Measuring Free-Living Physical Activity With Three Commercially Available Activity Monitors for Telemonitoring Purposes: Validation Study (Preprint)

BACKGROUND Remote monitoring of physical activity in patients with chronic conditions could be useful to offer care professionals real-time assessment of their patient’s daily activity pattern to adjust appropriate treatment. However, the validity of commercially available activity trackers that can be used for telemonitoring purposes is limited. OBJECTIVE The purpose of this study was to test usability and determine the validity of 3 consumer-level activity trackers as a measure of free-living activity. METHODS A usability evaluation (study 1) and validation study (study 2) were conducted. In study 1, 10 individuals wore one activity tracker for a period of 30 days and filled in a questionnaire on ease of use and wearability. In study 2, we validated three selected activity trackers (Apple Watch, Misfit Shine, and iHealth Edge) and a fourth pedometer (Yamax Digiwalker) against the reference standard (Actigraph GT3X) in 30 healthy participants for 72 hours. Outcome measures were 95% limits of agreement (LoA) and bias (Bland-Altman analysis). Furthermore, median absolute differences (MAD) were calculated. Correction for bias was estimated and validated using leave-one-out cross validation. RESULTS Usability evaluation of study 1 showed that iHealth Edge and Apple Watch were more comfortable to wear as compared with the Misfit Flash. Therefore, the Misfit Flash was replaced by Misfit Shine in study 2. During study 2, the total number of steps of the reference standard was 21,527 (interquartile range, IQR 17,475-24,809). Bias and LoA for number of steps from the Apple Watch and iHealth Edge were 968 (IQR −5478 to 7414) and 2021 (IQR −4994 to 9036) steps. For Misfit Shine and Yamax Digiwalker, bias was −1874 and 2004, both with wide LoA of (13,869 to 10,121) and (−10,932 to 14,940) steps, respectively. The Apple Watch noted the smallest MAD of 7.7% with the Actigraph, whereas the Yamax Digiwalker noted the highest MAD (20.3%). After leave-one-out cross validation, accuracy estimates of MAD of the iHealth Edge and Misfit Shine were within acceptable limits with 10.7% and 11.3%, respectively. CONCLUSIONS Overall, the Apple Watch and iHealth Edge were positively evaluated after wearing. Validity varied widely between devices, with the Apple Watch being the most accurate and Yamax Digiwalker the least accurate for step count in free-living conditions. The iHealth Edge underestimates number of steps but can be considered reliable for activity monitoring after correction for bias. Misfit Shine overestimated number of steps and cannot be considered suitable for step count because of the low agreement. Future studies should focus on the added value of remotely monitoring activity patterns over time in chronic patients.

Download Full-text

Neural Network–Based Algorithm for Adjusting Activity Targets to Sustain Exercise Engagement Among People Using Activity Trackers: Retrospective Observation and Algorithm Development Study

JMIR mhealth and uhealth ◽

10.2196/18142 ◽

2020 ◽

Vol 8 (9) ◽

pp. e18142

Author(s):

Ramin Mohammadi ◽

Mursal Atif ◽

Amanda Jayne Centi ◽

Stephen Agboola ◽

Kamal Jethwani ◽

...

Keyword(s):

