Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A Heterogeneous Graph Neural Network Framework for Multi-Horizon Stroke Mortality Prediction.

Aabila Tharzeen, Alireza Vafaei Sadr, Nazli Radfar

    Medrxiv : the Preprint Server for Health Sciences
    |June 22, 2026
    PubMed
    Summary

    Graph-based machine learning accurately predicts stroke mortality across multiple timeframes by analyzing electronic health records. This approach improves upon traditional models, offering better insights into patient outcomes.

    Related Concept Videos

    Hemorrhagic Stroke l: Introduction01:17

    Hemorrhagic Stroke l: Introduction

    A hemorrhagic stroke is an acute neurological event that occurs when a weakened cerebral blood vessel ruptures, allowing blood to accumulate within or around the brain. The sudden release of blood forms a focal hematoma that increases intracranial pressure, displaces neural tissue, and can obstruct cerebrospinal fluid pathways. These effects may be compounded by intraventricular extension of the hemorrhage, cerebral edema, or compression of adjacent structures, all of which contribute to...

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Evaluating and Mitigating Carbon Dioxide Equivalent Emissions in Stroke Management: A Modeling Study.

    Journal of the American Heart Association·2026
    Same author

    Timing of Antithrombotic Therapy Among Patients With Hemorrhagic Transformation After an Ischemic Stroke.

    Journal of the American Heart Association·2026
    Same author

    Evaluation of steroids for acute COVID in the prevention of long COVID in children: An EHR and pediatric cohort study from the RECOVER Initiative.

    PloS one·2026
    Same author

    Effectiveness of Physical Activity Interventions Using Wearables and Smartphone Applications for Individuals With Cardiovascular Diseases and Stroke: A Systematic Review and Meta-Analysis.

    Journal of the American Heart Association·2026
    Same author

    Translating AI to the Bedside with Physician Buy-In: Recommendations from a Meta-Analysis and Systematic Review of the Literature.

    Bioengineering (Basel, Switzerland)·2025
    Same author

    Tenecteplase Versus Alteplase for First-Pass Reperfusion in Basilar Artery Occlusion Stroke Thrombectomy.

    Annals of neurology·2025

    Area of Science:

    • Medical Informatics
    • Machine Learning
    • Clinical Prediction Models

    Background:

    • Traditional machine learning for stroke mortality prediction often overlooks the complex relationships within electronic health records (EHRs).
    • These models typically analyze each prediction timeframe independently, limiting their comprehensive understanding of patient outcomes.

    Purpose of the Study:

    • To develop and evaluate a novel graph-based machine learning model for predicting stroke mortality across multiple time horizons.
    • To assess the effectiveness of heterogeneous graph neural networks in capturing relational EHR data for improved mortality prediction.

    Main Methods:

    • Developed Stroke Temporal Heterogeneous Graph (StrokeTHG), a graph neural network model utilizing EHR data.
    • Incorporated temporal encoding and encoded relationships between EHR entities (patients, diagnoses, comorbidities).

    Related Experiment Videos

  • Compared StrokeTHG against traditional models like Logistic Regression, Random Forest, and XGBoost.

    • Main Results:

    • StrokeTHG achieved superior Area Under the Receiver Operating Characteristic Curve (AUROC) scores (0.872, 0.878, 0.837) across 30-day, 90-day, and 1-year prediction horizons.
    • The model outperformed all tabular baselines, identifying more mortality cases at clinically relevant specificity levels.
    • Analysis highlighted the importance of phenotype-patient and admission-patient edges in the EHR graph for accurate prediction.

    Conclusions:

    • Heterogeneous graph representations of EHR data significantly enhance multi-horizon stroke mortality prediction compared to tabular models.
    • StrokeTHG offers a promising methodological framework adaptable for various EHR-based clinical research studies.
    • The model demonstrates improved predictive performance, particularly at actionable sensitivity thresholds.