Roughly one in three workers in Singapore report burnout—among the highest rates worldwide. Burnout and chronic fatigue impose significant economic burdens and create serious risks in jobs that require constant alertness. However, current methods for diagnosing fatigue and related mental health issues rely mainly on self-reported questionnaires, which are subjective, sporadic, and not well suited for real-time assessment.
Wearable devices could bridge this gap by continuously monitoring cardiovascular signals tied to the autonomic nervous system, but their accuracy drops sharply during everyday movement. Motion artifacts from muscle activity, body movement, and physiological interference overwhelm the faint heart and blood pressure signals, while existing solutions usually target only a single noise source or limited frequency range.
A team led by Professor Ho Ghim Wei at the National University of Singapore, with Research Fellow Dr. Tian Guo as first author, developed a metahydrogel platform combined with AI-driven signal processing that simultaneously filters multiple sources of motion noise.
The system achieves an electrocardiograph (ECG) signal-to-noise ratio of 37.36 dB and blood pressure error as low as 3 mmHg during movement—meeting ISO clinical standards and surpassing current commercial trackers. Paired with machine learning, it can classify fatigue levels with 92% accuracy, enabling objective, continuous mental health monitoring in real-world conditions.
Eliminating Noise At Its Source
Instead of relying only on software to clean noisy data, the team addressed the issue at the sensor–skin interface. Their metahydrogel artifact-mitigating platform (MAP) integrates two filtering strategies into a single material.
Within the hydrogel, nanoparticles self-assemble into periodic bands that scatter and absorb mechanical vibrations, similar to soundproofing that traps and blocks noise across specific frequency ranges.
At the same time, a biocompatible glycerol–water electrolyte regulates ion movement through the gel, allowing low-frequency heart signals (below 30 Hz) to pass while suppressing higher-frequency muscle noise. A machine-learning denoising algorithm then removes any remaining irregular noise while preserving key physiological signals.
The platform is soft enough to match the mechanical properties of biological tissue, breathable with a water vapor transmission rate higher than that of human skin, and resilient under repeated stretching.
By pairing enhanced hardware with intelligent algorithms, the system significantly improves ECG clarity, raising signal quality from 5.19 dB to 37.36 dB. This cleaner signal enables more reliable detection of key ECG peaks, increasing peak-detection accuracy from 52% to 93% and making it easier to distinguish fatigue-related patterns from normal heart rhythms.
“Compared to current commercial devices, our metahydrogel platform delivers superior performance, especially during movement when artifact suppression is essential. Typical smartwatches achieve ECG signal-to-noise ratios of 10–20 dB, which can drop by about 40% during motion due to artifacts and unstable contact. Our system maintains around 37 dB during everyday activities,” said Dr. Tian.
From Reliable Signals to Decoding Mental States
Because fatigue affects the autonomic nervous system, it leaves detectable signs in heart rate variability, blood pressure patterns, and ECG features—but only if these signals are captured clearly during everyday activities.
The team developed a fully integrated, flexible MAP wearable with wireless connectivity and used it to monitor participants over several days, including simulated driving tasks designed to induce fatigue.
With high-quality cardiovascular data from the hydrogel sensor, a deep-learning system classified fatigue levels with 92% accuracy, compared to 64% using data collected without MAP. The system also met ISO 81060-2 gold-standard requirements for blood pressure monitoring.
Beyond fatigue detection, MAP effectively reduced artifacts in multiple biosignal types—heart sounds, respiratory sounds, voice, brain waves, and eye movements—demonstrating its potential for broader neurophysiological and mental health monitoring.
Advancing Mental Health Monitoring In Real-World Settings
The team spent roughly four years developing the foundational sensing technologies before arriving at the metahydrogel concept about two and a half years ago.
Designing and fabricating the platform took around a year, during which they created a library of metahydrogels with different material systems to target noise across various frequency ranges. An additional year was devoted to system integration and application validation, including exploring its potential for mental health monitoring.
“We aim to collaborate closely with mental health clinicians to understand which physiological signals are most relevant in real-world settings and what level of accuracy is needed for clinical use. Their expertise can help us connect the data to meaningful pathological indicators,” said Prof Ho.
On the industry side, the team is seeking partners to enhance device consistency and scalability.
“Our current material synthesis and system fabrication remain largely lab-based. Partnering with industry will help us optimize manufacturing processes and move the platform toward practical, product-ready implementation,” she added.
Read the original article on: Medical Xpress
Read more: Scientists engineer a kidney that can match any blood type