Identification of Children's Activity Type with Accelerometer-Based Neural Networks

Measurement Properties of Wearable Kinematic-Based Data Collection Systems to Evaluate Ball Kicking in Soccer: A Systematic Review with Evidence Gap Map

Kinematic assessment of ball kicking may require significant human effort (e.g., traditional vision-based tracking systems). Wearables offer a potential solution to reduce processing time. This systematic review collated measurement properties (validity, reliability, and/or accuracy) of wearable kinematic-based technology systems used to evaluate soccer kicking. Seven databases were searched for studies published on or before April 2024. The protocol was previously published and followed the PRISMA 2020 statement. The data items included any validity, reliability, and/or accuracy measurements extracted from the selected articles. Twelve articles (1011 participants) were included in the qualitative synthesis, showing generally (92%) moderate methodological quality. The authors claimed validity (e.g., concurrent) in seven of the eight studies found on the topic, reliability in two of three, and accuracy (event detection) in three of three studies. The synthesis method indicated moderate evidence for the concurrent validity of the MPU-9150/ICM-20649 InvenSense and PlayerMaker™ devices. However, limited to no evidence was identified across studies when considering wearable devices/systems, measurement properties, and specific outcome variables. To conclude, there is a knowledge base that may support the implementation of wearables to assess ball kicking in soccer practice, while future research should further evaluate the measurement properties to attempt to reach a strong evidence level.

Accelerometer techniques for capturing human movement validated against direct observation: a scoping review

Objective.Accelerometers are devices commonly used to measure human physical activity and sedentary time. Accelerometer capabilities and analytical techniques have evolved rapidly, making it difficult for researchers to keep track of advances and best practices for data processing and analysis. The objective of this scoping review is to determine the existing methods for analyzing accelerometer data for capturing human movement which have been validated against the criterion measure of direct observation.Approach.This scoping review searched 14 academic and 5 grey databases. Two independent raters screened by title and abstract, then full text. Data were extracted using Microsoft Excel and checked by an independent reviewer.Mainresults.The search yielded 1039 papers and the final analysis included 115 papers. A total of 71 unique accelerometer models were used across a total of 4217 participants. While all studies underwent validation from direct observation, most direct observation occurred live (55%) or using recordings (42%). Analysis techniques included machine learning (ML) approaches (22%), the use of existing cut-points (18%), receiver operating characteristic curves to determine cut-points (14%), and other strategies including regressions and non-ML algorithms (8%).Significance.ML techniques are becoming more prevalent and are often used for activity identification. Cut-point methods are still frequently used. Activity intensity is the most assessed activity outcome; however, both the analyses and outcomes assessed vary by wear location. This scoping review provides a comprehensive overview of accelerometer analysis and validation techniques using direct observation and is a useful tool for researchers using accelerometers.

An End-to-End Deep Learning Pipeline for Football Activity Recognition Based on Wearable Acceleration Sensors

Action statistics in sports, such as the number of sprints and jumps, along with the details of the corresponding locomotor actions, are of high interest to coaches and players, as well as medical staff. Current video-based systems have the disadvantage that they are costly and not easily transportable to new locations. In this study, we investigated the possibility to extract these statistics from acceleration sensor data generated by a previously developed sensor garment. We used deep learning-based models to recognize five football-related activities (jogging, sprinting, passing, shooting and jumping) in an accurate, robust, and fast manner. A combination of convolutional (CNN) layers followed by recurrent (bidirectional) LSTM layers achieved up to 98.3% of accuracy. Our results showed that deep learning models performed better in evaluation time and prediction accuracy than traditional machine learning algorithms. In addition to an increase in accuracy, the proposed deep learning architecture showed to be 2.7 to 3.4 times faster in evaluation time than traditional machine learning methods. This demonstrated that deep learning models are accurate as well as time-efficient and are thus highly suitable for cost-effective, fast, and accurate human activity recognition tasks.

