Machine Learning-Based Classification of Wheelchair Task Intensity for Injury Risk Prediction

For the manual wheelchair (MWC) user, loss of lower extremity function often places the burden for mobility and activities of daily living on the

There is a high prevalence of shoulder problems in manual wheelchair users (MWUs) with a spinal cord injury. How shoulder load relates to shoulder problems remains unclear. This study aimed to develop a machine-learning-based methodology to estimate the shoulder load in wheelchair-related activities of daily living using wearable sensors. Ten able-bodied participants equipped with five inertial measurement units (IMU) on their thorax, right arm, and wheelchair performed activities exemplary of daily life of MWUs. Electromyography (EMG) was recorded from the long head of the biceps and medial part of the deltoid. A neural network was trained to predict the shoulder load based on IMU and EMG data. Different cross-validation strategies, sensor setups, and model architectures were examined. The predicted shoulder load was compared to the shoulder load determined with musculoskeletal modeling. A subject-specific biLSTM model trained on a sparse sensor setup yielded the most promising results (mean correlation coefficient = 0.74 ± 0.14, relative root-mean-squared error = 8.93% ± 2.49%). The shoulder-load profiles had a mean similarity of 0.84 ± 0.10 over all activities. This study demonstrates the feasibility of using wearable sensors and neural networks to estimate the shoulder load in wheelchair-related activities of daily living.

This study aimed to develop and evaluate the ARM (arm repetitive movement) algorithm using inertial measurement unit (IMU) data to assess repetitive arm motion in manual wheelchair (MWC) users in real-world settings. The algorithm was tested on community data from four MWC users with spinal cord injury and compared with video-based analysis. Additionally, the algorithm was applied to in-home and free-living environment data from two and sixteen MWC users, respectively, to assess its utility in quantifying differences across activities of daily living and between dominant and non-dominant arms. The ARM algorithm accurately estimated active and resting times (>98%) in the community and confirmed asymmetries between dominant and non-dominant arm usage in in-home and free-living environment data. Analysis of free-living environment data revealed that the total resting bout time was significantly longer (P = 0.049) and total active bout time was significantly shorter (P = 0.011) for the non-dominant arm. Analysis of active bouts longer than 10 seconds showed higher total time (P = 0.015), average duration (P = 0.026), and number of movement cycles per bout (P = 0.020) for the dominant side. These findings support the feasibility of using the IMU-based ARM algorithm to assess repetitive arm motion and monitor shoulder disorder risk factors in MWC users during daily activities.

This study aimed to develop and evaluate the ARM (arm repetitive movement) algorithm using inertial measurement unit (IMU) data to assess repetitive arm motion in manual wheelchair (MWC) users in real-world settings. The algorithm was tested on community data from four MWC users with spinal cord injury and compared with video-based analysis. Additionally, the algorithm was applied to in-home and free-living environment data from two and sixteen MWC users, respectively, to assess its utility in quantifying differences across activities of daily living and between dominant and non-dominant arms. The ARM algorithm accurately estimated active and resting times (>98%) in the community and confirmed asymmetries between dominant and non-dominant arm usage in in-home and free-living environment data. Analysis of free-living environment data revealed that the total resting bout time was significantly longer (P = 0.049) and total active bout time was significantly shorter (P = 0.011) for the non-dominant arm. Analysis of active bouts longer than 10 seconds showed higher total time (P = 0.015), average duration (P = 0.026), and number of movement cycles per bout (P = 0.020) for the dominant side. These findings support the feasibility of using the IMU-based ARM algorithm to assess repetitive arm motion and monitor shoulder disorder risk factors in MWC users during daily activities.

