Michael Sung

LiveNet

The research in this thesis will demonstrate that we can use simple non-invasive physiological and contextual sensing using the LiveNet system to accurately classify a variety of physiological conditions. First, it will be demonstrated that non-invasive sensing can be correlated, and thus serve as a feature substitutes, to physiological data from more invasive and unwieldy forms of medical monitoring devices while still providing a high level of diagnostic power.

From this foundation, the LiveNet System is deployed in a number of studies to quantify physiological and affective state. First, a number of classifiers for important contextual cues such as activity state will be developed from basic physiological sensing. It is then demonstrated that the LiveNet System can be used to classify a number of physiological state and pathological conditions and that is robust in the presence of noise, motion artifacts, and other adverse conditions found in real-world situations. These studies include an on-going cold exposure and core body temperature classification study for real-time soldier monitoring systems and an epilepsy study that can classify a variety of epileptic convulsions and muscle spasms. We will also demonstrate that the same methodology can also be used to determine "soft" variables such as depression state. From there, we will demonstrate that we can develop real-time implementations of these classifiers to develop proactive health monitors that can provide instantaneous feedback.

The LiveNet System also opens up the door for long-term continuous monitoring applications for physiological trends that vary slowly with time. This thesis will show how long-term physiology changes can be captured to objectively measure medical treatment and medication efficacy in clinical depressed patients. Likewise, long-term modeling and trending of everyday behavior will be conducted based on simple physiologic and contextual information, demonstrating the possibility of predicting high-level human behavior based solely on ambulatory mobile sensing technology.

Because the LiveNet system is a distributed system that is capable of streaming data to arbitrary endpoints, we can imagine a variety of telemedicine applications where real-time feedback of important health state can be relayed both to individuals as well as to doctors and other care-givers. However, the analogy of sharing health information can be generalized beyond the specific critical health monitoring domain for such things as soldier systems and hospital networks to what I dub "Lifescape Networking," or the sharing of long-term health and physiology information with peer groups in a transparent and meaningful manner. Associated benefits include the sense of telepresence or virtual proximity of peer groups through multimodal data streams (including audio/video, physiological, and other contextual cues) and "Peer-to-Peer" pressure for eliciting healthy behavior in groups via the explicitly laid out social support infrastructure.

The LiveNet system embodies a flexible system infrastructure capable of a variety of individual and group-based context-aware healthcare applications. As the various applications and studies explored in this thesis will demonstrate, there is great promise for this system to be able to allow both individuals and groups of people to manage their health state and support each other more effectively.