Enhancing Alzheimer’s Disease Detection with Chaotic Moth Flame Optimization Algorithm: A Feature Selection Approach

Advances in medical science have led to a better understanding of Alzheimer's disease (AD), as shown by the clear description of AD using numerical, medical & MRI architectural imagery details. Early diagnosis of AD is possible using clinical data in addition its presence can be identified with huge correctness. Still, managing the massive and broadening AD data record is seemly an issue and utilising the unrefined data record unless removing redundant and duplicate data can significantly impact prediction or classification accuracy. Therefore, this research proposes an improved memetic metaheuristic behavioural founded feature selection framework called chaotic moth flame optimization algorithm (CMFOA). The suggested framework of MFO incorporates chaotic mapping to boost the search process, distinguishing it from the traditional form. CMFOA outperforms Ant colony optimization (ACO) and Firefly algorithm (FFA) in terms of ease, variable reduction, optimal efficiency, quick implementation, and rapid computing, resulting in improved subsets of characteristics for AD prediction. The study utilized the Alzheimer's disease neuroimaging initiative (ADNI) dataset to ascertain potential characteristics implicated in the diagnosis of Alzheimer's disease. The performance of the three feature subset selection approaches was validated using a deep neural network (DNN).The results demonstrated that CMFOA outperformed the other models in terms of accuracy.