Monitoring Activity and Gait in Children (MAGIC) using digital health technologies.

BACKGROUND: Digital health technologies (DHTs) can collect gait and physical activity in adults, but limited studies have validated these in children. This study compared gait and physical activity metrics collected using DHTs to those collected by reference comparators during in-clinic sessions, to collect a normative accelerometry dataset, and to evaluate participants' comfort and their compliance in wearing the DHTs at-home. METHODS: The MAGIC (Monitoring Activity and Gait in Children) study was an analytical validation study which enrolled 40, generally healthy participants aged 3-17 years. Gait and physical activity were collected using DHTs in a clinical setting and continuously at-home. RESULTS: Overall good to excellent agreement was observed between gait metrics extracted with a gait algorithm from a lumbar-worn DHT compared to ground truth reference systems. Majority of participants either "agreed" or "strongly agreed" that wrist and lumbar DHTs were comfortable to wear at home, respectively, with 86% (wrist-worn DHT) and 68% (lumbar-worn DHT) wear-time compliance. Significant differences across age groups were observed in multiple gait and activity metrics obtained at home. CONCLUSIONS: Our findings suggest that gait and physical activity data can be collected from DHTs in pediatric populations with high reliability and wear compliance, in-clinic and in home environments. TRIAL REGISTRATION: ClinicalTrials.gov: NCT04823650 IMPACT: Digital health technologies (DHTs) have been used to collect gait and physical activity in adult populations, but limited studies have validated these metrics in children. The MAGIC study comprehensively validates the performance and feasibility of DHT-measured gait and physical activity in the pediatric population. Our findings suggest that reliable gait and physical activity data can be collected from DHTs in pediatric populations, with both high accuracy and wear compliance both in-clinic and in home environments. The identified across-age-group differences in gait and activity measurements highlighted their potential clinical value.

SciKit digital health package for accelerometry-measured physical activity: comparisons to existing solutions and investigations of age effects in healthy adults.

Introduction: Accelerometry has become increasingly prevalent to monitor physical activity due to its low participant burden, quantitative metrics, and ease of deployment. Physical activity metrics are ideal for extracting intuitive, continuous measures of participants' health from multiple days or weeks of high frequency data due to their fairly straightforward computation. Previously, we released an open-source digital health python processing package, SciKit Digital Health (SKDH), with the goal of providing a unifying device-agnostic framework for multiple digital health algorithms, such as activity, gait, and sleep. Methods: In this paper, we present the open-source SKDH implementation for the derivation of activity endpoints from accelerometer data. In this implementation, we include some non-typical features that have shown promise in providing additional context to activity patterns, and provide comparisons to existing algorithms, namely GGIR and the GENEActiv macros. Following this reference comparison, we investigate the association between age and derived physical activity metrics in a healthy adult cohort collected in the Pfizer Innovation Research Lab, comprising 7-14 days of at-home data collected from younger (18-40 years) and older (65-85 years) healthy volunteers. Results: Results showed that activity metrics derived with SKDH had moderate to excellent ICC values (0.550 to 1.0 compared to GGIR, 0.469 to 0.697 compared to the GENEActiv macros), with high correlations for almost all compared metrics (>0.733 except vs GGIR sedentary time, 0.547). Several features show age-group differences, with Cohen's d effect sizes >1.0 and p-values < 0.001. These features included non-threshold methods such as intensity gradient, and activity fragmentation features such as between-states transition probabilities. Discussion: These results demonstrate the validity of the implemented SKDH physical activity algorithm, and the potential of the implemented PA metrics in assessing activity changes in free-living conditions.

Can Gait Characteristics Be Represented by Physical Activity Measured with Wrist-Worn Accelerometers?

