RDET stacking classifier: a novel machine learning based approach for stroke prediction using imbalance data.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Neurology

Background:

• Stroke, caused by disrupted brain blood flow, leads to disability.
• Early stroke identification is critical for effective treatment and improved patient quality of life.
• Existing machine learning models face challenges with dataset imbalance and computational complexity.

Purpose of the Study:

• To develop an ensemble machine learning model for accurate stroke prediction.
• To reduce model parameters and computational complexity for efficient stroke diagnosis.
• To address class imbalance in stroke datasets using oversampling techniques.

Main Methods:

• Utilized Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic Sampling (ADAYSN) to handle class imbalance.
• Developed a stacking ensemble classifier combining Random Forest, Decision Tree, and Extra Tree classifiers.
• Employed k-fold cross-validation and hyperparameter tuning for model optimization.

Main Results:

• The proposed stacking ensemble classifier achieved 100% accuracy.
• The model demonstrated exceptional precision, recall, and F1-score.
• Outperformed nine other machine learning classifiers in stroke prediction accuracy.

Conclusions:

• The stacking ensemble approach offers a highly accurate and effective method for stroke prediction.
• This model can significantly enhance early diagnosis and patient management for stroke.
• The research provides a computationally efficient and robust solution for stroke prediction using machine learning.