Explainable Artificial Intelligence-based Prediction Of Poor Neurological Outcome From Head Computed Tomography In The Immediate Post-resuscitation Phase

This study applied an explainable machine learning model to head CT images obtained immediately after resuscitation to predict neurological outcomes at 1 month with high accuracy. The accuracy of the machine learning predictions was comparable to that of the gradient-weighted class activation mapping (Grad-CAM) predictions in terms of the ROC-AUC and higher than that of the Grad-CAM predictions in terms of the PR-AUC. This is the first study to directly predict neurological outcomes at 1 month using only head CT images obtained immediately after resuscitation.