Wearable accelerometers allow for continuous monitoring of function and behaviors in the participant's naturalistic environment. Devices are typically worn in different body locations depending on the concept of interest and endpoint under investigation. The lumbar and wrist are commonly used locations: devices placed at the lumbar region enable the derivation of spatio-temporal characteristics of gait, while wrist-worn devices provide measurements of overall physical activity (PA). Deploying multiple devices in clinical trial settings leads to higher patient burden negatively impacting compliance and data quality and increases the operational complexity of the trial. In this work, we evaluated the joint information shared by features derived from the lumbar and wrist devices to assess whether gait characteristics can be adequately represented by PA measured with wrist-worn devices. Data collected at the Pfizer Innovation Research (PfIRe) Lab were used as a real data example, which had around 7 days of continuous at-home data from wrist- and lumbar-worn devices (GENEActiv) obtained from a group of healthy participants. The relationship between wrist- and lumbar-derived features was estimated using multiple statistical methods, including penalized regression, principal component regression, partial least square regression, and joint and individual variation explained (JIVE). By considering multilevel models, both between- and within-subject effects were taken into account. This work demonstrated that selected gait features, which are typically measured with lumbar-worn devices, can be represented by PA features measured with wrist-worn devices, which provides preliminary evidence to reduce the number of devices needed in clinical trials and to increase patients' comfort. Moreover, the statistical methods used in this work provided an analytic framework to compare repeated measures collected from multiple data modalities.

Reliability of Wearable Sensors for Assessing Gait and Chair Stand Function at Home in People With Knee Osteoarthritis.

OBJECTIVE: To assess the reliability of wearable sensors for at-home assessment of walking and chair stand activities in people with knee osteoarthritis (OA). METHODS: Baseline data from participants with knee OA (n = 20) enrolled in a clinical trial of an exercise intervention were used. Participants completed an in-person laboratory visit and a video conference-enabled at-home visit. In both visits, participants performed walking and chair stand tasks while fitted with 3 inertial sensors. During the at-home visit, participants self-donned the sensors and completed 2 sets of acquisitions separated by a 15-minute break, when they removed and redonned the sensors. Participants completed a survey on their experience with the at-home visit. During the laboratory visit, researchers placed the sensors on the participants. Spatiotemporal metrics of walking gait and chair stand duration were extracted from the sensor data. We used intraclass correlation coefficients (ICCs) and the Bland-Altman plot for statistical analyses. RESULTS: For test-retest reliability during the at-home visit, all ICCs were good to excellent (0.85-0.95). For agreement between at-home and laboratory visits, ICCs were moderate to good (0.59-0.87). Systematic differences were noted between at-home and laboratory data due to faster task speed during the laboratory visits. Participants reported a favorable experience during the at-home visit. CONCLUSION: Our method of estimating spatiotemporal gait measures and chair stand duration function remotely was reliable, feasible, and acceptable in people with knee OA. Wearable sensors could be used to remotely assess walking and chair stand in participant's natural environments in future studies.

SciKit Digital Health: Python Package for Streamlined Wearable Inertial Sensor Data Processing.

Wearable inertial sensors are providing enhanced insight into patient mobility and health. Significant research efforts have focused on wearable algorithm design and deployment in both research and clinical settings; however, open-source, general-purpose software tools for processing various activities of daily living are relatively scarce. Furthermore, few studies include code for replication or off-the-shelf software packages. In this work, we introduce SciKit Digital Health (SKDH), a Python software package (Python Software Foundation) containing various algorithms for deriving clinical features of gait, sit to stand, physical activity, and sleep, wrapped in an easily extensible framework. SKDH combines data ingestion, preprocessing, and data analysis methods geared toward modern data science workflows and streamlines the generation of digital endpoints in "good practice" environments by combining all the necessary data processing steps in a single pipeline. Our package simplifies the construction of new data processing pipelines and promotes reproducibility by following a convention over configuration approach, standardizing most settings on physiologically reasonable defaults in healthy adult populations or those with mild impairment. SKDH is open source, as well as free to use and extend under a permissive Massachusetts Institute of Technology license, and is available from GitHub (PfizerRD/scikit-digital-health), the Python Package Index, and the conda-forge channel of Anaconda.

Wearables and Deep Learning Classify Fall Risk From Gait in Multiple Sclerosis.