Upper extremity pain and injury are among the most common musculoskeletal complications manual wheelchair users face. Assessing the temporal parameters of manual wheelchair propulsion, such as propulsion duration, cadence, push duration, and recovery duration, is essential for providing a deep insight into the mobility, level of activity, energy expenditure, and cumulative exposure to repetitive tasks and thus providing personalized feedback. The purpose of this paper is to investigate the use of inertial measurement units (IMUs) to estimate these temporal parameters by identifying the start and end time of hand contact with the push-rim during each propulsion cycle. We presented a model based on data collected from 23 participants (14 males and 9 females, including 9 experienced manual wheelchair users) to guarantee the reliability and generalizability of our method. The obtained outcomes from our IMU-based model were then compared against an instrumented wheelchair (SMARTWheel) as a reference criterion. The results illustrated that our model was able to accurately detect hand contact and hand release and predict temporal parameters, including the push duration and recovery duration in manual wheelchair users, with the mean error ± standard deviation of 10 ± 60 milliseconds and -20 ± 80 milliseconds, respectively. The findings of this study demonstrate the potential of hand-mounted IMUs as a reliable and objective tool for analyzing temporal parameters in manual wheelchair propulsion. IMUs offer significant strides towards inclusivity and accessibility due to their portability and user-friendliness and can democratize health monitoring of manual wheelchair users by making it accessible to a broader range of users compared to traditional technologies.

Estimation of manual wheelchair-based activities in the free-living environment using a neural network model with inertial body-worn sensors

Background: Manual wheelchair (MWC) users with spinal cord injuries (SCI) are at a significantly higher risk of experiencing rotator cuff pathology than able-bodied individuals. A deeper understanding of where the arm is used dynamically within the humeral workspace during daily life may help explain why MWC users have higher shoulder pathology rates than able-bodied individuals. The purpose of this study was to report the daily percentage and consecutive durations MWC users and matched able-bodied individuals (controls) spent static and dynamic across the humeral elevation workspace.Methods: MWC users with SCI and controls wore three inertial measurement units on their bilateral arms and torso for 1 or 2 days. The percentages of time and average consecutive duration individuals were static or dynamic while in five humeral elevation ranges (0–30°, 30–60°, 60–90°, 90–120°, and >120°) were calculated and compared between cohorts.Results: Forty-four MWC users (10 females, age: 42.8 ± 12.0, time since injury: 12.3 ± 11.5) and 44 age- and sex-matched controls were enrolled. The MWC cohort spent significantly more time dynamic in 60–90° (p = 0.039) and 90–120° (p = 0.029) and had longer consecutive dynamic periods in 30–60° (p = 0.001), 60–90° (p = 0.027), and 90–120° (p = 0.043) on the dominant arm. The controls spent significantly more time dynamic in 0–30° of humeral elevation (p < 0.001) on both arms. Although the average consecutive static durations were comparable between cohorts across all humeral elevation ranges, the MWC cohort spent a significantly higher percentage of their day static in 30–60° of humeral elevation than controls (dominant: p = 0.001, non-dominant: p = 0.01). The MWC cohort had a moderate association of increased age with decreased time dynamic in 30–60° for both arms.Discussion: Remote data capture of arm use during daily life can aid in understanding how arm function relates to shoulder pathology that follows SCI and subsequent MWC use. MWC users spent more time dynamic in higher elevations than controls, and with age, dynamic arm use decreased in the 30–60° humeral elevation range. These results may exemplify effects of performing activities from a seated position and of age on mobility.

Quantitative evaluation of motor function is of great demand for monitoring clinical outcome of applied interventions and further guiding the establishment of therapeutic protocol. This study proposes a novel framework for evaluating upper limb motor function based on data fusion from inertial measurement units (IMUs) and surface electromyography (EMG) sensors. With wearable sensors worn on the tested upper limbs, subjects were asked to perform eleven straightforward, specifically designed canonical upper-limb functional tasks. A series of machine learning algorithms were applied to the recorded motion data to produce evaluation indicators, which is able to reflect the level of upper-limb motor function abnormality. Sixteen healthy subjects and eighteen stroke subjects with substantial hemiparesis were recruited in the experiment. The combined IMU and EMG data yielded superior performance over the IMU data alone and the EMG data alone, in terms of decreased normal data variation rate (NDVR) and improved determination coefficient (DC) from a regression analysis between the derived indicator and routine clinical assessment score. Three common unsupervised learning algorithms achieved comparable performance with NDVR around 10% and strong DC around 0.85. By contrast, the use of a supervised algorithm was able to dramatically decrease the NDVR to 6.55%. With the proposed framework, all the produced indicators demonstrated high agreement with the routine clinical assessment scale, indicating their capability of assessing upper-limb motor functions. This study offers a feasible solution to motor function assessment in an objective and quantitative manner, especially suitable for home and community use.

