Mingjing Yang

iGAIT: An interactive accelerometer based gait analysis system

This paper presents a software program (iGAIT) developed in MATLAB, for the analysis of gait patterns extracted from accelerometer recordings. iGAIT provides a user-friendly graphical interface to display and analyse gait acceleration data recorded by an accelerometer attached to the lower back of subjects. The core function of iGAIT is gait feature extraction, which can be used to derive 31 features from acceleration data, including 6 spatio-temporal features, 7 regularity and symmetry features, and 18 spectral features. Features extracted are summarised and displayed on screen, as well as an option to be stored in text files for further review or analysis if required. Another unique feature of iGAIT is that it provides interactive functionality allowing users to manually adjust the analysis process according to their requirements. The system has been tested under Window XP, Vista and Window 7 using three different types of accelerometer data. It is designed for analysis of accelerometer data recorded with sample frequencies ranging from 5 Hz to 200 Hz.

Machine Learning and Statistical Approaches to Support the Discrimination of Neuro-degenerative Diseases Based on Gait Analysis

Amyotrophic lateral sclerosis, Parkinson’s disease and Huntington’s disease are three neuro-degenerative diseases. In all these diseases, severe disturbances of gait and gait initiation are frequently reported. In this paper, we explore the feasibility of using machine learning and statistical approaches to support the discrimination of these three diseases based on gait analysis. A total of three supervised classification methods, namely support vector machine, KStar and Random Forest, were evaluated on a publicly-available gait dataset. The results demonstrate that it is feasible to apply computational classification techniques in characterise these three diseases with the features extracted from gait cycles. Results obtained show that using selected 4 features based on maximum relevance and minimum redundancy strategy can achieve reasonably high classification accuracy while 5 features can achieve the best performance. The continual increase of the number of features does not significantly improve classification performance.

Feature selection and construction for the discrimination of neurodegenerative diseases based on gait analysis

Gait disorder is one symptom of neurodegenerative disease. Using wearable motion sensors to monitor the motor function of patients with neurodegenerative disease has attracted more attention. Research has shown that machine learning techniques can be applied to the classification of neurodegenerative diseases from the gait data recorded by footswitches. In order to identify the most valuable features from 10 raw temporal variables extracted from gait cycles to improve the classification performance, we examine four types of feature selection and construction methods, namely, maximum signal-to-noise ratio based feature selection method, maximum signal-to-noise ratio combined with minimum correlation based feature selection method, maximum prediction power combined with minimum correlation based feature selection method and principal component analysis. Results show that using a set of four features, a relatively high prediction performance has been achieved with classification accuracies ranging from 79.04% to 93.96%. The continual increase of the number of features does not significantly contribute to the improvement of classification performance. This is consistent with clustering-based feature analysis.

A machine learning approach to assessing gait patterns for Complex Regional Pain Syndrome

Complex Regional Pain Syndrome (CRPS) is a condition that causes a long-term burning pain in a limb or part of a limb and it can cause various degrees of the physical functional performance deterioration. Objective assessment of physical functional performance of patients is one critical component to evaluate the therapy outcome for CRPS. This paper aims to investigate the feasibility of assessing the physical performance of patients with Complex Regional Pain Syndrome based on the analysis of gait data recorded by an accelerometer in short walking distances. Ten subjects with CRPS and ten control subjects were recruited. Thirty three features were extracted from each recording. A machine learning method, Multilayer perceptron neural-networks (MLP), was applied to classify the normal and abnormal gait patterns from data obtained on a 2.4m performance evaluation test. The best classification accuracy (99.38%) was achieved using 3 features selected by a step-wise-forward method. To further validate its performance, an independent test set including 14 cases extracted from data obtained on a 20 m performance evaluation test was adopted. A prediction accuracy of 85.7% was obtained.

Assessing accelerometer based gait features to support gait analysis for people with complex regional pain syndrome

In this paper, we explored the feasibility of analysing gait patterns during the Short Physical Performance Battery test by using an accelerometer to record the movement of the subject. 12 subjects with Complex Regional Pain Syndrome (CRPS) and 10 control subjects were recruited in this study. 21 gait features including temporal, frequency, regularity and symmetric information were extracted from each recording. The differences of each feature value on control subjects and patient subjects were assessed and compared. Features were selected based on the signal to noise ratio (SNR) ranking. Multilayer perceptron neural-networks were employed to differentiate between the normal and abnormal gait patterns. The result shows when using five features the best classification accuracy (97.5%) was achieved. It is feasible to discriminate the patients with CRPS from the control subjects using a small set of gait features extracted from walking acceleration data recorded during the SPPB test.

