- A student in the Department of Artificial Intelligence developed a cutting-edge model that uses search engine queries reflecting real-time public interest to predict COVID-19 case numbers

- Research published in JMIR, a top healthcare technology journal with an impact factor of 7.4 (JCR top 2.4%)

△  From left to right: Ahn Seong-ho (undergraduate student, co-first author, Department of Artificial Intelligence, Catholic University of Korea), Prof. Lim Kwang-il (co-first author, Uijeongbu St. Mary’s Hospital, Catholic University of Korea), Won Hyun-sik, M.S. (co-author, Department of Artificial Intelligence), Prof. Kim Kang-min (co-corresponding author, Department of Artificial Intelligence), and Prof. Jeong Dong-hwa (co-corresponding author, Department of Artificial Intelligence)

 A research team led by undergraduate students from the Department of Artificial Intelligence at The Catholic University of Korea has developed AI technology capable of accurately predicting COVID-19 case numbers. The model uses advanced word embedding-based query expansion to identify and analyze search engine queries that reflect changing public interest over time. 

 The study was a collaborative effort between Professor Jeong Dong-hwa and Professor Kim Kang-min from the Department of Artificial Intelligence and Professor Lim Kwang-il from Uijeongbu St. Mary’s Hospital. Undergraduate student Ahn Seong-ho and Professor Lim Kwang-il were named co-first authors of the study, which was published in the prestigious Journal of Medical Internet Research (JMIR), a top healthcare technology journal with an impact factor of 7.4 (JCR top 2.4%).

 The research addresses a key limitation in existing models that rely on expert-selected search queries. Instead, the team developed an AI-based method to automatically identify search engine queries using COVID-19-related news articles, which are more representative of public concerns. Recognizing that news articles often reflect public concerns, the research team developed a word embedding-based query expansion method that uses text data from COVID-19-related news articles to identify relevant search engine queries. With this approach, the team created an AI model that automatically selects search engine queries strongly correlated with case numbers every four months. These queries were then used to predict COVID-19 case numbers 1 to 14 days in advance. 

 The results showed that this AI model was significantly more accurate than traditional methods relying on expert-selected queries. It performed particularly well in predicting case numbers 9 to 11 days ahead. When tested on past outbreaks such as MERS and SARS, the model also demonstrated high accuracy, underscoring its potential as a reliable tool for forecasting the spread of infectious diseases.

 Ahn Seong-ho, a senior AI student, shared, “This technology allows us to reflect real-time changes in public interest without needing expert intervention. It offers a faster, more efficient way to predict the spread of infectious diseases. I hope it becomes a valuable tool for future epidemic forecasting and advancements in medical AI.” 

 Professor Jeong Dong-hwa added, “Ahn Seong-ho, one of our first-generation AI students, successfully expanded his undergraduate research scholarship project, applying what he learned in the classroom to real-world research. Publishing in a top-tier journal demonstrates the strength of his academic dedication and our department’s structured education. This study is a great example of how undergraduates can achieve high-quality research outcomes.” 

 This research was funded by grants from several key programs, including the First Research Grant from the Korea Research Foundation, led by Professor Jeong Dong-hwa of the Department of Artificial Intelligence; the ICT Talent Development Program for Undergraduate-Master’s Integration, supported by the Ministry of Science and ICT and managed by the Institute of Information & Communications Technology Planning & Evaluation (IITP); and the Excellent Young Researcher Program from the Korea Research Foundation, led by Professor Kim Kang-min.

△ Diagram of the query expansion-based COVID-19 prediction framework