Related Experiment Video
Updated: Aug 9, 2026
10:11
Fundus Photography as a Convenient Tool to Study Microvascular Responses to Cardiovascular Disease Risk Factors in Epidemiological Studies
Published on: October 22, 2014
Predicting stroke risk using retinal imaging with artificial intelligence: a scoping review of current evidence
Wagner Rios-Garcia1,2, Abigail D Via-Y-Rada-Torres3, Linda Salinas-Díaz4
1Research Network on Digital Health, Artificial Intelligence, and Education (NET- IA WORLD), Lima, Peru. wagner16rg@gmail.com.
Summary
Artificial intelligence (AI) combined with retinal imaging shows promise for predicting stroke risk by analyzing eye vascular changes. However, more robust validation is needed for clinical use.
Area of Science:
- Ophthalmology
- Neurology
- Medical Imaging
- Artificial Intelligence
Background:
- Stroke is a major cause of death and disability globally.
- Current stroke risk prediction tools have moderate accuracy.
- Retinal imaging offers a non-invasive method to assess cerebrovascular health.
Purpose of the Study:
- To conduct a scoping review of studies using artificial intelligence (AI) and retinal imaging for stroke risk prediction.
- To evaluate the methodologies and predictive performance of these AI-based approaches.
Main Methods:
- Scoping review following Joanna Briggs Institute and PRISMA-ScR guidelines.
- Searches in PubMed/MEDLINE, Scopus, and Web of Science.
- Inclusion of quantitative human studies reporting predictive performance metrics.
Main Results:
- Eleven studies were included, primarily using deep learning on fundus photography or OCT angiography.
- Predictive performance varied (AUC 0.719-0.987), with retinal vascular features being key predictors.
- Models integrating retinal imaging and clinical data showed improved discrimination, but most studies had high bias and lacked external validation.
Conclusions:
- AI-based retinal imaging has potential for stroke risk prediction.
- Current evidence is limited by methodological heterogeneity and insufficient external validation.
- Further prospective, externally validated studies are necessary for clinical implementation.
