<p>Memory and cognitive function are impacted by Alzheimer’s disease (AD), a neurodegenerative condition that progresses over time. A person’s prognosis and quality of life may improve with early diagnosis and prompt intervention and treatment for AD. Deep learning-based automatic and early detection algorithms have the potential to be more scalable and efficient, and they can potentially be used to screen more individuals for the disease. In this research, we propose a novel enhanced deep learning-based approach for AD diagnosis. The Enhanced MobileNetV2 model is introduced to classify the multi-class classifications of AD. To enhance classification performance, the local and global features of the images are extracted using the Modified ResNet-50 model. After that, the significant features are chosen using the Enhanced Walrus Optimization Algorithm. Extensive experimentation is performed using a publicly available Kaggle AD MRI Dataset. The proposed model successfully classified the number of brain MRI images into No Impairment (NI), Very Mild Impairment (VMI), Mild Impairment (MI), and Moderate Impairment (MOI). The proposed model achieves 99.23% accuracy, 98.63% precision, 99.11% recall, and 98.81% F1-score on the multi-class AD, outperforming the state-of-the-art methods. Evaluation measures strongly validate the robustness and efficacy of the proposed model for effective AD diagnosis. Healthcare providers can provide individualized treatment for AD patients with the help of the proposed enhanced model.</p>