Introduction

The ability to walk over various surfaces such as cobblestones, slopes or stairs is a very patient centric and clinically meaningful mobility outcome. Current wearable sensors only measure step counts or walking speed regardless of such context relevant for assessing gait function. This study aims to improve deep learning (DL) models to classify surfaces of walking by altering and comparing model features and sensor configurations.

Introduction

Transosseous flexion-distraction injuries of the spine typically require surgical intervention by stabilizing the fractured vertebra during healing with a pedicle-screw-rod constructs. As healing is taking place the load shifts from the implant back to the spine. Monitoring the load-induced deflection of the rods over time would allow quantifiable postoperative assessment of healing progress without the need for radiation exposure or frequent hospital visits. This approach, previously demonstrated to be effective in assessing fracture healing in long bones and monitoring posterolateral spinal fusion in sheep, is now being investigated for its potential in evaluating lumbar vertebra transosseous fracture healing.

Gait measurements can vary due to various intrinsic and extrinsic factors, and this variability becomes more pronounced using inertial sensors in a free-living environment. Therefore, identifying and quantifying the sources of variability is essential to ensure measurement reliability and maintain data quality.

This study aimed to determine the variability of daily accelerations recorded by an inertial sensor in a group of healthy individuals. Ten participants, four males and six females, with a mean age of 50 years (range: 29–61) and BMI of 26.9 kg/m² (range: 21.4–36.8), were included. A single accelerometer continuously recorded lower limb accelerations over two weeks. We extracted and analyzed the accelerations of three consecutive strides within walking bouts if the time difference between the bouts was more than two hours. Multivariate mixed-effects modeling was performed on both the discretized acceleration waveforms at 101 points (0–100) and the harmonics of the signals in the frequency domain to determine the variance components for different subjects, days, bouts, and steps as the random effect variables. Intraclass correlation coefficients (ICCs) were calculated for between-day, between-bout, and between-step comparisons.

The results showed that the ICCs for the between-day, between-bout, and between-step comparisons were 0.73, 0.82, 0.99 for the vertical axis; 0.64, 0.75, 0.99 for the anteroposterior axis; and 0.55, 0.96, 0.97 for the mediolateral axis. For the signal harmonics, the respective ICCs were 0.98, 0.98, 0.99 for the vertical axis; 0.54, 0.93, 0.98 for the anteroposterior axis; and 0.69, 0.78, 0.95 for the mediolateral axis.

Overall, this study demonstrated that accelerations recorded continuously for multiple days in a free-living environment exhibit high variability, mainly between days, and some variability arising from differences between walking bouts during different times within days. However, reliable and repeatable gait measurements can be obtained by identifying and quantifying the sources of variability.

Wearable inertial sensors can detect abnormal gait associated with knee or hip osteoarthritis (OA). However, few studies have compared sensor-derived gait parameters between patients with hip and knee OA or evaluated the efficacy of sensors suitable for remote monitoring in distinguishing between the two. Hence, our study seeks to examine the differences in accelerations captured by low-frequency wearable sensors in patients with knee and hip OA and classify their gait patterns.

We included patients with unilateral hip and knee OA. Gait analysis was conducted using an accelerometer ipsilateral with the affected joint on the lateral distal thighs. Statistical parametric mapping (SPM) was used to compare acceleration signals. The k-Nearest Neighbor (k-NN) algorithm was trained on 80% of the signals' Fourier coefficients and validated on the remaining 20% using 10-fold cross-validation to classify the gait patterns into hip and knee OA.

We included 42 hip OA patients (19 females, age 70 [63–78], BMI of 28.3 [24.8–30.9]) and 59 knee OA patients (31 females, age 68 [62–74], BMI of 29.7 [26.3–32.6]). The SPM results indicated that one cluster (12–20%) along the vertical axis had accelerations exceeding the critical threshold of 2.956 (p=0.024). For the anteroposterior axis, three clusters were observed exceeding the threshold of 3.031 at 5–19% (p = 0.0001), 39–54% (p=0.00005), and 88–96% (p = 0.01). Regarding the mediolateral axis, four clusters were identified exceeding the threshold of 2.875 at 0–9% (p = 0.02), 14–20% (p=0.04), 28–68% (p < 0.00001), and 84–100% (p = 0.004). The k-NN model achieved an AUC of 0.79, an accuracy of 80%, and a precision of 85%.

