Related Experiment Video
Updated: Sep 14, 2025
07:31
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
7.2K
Predicting delirium in critically Ill COVID-19 patients using EEG-derived data: a machine learning approach.
Ana Viegas1,2,3,4,5, Cristiana P Von Rekowski6,7,8, Rúben Araújo6,7,8
1NMS - NOVA Medical School, FCM - Faculdade de Ciências Médicas, Universidade NOVA de Lisboa, Campo Dos Mártires da Pátria 130, 1169-056, Lisbon, Portugal. a2020449@nms.unl.pt.
Geroscience
|July 23, 2025
Summary
Machine learning models using electroencephalography (EEG) show promise for predicting delirium in critically ill COVID-19 patients. Increased theta activity on EEG is a key predictor, but models require further refinement for optimal accuracy.
Area of Science:
- Neuroscience
- Critical Care Medicine
- Computational Biology
Background:
- Delirium is a frequent complication in critically ill patients, especially those with SARS-CoV-2 infection, increasing morbidity and mortality.
- Early detection of delirium risk is vital for timely interventions and improved patient outcomes.
Purpose of the Study:
- To investigate the efficacy of electroencephalography (EEG) combined with machine learning (ML) models for predicting delirium in critically ill patients with SARS-CoV-2 infection.
- To identify specific EEG features that can serve as predictors of delirium.
Main Methods:
- A prospective observational cohort study involving 70 critically ill patients with SARS-CoV-2 infection.
- Development of ML models using EEG data alone, and integrated demographic, clinical, laboratory, and EEG data.
- Analysis of EEG data before and after delirium diagnosis in a subset of patients.
Main Results:
- Increased theta activity on EEG was identified as a consistent predictor of delirium.
- An EEG-only ML model achieved an AUC of 0.733.
- Integrating clinical and demographic data improved predictive performance (AUC=0.825).
- A model analyzing pre- and post-delirium EEG data achieved the highest accuracy (AUC=0.950).
Conclusions:
- EEG-based ML models demonstrate potential for delirium prediction in critically ill COVID-19 patients.
- Increased theta activity is a significant EEG biomarker for delirium.
- Further model refinement is necessary to enhance predictive accuracy, sensitivity, and specificity.
