Imagine how much easier life would be if a computer dashboard could predict when a trip to the local supermarket, or hair salon, increased the risk of COVID-19?
A new computer dashboard being designed by Florida Atlantic University intends to do just that, sorting through vasts amount of data and using algorithms — just like Google and Netflix— to predict outcomes.
WLRN is committed to providing the trusted news and local reporting you rely on. Please keep WLRN strong with your support today. Donate now. Thank you.
“It’s really like a doctor,” said Xingquan Zhu, a computer engineer overseeing the work. “The majority of dashboards just represent the numbers. But we try to figure out what’s happening behind those, and then we will be able to produce some sort of estimation on the risk.”
The idea for the dashboard came from the onslaught of information produced as researchers race to understand the new disease and the evolving pandemic.
“There is little knowledge about its outbreak and spread patterns, or the impact of viral evolution, demography, social behavior, cultural differences, and quarantine policies,” FAU computer engineer Stella Batalama said in a statement.
That’s sometimes led to confusion and contradictions in guidance, she said.
The dashboard works by grouping characteristics — whether it’s age, location, family size — into cohorts and then estimating risk. It won’t be able to determine individual risk, but it will be able to estimate generalities, Zhu said. For example, in South Florida as cases rise, it can calculate the risk of behavior like visiting restaurants.
“Its mission is actually to help you to see what should I reduce,” he said. “We will probably not give you a number, but give you some sort of suggestion like, yes or no if I frequently shop. If you reduce your shopping, the risk will likely be reduced.”
Zhu said the dashboard will comb through the data to predict outcomes based on the cohorts. It will also mine social media, including Twitter and Reddit, for patterns that can help guide public health policy.
The project also involves creating more searchable databases for researchers, he said. Finding common patterns can be difficult with so much information being produced all over the world. Even basic terminology poses problems.
“We have to hunt for a way to describe it,” Zhu said. “Even for mugs. Some people might say a mug and some say a cup.”
In terms of advancing science, that part of the project perhaps provides the most promise by allowing researchers to build better models to predict outcomes with the disease.
The team, Zhu said, hopes to have a prototype available by August. The project is being funded by a $90,000 grant from the National Science Foundation.