Falls are a significant problem for persons with multiple sclerosis (PwMS). Yet fall prevention interventions are not often prescribed until after a fall has been reported to a healthcare provider. While still nascent, objective fall risk assessments could help in prescribing preventative interventions. To this end, retrospective fall status classification commonly serves as an intermediate step in developing prospective fall risk assessments. Previous research has identified measures of gait biomechanics that differ between PwMS who have fallen and those who have not, but these biomechanical indices have not yet been leveraged to detect PwMS who have fallen. Moreover, they require the use of laboratory-based measurement technologies, which prevent clinical deployment. Here we demonstrate that a bidirectional long short-term (BiLSTM) memory deep neural network was able to identify PwMS who have recently fallen with good performance (AUC of 0.88) based on accelerometer data recorded from two wearable sensors during a one-minute walking task. These results provide substantial improvements over machine learning models trained on spatiotemporal gait parameters (21% improvement in AUC), statistical features from the wearable sensor data (16%), and patient-reported (19%) and neurologist-administered (24%) measures in this sample. The success and simplicity (two wearable sensors, only one-minute of walking) of this approach indicates the promise of inexpensive wearable sensors for capturing fall risk in PwMS.

Assessment of Sit-to-Stand Transfers during Daily Life Using an Accelerometer on the Lower Back.

The ability to perform sit-to-stand (STS) transfers has a significant impact on the functional mobility of an individual. Wearable technology has the potential to enable the objective, long-term monitoring of STS transfers during daily life. However, despite several recent efforts, most algorithms for detecting STS transfers rely on multiple sensing modalities or device locations and have predominantly been used for assessment during the performance of prescribed tasks in a lab setting. A novel wavelet-based algorithm for detecting STS transfers from data recorded using an accelerometer on the lower back is presented herein. The proposed algorithm is independent of device orientation and was validated on data captured in the lab from younger and older healthy adults as well as in people with Parkinson's disease (PwPD). The algorithm was then used for processing data captured in free-living conditions to assess the ability of multiple features extracted from STS transfers to detect age-related group differences and assess the impact of monitoring duration on the reliability of measurements. The results show that performance of the proposed algorithm was comparable or significantly better than that of a commercially available system (precision: 0.990 vs. 0.868 in healthy adults) and a previously published algorithm (precision: 0.988 vs. 0.643 in persons with Parkinson's disease). Moreover, features extracted from STS transfers at home were able to detect age-related group differences at a higher level of significance compared to data captured in the lab during the performance of prescribed tasks. Finally, simulation results showed that a monitoring duration of 3 days was sufficient to achieve good reliability for measurement of STS features. These results point towards the feasibility of using a single accelerometer on the lower back for detection and assessment of STS transfers during daily life. Future work in different patient populations is needed to evaluate the performance of the proposed algorithm, as well as assess the sensitivity and reliability of the STS features.

Validation of Novel Relative Orientation and Inertial Sensor-to-Segment Alignment Algorithms for Estimating 3D Hip Joint Angles.

Wearable sensor-based algorithms for estimating joint angles have seen great improvements in recent years. While the knee joint has garnered most of the attention in this area, algorithms for estimating hip joint angles are less available. Herein, we propose and validate a novel algorithm for this purpose with innovations in sensor-to-sensor orientation and sensor-to-segment alignment. The proposed approach is robust to sensor placement and does not require specific calibration motions. The accuracy of the proposed approach is established relative to optical motion capture and compared to existing methods for estimating relative orientation, hip joint angles, and range of motion (ROM) during a task designed to exercise the full hip range of motion (ROM) and fast walking using root mean square error (RMSE) and regression analysis. The RMSE of the proposed approach was less than that for existing methods when estimating sensor orientation ( 12 . 32 ∘ and 11 . 82 ∘ vs. 24 . 61 ∘ and 23 . 76 ∘ ) and flexion/extension joint angles ( 7 . 88 ∘ and 8 . 62 ∘ vs. 14 . 14 ∘ and 15 . 64 ∘ ). Also, ROM estimation error was less than 2 . 2 ∘ during the walking trial using the proposed method. These results suggest the proposed approach presents an improvement to existing methods and provides a promising technique for remote monitoring of hip joint angles.