Classification of short time series in early Parkinson s disease with deep learning of fuzzy recurrence plots

Abstract

There are many techniques using sensors and wearable devices for detecting and monitoring patients with Parkinson's disease (PD). A recent development is the utilization of human interaction with computer keyboards for analyzing and identifying motor signs in the early stages of the disease. Current designs for classification of time series of computer-key hold durations recorded from healthy control and PD subjects require the time series of length to be considerably long. With an attempt to avoid discomfort to participants in performing long physical tasks for data recording, this paper introduces the use of fuzzy recurrence plots of very short time series as input data for the machine training and classification with long short-term memory (LSTM) neural networks. Being an original approach that is able to both significantly increase the feature dimensions and provides the property of deterministic dynamical systems of very short time series for information processing carried out by an LSTM layer architecture, fuzzy recurrence plots provide promising results and outperform the direct input of the time series for the classification of healthy control and early PD subjects.

The Role of Neural Network for the Detection of Parkinson’s Disease: A Scoping Review

“…Around half of the included studies used convolution neural networks ( n = 37); afterward, other neural networks ( n = 31) were implemented in the included studies, followed by artificial neural networks (ANNs) ( n = 10), recurrent neural networks (RNNs) ( n = 9), and fuzzy neural networks (FNNs), as shown in Table 3 . In the end, the most imitated neural network architecture in the included studies was LSTM ( n = 11) [ 6 , 34 , 36 , 38 , 40 , 65 , 70 , 74 , 77 , 80 , 83 ], VGG ( n = 3) [ 18 , 27 , 58 ], and DNN ( n = 6) [ 34 , 35 , 60 , 91 , 92 , 103 ]. Recently, with the developments of new techniques such as convolutional neural network [ 101 ] and transfer learning [ 63 ], deep learning gained significant advances in the computer vision tasks, e.g., ImageNet [ 77 ].…”

Time–frequency time–space LSTM for robust classification of physiological signals

“…Because LSTM networks can capture long-term temporal dependencies, they have been applied to provide solutions for many difficult problems in bioinformatics and computational biology 4 . As a state-of-the-art method for learning physiological models for disease prediction, many applications of LSTM and other deep-learning networks have recently been reported in literature, such as classifying electroencephalogram (EEG) signals in emotion, motor imagery, mental workload, seizure, sleep stage, and event related potentials 5 , non-EEG signals in Parkinson’s disease (PD) 6 , learning and synthesis of respiration, electromyograms, and electrocardiograms (ECG) signals 7 , decoding of gait phases using EEG 8 , and early prediction of stress, health, and mood using wearable sensor data 9 .…”

“…While an RP is a binary visualization of recurrences of states of a dynamical system at certain pairs of time, a fuzzy recurrence plot (FRP) 25 displays the visualization as a grayscale image. Because of being much richer in texture than RPs, the technique of FRPs of time series is a preferred approach for texture analysis and has been successfully applied to extract texture features for pattern recognition, including classification of PD and control subjects using deep learning 6 , 26 , tensor decomposition 27 , and SVMs 28 ; and other neuro-degenerative diseases 29 .…”

PFC: A Novel Perceptual Features-Based Framework for Time Series Classification

“…Time series classification [2][3][4] is an essential task that has attracted widespread attention. Normally, time series classification refers to assign time series patterns to a specific category, for example, judge whether it will rain or not through a series of temperature data [5] or determine whether the patient has Parkinson's disease through a period of physiological data [6,7]. Dau et al [8] proposed UCR Time Series Classification Archive (UCR) for this task, including 128 datasets from different fields such as ECG, Sensor, and Image.…”