Browse

A balance between the feasibility and validity of measures is an important consideration for physical activity (PA) research—particularly in school-based research with youth. The present study extends previously tested calibration methods to develop and test new equations for an online version of the youth activity profile (YAP) tool, a self-report tool designed for school applications. Data were collected across different regions and seasons to develop more robust, generalizable equations. The study involved a total of 717 youth from 33 schools (374 elementary [ages 9–11 years], 224 middle [ages 11–14 years], and 119 high school [ages 14–18 years]) in two different states in the United States. Participants wore a Sensewear monitor for a full week and then completed the online YAP at school to report PA and sedentary behaviors in school and at home. Accelerometer data were processed using an R-based segmentation program to compute PA and sedentary behavior levels. Quantile regression models were used with half of the sample to develop item-specific YAP calibration equations, and these were cross validated with the remaining half of the sample. Computed values of mean absolute percentage error ranged from 15 to 25% with slightly lower error observed for the middle school sample. The new equations had improved precision compared with the previous versions when tested on the same sample. The online version of the YAP provides an efficient and effective way to capture school level estimates of PA and sedentary behaviors in youth.

The response to COVID-19 resulted in behavioral restrictions to tackle the spread of infection. Initial data indicates that step counts were impacted by lockdown restrictions; however, there is little evidence regarding changes of light and moderate to vigorous physical activity (MVPA) behavioral intensities. In this study, participants were asked to provide longitudinal wearable data from Fitbit devices over a period of 30 weeks, from December 2019 to June 2020. Self-assessed key worker status was captured, along with wearable estimates of steps, light activity, and MVPA. Bayesian change point analyses of data from 97 individuals found that there was a sharp decrease of 1,473 steps (95% credible interval [CI] [−2,218, −709]) and light activity minutes (41.9; 95% CI [−54.3, −29.3]), but an increase in MVPA minutes (11.7; 95% CI [2.9, 19.4]) in the mean weekly totals for nonkey workers. For the key workers, the total number of steps (207; 95% CI [−788, 1,456]) and MVPA minutes increased (20.5; 95% CI [12.6, 28.3]) but light activity decreased by an average of 46.9 min (95% CI [−61.2, −31.8]). Interestingly, the change in steps was commensurate with that observed during Christmas (1,458; 95% CI [−2,286, −554]) for nonkey workers and behavioral changes occurred at different time points and rates depending on key worker status. Results indicate that there were clear behavioral modifications before and during the initial COVID-19 lockdown period, and future research should assess whether any behavioral modifications were sustained over time.

Background: Studies measuring physical activity (PA) and sedentary behavior on a 24/7 basis are scarce. The present study assessed the feasibility of using an accelerometer at the hip while awake and at the wrist while sleeping to describe 24/7 patterns of physical behavior in working-aged adults by age, sex, and fitness. Methods: The study was based on the FinFit 2017 study where the physical behavior of 20- to 69-year-old Finns was assessed 24/7 by triaxial accelerometer (UKKRM42; UKK Terveyspalvelut Oy, Tampere, Finland). During waking hours, the accelerometer was kept at the right hip and, during time in bed, at the nondominant wrist. PA variables were based on 1-min exponential moving average of mean amplitude deviation of the resultant acceleration signal analyzed in 6-s epochs. The angle for the posture estimation algorithm was used to identify sedentary behavior and standing. Evaluation of time in bed was based on the wrist movement. Fitness was estimated by the 6-min walk test. Results: A total of 2,256 eligible participants (mean age 49.5 years, SD = 13.5, 59% women) wore the accelerometer at the hip 15.7 hr/day (SD = 1.4) and at the wrist 8.3 hr/day (SD = 1.4). Sedentary behavior covered 9 hr 18 min/day (SD = 1.8 hr/day), standing nearly 2 hr/day (SD = 0.9), light PA 3.7 hr/day (SD = 1.3), and moderate to vigorous PA 46 min/day (SD = 26). Participants took 7,451 steps per day (SD = 2,962) on average. Men were most active around noon, while women had activity peaks at noon and at early evening. The low-fit tertile took 1,186 and 1,747 fewer steps per day than the mid- and high-fit tertiles (both p < .001). Conclusions: One triaxial accelerometer with a two wear-site approach provides a feasible method to characterize hour-by-hour patterns of physical behavior among working-aged adults.

