Estimation of Walking Exercise Intensity Using 3-D Acceleration Sensor

This paper describes a method to estimate the measure of human energy expenditure referred to as metabolic equivalents (METs). The measure is defined by an index of the amount of energy a person expends through normal daily activities. The average MET values are defined by the type of activity; for example, brisk walking corresponds to 4 METs. The index lists 107 activities and the corresponding MET values. However, to determine the energy consumed in a day, each type of activity and its duration must be noted, which is inconvenient and time consuming. Here, we consider a quantitative method to estimate the METs from the acceleration of waist motion in our activities. We consider a simple physical model of amoving human body, define the power output for an activity, and then relate the maximum oxygen intake rate per minute, which is proportional to the energy expenditure per minute with the power output. Based on these factors, we developed an automatic MET estimator using a 3-D accelerometer. The system was examined during various selected activities, and the validity of the MET estimation method was tested in activities, excluding very intense activities. The exercise intensity for walking at various speeds was estimated with an average error of -0.6 METs.