ParkINN: An Integrated Neural Network Model for Parkinson Detection

Abstract

One common neurological condition Parkinson is one of the diseases which might make it difficult for a patient to live a regular life like other people. It is a progressive neurodegenerative condition that is difficult to detect in the early stages. Traditional EEG-based PD diagnosis relies on arduous, time-consuming feature extraction that is done by hand. The ParkINN (Parkinson Identification Neural Network) has been proposed as a new EEG-based network for Parkinson’s screening that can quickly identify patients suffering from Parkinson’s or early stages of Parkinson’s. The suggested approach uses windowing and long-short term memory (LSTM) architectures for sequence learning, as well as 3 Dimensional Convolutional Neural Networks (CNN) for temporal learning of the EEG signal. The accuracy rate of the proposed 3D CNN-LSTM model is 94.64 percent, which is higher than the findings of the majority of other work in this area.