Options
O'Reilly, Martin
Preferred name
O'Reilly, Martin
Official Name
O'Reilly, Martin
Research Output
Now showing 1 - 10 of 12
- PublicationMobile App to Streamline the Development of Wearable Sensor-Based Exercise Biofeedback Systems: System Development and EvaluationBackground: Biofeedback systems that use inertial measurement units (IMUs) have been shown recently to have the ability toobjectively assess exercise technique. However, there are a number of challenges in developing such systems; vast amounts ofIMU exercise datasets must be collected and manually labeled for each exercise variation, and naturally occurring techniquedeviations may not be well detected. One method of combatting these issues is through the development of personalized exercisetechnique classifiers.Objective: We aimed to create a tablet app for physiotherapists and personal trainers that would automate the development ofpersonalized multiple and single IMU-based exercise biofeedback systems for their clients. We also sought to complete apreliminary investigation of the accuracy of such individualized systems in a real-world evaluation.Methods: A tablet app was developed that automates the key steps in exercise technique classifier creation through synchronizingvideo and IMU data collection, automatic signal processing, data segmentation, data labeling of segmented videos by an exerciseprofessional, automatic feature computation, and classifier creation. Using a personalized single IMU-based classification system,15 volunteers (12 males, 3 females, age: 23.8 [standard deviation, SD 1.8] years, height: 1.79 [SD 0.07] m, body mass: 78.4 [SD9.6] kg) then completed 4 lower limb compound exercises. The real-world accuracy of the systems was evaluated.Results: The tablet app successfully automated the process of creating individualized exercise biofeedback systems. Thepersonalized systems achieved 89.50% (1074/1200) accuracy, with 90.00% (540/600) sensitivity and 89.00% (534/600) specificityfor assessing aberrant and acceptable technique with a single IMU positioned on the left thigh.Conclusions: A tablet app was developed that automates the process required to create a personalized exercise techniqueclassification system. This tool can be applied to any cyclical, repetitive exercise. The personalized classification model displayedexcellent system accuracy even when assessing acute deviations in compound exercises with a single IMU.
304 - PublicationUse of body worn sensors to predict ankle injuries using screening toolsBackground The Single Leg Squat (SLS) is an important screening tool in predicting those at an increased risk of ankle injuries as it relates to landing, running and cutting tasks. However, clinical analysis of this exercise is often completed visually with relatively poor intra-rater reliability. More detailed analysis of SLS completed in biomechanics laboratories is time-consuming and costly. Recent developments in body worn sensors may allow for quick assessments that produce valid and reliable data.Objective To explore a model for leveraging data obtained from wearable sensors to aid in ankle injury risk factor screening.Design A single case study design, with qualitative analysis of quantitative data.Setting University research laboratory.Participants A single participant (female, age = 24 years; height = 158 cm, body mass = 47 kg) was chosen. The participant was familiar with the SLS exercise and had completed it as part of their exercise routine for the past year.Interventions The participant completed 10 left SLS repetitions. These were recorded using the sensors and repetitions where the participant lost balance were noted. Loss of balance was defined as when the subject was unable to maintain single leg stance during the downward or upward phase of the movement and placed their other foot on the ground for support.Main outcome measurements Visual analysis showed signals from the wearable sensors (accelerometer Y and gyroscope Z) were altered when the participant lost their balance compared to signals obtained when the participant maintained balance.Conclusions These preliminary results indicate that body worn sensors may be able to automatize screening tools such as the SLS. An automated system for characterising and quantifying deviations from good form could be developed to aid clinicians and researchers. Such a system would provide objective and reliable data to clinicians and allow researchers to analyse movements quicker and in a more naturalistic setting.
