The impact of an m-Health financial incentives program on the physical activity and diet of Australian truck drivers

Background

Chronic diseases are high in truck drivers and have been linked to work routines that promote inactivity and poor diets. This feasibility study examined the extent to which an m-Health financial incentives program facilitated physical activity and healthy dietary choices in Australian truck drivers.

Methods

Nineteen men (mean [SD] age = 47.5 [9.8] years; BMI = 31.2 [4.6] kg/m²) completed the 20-week program, and used an activity tracker and smartphone application (Jawbone UP™) to regulate small positive changes in occupational physical activity, and fruit, vegetable, saturated fat and processed/refined sugar food/beverage choices. Measures (baseline, end-program, 2-months follow-up; April–December 2014) were accelerometer-determined proportions of work time spent physically active, and a workday dietary questionnaire. Statistical (repeated measures ANOVA) and thematic (interviews) analyses assessed program impact.

Results

Non-significant increases in the mean proportions of work time spent physically active were found at end-program and follow-up (+1%; 7 mins/day). Fruit (p = 0.023) and vegetable (p = 0.024) consumption significantly increased by one serve/day at end-program. Non-significant improvements in saturated fat (5%) and processed/refined sugar (1%) food/beverage choices were found at end-program and follow-up. Overall, 65% (n = 11) of drivers demonstrated positive changes in physical activity, and at least one dietary choice (e.g. saturated fat) at follow-up. Drivers found the financial incentives component of the program to be a less effective facilitator of change than the activity tracker and smartphone application, although this technology was easier to use for monitoring of physical activity than healthy dietary choices.

Conclusions

Not all drivers benefitted from the program. However, positive changes for different health behaviours were observed in the majority of participants. Outcomes from this feasibility study inform future intervention development for studies with larger samples.

Trial registration

ANZCTR12616001513404. Registered November 2nd, 2016 (retrospectively registered).

Physical Activity Recognition Using Posterior-Adapted Class-Based Fusion of Multiaccelerometer Data

This paper proposes the use of posterior-adapted class-based weighted decision fusion to effectively combine multiple accelerometer data for improving physical activity recognition. The cutting-edge performance of this method is benchmarked against model-based weighted fusion and class-based weighted fusion without posterior adaptation, based on two publicly available datasets, namely PAMAP2 and MHEALTH. Experimental results show that: 1) posterior-adapted class-based weighted fusion outperformed model-based and class-based weighted fusion; 2) decision fusion with two accelerometers showed statistically significant improvement in average performance compared to the use of a single accelerometer; 3) generally, decision fusion from three accelerometers did not show further improvement from the best combination of two accelerometers; and 4) a combination of ankle and wrist located accelerometers showed the best overall performance compared to any combination of two or three accelerometers.

Patterns and correlates of physical activity in adolescents in Dhaka city, Bangladesh

OBJECTIVES

Despite the widely acknowledged public health importance of physical activity (PA), few studies have examined levels of PA in Bangladesh. The purpose of this study was to investigate the patterns and correlates of PA in adolescents in Bangladesh.

STUDY DESIGN

Cross-sectional survey.

METHODS

A total of 798 students, aged 13-17 years; 48% girls, from eight purposively selected secondary schools in Dhaka city, Bangladesh completed a self-administered questionnaire including the 3-Day PA Recall. Parents completed a separate questionnaire to provide household/family-level data. Multilevel generalized linear modelling was used to identify the correlates of PA for boys and girls.

RESULTS

Two-thirds (66%) of the adolescents met the recommendations of 60 min/day of moderate to vigorous PA (MVPA) daily, with more boys than girls (76% and 55%, respectively). The most common activities reported were walking for travel (42%), cricket (33%) and household chores (30%). Multivariable modelling showed that girls' PA was positively associated with mother's education level, walking to school, involvement in school sports and having home sports equipment. Boys' PA was positively associated with mother's employment, having home sports equipment, having a playground at school and walking to school.

CONCLUSIONS

One third of adolescents in Bangladesh were insufficiently active with girls less active than boys. Walking to school and access to sports facilities including playgrounds and home equipment may be important to promote activity among Bangladeshi adolescents, with special attention to the girls.