Manual wheelchair propulsion is a physically demanding task associated with upper extremity pain and pathology. Shoulder pain is reported in over 25% of pediatric manual wheelchairs users, and this number rises over the lifespan. Upper extremity biomechanics in adults has been associated with shoulder pain and pathology; however, few studies have investigated upper extremity joint dynamics in children. Furthermore, sex may be a critical factor that is currently unexplored with regard to pediatric wheelchair mobility. To investigate differences in upper extremity joint dynamics between pediatric male and female manual wheelchair users with spinal cord injury (SCI) during wheelchair propulsion. Novel instrumented wheelchair hand-rims synchronized with optical motion capture were used to acquire upper extremity joint dynamics of 20 pediatric manual wheelchair users with SCI (11 males, 9 females). Thorax, sternoclavicular, acromioclavicular, glenohumeral, elbow, and wrist joint kinematics and kinetics were calculated during wheelchair propulsion. Linear mixed models were used to assess differences between sexes. Females exhibited significantly greater peak forearm pronation (p = .007), normalized wrist lateral force (p = .03), and normalized elbow posterior force (p = .04) than males. Males exhibited significantly greater peak sternoclavicular joint retraction (p < .001) than females. No significant differences between males and females were observed for the glenohumeral joint (p > .012). This study found significant differences in upper extremity joint dynamics between sexes during manual wheelchair propulsion. Our results underscore the importance of considering sex when evaluating pediatric wheelchair mobility and developing comprehensive wheelchair training interventions for early detection and prevention of upper extremity pain and pathology.

There has been a proliferation of smartphones, smart watches, and wearable sensors, making them ubiquitous in our daily lives. Mobile sensors have found widespread use due to their ever-decreasing cost, ease of deployment, and ability to provide continuous monitoring as opposed to sensors installed at fixed locations. Various techniques have been proposed for fall detection, gait analysis, activity monitoring, and heart rate and sleep sensing by wearable sensors and mobile phones. Compared to works that use inertial measurement unit (IMU) data or static cameras installed in the environment, there has been relatively less work using egocentric videos, meaning providing the first-person view from wearable cameras. Moreover, most of the existing studies on egocentric videos are based on only one sensor modality, namely the camera. There have been even fewer approaches that combine egocentric video data with IMU data. In this chapter, we will describe three different applications using wearable cameras together with IMU data. First, we will present an overview of a fall detection system using wearable devices, e.g., smartphones and tablets, equipped with cameras and accelerometers. Since the portable device is worn by the subject, monitoring is not limited to confined areas, and extends to wherever the subject may travel, as opposed to static sensors installed in certain rooms. Second, we will present an autonomous and robust method for counting footsteps, and tracking and calculating stride length by using both accelerometer and camera data from smartphones or Google™ glass. To provide higher precision, instead of using a preset stride length, the proposed method calculates the distance traveled with each step by using the camera data. This method is compared with the commercially available accelerometer-based step counter apps. The results show that the proposed method provides a significant increase in accuracy, and has the lowest average error rate both in number of steps taken and the distance traveled. Finally, we will provide an overview of a robust and autonomous method to detect activities with more details and context by using accelerometer and egocentric video data obtained from a smartphone.

