Classification of Alzheimer Disease with Molecular Communication Systems using LSTM

Abstract

Today, there are many diseases caused by cell or inter molecular communication. For example, a communication disorder in the nerve nano-network can cause very serious nervous system-related diseases such as Multiple Sclerosis (MS), Alzheimer's and Paralysis. Understanding these diseases caused by communication is very important in order to develop innovative treatment methods inspired by information technologies. In addition, many advanced environmental and industrial nano-sensor networks such as the development of biologically inspired Molecular Communication systems (MCs), cellular-accurate health monitoring systems, many medical applications such as the development of communication-capable nano-implants for nervous system diseases. Nano networks focused on communication between nano-sized devices (Nano Machines) is a new communication concept which is known as MCs in literature. In this study, on the contrary to the literature, a new Long Short-Term Memory (LSTM) based MC model has been used to analyse the proposed system. After obtained the number of received molecules for different number of Amyloid Beta (Aβ) which causes Alzheimer’, a new method based on the LSTM model of deep learning is used for the classification of Aβ. Finally it is obtained that when the number of Aβ increases, the number of received molecules decrease. On a data set with five classes, experiments are conducted using LSTM. The proposed model's accuracy, precision, and sensitivity values are obtained as 97.05, 98.59 and 98.54 percent, respectively. The categorization procedure of the findings generated from the designed model appears to be performing well.