ABSTRACT
Physical activity (PA) is a body movement generated by skeletal muscle actions that increase energy expenditure above resting levels. This chapter highlights research on the multisensor data fusion technique applied to PA assessment. PA assessment typically involves two main steps. Activities as "signals" are first captured by the Measurement system, which consists of sensing elements, data conversion, and data transmission. Various body-worn sensors, including accelerometers, foot pressure sensors, and physiological sensors such as respiration, heart rate, and skin temperature sensors, have been used for the PA assessment. Time- and frequency-domain features are commonly used for PA assessment. Support vector machines (SVMs) transform the original multisensory signal features into a different and usually higher feature space/domain, where the relationship between the multisensor data and PA is modeled. The regression version of the SVM, support vector regression (SVR), can be implemented to predict energy expenditure associated with each activity.
