External validation of machine learning models for predicting prehospital delay in acute ischemic stroke: a retrospective cohort study - Takeaways - MDSpire
Coming Soon: Introducing MDSpire News. Learn more
Conexiant’s news site is now MDSpire News. Learn more

External Assessment of Machine Learning Algorithms for Anticipating Prehospital Delays in Acute Ischemic Stroke: A Retrospective Cohort Analysis

  • By

  • Shan Zeng

  • Aishanjiang Yusufujiang

  • Hui Dang

  • Li Zhu

  • Abudukeyoumu Yasheng

  • Jingjing Wang

  • Wei Yan

  • Huijuan Yu

  • Gulijanat Mamattursun

  • Maierpu· Aini

  • Di Kang

  • Hongyan Li

  • September 10, 2026

Share

  • 1

    The study externally validated eight prediction algorithms for prehospital delays in acute ischemic stroke across two hospitals in Kashgar, China.

  • 2

    Prehospital delay affected 80.9% of patients in Hospital 1 and 83.2% in Hospital 2, defined as an onset-to-door interval exceeding 4.5 hours.

  • 3

    Logistic regression showed an AUC of 0.774 without SMOTE, outperforming the SMOTE-trained model, which had an AUC of 0.756.

  • 4

    XGBoost achieved the highest validation AUC of 0.771, but its advantage over logistic regression was not statistically significant.

  • 5

    The study's framework is intended for retrospective health-system profiling and quality improvement, not for individual pre-arrival prediction.

Original Source(s)

Related Content