In a recent display of academic and technological prowess, Professor Kim Sung-hoon and his Biomedical Signal Research Team from Asan Medical Center (AMC) have clinched the top spot in the PhysioNet Challenge 2023 held in Atlanta, Georgia, from Oct. 1-4.

The competition is organized by PhysioNet, an open-source medical data provider institution jointly run by the Massachusetts Institute of Technology (MIT) and Harvard Medical School.

Since its inception in 2000, PhysioNet has been hosting this annual competition to evaluate the performance of AI algorithms utilizing biomedical signal data. The competition uses real clinical data from five hospitals in the U.S.

This year's competition focused on predicting neurological recovery using electroencephalography (EEG) signals from patients in a comatose state after cardiac arrest.

The competition was divided into two main categories -- a hackathon, which required on-site algorithm development within a restricted timeframe, and a challenge that ranked algorithms developed over three months.

Professor Kim's team, leveraging a pre-trained deep learning model optimized for biomedical signals and data augmentation techniques, secured first place in the hackathon and second in the challenge segment among the 113 teams from around the world that participated.

"Unlike conventional methods, which manually extract features, our deep learning technique that automatically learns clinical signal characteristics garnered significant global interest," Professor Kim said. "We will continuously explore new techniques applicable to prognosis predictions to best assist medical professionals and patients."

The findings of their analysis were also recently presented at the CinC (Computing in Cardiology) conference, which was held in conjunction with the PhysioNet Challenge.