UNIVERSITY OF PITTSBURGH THE
Human-observer based methods for measuring human motion are labor intensive, qualitative, and difficult to standardize across laboratories, clinical settings, and over time. Moreover, many conditions that affect normal human movements must be diagnosed during short visits to the clinician, complicating the process. Advances in wearable and wireless sensor networks have opened up new opportunities for monitoring and coaching in health care systems. During the timeframe of this grant, we will develop a cyber-physical system composed of accelerometers and novel machine learning algorithms to analyze data in the context of a set of driving health care applications.
To make the accelerometer information useful to doctors and patients, modeling and analysis is needed to extract, represent, and parse subtle human motion, body gestures, and pyschophysiological indicators from the stream of accelerometer data. Toward this goal, we will develop novel machine learning algorithms for temporal segmentation, recognition, and classification of subtle human motion events. These techniques will allow quantification of human motion and improved monitoring and assessment of medical conditions. In addition, our systems will provide full time monitoring and analysis using a lightweight wearable system.