Decision Trees for Detection of Activity Intensity in Youth with Cerebral Palsy

PURPOSE

To develop and test decision tree (DT) models to classify physical activity (PA) intensity from accelerometer output and Gross Motor Function Classification System (GMFCS) classification level in ambulatory youth with cerebral palsy (CP) and compare the classification accuracy of the new DT models to that achieved by previously published cut points for youth with CP.

METHODS

Youth with CP (GMFCS levels I-III) (N = 51) completed seven activity trials with increasing PA intensity while wearing a portable metabolic system and ActiGraph GT3X accelerometers. DT models were used to identify vertical axis (VA) and vector magnitude (VM) count thresholds corresponding to sedentary (SED) (<1.5 METs), light-intensity PA (LPA) (≥1.5 and <3 METs) and moderate-to-vigorous PA (MVPA) (≥3 METs). Models were trained and cross-validated using the "rpart" and "caret" packages within R.

RESULTS

For the VA (VA_DT) and VM DT (VM_DT), a single threshold differentiated LPA from SED, whereas the threshold for differentiating MVPA from LPA decreased as the level of impairment increased. The average cross-validation accuracies for the VC_DT were 81.1%, 76.7%, and 82.9% for GMFCS levels I, II, and III. The corresponding cross-validation accuracies for the VM_DT were 80.5%, 75.6%, and 84.2%. Within each GMFCS level, the DT models achieved better PA intensity recognition than previously published cut points. The accuracy differential was greatest among GMFCS level III participants, in whom the previously published cut points misclassified 40% of the MVPA activity trials.

CONCLUSIONS

The GMFCS-specific cut points provide more accurate assessments of MVPA levels in youth with CP across the full spectrum of ambulatory ability.

Influence of the day care, home and neighbourhood environment on young children's physical activity and health: protocol for the PLAYCE observational study

INTRODUCTION

The early years are a critical period in a child's health and development, yet most preschool children fail to meet physical activity guidelines. Outside of the home and neighbourhood, children spend a large proportion of time within early childhood education and care (ECEC) services such as long day care. Research is required to determine how the design of day care outdoor (and indoor) spaces provides opportunities or constraints for physical activity. A significant evidence gap surrounds what objectively measured attributes of the home and neighbourhood environment influence preschoolers' physical activity. The PLAY Spaces & Environments for Children's Physical Activity (PLAYCE) study will empirically investigate the relative and cumulative influence of the day care, home and neighbourhood environment on preschoolers' physical activity.

METHODS AND ANALYSIS

The PLAYCE study is a cross-sectional observational study (April 2015 to April 2018) of 2400 children aged 2-5 years attending long day care in metropolitan Perth, Western Australia. Accelerometers will measure physical activity with indoor physical activity measured using radio frequency identification. Global positioning systems will be used to determine outdoor location of physical activity around the home and neighbourhood for a subsample (n=310). The day care environment will be objectively measured using a validated audit tool. Other potential individual, social and physical environmental influences on preschoolers' physical activity will be collected by geographic information systems measures, parent and day care educator surveys.

ETHICS AND DISSEMINATION

Ethical approval has been granted by The University of Western Australia Human Ethics Research Committee, approval number RA/4/1/7417. Findings will be published in international peer-reviewed journals and presented at international conferences. Key findings will be disseminated to stakeholders, collaborators, policymakers and practitioners working in the ECEC sector. Day care centre directors and parents will be given a summary report of the key findings.

State of the Art Reviews: Measurement of Physical Activity in Children and Adolescents

To date, a wide range of methods has been used to measure physical activity in children and adolescents. These include self-report methods such as questionnaires, activity logs, and diaries as well as objective measures of physical activity such as direct observation, doubly labeled water, heart rate monitoring, accelerometers, and pedometers. The purpose of this review is to overview the methods currently being used to measure physical activity in children and adolescents. For each measurement approach, new developments and/or innovations are identified and discussed. Particular attention is given to the use of accelerometers and the calibration of accelerometer output to units of energy expenditure to developing children.