In conclusion, the Fourier coefficients of the signals recorded by wearable sensors can effectively discriminate the gait patterns of knee and hip OA. In addition, the most remarkable differences in the time domain were observed along the mediolateral axis.

Aims

Research on hip biomechanics has analyzed femoroacetabular contact pressures and forces in distinct hip conditions, with different procedures, and used diverse loading and testing conditions. The aim of this scoping review was to identify and summarize the available evidence in the literature for hip contact pressures and force in cadaver and in vivo studies, and how joint loading, labral status, and femoral and acetabular morphology can affect these biomechanical parameters.

Physiotherapy is a critical element in successful conservative management of low back pain (LBP). The aim of this study was to develop and evaluate a system with wearable inertial sensors to objectively detect sitting postures and performance of unsupervised exercises containing movement in multiple planes (flexion, extension, rotation).

A set of 8 inertial sensors were placed on 19 healthy adult subjects. Data was acquired as they performed 7 McKenzie low-back exercises and 3 sitting posture positions. This data was used to train two models (Random Forest (RF) and XGBoost (XGB)) using engineered time series features. In addition, a convolutional neural network (CNN) was trained directly on the time series data. A feature importance analysis was performed to identify sensor locations and channels that contributed most to the models. Finally, a subset of sensor locations and channels was included in a hyperparameter grid search to identify the optimal sensor configuration and the best performing algorithm(s) for exercise classification. Models were evaluated using F1-score in a 10-fold cross validation approach.

The optimal hardware configuration was identified as a 3-sensor setup using lower back, left thigh, and right ankle sensors with acceleration, gyroscope, and magnetometer channels. The XBG model achieved the highest exercise (F1=0.94±0.03) and posture (F1=0.90±0.11) classification scores. The CNN achieved similar results with the same sensor locations, using only the accelerometer and gyroscope channels for exercise classification (F1=0.94±0.02) and the accelerometer channel alone for posture classification (F1=0.91±0.03).

This study demonstrates the potential of a 3-sensor lower body wearable solution (e.g. smart pants) that can identify proper sitting postures and exercises in multiple planes, suitable for low back pain. This technology has the potential to improve the effectiveness of LBP rehabilitation by facilitating quantitative feedback, early problem diagnosis, and possible remote monitoring.

Patient-reported outcome measures (PROMs) have failed to highlight differences in function or outcome when comparing knee replacement designs and implantation techniques. Ankle-worn inertial measurement units (IMUs) can be used to remotely measure and monitor the bi-lateral impact load of patients, augmenting traditional PROMs with objective data. The aim of this study was to compare IMU-based impact loads with PROMs in patients who had undergone conventional total knee arthroplasty (TKA), unicompartmental knee arthroplasty (UKA), and robotic-assisted TKA (RA-TKA).

77 patients undergoing primary knee arthroplasty (29 RA-TKA, 37 TKA, and 11 UKA) for osteoarthritis were prospectively enrolled. Remote patient monitoring was performed pre-operatively, then weekly from post-operative weeks two to six using ankle-worn IMUs and PROMs. IMU-based outcomes included: cumulative impact load, bone stimulus, and impact load asymmetry. PROMs scores included: Oxford Knee Score (OKS), EuroQol Five-dimension with EuroQol visual analogue scale, and the Forgotten Joint Score.

On average, patients showed improved impact load asymmetry by 67% (p=0.001), bone stimulus by 41% (p<0.001), and cumulative impact load by 121% (p=0.035) between post-operative week two and six. Differences in IMU-based outcomes were observed in the initial six weeks post-operatively between surgical procedures. The mean change scores for OKS were 7.5 (RA-TKA), 11.4 (TKA), and 11.2 (UKA) over the early post-operative period (p=0.144). Improvements in OKS were consistent with IMU outcomes in the RA-TKA group, however, conventional TKA and UKA groups did not reflect the same trend in improvement as OKS, demonstrating a functional decline.