253 - PublicationInertial sensory data provides depth to clinical measures of dynamic balance(BMJ, 2017-09-17)
; ; ; ; Objectives: Establish the role a single inertial sensor may play in the objective quantification of dynamic postural stability following acute ankle injuries.Background The Y Balance test (YBT) is one of the most commonly utilised clinical dynamic balance assessments. Research has demonstrated the utility of the YBT in identifying balance deficits in those with acute ankle injuries and chronic ankle instability. However, reach distances fail to provide information relating to the quality of balance strategy and dynamic stability. Motion capture systems are often employed to provide micro-level detail pertaining to an individuals postural stability. However, such systems are expensive, lack accessibility, hinder natural movement and require extensive processing expertise. The addition of inertial sensors may allow for the inexpensive, accessible quantification of postural stability in an unconstrained environment.Case Description Forty-two elite under-20 rugby union players were recruited as part of a wider study. Two athletes were identified to have sustained acute ankle injuries two weeks previously; one lateral ankle sprain and one deltoid ligament sprain. A single inertial sensor was mounted at the level of the 4th lumbar vertebra. Participants completed four practice YBTs bilaterally, prior to completing 3 recorded YBTs. Reach distance and inertial sensor data were recorded for each reach excursion.Outcomes When compared to the group mean, both athletes demonstrated no clinically meaningful reduction in reach distances for all three reach directions. However, both athletes demonstrated a higher 95% ellipsoid volume of sway than the healthy control group for all three directions of the YBT when completed on their affected limb.Conclusions Preliminary analysis suggests that inertial sensor data may provide information relating to the quality of dynamic postural stability following an acute ankle injury. Further investigation is required to establish the role that such measures may play in the assessment and management of ankle injuries.77 - PublicationEvaluating the use of machine learning in the assessment of joint angle using a single inertial sensorIntroduction: Joint angle measurement is an important objective marker in rehabilitation. Inertial measurement units may provide an accurate and reliable method of joint angle assessment. The objective of this study was to assess whether a single sensor with the application of machine learning algorithms could accurately measure hip and knee joint angle, and investigate the effect of inertial measurement unit orientation algorithms and person-specific variables on accuracy. Methods: Fourteen healthy participants completed eight rehabilitation exercises with kinematic data captured by a 3D motion capture system, used as the reference standard, and a wearable inertial measurement unit. Joint angle was calculated from the single inertial measurement unit using four machine learning models, and was compared to the reference standard to evaluate accuracy. Results: Average root-mean-squared error for the best performing algorithms across all exercises was 4.81 (SD ¼ 1.89). The use of an inertial measurement unit orientation algorithm as a pre-processing step improved accuracy; however, the addition of person-specific variables increased error with average RMSE 4.99 (SD ¼ 1.83). Conclusions: Hip and knee joint angle can be measured with a good degree of accuracy from a single inertial measurement unit using machine learning. This offers the ability to monitor and record dynamic joint angle with a single sensor outside of the clinic.
249 - PublicationA Wearable Sensor-Based Exercise Biofeedback System: Mixed Methods Evaluation of Formulift(JMIR Publications, 2018-01-31)
; ; ; Background: Formulift is a newly developed mobile health (mHealth) app that connects to a single inertial measurement unit (IMU) worn on the left thigh. The IMU captures users movements as they exercise, and the app analyzes the data to count repetitions in real time and classify users exercise technique. The app also offers feedback and guidance to users on exercising safely and effectively. Objective: The aim of this study was to assess the Formulift system with three different and realistic types of potential users (beginner gym-goers, experienced gym-goers, and qualified strength and conditioning [S&C] coaches) under a number of categories: (1) usability, (2) functionality, (3) the perceived impact of the system, and (4) the subjective quality of the system. It was also desired to discover suggestions for future improvements to the system. Methods: A total of 15 healthy volunteers participated (12 males; 3 females; age: 23.8 years [SD 1.80]; height: 1.79 m [SD0.07], body mass: 78.4 kg [SD 9.6]). Five participants were beginner gym-goers, 5 were experienced gym-goers, and 5 were qualified and practicing S&C coaches. IMU data were first collected from each participant to create individualized exercise classifiers for them. They then completed a number of non exercise-related tasks with the app. Following this, a workout was completed using the system, involving squats, dead lifts, lunges, and single-leg squats. Participants were then interviewed about their user experience and completed the System Usability Scale (SUS) and the user version of the Mobile Application Rating Scale (uMARS). Thematic analysis was completed on all interview transcripts, and survey results were analyzed. Results: Qualitative and quantitative analysis found the system has good to excellent usability. The system achieved a mean (SD) SUS usability score of 79.2 (8.8). Functionality was also deemed to be good, with many users reporting positively on the systems repetition