Background: Machine learning has been used for classification of physical behavior bouts from hip-worn accelerometers; however, this research has been limited due to the challenges of directly observing and coding human behavior “in the wild.” Deep learning algorithms, such as convolutional neural networks (CNNs), may offer better representation of data than other machine learning algorithms without the need for engineered features and may be better suited to dealing with free-living data. The purpose of this study was to develop a modeling pipeline for evaluation of a CNN model on a free-living data set and compare CNN inputs and results with the commonly used machine learning random forest and logistic regression algorithms. Method: Twenty-eight free-living women wore an ActiGraph GT3X+ accelerometer on their right hip for 7 days. A concurrently worn thigh-mounted activPAL device captured ground truth activity labels. The authors evaluated logistic regression, random forest, and CNN models for classifying sitting, standing, and stepping bouts. The authors also assessed the benefit of performing feature engineering for this task. Results: The CNN classifier performed best (average balanced accuracy for bout classification of sitting, standing, and stepping was 84%) compared with the other methods (56% for logistic regression and 76% for random forest), even without performing any feature engineering. Conclusion: Using the recent advancements in deep neural networks, the authors showed that a CNN model can outperform other methods even without feature engineering. This has important implications for both the model’s ability to deal with the complexity of free-living data and its potential transferability to new populations.

Background: Thigh-worn accelerometers have established reliability and validity for measurement of free-living physical activity-related behaviors. However, comparisons of methods for measuring sleep and time in bed using the thigh-worn accelerometer are rare. The authors compared the thigh-worn accelerometer algorithm that estimates time in bed with the output of a sleep diary (time in bed and time asleep). Methods: Participants (N = 5,498), from the 1970 British Cohort Study, wore an activPAL device on their thigh continuously for 7 days and completed a sleep diary. Bland–Altman plots and Pearson correlation coefficients were used to examine associations between the algorithm derived and diary time in bed and asleep. Results: The algorithm estimated acceptable levels of agreement with time in bed when compared with diary time in bed (mean bias of −11.4 min; limits of agreement −264.6 to 241.8). The algorithm-derived time in bed overestimated diary sleep time (mean bias of 55.2 min; limits of agreement −204.5 to 314.8 min). Algorithm and sleep diary are reasonably correlated (ρ = .48, 95% confidence interval [.45, .52] for women and ρ = .51, 95% confidence interval [.47, .55] for men) and provide broadly comparable estimates of time in bed but not for sleep time. Conclusions: The algorithm showed acceptable estimates of time in bed compared with diary at the group level. However, about half of the participants were outside of the ±30 min difference of a clinically relevant limit at an individual level.

Little is known about how sedentary behavior (SB) metrics derived from hip- and thigh-worn accelerometers agree for older adults. Thigh-worn activPAL (AP) micro monitors were concurrently worn with hip-worn ActiGraph (AG) GT3X+ accelerometers (with SB measured using the 100 counts per minute [cpm] cut point; AG_100cpm) by 953 older adults (age 77 ± 6.6, 54% women) for 4–7 days. Device agreement for sedentary time and five SB pattern metrics was assessed using mean error and correlations. Logistic regression tested associations with four health outcomes using standardized (i.e., z scores) and unstandardized SB metrics. Mean errors (AP − AG_100cpm) and 95% limits of agreement were: sedentary time −54.7 [−223.4, 113.9] min/day; time in 30+ min bouts 77.6 [−74.8, 230.1] min/day; mean bout duration 5.9 [0.5, 11.4] min; usual bout duration 15.2 [0.4, 30] min; breaks in sedentary time −35.4 [−63.1, −7.6] breaks/day; and alpha −.5 [−.6, −.4]. Respective Pearson correlations were: .66, .78, .73, .79, .51, and .40. Concordance correlations were: .57, .67, .40, .50, .14, and .02. The statistical significance and direction of associations were identical for AG_100cpm and AP metrics in 46 of 48 tests, though significant differences in the magnitude of odds ratios were observed among 13 of 24 tests for unstandardized and five of 24 for standardized SB metrics. Caution is needed when interpreting SB metrics and associations with health from AG_100cpm due to the tendency for it to overestimate breaks in sedentary time relative to AP. However, high correlations between AP and AG_100cpm measures and similar standardized associations with health outcomes suggest that studies using AG_100cpm are useful, though not ideal, for studying SB in older adults.