Our data illustrate that PROMs do not necessarily align with patient function, with some patients reporting good PROMs, yet show a decline in cumulative impact load or load asymmetry. These data also provide evidence for a difference in the functional outcome of TKA and UKA patients that might be overlooked by using PROMs alone.

Literature surrounding artificial intelligence (AI)-related applications for hip and knee arthroplasty has proliferated. However, meaningful advances that fundamentally transform the practice and delivery of joint arthroplasty are yet to be realized, despite the broad range of applications as we continue to search for meaningful and appropriate use of AI. AI literature in hip and knee arthroplasty between 2018 and 2021 regarding image-based analyses, value-based care, remote patient monitoring, and augmented reality was reviewed. Concerns surrounding meaningful use and appropriate methodological approaches of AI in joint arthroplasty research are summarized. Of the 233 AI-related orthopaedics articles published, 178 (76%) constituted original research, while the rest consisted of editorials or reviews. A total of 52% of original AI-related research concerns hip and knee arthroplasty (n = 92), and a narrative review is described. Three studies were externally validated. Pitfalls surrounding present-day research include conflating vernacular (“AI/machine learning”), repackaging limited registry data, prematurely releasing internally validated prediction models, appraising model architecture instead of inputted data, withholding code, and evaluating studies using antiquated regression-based guidelines. While AI has been applied to a variety of hip and knee arthroplasty applications with limited clinical impact, the future remains promising if the question is meaningful, the methodology is rigorous and transparent, the data are rich, and the model is externally validated. Simple checkpoints for meaningful AI adoption include ensuring applications focus on: administrative support over clinical evaluation and management; necessity of the advanced model; and the novelty of the question being answered.

Cite this article: Bone Joint J 2022;104-B(12):1292–1303.

Hip instability is one of the most common causes for total hip arthroplasty (THA) revision surgery. Studies have indicated that lumbar fusion (LF) surgery is a risk factor for hip dislocation. Instrumented spine fusion surgery decreases pelvic tilt, which might lead to an increase in hip motion to accommodate this postural change. To the best of our knowledge, spine-pelvis-hip kinematics during a dynamic activity in patients that previously had both a THA and LF have not been investigated. Furthermore, patients with a combined THA and LF tend to have greater disability. The purpose was to examine spine-pelvis-hip kinematics during a sit to stand task in patients that have had both THA and LF surgeries and compare it to a group of patients that had a THA with no history of spine surgery. The secondary purpose was to compare pain, physical function, and disability between these patients. This cross-sectional study recruited participants that had a combined THA and LF (n=10; 6 females, mean age 73 y) or had a THA only (n=11; 6 females, mean age 72 y). Spine, pelvis, and hip angles were measured using a TrakSTAR motion capture system sampled at 200 Hz. Sensors were mounted over the lateral thighs, base of the sacrum, and the spinous process of the third lumbar,12th thoracic, and ninth thoracic vertebrae. Participants completed 10 trials of a standardized sit-to-stand-to-sit task. Hip, pelvis, lower lumbar, upper lumbar, and lower thoracic sagittal joint angle range of motion (ROM) were calculated over the entire task. In addition, pain, physical function, and disability were measured with clinical outcomes: Hip Disability Osteoarthritis Outcome Score (pain and physical function), Oswestry Low Back Disability Questionnaire (disability), and Harris Hip Score (pain, physical function, motion). Physical function performance was measured using 6-Minute Walk Test, Stair Climb Test, and 30s Chair Test. Angle ROMs during the sit-to-stand-to-sit task and clinical outcomes were compared between THA+LF and THA groups using independent t-tests and effect sizes (d). The difference in hip ROM was approaching statistical significance (p=0.07). Specifically, the THA+LF group had less hip ROM during the sit-to-stand-to-sit task than the THA only group (mean difference=11.17, 95% confidence interval=-1.13 to 23.47), which represented a large effect size (d=0.83). There were no differences in ROM for pelvis (p=0.54, d=0.28) or spinal (p=0.14 to 0.97; d=0.02 to 0.65) angles between groups. The THA+LF group had worse clinical outcomes for all measures of pain, physical function, and disability (p=0.01 to 0.06), representing large effect sizes (d=0.89 to 2.70). Hip ROM was not greater in the THA+LF group, and thus this is unlikely a risk factor for hip dislocation during this specific sit-to-stand-to-sit task. Other functional tasks that demand greater excursions in the joints should be investigated. Furthermore, the lack of differences in spinal and pelvis ROM were likely due to the task and the THA+LF group had spinal fusions at different levels. Combined THA+LF results in worse clinical outcomes and additional rehabilitation is required for these patients

