Browse

You are looking at 1 - 10 of 174 items for :

  • Journal for the Measurement of Physical Behaviour x
  • Refine by Access: All Content x
Clear All
Free access

The Intrinsic Properties of ActiGraph Counts and Alternatives

Jan Christian Brønd, Niels Christian Møller, and Anders Grøntved

There are currently several methods available to generate summary measures from acceleration, while ActiGraph (AG) counts as the first method to be used at large scale. The recent disclosure of the AG counts method exposes its intrinsic properties, which has not been accessible before. The intrinsic properties are the raw acceleration processing elements like filtering, rectification, or dead-band elimination, which are used to estimate physical activity intensity. The aim of this technical note is to compare the intrinsic properties of AG counts method with five alternatives (Euclidean Norm Minus One, mean average deviation, Activity Index, Rate of Change Accelerometry Movement, and Monitor-Independent Movement Summary) and how rescaling of AG counts and Monitor-Independent Movement Summary/minute into the International System of Units can be used to harmonize all summary measures and facilitate direct comparison. A total of 12 intrinsic properties are compared, and the overview demonstrates that there is large diversity regarding the specific intrinsic property elements being included, and with Monitor-Independent Movement Summary to be the only summary measure, which has been developed considering all elements. The harmonized output generated from all summary methods is highly comparable within common activities, but to obtain a robust summary measure recorded in subjects during free-living conditions, more research is warranted to evaluate the effect of the different intrinsic properties.

Restricted access

Distinguishing Passive and Active Standing Behaviors From Accelerometry

Robert J. Kowalsky, Herman van Werkhoven, Marco Meucci, Tyler D. Quinn, Lee Stoner, Christopher M. Hearon, and Bethany Barone Gibbs

Purpose: To investigate whether active standing can be identified separately from passive standing via accelerometry data and to develop and test the accuracy of a machine-learning model to classify active and passive standing. Methods: Ten participants wore a thigh-mounted activPAL monitor and stood for three 5-min periods in the following order: (a) PASSIVE: standing with no movement; (b) ACTIVE: five structured weight-shifting micromovements in the medial–lateral, superior–inferior, and anterior–poster planes while standing; and (c) FREE: participant’s choice of active standing. Averages of absolute resultant acceleration values in 15-s epochs were compared via analysis of variance (Bonferroni adjustment for pairwise comparisons) to confirm the dichotomization ability of the standing behaviors. Absolute resultant acceleration values and SDs in 2- and 5-s epochs were used to develop a machine-learning model using leave-one-subject-out cross validation. The final accuracy of the model was assessed using the area under the curve from a receiver operating characteristic curve. Results: Comparison of resultant accelerations across the three conditions (PASSIVE, ACTIVE, and FREE) resulted in a significant omnibus difference, F(2, 19) = [116], p < .001, η2 = .86, and in all pairwise post hoc comparisons (all p < .001). The machine-learning model using 5-s epochs resulted in 94% accuracy for the classification of PASSIVE versus ACTIVE standing. Model application to the FREE data resulted in an absolute average difference of 4.8% versus direct observation and an area under the curve value of 0.71. Conclusions: Active standing in three planes of movement can be identified from thigh-worn accelerometry via a machine-learning model, yet model refinement is warranted.

Restricted access

Criterion Validity of Commonly Used Sedentary Behavior Questionnaires to Measure Total Sedentary Time in Adults

