Funding Source
National Science Foundation, 2442593
Project Period
7/1/2025 – 6/30/2030
Principal Investigator
Varun Mishra, PhD (Northeastern University)
Other Project Staff
Project Summary
This project aims to develop intelligent, context-aware systems that improve how digital health interventions are delivered through smartphones and wearable devices. By modeling a person’s external and internal contextual state—such as activity, location, and physiological arousal—the system can determine the most effective time, type, and delivery method for support. The goal is to enhance long-term engagement with digital interventions by making them more timely, relevant, and less burdensome. The project integrates mobile sensing, human-centered computing, and machine learning—specifically, probabilistic modeling and reinforcement learning—to first define a robust framework to model and forecast contextual states and understand when individuals are most receptive to digital health support. The resulting algorithms and tools will be evaluated in real-world settings and shared as open-source resources for researchers and intervention designers working in digital health.
Public Health Relevance
Digital interventions have the potential to support large-scale, cost-effective care for mental and behavioral health conditions, such as stress, anxiety, and physical inactivity. However, their long-term effectiveness is often limited by poor user engagement. This project directly addresses that barrier by developing systems that personalize when, how, and what support is delivered, based on a person’s real-time context. These advances will help make digital health interventions more timely, relevant, and less burdensome—ultimately improving accessibility and health outcomes across diverse populations. The tools and methods developed in this work will support behavioral scientists and health professionals in designing smarter, more effective digital interventions for public health applications.