during pandemics.