Frequency and Time Domain EEG Analysis for Prognostication of Postanoxic Comatose Patients

As part of the George B. Moody PhysioNet Challenge 2023, our team (am_vision) presents a novel approach to prognosticate the outcomes of postanoxic comatose patients based on frequency domain and time domain features by using electroencephalogram (EEG) recordings. Frequency domain features include spectral entropy, power distribution, and dominant frequency. In the Time domain, zero crossing rate, energy, and entropy were calculated. We also utilized demographic information: Age, Sex, Return of Spontaneous Circulation (ROSC), Out-of-Hospital Cardiac Arrest (OHCA), Shockable Rhythm, and Targeted Temperature Management (TTM) to train XGBoost regressor and classifier to predict both the Cerebral Performance Category (CPC) score and the Outcome respectively. Our proposed method received a challenge score of 0.485 (ranked 18
^th
out of 36), an outcome accuracy of 0.81 (ranked 4th out of 36), and an outcome F-measure of 0.769 (ranked 5
^th
out of 36) on the test set.