University of Houston

The importance of understanding modern walking behavior and designing effective technology-mediated interventions cannot be overestimated. Suffice to say that the fall of walking activity coincides with the rise of the sedentary lifestyle and the obesity epidemic in modern societies. Studying quantitatively walking behavior has been distinctly difficult, due to the large number of subjects needed, the 24/7 nature of monitoring required, and the desired length of the observation window. In my Ph.D. work I have introduced a new method to conduct large scale physical activity studies, achieving maximum outreach and unprecedented longitudinal horizon at a minimal cost. This method is based on well designed smartphone (iPhone) applications that are freely available through communal portals, such as the App Store. An application of this kind quantifies calories expended through walking (the most ubiquitous physical activity) via interpretation of the phone's accelerometer values. A major problem I had to address in this respect was positional calibration. Accelerometer values can reliably translate to caloric values via indirect calorimetry calibration only when the accelerometer (smartphone) is worn on the waist. This is not always the case, and I developed a statistics-based method that adjusts any other body position to an equivalent waist position for accelerometer calibration purposes. Partly due to measurement reliability afforded by the positional calibration method and partly due to simple and appealing interface design (the end result of a repetitive process), thousands of users around the globe took to my application (Walk n' Play) the last two years providing a trove of anonymous data. These data provide a window to human walking behavior and paint clear quantitative profiles of ideal and non-ideal users, opening intervention to hard science rather than speculation.