Balanced: a randomised trial examining the efficacy of two self-monitoring methods for an app-based multi-behaviour intervention to improve physical activity, sitting and sleep in adults

BACKGROUND

Many adults are insufficiently physically active, have prolonged sedentary behaviour and report poor sleep. These behaviours can be improved by interventions that include education, goal setting, self-monitoring, and feedback strategies. Few interventions have explicitly targeted these behaviours simultaneously or examined the relative efficacy of different self-monitoring methods.

METHODS/DESIGN

This study aims to compare the efficacy of two self-monitoring methods in an app-based multi-behaviour intervention to improve objectively measured physical activity, sedentary, and sleep behaviours, in a 9 week 2-arm randomised trial. Participants will be adults (n = 64) who report being physically inactive, sitting >8 h/day and frequent insufficient sleep (≥14 days out of last 30). The "Balanced" intervention is delivered via a smartphone 'app', and includes education materials (guidelines, strategies to promote change in behaviour), goal setting, self-monitoring and feedback support. Participants will be randomly allocated to either a device-entered or user-entered self-monitoring method. The device-entered group will be provided with a activity tracker to self-monitor behaviours. The user-entered group will recall and manually record behaviours. Assessments will be conducted at 0, 3, 6, and 9 weeks. Physical activity, sedentary behaviour and sleep-wake behaviours will be measured using the wrist worn Geneactiv accelerometer. Linear mixed models will be used to examine differences between groups and over time using an alpha of 0.01.

DISCUSSION

This study will evaluate an app-based multi-behavioural intervention to improve physical activity, sedentary behaviour and sleep; and the relative efficacy of two different approaches to self-monitoring these behaviours. Outcomes will provide information to inform future interventions and self-monitoring targeting these behaviours.

TRIAL REGISTRATION

ACTRN12615000182594 (Australian New Zealand Clinical Trials Registry. Registry URL: www.anzctr.org.au ; registered prospectively on 25 February 2015).

Evaluation of a Physical Activity Intervention for Adults With Brain Impairment

Background. Individuals with brain impairment (BI) are less active than the general population and have increased risk of chronic disease. Objective. This controlled trial evaluated the efficacy of a physical activity (PA) intervention for community-dwelling adults with BI. Methods. A total of 43 adults with BI (27 male, 16 female; age 38.1 ± 11.9 years; stage of change 1-3) who walked as their primary means of locomotion were allocated to an intervention (n = 23) or control (n = 20) condition. The intervention comprised 10 face-to-face home visits over 12 weeks, including a tailored combination of stage-matched behavior change activities, exercise prescription, community access facilitation, and relapse prevention strategies. The control group received 10 face-to-face visits over 12 weeks to promote sun safety, healthy sleep, and oral health. Primary outcomes were daily activity counts and minutes of moderate-to-vigorous-intensity PA (MVPA) measured with the ActiGraph GT1M at baseline (0 weeks), postintervention (12 weeks) and follow-up (24 weeks). Between-group differences were evaluated for statistical significance using repeated-measures ANOVA. Results. MVPA for the intervention group increased significantly from baseline to 12 weeks (20.8 ± 3.1 to 31.2 ± 3.1 min/d; P = .01), but differences between baseline and 24 weeks were nonsignificant (20.8 ± 3.1 to 25.3 ± 3.2 min/d; P = .28). MVPA changes for the control group were negligible and nonsignificant. Between-group differences for change in MVPA were significant at 12 weeks ( P = .03) but not at 24 weeks ( P = .49). Conclusion. The 12-week intervention effectively increased adoption of PA in a sample of community-dwelling adults with BI immediately after the intervention but not at follow-up. Future studies should explore strategies to foster maintenance of PA participation.

Field evaluation of a random forest activity classifier for wrist-worn accelerometer data

OBJECTIVES

Wrist-worn accelerometers are convenient to wear and associated with greater wear-time compliance. Previous work has generally relied on choreographed activity trials to train and test classification models. However, validity in free-living contexts is starting to emerge. Study aims were: (1) train and test a random forest activity classifier for wrist accelerometer data; and (2) determine if models trained on laboratory data perform well under free-living conditions.

DESIGN

Twenty-one participants (mean age=27.6±6.2) completed seven lab-based activity trials and a 24h free-living trial (N=16).

