Abstract
This paper presents methods to choose individuals to test for infection during a pandemic such as COVID-19, characterized by high contagion and presence of asymptomatic carriers. The smart-testing ideas presented here are motivated by active learning and multi-armed bandit techniques in machine learning. Our active sampling method works in conjunction with quarantine policies, can handle different objectives, and is dynamic and adaptive in the sense that it continually adapts to changes in real-time data. The bandit algorithm uses contact tracing, location-based sampling and random sampling in order to select specific individuals to test. Using a data-driven agent-based model simulating New York City we show that the algorithm samples individuals to test in a manner that rapidly traces infected individuals. Experiments also suggest that smart-testing can significantly reduce the death rates as compared with current methods, with or without vaccination. While smart testing strategies can help mitigate disease spread, there could be unintended consequences with fairness or bias when deployed in real-world settings. To this end we show how procedural fairness can be incorporated into our method and present results that show that this can be done without hurting the effectiveness of the mitigation that can be achieved.
Original language | English |
---|---|
Pages (from-to) | 120-144 |
Number of pages | 25 |
Journal | Information Systems Research |
Volume | 35 |
Issue number | 1 |
DOIs | |
State | Published - Mar 2024 |
Keywords
- COVID-19
- active sampling
- agent-based models
- algorithmic fairness
- multi-armed bandits
- non-stationarity