Neural Network ◽

Physical Activity ◽

Machine Learning ◽

Learning Model ◽

Step Count ◽

Daily Step ◽

Activity Tracker ◽

Activity Trackers ◽

Machine Learning Model ◽

Daily Step Count

Background It is well established that lack of physical activity is detrimental to the overall health of an individual. Modern-day activity trackers enable individuals to monitor their daily activities to meet and maintain targets. This is expected to promote activity encouraging behavior, but the benefits of activity trackers attenuate over time due to waning adherence. One of the key approaches to improving adherence to goals is to motivate individuals to improve on their historic performance metrics. Objective The aim of this work was to build a machine learning model to predict an achievable weekly activity target by considering (1) patterns in the user’s activity tracker data in the previous week and (2) behavior and environment characteristics. By setting realistic goals, ones that are neither too easy nor too difficult to achieve, activity tracker users can be encouraged to continue to meet these goals, and at the same time, to find utility in their activity tracker. Methods We built a neural network model that prescribes a weekly activity target for an individual that can be realistically achieved. The inputs to the model were user-specific personal, social, and environmental factors, daily step count from the previous 7 days, and an entropy measure that characterized the pattern of daily step count. Data for training and evaluating the machine learning model were collected over a duration of 9 weeks. Results Of 30 individuals who were enrolled, data from 20 participants were used. The model predicted target daily count with a mean absolute error of 1545 (95% CI 1383-1706) steps for an 8-week period. Conclusions Artificial intelligence applied to physical activity data combined with behavioral data can be used to set personalized goals in accordance with the individual’s level of activity and thereby improve adherence to a fitness tracker; this could be used to increase engagement with activity trackers. A follow-up prospective study is ongoing to determine the performance of the engagement algorithm.

Download Full-text

Validity, Reliability and Sensitivity to Change of Three Consumer-grade Activity Trackers in Controlled and Free-living Conditions among Older Adults (Preprint)

10.2196/preprints.27155 ◽

2021 ◽

Author(s):

Kaja Kastelic ◽

Marina Dobnik ◽

Stefan Loefler ◽

Christian Hofer ◽

Nejc Šarabon

Keyword(s):

Physical Activity ◽

Older Adults ◽

Intraclass Correlation ◽

Sleep Time ◽

Living Conditions ◽

Percentage Error ◽

Sensitivity To Change ◽

Free Living ◽

Activity Trackers ◽

Free Living Conditions

BACKGROUND Wrist worn consumer-grade activity trackers are popular devices, developed mainly for personal use, but with the potential to be used also for clinical and research purposes. OBJECTIVE The objective of this study was to explore the validity, reliability and sensitivity to change of movement behaviours metrics from three popular activity trackers (POLAR Vantage M, Garmin Vivosport and Garmin Vivoactive 4s) in controlled and free-living conditions when worn by older adults. METHODS Participants (n = 28; 74 ± 5 years) underwent a videotaped laboratory protocol while wearing all three activity trackers. On a separate occasion, participants wore one (randomly assigned) activity tracker and a research grade physical activity monitor ActiGraph wGT3X-BT simultaneously for six consecutive days for comparisons. RESULTS Both Garmin activity trackers showed excellent performance for step counts, with mean absolute percentage error (MAPE) below 20 % and intraclass correlation coefficient (ICC2,1) above 0.90 (P < .05), while Polar Vantage M substantially over counted steps (MAPE = 84 % and ICC2,1 = 0.37 for free-living conditions). MAPE for sleep time was within 10 % for all the trackers tested, while far beyond 20 % for all the physical activity and calories burned outputs. Both Garmin trackers showed fair agreement (ICC2,1 = 0.58–0.55) for measuring calories burned when compared with ActiGraph. CONCLUSIONS Garmin Vivoactive 4s showed overall best performance, especially for measuring steps and sleep time in healthy older adults. Minimal detectible change was consistently lower for an average day measures than for a single day measure, but still relatively high. The results provided in this study could be used to guide choice on activity trackers aiming for different purposes – individual use/care, longitudinal monitoring or in clinical trial setting.

Download Full-text

Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data

Sensors ◽

10.3390/s19204504 ◽

2019 ◽

Vol 19 (20) ◽

pp. 4504 ◽

Cited By ~ 1

Author(s):

Petra Jones ◽

Evgeny M. Mirkes ◽

Tom Yates ◽

Charlotte L. Edwardson ◽

Mike Catt ◽

...

