40 years of actigraphy in sleep medicine and current state of the art algorithms.

For the last 40 years, actigraphy or wearable accelerometry has provided an objective, low-burden and ecologically valid approach to assess real-world sleep and circadian patterns, contributing valuable data to epidemiological and clinical insights on sleep and sleep disorders. The proper use of wearable technology in sleep research requires validated algorithms that can derive sleep outcomes from the sensor data. Since the publication of the first automated scoring algorithm by Webster in 1982, a variety of sleep algorithms have been developed and contributed to sleep research, including many recent ones that leverage machine learning and / or deep learning approaches. However, it remains unclear how these algorithms compare to each other on the same data set and if these modern data science approaches improve the analytical validity of sleep outcomes based on wrist-worn acceleration data. This work provides a systematic evaluation across 8 state-of-the-art sleep algorithms on a common sleep data set with polysomnography (PSG) as ground truth. Despite the inclusion of recently published complex algorithms, simple regression-based and heuristic algorithms demonstrated slightly superior performance in sleep-wake classification and sleep outcome estimation. The performance of complex machine learning and deep learning models seem to suffer from poor generalization. This independent and systematic analytical validation of sleep algorithms provides key evidence on the use of wearable digital health technologies for sleep research and care.

Stepping towards More Intuitive Physical Activity Metrics with Wrist-Worn Accelerometry: Validity of an Open-Source Step-Count Algorithm.

Stepping-based targets such as the number of steps per day provide an intuitive and commonly used method of prescribing and self-monitoring physical activity goals. Physical activity surveillance is increasingly being obtained from wrist-worn accelerometers. However, the ability to derive stepping-based metrics from this wear location still lacks validation and open-source methods. This study aimed to assess the concurrent validity of two versions (1. original and 2. optimized) of the Verisense step-count algorithm at estimating step-counts from wrist-worn accelerometry, compared with steps from the thigh-worn activPAL as the comparator. Participants (n = 713), across three datasets, had >24 h continuous concurrent accelerometry wear on the non-dominant wrist and thigh. Compared with activPAL, total daily steps were overestimated by 913 ± 141 (mean bias ± 95% limits of agreement) and 742 ± 150 steps/day with Verisense algorithms 1 and 2, respectively, but moderate-to-vigorous physical activity (MVPA) steps were underestimated by 2207 ± 145 and 1204 ± 103 steps/day in Verisense algorithms 1 and 2, respectively. In summary, the optimized Verisense algorithm was more accurate in detecting total and MVPA steps. Findings highlight the importance of assessing algorithm performance beyond total step count, as not all steps are equal. The optimized Verisense open-source algorithm presents acceptable accuracy for derivation of stepping-based metrics from wrist-worn accelerometry.

Stepping up with GGIR: Validity of step cadence derived from wrist-worn research-grade accelerometers using the verisense step count algorithm.

The Verisense Step Count Algorithm facilitates generation of steps from wrist-worn accelerometers. Based on preliminary evidence suggesting a proportional bias with overestimation at low steps/day, but underestimation at high steps/day, the algorithm parameters have been revised. We aimed to establish validity of the original and revised algorithms relative to waist-worn ActiGraph step cadence. We also assessed whether step cadence was similar across accelerometer brand and wrist. Ninety-eight participants (age: 58.6±11.1 y) undertook six walks (~500 m hard path) at different speeds (cadence: 92.9±9.5-127.9±8.7 steps/min) while wearing three accelerometers on each wrist (Axivity, GENEActiv, ActiGraph) and an ActiGraph on the waist. Of these, 24 participants also undertook one run (~1000 m). Mean bias for the original algorithm was -21 to -26.1 steps/min (95% limits of agreement (LoA) ~±65 steps/min) and mean absolute percentage error (MAPE) 17-22%. This was unevenly distributed with increasing error as speed increased. Mean bias and 95%LoA were halved with the revised algorithm parameters (~-10 to -12 steps/min, 95%LoA ~30 steps/min, MAPE ~10-12%). Performance was similar across brand and wrist. The revised step algorithm provides a more valid measure of step cadence than the original, with MAPE similar to recently reported wrist-wear summary MAPE (7-11%).

Coordination and Symmetry Patterns During the Drop Vertical Jump in People With Chronic Ankle Instability and Lateral Ankle Sprain Copers.