counting, technique classification, and feedback. A number of bugs were found, and other suggested changes to the system were also made. The overall subjective quality of the app was good, with a median star rating of 4 out of 5 (interquartile range, IQR: 3-5). Participants also reported that the system would aid their technique, provide motivation, reassure them, and help them avoid injury. Conclusions: This study demonstrated an overall positive evaluation of Formulift in the categories of usability, functionality, perceived impact, and subjective quality. Users also suggested a number of changes for future iterations of the system. These findings are the first of their kind and show great promise for wearable sensor-based exercise biofeedback systems.303Scopus© Citations 14 - PublicationReliability, Validity and Utility of Inertial Sensor Systems for Postural Control Assessment in Sport Science and Medicine Applications: A Systematic ReviewBackground Recent advances in mobile sensing and computing technology have provided a means to objectively and unobtrusively quantify postural control. This has resulted in the rapid development and evaluation of a series of wearable inertial sensor-based assessments. However, the validity, reliability and clinical utility of such systems is not fully understood. Objectives This systematic review aims to synthesise and evaluate studies that have investigated the ability of wearable inertial sensor systems to validly and reliably quantify postural control performance in sports science and medicine applications. Methods A systematic search strategy utilising the PRISMA guidelines was employed to identify eligible articles through ScienceDirect, Embase and PubMed databases. In total, 47 articles met the inclusion criteria and were evaluated and qualitatively synthesised under two main headings: measurement validity and measurement reliability. Furthermore, studies that investigated the utility of these systems in clinical populations were summarised and discussed. Results After duplicate removal, 4374 articles were identified with the search strategy, with 47 papers included in the final review. In total, 28 studies investigated validity in healthy populations, and 15 studies investigated validity in clinical populations; 13 investigated the measurement reliability of these sensor-based systems. Conclusions The application of wearable inertial sensors for sports science and medicine postural control applications is an evolving field. To date, research has primarily focused on evaluating the validity and reliability of a heterogeneous set of assessment protocols, in a laboratory environment. While researchers have begun to investigate their utility in clinical use cases such as concussion and musculoskeletal injury, most studies have leveraged small sample sizes, are of low quality and use a variety of descriptive variables, assessment protocols and sensor-mounting locations. Future research should evaluate the clinical utility of these systems in large high-quality prospective cohort studies to establish the role they may play in injury risk identification, diagnosis and management. This systematic review was registered with the International Prospective Register of Systematic Reviews on 10 August 2018 (PROSPERO registration: CRD42018106363): https://www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=106363.
964Scopus© Citations 45 - PublicationInertial Sensor Technology Can Capture Changes in Dynamic Balance Control during the Y Balance TestIntroduction: The Y Balance Test (YBT) is one of the most commonly utilised clinical dynamicbalance assessments. Research has demonstrated the utility of the YBT in identifying balancedeficits in individuals following lower limb injury. However, quantifying dynamic balancebased on reach distances alone fails to provide potentially important information related tothe quality of movement control and choice of movement strategy during the reaching action.The addition of an inertial sensor to capture more detailed motion data may allow for the inexpensive,accessible quantification of dynamic balance control during the YBT reach excursions.As such, the aim of this study was to compare baseline and fatigued dynamic balancecontrol, using reach distances and 95EV (95% ellipsoid volume), and evaluate the ability of95EV to capture alterations in dynamic balance control, which are not detected by YBT reachdistances. Methods: As part of this descriptive laboratory study, 15 healthy participants completedrepeated YBTs at 20, 10, and 0 min prior to and following a modified 60-s Wingate testthat was used to introduce a short-term reduction in dynamic balance capability. Dynamicbalance was assessed using the standard normalised reach distance method, while dynamicbalance control during the reach attempts was simultaneously measured by means of the95EV derived from an inertial sensor, worn at the level of the 4th lumbar vertebra. Results:Intraclass correlation coefficients for the inertial sensor-derived measures ranged from 0.76to 0.92, demonstrating strong intrasession test-retest reliability. Statistically significant altera-tions (p < 0.05) in both reach distance and the inertial sensor-derived 95EV measure wereobserved immediately post-fatigue. However, reach distance deficits returned to baseline levelswithin 10 min, while 95EV remained significantly increased (p < 0.05) beyond 20 min forall 3 reach distances. Conclusion: These findings demonstrate the ability of an inertial sensorderivedmeasure to quantify alterations in dynamic balance control, which are not capturedby traditional reach distances alone. This suggests that the addition of an inertial sensor tothe YBT may provide clinicians and researchers with an accessible means to capture subtlealterations in motor function in the clinical setting.