Thermal sensors have been used in bracing research as self-reported diaries are inaccurate. Little is known about new low-profile sensors, optimal location within a brace, locational thermal micro-climate and effect of brace lining. Our objective is to Determine an optimal temperature threshold for sensor-measured and true wear time agreement. Identify optimal sensor location. Assess all factors to determine the best sensor option for the Bracing AdoleScent Idiopathic Scoliosis (BASIS) multicentre RCT.

Seven Orthotimer and five iButton (DS1925L) sensors were synchronised to record temperature at five-minute intervals. Three healthy participants donned a rigid spinal brace, embedded with both sensors across four anatomical locations (abdomen/axilla/lateral-gluteal/sacral). Universal-coordinated-time wear protocols were performed in/out-doors. Intraclass correlation coefficient (ICC) assessed sensor-measured and true wear time agreement at thresholds 15–36oC.

Optimal thresholds, determined by largest ICC estimate: Orthotimer: Abdomen=26oC, axilla=27oC, lateral-gluteal=24.5oC, sacral=22.5oC. iButton: Abdomen=26oC, axilla=27oC, lateral-gluteal=23.5oC, sacral=23.5oC. Warm-up time and error at optimal thresholds increased for moulded sensors covered with 6mm lining.

Location: anterior abdominal wall. Excellent reliability and higher optimal thresholds, less likely to be exceeded by ambient temperature; not a pressure area. Sensor: iButton, longer battery life and larger memory than Orthotimer; allows recording at 10 min intervals for life of brace. Orthotimer only able to record every 30 mins, increasing error between true and measured wear time; Orthotimer needs 6-monthly data download. Threshold: 26oC is optimal threshold to balance warm-up and cool-down times for accurately measuring wear time. Sensor should not be covered by lining foam as this significantly prolongs warm-up time.

Aims

Intraoperative pressure sensors allow surgeons to quantify soft-tissue balance during total knee arthroplasty (TKA). The aim of this study was to determine whether using sensors to achieve soft-tissue balance was more effective than manual balancing in improving outcomes in TKA.

Introduction

Wearable sensors are promising tools for fast clinical gait evaluations in individuals with osteoarthritis (OA) of the knee and hip. However, gait assessments with wearable sensor are often limited to relatively simple straight-ahead walking paradigms. Parameters reflecting more complex and relevant aspects of gait, including dual-tasking, turning, and compensatory upper body motion are often overlooked in literature. The aim of this study was to investigate turning, dual-task performance, and upper body motion in individuals with knee or hip OA in addition to spatiotemporal gait parameters, taking shared covariance between gait parameters into account.

Participation in a physical therapy program is considered one of the greatest predictors for successful conservative management of common shoulder disorders, however, adherence to standard exercise protocols is often poor (around 50%) and typically worse for unsupervised home exercise programs. Currently, there are limited tools available for objective measurement of adherence and performance of shoulder rehabilitation in the home setting. The goal of this study was to develop and evaluate the potential for performing home shoulder physiotherapy monitoring using a commercial smartwatch. We hypothesize that shoulder physiotherapy exercises can be classified by analyzing the temporal sequence of inertial sensor outputs from a smartwatch worn on the extremity performing the exercise.

