Physical Activity Trend eXtraction: A Framework for Extracting Moderate-Vigorous Physical Activity Trends From Wearable Fitness Tracker Data

Change in adaptive and maladaptive exercise and objective physical activity throughout CBT for individuals with eating disorders

Maladaptive exercise (i.e., exercise that compensates for binge eating or is used to avoid negative consequences of not exercising-like weight gain) is a common eating disorder (ED) behavior, yet, some individuals with EDs only engage in adaptive exercise. CBT for EDs targets reducing maladaptive exercise but does not address adaptive exercise. Thus, research is limited on how adaptive and maladaptive exercise are impacted during CBT for EDs. The current study examined how assessor-rated adaptive and maladaptive exercise and objectively measured physical activity changed over a 12-week CBT treatment among adults with transdiagnostic binge eating and restrictive eating that did and did not engage in maladaptive exercise at the start of treatment (n = 13 non-maladaptive exercise group, n = 17 maladaptive exercise group). The overall amount of adaptive and maladaptive exercise was measured via the Eating Disorder Examination Interview and objectively measured physical activity (e.g., step count, minutes of moderate-to-vigorous physical activity [MVPA]) was measured via a wrist-worn fitness tracker. Throughout treatment, the overall amount of adaptive exercise did not significantly change for both groups, but the overall amount of maladaptive exercise significantly decreased in the maladaptive exercise group. Step count did not significantly change for both groups, but minutes of MVPA significantly increased over treatment for the non-maladaptive exercise group. Increases in step count and minutes of MVPA were not associated with ED symptom changes in either group. These results enhance an understanding of exercise changes during transdiagnostic CBT-based ED treatment for individuals with varying baseline exercise profiles.Level of evidence: Level 1, randomized controlled trial.

Data Imputation and Body Weight Variability Calculation Using Linear and Nonlinear Methods in Data Collected From Digital Smart Scales: Simulation and Validation Study

Background

Body weight variability (BWV) is common in the general population and may act as a risk factor for obesity or diseases. The correct identification of these patterns may have prognostic or predictive value in clinical and research settings. With advancements in technology allowing for the frequent collection of body weight data from electronic smart scales, new opportunities to analyze and identify patterns in body weight data are available.

Objective

This study aims to compare multiple methods of data imputation and BWV calculation using linear and nonlinear approaches

Methods

In total, 50 participants from an ongoing weight loss maintenance study (the NoHoW study) were selected to develop the procedure. We addressed the following aspects of data analysis: cleaning, imputation, detrending, and calculation of total and local BWV. To test imputation, missing data were simulated at random and using real patterns of missingness. A total of 10 imputation strategies were tested. Next, BWV was calculated using linear and nonlinear approaches, and the effects of missing data and data imputation on these estimates were investigated.

Results

Body weight imputation using structural modeling with Kalman smoothing or an exponentially weighted moving average provided the best agreement with observed values (root mean square error range 0.62%-0.64%). Imputation performance decreased with missingness and was similar between random and nonrandom simulations. Errors in BWV estimations from missing simulated data sets were low (2%-7% with 80% missing data or a mean of 67, SD 40.1 available body weights) compared with that of imputation strategies where errors were significantly greater, varying by imputation method.

Conclusions

The decision to impute body weight data depends on the purpose of the analysis. Directions for the best performing imputation methods are provided. For the purpose of estimating BWV, data imputation should not be conducted. Linear and nonlinear methods of estimating BWV provide reasonably accurate estimates under high proportions (80%) of missing data.

Activity Tracker-Based Metrics as Digital Markers of Cardiometabolic Health in Working Adults: Cross-Sectional Study

Background

Greater adoption of wearable devices with multiple sensors may enhance personalized health monitoring, facilitate early detection of some diseases, and further scale up population health screening. However, few studies have explored the utility of data from wearable fitness trackers in cardiovascular and metabolic disease risk prediction.

Objective

This study aimed to investigate the associations between a range of activity metrics derived from a wearable consumer-grade fitness tracker and major modifiable biomarkers of cardiometabolic disease in a working-age population.

Methods

This was a cross-sectional study of 83 working adults. Participants wore Fitbit Charge 2 for 21 consecutive days and went through a health assessment, including fasting blood tests. The following clinical biomarkers were collected: BMI, waist circumference, waist-to-hip ratio, blood pressure, triglycerides (TGs), high-density lipoprotein (HDL) and low-density lipoprotein cholesterol, and blood glucose. We used a range of wearable-derived metrics based on steps, heart rate (HR), and energy expenditure, including measures of stability of circadian activity rhythms, sedentary time, and time spent at various intensities of physical activity. Spearman rank correlation was used for preliminary analysis. Multiple linear regression adjusted for potential confounders was used to determine the extent to which each metric of activity was associated with continuous clinical biomarkers. In addition, pairwise multiple regression was used to investigate the significance and mutual dependence of activity metrics when two or more of them had significant association with the same outcome from the previous step of the analysis.

Results

The participants were predominantly middle aged (mean age 44.3 years, SD 12), Chinese (62/83, 75%), and male (64/83, 77%). Blood biomarkers of cardiometabolic disease (HDL cholesterol and TGs) were significantly associated with steps-based activity metrics independent of age, gender, ethnicity, education, and shift work, whereas body composition biomarkers (BMI, waist circumference, and waist-to-hip ratio) were significantly associated with energy expenditure–based and HR-based metrics when adjusted for the same confounders. Steps-based interdaily stability of circadian activity rhythm was strongly associated with HDL (beta=5.4 per 10% change; 95% CI 1.8 to 9.0; P=.005) and TG (beta=−27.7 per 10% change; 95% CI −48.4 to −7.0; P=.01). Average daily steps were negatively associated with TG (beta=−6.8 per 1000 steps; 95% CI −13.0 to −0.6; P=.04). The difference between average HR and resting HR was significantly associated with BMI (beta=−.5; 95% CI −1.0 to −0.1; P=.01) and waist circumference (beta=−1.3; 95% CI −2.4 to −0.2; P=.03).

Conclusions

Wearable consumer-grade fitness trackers can provide acceptably accurate and meaningful information, which might be used in the risk prediction of cardiometabolic disease. Our results showed the beneficial effects of stable daily patterns of locomotor activity for cardiometabolic health. Study findings should be further replicated with larger population studies.