227Scopus© Citations 16 - PublicationInterpretable Time Series Classification using Linear Models and Multi-resolution Multi-domain Symbolic Representations(Springer, 2019-05-21)
; ; ; ; The time series classification literature has expanded rapidly over the last decade, with many new classification approaches published each year. Prior research has mostly focused on improving the accuracy and efficiency of classifiers, with interpretability being somewhat neglected. This aspect of classifiers has become critical for many application domains and the introduction of the EU GDPR legislation in 2018 is likely to further emphasize the importance of interpretable learning algorithms. Currently, state-of-the-art classification accuracy is achieved with very complex models based on large ensembles (COTE) or deep neural networks (FCN). These approaches are not efficient with regard to either time or space, are difficult to interpret and cannot be applied to variable-length time series, requiring pre-processing of the original series to a set fixed-length. In this paper we propose new time series classification algorithms to address these gaps. Our approach is based on symbolic representations of time series, efficient sequence mining algorithms and linear classification models. Our linear models are as accurate as deep learning models but are more efficient regarding running time and memory, can work with variable-length time series and can be interpreted by highlighting the discriminative symbolic features on the original time series. We advance the state-of-the-art in time series classification by proposing new algorithms built using the following three key ideas: (1) Multiple resolutions of symbolic representations: we combine symbolic representations obtained using different parameters, rather than one fixed representation (e.g., multiple SAX representations); (2) Multiple domain representations: we combine symbolic representations in time (e.g., SAX) and frequency (e.g., SFA) domains, to be more robust across problem types; (3) Efficient navigation in a huge symbolic-words space: we extend a symbolic sequence classifier (SEQL) to work with multiple symbolic representations and use its greedy feature selection strategy to effectively filter the best features for each representation. We show that our multi-resolution multi-domain linear classifier (mtSS-SEQL+LR) achieves a similar accuracy to the state-of-the-art COTE ensemble, and to recent deep learning methods (FCN, ResNet), but uses a fraction of the time and memory required by either COTE or deep models. To further analyse the interpretability of our classifier, we present a case study on a human motion dataset collected by the authors. We discuss the accuracy, efficiency and interpretability of our proposed algorithms and release all the results, source code and data to encourage reproducibility.897Scopus© Citations 46 - PublicationTechnology in Rehabilitation: Comparing Personalised and Global Classification Methodologies in Evaluating the Squat Exercise with Wearable IMUs(Thieme, 2017)
; ; ; ; BACKGROUND: The barbell squat is a popularly used lower limb rehabilitation exercise. It is also an integral exercise in injury risk screening protocols. To date athlete/patient technique has been assessed using expensive laboratory equipment or subjective clinical judgement; both of which are not without shortcomings. Inertial measurement units (IMUs) may offer a low cost solution for the objective evaluation of athlete/patient technique. However, it is not yet known if global classification techniques are effective in identifying naturally occurring, minor deviations in barbell squat technique.OBJECTIVES: The aims of this study were to: (a) determine if in combination or in isolation, IMUs positioned on the lumbar spine, thigh and shank are capable of distinguishing between acceptable and aberrant barbell squat technique; (b) determine the capabilities of an IMU system at identifying specific natural deviations from acceptable barbell squat technique; and (c) compare a personalised (N=1) classifier to a global classifier in identifying the above. METHODS: Fifty-five healthy volunteers (37 males, 18 females, age = 24.21 +/- 5.25 years, height = 1.75 +/- 0.1 m, body mass = 75.09 +/- 13.56 kg) participated in the study. All participants performed a barbell squat 3-repetition maximum max strength test. IMUs were positioned