Affiliations:
Management Information System Department, University of Buraimi, Al Buraimi, Oman
This study presents the development of TAER_Robot, an explainable AI (XAI)-based medical assistant for predicting Alzheimer’s Disease (AlzD). The main aim is to integrate Machine Learning (ML) models with explanation techniques to build an accurate and interpretable risk assessment system. The research explores how age, cognitive function, and lifestyle factors influence prediction results, using a dataset of 2,149 records with 33 features such as age, gender, BMI, smoking, and alcohol use. Data preprocessing involved normalization, categorical encoding, and handling missing values. The dataset was split into training and testing sets at ratios of 80/20, 70/30, and 60/40 to identify the best configuration. Random Forest, CatBoost, and XGBoost were used as core ML models, while SHAP and LIME provided interpretability. LightGBM achieved the highest performance, with 95.6% accuracy and a 0.955 ROC-AUC score, exceeding previous models. Further testing confirmed system reliability with up to 94.1% accuracy. TAER_Robot enhances early-stage AlzD prediction by offering both strong performance and transparent decision-making, contributing to the improvement of AI-supported clinical decision systems.
Alzheimer’s prediction, Explainable AI, Machine learning, Risk assessment, Clinical decision
https://doi.org/10.21833/ijaas.2025.10.014
Shannaq, B. (2025). A smart medical assistant robot for explainable AI-based Alzheimer’s disease prediction using big data analytics. International Journal of Advanced and Applied Sciences, 12(10), 118–128. https://doi.org/10.21833/ijaas.2025.10.014