Low-power metabolic equivalents estimation algorithm using adaptive acceleration sampling

Annu Int Conf IEEE Eng Med Biol Soc. 2016 Aug:2016:1878-1881. doi: 10.1109/EMBC.2016.7591087.

Abstract

This paper describes a proposed low-power metabolic equivalent estimation algorithm that can calculate the value of metabolic equivalents (METs) from triaxial acceleration at an adaptively changeable sampling rate. This algorithm uses four rates of 32, 16, 8 and 4 Hz. The mode of switching them is decided from synthetic acceleration. Applying this proposed algorithm to acceleration measured for 1 day, we achieved the low root mean squared error (RMSE) of calculated METs, with current consumption that was 41.5 % of the value at 32 Hz, and 75.4 % of the value at 16 Hz.

MeSH terms

  • Acceleration*
  • Adult
  • Algorithms*
  • Decision Trees
  • Exercise / physiology
  • Female
  • Humans
  • Male
  • Metabolic Equivalent*
  • Young Adult