Manual wheelchair users (MWUs) are prone to a sedentary life that can negatively affect their physical and cardiovascular health, making regular assessment important to identify appropriate interventions and lifestyle modifications. One mean of assessing MWUs’ physical health is the 6 min push test (6MPT), where the user propels themselves as far as they can in six minutes. However, reliance on observer input introduces subjectivity, while limited quantitative data inhibit comprehensive assessment. Incorporating sensors into the 6MPT can address these limitations. Here, ten MWUs performed the 6MPT with additional sensors: two inertial measurement units (IMUs)—one on the wheelchair and one on the wrist together with a heart rate wristwatch. The conventional measurements of distance and laps were recorded by the observer, and the IMU data were used to calculate laps, distance, speed, and cadence. The results demonstrated that the IMU can provide the metrics of the traditional 6MPT with strong significant correlations between calculated laps and observer lap counts (r = 0.947, p < 0.001) and distances (r = 0.970, p < 0.001). Moreover, heart rate during the final minute was significantly correlated with calculated distance (r = 0.762, p = 0.017). Enhanced 6MPT assessment can provide objective, quantitative, and comprehensive data for clinicians to effectively inform interventions in rehabilitation.

Basketball is a major sport worldwide among different age groups that leads to a high frequency of injuries at multiple body sites. Upper and lower extremities and lower back are common pain sites in basketball players; however, there is little information about the relationship between upper or lower extremity pain and lower back pain. This study elucidated the associations between upper extremity (shoulder and elbow) pain and lower back pain (LBP) among young basketball players. We conducted a cross-sectional study using self-reported questionnaires mailed to 25,669 young athletes; the final study population comprised 590 basketball players, and their median age was 13 years (range: 6-15 years). The point prevalence rates of lower back, shoulder, elbow, and upper extremity pain among young basketball players were 12.9% (76/590), 4.6% (27/590), 2.7% (16/590), and 7.1% (42/590), respectively. Multivariate logistic regression analyses revealed that upper extremity pain was significantly associated with LBP (adjusted odds ratio [OR]: 7.86; 95% confidential interval [CI], 3.93-15.72). Shoulder pain was significantly associated with training per week (> 4 days) (adjusted OR: 4.15; 95% CI: 1.29-13.40) and LBP (adjusted OR: 13.77; 95% CI: 5.70-33.24). This study indicates that upper extremity and shoulder pain is associated with LBP among young basketball players. Assessing for lower back pain, as well as elbow and/or shoulder pain, may help prevent severe injuries in young basketball players. In conclusion, parents and coaches should be properly re-educated to help improve lower back, upper extremity, and shoulder pain among young basketball players.

IntroductionPhysics simulation has emerged as a promising approach to generate virtual Inertial Measurement Unit (IMU) data, offering a solution to reduce the extensive cost and effort of real-world data collection. However, the fidelity of virtual IMU depends heavily on the quality of the source motion data, which varies with motion capture setups. We hypothesize that improving virtual IMU fidelity is crucial to fully harness the potential of physics simulation for virtual IMU data generation in training Human Activity Recognition (HAR) models.MethodTo investigate this, we introduce WIMUSim, a 6-axis wearable IMU simulation framework designed to accurately parameterize real IMU properties when deployed on people. WIMUSim models IMUs in wearable sensing using four key parameters: Body (skeletal model), Dynamics (movement patterns), Placement (device positioning), and Hardware (IMU characteristics). Using these parameters, WIMUSim simulates virtual IMU through differentiable vector manipulations and quaternion rotations. A key novelty enabled by this approach is the identification of WIMUSim parameters using recorded real IMU data through gradient descent-based optimization, starting from an initial estimate. This process enhances the fidelity of the virtual IMU by optimizing the parameters to closely mimic the recorded IMU data. Adjusting these identified parameters allows us to introduce physically plausible variabilities.ResultsOur fidelity assessment demonstrates that WIMUSim accurately replicates real IMU data with optimized parameters and realistically simulates changes in sensor placement. Evaluations using exercise and locomotion activity datasets confirm that models trained with optimized virtual IMU data perform comparably to those trained with real IMU data. Moreover, we demonstrate the use of WIMUSim for data augmentation through two approaches: Comprehensive Parameter Mixing, which enhances data diversity by varying parameter combinations across subjects, outperforming models trained with real and non-optimized virtual IMU data by 4–10 percentage points (pp); and Personalized Dataset Generation, which customizes augmented datasets to individual user profiles, resulting in average accuracy improvements of 4 pp, with gains exceeding 10 pp for certain subjects.DiscussionThese results underscore the benefit of high-fidelity virtual IMU data and WIMUSim's utility in developing effective data generation strategies, alleviating the challenge of data scarcity in sensor-based HAR.