Keyword(s):

Physical Activity ◽

Cross Validation ◽

Accelerometer Data ◽

Activity Type ◽

Free Living ◽

Independent Dataset ◽

Clustering Model ◽

Activity Types ◽

Unsupervised Approach ◽

Viable Approach

Few methods for classifying physical activity from accelerometer data have been tested using an independent dataset for cross-validation, and even fewer using multiple independent datasets. The aim of this study was to evaluate whether unsupervised machine learning was a viable approach for the development of a reusable clustering model that was generalisable to independent datasets. We used two labelled adult laboratory datasets to generate a k-means clustering model. To assess its generalised application, we applied the stored clustering model to three independent labelled datasets: two laboratory and one free-living. Based on the development labelled data, the ten clusters were collapsed into four activity categories: sedentary, standing/mixed/slow ambulatory, brisk ambulatory, and running. The percentages of each activity type contained in these categories were 89%, 83%, 78%, and 96%, respectively. In the laboratory independent datasets, the consistency of activity types within the clusters dropped, but remained above 70% for the sedentary clusters, and 85% for the running and ambulatory clusters. Acceleration features were similar within each cluster across samples. The clusters created reflected activity types known to be associated with health and were reasonably robust when applied to diverse independent datasets. This suggests that an unsupervised approach is potentially useful for analysing free-living accelerometer data.

Download Full-text

Validity of hip-mounted uniaxial accelerometry with heart-rate monitoring vs. triaxial accelerometry in the assessment of free-living energy expenditure in young children: the IDEFICS Validation Study

Journal of Applied Physiology ◽

10.1152/japplphysiol.01290.2011 ◽

2012 ◽

Vol 113 (10) ◽

pp. 1530-1536 ◽

Cited By ~ 20

Author(s):

Robert Ojiambo ◽

Kenn Konstabel ◽

Toomas Veidebaum ◽

John Reilly ◽

Vera Verbestel ◽

...

Keyword(s):

Physical Activity ◽

Heart Rate ◽

Energy Expenditure ◽

Young Children ◽

Validation Study ◽

Longitudinal Axis ◽

Activity Level ◽

Free Living ◽

Model Combining ◽

Sedentary Activities

One of the aims of Identification and Prevention of Dietary- and Lifestyle-Induced Health Effects in Children and Infants (IDEFICS) validation study is to validate field measures of physical activity (PA) and energy expenditure (EE) in young children. This study compared the validity of uniaxial accelerometry with heart-rate (HR) monitoring vs. triaxial accelerometry against doubly labeled water (DLW) criterion method for assessment of free-living EE in young children. Forty-nine European children (25 female, 24 male) aged 4–10 yr (mean age: 6.9 ± 1.5 yr) were assessed by uniaxial ActiTrainer with HR, uniaxial 3DNX, and triaxial 3DNX accelerometry. Total energy expenditure (TEE) was estimated using DLW over a 1-wk period. The longitudinal axis of both devices and triaxial 3DNX counts per minute (CPM) were significantly ( P < 0.05) associated with physical activity level (PAL; r = 0.51 ActiTrainer, r = 0.49 uniaxial-3DNX, and r = 0.42 triaxial Σ3DNX). Eight-six percent of the variance in TEE could be predicted by a model combining body mass (partial r2 = 71%; P < 0.05), CPM-ActiTrainer (partial r2 = 11%; P < 0.05), and difference between HR at moderate and sedentary activities (ModHR − SedHR) (partial r2 = 4%; P < 0.05). The SE of TEE estimate for ActiTrainer and 3DNX models ranged from 0.44 to 0.74 MJ/days or ∼7–11% of the average TEE. The SE of activity-induced energy expenditure (AEE) model estimates ranged from 0.38 to 0.57 MJ/day or 24–26% of the average AEE. It is concluded that the comparative validity of hip-mounted uniaxial and triaxial accelerometers for assessing PA and EE is similar.

Download Full-text

Convergent Validity of the Fitbit Charge 2 to Measure Sedentary Behavior and Physical Activity in Overweight and Obese Adults

Journal for the Measurement of Physical Behaviour ◽

10.1123/jmpb.2020-0014 ◽

2020 ◽

pp. 1-8

Author(s):

Joanne A. McVeigh ◽

Jennifer Ellis ◽

Caitlin Ross ◽

Kim Tang ◽

Phoebe Wan ◽

...

