DESCRIPTION (provided by applicant): To objectively measure physical activity, researchers often use small waist-borne accelerometers such as the Actigraph. These devices have internal clocks and large memory capacity, making them well suited for tracking the "pattern" of activity, and capturing multiple short bouts of activity performed over the course of a day. However, previous methods of analyzing accelerometer data had limited accuracy for predicting energy expenditure and time spent in different intensity categories. Recently, we developed a new method of analyzing data that improves upon the prediction of EE. This new method distinguishes walking or jogging from intermittent lifestyle activity, based on the variability in counts between successive epochs. It utilizes the fact that walking and running are rhythmic forms of locomotion with consistent counts across time, whereas housework, gardening, sports, etc. are much more variable. By constructing two different regression lines (one representing locomotor activities, and one representing intermittent, lifestyle activities) one can vastly improve upon the prediction of energy expenditure during 10-min bouts of continuous activity. Thus, "proof of concept" has been shown for the new technique, but some limitations still exist. The current grant proposal outlines a series of steps that need to be accomplished before the new method is finally developed and ready for implementation. The specific aims of this two-year project are: (1) to further refine and develop the new method to permit more accurate predictions of EE and bout duration, (2) to validate the new method in a group of adults 20-60 yrs of age (n=64) who will perform structured activities in 10-minute bouts, and (3) to validate the refined new method of predicting EE and "time spent in various intensity categories" in a six-hour study of free-living activity. The overall goal of the project is to develop algorithms for predicting EE and time spent in activity, disseminate them in peer-reviewed journal articles, and incorporate them into the Actigraph company's software. The proposed study will lead to an improved method of measuring physical activity in free-living individuals using a device that is already accepted and widely available. The improved method could then be used in future studies of the relationships between physical activity and health outcomes.