An efficient GS-RBFN framework for early prediction and classification of ad

A progressive brain disorder, which eventually destroys memory cells, is termed Alzheimer’s Disease (AD). AD causes memory loss and other regular activities. Due to the variations in cytoarchitecture, the categorical labeling of various tissues presents a difficult task in AD classification. For addressing this challenge, this paper proposes a new GELU and SWISH-based Radial Basis Function Network (GS-RBFN)-centric early prediction and classification of AD. For classifying AD into Mild Cognitive Impairment (MCI), AD, and Control Normal (CN), the proposed model deploys image pre-processing, segmentation, morphological operation, data augmentation, image representation extraction, feature selection, and classification steps. Primarily, images are gathered from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. Next, by utilizing normalization, skull removal, and spatial smoothing approaches, the images are pre-processed. Then, by using the Brownian Log Scaling Archimedes Optimization-based Watershed Segmentation (BLSAOWS), significant brain tissues are segmented. After that, using morphological operations, the segmented images are enhanced. Next, for obtaining different formations of the segmented images, a data augmentation process is deployed. Subsequently, the image features are extracted, and the best features are chosen utilizing the Base Switch Rule Infimum and Supremum-centric Rock Hyrax Swarm Optimization (BSRISRHSO) algorithm. Lastly, utilizing a new GS-RBFN classifier, the AD is classified. Through the experimental analysis, the proposed model’s efficiency is determined. Thus, the proposed GS-RBFN proficiently predicts AD individuals with an accuracy, precision, and sensitivity of 98.45%, 98.44%, and 98.44%, respectively. The proposed GS-RBFN achieved a less computation time of 14876 ms. Furthermore, the proposed BSRISRHSO obtained a minimum feature selection time of 24012 ms. The Proposed BLSAOWS acquired a high efficiency of 98%. Also, the proposed model acquired superior accuracy that outperformed all baseline techniques. Thus, the experimental results revealed that the research methodology obtained more impressive outcomes in AD prediction.