on participants' lumbar spine, both shanks and both thighs; these were utilized to record tri-axial accelerometer, gyroscope and magnetometer data during all repetitions of the barbell squat exercise. Technique was assessed and labelled by a Chartered Physiotherapist using an evaluation framework. Features were extracted from the labelled IMU data. These features were used to train and evaluate both global and personalised random forests classifiers.RESULTS: Global classification techniques produced poor accuracy (AC), sensitivity (SE) and specificity (SP) scores in binary classification even with a 5 IMU set-up in both binary (AC: 64%, SE: 70%, SP: 28%) and multi-class classification (AC: 59%, SE: 24%, SP: 84%). However, utilising personalised classification techniques even with a single IMU positioned on the left thigh produced good binary classification scores (AC: 81%, SE: 81%, SP: 84%) and moderate-to-good multi-class scores (AC: 69%, SE: 70%, SP: 89%).CONCLUSIONS: There are a number of challenges in developing global classification exercise technique evaluation systems for rehabilitation exercises such as the barbell squat. Building large, balanced data sets to train such systems is difficult and time intensive. Minor, naturally occurring deviations may not be detected utilising global classification approaches. Personalised classification approaches allow for higher accuracy and greater system efficiency for end-users in detecting naturally occurring barbell squat technique deviations. Applying this approach also allows for a single-IMU set up to achieve similar accuracy to a multi-IMU setup, which reduces total system cost and maximises system usability.418Scopus© Citations 11 - PublicationAssociation of dynamic balance with sports related concussion: a prospective cohort study(Sage, 2018-12-03)
; ; ; ; Background: Concussion is one of the most common sports-related injuries, with little understood about the modifiable and non-modifiable risk factors. Researchers have yet to evaluate the association between modifiable sensorimotor function variables and concussive injury. Purpose: To investigate the association between dynamic balance performance, a discrete measure of sensorimotor function, and concussive injuries. Study Design: Cohort study (diagnosis); Level of evidence, 3. Methods: A total of 109 elite male rugby union players were baseline tested in dynamic balance performance while wearing an inertial sensor and prospectively followed during the 2016-2017 rugby union season. The sample entropy of the inertial sensor gyroscope magnitude signal was derived to provide a discrete measure of dynamic balance performance. Logistic regression modeling was then used to investigate the association among the novel digital biomarker of balance performance, known risk factors of concussion (concussion history, age, and playing position), and subsequent concussive injury. Results: Participant demographic data (mean 6 SD) were as follows: age, 22.6 6 3.6 years; height, 185 6 6.5 cm; weight, 98.9 612.5 kg; body mass index, 28.9 6 2.9 kg/m2; and leg length, 98.8 6 5.5 cm. Of the 109 players, 44 (40.3%) had a history of concussion, while 21 (19.3%) sustained a concussion during the follow-up period. The receiver operating characteristic analysis for the anterior sample entropy demonstrated a statistically significant area under the curve (0.64; 95% CI, 0.52-0.76; P \ .05), with the cutoff score of anterior sample entropy 1.2, which maximized the sensitivity (76.2%) and specificity (53.4%) for identifying individuals who subsequently sustained a concussion. Players with suboptimal balance performance at baseline were at a 2.81-greater odds (95% CI, 1.02-7.74) of sustaining a concussion during the rugby union season than were those with optimal balance performance, even when controlling for concussion history. Conclusion: Rugby union players who possess poorer dynamic balance performance, as measured by a wearable inertial sensor during the Y balance test, have a 3-times-higher relative risk of sustaining a sports-related concussion, even when controlling for history of concussion. These findings have important implications for research and clinical practice, as it identifies a potential modifiable risk factor. Further research is required to investigate this association in a large cohort consisting of males and females across a range of sports.441Scopus© Citations 16