Twenty healthy adult subjects with no prior shoulder disorders performed seven exercises from a standard evidence-based rotator cuff physiotherapy protocol: pendulum, abduction, forward elevation, internal/external rotation and trapezius extension with a resistance band, and a weighted bent-over row. Each participant performed 20 repetitions of each exercise bilaterally under the supervision of an orthopaedic surgeon, while 6-axis inertial sensor data was collected at 50 Hz from an Apple Watch. Using the scikit-learn and keras platforms, four supervised learning algorithms were trained to classify the exercises: k-nearest neighbour (k-NN), random forest (RF), support vector machine classifier (SVC), and a deep convolutional recurrent neural network (CRNN). Algorithm performance was evaluated using 5-fold cross-validation stratified first temporally and then by subject.

Categorical classification accuracy was above 94% for all algorithms on the temporally stratified cross validation, with the best performance achieved by the CRNN algorithm (99.4± 0.2%). The subject stratified cross validation, which evaluated classifier performance on unseen subjects, yielded lower accuracies scores again with CRNN performing best (88.9 ± 1.6%).

This proof-of concept study demonstrates the feasibility of a smartwatch device and machine learning approach to more easily monitor and assess the at-home adherence of shoulder physiotherapy exercise protocols. Future work will focus on translation of this technology to the clinical setting and evaluating exercise classification in shoulder disorder populations.

Whilst home-based exercise rehabilitation plays a key role in determining patient outcomes following orthopaedic intervention (e.g. total knee replacement), it is very challenging for clinicians to objectively monitor patient progress, attribute functional improvement (or lack of) to adherence/non-adherence and ultimately prescribe personalised interventions. This research aimed to identify whether 4 knee rehabilitation exercises could be objectively distinguished from each other using lower body inertial measurement units (IMUs) and principle components analysis (PCA) in the hope to facilitate objective home monitoring of exercise rehabilitation.

5 healthy participants performed 4 repetitions of 4 exercises (knee flexion in sitting, knee extension, single leg step down and sit to stand) whilst wearing lower body IMU sensors (Xsens, Holland; sampling at 60 Hz). Anthropometric measurements and a static calibration were combined to create the biomechanical model, with 3D hip, knee and ankle angles computed using the Euler sequence ZXY. PCA was performed on time normalised (101 points) 3D joint angle data which reduced all joint angle waveforms into new uncorrelated PCs via an orthogonal transformation. Scatterplots of PC1 versus PC2 were used to visually inspect for clustering between the PC values for the 4 exercises. A one-way ANOVA was performed on the first 3 PC values for the 9 variables under analysis. Games-Howell post hoc tests identified variables that were significantly different between exercises.

All exercises were clearly distinguishable using the PC scatterplot representing hip flexion-extension waveforms. ANOVA results revealed that PC1 for the knee flexion angle waveform was the only PC value statistically different across all exercises.

Findings demonstrate clear potential to objectively distinguish between different knee rehabilitation exercises using IMU sensors and PCA. Flexion-extension angles at the hip and knee appear most suited for accurate separation, which will be further investigated on patient data and additional exercises.

Immune response in periprosthetic joint infection (PJI) is diverse. Resident macrophage and/or wandering monocyte are superb guardians to sense microbial attacks, take invaders and alarm the danger. Neutrophils are refined but momentary fighters to kill microbes with projectile weapons as well as predation. The swift action is usually effective at the forefront to prevent expansion of infectious foci. However, such characteristics often evokes overshooting via self-defeating of pus, thus leading to crucial soft tissue damage in the acute phase. Intervention of monocyte/macrophages follow and act as wise organizers. In addition, stromal fibroblasts also act in front for host defence. They equip innate immune sensors (TLRs, NLRs), which can sense dangers and trigger off inflammatory response, but also is usually self-regulated. These sensors not only interact each other, but also have possible contribution to selective autophagy (xenophagy and lysophagy) in PJI. In this presentation, overview of pathology in PJI will be summarized with a special attention to innate immune sensors (TLRs and NLRs), and selective autophagy.

