The purpose of this study was to conduct a comprehensive evaluation of the ActiGraph GT3X+ (AG) and activPAL (AP) for assessing time spent in sedentary behaviors (SB) in youth using structured and free-living activities. Forty-four participants (M age, 12.7±0.8 yrs) completed up to eight structured activities and approximately 2 hrs of free-living activity while wearing an AG (right hip) and AP (right thigh). A Cosmed K4b2 was used for measured energy expenditure (METy; activity VO2 ÷ resting VO2). Direct observation was used during the structured activities. SB time was estimated using the inclinometer function of the AP and AG, and count thresholds with AG (<75 vector magnitude [VM] counts/10-s; <25 vertical axis [VA] counts/10-s; and <50, 100, 150, and 200 VA counts/min). For the structured activities, the AG inclinometer and AP correctly classified supine rest about 45% of the time, seated activities 54.6% and 65.1% of the time, respectively, and walking and running >96% of the time. For the free-living measurement, the VA <25 counts/10-s had the lowest RMSE (20.6 min), while the VM <75 counts/10-s had the lowest MAPE (69.2%). The AG inclinometer was within 0.2 minutes of measured time, but had the highest MAPE (107.1%). The AP was within 1.6 minutes of measured time, but had the highest RMSE (28.5 minutes). Compared to measured SB time, the VA <25 counts/10-s and VM <75 counts/10-s provided the most precise estimates of SB during free-living activity. Further refinement is needed to improve the AP and AG posture estimates.
Scott E. Crouter, Paul R. Hibbing, and Samuel R. LaMunion
Paul R. Hibbing, Seth A. Creasy, and Jordan A. Carlson
Physical behaviors (e.g., sleep, sedentary behavior, and physical activity) often occur in sustained bouts that are punctuated with brief interruptions. To detect and classify these interrupted bouts, researchers commonly use wearable devices and specialized algorithms. Most algorithms examine the data in chronological order, initiating and terminating bouts whenever specific criteria are met. Consequently, the bouts may encapsulate or overlap with later periods that also meet the activation and termination criteria (i.e., alternative bout solutions). In some cases, it is desirable to compare these alternative bout solutions before making a final classification. Thus, comparison-focused algorithms are needed, which can be used in isolation or in concert with their chronology-focused counterparts. In this technical note, we present a comparison-focused algorithm called CRIB (Clustered Recognition of Interrupted Bouts). It uses agglomerative hierarchical clustering to facilitate the comparison of different bout solutions, with the final classification being made in favor of the smallest number of bouts that comply with user-specified criteria (i.e., limits on the number, individual duration, and cumulative duration of interruptions). For demonstration, we use CRIB to assess bouts of moderate to vigorous physical activity in accelerometer data from the National Health and Nutrition Examination Survey, and we include a comparison against results from two established chronology-focused algorithms. Our discussion explores strengths and limitations of CRIB, as well as potential considerations and applications for using it in future studies. An online vignette (https://github.com/paulhibbing/PBpatterns/blob/main/vignettes/CRIB.pdf) is available to assist users with implementing CRIB in R.
Paul R. Hibbing, Nicholas R. Lamoureux, Charles E. Matthews, and Gregory J. Welk
Physical behavior can be assessed using a range of competing methods. The Free-Living Activity Study for Health (FLASH) is an ongoing study that facilitates the comparison of such methods. The purpose of this report is to describe the FLASH, with a particular emphasis on a subsample of participants who have consented to have their deidentified data released in a shared repository. Participants in the FLASH wear seven physical activity monitors for a 24-hr period and then complete a detailed recall using the Activities Completed Over Time in 24-hr online assessment tool. The participants can optionally agree to be video recorded for 30–60 min, which allows for direct observation as a criterion indicator of their behavior during that period. As of version 0.1.0, the repository includes data from 38 participants, and the sample size will grow as data are collected, processed, and released in future versions. The repository makes it possible to combine sensor data (e.g., from ActiGraph and SenseWear) with minute-by-minute contextual data (from the Activities Completed Over Time in 24-hr recall system), which enables the FLASH to generate benchmark data for a wide range of future research. The repository itself provides an example of how a powerful open-source tool (GitHub) can be used to share data and code in a way that encourages communication and collaboration among a variety of scientists (e.g., algorithm developers and end users). The FLASH data set will provide long-term benefits to researchers interested in advancing the science of physical behavior monitoring.
