Published Research
[ IEEE Xplore ]Gradient Boosting-Based Framework for Alzheimer's Disease Prediction Using Clinical and Lifestyle Data
Designed an interpretable ensemble ML pipeline using XGBoost and LightGBM on 2,149 patient records spanning 35 clinical, cognitive, and lifestyle features. Achieved 95.35% accuracy and 0.948 ROC-AUC through ensemble learning and stratified cross-validation, applying SHAP to interpret key predictive features including MMSE, ADL, and Functional Assessment scores.
Core Methodologies & Focus Areas
95.35%
Accuracy
0.948
ROC-AUC
2,149 Records
Dataset Size
35 Attributes
Features
