Explainable AI-Driven Diagnosis of Alzheimer's Disease Using SMOTEENN and SHAP Interpretability
Year: 2026
Type: conference-paper
Abstract: Alzheimer's disease (AD) is the most prevalent category of dementia and a progressive neurodegenerative disorder that gradually affects a person's potential for memory and thinking, so prompt and accurate detection is necessary for effective treatment. The research proposes an ensemble machine learning (ML) model with explainable artificial intelligence (XAI) for the diagnosis of prior AD. Used an Alzheimer's disease dataset from Kaggle. To improve the model performance consistency, a well-performing preprocessing pipeline was adopted, such that simple imputation strategies to estimate missing values, Z-score normalization for feature scaling, relevant class-balancing techniques such as SMOTEENN, recursive feature elimination (RFE) is employed for optimal feature selection. Applied various machine learning models, including LightGBM, CatBoost, Logistic Regression, Support Vector Machine, XGBoost, Decision Tree, AdaBoost, K-Nearest Neighbors, and Random Forest. An ensemble model was created by combining LightGBM, CatBoost, and SVM, and it proved to be the most effective performance, achieving 97.3% accuracy on the database. Despite this, the implemented framework illustrated robust predictive achievement, the results are based on a single publicly available dataset and should be interpreted as preliminary evidence. Further validation using independent clinical datasets is required to assess generalizability. Also, use XAI (SHAP) to leverage the model's prediction mechanism and transferability. So, the proposed ensemble model easily finds the early key risk factor of these diseases, and it can be integrated into real-world health care.
Institutions: Southeast University, Maharishi International University
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W7204122012