FAU Researchers to Create AI-Powered COVID-19 Risk Assessment Tool
As the pandemic rears its head in the United States, attention is turning once again to the relative risk of infection in various areas across the country. Now, a team of researchers from Florida Atlantic University (FAU) have received a $90,000 grant through the National Science Foundation (NSF) RAPID program to build an AI-powered, web-based tool for COVID-19 modeling and risk evaluation.
“COVID-19 is an evolving epidemic and there is little knowledge about its outbreak and spread patterns, or the impact of viral evolution, demography, social behavior, cultural differences, and quarantine policies regarding these outbreaks,” said Stella Batalama, dean of FAU‘s College of Engineering and Computer Science. “As the battle against COVID-19 continues, a deluge of information is being produced. As a result, the dramatic outbreak differences with respect to diverse geographies, regional policies, and cultural groups is raising confusion, contradictions, and inconsistencies in disease outbreak modeling.”
The first step of the project will be a broad knowledge base of COVID-19 data, built to help users understand relationships between a wide range of variables and the virus’ spread and mutations. With that knowledge base in hand, the researchers will then develop a deep neural network-based prediction tool that uses information about demographics, infections, policies and more to evaluate risks. To do this, it will use a networked graph to represent entities and their relationships with one another.
“Supported by the knowledge base, the public will be able to employ information to estimate their infection risk level using social and behavioral information such as their family size, shopping patterns, and dining patterns, as well as local authority policies such as school, restaurant, and movie theater closures and night time curfews,” said Xingquan (Hill) Zhu, principal investigator of the grant and a professor at FAU. “They also will have access to demographic information such as population age, density and income, as well as health conditions like heart disease incidence, cancer prevalence, and substance misuse. Public health officials and the public-at-large also will be able to access regional virus conditions such as the number of infection cases in the area studied and infection rate.”
The researchers hope that the knowledge base and AI tool will lead to better-informed policy decisions.
“Academia, news agencies, and governments are continuously publishing advances in the understanding of the virus’ clinical pathologies, its genome sequences, and relevant administrative policies and actions taken. In addition, the public also responds to the changing environments through social media sites or other online sources, resulting in real-time social sensing opportunities,” said Xingquan (Hill) Zhu. “This is why a knowledge base of COVID-19 using machine learning is so crucial for us to model and understand the spread of COVID-19, and eventually mitigate the negative effects of the virus on public health, society, and the economy.”