CAREER: Closed-loop Health Behavior Interventions in Multi-device Environments

Suny At Stony Brook, Stony Brook NY

Investigators

Abstract

Motivated by the rising caregiver burden and challenges in remote health behavior monitoring, the proposed research will enable effective assistive interventions in response to dynamically changing health behaviors for target populations. To be effective and impactful, assistive mechanisms need to capture and respond to the subtle and changing context of the human. Human behaviors, however, are challenging to learn due to their complexity and the constantly changing physical, social, and environmental context. Recently, wearables have emerged to fill this gap as users are adopting a variety of devices to help them monitor health related parameters. Given their ubiquity, wearables are positioned ideally to deliver persuasive content aimed at improving users’ health outcomes. However, there is a need for a holistic approach to infer human health behaviors, even as the user's context and the devices measuring their behavior vary over time. The proposed research has the potential to transform human health outcomes by capturing and responding to fine-grained behavioral information continuously, inexpensively, and unobtrusively. This human-in-the-loop system will facilitate rapid development of Health applications by providing the foundations for using adaptive and personalized interventions for diverse health populations to enable assistive care for all. The objective of this research is to develop human-in-the-loop cyber-physical systems that can model human behaviors and enable assistive interventions in sparse multi-device environments. This research will engender: (i) modeling human motion in sparse multi-device environments; (ii) learning motion-derived behavioral measures (verbal, physical, and psychological); (iii) a human-in-the-loop model that delivers interventions to the human and solicits their feedback when needed; and (iv) development and evaluation of the proposed techniques with target health populations. Our key idea is to develop novel techniques for learning coarse and fine-grained human motion in sparse multi-device environments, and infer physical, verbal, and psychological behaviors from human motion. This ultimately feeds into a human-in-the-loop CPS model to deliver the right interventions at the right time for target behavioral outcomes. The research outcomes from this work will be integrated into our comprehensive education plan and will influence pedagogy at the intersection of multiple disciplines. 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.