8th International Conference on Body Area Networks

Research Article

Energy Expenditure Estimation using Smartphone Body Sensors

  • @INPROCEEDINGS{10.4108/icst.bodynets.2013.253699,
        author={Amit Pande and Yunze Zeng and Aveek Das and Prasant Mohapatra and Sheridan Miyamoto and Edmund Seto and Erik Henricson and Jay Han},
        title={Energy Expenditure Estimation using Smartphone Body Sensors},
        proceedings={8th International Conference on Body Area Networks},
        keywords={accelerometer barometer energy expenditure artificial neural networks},
  • Amit Pande
    Yunze Zeng
    Aveek Das
    Prasant Mohapatra
    Sheridan Miyamoto
    Edmund Seto
    Erik Henricson
    Jay Han
    Year: 2013
    Energy Expenditure Estimation using Smartphone Body Sensors
    DOI: 10.4108/icst.bodynets.2013.253699
Amit Pande1,*, Yunze Zeng1, Aveek Das1, Prasant Mohapatra1, Sheridan Miyamoto2, Edmund Seto3, Erik Henricson2, Jay Han2
  • 1: University of California, Davis, CA, USA
  • 2: UC Davis School of Medicine, Sacramento, CA, USA
  • 3: University of California, Berkeley
*Contact email: pande@ucdavis.edu


Energy Expenditure Estimation (EEE) is an important step in tracking personal activity and preventing chronic diseases such as obesity, diabetes and cardiovascular diseases. Accurate and online EEE utilizing small wearable sensors is a difficult task, primarily because most existing schemes work offline or using heuristics. In this work, we focus on accurate EEE for tracking ambulatory activities (walking, standing, climbing upstairs or downstairs) of a common smartphone user. We used existing smartphone sensors (accelerometer and barometer sensor), sampled at low frequency, to accurately detect EEE. Using Artificial Neural Networks, a machine learning technique, we build a generic regression model for EEE that yields upto 89% correlation with actual Energy Expenditure (EE). Using barometer data, in addition to accelerometry is found to significantly improve EEE performance (upto 15%). We compare our results against state-of-the-art Calorimetry Equations (CE) and consumer electronics devices (Fitbit and Nike+ Fuel Band). We were able to demonstrate the superior accuracy achieved by our algorithm. The results were calibrated against COSMED K4b2 calorimeter readings.