METHODS

Participants wore a GENEActiv monitor on the non-dominant wrist. Classification models recognising four activity classes (sedentary, stationary+, walking, and running) were trained using time and frequency domain features extracted from 10-s non-overlapping windows. Model performance was evaluated using leave-one-out-cross-validation. Models were implemented using the randomForest package within R. Classifier accuracy during the 24h free living trial was evaluated by calculating agreement with concurrently worn activPAL monitors.

RESULTS

Overall classification accuracy for the random forest algorithm was 92.7%. Recognition accuracy for sedentary, stationary+, walking, and running was 80.1%, 95.7%, 91.7%, and 93.7%, respectively for the laboratory protocol. Agreement with the activPAL data (stepping vs. non-stepping) during the 24h free-living trial was excellent and, on average, exceeded 90%. The ICC for stepping time was 0.92 (95% CI=0.75-0.97). However, sensitivity and positive predictive values were modest. Mean bias was 10.3min/d (95% LOA=-46.0 to 25.4min/d).

CONCLUSIONS

The random forest classifier for wrist accelerometer data yielded accurate group-level predictions under controlled conditions, but was less accurate at identifying stepping verse non-stepping behaviour in free living conditions Future studies should conduct more rigorous field-based evaluations using observation as a criterion measure.

Developmental Trends in the Energy Cost of Physical Activities Performed by Youth

BACKGROUND

Published energy cost data for children and adolescents are lacking. The purpose of this study was to measure and describe developmental trends in the energy cost of 12 physical activities commonly performed by youth.

METHODS

A mixed age cohort of 209 participants completed 12 standardized activity trials on 4 occasions over a 3-year period (baseline, 12-months, 24-months, and 36-months) while wearing a portable indirect calorimeter. Bayesian hierarchical regression was used to link growth curves from each age cohort into a single curve describing developmental trends in energy cost from age 6 to 18 years.

RESULTS

For sedentary and light-intensity household chores, YOUTH METs (METy) remained stable or declined with age. In contrast, METy values associated with brisk walking, running, basketball, and dance increased with age.

CONCLUSIONS

The reported energy costs for specific activities will contribute to efforts to update and expand the youth compendium.

Chronic disease risks and use of a smartphone application during a physical activity and dietary intervention in Australian truck drivers

OBJECTIVE

This study examined chronic disease risks and the use of a smartphone activity tracking application during an intervention in Australian truck drivers (April-October 2014).

METHODS

Forty-four men (mean age=47.5 [SD 9.8] years) completed baseline health measures, and were subsequently offered access to a free wrist-worn activity tracker and smartphone application (Jawbone UP) to monitor step counts and dietary choices during a 20-week intervention. Chronic disease risks were evaluated against guidelines; weekly step count and dietary logs registered by drivers in the application were analysed to evaluate use of the Jawbone UP.

RESULTS

Chronic disease risks were high (e.g. 97% high waist circumference [≥ 94 cm]). Eighteen drivers (41%) did not start the intervention; smartphone technical barriers were the main reason for drop out. Across 20-weeks, drivers who used the Jawbone UP logged step counts for an average of 6 [SD 1] days/week; mean step counts remained consistent across the intervention (weeks 1-4=8,743[SD 2,867] steps/day; weeks 17-20=8,994[SD 3,478] steps/day). The median number of dietary logs significantly decreased from start (17 [IQR 38] logs/weeks) to end of the intervention (0 [IQR 23] logs/week; p<0.01); the median proportion of healthy diet choices relative to total diet choices logged increased across the intervention (weeks 1-4=38[IQR 21]%; weeks 17-20=58[IQR 18]%).

CONCLUSIONS

Step counts were more successfully monitored than dietary choices in those drivers who used the Jawbone UP.

IMPLICATIONS

Smartphone technology facilitated active living and healthy dietary choices, but also prohibited intervention engagement in a number of these high-risk Australian truck drivers.

Maternal and paternal support for physical activity and healthy eating in preschool children: a cross-sectional study

Background

Parental support is a key influence on children’s health behaviours; however, no previous investigation has simultaneously explored the influence of mothers’ and fathers’ social support on eating and physical activity in preschool-aged children. This study evaluated the singular and combined effects of maternal and paternal support for physical activity (PA) and fruit and vegetable consumption (FV) on preschoolers’ PA and FV.

