Ερευνητικό Σεμινάριο

ΑΝΑΚΟΙΝΩΣΗ

Ερευνητικό Σεμινάριο της Σχολής Επιστημών και Τεχνολογίας της Πληροφορίας
Δευτέρα 7 Δεκεμβρίου 2020, ώρα 14:00(ακριβώς) - 15:00

(Η ημερίδα θα διεξαχθεί διαδικτυακά μέσω της πλατφόρμας MS Teams)

"Deploying a Data-Driven COVID-19 Screening Policy at the Greek Border"
Κίμων Δρακόπουλος, Assistant Professor of Data Sciences and Operations, Marshall School of Business, University of Southern California

Περίληψη

In collaboration with the Greek government, we designed and deployed a  nation-wide COVID-19 screening protocol for travelers to Greece. The goals of the protocol  were to combine limited demographic information about arriving travelers with screening  results from recently tested travelers to i) judiciously allocate Greece’s limited testing budget to identify  asymptomatic, infected travelers and ii) quickly identify hotspots and spikes in other nations to inform  immigration/border policies in real-time. This talk details i) the operations of our designed system (including  border screening, database management, closed-loop feedback, and liasing with contact-tracing  teams) ii) a novel, batched,
contextual bandit algorithm tailored to the unique features of this  problem and iii) an empirical assessment of the benefits of the deployed system from the summer/fall  2020, showing that targeted testing based on traveler’s features essentially doubles the  effectiveness compared to random testing and static greylisting. That is, in a country with  daily budget of 7500 tests, targeting is as effective as Radom sampling with 14,850 tests, a  number that at the time was
effectively the testing capacity of the whole country.

Βιογραφικό
————————

Kimon Drakopoulos is an Assistant Professor of Data Sciences and Operations at USC Marshall School of Business, where he researches complex networked  systems, information design and information economics. He completed his Ph.D. in the Laboratory for Information and Decision Systems at MIT, focusing on  the analysis and control of contagion within networks. His current research revolves  around controlling contagion, epidemic or informational as well as the use of information  as a lever to improve operational outcomes in the context of testing allocation, fake news propagation and belief polarization.