Artificial intelligence (AI) has become a hot topic in recent months because of its ability to rapidly perform various complex tasks. But, it has caught a negative connotation in academia due to its use by students. However, researchers in Oklahoma State University’s College of Arts and Sciences use a subfield of AI —machine learning — for positively impactful research studying complex data sets. DETECTING DISEASE Dr. Lucas Stolerman, assistant professor in the Department of Mathematics, conducted research using digital traces to build prospective and real-time county-level early warning systems using machine learning to anticipate COVID-19 outbreaks in the U.S. During his postdoctoral year at the Boston Children’s Hospital and HarvardMedical School, Stolerman initially began conducting this research under Dr. Mauricio Santillana and in collaboration with Leonardo Clemente, a doctoral student. During his first year on faculty at OSU, the project was finalized. “Our goal was to detect potential COVID-19 outbreaks in U.S. counties ahead of time,” Stolerman said. “To this end, we created a machine learningbased early warning system leveraging internet-based digital data.” The early warning system provides an alert when the digital data increases STORY SYDNEY TRAINOR | PHOTOS JASON WALLACE AND ASHLEY RAILEY Artificial Assistance Oklahoma State University CAS faculty conduct research using AI machine learning technologies up to a certain level. The digital datasets included Twitter microblogs and Google trends terms including “Covid,” “Covid19,” “How Long Does Covid Last,” “Covid Symptoms,” “Fever,” “Cough” and others. By training the models over historical data, the researchers could then perform out-of-sample predictions and assess their method’s ability to correctly predict COVID-19 outbreaks. “COVID-19 outbreaks result from a complex combination of multiple factors, including viral mutations, social interactions, vaccination and human behavior. A powerful message from our study is that digital traces can help detect when those outbreaks will take place in U.S. counties,” Stolerman said. The figure above shows two datasets from Dr. Lucas Stolerman’s research: one is the Google search volume over time for the term “COVID” in Los Angeles (in blue). The other dataset is the actual number of COVID cases in the same region (in red). Comparing these time series shows an increase (blue circle) in Google searches preceding an increase in COVID-19 cases (red circle). Los Angeles (2020) Google (COVID) Incidence 0 11-Jan 22-Feb 04-Apr 16-May 27-Jun 08-Aug 100 200 0 2 4 6 105 36 CONNECT 2023
RkJQdWJsaXNoZXIy MjAxMjk=