Madeline E. Shivgulam, Derek S. Kimmerly, and Myles W. O’Brien

Background: Self-report questionnaires are a fast and cost-efficient method to determine habitual sedentary time (sitting/lying time while awake), but their accuracy versus thigh-worn accelerometry (criterion), which can distinguish between sitting and standing postures, is unclear. While the validity of sedentary questionnaires has previously been evaluated, they have not been investigated simultaneously in the same sample population. We tested the hypothesis that common sedentary questionnaires underpredict habitual sedentary time compared with an objective, monitor-based assessment. Methods: Ninety-three participants (30 ± 18 years, 59 females) wore the activPAL inclinometer on the midthigh 24 hr per day for 6.9 ± 0.4 days and completed the SIT-Q, Sedentary Behavior Questionnaire (SBQ), International Physical Activity Questionnaire (IPAQ), and Physical Activity and Sedentary Behavior Questionnaire (PASB-Q). Results: In comparison to the activPAL (9.9 ± 1.9 hr/day), the SIT-Q measured more time (12.9 ± 5.4 hr/day), but the SBQ (7.5 ± 3.3 hr/day), IPAQ (7.4 ± 3.0 hr/day), and PASB-Q (6.6 ± 3.0 hr/day) measured less time (all p < .001). The SIT-Q was positively and weakly correlated (ρ = .230 [95% confidence interval: .020, .422], p = .028) with the activPAL, but the SBQ, IPAQ, and PASB-Q were not (all ps > .760). Equivalence testing demonstrated poor equivalence for the SIT-Q (±40%), SBQ (±31%), IPAQ (±36%), and PASB-Q (±29%). The SIT-Q (β = −1.36), SBQ (β = −0.97), and IPAQ (β = −0.78) exhibited a negative proportional bias (all ps < .002). Conclusions: In summary, the SIT-Q, SBQ, IPAQ, and PASB-Q demonstrated poor validity. Researchers and health promoters should be cautious when implementing these self-report sedentary time questionnaires, as they may not reflect the true sedentary activity and negatively impact study results.

Free access

agcounts: An R Package to Calculate ActiGraph Activity Counts From Portable Accelerometers

Brian C. Helsel, Paul R. Hibbing, Robert N. Montgomery, Eric D. Vidoni, Lauren T. Ptomey, Jonathan Clutton, and Richard A. Washburn

Portable accelerometers are used to capture physical activity in free-living individuals with the ActiGraph being one of the most widely used device brands in physical activity and health research. Recently, in February 2022, ActiGraph published their activity count algorithm and released a Python package for generating activity counts from raw acceleration data for five generations of ActiGraph devices. The nonproprietary derivation of the ActiGraph count improved the transparency and interpretation of accelerometer device-measured physical activity, but the Python release of the count algorithm does not integrate with packages developed by the physical activity research community using the R Statistical Programming Language. In this technical note, we describe our efforts to create an R-based translation of ActiGraph’s Python package with additional extensions to make data processing easier and faster for end users. We call the resulting R package agcounts and provide an inside look at its key functionalities and extensions while discussing its prospective impacts on collaborative open-source software development in physical behavior research. We recommend that device manufacturers follow ActiGraph’s lead by providing open-source access to their data processing algorithms and encourage physical activity researchers to contribute to the further development and refinement of agcounts and other open-source software.

Restricted access

Comparing Accelerometer Processing Metrics and Hyperparameter Optimization for Physical Activity Classification Accuracy Using Machine Learning Methods

Sumayyah Bamidele Musa, Arnab Barua, Kevin G. Stanley, Fabien A. Basset, Hiroshi Mamyia, Kevin Mongeon, and Daniel Fuller

Background: Physical activity (PA) is a crucial factor in maintaining good health and preventing chronic diseases. However, accurately measuring PA is challenging. Euclidean Norm Minus One (ENMO), ActiGraph Counts, and Monitor-Independent Movement Summary (MIMS) units are processing metrics used to classify PA through accelerometry, but they employ different methods to calculate activity levels. This study aimed to compare ENMO, ActiGraph Counts, and MIMS accelerometer metrics using machine learning algorithms. Methods: Data from a smartphone accelerometer were collected from 50 participants who held the smartphone in their right hand while completing six activities. The data were used to generate ENMO, ActiGraph Counts, and MIMS acceleration metrics. Random Forest, K-Nearest Neighbor, and Support Vector Machine algorithms were applied to the data to classify PA into different levels of activity intensity and types. The algorithms’ performance was evaluated using various metrics such as accuracy, precision, and recall. Results: The results showed that both the Random Forest and K-Nearest Neighbor algorithms performed well, achieving above 80% accuracy in classifying PA into different intensity levels and types. Both the ENMO and MIMS metrics proved more accurate than ActiGraph Counts in classifying moderate to vigorous PA. Conclusions: This study provides evidence that both ENMO and MIMS metrics can accurately measure PA with accelerometry, and machine learning algorithms can classify the activity into different intensity levels. These metrics and methods are valuable tools for monitoring PA and understanding the relationship between PA and health outcomes.