Paul R. Hibbing, Samuel R. LaMunion, Haileab Hilafu, and Scott E. Crouter
Background: Bout detection algorithms are used to segment data from wearable sensors, but it is challenging to assess segmentation correctness. The purpose of this study was to present and demonstrate the Transition Pairing Method (TPM), a new method for evaluating the performance of bout detection algorithms. Methods: The TPM compares predicted transitions to a criterion measure in terms of number and timing. A true positive is defined as a predicted transition that corresponds with one criterion transition in a mutually exclusive pair. The pairs are established using an extended Gale-Shapley algorithm, and the user specifies a maximum allowable within-pair time lag, above which pairs cannot be formed. Unpaired predictions and criteria are false positives and false negatives, respectively. The demonstration used raw acceleration data from 88 youth who wore ActiGraph GT9X monitors (right hip and non-dominant wrist) during simulated free-living. Youth Sojourn bout detection algorithms were applied (one for each attachment site), and the TPM was used to compare predicted bout transitions to the criterion measure (direct observation). Performance metrics were calculated for each participant, and hip-versus-wrist means were compared using paired t-tests (α = 0.05). Results: When the maximum allowable lag was 1-s, both algorithms had recall <20% (2.4% difference from one another, p < .01) and precision <10% (1.4% difference from one another, p < .001). That is, >80% of criterion transitions were undetected, and >90% of predicted transitions were false positives. Conclusion: The TPM improves on conventional analyses by providing specific information about bout detection in a standardized way that applies to any bout detection algorithm.
Nicholas R. Lamoureux, Paul R. Hibbing, Charles Matthews, and Gregory J. Welk
Accelerometry-based monitors are commonly utilized to evaluate physical activity behavior, but the lack of contextual information limits the interpretability and value of the data. Integration of report-based with monitor-based data allows the complementary strengths of the two approaches to be used to triangulate information and to create a more complete picture of free-living physical behavior. This investigation utilizes data collected from the Free-Living Activity Study for Health to test the feasibility of annotating monitor data with contextual information from the Activities Completed Over Time in 24-hr (ACT24) previous-day recall. The evaluation includes data from 134 adults who completed the 24-hr free-living monitoring protocol and retrospective 24-hr recall. Analyses focused on the relative agreement of energy expenditure estimates between ACT24 and two monitor-based methods (ActiGraph and SenseWear Armband). Daily energy expenditure estimates from ACT24 were equivalent to the reference device-based estimate. Minute-level agreement of energy expenditure between ACT24 and device-based methods was moderate and was similar to the agreement between two different monitor-based methods. This minute-level agreement between ACT24 and device-based methods demonstrates the feasibility and utility of integrating self-report with accelerometer data to provide richer context on the monitored behaviors. This type of integration offers promise for advancing the assessment of physical behavior by aiding in data interpretation and providing opportunities to improve physical activity assessment methods under free-living conditions.
Elizabeth L. Stegemöller, Joshua R. Tatz, Alison Warnecke, Paul Hibbing, Brandon Bates, and Andrew Zaman
Auditory cues, including music, are commonly used in the treatment of persons with Parkinson’s disease. Yet, how music style and movement rate modulate movement performance in persons with Parkinson’s disease have been neglected and remain limited in healthy young populations. The purpose of this study was to determine how music style and movement rate influence movement performance in healthy young adults. Healthy participants were asked to perform repetitive finger movements at two pacing rates (70 and 140 beats per minute) for the following conditions: (a) a tone only, (b) activating music, and (c) relaxing music. Electromyography, movement kinematics, and variability were collected. Results revealed that the provision of music, regardless of style, reduced amplitude variability at both pacing rates. Intermovement interval was longer, and acceleration variability was reduced during both music conditions at the lower pacing rate only. These results may prove beneficial for designing therapeutic interventions for persons with Parkinson’s disease.
Adrian Ortega, Bethany Forseth, Paul R. Hibbing, Chelsea Steel, and Jordan A. Carlson
Purpose: We investigated convergent validity of commonly used ActiGraph scoring methods with various activPAL scoring methods in youth and adults. Methods: Youth and adults wore an ActiGraph and activPAL simultaneously for 1–3 days. We compared moderate to vigorous physical activity (MVPA) estimates from the ActiGraph Evenson 15-s (youth) and Freedson 60-s (adult) cut-point scoring methods and four activPAL scoring methods based on metabolic equivalents (METs), step counts, vertical axis counts, and vector magnitude counts. All activPAL methods were applied to 15-s epochs for youth and 60-s epochs for adults, and the METs method was also applied to 1-s epochs. Epoch-level agreement was examined with classification tests (sensitivity, positive predictive value, and F1) using the ActiGraph methods as the referent. Day-level agreement was examined using tests of mean error, mean absolute error, and Spearman correlations. Results: Relative to ActiGraph methods, which indicated a mean MVPA of 41 min/day for youth and 24 min/day for adults, the activPAL METs method applied to 15-s epochs in youth and 60-s epochs in adults yielded the most comparable estimates of MVPA. Daily MVPA estimated from all other activPAL scoring methods generally had poor agreement with ActiGraph methods in youth and adults. Conclusion: When using the same epoch lengths between monitors, MVPA estimation via the activPAL METs scoring method appears to have good comparability to ActiGraph cut points at the group-level and moderate comparability at the individual-level in youth and adults. When using this scoring method, the activPAL appears to be appropriate for measuring daily minutes of MVPA in youth and adults.