Transfer Learning Improves Accelerometer-Based Child Activity Recognition via Subject-Independent Adult-Domain Adaption

Wearable activity recognition can collate the type, intensity, and duration of each childs physical activity profile, which is important for exploring underlying adolescent health mechanisms. Traditional machine-learning-based approaches require large labeled data sets; however, child activity data sets are typically small and insufficient. Thus, we proposed a transfer learning approach that adapts adult-domain data to train a high-fidelity, subject-independent model for child activity recognition. Twenty children and twenty adults wore an accelerometer wristband while performing walking, running, sitting, and rope skipping activities. Activity classification accuracy was determined via the traditional machine learning approach without transfer learning and with the proposed subject-independent transfer learning approach. Results showed that transfer learning increased classification accuracy to 91.4% as compared to 80.6% without transfer learning. These results suggest that subject-independent transfer learning can improve accuracy and potentially reduce the size of the required child data sets to enable physical activity monitoring systems to be adopted more widely, quickly, and economically for children and provide deeper insights into injury prevention and health promotion strategies.

High-intensity activity is more strongly associated with metabolic health in children compared to sedentary time: a cross-sectional study of the I.Family cohort

Background

Physical activity (PA) during childhood is important for preventing future metabolic syndrome (MetS). To examine the relationship between PA and MetS in more detail, accurate measures of PA are needed. Previous studies have only utilized a small part of the information available from accelerometer measured PA. This study investigated the association between measured PA and MetS in children with a new method for data processing and analyses that enable more detailed interpretation of PA intensity level.

Methods

The association between PA pattern and risk factors related to MetS was investigated in a cross- sectional sample of children (n = 2592, mean age 10.9 years, 49.4% male) participating in the European multicenter I. Family study. The risk factors examined include body mass index, blood pressure, high-density lipoprotein cholesterol, insulin resistance and a combined risk factor score (MetS score). PA was measured by triaxial accelerometers and raw data was processed using the 10 Hz frequency extended method (FEM). The PA output was divided into an intensity spectrum and the association with MetS risk factors was analyzed by partial least squares regression.

Results

PA patterns differed between the European countries investigated, with Swedish children being most active and Italian children least active. Moderate intensity physical activity was associated with lower insulin resistance (R² = 2.8%), while vigorous intensity physical activity was associated with lower body mass index (R² = 3.6%), MetS score (R² = 3.1%) and higher high-density lipoprotein cholesterol (R² = 2.3%). PA of all intensities was associated with lower systolic- and diastolic blood pressure, although the associations were weaker than for the other risk factors (R² = 1.5% and R² = 1.4%). However, the multivariate analysis implies that the entire PA pattern must be considered. The main difference in PA was observed between normal weight and overweight children.

Conclusions

The present study suggests a greater importance of more PA corresponding to an intensity of at least brisk walking with inclusion of high-intense exercise, rather than a limited time spent sedentary, in the association to metabolic health in children. The methods of data processing and statistical analysis enabled accurate analysis and interpretation of the health benefits of high intensity PA that have not been shown previously.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12966-021-01156-1.

Using accelerometry to classify physical activity intensity in older adults: What is the optimal wear-site?

Abstract This study aimed to determine the optimal accelerometer wear-site specific cut-points for discrimination of the sedentary time, light physical activity and moderate-to-vigorous physical activity (MVPA) in older adults. Twenty-three adults (14 females) aged 55-77 years wore a GENEActiv accelerometer on their non-dominant wrist, dominant wrist, waist and dominant ankle whilst undertaking eight, five-minute bouts of activity: lay supine, seated reading, slow walking, medium walking, fast walking, folding laundry, sweeping and stationary cycling. VO₂ was assessed concurrently using indirect calorimetry. Receiver-operating-characteristic (ROC) analyses were used to derive wear-site specific cut-points for classifying intensity. Indirect calorimetry indicated that being lay supine and seated reading were classified as sedentary (<1.5 METs), laundry as light (1.51-2.99 METs) and sweeping, slow, medium and fast walking and cycling all classified as moderate intensity (>3 METs). Areas under ROC curves indicated that the classification of sedentary activity was good for the non-dominant wrist and excellent for all other wear sites. Classification of MVPA was excellent for the waist and ankle, good for the waist and poor for the dominant and non-dominant wrists. Overall, the ankle location performed better than in other locations. Ankle-worn accelerometry appears to provide the most suitable wear-site to discriminate between sedentary time and MVPA in older adults.

Activity Mapping of Children in Play Using Multivariate Analysis of Movement Events

Supplemental digital content is available in the text.

