Related Experiment Video
Updated: Dec 11, 2025
06:37
Electroencephalography Network Indices as Biomarkers of Upper Limb Impairment in Chronic Stroke
Published on: July 14, 2023
1.2K
Prediction of sleep-disordered breathing after stroke
Devin L Brown1, Kevin He2, Sehee Kim2
1Stroke Program, University of Michigan, United States.
Sleep Medicine
|August 25, 2020
Summary
Predicting sleep-disordered breathing after stroke using baseline characteristics is not yet reliable. Objective testing remains essential for accurate diagnosis in stroke survivors with potential sleep apnea.
Area of Science:
- Neurology
- Sleep Medicine
- Biostatistics
Background:
- Sleep-disordered breathing (SDB) is common after stroke and linked to worse outcomes.
- Current diagnosis relies on objective testing to identify SDB in stroke patients.
Purpose of the Study:
- To assess the utility of a statistical model using baseline characteristics to predict post-stroke SDB.
- To evaluate machine learning performance in identifying SDB risk factors after ischemic stroke.
Main Methods:
- A population-based study included 1330 participants post-ischemic stroke (2010-2018).
- Home sleep apnea tests were used for SDB screening (Respiratory Event Index ≥10).
- A random forest classifier analyzed demographics, stroke severity, clinical measures, and sleep symptoms to predict SDB.
Main Results:
- SDB was present in 67% of the study sample (n=891).
- The random forest model achieved an Area Under the Curve of 0.75.
- The model correctly classified 72.5% of validation samples.
Conclusions:
- Machine learning models using baseline data showed only fair predictive performance for post-stroke SDB.
- Objective diagnostic tests are still necessary for accurate SDB identification in stroke patients.
- Further research may refine predictive models for sleep-disordered breathing after stroke.
