Clinically oriented dual-tier screening for post-stroke epilepsy with interpretable machine learning in a severely imbalanced cohort - Takeaways - MDSpire

Clinically oriented dual-tier screening for post-stroke epilepsy with interpretable machine learning in a severely imbalanced cohort

  • By

  • Lijun Wu

  • Han Wu

  • Lingling Xu

  • Jingze Li

  • Jian Wang

  • May 21, 2026

  • 0 min

Share

  • 1

    Post-stroke epilepsy occurs in 4.4% of stroke patients, complicating recovery and management despite its low incidence.

  • 2

    A dual-tier screening framework was developed to address severe class imbalance in predicting post-stroke epilepsy.

  • 3

    The primary model achieved a macro-area under the curve of 0.996, indicating excellent overall performance for risk stratification.

  • 4

    Key predictors of post-stroke epilepsy included neurological severity, hypertension, lactate, D-dimer, and aspartate aminotransferase.

  • 5

    The study emphasizes the need for interpretable models to support clinical decision-making in post-stroke epilepsy management.

Original Source(s)

Clinically oriented dual-tier screening for post-stroke epilepsy with interpretable machine learning in a severely imbalanced cohort

  • by Lijun Wu, Han Wu, Lingling Xu, Jingze Li, Jian Wang

  • May 21, 2026

Related Content

Magnetic resonance imaging and clinical prediction of intracranial atherosclerotic large vessel occlusion in acute ischemic stroke treated with endovascular thrombectomy

Top 10 States With Rising Demand for Physicians

These 10 states report physician shortages, projected physician supply gaps, or physician-to-population ratios below national benchmarks.

Not All Poststroke Cognitive Patterns Plateau

Researchers identify divergent recovery, persistence, and decline patterns across cognitive domains following stroke.