Restricted access

Examining the Agreement Between the activPAL micro4 and ActiGraph GT9X Accelerometers on Daily Movement Behaviors Among Adults With Total Knee Replacement

Katherine E. DeVivo, Chih-Hsiang Yang, and Christine A. Pellegrini

Objective: The primary purpose was to examine the agreement in sedentary, light, and moderate to vigorous minutes and step counts between the activPAL micro4 and ActiGraph GT9X in adults following total knee replacement. A secondary purpose was to examine the agreement between the activPAL micro4 and ActiGraph GT9X accelerometers at two different time points after surgery (∼1 and 3 months). Methods: Participants in a randomized trial wore ActiGraph GT9X and activPAL micro4 monitors simultaneously for 7 days at ∼1 and 3 months after total knee replacement. The intraclass correlations for time spent in sedentary behavior, light, and moderate to vigorous activity in addition to step counts were estimated to determine consistency between the two monitors. Bland–Altman plot demonstrated the 95% limits of agreement between the monitors at both time points. Results: A total of 480 observations (days) were used from 41 participants (64.9 ± 7 years, 32.4 ± 6.5 kg/m2, 75.6% White, 61% female). The intraclass correlations between the ActiGraph GT9X and activPAL micro4 accelerometers was .839 for sedentary behavior, .853 for light activity, .806 for moderate to vigorous activity, and .937 for steps. The 95% confidence intervals of intraclass correlations between time points indicate a significantly higher agreement between the monitors at 3 months as compared with 1 month. Conclusion: The results suggest that either the ActiGraph GT9X or the activPAL micro4 accelerometers may be used for measuring sedentary, light, and moderate to vigorous minutes and step counts in adults after knee replacement.

Free access

From Research to Application of Wearable-Derived Digital Health Measures—A Perspective From ActiGraph

Jeremy Wyatt and Christine C. Guo

ActiGraph counts were first conceptualized in 1996 to provide an accelerometer-derived metric that can quantify physical activity based on intensity. ActiGraph incorporated this metric into its product suite, enabling its wide adoption in research studies. Over the last 20 years, ActiGraph activity counts have become one of the most common metrics and building blocks of health outcome measures used in wearable research, with >24,000 journal articles published (based on Google Scholar search in 2023). Recently, this field of research is increasingly moving toward clinical application where wearable-derived metrics are growing in industry-sponsored clinical trials, including several use cases endorsed by the regulatory authorities. We celebrate this emerging trend as these patient-generated measures help reduce trial burden and enhance the meaningfulness of developed medical products to the patients. However, true adoption of digital measures in industry research is only in its infancy and still faces many challenges. As a digital health technology provider, ActiGraph has launched several strategic initiatives to support the research community to overcome these challenges and accelerate the translation of research to clinical application. The open-source release of the ActiGraph count algorithm was one of those initiatives. In this commentary, we take the opportunity to share our perspective in supporting the research community with this metric over the last 20 years, the motivation for making this open source, and what we are building to accelerate clinical adoption and realize the promise of better patient care.

Restricted access

Measurement Reactivity in Ecological Momentary Assessment Studies of Movement-Related Behaviors

Jaclyn P. Maher, Danielle Arigo, Kiri Baga, Gabrielle M. Salvatore, Kristen Pasko, Brynn L. Hudgins, and Laura M. König