BACKGROUND: The drop vertical jump (DVJ) task has previously been used to identify movement patterns associated with a number of injury types. However, no current research exists evaluating people with chronic ankle instability (CAI) compared with people coping with lateral ankle sprain (LAS) (referred to as "LAS copers") during this task. OBJECTIVE: The aim of this study was to identify the coping movement and motor control patterns of LAS copers in comparison with individuals with CAI during the DVJ task. DESIGN: This was a case-control study. METHODS: Seventy individuals were recruited at convenience within 2-weeks of sustaining a first-time acute LAS injury. One year following recruitment, these individuals were stratified into 2 groups: 28 with CAI and 42 LAS copers. They attended the testing laboratory to complete a DVJ task. Three-dimensional kinematic and sagittal-plane kinetic profiles were plotted for the lower extremity joints of both limbs for the drop jump phase (phase 1) and drop landing phase (phase 2) of the DVJ. The rate of impact modulation relative to body weight during both phases of the DVJ also was determined. RESULTS: Compared with LAS copers, participants with CAI displayed significant increases in hip flexion on their "involved" limb during phase 1 of the DVJ (23° vs 18°) and bilaterally during phase 2 (15° vs 10°). These movement patterns coincided with altered moment-of-force patterns at the hip on the "uninvolved" limb. LIMITATIONS: It is unknown whether these movement and motor control patterns preceded or occurred as a result of the initial LAS injury. CONCLUSIONS: Participants with CAI displayed hip-centered changes in movement and motor control patterns during a DVJ task compared with LAS copers. The findings of this study may give an indication of the coping mechanism underlying outcome following initial LAS injury.

Validation of temporal gait metrics from three IMU locations to the gold standard force plate.

The purpose of this work is to compare temporal gait parameters from three different IMU locations to the gold standard force platform. 33 subjects (12 F, 21 M) performed twenty gait trials each while wearing inertial measurement units (IMUs) on the trunk, both shanks and both feet. Data was simultaneously collected from a laboratory embedded force plate. Step times were derived from the raw IMU data at the three IMU locations using methods that have been shown to be accurate. Step times from all locations were valid compared to the force plate. Foot IMU step time was the most accurate (Pearson = .991, CI width = 3.00e2), the trunk IMU was the next most accurate (Pearson = .974, CI width = 4.85e2) and shank step time was the least accurate (Pearson = .958, CI width = 6.80e2). All three sensing locations result in valid estimations of step time compared to the gold standard force plate. These results suggest that the foot location would be most appropriate for clinical applications where very precise temporal parameter detection is required.

Coordination and symmetry patterns during the drop vertical jump, 6-months after first-time lateral ankle sprain.

To evaluate the adaptive movement and motor control patterns of a group with a 6-month history of first-time lateral ankle sprain (LAS) injury during a drop vertical jump (DVJ) task. Fifty-one participants with a 6-month history of first-time acute LAS injury and twenty controls performed a DVJ task. 3D kinematic and sagittal plane kinetic profiles were plotted for the lower extremity joints of both limbs for the drop jump (phase 1) and drop landing (phase 2) phases of the DVJ. Inter-limb symmetry and the rate of impact modulation (RIM) relative to bodyweight (BW) during both phases of the DVJ were also determined. LAS participants displayed bilateral increases in knee flexion and an increase in ankle inversion during phases 1 and 2, respectively. They also displayed reduced ankle plantar flexion on their injured limb during both phases of the DVJ (p < 0.05); increased inter-limb asymmetry of RIM was noted for both phases of the DVJ, while the moment-of-force profile exhibited bilaterally greater hip extensor dominance during phase 1. Participants with a 6-month history of first-time LAS display some movement patterns consistent with those observed in chronic ankle instability populations during similar tasks.

Dual-task and electrophysiological markers of executive cognitive processing in older adult gait and fall-risk.

The role of cognition is becoming increasingly central to our understanding of the complexity of walking gait. In particular, higher-level executive functions are suggested to play a key role in gait and fall-risk, but the specific underlying neurocognitive processes remain unclear. Here, we report two experiments which investigated the cognitive and neural processes underlying older adult gait and falls. Experiment 1 employed a dual-task (DT) paradigm in young and older adults, to assess the relative effects of higher-level executive function tasks (n-Back, Serial Subtraction and visuo-spatial Clock task) in comparison to non-executive distracter tasks (motor response task and alphabet recitation) on gait. All DTs elicited changes in gait for both young and older adults, relative to baseline walking. Significantly greater DT costs were observed for the executive tasks in the older adult group. Experiment 2 compared normal walking gait, seated cognitive performances and concurrent event-related brain potentials (ERPs) in healthy young and older adults, to older adult fallers. No significant differences in cognitive performances were found between fallers and non-fallers. However, an initial late-positivity, considered a potential early P3a, was evident on the Stroop task for older non-fallers, which was notably absent in older fallers. We argue that executive control functions play a prominent role in walking and gait, but the use of neurocognitive processes as a predictor of fall-risk needs further investigation.

Lower Limb Interjoint Postural Coordination One Year after First-Time Lateral Ankle Sprain.

