What question did this study set out to answer?

Develop and externally validate an hourly risk score derived from routine vital signs to enable early detection of delayed cerebral ischemia following aneurysmal subarachnoid hemorrhage.

February 18, 2021StrokeOpen Access

Dynamic Detection of Delayed Cerebral Ischemia

View Full Paper

Ask AI

Bookmark

Authors

MMMurad MegjhaniColumbia University Irving Medical Center KTKalijah TerilliStanford University MWMiriam WeissUniversity of Bern

Discussion

Member takes

Implication

Cohort study reveals dynamic vital sign models predict delayed cerebral ischemia up to 12 hours early in subarachnoid hemorrhage patients, highlighting potential to prevent brain damage.

Key Points

Develop and externally validate an hourly risk score derived from routine vital signs to enable early detection of delayed cerebral ischemia following aneurysmal subarachnoid hemorrhage.
Trained an ensemble machine learning model combining L2-regularized logistic regression, random forest, and support vector machines on N=310 patients using demographics and 22 time-varying vital sign features.
Computed time-varying measures including means, standard deviations, and cross-correlations between heart rate time series and other routine vitals (blood pressure, respiratory rate, oxygen saturation).
Evaluated model performance using area under the receiver operating characteristic curves and validated hourly risk score generation across two external cohorts (n=38 and n=40).
In the training cohort, 101 of 310 patients (33%) developed delayed cerebral ischemia at a median onset of 6 days (interquartile range, 5–8 days).
The classification model achieved an area under the receiver operating characteristic curve of 0.83 (interquartile range, 0.76–0.83) prior to clinical onset.
When tested on external cohorts, the algorithm correctly identified 64% and 91% of events up to 12 hours prior to clinical diagnosis, yielding 2.7 and 1.6 true alerts per false alert.

Cite This Study

Megjhani et al. (2021) studied this question.

synapsesocial.com/papers/6a881d23f8cd0ec04aeb3057 https://doi.org/10.1161/strokeaha.120.032546

View Full Paper

Ask AI

Bookmark

Implication

Cohort study reveals dynamic vital sign models predict delayed cerebral ischemia up to 12 hours early in subarachnoid hemorrhage patients, highlighting potential to prevent brain damage.

Dynamic Detection of Delayed Cerebral Ischemia

Discussion

Member takes

Implication

Key Points

Cite This Study

Dynamic Detection of Delayed Cerebral Ischemia

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider