Development and Internal Validation of a Clinical Imaging-Based Stroke Prediction Model in a Community Cohort

Area of Science:

• Neurology
• Epidemiology
• Medical Imaging

Background:

• Existing stroke prediction models often overestimate risk and neglect neuroimaging.
• There is a need for improved stroke risk assessment in community settings.

Purpose of the Study:

• To develop a novel stroke prediction model using comprehensive neuroimaging data.
• To enhance stroke risk prediction accuracy in a community-based population.

Main Methods:

• Prospective cohort study of 1586 participants in rural China.
• Least absolute shrinkage and selection operator regression for predictor selection.
• Cox proportional hazards model with bootstrap validation.

Main Results:

• The final model identified smoking, diabetes, cerebral small-vessel disease, and intracranial artery stenosis as key predictors.
• The model demonstrated strong discrimination (AUC 0.88 for 5-year risk) and good calibration.
• A nomogram was developed for individualized stroke risk assessment.

Conclusions:

• The new model accurately predicts 5- and 7-year stroke risk by integrating clinical and neuroimaging features.
• Management of diabetes, smoking, and cerebrovascular disease is crucial for primary stroke prevention.