"What does the left hand tell us? - Comparing acceleration and GPS data to detect movement patterns"

Study background: Movement patterns in normal life settings are an interesting topic by itself e.g. for health research issues. But they get an added research value, if they can be linked to parallel psychosomatic changes like electrodermal or cardiovascular responses and -  if data collection can be accomplished not only with clinical but also with random population samples. A prerequisite for both aspects is that data collection by acceleration sensors can be conducted by an unobtrusive, highly efficient method with lowest costs to study participants. Electronic devices (with embedded microprocessor and sensors), which can be easily put on the wrist, seem to be the solution – preferentially on the left resp. non-active hand. But – which kind of movement patterns can be detected from left hand acceleration data? Relying on only one body point (the left ankle) data, is it possible to separate  gesture and body motion patterns resp. different body postures, like sitting, standing and walking?

 Method: In two studies people weared an electronic textile band, with embedded microprocessor and tri-axial acceleration sensor, which measured acceleration every second (sampling rate 1 Hz). Participants got also a GPS logger, which simultaneously captured data on geographic position as well on speed of movement at the same sampled rate. In the first study, 10 people visited a quarter artist festival, walking around and standing for watching events they were interested in. By changing geographical position and speed, standing and walking resp. fast and slower walking patterns are easily defined. Acceleration data are used to identify these movement patterns. The second study was conducted with 30 participants of city sightseeing tours, wearing the electronic band and a GPS logger. The sightseeing tour was a mix of sitting in the bus and travelling through the city and of leaving the bus for a walk to and into specific places and buildings. With these data, gestures while sitting in a transportation vehicle plus walking around are examined by comparing acceleration and GPS data trajectories.

  Results: It is found, that general movement patterns can be detected by the ankle measurement acceleration data, namely standing and walking, but also sitting. It could be also shown that the specific arrangement of the acceleration sensor at the underside of the wrist was very effective in detecting posture change patterns. By observing reversals in the relative distance of dimensionspecific acceleration values, a change from a sitting to a standing position was clearly identifiable.

  Discussion: Acceleration of the left resp. non-active ankle is a mixture of different body movement dimensions, trunk, arms and hands. Comparing the acceleration data with concomitant movement data by GPS showed the possibility of separating general movement patterns by this one source acceleration data. But more sophisticated statistical modelling, and more specific field experiments in every day life settings are needed to provide reliable indication of motility patterns by acceleration profiles.

 

Author:

Dr. Georgios Papastefanou

 

Affiliation: gesis-zuma Center for Survey Research and Methodology, Mannheim , Germany ,

B2,1 – 68159 Mannheim

www.gesis.org

georgios.papastefanou@gesis.org

International Conference on Ambulatory Monitoring of Physical Activity and Movement (ICAMPAM), Rotterdam, 21.5. - 24.5.2008

http://www.icampam.org/