Boosted Model of LSTM-RNN for Alzheimer Disease Prediction at their Early Stages

Abstract

Due to progressive cognitive decline of memory consequences as a degenerative disease known as Alzheimer which affects severely on elderly peoples. This paper devised a Boosted LSTM-RNN (BLSTM-RNN) to discover the Alzheimer Disease (AD) in their earlier stages. Diagnosing it in early stage provide will greatly helps the health professionals to improve the life of the victims in future and their vigorous affect can be gradually controlled. This paper used the variant of deep learning approach known as Recurrent Neural Network (RNN) which is gated with bidirectional LSTM to learn the pattern of cognitive impairments. This work, extracts deep features of the Alzheimer Disease Neuroimaging Initiative (ADNI) dataset and passed them into the LSTM layer to built a boosted model for alzheimerprediction. The deployment of LSTM greatly assists to predict the datasets more precisely by adapting the ability of relatively insensitive to gap length and it can able to remember huge number of patterns for a long period of time. The simulation study also proved that the accuracy in prediction of cognitive stage of the alzheimer patients in earlier stages and produces less error rate compared to the SVM, ANN and standard RNN classification models.