An Intelligent Metaheuristic-Assisted Feedforward Neural Network Framework for Accurate Alzheimer’s Disease Classification Using Medical Diagnostic Data

Alzheimer's disease (AD) is a progressive neurodegenerative disease which represents a major burden of global health. The early and accurate classification of the stages of AD is crucial for early clinical intervention. In this paper, an intelligent hybrid system combining a Feedforward Neural Network (FNN) and a metaheuristic optimization method known as Grey Wolf Optimizer (GWO) is proposed for accurate classification of Alzheimer's disease based on multi-modal medical diagnostic information. To overcome the drawbacks of gradient-based learning such as falling into local minima and convergence slowdown, the GWO is used to optimize the FNN's hyperparameters and synaptic weights. The proposed GWO-FNN model is tested on two datasets: The Open Access Series of Imaging Studies (OASIS) dataset and Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. The experimental results show that the GWO-FNN outperforms the conventional machine learning models and the standalone neural networks, with a classification accuracy of 97.4%, sensitivity of 96.8%, specificity of 97.9% and AUC of 0.987. The results highlight the effectiveness of Deep Optimization using a Metaheuristic approach in clinical neurodiagnostic system.