3D measurement of joint angles so far has only been possible using marker-based movement analysis, and therefore has not been applied in (larger scale) clinical practice (performance test) and even less so in the free field (activity monitoring). 3D joint angles could provide useful additional information in assessing the risk of anterior cruciate ligament injury using a vertical drop jump or in assessing knee range of motion after total knee arthroplasty. We developed a tool to measure dynamic 3D joint angles using 6 inertial sensors, attached to left and right shank, thigh and pelvis. The same sensors have been used for activity identification in a previous study. To validate the setup in a pilot study, we measured 3D knee and hip angles using the sensors and a Vicon movement lab simultaneously in 3 subjects. Subjects performed drop jumps, squats and ran on the spot. The mean error between Vicon and sensor measurement for the maximum joint angles was 3, 7 and 8 degrees for knee flexion, ad/abduction and rotation respectively, and 9, 7 and 10 degrees for hip flexion, ad/abduction and rotation respectively. No calibration movements were required. A major part of the inaccuracy was caused by soft tissue effects and can partly be resolved by improved sensor attachment. These pilot results show that it is feasible to measure 3D joint angles continuously using unobtrusive light-weight sensors. No movement lab is necessary and therefore the measurements can be done in a free field setting, e.g. at home or during training at a sport club. A more extensive validation study will be performed in the near future.

Introduction & Aims

The traditional method of soft-tissue balancing during TKA is subjective in nature, and stiffness and instability are common indications for revision, suggesting that TKA balancing by subjective assessment is suboptimal. This study examines the intraoperative mediolateral loads measured with a nanosensor-enabled tibial insert trial and the sequential balancing steps used to achieve quantitative balance. Data obtained from a prospective multicenter study was assessed to determine the effect of targeted ligament release on intra-articular loading, and to understand which types of releases are necessary to achieve quantified ligament balance.

Introduction

Following primary total knee arthroplasty (TKA), patients experience pain relief and report improved physical function and activity. However, there is paucity of evidence that patients are truly more active in daily life after TKA. The aims of this study were: 1) to prospectively measure physical activity with a wearable motion sensor before and after TKA; 2) to compare patient-reported levels of physical activity with objectively assessed levels of physical activity before and after TKA; 3) to investigate whether differences in physical activity after TKA are related to levels of physical function.

Background

Shoulder pain limits range of motion (ROM) and reduces performing activities of daily living (ADL). Objective assessment of shoulder function could be of interest for diagnosing shoulder pathology or functional assessment of the shoulder after therapy.

The feasibility of 2 wearable inertial sensors for functional assessment to differentiate between healthy subjects and patients with unilateral shoulder pathology is investigated using parameters as asymmetry.

One out of every five total knee arthroplasty (TKA) recipients is unhappy with the outcome of their surgery. As the number of TKA candidates continues to increase, so, too, will the dissatisfied patient population. These statistics should not be acceptable to the surgeons, hospitals, and patients implicated in this elective procedure. There are many contributing factors to patient dissatisfaction, paramount among them being post-operative levels of functionality and pain. Therefore, in an attempt to increase function and decrease pain levels through soft-tissue management, sensor-assisted TKA outcomes were compared with manual TKA outcomes.

One hundred and fourteen primary TKA patients were evaluated: 57 sensor-assisted TKA patients; 57 manual TKA patients. All procedures were performed by the same surgeon. In order to reduce confounding variables, all patients were matched for: age, gender distribution, BMI, marital status, smoking proclivity, pre-operative ROM, pre-operative alignment, and employment status. Outcomes scores were captured pre-operatively, and at the 6-month interval, including Knee Society Score metrics and the Oxford score, as well as 6-month ROM. The sensor device used in this analysis is inserted into the tibial component, during the trialing, and displays loading values in the medial and lateral compartments (lbf.), and also displays the medial and lateral center of load location. In the sensor-assisted TKA group, balance was achieved for all patients, as previously described in literature.

There was a statistically significant rate of improvement, for all outcomes measures, in the sensor-assisted TKA group when compared with the manual group (Figure 1). In addition to rate of improvement, there was also a significant trend towards a significance in ROM in the sensor-assisted group, as a stand-alone dependent variable (P = 0.002).

By the 6-month follow-up interval, patients in receipt of a sensor-assisted TKA reported greater improvement in function and less pain than the patients in the manual TKA group. This data suggests that soft-tissue balance may contribute to faster recovery, as reported by the patient. Because pain and function play an integral role in patient satisfaction, further follow-up might yield higher satisfaction in the sensor-assisted patient group, which is consistent with previously published observations.