Understanding development of bimanual upper limb (UL) activities in both typical and atypical conditions in children is important for: i) tailoring rehabilitation programs, ii) monitoring progress, iii) determining outcomes and iv) evaluating effectiveness of treatment/rehabilitation. Recent technological advances, such as wearable sensors, offer possibilities to perform standard medical monitoring. Body-worn motion sensors, mainly accelerometers, have shown very promising results but, so far, these studies have mainly focused on adults. The main aim of this review was to report the evidence of UL activity of both typically developing (TD) children and children with neurodevelopmental disorders (NDDs) that are reliably reported and comparable, using a combination of multiple wearable inertial sensors, both in laboratory and natural settings. Articles were selected from three research databases (PubMed, Web of Science and EBSCO). Included studies reported data on children aged 0–20 years old simultaneously wearing at least two inertial sensors on upper extremities. The collected and reported data were relevant in order to describe the amount of physical activity performed by the two ULs separately. A total of 21 articles were selected: 11 including TD, and 10 regarding NDDs. For each article, a review of both clinical and technical data was performed. We considered inertial sensors used for following aims: (i) to establish activity intensity cut-points; (ii) to investigate validity and reliability of specified markers, placement and/or number of inertial sensors; (iii) to evaluate duration and intensity of natural UL movements, defined motor tasks and tremor; and (iv) to assess efficacy of certain rehabilitation protocols. Our conclusions were that inertial sensors are able to detect differences in use between both hands and that all reviewed studies support use of accelerometers as an objective outcome measure, appropriate in assessing UL activity in young children with NDDs and determining intervention effectiveness. Further research on responsiveness to interventions and consistency with use in real-world settings is needed. This information could be useful in planning UL rehabilitation strategies.

Shoulder tendon pathology is 10 times more likely in chronic manual wheelchair (MWC) users than in controls [1]. The increase in pathology is often attributed to a narrowing of the subacromial space, which is smallest between 30-60° of humeral elevation (HE) [2]. MWC users spend significantly more time in 30-60° of HE than controls [3]; however, their arm activity while in this workspace is unknown. PURPOSE: To determine the active and sedentary time of the arms for MWC users and controls while in 30-60° of HE. METHODS: Under IRB approval, participants wore three wireless inertial measurement units (Emerald, APDM, Inc.; 128 Hz) on their bilateral upper arms and torso for one to two days. Custom MATLAB (MathWorks, Inc.) code calculated the HE [3] and acceleration-based activity levels [4] of both arms for each second. The percentage of daily wear time each participant spent in sedentary and active time in 30-60° of HE was calculated for each arm. Separate paired t-tests were used to determine differences between cohorts (α<0.05). RESULTS: 34 MWC users (sex: 6f, age: 43 ± 13, injury level: C6-L1, years since injury 11 ± 11) and 34 controls (sex: 6f, age: 43 ± 13) were enrolled. MWC users and controls spent similar amounts of time active; however MWC users spent a significantly higher percentage of time sedentary.CONCLUSIONS: Although MWC users spend more time in 30-60° of humeral elevation, the majority of this time is sedentary, emphasizing the importance of understanding other factors such as arm loading and velocity of movement in this population. MWC users may be loading their arms more while in sedentary (i.e., resting condition) and active (i.e., propulsion) conditions, which may contribute to the increase in pathology. [1] Akbar M et al., 2010. JBJS, 92:23-30.[2] Larence R et al., 2017. J Orthop Res, 35:2329-37. [3] Goodwin B et al., 2019. Under Review. [4] Lugade V et al., 2014. Med Eng Phys, 36:169-76. Supported by NIH R01HD84423-01 and NCATS UL1 TR002377