Browse

Understanding the simultaneous patterning of accelerometer-measured physical activity and sedentary behavior (physical behaviors) can inform targeted interventions. This cross-sectional study described multicomponent patterns and correlates of physical behaviors using accelerometry among diverse postmenopausal women. The Women’s Health Accelerometry Collaboration combined two United States-based cohorts of postmenopausal women with similar accelerometry protocols and measures. Women (n = 22,612) aged 62–97 years enrolled in the Women’s Health Study (n = 16,742) and the Women’s Health Initiative Objective Physical Activity and Cardiovascular Health Study (n = 5,870) wore an ActiGraph GT3X+ accelerometer on their hip for 1 week. Awake-time accelerometry data were summarized using the accelerometer activity index into sedentary behavior, light (low, high), and moderate to vigorous physical activity. Latent class analysis was used to classify physical behavior hour-by-hour. Five unique patterns were identified with higher total volume of physical activity and lower sedentary behavior with each successively higher class number based on percentage of the day in physical activity/sedentary behavior per hour over 7 days. The percentage assignment was 16.3% Class 1, 33.9% Class 2, 20.2% Class 3, 18.0% Class 4, and 11.7% Class 5. The median posterior probabilities ranged from 0.99 to 1.00. Younger age, higher education and general health, normal weight, never smokers, weekly drinking, and faster self-reported walking speed generally had higher-class assignments compared with their counterparts. History of diabetes and cardiovascular disease generally had lower-class assignments compared with those without these conditions. These results can inform targeted interventions based on common patterns of physical behaviors by time of day among postmenopausal women.

Background: The aim of this study was to systematically review the findings for validity, reliability, and acceptability of device-based measures of office-based physical activity and/or sedentary time in an office context to evaluate workplace interventions. Methods: The review was conducted according to the Preferred Reporting Items for Systematic Reviews and Analysis guidelines. Five electronic databases (PubMed, EMBASE, CINAHL, Cochrane, and Web of Science) were searched (inception to December 2023). Keywords included population (e.g., workers), type of measure (e.g., device-based), measurement constructs (e.g., validity), context (e.g., office), and behavior (e.g., sitting). Two authors screened titles, abstracts, and full texts independently with disagreements resolved by a third author. Findings were reported using narrative synthesis, and COnsensus-based standards for the Selection of health status Measurement INstruments was used for quality assessment. Results: In total, 2,299 articles were identified, with 16 articles retained. These reported 21 measurement protocols (nine in free-living settings) assessing eight worn, four remote, and one combined method. Sixteen protocols assessed office sitting, with standing (n = 8), moving (n = 11), postural transitions (n = 7), and location (n = 2) also assessed. Participant sample sizes ranged from one to 42 (median = 13). Criterion validity was assessed in all 21 protocols, with lower limb–worn measures of sitting, and worn and remote measures of location reporting the highest validity/accuracy compared with the ground truth (good to excellent). Only two articles reported acceptability (good acceptability), with none reporting reliability. Conclusions: There is evidence of valid device-based measures of office behavior (particularly sitting and location of workers), but this has largely been obtained in laboratory settings and/or with small samples. Larger studies in more varied free-living settings, potentially using multiples sources of data and assessing acceptability, are required.

Measuring infants’ (0–2 years) physical activity is a growing area of research globally. Accelerometers have been widely used to measure older children’s and adults’ physical activity. An increasing number of studies have used accelerometers as a way to measure infant physical activity, which has resulted in the application of a variety of methods. The purpose of this scoping review is to synthesize the published literature on accelerometer methodology to measure daytime physical activity among infants (0–2 years). A systematic search of five online databases using carefully selected key terms was conducted to compile relevant literature. The results of the online database searches were screened for inclusion in the scoping review. In total, 105 articles met the inclusion criteria of using accelerometers to measure infants’ physical activity. The methodologies used in the included studies were categorized by age groups: <1 month, 1–6 months, >6–12 months, >12–18 months, >18–24 months, and longitudinal (i.e., multiple measurements taken across the previously mentioned age groups). Accelerometry methodologies (e.g., wear location, number of devices, device initialization) and study design qualities (e.g., outcome of interest and location of data collection) varied widely between and within the various age groups. Accelerometer brand or type of device demonstrated greatest variation across included studies. However, ActiGraph devices to measure physical activity within free-living environments were the most common. This review provides evidence of the need for researchers to ensure the methodology used is reported in detail in order to help develop methodology that can accurately assess infant daytime movement.

