Berkers and colleagues went prospectively into the tricky territory of predicting outcome in coma, using a task-state EEG paradigm most clinicians barely know exists. Machine learning on EEG signals gave solid AUCs across several timepoints—highlighting a signal, not a miracle. How much this tactically improves family counseling or decision-making is murky; the true test will be routine ICU adoption and whether it alters care or merely sharpens grim prognostication.
EEG and Familiar Names: Signal and Limits in ICU Prognosis for Comatose Patients
Does EEG response to familiar names meaningfully improve prediction of long-term outcome in comatose ICU patients?