Susan Park, Lindsay P. Toth, Paul R. Hibbing, Cary M. Springer, Andrew S. Kaplan, Mckenzie D. Feyerabend, Scott E. Crouter, and David R. Bassett
It has become common to wear physical activity monitors on the wrist to estimate steps per day, but few studies have considered step differences between monitors worn on the dominant and non-dominant wrists. Purpose: The purpose of this study was to compare four step counting methods on the dominant versus non-dominant wrist using the Fitbit Charge (FC) and ActiGraph GT9X (GT9X) across all waking hours of one day. Methods: Twelve participants simultaneously wore two monitors (FC and GT9X) on each wrist during all waking hours for an entire day. GT9X data were analyzed with three step counting methods: ActiLife algorithm with default filter (AG-noLFE), ActiLife algorithm with low-frequency extension (AG-LFE), and the Moving Average Vector Magnitude (AG-MAVM) algorithm. A 2-way repeated measures ANOVA (method × wrist) was used to compare step counts. Results: There was a significant main effect for wrist placement (F(1,11) = 11.81, p = .006), with the dominant wrist estimating an average of 1,253 more steps than the non-dominant wrist. Steps differed between the dominant and non-dominant wrist for three of the step methods: AG-noLFE (1,327 steps), AG-LFE (2,247 steps), AG-MAVM (825 steps), and approached statistical significance for FC (613 steps). No significant method x wrist placement interaction was found (F(3,9) = 2.62, p = .115). Conclusion: Findings suggest that for step counting algorithms, it may be important to consider the placement of wrist-worn monitors since the dominant wrist location tended to yield greater step estimates. Alternatively, standardizing the placement of wrist-worn monitors could help to reduce the differences in daily step counts across studies.
Gregory J. Welk, Pedro F. Saint-Maurice, Philip M. Dixon, Paul R. Hibbing, Yang Bai, Gabriella M. McLoughlin, and Michael Pereira da Silva
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.
Chelsea Steel, Katie Crist, Amanda Grimes, Carolina Bejarano, Adrian Ortega, Paul R. Hibbing, Jasper Schipperijn, and Jordan A. Carlson
Objective: To investigate the convergent validity of a global positioning system (GPS)-based and two consumer-based measures with trip logs for classifying pedestrian, cycling, and vehicle trips in children and adults. Methods: Participants (N = 34) wore a Qstarz GPS tracker, Fitbit Alta, and Garmin vivosmart 3 on multiple days and logged their outdoor pedestrian, cycling, and vehicle trips. Logged trips were compared with device-measured trips using the Personal Activity Location Measurement System (PALMS) GPS-based algorithms, Fitbit’s SmartTrack, and Garmin’s Move IQ. Trip- and day-level agreement were tested. Results: The PALMS identified and correctly classified the mode of 75.6%, 94.5%, and 96.9% of pedestrian, cycling, and vehicle trips (84.5% of active trips, F1 = 0.84 and 0.87) as compared with the log. Fitbit and Garmin identified and correctly classified the mode of 26.8% and 17.8% (22.6% of active trips, F1 = 0.40 and 0.30) and 46.3% and 43.8% (45.2% of active trips, F1 = 0.58 and 0.59) of pedestrian and cycling trips. Garmin was more prone to false positives (false trips not logged). Day-level agreement for PALMS and Garmin versus logs was favorable across trip modes, though PALMS performed best. Fitbit significantly underestimated daily cycling. Results were similar but slightly less favorable for children than adults. Conclusions: The PALMS showed good convergent validity in children and adults and were about 50% and 27% more accurate than Fitbit and Garmin (based on F1). Empirically-based recommendations for improving PALMS’ pedestrian classification are provided. Since the consumer devices capture both indoor and outdoor walking/running and cycling, they are less appropriate for trip-based research.