Ann Based Early Detection of Alzheimer Disease on Selected Features

Abstract

Alzheimer's Disease (AD) is a type of dementia, also called cognitive impairment. In cases where measures are not taken against the disease, it may result in a decrease in the quality of life of the person and result in very serious consequences. While it presents with neurological consequences such as decreased functions of thinking and memory, it may result in death in advanced cases. The fact that the treatment is not completely possible makes the place of early diagnosis and intervention important for AD. As a result of the researches carried out in the study, it was seen that there are many studies and scientific content within the framework of AD. A method for early diagnosis of the disease was evaluated by using an open source shared dataset, which includes some disease-specific values and demographic characteristics. By using Artificial Neural Networks (ANN) model, which is one of the machine learning methods, it is aimed to be useful for other studies to take precautions for early detection of the disease. With the ANN, which was classified as dementia and non-dementia individuals, Root Mean Square Error (RMSE) value 0.2302, Mean Absolute Error (MAE) value 0.1899 and accuracy rate of 98.5% was obtained.