Enhancing Performance and User Engagement in Everyday Stress Monitoring: A Context-Aware Active Reinforcement Learning Approach

TL;DR

The paper tackles efficient, personalized stress monitoring in everyday life by integrating context-aware Deep Q-learning with smartphone-derived contextual data and smartwatch PPG signals. It advances a three-layer ZotCare architecture for offline data collection and online real-time EMA triggering, demonstrating substantial gains in labeling efficiency (up to 88% fewer EMAs offline) and stress detection performance (up to 21% improvement; 4% gain from context) and further achieving an 11% online improvement in F1 and a 10% personalized AUC-ROC boost. The work highlights the importance of real-time context, user responsiveness, and personalization in reducing user burden while improving label quality and model accuracy. These findings have practical implications for deploying real-time, user-centric stress monitoring systems in everyday environments, with potential applicability to other context-rich health sensing tasks.

Abstract

In today's fast-paced world, accurately monitoring stress levels is crucial. Sensor-based stress monitoring systems often need large datasets for training effective models. However, individual-specific models are necessary for personalized and interactive scenarios. Traditional methods like Ecological Momentary Assessments (EMAs) assess stress but struggle with efficient data collection without burdening users. The challenge is to timely send EMAs, especially during stress, balancing monitoring efficiency and user convenience. This paper introduces a novel context-aware active reinforcement learning (RL) algorithm for enhanced stress detection using Photoplethysmography (PPG) data from smartwatches and contextual data from smartphones. Our approach dynamically selects optimal times for deploying EMAs, utilizing the user's immediate context to maximize label accuracy and minimize intrusiveness. Initially, the study was executed in an offline environment to refine the label collection process, aiming to increase accuracy while reducing user burden. Later, we integrated a real-time label collection mechanism, transitioning to an online methodology. This shift resulted in an 11% improvement in stress detection efficiency. Incorporating contextual data improved model accuracy by 4%. Personalization studies indicated a 10% enhancement in AUC-ROC scores, demonstrating better stress level differentiation. This research marks a significant move towards personalized, context-driven real-time stress monitoring methods.

Figures (6)

• Figure 1: System Architecture - Offline Study
• Figure 2: Number of queries needed to reach a certain performance level during personalization.
• Figure 3: Presonalization Recall in Previous Work
• Figure 4: System Architecture - Online Study
• Figure 5: Distribution of Stress Labels