Purpose

(i) To develop an automated measurement technique for the assessment of both the form and intensity of physical activity undertaken by children during play. (ii) To profile the varying activity across a cohort of children using a multivariate analysis of their movement patterns.

Methods

Ankle-worn accelerometers were used to record 40 min of activity during a school recess, for 24 children over five consecutive days. Activity events of 1.1 s duration were identified within the acceleration time trace and compared with a reference motif, consisting of a single walking stride acceleration trace, obtained on a treadmill operating at a speed of 4 km h⁻¹. Dynamic time warping of motif and activity events provided metrics of comparative movement duration and intensity, which formed the data set for multivariate mapping of the cohort activity using a principal component analysis (PCA).

Results

The two-dimensional PCA plot provided clear differentiation of children displaying diverse activity profiles and clustering of those with similar movement patterns. The first component of the PCA correlated to the integrated intensity of movement over the 40-min period, whereas the second component informed on the temporal phasing of activity.

Conclusions

By defining movement events and then quantifying them by reference to a motion-standard, meaningful assessment of highly varied activity within free play can be obtained. This allows detailed profiling of individual children’s activity and provides an insight on social aspects of play through identification of matched activity time profiles for children participating in conjoined play.

Estimating Physical Activity in Children Aged 8-11 Years Using Accelerometry: Contributions From Fundamental Movement Skills and Different Accelerometer Placements

Accelerometers are widely used to assess physical activity, but it is unclear how effective accelerometers are in capturing fundamental movement skills in children. This study examined the energy expenditure during different physical activities (PA) and calibrated triaxial accelerometry, worn at the wrist, waist and ankle, during children's PA with attention to object control movement skills and cycling. Thirty children (14 girls) aged 8 to 11 years wore a GENEActiv accelerometer on their non-dominant wrist, dominant wrist, waist and ankle. Children undertook eight, 5-min bouts of activity comprising being lay supine, playing with Lego, slow walking, medium walking, medium paced running, overarm throwing and catching, instep passing a football and cycling at 35 W. VO₂ was assessed concurrently using indirect calorimetry. Indirect calorimetry indicated that being lay supine and playing with Lego were classified as sedentary in nature (<1.5 METs), slow paced walking, medium placed walking and throwing and catching were classified as light (1.51-2.99 METs) and running, cycling and instep passing were classified as moderate intensity (>3 METs). ROC curve analysis indicated that discrimination of sedentary activity was excellent for all placements although the ankle performed better than other locations. This pattern was replicated for moderate physical activity (MPA) where the ankle performed better than other locations. Data were reanalyzed removing cycling from the data set. When this analysis was undertaken discrimination of sedentary activity remained excellent for all locations. For MPA discrimination of activity was considered good for waist and ankle placement and fair for placement on either wrist. The current study is the first to quantify energy expenditure in object control fundamental movement skills via indirect calorimetry in children aged 8-11 years whilst also calibrating GENEActiv accelerometers worn at four body locations. Results suggest throwing and catching is categorized as light intensity and instep kicking a football moderate intensity, resulting in energy expenditure equivalent to slow or medium paced walking or cycling and running, respectively. Ankle worn accelerometry appears to provide the most suitable wear location to quantify MPA including ambulatory activity, object control skills and cycling, in children aged 8-11 years.

Calibration and validation of accelerometer-based activity monitors: A systematic review of machine-learning approaches

BACKGROUND

Objective measures using accelerometer-based activity monitors have been extensively used in physical activity (PA) and sedentary behavior (SB) research. To measure PA and SB precisely, the field is shifting towards machine learning-based (ML) approaches for calibration and validation of accelerometer-based activity monitors. Nevertheless, various parameters regarding the use and development of ML-based models, including data type (raw acceleration data versus activity counts), sampling frequency, window size, input features, ML technique, accelerometer placement, and free-living settings, affect the predictive ability of ML-based models. The effects of these parameters on ML-based models have remained elusive, and will be systematically reviewed here. The open challenges were identified and recommendations are made for future studies and directions.

METHOD

We conducted a systematic search of PubMed and Scopus databases to identify studies published before July 2017 that used ML-based techniques for calibration and validation of accelerometer-based activity monitors. Additional articles were manually identified from references in the identified articles.

RESULTS

A total of 62 studies were eligible to be included in the review, comprising 48 studies that calibrated and validated ML-based models for predicting the type and intensity of activities, and 22 studies for predicting activity energy expenditure.

