One of the ongoing debates during the COVID-19 pandemic has been which non-pharmaceutical interventions prevent disease spread and how stringent they should be. To help answer this question, XPRIZE and New Jersey-based Cognizant are offering $500,000 to the team that develops the best data tools to select the most balanced approaches.
Now a UC San Diego computer science and pediatrics professor, Debashis Sahoo, is one step closer to the prize after moving into the second round of the competition.
The XPRIZE Pandemic Response Challenge, which started with more than 100 teams, seeks to give governments better ways to fight COVID as well as future pandemics without destroying their economies.
“The plan is to come up with a range of options,” said Sahoo, the principal investigator at the Boolean Lab. “If a government wants to be more stringent, we should be able to give them a plan. If they want to be less stringent, we should be able to give them useful options there as well.”
The competition is divided into two phases. In the first, teams were asked to predict new daily cases based on the Oxford University dataset, which records how different communities around the world intervened and whether those actions stopped the virus. During the second phase, teams are developing develop data tools to identify the best strategies to reduce COVID cases.
Sahoo’s group is one of 48 teams to make the second round. The winner will be announced on February 26. But even before the competition is complete, the team is finding useful insights.
“When we experiment on different intervention plans, the naive approach is, if cases are too high, we want to be little more stringent, and if cases are low we want to be less stringent,” said Sahoo. “But that strategy really doesn't work. It has to be more dynamic, because once you put an intervention in place, the effect may not be evident for two weeks.”
The team also found that employing aggressive shutdown strategies in short bursts, even a single day, can have a positive impact, a surprising finding. In addition, holidays and workplaces play significant roles in COVID spread. Sahoo believes these data approaches will help governments make the most appropriate decisions to manage COVID and possibly future pandemics.