Modeling and Early Diagnosis of Alzheimer’s Disease Using Recurrent Neural Network

The Alzheimer’s is a common and predominant form of disease in old age people which is progressive (with time) by nature. This can be dangerous if not diagnosed and treated during the early stages. In the present paper, the authors have proposed a novel work for the detection of Alzheimer’s disease (AD) by incorporating long-short term memory (LSTM) algorithm with recurrent neural network-based architecture. The medical practitioners need an efficient and fast artificial learning techniques to diagnose the progressive nature of brain diseases such as AD which requires accurate modeling and diagnosis. The research work in this paper has been presented in two stages: the first stage is modeling of AD, and the second stage is diagnosis of the progressive AD by calculating time-series data from the patient’s visit at an interval of 6 months up to 48 months. In the first stage, a logistic regression technique-based model is developed (as proposed in our previous work) for the diagnosis of stable mild cognitive impairment (sMCI)/progressive mild cognitive impairment(pMCI) and prediction of progression time-frame from sMCI to pMCI or pMCI to AD. Furthermore, in the second stage, the developed model has been utilized for the detection of AD and prediction of its progression time by incorporating the LSTM algorithm. The simulation results show the effectiveness and accuracy of the proposed technique. It has been found that the accuracy is 94.79%, precision is 93.87%, and recall is 93.87%.