Related Experiment Video
Updated: Jul 13, 2026
Abbiategrasso Brain Bank Protocol for Collecting, Processing and Characterizing Aging Brains
Published on: June 3, 2020
A qEEG-based prognostic model for cognitive impairment after ischemic stroke: Development and internal validation
Vy Tuong Nguyen1,2, Trang Ho Thu Quach3, Hiep Do4
1Neurology Department, University of Medicine and Pharmacy at Ho Chi Minh City, Vietnam.
Background:
Cognitive impairment is a frequent and disabling consequence of ischemic stroke, yet early prognostic tools are limited. Quantitative electroencephalography (qEEG) offers a portable, low-cost measure of post-stroke cortical dysfunction. We aimed to develop a prognostic model for six-month cognitive impairment after first-ever ischemic stroke.
Methods:
In this single-centre prospective cohort study, patients with first-ever supratentorial ischemic stroke (NIHSS < 15) were enrolled within 7 days of onset at a tertiary hospital in Vietnam. Resting-state 19-channel EEG and brain MRI were acquired at baseline. Four region-specific qEEG parameters were derived for five cortical regions and globally: Delta/Alpha Ratio (DAR), Delta-Theta/Alpha-Beta Ratio, Peak Alpha Frequency (PAF), and Dominant Frequency Variability. The primary outcome was cognitive impairment at six months (education-adjusted MoCA < 26). Predictors were selected by AIC-guided forward stepwise logistic regression. Internal validation used a selection-aware bootstrap (1000 replicates), with multiple imputation and sensitivity analyses. Reporting followed TRIPOD and STROBE guidelines.
Results:
The analytic cohort comprised 96 participants; 44 (46%) developed cognitive impairment at six months. Higher occipital DAR (OR 2.09, 95% CI 1.45-3.39), lower frontal PAF (OR 0.46, 95% CI 0.27-0.74), and education below high school (OR 7.16, 95% CI 1.89-34.18) increased risk of cognitive impairment. Infarct volume was associated univariably but did not survive multivariable selection. Apparent AUC was 0.937 (95% CI 0.890-0.984); optimism-corrected AUC was 0.891 (0.861-0.945), with good calibration (Brier score 0.098). Effects were preserved across multiple imputation and alternative MoCA cutoffs.
Conclusions:
In this single-centre development cohort, a parsimonious three-variable model combining occipital DAR, frontal PAF, and educational attainment showed good internal discrimination and calibration for six-month cognitive impairment after first-ever ischemic stroke. These variables should be regarded as candidate predictors rather than an established prognostic tool; external multicentre validation against clinically adjudicated endpoints is the necessary next step before any clinical interpretation.
