Efficient Deep Learning for Detecting Alzheimer’s Disease Staging: An Ensemble Strategy Using MobileNetV3 and ShuffleNetV2

The progressive neurodegenerative disease known as Alzheimer’s disease (AD) necessitates an early and precise diagnosis in order to effectively treat it. Conventional diagnostic techniques depend on the manual evaluation of medical imaging, which is laborious and subject to error. By merging MobileNetV3 with ShuffleNetV2, this study suggests a lightweight deep learning system for automated Alzheimer’s disease staging into which the model is trained to classify brain MRI scans. To improve classification performance, MobileNetV3 and ShuffleNetV2, which are renowned for their effectiveness and high accuracy, are combined utilizing an ensemble approach. When tested on a publicly accessible Alzheimer’s MRI dataset, the suggested framework outperforms traditional deep learning models while retaining a small CPU footprint, with an accuracy of over $92 \%$. The findings show that the ensemble approach is ideal for real-time clinical applications and deployment on edge devices because it offers quick, dependable, and resource-efficient Alzheimer’s disease categorization.