Activity trackers provide real-time sedentary behavior (SB) and physical activity (PA) data enabling feedback to support behavior change. The validity of activity trackers in an obese population in a free-living environment is largely unknown. This study determined the convergent validity of the Fitbit Charge 2 in measuring SB and PA in overweight adults. The participants (n = 59; M ± SD: age = 48 ± 11 years; body mass index = 34 ± 4 kg/m²) concurrently wore a Charge 2 and ActiGraph GT3X+ accelerometer for 8 days. The same waking wear periods were analyzed, and standard cut points for GT3X+ and proprietary algorithms for the Charge 2, together with a daily step count, were used. Associations between outputs, mean difference (MD) and limits of agreement (LOA), and relative differences were assessed. There was substantial association between devices (intraclass correlation coefficients from .504, 95% confidence interval [.287, .672] for SB, to .925, 95% confidence interval [.877, .955] for step count). In comparison to the GT3X+, the Charge 2 overestimated SB (MD = 37, LOA = −129 to 204 min/day), moderate to vigorous PA (MD = 15, LOA = −49 to 79 min/day), and steps (MD = 1,813, LOA = −1,066 to 4,691 steps/day), and underestimated light PA (MD = −32, LOA = −123 to 58 min/day). The Charge 2 may be a useful tool for self-monitoring of SB and PA in an overweight population, as mostly good agreement was demonstrated with the GT3X+. However, there were mean and relative differences, and the implications of these need to be considered for overweight adult populations who are already at risk of being highly sedentary and insufficiently active.

Background: The evidence base regarding validity of wearable fitness trackers for assessment and/or modification of physical activity behavior is evolving. Accurate assessment of moderate- to vigorous-intensity physical activity (MVPA) is important for measuring adherence to physical activity guidelines in the United States and abroad. Therefore, this systematic review synthesizes the state of the validation literature regarding wearable trackers and MVPA. Methods: A systematic search of the PubMed, Scopus, SPORTDiscus, and Cochrane Library databases was conducted through October 2019 (PROSPERO registration number: CRD42018103808). Studies were eligible if they reported on the validity of MVPA and used devices from Fitbit, Apple, or Garmin released in 2012 or later or available on the market at the time of review. A meta-analysis was conducted on the correlation measures comparing wearables with the ActiGraph. Results: Twenty-two studies met the inclusion criteria; all used a Fitbit device; one included a Garmin model and no Apple-device studies were found. Moderate to high correlations (.7–.9) were found between MVPA from the wearable tracker versus criterion measure (ActiGraph n = 14). Considerable heterogeneity was seen with respect to the specific definition of MVPA for the criterion device, the statistical techniques used to assess validity, and the correlations between wearable trackers and ActiGraph across studies. Conclusions: There is a need for standardization of validation methods and reporting outcomes in individual studies to allow for comparability across the evidence base. Despite the different methods utilized within studies, nearly all concluded that wearable trackers are valid for measuring MVPA.

Automatic tagging of video recordings of sports matches and training sessions can be helpful to coaches and players and provide access to structured data at a scale that would be unfeasible if one were to rely on manual tagging. Recognition of different actions forms an essential part of sports video tagging. In this paper, the authors employ machine learning techniques to automatically recognize specific types of volleyball actions (i.e., underhand serve, overhead pass, serve, forearm pass, one hand pass, smash, and block which are manually annotated) during matches and training sessions (uncontrolled, in the wild data) based on motion data captured by inertial measurement unit sensors strapped on the wrists of eight female volleyball players. Analysis of the results suggests that all sensors in the inertial measurement unit (i.e., magnetometer, accelerometer, barometer, and gyroscope) contribute unique information in the classification of volleyball actions types. The authors demonstrate that while the accelerometer feature set provides better results than other sensors, overall (i.e., gyroscope, magnetometer, and barometer) feature fusion of the accelerometer, magnetometer, and gyroscope provides the bests results (unweighted average recall = 67.87%, unweighted average precision = 68.68%, and κ = .727), well above the chance level of 14.28%. Interestingly, it is also demonstrated that the dominant hand (unweighted average recall = 61.45%, unweighted average precision = 65.41%, and κ = .652) provides better results than the nondominant (unweighted average recall = 45.56%, unweighted average precision = 55.45, and κ = .553) hand. Apart from machine learning models, this paper also discusses a modular architecture for a system to automatically supplement video recording by detecting events of interests in volleyball matches and training sessions and to provide tailored and interactive multimodal feedback by utilizing an HTML5/JavaScript application. A proof of concept prototype developed based on this architecture is also described.

Traditional physical activity tracking tools, such as self-report questionnaires, are inherently subjective and vulnerable to bias. Physical activity tracking technology, such as activity tracking wristbands, is becoming more reliable and readily available. As such, researchers are employing these objective measurement tools in both observational- and intervention-based studies. There remains a gap in the literature on how to properly select activity tracking wristbands for research, specifically for the older adult population. This paper outlines considerations for choosing the most appropriate wrist-worn wearable device for use in research with older adults. Device features, outcome measures, population, and methodological considerations are explored.