par Reyna, Matthew A;Amorim, Edilberto;Sameni, Reza;Weigle, James;Elola, Andoni;Rad, Ali Bahrami;Seyedi, Salman;Kwon, Hyeokhyen;Zheng, Wei Long;Ghassemi, Mohammad Mahdi;van Putten, Michel J A M M.J.A.M.;Hofmeijer, Jeannette;Gaspard, Nicolas
;Sivaraju, Adithya;Herman, Susan S.T.;Lee, Jongwoo J.W.;Westover, Michael Brandon;Clifford, Gari G.D.
Référence Computing in Cardiology
Publication Publié, 2023-03-01

Référence Computing in Cardiology
Publication Publié, 2023-03-01
Article révisé par les pairs
Résumé : | The George B. Moody PhysioNet Challenge 2023 invited teams to develop algorithmic approaches for predicting the recovery of comatose patients after cardiac arrest. A patient's prognosis after the return of spontaneous circulation informs treatment, including the continuation or withdrawal of life support. Brain monitoring with an electroencephalogram (EEG) can improve the objectivity of a prognosis, but EEG interpretation requires clinical expertise. The algorithmic analysis of EEGs can potentially improve the accuracy and accessibility of prognoses, but existing work is limited by small and homogeneous datasets. The PhysioNet Challenge 2023 contributed to addressing these problems. It introduced the International Cardiac Arrest REsearch consortium (I-CARE) dataset, which is a large, multi-center collection of EEGs, other physiological data, and clinical outcomes, with over 57,000 hours of data from 1,020 patients from seven hospitals. It required teams to submit their complete training and inference code to improve the reproducibility and generalizability of their research. A total of 111 teams participated in the Challenge, contributing diverse approaches from academic, clinical, and industry participants worldwide. |