EEG-Based Cardiac Arrest Outcome Estimation with Highly Interpretable Features

The following is the solution proposed by the team UPFantastic for the Predicting Neurological Recovery from Coma After Cardiac Arrest: The George B. Moody PhysioNet Challenge 2023. The team was unable to be scored on the test set for the official phase of the challenge. Nevertheless, in our view, the methods proposed merit consideration, since the approach focused on interpretable features and analysis of the relevance and distribution of the data. We used synchrony and relative EEG band power measurements, along with patient data to feed a tree-based machine learning algorithm. We also used Multiple Kernel Learning in a subset to analyze the tendency of the patients during the monitoring in a dimensional reduced space.