MRI-Based Biomarkers for Early Detection and Classification of Alzheimer Disease using Machine Learning

Older adults suffer from Alzheimer's Disease which intensifies as a neurodegenerative condition and raises both fatal outcomes and worsening dementia progression. The correct identification of Alzheimer's Disease along with its early detection remains essential because the current diagnostic methods show limited validity. MRI shows its effectiveness through both local brain region and overall brain area tissue atrophy assessment for AD patients. Binary classifiers based on Machine Learning (ML) models processing biomarkers extracted from MRI data improve clinical decision accuracy because they enable better-informed diagnosis. This research creates an AI-based diagnostic system which uses the OASIS MRI dataset to perform three cognitive status categories: Nondemented, Demented, and Converted. This last category identifies subjects whose brain condition evolved from nondemented to demented over time. The system utilizes Random Forest as well as AdaBoost alongside SVM and KNN and LR models for its operations. The classification accuracy Reached 96% for Random Forest, SVM and Logistic Regression while their AUC scores reached 0.9906, 0.9898, 0.9935 respectively. The experimental results displayed AdaBoost next to KNN for accuracy with 94.67% while having AUC scores of 0.9767 and 0.9938 respectively. AI-driven MRI analysis demonstrates strong potential to detect early AD while classifying patients before it advances to an advanced stage through efficient interventions.