CONCLUSIONS

It appears that various ML-based techniques together with raw acceleration data sampled at 20-30 Hz provide the opportunity of predicting the type and intensity of activities, as well as activity energy expenditure with comparable overall predictive accuracies regardless of accelerometer placement. However, the high predictive accuracy of laboratory-calibrated models is not reproducible in free-living settings, due to transitive and unseen activities together with differences in acceleration signals.

Identifying Free-Living Physical Activities Using Lab-Based Models with Wearable Accelerometers

The purpose of this study was to classify, and model various physical activities performed by a diverse group of participants in a supervised lab-based protocol and utilize the model to identify physical activity in a free-living setting. Wrist-worn accelerometer data were collected from (N=152) adult participants; age 18–64 years, and processed the data to identify and model unique physical activities performed by the participants in controlled settings. The Gaussian mixture model (GMM) and the hidden Markov model (HMM) algorithms were used to model the physical activities with time and frequency-based accelerometer features. An overall model accuracy of 92.7% and 94.7% were achieved to classify 24 physical activities using GMM and HMM, respectively. The most accurate model was then used to identify physical activities performed by 20 participants, each recorded for two free-living sessions of approximately six hours each. The free-living activity intensities were estimated with 80% accuracy and showed the dominance of stationary and light intensity activities in 36 out of 40 recorded sessions. This work proposes a novel activity recognition process to identify unsupervised free-living activities using lab-based classification models. In summary, this study contributes to the use of wearable sensors to identify physical activities and estimate energy expenditure in free-living settings.

Physical activity characterization: does one site fit all?

BACKGROUND

It is evident that a growing number of studies advocate a wrist-worn accelerometer for the assessment of patterns of physical activity a priori, yet the veracity of this site rather than any other body-mounted location for its accuracy in classifying activity is hitherto unexplored.

OBJECTIVE

The objective of this review was to identify the relative accuracy with which physical activities can be classified according to accelerometer site and analytical technique.

METHODS

A search of electronic databases was conducted using Web of Science, PubMed and Google Scholar. This review included studies written in the English language, published between database inception and December 2017, which characterized physical activities using a single accelerometer and reported the accuracy of the technique.

RESULTS

A total of 118 articles were initially retrieved. After duplicates were removed and the remaining articles screened, 32 full-text articles were reviewed, resulting in the inclusion of 19 articles that met the eligibility criteria.

CONCLUSION

There is no 'one site fits all' approach to the selection of accelerometer site location or analytical technique. Research design and focus should always inform the most suitable location of attachment, and should be driven by the type of activity being characterized.

Relationship between Physical Activity, Screen Time and Weight Status among Young Adolescents

It is well established that lack of physical activity and high bouts of sedentary behaviour are now associated with all-cause and cardiovascular mortality. The purpose of this study was to investigate the relationship between physical activity participation, overall screen time and weight status amongst early Irish adolescent youth. Participants were a sample of 169 students: 113 boys (mean age = 12.89 ± 0.34 years) and 56 girls (mean age = 12.87 ± 0.61 years). The data gathered in the present study included physical activity (accelerometry), screen time (self-report) and anthropometric measurements. Overweight and obese participants accumulated significantly more minutes of overall screen time daily compared to their normal-weight counterparts. A correlation between physical activity and daily television viewing was evident among girls. No significant interaction was apparent when examining daily physical activity and overall screen time in the prediction of early adolescents’ body mass index. Results suggest the importance of reducing screen time in the contribution towards a healthier weight status among adolescents. Furthermore, physical activity appears largely unrelated to overall screen time in predicting adolescent weight status, suggesting that these variables may be independent markers of health in youth. The existing relationship for girls between moderate-to-vigorous physical activity and time spent television viewing may be a potential area to consider for future intervention design with adolescent youth.

Systematic Review of Childhood Sedentary Behavior Questionnaires: What do We Know and What is Next?

BACKGROUND

Accurate measurement of child sedentary behavior is necessary for monitoring trends, examining health effects, and evaluating the effectiveness of interventions.

OBJECTIVES

We therefore aimed to summarize studies examining the measurement properties of self-report or proxy-report sedentary behavior questionnaires for children and adolescents under the age of 18 years. Additionally, we provided an overview of the characteristics of the evaluated questionnaires.