Methods

A random sample comprising 173 parent–child dyads completed validated scales assessing maternal and paternal instrumental support and child PA and FV behaviour. Pearson correlations, controlling for child age, parental age, and parental education, were used to evaluate relationships between maternal and paternal support and child PA and FV. K-means cluster analysis was used to identify families with distinct patterns of maternal and paternal support for PA and FV, and one-way ANOVA examined the impact of cluster membership on child PA and FV.

Results

Maternal and paternal support for PA were positively associated with child PA (r = 0.37 and r = 0.36, respectively; P < 0.001). Maternal but not paternal support for FV was positively associated with child FV (r = 0.35; P < 0.001). Five clusters characterised groups of families with distinct configurations of maternal and paternal support for PA and FV: 1) above average maternal and paternal support for PA and FV, 2) below average maternal and paternal support for PA and FV, 3) above average maternal and paternal support for PA but below average maternal and paternal support for FV, 4) above average maternal and paternal support for FV but below average maternal and paternal support for PA, and 5) above average maternal support but below average paternal support for PA and FV. Children from families with above average maternal and paternal support for both health behaviours had higher PA and FV levels than children from families with above average support for just one health behaviour, or below average support for both behaviours.

Conclusions

The level and consistency of instrumental support from mothers and fathers for PA and FV may be an important target for obesity prevention in preschool-aged children.

Validity of the OMNI rating of perceived exertion scale for children and adolescents with cerebral palsy

AIM

This study evaluated the validity of the OMNI Walk/Run Rating of Perceived Exertion (OMNI-RPE) scores with heart rate and oxygen consumption (VO₂) for children and adolescents with cerebral palsy (CP).

METHOD

Children and adolescents with CP, aged 6 to 18 years and Gross Motor Function Classification System (GMFCS) levels I to III completed a physical activity protocol with seven trials ranging in intensity from sedentary to moderate-to-vigorous. VO₂ and heart rate were recorded during the physical activity trials using a portable indirect calorimeter and heart rate monitor. Participants reported OMNI-RPE scores for each trial. Concurrent validity was assessed by calculating the average within-subject correlation between OMNI-RPE ratings and the two physiological indices.

RESULTS

For the correlational analyses, 48 participants (22 males, 26 females; age 12y 6mo, SD 3y 4mo) had valid bivariate data for VO₂ and OMNI-RPE, while 40 participants (21 males, 19 females; age 12y 5mo, SD 2y 9mo) had valid bivariate data for heart rate and OMNI-RPE. VO₂ (r=0.80; 95% CI 0.66-0.88) and heart rate (r=0.83; 95% CI 0.70-0.91) were moderately to highly correlated to OMNI-RPE scores. No difference was found for the correlation of physiological data and OMNI-RPE scores across the three GMFCS levels. The OMNI-RPE scores increased significantly in a dose-response manner (F(6,258) =116.1, p<0.001) as exercise intensity increased from sedentary to moderate-to-vigorous.

INTERPRETATION

OMNI-RPE is a clinically feasible option to monitor exercise intensity in ambulatory children and adolescents with CP.

Physical Activity, Obesity Status, and Blood Pressure in Preschool Children

OBJECTIVE

To examine the combined effects of physical activity and weight status on blood pressure (BP) in preschool-aged children.

STUDY DESIGN

The sample included 733 preschool-aged children (49% female). Physical activity was objectively assessed on 7 consecutive days by accelerometry. Children were categorized as sufficiently active if they met the recommendation of at least 60 minutes daily of moderate-to-vigorous physical activity (MVPA). Body mass index was used to categorize children as nonoverweight or overweight/obese, according to the International Obesity Task Force benchmarks. BP was measured using an automated BP monitor and categorized as elevated or normal using BP percentile-based cut-points for age, sex, and height.