INTRODUCTION: Longitudinal analyses of participants with a history of lateral ankle sprain are lacking. This investigation combined measures of lower limb interjoint coordination and stabilometry to evaluate static unipedal stance with the eyes open (condition 1) and closed (condition 2) in a group of participants with chronic ankle instability (CAI) compared to lateral ankle sprain "copers" (both recruited 12 months after sustaining an acute first-time lateral ankle sprain) and a group of noninjured controls. METHODS: Twenty-eight participants with CAI, 42 lateral ankle sprain "copers," and 20 noninjured controls completed three 20-s single-limb stance trials in conditions 1 and 2. An adjusted coefficient of multiple determination statistic was used to compare stance limb three-dimensional kinematic data for similarity to establish patterns of interjoint coordination. The fractal dimension of the stance limb center of pressure path was also calculated. RESULTS: Between-group analyses revealed that participants with CAI displayed notable increases in ankle-hip linked coordination compared with both lateral ankle sprain "copers" (-0.52 (1.05) vs 0.28 (0.9), P = 0.007) and controls (-0.52 (1.05) vs 0.63 (0.64), P = 0.006) in condition 1 and compared with controls only (0.62 (1.92) vs 0.1 (1.0) P = 0.002) in condition 2. Participants with CAI also exhibited a decrease in the fractal dimension of the center-of-pressure path during condition 2 compared with both controls and lateral ankle sprain "copers." CONCLUSIONS: Participants with CAI present with a hip-dominant strategy of eyes-open and eyes-closed static unipedal stance. This coincided with reduced complexity of the stance limb center of pressure path in the eyes-closed condition.

Context aware falls risk assessment: A case study comparison.

This paper describes three retrospective case studies to illustrate the potential clinical value of a system capable of capturing objective gait metrics and environment data from older adults with a history of falls while they go about their daily lives. Participants in this study wore an inertial sensor above each ankle and a wearable camera around their neck for seven consecutive days. Selected metrics are presented to illustrate scenarios where the data collected by the system could be of clinical value. Evidence suggests that obtaining objective gait metrics and environment data from older adults may not only allow healthcare professionals to assess gait more accurately, but also to design treatment plans and falls prevention strategies that are more specifically tailored to each individual.

Context focused older adult mobility and gait assessment.

This paper presents an initial overview of insights gained into how older adults mobilize in the home and community, based on data from inertial sensors which were worn by study participants over a 7-day period. The addition of a wearable camera provided additional contextual information which can be used to assess mobility and understand the factors that influence it in the free living environment. Seven days of data collected from a group of older adults who had experienced one or more falls in the previous six months was compared to that of a control group with no history of falling. Results showed that both groups spent relatively little time walking in challenging environmental conditions, and that the fallers spent significantly less time walking under regular conditions (no effect on gait) and outdoors. Analysis of gait metrics showed that the fallers were slightly slower in general, and more noticeable differences were observed when the participants were regrouped according to mobility levels determined from baseline assessments using traditional methods.

An ambulatory method of identifying anterior cruciate ligament reconstructed gait patterns.

The use of inertial sensors to characterize pathological gait has traditionally been based on the calculation of temporal and spatial gait variables from inertial sensor data. This approach has proved successful in the identification of gait deviations in populations where substantial differences from normal gait patterns exist; such as in Parkinsonian gait. However, it is not currently clear if this approach could identify more subtle gait deviations, such as those associated with musculoskeletal injury. This study investigates whether additional analysis of inertial sensor data, based on quantification of gyroscope features of interest, would provide further discriminant capability in this regard. The tested cohort consisted of a group of anterior cruciate ligament reconstructed (ACL-R) females and a group of non-injured female controls, each performed ten walking trials. Gait performance was measured simultaneously using inertial sensors and an optoelectronic marker based system. The ACL-R group displayed kinematic and kinetic deviations from the control group, but no temporal or spatial deviations. This study demonstrates that quantification of gyroscope features can successfully identify changes associated with ACL-R gait, which was not possible using spatial or temporal variables. This finding may also have a role in other clinical applications where small gait deviations exist.

Peak knee adduction moment during gait in anterior cruciate ligament reconstructed females.

