Automating Alzheimer’s Disease Prediction Using AI and Machine Learning Boosting Machines

Authors- Research Scholar Ms.A. Kamatchi, Associate Professor Dr. V. Maniraj

Abstract-Alzheimer’s disease (AD) is a progressive neurodegenerative condition that primarily affects the elderly. Although its symptoms start off mildly, they worsen as time goes on. While there is no cure for AD, early diagnosis can significantly mitigate its adverse effects. This study proposes a methodology called SMOTE-RF for AD prediction, utilizing machine learning (ML) algorithms. The performance of three algorithms—decision tree (DT), extreme gradient boosting (XGB), and random forest (RF)—is evaluated for this purpose. The experiments are conducted using the Open Access Series of Imaging Studies (OASIS) longitudinal dataset, which is available on Kaggle. To address class imbalance in the dataset, the synthetic minority oversampling technique (SMOTE) is applied. The experiments are carried out on both imbalanced and balanced datasets. On the imbalanced dataset, DT achieved an accuracy of 73.38%, XGB reached 83.88%, and RF obtained the highest accuracy of 87.84%. After balancing the dataset using SMOTE, DT achieved 83.15% accuracy, XGB reached 91.05%, and RF achieved the highest accuracy of 95.03%. The highest accuracy of 95.03% was attained with the SMOTE-RF model.

DOI: /10.61463/ijset.vol.13.issue2.240