Keyword(s):

Physical Activity ◽

Confidence Interval ◽

Sedentary Behavior ◽

Convergent Validity ◽

Intraclass Correlation ◽

Correlation Coefficients ◽

Living Environment ◽

Step Count ◽

Activity Trackers ◽

Relative Differences

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.

Download Full-text

Use of Featforward Mobile Phone App Associated with Decreased Cardiometabolic Risk Factors in Patients with Chronic Conditions (Preprint)

10.2196/preprints.11882 ◽

2018 ◽

Author(s):

Sara B Golas ◽

Ramya Palacholla ◽

Amanda Centi ◽

Odeta Dyrmishi ◽

Stephen Agboola ◽

...

Keyword(s):

Physical Activity ◽

Mobile Phone ◽

Chronic Conditions ◽

Primary Outcome ◽

Fasting Blood Glucose ◽

Intervention Group ◽

Step Count ◽

Control Group ◽

Activity Trackers ◽

Secondary Outcomes

BACKGROUND Physical inactivity is one of the leading risk factors contributing to rising rates of chronic diseases and has been associated with deleterious health outcomes in patients with chronic disease conditions. FeatForward is a mobile phone app designed to encourage patients with cardiometabolic risk (CMR) factors to increase their levels of physical activity. OBJECTIVE To evaluate the effect of the FeatForward mobile phone app on physical activity levels (primary outcome) and global CMR factors (secondary outcomes) in patients with chronic conditions. METHODS In this 6-month, 2-arm randomized controlled trial, adult participants endorsing at least 1 study-eligible condition (obesity, [pre-]diabetes, [pre-]hypertension) were enrolled and assigned to either the intervention group (FeatForward app and standard care) or control group (standard care only). The primary and secondary outcomes were, respectively, change from baseline in physical activity (step count) and CMR factors (weight, body mass index [BMI], waist circumference, glycated hemoglobin [HbA1c], fasting blood glucose, systolic/diastolic blood pressures, serum lipids, C-reactive protein [CRP]). CMR data were collected at 3 time-points: baseline, 3 months, and 6 months. Step count data were recorded continuously by patients’ study-issued activity trackers and collected in batches at 3 and 6 months. At study end, patients’ weekly average step counts (WAS) were calculated as total steps taken divided by days of step data (0-7) for each of 26 study weeks. Mixed-effects linear regression models evaluated change over time between groups for the primary outcome and secondary outcomes. All models controlled for baseline values. The step count model additionally controlled for proportion of days without data, defined as (7 – days of data) / 7. Analyses were conducted for both groups overall, and by disease cohort (obesity, diabetes, hypertension). RESULTS Step count and CMR data were analyzed for 128 intervention and 133 control patients. There were no demographic differences between groups. While there was an overall downward trend in WAS for both groups, the intervention group decreased significantly less than the control group, with a slope of -29.3 steps per week compared to controls’ -57.9 (P=.02). Intervention patients with obesity slightly increased their step count overtime, differing significantly from controls (slope of 0.9 vs -90.2; P<.001). Intervention patients significantly lowered their BMI per study month compared to controls (slopes -0.23 vs -0.02; P=.04). Additionally, intervention patients with hypertension significantly decreased weight (P=.003), BMI (P=.002), and CRP (P=.03) per month compared to the control group. Waist circumference, HbA1c, fasting blood glucose, blood pressure, and lipids did not differ significantly by group or disease cohort over time. CONCLUSIONS While it is common for patient engagement with physical activity trackers to decrease over the course of a study, patients using the FeatFoward app had a slower decline in physical activity compared to controls. Intervention patients experienced a reduction in their BMI from a mean of 34.3 to 33.4, compared to controls’ 34.8 to 35.0. Patients with hypertension experienced significant decreases in BMI, weight, and CRP compared to controls. Future analyses will evaluate the impact of app engagement levels on step counts and CMR factors for the intervention group.

Download Full-text