Early prediction of post-stroke interictal epileptiform discharges by development of an interpretable imbalance-aware predictive model

Interictal epileptiform discharges (IEDs) detected by EEG within 72 h of admission after acute stroke have been associated with poorer outcomes, yet EEG is not routinely performed in this setting. Identifying patients with a higher probability of early IEDs may help prioritize EEG assessment. This study primarily aimed to identify clinical and neuroimaging predictors of early IEDs and to develop an interpretable, imbalance-aware predictive model with a nomogram for individual risk stratification. We analyzed 647 acute stroke patients who underwent EEG within 72 h of admission. Clinical and neuroimaging variables were collected and features were extracted. Because of class imbalance, the training data were resampled using SMOTENC. Feature selection was performed using the ReliefF algorithm, and the selected variables were used to develop an interpretable model for predicting early IEDs occurrence. Model performance was evaluated on an independent test set. As a complementary clinical analysis, multivariable logistic regression was also performed to examine the association between early IEDs and 3-month functional outcome. IEDs were detected in 148 patients (23%). Patients with IEDs were older and more frequently female and had higher prevalence of atrial fibrillation, acute infectious conditions at admission, hemorrhagic stroke, greater neurological severity, and more extensive cortical involvement. The predictive model achieved an accuracy of 71% and an area under the curve (AUC) of 0.83 on the independent test set. The strongest predictors of IEDs occurrence were cortical involvement, female sex, and hypertension, followed by higher admission NIHSS, absence of dyslipidemia, older age, no reperfusion treatment, and hemorrhagic stroke. These variables were incorporated into a nomogram for individualized risk estimation. As a complementary clinical finding, multivariable analysis showed that early IEDs were independently associated with unfavorable 3-month functional outcome (OR = 2.206, 95% CI 1.318–3.690; p = 0.003). We developed an interpretable predictive model that uses routinely available clinical and neuroimaging features to estimate the risk of early post-stroke IEDs. The proposed model may help identify patients with a higher estimated risk of early IEDs and potentially inform EEG prioritization, thereby contributing to a more personalized approach to acute stroke management.

Authors

Institutions

Publication Details

Journal
BioMedical Engineering OnLine
Published
2026-11-18
DOI
https://doi.org/10.1186/s12938-026-01635-1
Primary Topic
Acute Ischemic Stroke Management
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Early prediction of post-stroke interictal epileptiform discharges by development of an interpretable imbalance-aware predictive model

Paolo Manganotti, Giovanni Furlanis, Katerina Iscra, Giulia Mazzon et al.
6 links
BioMedical Engineering OnLine
Acute Ischemic Stroke Management
article

Early prediction of post-stroke interictal epileptiform discharges by development of an interpretable imbalance-aware predictive model

Paolo Manganotti, Giovanni Furlanis, Katerina Iscra, Giulia Mazzon, Marinella Tomaselli, Miloš Ajčević, Marcello Naccarato, Gabriele Prandin, Alessandro Biscontin, Paola Caruso, E. Ricci, Michele Malesani
article en

Abstract

Interictal epileptiform discharges (IEDs) detected by EEG within 72 h of admission after acute stroke have been associated with poorer outcomes, yet EEG is not routinely performed in this setting. Identifying patients with a higher probability of early IEDs may help prioritize EEG assessment. This study primarily aimed to identify clinical and neuroimaging predictors of early IEDs and to develop an interpretable, imbalance-aware predictive model with a nomogram for individual risk stratification. We analyzed 647 acute stroke patients who underwent EEG within 72 h of admission. Clinical and neuroimaging variables were collected and features were extracted. Because of class imbalance, the training data were resampled using SMOTENC. Feature selection was performed using the ReliefF algorithm, and the selected variables were used to develop an interpretable model for predicting early IEDs occurrence. Model performance was evaluated on an independent test set. As a complementary clinical analysis, multivariable logistic regression was also performed to examine the association between early IEDs and 3-month functional outcome. IEDs were detected in 148 patients (23%). Patients with IEDs were older and more frequently female and had higher prevalence of atrial fibrillation, acute infectious conditions at admission, hemorrhagic stroke, greater neurological severity, and more extensive cortical involvement. The predictive model achieved an accuracy of 71% and an area under the curve (AUC) of 0.83 on the independent test set. The strongest predictors of IEDs occurrence were cortical involvement, female sex, and hypertension, followed by higher admission NIHSS, absence of dyslipidemia, older age, no reperfusion treatment, and hemorrhagic stroke. These variables were incorporated into a nomogram for individualized risk estimation. As a complementary clinical finding, multivariable analysis showed that early IEDs were independently associated with unfavorable 3-month functional outcome (OR = 2.206, 95% CI 1.318–3.690; p = 0.003). We developed an interpretable predictive model that uses routinely available clinical and neuroimaging features to estimate the risk of early post-stroke IEDs. The proposed model may help identify patients with a higher estimated risk of early IEDs and potentially inform EEG prioritization, thereby contributing to a more personalized approach to acute stroke management.

BioMedical Engineering OnLine
University of Trieste (IT), IRCCS Materno Infantile Burlo Garofolo (IT), Ospedale di Cattinara (IT), Azienda Sanitaria Universitaria Integrata di Trieste (IT)
Good health and well-being
Acute Ischemic Stroke Management
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon