CAREER: Advances in Monitoring Human Performance: Moving Wearable Technology from the Expert to Nonexpert User

Massachusetts Institute Of Technology, Cambridge MA

Investigators

Abstract

Wearable computing technology is rapidly proliferating and playing an increasing role in our daily lives. The PI's focus in this research is on developing technology for the monitoring by non-experts of human performance within the context of stroke rehabilitation, which will be applicable to a wide spectrum of applications. Robust performance metrics that can be interpreted by a non-expert would enable people to track their well-being in a manner not currently possible. With technology in the home environment, there is the potential to better engage the user in self-monitoring, to increase motivation, and to improve motion strategies for activities related to well-being. From the healthcare professional's perspective, wearable technology in the home could allow the clinician to change the balance of time so as to emphasize educating and working with the patient on enabling tasks, because the wearable technology would provide information on compliance history and progress. The longitudinal data from the sensors would also permit improved evaluation of patient-specific dose-response sensitivity. The human-centered research methodology implemented here will also provide new insights into systems modeling heuristics, in particular how to formalize relationships between the human and computer entities of the systems architecture. Because the research will involve both healthy and stroke participant groups, project outcomes will include a novel database with participant demographics, expert outcome measures, and daily home task performance which will permit the advancement of new algorithms and will provide a way to compare algorithm performance across populations. The PI argues that higher fidelity motion sensing is the key to empowering improved human performance, goal monitoring, and well-being. To this end, in this project she will extend the capabilities of wearable motion sensing technology through advances in dynamic system modeling and signal processing to account for the underlying variability in motion and compliant structure of the individual. A cyber-human platform will be developed for those with limited knowledge in sensor technology and physiological systems (non-experts), through analysis of performance metrics and decision-making interfaces with the end user in mind. The effort will involve three related thrusts that will be demonstrated within the context of stroke rehabilitation: characterization of variability for relevant tasks in a natural environment; application of estimation algorithms and investigation of performance metrics robust to uncertainties in the natural environment; and evaluation of decision-making interfaces synergistic with the expected end user. The PI will implement novel estimation and calibration algorithms to inform performance metric generation, and will integrate these parameters into a user interface that is evaluated in human studies as a platform for decision making across expertise level. By bridging biomechanics and control theory, new capabilities will be enabled for wearable motion-sensing devices that integrate relevant nonlinear models with the appropriate stochasticity, which in turn will lead to exciting research opportunities for the biomechanics community to understand motor behavior in natural settings, as well as adaptations and extensions in control theory from a methods perspective due to new challenges in maintaining calibrations for systems with compliance and underlying variability.