Technology of neurological disease recognition using gated recurrent unit neural network and internet of things

Abstract

In this paper, authors proposed a neurological disease recognition technique using gated recurrent unit neural network and supporting Internet of Things (IoT), which was checked by taking Alzheimer’s disease (AD) and Parkinson’s disease (PD) as examples. In this method first pre-emphasized and denoised the voice data, then segmented the voice signals with a sliding fixed window using the Hamming window function. Then we were extracted the eGeMAPSv02 voice features from the window signal, fed the features into the gated recurrent unit neural network model for its training, testing and achieve the disease diagnosis. The results of the study showed that despite the limited generalization ability of the gated recurrent unit model, it can still efficiently achieve voice recognition detection of a portion of neurological diseases. The model is implemented on the basis of the IoT platform for building a subsystem of IT diagnostics of patients as part of the smart city project. The code is stored in https://github.com/ HkThinker/Technology-of-neural-disease-recognition.