Assessing Gait Patterns of Healthy Adults Climbing Stairs Employing Machine Learning Techniques

So far, stair climbing has not been studied as extensively as gait has, although the significance of the prevention of falling on stairs has been well recognized. Based on acceleration data taken from 25 healthy subjects climbing up and down a set of 13 stairs with an accelerometer placed on the lumbo‐sacral joint, this paper aims to assess gait patterns of younger and older adults climbing stairs using a machine learning approach. A total of 14 gait features were extracted and analyzed. The performance of six representative classification models: Multilayer Perceptron (MLP), KStar, Support Vector Machine (SVM), Naïve Bayesian (NB), C4.5 Decision Trees, and Random Forests were evaluated in terms of their ability to discriminate between younger and older adults climbing up‐ and downstairs. MLP was found to provide the highest accuracy for classification. Accuracy of 95.7% was found for classifying a subject walking either up or down the stairs and an accuracy of 80.6% for classifying whether the subject was younger or older. An evaluation of individual features showed poor performance of classification for younger and older subjects climbing up‐ and downstairs, and in most cases failed to distinguish between the two classes. To access which set of features derived from a triaxial accelerometer can better describe the performance differences between younger and older adults climbing up‐ and downstairs, two feature selection algorithms, sequential feature selection and correlation‐based feature selection, were implemented. Results show that 10 features derived from correlation‐based feature selection were able to produce a 96.8% accuracy for classification between subjects climbing up and down. A subset of seven features achieved a performance of 84.9% accuracy for classification between younger and older subjects.

Machine learning and statistical approaches to assessing gait patterns of younger and older healthy adults climbing stairs

The following study explores the methods for activity recognition of younger and older adults climbing stairs. There is a correlation to health and the level of activity of an individual, which has captured interest in this field in computing science to determine the level of activity of an individual. The focus of the study is the classification of younger and older gait patterns climbing up and down a set of 13 stairs. From using acceleration data from an accelerometer placed on the lumbro-sacral joint; 14 gait features are extracted and analysed. The machine learning algorithms focused in this study are Multilayer Perceptron (MLP), KStar and Support Vector Machine (SVM). An evaluation of performance among three machine learning algorithms was carried out. MLP was found to provide the highest in accuracy for classification. Accuracy of 95.7% was found for classifying a subject walking either up or down the stairs and an accuracy of 80.6% for classifying whether the subject was younger or older. An evaluation of individual features showed poor performance of classification for young and older subjects climbing up and down stairs, and at most cases failed to distinguish between the two classes. However, subsets of features were created using a sequential feature selection algorithm based on feature ranking and individual feature performance. The performance of each subset was recorded and a subset of the top four features achieved an accuracy of 81.7% for classification between young and older subjects. In comparison, 13 features were required to obtain the best performance of 95.7% to distinguish between up and down classes.

Combining feature ranking with PCA: An application to gait analysis

Feature reduction is an effective way to improve the classification performance when machine learning methods are used in gait analysis. In this paper, we proposed a novel hybrid feature reduction method (MSNR&PCA) based on the combination of feature ranking with principle component analysis (PCA). Three feature reduction methods, namely, feature ranking based the value of signal to noise ratio (MSNR), PCA and the proposed hybrid approach (MSNR&PCA), were examined in two gait analysis problems. One gait analysis problem is to differentiate the patients with Neurodegenerative disease from the controls based on the gait data collected by footswitches. The other problem is to discriminate the patients with complex regional pain syndrome (CRPS) from controls based on the gait data collected by an accelerometer. Results showed that the proposed MSNR&PCA achieved best classification performance in two gait datasets. In footswitch data, the highest accuracy (81.78%) was obtained using a feature subset with 4 features generated from original 10 features by MSNR&PCA. In the accelerometer dataset, classification with three features generated from 17 features by MSNR&PCA achieved the best performance with an accuracy of 100%.

Statistical and Machine Learning Approach to Assessing the Environmental Impact on Walking Patterns

It has been recongised that subjects may change their gait pattern when walking in different environments. This paper investigated the impact of walking environments on gait monitoring and analysis. A tri-axial accelerometer attached to subjects lower back was used to record gait pattern while walking in 5 different urban environments (quiet street, busy street, cobbled street, dark street and checkerboard floor). Forty-one young students participated the experiment. For each trial, a total of 33 gait features were extracted, of which 11 were derived from the entire walking trial and 22 were computed for each stride cycle. Statistics analysis showed that 7 out of 11 features extracted from each trial were significantly different across the five environments. The obtained results suggested that different environments have various impacts on gait features extracted from accelerometer data. To further access the impact, a multi-layer perceptrons based hierarchical classification approach was proposed to discriminate stride cycles taken from different walking environments. The classification accuracy of each level ranged from 98.26% to 65.62% with the discrimination of walking in quiet environment achieving the best performance.