Project

Accelerometer-based activity monitoring

Introduction

There exists no reliable, non-invasive, portable, inexpensive and widely validated energy expenditure measurement sensor. When one considers that statistics report that a frightening percentage of today's youth and adults as being overweight and that a large portion of our elderly suffer from conditions that inhibit their physical activity the need for a reliable, objective and portable energy measurement system is self evident. 

Recent researchers in activity monitoring have published exciting results in this form of energy expenditure estimation. The favoured means of measuring this is with microelectromechanical systems (MEMS) accelerometers. Accelerometers are sensors which can detect changes in motion along a directional axis.

Indirect calorimetry (IC) is the gold standard in energy expenditure measurement. The heat that a living organism produces is a function of the ratio of O2 and CO2 it expresses, and IC exploits this fact to regress the heat energy to an energy expenditure measurement, i.e. kilo calories, kilo Joules etc. In general an IC monitor is large, very expensive, nonportable and requires a gas mask to measure the gaseous content of the breath.

 Aims

The aim of this project is to approximate real-time energy expenditure through analysis of accelerations. A number of situations may benefit from an accurate and mobile energy expenditure module such as this, and a short list of potential applications is provided below –

  • Home monitoring of the elderly
  • Monitoring of mobility of post procedure patients
  • Assessment of exertion of daily exercise

Accelerometers provide the means to perform this energy expenditure estimation, but they may also be used in the analysis of the gait of the stride of a subject, i.e. analysis of the motion of limbs during exercise and any form of locomotion, which can provide information on the quality of movement of a subject.

Procedure

A treadmill was set up to operate from 3kmph to 7kmph at 1kmph increments. Volunteers were requested to walk on it at these speeds and five accelerometers were placed on the body to record acceleration data. The accelerometers were placed on the ankle, knee, waist, wrist and arm and were attached to the body via custom-made neoprene straps. The accelerations that were sensed were recorded on a nearby PC via a Bluetooth radio.

To begin with algorithms developed by researchers in the area of accelerometer-based energy expenditure regression were used and their results were compared to reference levels recorded by the indirect calorimeter that was used, and the R2 values are presented in the table below (further algorithms will be implemented over time).

 

Algorithm

R2

Chen nonlinear

0.92

Chen linear

0.88

Bouten

0.85

Crouter

0.81

Custom made energy expenditure algorithms will then be designed based on the strengths of the algorithms that have been implemented, and research into the optimal location of the sensor on the body will also be investigated; some researchers place the sensor on the waist or wrist and there is no clear indication which location is optimal.

The effects of cooperating sensors, be they subsequent accelerometers or fresh brands of sensors, on the estimated energy expenditure levels will then be determined and the algorithms modified accordingly.

Top