Related Experiment Video
Updated: Dec 12, 2025
A Thrombotic Stroke Model Based On Transient Cerebral Hypoxia-ischemia
Published on: August 18, 2015
Machine Learning Prediction of Stroke Mechanism in Embolic Strokes of Undetermined Source
Hooman Kamel1, Babak B Navi1, Neal S Parikh1
1Clinical and Translational Neuroscience Unit, Department of Neurology, Feil Family Brain and Mind Research Institute (H.K., B.B.N., N.S.P., A.E.M., C.I.), Weill Cornell Medical College, New York.
Insights
Machine learning accurately identified that 44% of undetermined source strokes (ESUS) likely stem from cardiac embolism, aiding in better stroke cause determination.
Area of Science:
- Neurology
- Cardiology
- Artificial Intelligence
Background:
- Embolic strokes of undetermined source (ESUS) account for one-fifth of ischemic strokes.
- Distinguishing between cardioembolic and non-cardioembolic causes is crucial but challenging.
- The precise proportions of these underlying causes in ESUS remain unknown.
Purpose of the Study:
- To develop and apply a machine-learning algorithm to predict the proportion of cardioembolic strokes within the ESUS population.
- To differentiate between cardioembolic and non-cardioembolic etiologies in ESUS cases.
- To validate the algorithm's predictions against clinical outcomes like atrial fibrillation detection.
Main Methods:
- A machine-learning classifier was trained on a dataset of stroke cases with known etiologies, incorporating demographics, comorbidities, lab results, and echocardiograms.
- Ensemble methods including XGBoost, random forests, and adaptive splines were employed, with hyperparameter tuning via cross-validation.
- The validated algorithm was applied to a separate cohort of ESUS cases to estimate the prevalence of cardiac embolism.
Main Results:
- The machine-learning classifier achieved high accuracy (AUC 0.85) in distinguishing cardioembolic from non-cardioembolic strokes.
- The algorithm predicted that 44% of ESUS cases have a cardiac embolism source (95% CI, 39%-49%).
- Higher predicted probabilities of cardiac embolism in ESUS patients correlated with later detection of atrial fibrillation and specific clinical factors.
Conclusions:
- Machine learning provides an effective tool for estimating the proportion of cardioembolic strokes within the ESUS population.
- Approximately 44% of ESUS cases are predicted to be cardioembolic, highlighting the importance of cardiac evaluation.
- This approach offers a method to indirectly identify potential cardioembolic sources in stroke patients.
Background And Purpose:
One-fifth of ischemic strokes are embolic strokes of undetermined source (ESUS). Their theoretical causes can be classified as cardioembolic versus noncardioembolic. This distinction has important implications, but the categories' proportions are unknown.
Methods:
Using data from the Cornell Acute Stroke Academic Registry, we trained a machine-learning algorithm to distinguish cardioembolic versus non-cardioembolic strokes, then applied the algorithm to ESUS cases to determine the predicted proportion with an occult cardioembolic source. A panel of neurologists adjudicated stroke etiologies using standard criteria. We trained a machine learning classifier using data on demographics, comorbidities, vitals, laboratory results, and echocardiograms. An ensemble predictive method including L1 regularization, gradient-boosted decision tree ensemble (XGBoost), random forests, and multivariate adaptive splines was used. Random search and cross-validation were used to tune hyperparameters. Model performance was assessed using cross-validation among cases of known etiology. We applied the final algorithm to an independent set of ESUS cases to determine the predicted mechanism (cardioembolic or not). To assess our classifier's validity, we correlated the predicted probability of a cardioembolic source with the eventual post-ESUS diagnosis of atrial fibrillation.
Results:
Among 1083 strokes with known etiologies, our classifier distinguished cardioembolic versus noncardioembolic cases with excellent accuracy (area under the curve, 0.85). Applied to 580 ESUS cases, the classifier predicted that 44% (95% credibility interval, 39%-49%) resulted from cardiac embolism. Individual ESUS patients' predicted likelihood of cardiac embolism was associated with eventual atrial fibrillation detection (OR per 10% increase, 1.27 [95% CI, 1.03-1.57]; c-statistic, 0.68 [95% CI, 0.58-0.78]). ESUS patients with high predicted probability of cardiac embolism were older and had more coronary and peripheral vascular disease, lower ejection fractions, larger left atria, lower blood pressures, and higher creatinine levels.
Conclusions:
A machine learning estimator that distinguished known cardioembolic versus noncardioembolic strokes indirectly estimated that 44% of ESUS cases were cardioembolic.