Accelerometers have been at the forefront of free-living activity capture for decades, and accordingly ActiGraph the largest distributor. Historically, limitations in data storage and battery power led to the use of summary metrics, which have been termed activity counts. Recently, ActiGraph publicly released their count-based algorithm, marking a notable development in the field. This study aimed to assess and compare activity counts generated through different processing techniques (ActiLife and open-source), filters that are available through ActiGraph count generation (normal- and low-frequency extension), and data from various ActiGraph models and GENEActiv devices. We evaluated ActiGraph GT3X+ (n = 8), ActiGraph wGT3X-BT (n = 10), ActiGraph GT9X (n = 8; primary and secondary sensors), OPAL (n = 6), and GENEActiv (n = 5), subjected to oscillations across their full dynamic range (0.005–8 G) using a multiaxis shaker table. Results indicated that the low-frequency extension produced significantly higher counts compared to the normal frequency across the devices and processing techniques. Notably, open-source counts (R and Python) were statistically equivalent to ActiLife-generated counts (p < .05) for the GT9X, wGT3X-BT, and the GT3X+. Overall, many of the counts generated by different ActiGraph models were statistically equivalent or had mean differences <5.03 counts. Conversely, the GENEActiv, OPAL, and GT9X secondary monitor exhibited significantly higher responses than the other ActiGraph models at higher frequencies with mean differences ranging from 55.50 to 104.91 counts. This study provides insights into accelerometer data processing methods and highlights the comparability of counts across different devices and techniques.

Activity counts have been used for over two decades with over 22,000 published scientific papers in public health and clinical research. ActiGraph recently released the algorithm for computing counts from raw accelerometer data as an open-source Python library, which is now ported by researchers to other languages, notably R. The current commentary presents historical overview of ActiGraph counts, and its development and evolution as a measure of physical activity. Further, we provide general recommendations on extracting counts from raw accelerometer data and discuss specific considerations with respect to device types, resampling, nonwear, axes orientations, and epoch length that may influence counts. Last, we provide a tutorial on how to use ActiGraph’s open-source Python library, agcounts, for consistent, accurate, and reproducible count. We expect this commentary will provide familiarity and transparency needed to adopt and produce activity counts in a consistent manner, allowing researchers to conduct statistical comparisons across multiple data sets and studies.

Background: Walking-based metrics, including step count and total time walking, are easily interpretable measures of physical activity. Algorithms can estimate steps from accelerometry, which increasingly is measured with accelerometers located on the wrist. However, many existing step counting algorithms have not been validated in free-living settings, exhibit high error rates, or cannot be used without proprietary software. We compare the performance of several existing open-source step counting algorithms on three publicly available data sets, including one with free-living data. Methods: We applied five open-source algorithms: Adaptive Empirical Pattern Transformation, Oak, Step Detection Threshold, Verisense, and stepcount, and one proprietary algorithm (ActiLife) to three publicly available data sets with ground truth step counts: Clemson Ped-Eval, Movement Analysis in Real-World Environments Using Accelerometers, and OxWalk. We evaluate F1 score, precision, recall, mean absolute percent error (MAPE), and mean bias for each algorithm and setting. Results: The machine learning-based stepcount algorithm exhibited the highest F1 score (0.89 ± 0.11) and lowest MAPE (8.6 ± 9%) across all data sets and had the best, or comparable, F1 scores and MAPE in each individual data set. All algorithms performed worse with respect to both F1 score and MAPE in free-living compared with regular walking scenarios, and stepcount and Verisense were most sensitive to sampling frequency of input data. Conclusion: Machine learning-based algorithms, including stepcount, are a promising avenue for step counting. More free-living accelerometry data sets with ground truth step counts are needed for testing, validation, and continued refinement of algorithms.