Lifestyle data collection systems

The purpose of this research is to enhance the collection of lifestyle data in three different aspects: positioning, monitoring, and reporting. Positioning of the user is done by using wireless local area network (WLAN) fingerprint methods, which are typically used for indoor localization. These methods are able to locate the individual by comparing fingerprints, which are vectors that hold received signal strength (RSS) measurements from available access points (AP). A typical problem that affects the accuracy and robustness of these fingerprint methods is that RSS measurements are often unstable due to WLAN card, AP, or medium-related transient effects. This research attempts to develop a more resilient method for indoor positioning by mitigating the impact that these transient effects have in the accuracy of positioning. Monitoring of the user's physical activities is done via smartphone integrated motion sensors. This type of monitoring has been done before by analyzing the accelerometer and gyroscope signals from the smartphone. However, most of this research still remains in the laboratory due to usability issues. This research attempts to develop a more usable framework for physical activity monitoring by smartphones. Specifically, the focus is on improving recognition when the data comes from subjects that have not been used to train the system, and which are wearing the smartphone in an unknown orientation, because random orientation of the smartphone poses a problem for recognition of the physical activity being performed. The set of activities to be recognized has also been chosen carefully, with the purpose of recognizing useful activities, as opposed to simply choosing a set of activities that in combination may lead to accuracy numbers higher than those obtained by other researchers for a different activity set. Reporting of lifestyle data is done by the user via text messaging. Automated Short Message Service (SMS) systems have been used before for user data collection. However, most automated messaging systems are limited to simplistic conversations, mainly due to the limitations of the SMS technology (i.e., delays, message length, encoding). These technical difficulties hinder the development of systems that could otherwise be able to carry more complex but useful conversations with a human. In this case, this research attempts to develop a set of strategies that contribute to the design, implementation, testing, and scalability of automated text messaging systems.