AI-Powered Wearable Sensor Detects Fatigue and Stress via Body Signals

Researchers at the National University of Singapore (NUS) have developed a wearable biosensor platform capable of objective and continuous monitoring of fatigue and stress, even while the user is moving. The system, which combines a novel hydrogel material with AI-based signal processing, aims to provide a real-time alternative to traditional diagnostic methods.

According to reports published on March 30, 2026, the technology addresses a significant gap in how fatigue and burnout are currently assessed. Most existing diagnostic approaches rely on subjective questionnaires, which researchers note are not suitable for continuous or real-time assessment. This is particularly critical in professions that require sustained alertness, where fatigue and burnout pose substantial risks.

Overcoming Motion Artefacts in Biosensors

A primary challenge for wearable devices is the delivery of reliable data during daily activities. Standard biosensors often struggle with motion artefacts, which are distortions caused by body movement, muscle activity, and physiological interference. These artefacts frequently overwhelm the faint signals required for accurate cardiovascular monitoring, such as blood pressure measurements and electrocardiograms (ECG).

While existing solutions often address only limited noise sources, the NUS team developed a system to suppress multiple types of interference simultaneously. Rather than relying exclusively on software to clean up noisy data after it has been collected, the researchers focused on filtering noise at the sensor-body interface itself.

The Metahydrogel Platform and AI Integration

At the center of this technology is a metahydrogel artefact-mitigating platform (MAP). This platform utilizes a soft, skin-like hydrogel sensor that demonstrates superior performance during movement, where reducing signal noise is most critical.

The system integrates this advanced material with AI-driven signal processing to decode body signals. By improving the clarity of the signal at the source, the sensor can more reliably detect key ECG peaks. This technical improvement has resulted in a significant increase in peak-detection accuracy, rising from 52% to 93%.

This increased accuracy allows the system to better distinguish between fatigue-related patterns and normal heart rhythms, providing a more precise reading of the user’s physiological state while they are on the move.

Clinical Significance and Research Findings

The ability to objectively monitor cardiovascular signals in real-time allows for a more data-driven understanding of stress and exhaustion. By utilizing the metahydrogel platform, the sensor can maintain high performance under real-world motion conditions that typically render other wearables inaccurate.

The findings regarding this wearable biosensor platform were published in the journal Nature Sensors.

The development of this sensor represents a shift toward objective monitoring for mental and physical wellbeing, moving away from the limitations of self-reported data. By tackling the problem of signal interference at the physical interface between the sensor and the skin, the NUS researchers have created a more stable method for decoding the body’s stress and fatigue signals during active use.