Nour Zawawi, Heba Gamal Saber, Mohamed Hashem and Tarek F.Gharib, Ain Shams University, Egypt

Alzheimer's disease (AD) is a degenerative brain ailment that affects millions worldwide. It is the most common form of dementia. Patients with an early diagnosis of Alzheimer's disease have a strong chance of preventing additional brain damage by halting nerve cell death. At the same time, it begins to progress several years before any symptoms appear. The variety of data is the biggest problem encountered during diagnosis. Neurological examination, brain imaging, and often asked questions from his connected closed relatives are the three forms of data that a neurologist or geriatrics employs to diagnose patients. One of the biggest questions which need answering is the choice of a convenient feature.

The main objective of this paper is to help neurologists or geriatricians diagnose patient conditions. It proposes a new hybrid model for features extracted from medical data. It discusses AD's early diagnosis and progression for all features considered in the diagnosis and their complex interactions. It proves to have the best accuracy when compared with the state-ofthe-art algorithm. Also, it proves to be more accurate against some recent research ideas. It got 95% in all cases, considering this work focused more on increasing the number of instances in comparison.

Alzheimer's disease, Diagnosis, Prediction, Classification, Feature Selection.