Development of a multi-wear-site, deep learning-based physical activity intensity classification algorithm using raw acceleration data.

BACKGROUND: Accelerometers are widely adopted in research and consumer devices as a tool to measure physical activity. However, existing algorithms used to estimate activity intensity are wear-site-specific. Non-compliance to wear instructions may lead to misspecifications. In this study, we developed deep neural network models to classify device placement and activity intensity based on raw acceleration data. Performances of these models were evaluated by making comparisons to the ground truth and results derived from existing count-based algorithms. METHODS: 54 participants (26 adults 26.9±8.7 years; 28 children, 12.1±2.3 years) completed a series of activity tasks in a laboratory with accelerometers attached to each of their hip, wrist, and chest. Their metabolic rates at rest and during activity periods were measured using the portable COSMED K5; data were then converted to metabolic equivalents, and used as the ground truth for activity intensity. Deep neutral networks using the Long Short-Term Memory approach were trained and evaluated based on raw acceleration data collected from accelerometers. Models to classify wear-site and activity intensity, respectively, were evaluated. RESULTS: The trained models correctly classified wear-sites and activity intensities over 90% of the time, which outperformed count-based algorithms (wear-site correctly specified: 83% to 85%; wear-site misspecified: 64% to 75%). When additional parameters of age, height and weight of participants were specified, the accuracy of some prediction models surpassed 95%. CONCLUSIONS: Results of the study suggest that accelerometer placement could be determined prospectively, and non-wear-site-specific algorithms had satisfactory accuracies. The performances, in terms of intensity classification, of these models also exceeded typical count-based algorithms. Without being restricted to one specific wear-site, research protocols for accelerometers wear could allow more autonomy to participants, which may in turn improve their acceptance and compliance to wear protocols, and in turn more accurate results.

Examining the criterion validity of two scalable, information technology-based systems designed to measure the quantity and quality of movement behaviours of children from Hong Kong primary schools: a cross-sectional validation study.

OBJECTIVES: Research has shown that having adequate quantity and quality of physical activity can contribute to the health and well-being of children. Nonetheless, existing tools to measure these constructs in children have limitations in terms of their objectivity and scalability. In this study, we provide criterion validity evidence of two systems built on commercially available sensors (ie, gyroscopes and infrared cameras), designed to measure children's moderate-to-vigorous physical activity and fundamental movement skill proficiencies. DESIGN: Cross-sectional. SETTING: Primary schools in Hong Kong. PARTICIPANTS: Data from 30 (age=8.55±1.25 years) and 1174 (age=9.15±1.63 years) children were included for the validation of physical activity and fundamental movement skills measures, respectively. Children's outcomes were simultaneously measured using the developed systems and existing, well-established measures (accelerometers and expert ratings). RESULTS: We found a strong correlation between physical activity outcomes measured using our developed system and accelerometers (Pearson r=0.795). Motor skill proficiency scored using our real-time rating system had strong agreement with expert ratings (percentage agreement=84%-94%, kappa=0.661 to 0.859). DISCUSSION: Results of the current study supported the application of the respective systems in physical education and large-scale research studies. Collection of such data at mass levels could help researchers depict the complex relation between children's quantity and quality of physical activity.