AI Predicts Overseas COVID-19 Cases Arriving in Korea

On the 24th, Iraqi dispatched workers who arrived at Incheon Airport aboard the Air Force aerial refueling aircraft 'KC-330' are exiting the arrival hall. / Photo by Mun Ho-nam munonam@

[Asia Economy Reporter Junho Hwang] Domestic researchers have developed an artificial intelligence (AI) capable of predicting the number of overseas confirmed cases entering from abroad. With higher accuracy than existing AI, it is expected to enable efficient monitoring of inflow trends. In particular, it can predict the situation up to two weeks ahead, contributing to the efficient allocation of quarantine resources.

On the 19th, the research team led by Professor Jaegil Lee of the Department of Industrial and Systems Engineering at the Korea Advanced Institute of Science and Technology (KAIST) announced the development of big data and AI technology to predict the number of COVID-19 overseas imported confirmed cases. The AI developed by the team utilizes big data such as the number of confirmed cases and deaths in various countries, the frequency of COVID-19 keyword searches in each country, the daily number of flights to Korea, and the number of roaming subscribers entering Korea from each country (KT) to predict the number of overseas imported confirmed cases for two weeks.

Subsequently, the research team predicted the number of overseas imported confirmed cases for the next two weeks using only about a month and a half of short-term training data, confirming that this model has up to 35% higher accuracy compared to existing time-series data-based machine learning or deep learning models.

The research team stated, "If the predicted values by existing AIs showed a difference of about two cases compared to the actual confirmed cases, the difference has been reduced to about one case."

Minseok Kim, a doctoral student and the first author, said, "This research is an example showing that the latest AI technology can be applied to COVID-19 quarantine," adding, "Using the AI developed this time, limited quarantine resources can be efficiently allocated."

Meanwhile, this research will be presented on the 24th in the 'AI for COVID-19' session at the international academic conference 'ACM KDD 2020.'

This content was produced with the assistance of AI translation services.