Measurement reactivity has implications for behavioral science, as it is crucial to determine whether changes in constructs of interest represent true change or are an artifact of assessment. This study investigated whether measurement reactivity occurs for movement-related behaviors, motivational antecedents of behavior, and associations between them. Data from ecological momentary assessment studies of older adults (n = 195) and women in midlife (n = 75) lasting 8–10 days with 5–6 prompts/day and ambulatory monitoring of movement were used for this secondary data analysis. To examine potential drop-off patterns indicative of measurement reactivity, multilevel models tested whether behavior, antecedents, and associations changed after the first or first 2 prompts compared with remaining prompts and the first, first 2, or first 3 days compared with remaining days. Older adults’ sedentary behavior was lower, and time spent upright and intentions and self-efficacy to stand/move were higher on the first 2 and first 3 days compared with remaining days. Associations between intentions and self-efficacy and subsequent sedentary behavior were weaker earlier in the study compared to later. For women in midlife, light physical activity was higher at the first and first 2 prompts compared with remaining prompts, and physical activity motivation was higher across all prompts and days tested. There was a stronger association between intended and observed minutes of moderate to vigorous physical activity on the first 2 days compared with remaining days. Measurement reactivity appeared as expected for movement-related behaviors and motivational antecedents, though changes in associations between these constructs are likely do not reflect measurement reactivity.

Restricted access

Characterizing ActiGraph’s Idle Sleep Mode in Free-Living Assessments of Physical Behavior

Samuel R. LaMunion, Robert J. Brychta, Joshua R. Freeman, Pedro F. Saint-Maurice, Charles E. Matthews, Asuka Ishihara, and Kong Y. Chen

ActiGraph’s idle sleep mode (ISM) is an optional battery- and memory-conserving feature believed to engage during periods of nonwear, inactivity, and sleep, but it has not been well studied in free-living environments. Thus, we investigated ISM during a 7-day assessment in a nationally representative sample of 13,649 participants (6–80 years) in the United States and found it engaged 43.6% ± 0.2% (mean ± SE) of the 24 hr per day. ISM engagement was highest (78.4% ± 0.2%) during early morning (00:00–05:59) and lowest (20.4% ± 0.3%) during afternoon (12:00–17:59), corresponding to quadrants of lowest and highest of movement, respectively. ISM engagement was also inversely correlated with daily activity across all participants (R = −.72, p < .001). When restricted to participants averaging ≥21 hr per day of wear (N = 10,482), ISM still engaged 39.5% ± 0.2% of the day and inversely correlated to daily activity (R = −.58, p < .001). These results suggest ISM engages in activity level-dependent temporal patterns. Additional research is needed to better inform analyses and interpretation of ISM-enabled data including whether it is appropriate to process them with existing methods that were developed and validated using data without ISM enabled. This issue may be particularly relevant for methods used to detect and score sleep, as ISM engaged during a substantial portion of the typical overnight sleep period in the 8-hr window between ≥22:00 and <06:00 (74.0% ± 12.6%, mean ± SD).

Restricted access

Comparability of 24-hr Activity Cycle Outputs From ActiGraph Counts Generated in ActiLife and RStudio

Alexander H.K. Montoye, Kimberly A. Clevenger, Benjamin D. Boudreaux, and Michael D. Schmidt

Data from ActiGraph accelerometers have long been imported into ActiLife software, where the company’s proprietary “activity counts” were generated in order to understand physical behavior metrics. In 2022, ActiGraph released an open-source method to generate activity counts from any raw, triaxial accelerometer data using Python, which has been translated into RStudio packages. However, it is unclear if outcomes are comparable when generated in ActiLife and RStudio. Therefore, the authors’ technical note systematically compared activity counts and related physical behavior metrics generated from ActiGraph accelerometer data using ActiLife or available packages in RStudio and provides example code to ease implementation of such analyses in RStudio. In addition to comparing triaxial activity counts, physical behavior outputs (sleep, sedentary behavior, light-intensity physical activity, and moderate- to vigorous-intensity physical activity) were compared using multiple nonwear algorithms, epochs, cut points, sleep scoring algorithms, and accelerometer placement sites. Activity counts and physical behavior outcomes were largely the same between ActiLife and the tested packages in RStudio. However, peculiarities in the application of nonwear algorithms to the first and last portions of a data file (that occurred on partial, first or last days of data collection), differences in rounding, and handling of counts values on the borderline of activity intensities resulted in small but inconsequential differences in some files. The hope is that researchers and both hardware and software manufacturers continue to push efforts toward transparency in data analysis and interpretation, which will enhance comparability across devices and studies and help to advance fields examining links between physical behavior and health.