METHODS

We performed systematic literature searches in the EMBASE, PubMed, and SPORTDiscus electronic databases. Studies had to report on at least one measurement property of a questionnaire assessing sedentary behavior. Questionnaire data were extracted using a standardized checklist, i.e. the Quality Assessment of Physical Activity Questionnaire (QAPAQ) checklist, and the methodological quality of the included studies was rated using a standardized tool, i.e. the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) checklist.

RESULTS

Forty-six studies on 46 questionnaires met our inclusion criteria, of which 33 examined test-retest reliability, nine examined measurement error, two examined internal consistency, 22 examined construct validity, eight examined content validity, and two examined structural validity. The majority of the included studies were of fair or poor methodological quality. Of the studies with at least a fair methodological quality, six scored positive on test-retest reliability, and two scored positive on construct validity.

CONCLUSION

None of the questionnaires included in this review were considered as both valid and reliable. High-quality studies on the most promising questionnaires are required, with more attention to the content validity of the questionnaires. PROSPERO registration number: CRD42016035963.

Estimating physical activity in youth using an ankle accelerometer

This study developed and validated a vector magnitude (VM) two-regression model (2RM) for use with an ankle-worn ActiGraph accelerometer. For model development, 181 youth (mean ± SD; age, 12.0 ± 1.5 yr) completed 30 min of supine rest and 2-7 structured activities. For cross-validation, 42 youth (age, 12.6 ± 0.8 yr) completed approximately 2 hr of unstructured physical activity (PA). PA data were collected using an ActiGraph accelerometer, (non-dominant ankle) and the VM was expressed as counts/5-s. Measured energy expenditure (Cosmed K4b²) was converted to youth METs (MET_y; activity VO₂ divided by resting VO₂). A coefficient of variation (CV) was calculated for each activity to distinguish continuous walking/running from intermittent activity. The ankle VM sedentary behavior threshold was ≤10 counts/5-s, and a CV≤15 counts/5-s was used to identify walking/running. The ankle VM2RM was within 0.42 MET_y of measured MET_y during the unstructured PA (P > 0.05). The ankle VM2RM was within 5.7 min of measured time spent in sedentary, LPA, MPA, and VPA (P > 0.05). Compared to the K4b², the ankle VM2RM provided similar estimates to measured values during unstructured play and provides a feasible wear location for future studies.

Activity Recognition in Youth Using Single Accelerometer Placed at Wrist or Ankle

PURPOSE

State-of-the-art methods for recognizing human activity using raw data from body-worn accelerometers have primarily been validated with data collected from adults. This study applies a previously available method for activity classification using wrist or ankle accelerometer to data sets collected from both adults and youth.

METHODS

An algorithm for detecting activity from wrist-worn accelerometers, originally developed using data from 33 adults, is tested on a data set of 20 youth (age, 13 ± 1.3 yr). The algorithm is also extended by adding new features required to improve performance on the youth data set. Subsequent tests on both the adult and youth data were performed using crossed tests (training on one group and testing on the other) and leave-one-subject-out cross-validation.

RESULTS

The new feature set improved overall recognition using wrist data by 2.3% for adults and 5.1% for youth. Leave-one-subject-out cross-validation accuracy performance was 87.0% (wrist) and 94.8% (ankle) for adults, and 91.0% (wrist) and 92.4% (ankle) for youth. Merging the two data sets, overall accuracy was 88.5% (wrist) and 91.6% (ankle).

CONCLUSIONS

Previously available methodological approaches for activity classification in adults can be extended to youth data. Including youth data in the training phase and using features designed to capture information on the activity fragmentation of young participants allows a better fit of the methodological framework to the characteristics of activity in youth, improving its overall performance. The proposed algorithm differentiates ambulation from sedentary activities that involve gesturing in wrist data, such as that being collected in large surveillance studies.

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.

Classification of accelerometer wear and non-wear events in seconds for monitoring free-living physical activity

OBJECTIVE

To classify wear and non-wear time of accelerometer data for accurately quantifying physical activity in public health or population level research.

DESIGN

A bi-moving-window-based approach was used to combine acceleration and skin temperature data to identify wear and non-wear time events in triaxial accelerometer data that monitor physical activity.

SETTING

Local residents in Swansea, Wales, UK.

