Related Experiment Video
Updated: May 31, 2026
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Multimodal machine learning for early risk stratification of post-stroke cognitive impairment
Xingyongpei Zheng1, Panpan Zhao2, Na Wang1
1Department of Neurology, The Affiliated Lianyungang Hospital of Xuzhou Medical University, Lianyungang, China.
Journal of Alzheimer'S Disease : JAD
|May 30, 2026
Summary
A new machine learning model accurately predicts post-stroke cognitive impairment (PSCI) in acute ischemic stroke (AIS) patients using clinical and imaging data. Early detection aids intervention to prevent dementia progression.
Area of Science:
- Neurology
- Artificial Intelligence
- Medical Informatics
Background:
- Post-stroke cognitive impairment (PSCI) is a significant complication of acute ischemic stroke (AIS), impacting recovery and quality of life.
- Early identification of high-risk individuals is crucial for timely intervention and preventing progression to dementia.
Purpose of the Study:
- To develop and validate a stacking-based multimodal machine learning model for predicting PSCI in AIS patients.
- Integrate clinical, demographic, and neuroimaging features for enhanced predictive accuracy.
Main Methods:
- Retrospective cohort study of 1070 AIS patients.
- Developed a stacking ensemble model combining XGBoost, Gradient Boosting, CatBoost, SVM, Logistic Regression, and LightGBM.
- Cognitive function assessed 3-6 months post-stroke; PSCI defined as z-score ≤ -2.0.
Main Results:
- The stacking model achieved high internal validation accuracy (98.13%) and AUC (0.9972).
- External validation demonstrated good performance with 81.00% accuracy and 0.9049 AUC.
- Key predictors included infarct volume, cortical lesions, medial temporal lobe atrophy, and NIHSS score.
Conclusions:
- The developed stacking-based multimodal model reliably predicts PSCI risk in AIS patients.
- This tool facilitates early detection and personalized interventions to prevent post-stroke dementia.
