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 Video

Updated: Jul 4, 2026

An R-Based Landscape Validation of a Competing Risk Model
05:37

An R-Based Landscape Validation of a Competing Risk Model

Published on: September 16, 2022

Development and External Validation of a Machine Learning Model for 10-Year Ischemic Stroke Risk Prediction in

Ahmed Khattab, Zhe Wang, Vinodh Srinivasasainagendra

    Medrxiv : the Preprint Server for Health Sciences
    |July 3, 2026
    PubMed
    Summary
    This summary is machine-generated.

    Related Concept Videos

    You might also read

    Related Articles

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

    Sort by
    Same author

    Implementing a Multi-Ancestry Polygenic Risk Score for Coronary Heart Disease in a Diverse Cohort.

    Genetics in medicine : official journal of the American College of Medical Genetics·2026
    Same author

    Reply to the comment on: The microbiota in axial spondyloarthritis: what have we learned from Mendelian randomisation studies?

    Clinical and experimental rheumatology·2026
    Same author

    Exploration of Regulatory Elements, MicroRNAs, and Copy Number Variation in Urogenital Chlamydia Reinfection in African American Women.

    International journal of molecular sciences·2026
    Same author

    Comment on: A Novel Mendelian Randomization Method With Binary Risk Factor and Outcome.

    Genetic epidemiology·2026
    Same author

    Integrating genomic medicine into primary care -examining perceptions of community advisory board members.

    Journal of community genetics·2026
    Same author

    Co-occurring clonal hematopoiesis exhibits strong selection and high leukemia risk.

    Nature communications·2026

    Machine-learning models for ischemic stroke risk prediction show improved accuracy using clinical, lab, and polygenic risk scores (PRS). External validation confirmed performance but highlighted the need for ancestry-specific recalibration, suggesting tailored approaches over universal race-aware models.

    Area of Science:

    • Cardiovascular disease research
    • Genetics and precision medicine
    • Health informatics

    Background:

    • Machine-learning models for ischemic stroke risk prediction often lack validation across diverse ancestral populations.
    • The specific contributions of polygenic risk scores (PRS) and self-reported race to these models remain unclear.

    Purpose of the Study:

    • To develop and externally validate a 10-year ischemic stroke risk prediction model.
    • To quantify the added value of laboratory data trajectories, PRS, and self-reported race/ethnicity across different populations.

    Main Methods:

    • A retrospective cohort study using the All of Us (AoU) Research Program for development and the BioMe Biobank for external validation.
    • XGBoost machine-learning models were developed in tiers, incorporating clinical features, laboratory trajectories, and 20 PRS.

    Related Experiment Videos

    Last Updated: Jul 4, 2026

    An R-Based Landscape Validation of a Competing Risk Model
    05:37

    An R-Based Landscape Validation of a Competing Risk Model

    Published on: September 16, 2022

  • Models were evaluated under both race-blind and race-aware specifications, assessing discrimination (AUROC) and calibration (O/E ratio).

    • Main Results:

    • The machine-learning model (M3) incorporating clinical, lab, and PRS significantly outperformed traditional stroke risk scores (Framingham, Pooled Cohort Equations) in the AoU cohort (ΔAUROC 0.16-0.18).
    • External validation in the BioMe cohort showed good discrimination but required recalibration, particularly for European American participants.
    • Polygenic risk score contribution was significant in the BioMe Hispanic cohort, and adding self-reported race provided small gains when combined with PRS.

    Conclusions:

    • A machine-learning ensemble integrating clinical, laboratory, and polygenic data offers superior ischemic stroke risk prediction compared to traditional scores.
    • External validation demonstrated the model's generalizability but underscored the necessity of site-specific recalibration for absolute risk predictions.
    • The findings support per-ancestry calibration strategies, as the impact of self-reported race overlapped with polygenic signals, suggesting limitations in universal race-aware model deployment.