Preliminary analysis of deep learning models for predicting Alzheimer’s disease progression

Abstract

Correct diagnosis of Alzheimer's disease (AD) is crucial to patient care, particularly in the early stages of the illness, as risk knowledge enables patients to take preventative actions before irreversible brain damage develops. The majority of machine detection techniques are restricted by congenital observations, despite the fact that numerous recent research have employed computers to diagnose AD. Early diagnosis of AD is possible, however it cannot be predicted because prediction is only useful up until the point at which the disease starts to show symptoms. Deep Learning (DL) has emerged as a popular method for AD early diagnosis. Recurrent neural networks, of which Convalutional Bidirectional long short-term memory (Conv-BiLSTM ) is a particular variety, may be able to relate prior knowledge to the current task. We suggest a prediction model based on Convalutional Bidirectional long short-term memory (Conv-BiLSTM after observing that a patient's temporal data may be useful for forecasting the course of their illness. In order to represent the temporal relationship between characteristics and the subsequent stage of Alzheimer's disease, an Convalutional Bidirectional long short-term memory (Conv-BiLSTM) network with fully linked layer and activation layers is constructed. The Results indicate that our model performs better than the majority of the current models.