Artificial Neural Networks to Predict Activity Type and Energy Expenditure in Youth

Trost, Stewart G.; Wong, Weng Keen; Pfeiffer, Karen A.; Zheng, Yujie

doi:10.1249/mss.0b013e318258ac11

Cited by 119 publications

(144 citation statements)

References 16 publications

Supporting

Mentioning

139

Contrasting

Order By: Relevance

“…The results are summarised in Table 2. In agreement with the results of Trost et al 6 recognition accuracy for the MLP was 88.4%. Recognition accuracy for the SOM and DLEN was higher than that observed for preschool children at 75.1% and 89.7%.…”

Section: Resultssupporting

confidence: 92%

“…As in the current study, activity trials were categorised into five activity classes; sedentary activities, light house-hold activities or games, moderate-to-vigorous games and sports, walking, and running. We hypothesised that the MLP model would provide similar recognition accuracy to that reported by Trost et al 6 and that the DLEN would provide higher recognition accuracy than the standard MLP or SOM. The results are summarised in Table 2.…”

Section: Resultsmentioning

confidence: 69%

“…The confusion matrices show that the MLP and SOM commonly confused walking and running with more generic classes such as light activities and games and moderate activities. < Insert Table 1 here > Accelerometry data is available from an earlier study evaluating the accuracy of a standard MLP in 5-15 year-old school-aged children and adolescents (n = 100) 6 . As in the current study, activity trials were categorised into five activity classes; sedentary activities, light house-hold activities or games, moderate-to-vigorous games and sports, walking, and running.…”

Section: Resultsmentioning

confidence: 99%

“…However, conventional regression-based approaches are limited in their ability to accurately predict energy expenditure across a wide range of activities 3,4,5 , because the relationship between accelerometer counts and energy expenditure (EE) differs according to the type of activity performed. Not surprisingly, cut-point methods exhibit 28%-45% misclassification of PA intensity in children and adolescents 3,5,6 . As accelerometry use is widespread, this level of misclassification has significant implications for understanding and promoting PA among children and adolescents internationally.…”

Section: Introductionmentioning

confidence: 99%

“…Trost and colleagues 6 developed and tested an ANN to classify PA type from second-bysecond hip-worn ActiGraph data in 5 to 15 year-olds. Participants completed 12 activity trials that were categorised into 5 activity types: sedentary, walking, running, light intensity house-hold activities or games, and moderate-to-vigorous games or sports.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Prediction of activity type in preschool children using machine learning techniques

Hagenbuchner

Cliff

Trost

et al. 2015

Journal of Science and Medicine in Sport

Self Cite

View full text Add to dashboard Cite

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, moderateto-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. KeywordsPhysical activity, pattern recognition, accelerometry, neural networks, exercise, validity Disciplines Medicine and Health Sciences | Social and Behavioral Sciences Publication DetailsHagenbuchner, M., Cliff, D. P., Trost, S. G., Van Tuc, N. & Peoples, G. E. (2015). Prediction of activity type in preschool children using machine learning techniques. Journal of Science and Medicine in Sport, 18 (4) AbstractObjectives: 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...

show abstract

Section: Resultssupporting

confidence: 92%

Section: Resultsmentioning

confidence: 69%

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Prediction of activity type in preschool children using machine learning techniques

Hagenbuchner

Cliff

Trost

et al. 2015

Journal of Science and Medicine in Sport

Self Cite

View full text Add to dashboard Cite

show abstract

A Visual Variability and Visuo‐Tactile Coordination Inspired Child Adaptation Mechanism for Wearable Age Group Recognition and Activity Recognition

Kang

Jiang

et al. 2022

Advanced Intelligent Systems

View full text Add to dashboard Cite

Exercise is critical to children's mental and physical health. However, improper exercise can be counterproductive. [1] Research shows that excessive exercise can cause stress fractures, ligamentous injuries, and knee articular cartilage damage in children, while lack of exercise can lead to obesity and even depression. [2] Therefore, it is necessary to provide children with professional coaches and health advisers. However, professional guidance is expensive and impractical in rural or underdeveloped areas where health infrastructure is weak and professional coaches are not available. [3] Recently, the issue became more obvious because the coronavirus pandemic quarantined children from their schools' athletic facilities, leaving them with less professional guidance from their teachers. [4] Fortunately, wearable devices (e.g., smartwatches or smart-bands) can provide motion monitoring and exercise advice to people who do not have access to professional coaches and health advisers at a low cost and with good accessibility. [5,6] However, children are often ignored in the design of these devices. Wearable devices and algorithms developed for children lack diversity and function, forcing them to use commercial adult devices or algorithms to get high-quality remote exercise guidance. [7] Utilizing adult devices or algorithms on children faces two severe problems. First, human physical characteristics show variability in different activities. It is hard to build an age group recognition model robust to multiple activities. Second, due to the physiological differences between children and adults, human activity recognition algorithms developed on adults work poorly on children, making motion monitoring inaccurate. [8] Previous research reports a 10.8%-27.1% recognition decrease when using a model trained by adult data to recognize child activity. [9] Providing exercise advice through wearable fitness devices requires three steps: 1) recognizing the user's current activity; 2) monitoring duration or intensity; and 3) providing advice based on the user's age group characteristics and data collected in step 1 and 2. Therefore, incorrect activity recognition and exercise standard misuse will lead to improper exercise advice and can cause excessive exercise or lack of exercise, which would finally lead to harm for the children. [10] In addition, lots of

show abstract

Classification of human physical activity based on raw accelerometry data via spherical coordinate transformation

Kos

Bogdan

Glynn

et al. 2020

Statistics in Medicine

View full text Add to dashboard Cite

Summary Human health is strongly associated with person's lifestyle and levels of physical activity. Therefore, characterization of daily human activity is an important task. Accelerometers have been used to obtain precise measurements of body acceleration. Wearable accelerometers collect data as a three‐dimensional time series with frequencies up to 100 Hz. Using such accelerometry signal, we are able to classify different types of physical activity. In our work, we present a novel procedure for physical activity classification based on the raw accelerometry signal. Our proposal is based on the spherical representation of the data. We classify four activity types: resting, upper body activities (sitting), upper body activities (standing), and lower body activities. The classifier is constructed using decision trees with extracted features consisting of spherical coordinates summary statistics, moving averages of the radius and the angles, radius variance, and spherical variance. The classification accuracy of our method has been tested on data collected on a sample of 47 elderly individuals who performed a series of activities in laboratory settings. The achieved classification accuracy is over 90% when the subject‐specific data are used and 84% when the group data are used. Main contributor to the classification accuracy is the angular part of the collected signal, especially spherical variance. To the best of our knowledge, spherical variance has never been previously used in the analysis of the raw accelerometry data. Its major advantage over other angular measures is its invariance to the accelerometer location shifts.

show abstract

Artificial Neural Networks to Predict Activity Type and Energy Expenditure in Youth

Cited by 119 publications

References 16 publications

Prediction of activity type in preschool children using machine learning techniques

Prediction of activity type in preschool children using machine learning techniques

A Visual Variability and Visuo‐Tactile Coordination Inspired Child Adaptation Mechanism for Wearable Age Group Recognition and Activity Recognition

Classification of human physical activity based on raw accelerometry data via spherical coordinate transformation

Contact Info

Product

Resources

About