This thesis examines the potential of improving early Alzheimer’s disease (AD) diagnosis by integrating multimodal data, combining 3D Magnetic Resonance Imaging (MRI) scans with clinical and genetic data. The objective is to construct a comprehensive Machine Learning based diagnostic tool that exceeds the capabilities of traditional single-data-type methodologies. Subjects are classified into three distinct categories: Cognitively Normal (CN), Mild Cognitive Impairment (MCI), and Alzheimer’s Disease (AD). Data for this study is sourced from the Alzheimer’s Disease and Neuroimaging Initiative (ADNI), a key player in AD research, enabling the experimentation with various state-of-the-art (SOTA) pre-trained networks tailored to the data. ACS Converter is used to convert the 2D models into 3D models without the loss of pre-trained weights to accommodate the 3D imaging data. This research explores both early and late fusion techniques for integrating data, with late fusion—merging modalities at advanced stages of the training process—demonstrating a higher diagnostic accuracy of 95.64%.