RESULTS

The prevalence of elevated systolic BP (SBP) and diastolic BP was 7.7% and 3.0%, respectively. The prevalence of overweight/obese was 32%, and about 15% of children did not accomplish the recommended 60 minutes of daily MVPA. After controlling for age and sex, overweight/obese children who did not meet the daily MVPA recommendation were 3 times more likely (OR 3.8; CI 1.6-8.6) to have elevated SBP than nonoverweight children who met the daily MVPA recommendation.

CONCLUSIONS

Overweight or obese preschool-aged children with insufficient levels of MVPA are at significantly greater risk for elevated SBP than their nonoverweight and sufficiently active counterparts.

Exploring Metrics to Express Energy Expenditure of Physical Activity in Youth

Background

Several approaches have been used to express energy expenditure in youth, but no consensus exists as to which best normalizes data for the wide range of ages and body sizes across a range of physical activities. This study examined several common metrics for expressing energy expenditure to determine whether one metric can be used for all healthy children. Such a metric could improve our ability to further advance the Compendium of Physical Activities for Youth.

Methods

A secondary analysis of oxygen uptake (VO₂) data obtained from five sites was completed, that included 947 children ages 5 to 18 years, who engaged in 14 different activities. Resting metabolic rate (RMR) was computed based on Schofield Equations [Hum Nutr Clin Nut. 39(Suppl 1), 1985]. Absolute oxygen uptake (ml.min^-1), oxygen uptake per kilogram body mass (VO₂ in ml.kg^-1.min^-1), net oxygen uptake (VO₂ – resting metabolic rate), allometric scaled oxygen uptake (VO₂ in ml.kg^-0.75.min^-1) and YOUTH-MET (VO₂.[resting VO₂] ^-1) were calculated. These metrics were regressed with age, sex, height, and body mass.

Results

Net and allometric-scaled VO₂, and YOUTH-MET were least associated with age, sex and physical characteristics. For moderate-to-vigorous intensity activities, allometric scaling was least related to age and sex. For sedentary and low-intensity activities, YOUTH-MET was least related to age and sex.

Conclusions

No energy expenditure metric completely eliminated the influence of age, physical characteristics, and sex. The Adult MET consistently overestimated EE. YOUTH-MET was better for expressing energy expenditure for sedentary and light activities, whereas allometric scaling was better for moderate and vigorous intensity activities. From a practical perspective, The YOUTH-MET may be the more feasible metric for improving of the Compendium of Physical Activities for Youth.

Psychometric properties of the modified RESIDE physical activity questionnaire among low-income overweight women

OBJECTIVES

This study explored the criterion-related validity and test-retest reliability of the modified RESIDential Environment physical activity questionnaire and whether the instrument's validity varied by body mass index, education, race/ethnicity, or employment status.

DESIGN

Validation study using baseline data collected for randomized trial of a weight loss intervention.

METHODS

Participants recruited from health departments wore an ActiGraph accelerometer and self-reported non-occupational walking, moderate and vigorous physical activity on the modified RESIDential Environment questionnaire. We assessed validity (n=152) using Spearman correlation coefficients, and reliability (n=57) using intraclass correlation coefficients.

RESULTS

When compared to steps, moderate physical activity, and bouts of moderate/vigorous physical activity measured by accelerometer, these questionnaire measures showed fair evidence for validity: recreational walking (Spearman correlation coefficients 0.23-0.36), total walking (Spearman correlation coefficients 0.24-0.37), and total moderate physical activity (Spearman correlation coefficients 0.18-0.36). Correlations for self-reported walking and moderate physical activity were higher among unemployed participants and women with lower body mass indices. Generally no other variability in the validity of the instrument was found. Evidence for reliability of RESIDential Environment measures of recreational walking, total walking, and total moderate physical activity was substantial (intraclass correlation coefficients 0.56-0.68).

CONCLUSIONS

Evidence for questionnaire validity and reliability varied by activity domain and was strongest for walking measures. The questionnaire may capture physical activity less accurately among women with higher body mass indices and employed participants. Capturing occupational activity, specifically walking at work, may improve questionnaire validity.

Step based physical activity guidelines for preschool-aged children

OBJECTIVE

Public health organizations recommend that preschool-aged children accumulate at least 3h of physical activity (PA) daily. Objective monitoring using pedometers offers an opportunity to measure preschooler's PA and assess compliance with this recommendation. The purpose of this study was to derive step-based recommendations consistent with the 3h PA recommendation for preschool-aged children.

