NUS Innovates AI-Powered Metahydrogel for Clinical-Grade Fatigue Detection

A research team led by Professor Ho Ghim Wei from the National University of Singapore's Department of Electrical and Computer Engineering has developed a metahydrogel platform integrated with AI-driven signal processing for enhanced fatigue detection. This groundbreaking technology promises to transform how fatigue and mental health conditions are diagnosed and monitored.

Addressing Current Diagnostic Limitations

This technology aims to address challenges in diagnosing fatigue and mental health conditions, which currently rely on subjective self-reported questionnaires. The inherent subjectivity of these methods often leads to inconsistencies and delays in diagnosis. Furthermore, existing wearable devices that track cardiovascular markers often experience signal degradation due to motion interference during everyday movement, compromising their accuracy and reliability.

Revolutionizing Wearable Accuracy

The NUS team's innovative system directly tackles these limitations. The new system is designed to suppress multiple sources of motion noise simultaneously, ensuring consistent data collection even during daily activities. It achieves remarkable accuracy, delivering an electrocardiograph (ECG) signal-to-noise ratio (SNR) of 37.36 dB and a blood pressure deviation of 3 mmHg during movement. Crucially, these accuracy levels meet ISO clinical-grade standards and exceed those of commercial trackers currently available, marking a significant leap forward in wearable health technology.

AI for Objective Fatigue Classification

Integrating this highly accurate sensor platform with artificial intelligence further enhances its capabilities. When combined with machine learning, the platform demonstrates 92 percent accuracy in classifying fatigue levels. This high level of precision allows for objective and reliable assessment of a user's fatigue state, moving beyond the limitations of self-reporting.

This development suggests a potential for objective, continuous mental health monitoring in various settings.

The findings from this pivotal research were officially published in Nature Sensors on March 24, 2026.