BACKGROUND: Recent work has shown that anterior cruciate ligament reconstructed patients exhibit an increased peak knee adduction moment during walking gait compared to healthy controls. An increased peak knee adduction moment has been suggested to be a potential mechanism of degeneration for knee osteoarthritis. The few studies in this area have not considered an exclusively female anterior cruciate ligament reconstructed group. This study tested the hypothesis that female anterior cruciate ligament-reconstructed patients would have higher peak knee adduction moments than controls. METHODS: Peak knee adduction moment during walking was compared between a group of anterior cruciate ligament reconstructed females and a group of female activity matched controls over ten 15m walking trials in a laboratory at a self-selected pace. FINDINGS: Peak knee adduction moment was lower for the anterior cruciate ligament reconstructed group (N = 17, M = 0.31 Nm/kg · m, SD = 0.08) than for the control group (N = 17, M = 0.41 Nm/kg · m, SD = 0.12; t(32) = 2.483, p = 0.010, one-tailed, eta squared effect size = 0.16). INTERPRETATION: A group of female anterior cruciate ligament reconstructed subjects did not exhibit a gait characteristic which has been suggested to be associated with knee osteoarthritis development and has been shown to be present in male and mixed sex anterior cruciate ligament reconstructed populations previously.

Does external walking environment affect gait patterns?

The objective of this work is to develop an understanding of the relationship between mobility metrics obtained outside of the clinic or laboratory and the context of the external environment. Ten subjects walked with an inertial sensor on each shank and a wearable camera around their neck. They were taken on a thirty minute walk in which they mobilized over the following conditions; normal path, busy hallway, rough ground, blind folded and on a hill. Stride time, stride time variability, stance time and peak shank rotation rate during swing were calculated using previously published algorithms. Stride time was significantly different between several of the conditions. Technological advances mean that gait variables can now be captured as patients go about their daily lives. The results of this study show that the external environment has a significant impact on the quality of gait metrics. Thus, context of external walking environment is an important consideration when analyzing ambulatory gait metrics from the unsupervised home and community setting.

Using a foot mounted accelerometer to detect changes in gait patterns.

The purpose of this study is to investigate how data from a foot mounted accelerometer can be used to detect motor pattern healthy subjects performed walking trails under two different conditions; normal and stiff ankle walking. Lower body kinematic data were collected as well as accelerometer data from both feet. An algorithm is presented which quantifies relevant swing phase characteristics from the foot accelerometer. Peak total acceleration during initial swing was significantly higher in the stiff ankle condition (M = 33.10, SD = 5.12) than in the normal walking condition (M = 29.47, SD = 5.75; t(7) = 4.32, p = .003, two-tailed). There was a large effect size (eta squared = 0.853). Time between peak acceleration during initial swing to foot strike was significantly shorter in the stiff ankle condition (M = 0.42, SD = 0.02) than in the normal condition (M = 0.44, SD = 0.03; t(7) = -2.54, p = .039, two- tailed). There was a large effect size (eta squared = 0.693). Simple to process metrics from tri-axial accelerometer data on the foot show potential to detect changes in ankle kinematic patterns.

A diagonal landing task to assess dynamic postural stability in ACL reconstructed females.

BACKGROUND: Previous research has used time to stabilization (TTS) from forward landing tasks to assess dynamic postural stability in ACL reconstructed (ACLR) athletes in order to identify impaired sensorimotor control and mechanical stability. This may not be an appropriate test due to the fact that research has suggested that ACL injury has a multi-planar mechanism of injury. The purpose of the present study was to compare TTS values from a forward land and a diagonal land to determine if diagonal landing TTS values are more sensitive to dynamic postural stability deficits in female ACLR athletes. METHODS: A group of ACL reconstructed female athletes and a group of female control athletes performed three forward lands and three diagonal lands onto a force-plate and remained still on one foot for 15s. TTS was calculated for the anterior-posterior and medial-lateral ground reaction forces as well as the resultant vector of both forces. RESULTS: All three TTS values were significantly increased in the ACLR group from the control group for the diagonal landing task. There was no difference in TTS values between the groups for the forward landing task. CONCLUSION: TTS values from a diagonal landing are more sensitive at detecting impaired dynamic postural stability in a group of female ACLR athletes compared to TTS values from a forward land. LEVEL OF EVIDENCE: III - Casecontrolled study.

Comparing adaptive algorithms to measure temporal gait parameters using lower body mounted inertial sensors.

The purpose of this research was to compare different adaptive algorithms in terms of their ability to determine temporal gait parameters based on data acquired from inertial measurement units (IMUs). Eight subjects performed 25 walking trials over a force plate under five different conditions; normal, fast, slow, simulated stiff ankle and simulated stiff knee walking. Data from IMUs worn on the shanks and on the feet were used to identify temporal gait features using three different adaptive algorithms (Green, Selles & Sabatini). Each method's ability to estimate temporal events was compared to the gold standard force plate method for stance time (Greene, r= .990, Selles, r= 0.865, Sabatini, r= 0.980) and double support time (Greene, r= .837, Selles, r= .583, Sabatini, r= .745). The Greene method of estimating gait events from inertial sensor data resulted in the most accurate stance and double support times.