PARTICIPANTS

50 participants aged under 16 years (n=23) and over 17 years (n=27) were recruited in two phases: phase 1: design of the wear/non-wear algorithm (n=20) and phase 2: validation of the algorithm (n=30).

METHODS

Participants wore a triaxial accelerometer (GeneActiv) against the skin surface on the wrist (adults) or ankle (children). Participants kept a diary to record the timings of wear and non-wear and were asked to ensure that events of wear/non-wear last for a minimum of 15 min.

RESULTS

The overall sensitivity of the proposed method was 0.94 (95% CI 0.90 to 0.98) and specificity 0.91 (95% CI 0.88 to 0.94). It performed equally well for children compared with adults, and females compared with males. Using surface skin temperature data in combination with acceleration data significantly improved the classification of wear/non-wear time when compared with methods that used acceleration data only (p<0.01).

CONCLUSIONS

Using either accelerometer seismic information or temperature information alone is prone to considerable error. Combining both sources of data can give accurate estimates of non-wear periods thus giving better classification of sedentary behaviour. This method can be used in population studies of physical activity in free-living environments.

Reliability and validity of the transport and physical activity questionnaire (TPAQ) for assessing physical activity behaviour

BACKGROUND

No current validated survey instrument allows a comprehensive assessment of both physical activity and travel behaviours for use in interdisciplinary research on walking and cycling. This study reports on the test-retest reliability and validity of physical activity measures in the transport and physical activity questionnaire (TPAQ).

METHODS

The TPAQ assesses time spent in different domains of physical activity and using different modes of transport for five journey purposes. Test-retest reliability of eight physical activity summary variables was assessed using intra-class correlation coefficients (ICC) and Kappa scores for continuous and categorical variables respectively. In a separate study, the validity of three survey-reported physical activity summary variables was assessed by computing Spearman correlation coefficients using accelerometer-derived reference measures. The Bland-Altman technique was used to determine the absolute validity of survey-reported time spent in moderate-to-vigorous physical activity (MVPA).

RESULTS

In the reliability study, ICC for time spent in different domains of physical activity ranged from fair to substantial for walking for transport (ICC = 0.59), cycling for transport (ICC = 0.61), walking for recreation (ICC = 0.48), cycling for recreation (ICC = 0.35), moderate leisure-time physical activity (ICC = 0.47), vigorous leisure-time physical activity (ICC = 0.63), and total physical activity (ICC = 0.56). The proportion of participants estimated to meet physical activity guidelines showed acceptable reliability (k = 0.60). In the validity study, comparison of survey-reported and accelerometer-derived time spent in physical activity showed strong agreement for vigorous physical activity (r = 0.72, p<0.001), fair but non-significant agreement for moderate physical activity (r = 0.24, p = 0.09) and fair agreement for MVPA (r = 0.27, p = 0.05). Bland-Altman analysis showed a mean overestimation of MVPA of 87.6 min/week (p = 0.02) (95% limits of agreement -447.1 to +622.3 min/week).

CONCLUSION

The TPAQ provides a more comprehensive assessment of physical activity and travel behaviours and may be suitable for wider use. Its physical activity summary measures have comparable reliability and validity to those of similar existing questionnaires.

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.

Neural network versus activity-specific prediction equations for energy expenditure estimation in children

The aim of this study was to compare the energy expenditure (EE) estimations of activity-specific prediction equations (ASPE) and of an artificial neural network (ANNEE) based on accelerometry with measured EE. Forty-three children (age: 9.8 ± 2.4 yr) performed eight different activities. They were equipped with one tri-axial accelerometer that collected data in 1-s epochs and a portable gas analyzer. The ASPE and the ANNEE were trained to estimate the EE by including accelerometry, age, gender, and weight of the participants. To provide the activity-specific information, a decision tree was trained to recognize the type of activity through accelerometer data. The ASPE were applied to the activity-type-specific data recognized by the tree (Tree-ASPE). The Tree-ASPE precisely estimated the EE of all activities except cycling [bias: −1.13 ± 1.33 metabolic equivalent (MET)] and walking (bias: 0.29 ± 0.64 MET; P < 0.05). The ANNEE overestimated the EE of stationary activities (bias: 0.31 ± 0.47 MET) and walking (bias: 0.61 ± 0.72 MET) and underestimated the EE of cycling (bias: −0.90 ± 1.18 MET; P < 0.05). Biases of EE in stationary activities (ANNEE: 0.31 ± 0.47 MET, Tree-ASPE: 0.08 ± 0.21 MET) and walking (ANNEE 0.61 ± 0.72 MET, Tree-ASPE: 0.29 ± 0.64 MET) were significantly smaller in the Tree-ASPE than in the ANNEE ( P < 0.05). The Tree-ASPE was more precise in estimating the EE than the ANNEE. The use of activity-type-specific information for subsequent EE prediction equations might be a promising approach for future studies.

