ERI: data-centric AI for wearable biosensors to enhance learning

The National Science Foundation Division of Electrical, Communications and Cyber Systems awarded the University of San Diego $199,999 on May 1, 2026, under the Engineering Research Initiation program (CFDA 47.041) to develop an artificial intelligence framework integrating wearable biosensors for enhanced classroom learning environments.

The project develops machine learning models to analyze physiological signals—heart rate, electrodermal activity, and oxygen levels—collected from wearable devices worn by volunteer students during classroom instruction. The work addresses signal quality challenges through techniques including cleaning, feature engineering, and data augmentation to overcome issues associated with limited and noisy biosignal datasets. AI models will identify temporal patterns in physiological data corresponding to student engagement and stress during learning activities. Outcomes include new methods for analyzing classroom biosignals, validated datasets for educational research, and a prototype application providing aggregated feedback to instructors, enabling data-driven adjustments to course pacing, instructional strategies, and classroom activities.

The award covers the period May 1, 2026, through April 30, 2028, with place of performance in San Diego, California. The University of San Diego, a private nonprofit Catholic university, is the recipient. All work proceeds under approved ethical and privacy safeguards for student participants.

Abstract

This project integrates wearable biosensors with an artificial intelligence framework to improve understanding of student engagement and stress in classroom environments. Physiological signals such as heart rate, electrodermal activity, and oxygen levels provide insights into cognitive load and emotional states during learning activities. Current approaches rarely translate these signals into practical guidance for instructors. The project investigates how biosignal data collected from wearable devices can support more responsive and effective teaching practices. By focusing on improving the reliability and interpretation of physiological data, the work seeks to provide instructors with meaningful feedback that can help them adjust course pacing, instructional strategies, and classroom activities. The research develops an AI framework for multimodal physiological signals from wearable devices in real classroom settings. Volunteer students will wear selected devices during instructional sessions to collect biosignals under approved ethical and privacy safeguards. The project will investigate techniques for improving data quality, including signal cleaning, feature engineering, and data augmentation to address challenges associated with limited and noisy biosignal datasets. AI/machine learning models will be developed to analyze temporal patterns in the physiological signals and identify changes associated with engagement and stress during classroom activities. The resulting models will support the development of a prototype application that provides aggregated feedback to instructors, enabling informed adjustments to teaching strategies. The outcomes include new methods for analyzing classroom biosignals, validated datasets for educational research, and tools that demonstrate how wearable sensing and data-driven analysis can support improved learning environments. 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.- subawards are not planned for this award.