Hybrid Harris Hawk Optimization (HHO): A Novel Framework for Alzheimer's Disease Prediction Using Neuroimaging Data

Alzheimer's disease is a progressive neurological disorder marked by memory loss and cognitive decline. Using neuroimaging data from the Alzheimer's Disease Neuroimaging Initiative dataset, this work examines the use of Harris Hawk Optimization in Alzheimer's disease prediction. HHO improves AD prediction accuracy with and without feature selection techniques. It is inspired by the hunting behaviour of Harris's hawks. Numerical evaluations of the results show significant gains in performance when feature selection is paired with HHO. In particular, without feature selection, the Gaussian Mixture Model classification obtains the best accuracy of 82%; with 800 selected features, forward selection raises the accuracy to 87.90%. These results highlight the prominence of feature selection in maximizing predictive performance and the effectiveness of HHO in AD prediction. The study offers insightful information on using neuroimaging data for early diagnosis and recognition of Alzheimer's disease.