Artificial neural networks to predict activity type and energy expenditure in youth

Previous studies have demonstrated that pattern recognition approaches to accelerometer data reduction are feasible and moderately accurate in classifying activity type in children. Whether pattern recognition techniques can be used to provide valid estimates of physical activity (PA) energy expenditure in youth remains unexplored in the research literature.

PURPOSE

The objective of this study is to develop and test artificial neural networks (ANNs) to predict PA type and energy expenditure (PAEE) from processed accelerometer data collected in children and adolescents.

METHODS

One hundred participants between the ages of 5 and 15 yr completed 12 activity trials that were categorized into five PA types: sedentary, walking, running, light-intensity household activities or games, and moderate-to-vigorous-intensity games or sports. During each trial, participants wore an ActiGraph GT1M on the right hip, and VO2 was measured using the Oxycon Mobile (Viasys Healthcare, Yorba Linda, CA) portable metabolic system. ANNs to predict PA type and PAEE (METs) were developed using the following features: 10th, 25th, 50th, 75th, and 90th percentiles and the lag one autocorrelation. To determine the highest time resolution achievable, we extracted features from 10-, 15-, 20-, 30-, and 60-s windows. Accuracy was assessed by calculating the percentage of windows correctly classified and root mean square error (RMSE).

RESULTS

As window size increased from 10 to 60 s, accuracy for the PA-type ANN increased from 81.3% to 88.4%. RMSE for the MET prediction ANN decreased from 1.1 METs to 0.9 METs. At any given window size, RMSE values for the MET prediction ANN were 30-40% lower than the conventional regression-based approaches.

CONCLUSIONS

ANNs can be used to predict both PA type and PAEE in children and adolescents using count data from a single waist mounted accelerometer.

Support vector machines classifiers of physical activities in preschoolers

The goal of this study is to develop, test, and compare multinomial logistic regression (MLR) and support vector machines (SVM) in classifying preschool-aged children physical activity data acquired from an accelerometer. In this study, 69 children aged 3-5 years old were asked to participate in a supervised protocol of physical activities while wearing a triaxial accelerometer. Accelerometer counts, steps, and position were obtained from the device. We applied K-means clustering to determine the number of natural groupings presented by the data. We used MLR and SVM to classify the six activity types. Using direct observation as the criterion method, the 10-fold cross-validation (CV) error rate was used to compare MLR and SVM classifiers, with and without sleep. Altogether, 58 classification models based on combinations of the accelerometer output variables were developed. In general, the SVM classifiers have a smaller 10-fold CV error rate than their MLR counterparts. Including sleep, a SVM classifier provided the best performance with a 10-fold CV error rate of 24.70%. Without sleep, a SVM classifier-based triaxial accelerometer counts, vector magnitude, steps, position, and 1- and 2-min lag and lead values achieved a 10-fold CV error rate of 20.16% and an overall classification error rate of 15.56%. SVM supersedes the classical classifier MLR in categorizing physical activities in preschool-aged children. Using accelerometer data, SVM can be used to correctly classify physical activities typical of preschool-aged children with an acceptable classification error rate.

