Delirium detection using GAMMA wave and machine learning: A pilot study.

Area of Science:

• Neuroscience
• Critical Care Medicine
• Biomedical Engineering

Background:

• Delirium affects up to 80% of critically ill older adults, leading to adverse outcomes like cognitive impairment and mortality.
• Current subjective assessment tools identify less than half of delirium cases.
• Objective measures like electroencephalogram (EEG) are needed but have been impractical.

Purpose of the Study:

• To evaluate machine learning methods for predicting delirium using gamma band EEG data.
• To assess the feasibility of a limited-lead, rapid-response handheld EEG device for delirium detection.

Main Methods:

• A prospective pilot study enrolled 13 critically ill participants (age ≥ 50) on mechanical ventilation.
• Machine learning models analyzed gamma band EEG data from a handheld device.
• Stepwise discriminant analysis was among the predictive models used.

Main Results:

• Machine learning models achieved over 70% accuracy in predicting delirium.
• Stepwise discriminant analysis demonstrated the best overall performance.
• Gamma band analysis from a handheld EEG device showed predictive potential.

Conclusions:

• A handheld, limited-lead EEG device shows promise for objective delirium prediction in critically ill older adults.
• Machine learning analysis of gamma band EEG data can improve delirium detection rates.
• Further research is needed to optimize cut points and confirm efficacy.