A Novel Algorithm for Activity State Recognition using Smartwatch Data

Ebrahim Nemati1, Daniyal Liaqat2, Md Mahbubur Rahman3, Jilong Kuang3

  • 1UCLA
  • 2University of Toronto
  • 3Samsung Research America

Details

12:00 - 13:45 | Mon 6 Nov | Auditorium Foyer, E1/E2, Upper Atrium Space | MLunch_Break.5

Session: Lunch, Posters and POC Technologies Demonstrations – Session I

Full Text

Digital Library Website

Abstract

This work presents a novel algorithm for recognizing activity states which are of interest for assessing the general well-being of cancer, frail and elderly patients. Using the novel idea of two-level classification, misclassification due to unwanted hand motion noise, which is a common source of error in wrist-worn sensing systems, is mitigated. The algorithm is verified using data from 20 subjects performing a sequence of related activities. It is shown that the proposed algorithm improves the accuracy value for the “activity state” which includes “sit”, “stand” and “move” by up to 8%.