A novel validated, imaging-based risk model may predict poststroke epilepsy more accurately than existing tools

1. The IsCHEMiA score is a validated, imaging-based risk model that more accurately predicts poststroke epilepsy than existing tools across multiple international cohorts.

2. Incorporation of infarct size and hemorrhagic transformation identifies high-risk patients who may benefit from closer monitoring and future antiepileptogenic interventions.

Evidence Rating Level: 2 (Good)

Study Rundown: Poststroke epilepsy is a common and morbid complication of ischemic stroke, yet existing risk scores incompletely capture neuroimaging predictors of epileptogenesis. This study developed and internationally validated the IsCHEMiA score, a six-item clinical and imaging-based tool to predict poststroke epilepsy after first-ever ischemic stroke. Using a prospective U.S. stroke registry and competing risk regression with mortality as a competing event, the authors identified infarct size ≥5 cm, cortical involvement, hemorrhagic transformation, early symptomatic seizures, middle cerebral artery involvement, and age younger than 65 years as independent predictors. The score was externally validated in three cohorts from Hong Kong and Japan, demonstrating strong discrimination (c-statistics 0.826–0.870) and good calibration. IsCHEMiA consistently outperformed existing SeLECT-based scores. Low scores predicted minimal seizure risk, whereas scores ≥8 identified patients with very high long-term epilepsy risk. The IsCHEMiA score provides a practical framework for individualized poststroke seizure risk stratification and future preventive trials.

Click to read the study in Neurology

Relevant Reading: Influence of seizures on stroke outcomes: a large multicenter study

In-Depth [Prospective Cohort Study]: 

Poststroke epilepsy (PSE) is a frequent and clinically significant complication of ischemic stroke, affecting approximately 4–6% of patients within the first year and up to 8% within five years. PSE is associated with worse functional recovery, reduced quality of life, and increased mortality. Despite this burden, there is currently insufficient evidence to support routine primary antiseizure medication prophylaxis after stroke. Existing prediction tools, such as the SeLECT and SeLECT2.0 scores, have provided risk stratification but do not incorporate several key neuroimaging features known to be associated with epileptogenesis. The authors therefore sought to develop and internationally validate a novel, imaging-based risk score to predict PSE following the first-ever acute ischemic stroke.

This study used a prospective stroke registry from Massachusetts General Hospital (MGH) to derive the prediction model, including adults admitted with first-ever ischemic stroke between 2016 and 2018. Patients with prior stroke, prior epilepsy, primary intracranial hemorrhage, concurrent antiseizure medication use, or insufficient follow-up were excluded. Neuroimaging was systematically reviewed to characterize infarct size, cortical involvement, vascular territory, and hemorrhagic transformation. Competing risk regression was performed with all-cause mortality as a competing event, and backward stepwise elimination using the Akaike Information Criterion was used to derive the final multivariable model. External validation was conducted in three independent cohorts from Hong Kong (Queen Mary Hospital and Ruttonjee Hospital) and Japan (National Cerebral and Cardiovascular Center).

The derivation cohort included 1,436 patients, and the validation cohorts contributed an additional 2,534 patients. Across the overall study population, PSE occurred in 5.5% of patients. Univariable analyses identified several predictors previously described in the literature, as well as novel neuroimaging factors. After multivariable adjustment, six independent predictors remained and were incorporated into the final IsCHEMiA score: infarct size ≥5 cm, cortical involvement, hemorrhagic transformation, early symptomatic seizures within seven days of stroke, middle cerebral artery territory involvement, and age younger than 65 years. Each variable was assigned a weighted point value based on its adjusted subdistribution hazard ratio.

The IsCHEMiA score demonstrated excellent discrimination in the derivation cohort, with a c-statistic of 0.870, and consistent performance in all validation cohorts (c-statistics ranging from 0.826 to 0.857). Calibration at one and three years after stroke was good, with close agreement between predicted and observed PSE risk. Importantly, the IsCHEMiA score outperformed both SeLECT and SeLECT2.0 in most cohorts. Risk stratification showed clinically meaningful gradients: an IsCHEMiA score of 3 predicted a low risk of PSE (approximately 2% at one year and 6% at five years), whereas scores of 8 or higher were associated with a very high risk (approximately 67% at one year and 78% at five years).

The authors conclude that IsCHEMiA is a practical, imaging-informed, and internationally validated tool for predicting PSE in modern stroke care. By identifying patients at high risk, the score may guide closer monitoring, targeted EEG use, and future trials of antiepileptogenic therapies, while reinforcing current recommendations against routine prophylactic antiseizure medication use in low-risk patients.

Image: PD

©2026 2 Minute Medicine, Inc. All rights reserved. No works may be reproduced without expressed written consent from 2 Minute Medicine, Inc. Inquire about licensing here. No article should be construed as medical advice and is not intended as such by the authors or by 2 Minute Medicine, Inc.