SCH: INT: Collaborative Research: A Data-Driven Approach for Enhancing Wearable Device Performance - A Study on Early Detection of Asthma Exacerbation

点击翻译按钮获取中文摘要

Advances on wearable devices have enabled the continuous sensing of a number of physiological parameters such as heart rate, heart rate variability, respiratory rate, activity levels, and coughing. These parameters can be used for a number of health applications, including prediction of asthma exacerbation, to achieve efficient management and prevention of severe symptoms. However, there have been significant challenges identified for the broad adoption of wearable devices, in particular ensuring reliable measurements and maximizing their battery-life. In current practice, clinical gold-standard devices can obtain reliable measurements in medical and controlled environments whereas wearable technologies target to be integrated into daily life and be reliable in unconstrained real-world conditions. As a result, most current procedures to evaluate asthma-related wearable devices often take place in controlled environments and do not capture the broad spectrum of scenarios that a device may be exposed to during an individual's daily use. These real-world scenarios can compromise data quality and usefulness of a device. In this project, the investigators aim to provide an innovative framework for characterizing the performance of wearable devices in the real-world based on contextual information of their usage, and aim to demonstrate the framework's value by enabling more reliable early detection of asthma exacerbations in young adults. The data produced by this award will be used as part of projects for undergraduate and graduate students. Demonstrations and video materials will be produced as part of the outreach efforts for K-12 and underrepresented communities.The investigators plan to achieve their scientific goals by focusing on three research thrusts. (1) Characterization of signal quality: A robust statistical framework will be developed to characterize signal quality in the real-world based on the context in which they are used. Context will be represented using activity, environmental and device-state information. The project will develop a supervised methodology using controlled in-lab experiments, and expand the framework to be unsupervised/ semi-supervised in order to be applicable to real-world conditions. (2) Development of a signal-quality and context-aware inference model for early asthma exacerbation: The characterization of signal quality will be used to develop more reliable inference pipelines. (3) Feedback to user and device: Users will be provided with easy-to-interpret and actionable feedback on the inference and any adjustments needed for the device. The effect of this feedback on signal quality and user satisfaction will be studied. The device will also receive feedback in the form of parameter settings associated with sampling and filtering that will ensure accurate levels of prediction while minimizing the power profile.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.