Tri-axial accelerometer analysis techniques for evaluating functional use of the extremities

Activity monitors provide an objective mechanism for evaluating patient function. It is unclear what similarities or unique information may be yielded using different analyses. Fifteen patients scheduled to undergo shoulder arthroplasty and fifteen matched control subjects wore tri-axial accelerometer activity monitors bilaterally at the lower (wrist) and upper (biceps) arm for 3days. Measures of central tendency, variance, sample entropy, and asymmetry were calculated. A novel technique to evaluate time distribution of activity intensity was also performed. Within both groups there was a difference in central tendency and variance when comparing dominant and non-dominant limbs for both the lower (

CONTROLS

Mean Activity, P<0.001; Max Activity, P<0.001;

PATIENTS

Mean Activity, P=0.044; Max Activity, P=0.009) and upper (

CONTROLS

Mean Activity, P<0.001; Max Activity, P=0.046;

PATIENTS

Mean Activity, P=0.002; Max Activity, P=0.049) arm. Within group differences were also present for lower arm entropy in both groups (CONTROLS, P<0.001; PATIENTS P=0.041), and at the upper arm for patients (P=0.003). There were differences between groups for the asymmetry index for both the lower (P=0.033) and upper arm (P=0.005), and maximum activity level of the lower arm (P=0.05). Between group differences were present for time distribution of activity intensity, as the involved upper arm of patients was inactive for a greater time than controls (P=0.013). These results highlight unique information provided by multiple analysis methods, and include a novel approach of evaluating the distribution of time spent across variable intensity activities.

Methods of Measurement in epidemiology: sedentary Behaviour

BACKGROUND

Research examining sedentary behaviour as a potentially independent risk factor for chronic disease morbidity and mortality has expanded rapidly in recent years.

METHODS

We present a narrative overview of the sedentary behaviour measurement literature. Subjective and objective methods of measuring sedentary behaviour suitable for use in population-based research with children and adults are examined. The validity and reliability of each method is considered, gaps in the literature specific to each method identified and potential future directions discussed.

RESULTS

To date, subjective approaches to sedentary behaviour measurement, e.g. questionnaires, have focused predominantly on TV viewing or other screen-based behaviours. Typically, such measures demonstrate moderate reliability but slight to moderate validity. Accelerometry is increasingly being used for sedentary behaviour assessments; this approach overcomes some of the limitations of subjective methods, but detection of specific postures and postural changes by this method is somewhat limited. Instruments developed specifically for the assessment of body posture have demonstrated good reliability and validity in the limited research conducted to date. Miniaturization of monitoring devices, interoperability between measurement and communication technologies and advanced analytical approaches are potential avenues for future developments in this field.

CONCLUSIONS

High-quality measurement is essential in all elements of sedentary behaviour epidemiology, from determining associations with health outcomes to the development and evaluation of behaviour change interventions. Sedentary behaviour measurement remains relatively under-developed, although new instruments, both objective and subjective, show considerable promise and warrant further testing.

Actigraph accelerometer-defined boundaries for sedentary behaviour and physical activity intensities in 7 year old children

Background

Accurate objective assessment of sedentary and physical activity behaviours during childhood is integral to the understanding of their relation to later health outcomes, as well as to documenting the frequency and distribution of physical activity within a population.

Purpose

To calibrate the Actigraph GT1M accelerometer, using energy expenditure (EE) as the criterion measure, to define thresholds for sedentary behaviour and physical activity categories suitable for use in a large scale epidemiological study in young children.

Methods

Accelerometer-based assessments of physical activity (counts per minute) were calibrated against EE measures (kcal.kg⁻¹.hr⁻¹) obtained over a range of exercise intensities using a COSMED K4b² portable metabolic unit in 53 seven-year-old children. Children performed seven activities: lying down viewing television, sitting upright playing a computer game, slow walking, brisk walking, jogging, hopscotch and basketball. Threshold count values were established to identify sedentary behaviour and light, moderate and vigorous physical activity using linear discriminant analysis (LDA) and evaluated using receiver operating characteristic (ROC) curve analysis.

Results

EE was significantly associated with counts for all non-sedentary activities with the exception of jogging. Threshold values for accelerometer counts (counts.minute⁻¹) were <100 for sedentary behaviour and ≤2240, ≤3840 and ≥3841 for light, moderate and vigorous physical activity respectively. The area under the ROC curves for discrimination of sedentary behaviour and vigorous activity were 0.98. Boundaries for light and moderate physical activity were less well defined (0.61 and 0.60 respectively). Sensitivity and specificity were higher for sedentary (99% and 97%) and vigorous (95% and 91%) than for light (60% and 83%) and moderate (61% and 76%) thresholds.

Conclusion

The accelerometer cut points established in this study can be used to classify sedentary behaviour and to distinguish between light, moderate and vigorous physical activity in children of this age.