Related Experiment Video
Updated: Jul 26, 2026
09:48
Quantification of Neurovascular Protection Following Repetitive Hypoxic Preconditioning and Transient Middle Cerebral Artery Occlusion in Mice
Published on: May 4, 2015
13.8K
PRERISK: A Personalized, Artificial Intelligence-Based and Statistically-Based Stroke Recurrence Predictor for
Giorgio Colangelo1,2, Marc Ribo1,3, Estefanía Montiel2
1Vall d'Hebron Research Institute, Passeig de la Vall d'Hebron, Barcelona, Spain (G.C., M. Ribo, M.O.-G., M.M., Á.G.-T., M. Requena, J.P., J.J., D.R.-L., N.R.-V., F.R., B.T., C.A.M., M. Rubiera).
Stroke
|March 28, 2024
Summary
Predicting stroke recurrence is challenging. PRERISK, a new machine learning tool, offers personalized stroke recurrence risk prediction, potentially improving patient self-care and outcomes.
Area of Science:
- Neurology
- Data Science
- Public Health
Background:
- Individualized stroke recurrence prediction is difficult but crucial for patient self-care.
- A novel statistical and machine learning classifier, PRERISK, was developed to address this challenge.
Purpose of the Study:
- To develop and validate a machine learning model for predicting individual stroke recurrence risk over time.
- To compare the performance of machine learning models against traditional Cox regression.
Main Methods:
- Analysis of clinical and socioeconomic data from 41,975 patients diagnosed with stroke in Catalonia, Spain (2014-2020).
- Development of supervised machine learning models to predict early, late, and long-term stroke recurrence.
- Evaluation of model accuracy using C statistics and area under the receiver operating characteristic curve (AUC).
Main Results:
- 16.21% of patients experienced stroke recurrence within a median follow-up of 2.69 years.
- Key predictors identified include time from previous stroke, Barthel Index, atrial fibrillation, dyslipidemia, age, diabetes, and sex.
- Machine learning models demonstrated statistically significant improvements over Cox regression for predicting recurrence at 90 days, 91-365 days, and >365 days (AUCs: 0.76, 0.60, 0.71 respectively).
Conclusions:
- PRERISK offers a novel, personalized, and accurate approach to predicting stroke recurrence risk.
- The model's ability to incorporate dynamic risk factor control holds potential for improved patient management.
