Description:
BRIEF DESCRIPTION
A contactless smart ring that optically tracks stress-related physiology and uses on-device AI for continuous stress monitoring.
BACKGROUND
Continuous stress monitoring could support personal wellness, workplace safety, and clinical care, but conventional electrodermal activity (EDA) sensors depend on skin-contact electrodes and conductive gels. Over extended wear, gel drying can cause baseline drift and declining signal quality, while motion and thermoregulatory sweating can confound stress measurements. Existing approaches may also rely on a single physiological signal or external processing. A discreet wearable that can maintain signal quality during long-term use, distinguish stress from common confounders, and process data locally could enable more practical everyday stress monitoring.
INNOVATION
USC researchers have developed an optodermal activity (ODA) ring that uses light to monitor stress-related sweat gland activity without gel-based electrodes. The optical platform also captures cardiovascular signals and combines them with temperature and motion measurements for multimodal stress assessment. In testing, ODA measurements showed an average 86% correlation with EDA, while multimodal classification reached 92% accuracy for short-term stress vs. relaxation detection and 79% after 12 hours. A lightweight neural network can run directly on the ring, enabling real-time classification with reduced dependence on continuous raw-data transmission.
ADVANTAGES
- Contactless, gel-free sensing: avoids gel drying and electrode-contact limitations that can degrade conventional EDA measurements during extended wear
- Improved long-term stress classification: multimodal ODA-ring data achieved 79% binary accuracy after 12 hours versus 57% for gel-based EDA
- Distinguishes common confounders: four-class classification of stress, relaxation, motion, and thermal sweating reached 100% accuracy in short-term testing and 89.72% after 12 hours
- On-device intelligence: edge-AI inference achieved 87.53% four-state classification accuracy and can reduce latency, wireless data transmission, power use, and exposure of raw physiological data
STAGE OF DEVELOPMENT
- Functional ring prototype developed
- Proof-of-concept studies conducted in healthy volunteers