METHOD

The study sample comprised 916 preschool-aged children, aged 3 to 6years (mean age=5.0±0.8years). Children were recruited from kindergartens located in Portugal, between 2009 and 2013. Children wore an ActiGraph GT1M accelerometer that measured PA intensity and steps per day simultaneously over a 7-day monitoring period. Receiver operating characteristic (ROC) curve analysis was used to identify the daily step count threshold associated with meeting the daily 3hour PA recommendation.

RESULTS

A significant correlation was observed between minutes of total PA and steps per day (r=0.76, p<0.001). The optimal step count for ≥3h of total PA was 9099 steps per day (sensitivity (90%) and specificity (66%)) with area under the ROC curve=0.86 (95% CI: 0.84 to 0.88).

CONCLUSION

Preschool-aged children who accumulate less than 9000 steps per day may be considered Insufficiently Active.

Machine learning for activity recognition: hip versus wrist data

PROBLEM ADDRESSED

Wrist-worn accelerometers are associated with greater compliance. However, validated algorithms for predicting activity type from wrist-worn accelerometer data are lacking. This study compared the activity recognition rates of an activity classifier trained on acceleration signal collected on the wrist and hip.

METHODOLOGY

52 children and adolescents (mean age 13.7  ±  3.1 year) completed 12 activity trials that were categorized into 7 activity classes: lying down, sitting, standing, walking, running, basketball, and dancing. During each trial, participants wore an ActiGraph GT3X+ tri-axial accelerometer on the right hip and the non-dominant wrist. Features were extracted from 10-s windows and inputted into a regularized logistic regression model using R (Glmnet + L1).

RESULTS

Classification accuracy for the hip and wrist was 91.0% ± 3.1% and 88.4% ± 3.0%, respectively. The hip model exhibited excellent classification accuracy for sitting (91.3%), standing (95.8%), walking (95.8%), and running (96.8%); acceptable classification accuracy for lying down (88.3%) and basketball (81.9%); and modest accuracy for dance (64.1%). The wrist model exhibited excellent classification accuracy for sitting (93.0%), standing (91.7%), and walking (95.8%); acceptable classification accuracy for basketball (86.0%); and modest accuracy for running (78.8%), lying down (74.6%) and dance (69.4%).Potential Impact: Both the hip and wrist algorithms achieved acceptable classification accuracy, allowing researchers to use either placement for activity recognition.

Prediction of activity type in preschool children using machine learning techniques

OBJECTIVES

Recent research has shown that machine learning techniques can accurately predict activity classes from accelerometer data in adolescents and adults. The purpose of this study is to develop and test machine learning models for predicting activity type in preschool-aged children.

DESIGN

Participants completed 12 standardised activity trials (TV, reading, tablet game, quiet play, art, treasure hunt, cleaning up, active game, obstacle course, bicycle riding) over two laboratory visits.

METHODS

Eleven children aged 3-6 years (mean age=4.8±0.87; 55% girls) completed the activity trials while wearing an ActiGraph GT3X+ accelerometer on the right hip. Activities were categorised into five activity classes: sedentary activities, light activities, moderate to vigorous activities, walking, and running. A standard feed-forward Artificial Neural Network and a Deep Learning Ensemble Network were trained on features in the accelerometer data used in previous investigations (10th, 25th, 50th, 75th and 90th percentiles and the lag-one autocorrelation).

RESULTS

Overall recognition accuracy for the standard feed forward Artificial Neural Network was 69.7%. Recognition accuracy for sedentary activities, light activities and games, moderate-to-vigorous activities, walking, and running was 82%, 79%, 64%, 36% and 46%, respectively. In comparison, overall recognition accuracy for the Deep Learning Ensemble Network was 82.6%. For sedentary activities, light activities and games, moderate-to-vigorous activities, walking, and running recognition accuracy was 84%, 91%, 79%, 73% and 73%, respectively.

CONCLUSIONS

Ensemble machine learning approaches such as Deep Learning Ensemble Network can accurately predict activity type from accelerometer data in preschool children.