Abstract: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.

Analysis of Personal Life Patterns Using Accelerometer-Based Wearable Devices

An Automatic User-Adapted Physical Activity Classification Method Using Smartphones.

SmartCare: Energy-efficient Long-Term Physical Activity Tracking Using Smartphones

Classifying Tasks Performed by Electrical Line Workers Using a Wrist-Worn Sensor: A Data Analytic Approach

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

Real-Time Physical Activity Classification And Tracking Using Wearble Sensors

Activity Recognition on an Accelerometer Embedded Mobile Phone with Varying Positions and Orientations

Predictive Walking-Age Health Analyzer

Bring Gait Lab to Everyday Life: Gait Analysis in Terms of Activities of Daily Living

Experimental Analysis Of Artificial Neural Networks Performance For Physical Activity Recognition Using Belt And Wristband Devices

Implementation of a Real-Time Human Activity Classifier Using a Triaxial Accelerometer and Smartphone

Physical Activity Recognition by Utilising Smartphone Sensor Signals

Experimental Analysis of Artificial Neural Networks Performance for Accessing Physical Activity Recognition in Daily Life

Sensor Positioning for Activity Recognition Using Wearable Accelerometers.

Precise Measurement of Physical Activities and High-Impact Motion: Feasibility of Smart Activity Sensor System

Real-time Dynamic Data Analysis Model Based on Wearable Smartband

Gait Assessment of Younger and Older Adults with Portable Motion-Sensing Methods: A User Study.

Quantitative Analysis of Human Daily Activities Based on a Wireless Sensor System

Real-time Human Activity Classification by Accelerometer Embedded Wearable Devices.

Home health care system for the Elderly based on IMU wearable device

Recognizing Manual Activities Using Wearable Inertial Measurement Units: Clinical Application for Outcome Measurement