Nanoengineered wrist sensor detects driver fatigue through pulse wave analysis

Apr 20, 2026

A nanoengineered wrist-worn triboelectric sensor captures arterial pulse waves under real-world conditions and uses machine learning to classify driver fatigue with up to 98 percent accuracy.

(Nanowerk News) Researchers have developed a wrist-worn sensor that reads arterial pulse waves accurately enough to flag driver fatigue in real time, even under the imperfect contact conditions typical of everyday wearable use. The nanoengineered triboelectric device, paired with machine learning, classified fatigue-related states with accuracy reaching 98 percent in testing. The study, published in Microsystems & Nanoengineering (“Optimized stress transfer interfaces enabled wearable nano-electronics for fatigue driving monitoring”), was led by researchers at Xi’an Jiaotong-Liverpool University, Soochow University, and the University of Liverpool.

Key Findings

Monitoring pulse waves from the wrist offers a noninvasive route to tracking cardiovascular health and alertness, but the signal itself is weak. Strap pressure, uneven skin surfaces, and gaps between sensor and skin all distort readings. Previous triboelectric and mechanical sensors addressed some of these problems through surface microstructures, yet many still lost performance when pressed against the skin or failed to capture the full shape of each pulse wave. Device structures and working mechanisms of interfacial engineering-based triboelectric sensor. (Image: Reproduced from DOI:10.1038/s41378-025-01107-x, CC BY) (click on image to enlarge) The IETS addresses these issues at two separate interfaces. At the boundary between the sensor and the skin, an array of piezo-frustums fills microscopic gaps, creating additional pathways for stress transfer while simultaneously generating piezoelectric charges. At the triboelectric layer, mountain-like microstructures provide multiple stress-concentration points that keep the sensor responsive even under sustained pressure. This dual-interface architecture produces what the researchers describe as a mechano-electric coupling effect that amplifies the electrical signal from faint pulse beats. In quantitative terms, the device recorded a detection limit of 2 Pa, a response time of 70 milliseconds, and a detection range extending to 110 kPa. Under a sustained preload of 10 kPa, it resolved three distinct peaks in each pulse wave cycle, detail that simpler sensor structures could not maintain. These waveform features carry clinically relevant information, since changes in pulse shape can reflect shifts in arterial stiffness, blood pressure, and autonomic nervous system activity. The sensor was built into a smart wrist strap and connected via Bluetooth to a mobile application. The system extracted heart rate variability features from the pulse signal, converted them for analysis, and fed them into a one-dimensional convolutional neural network trained to distinguish fatigue-related states. In one subject, classification accuracy reached 98 percent. The combination of hardware precision and software analysis allowed the device to assess both cardiovascular condition and fatigue level from the same data stream. “This is the kind of wearable that does more than record a signal,” the study notes. “It keeps working when real life gets in the way — when skin is uneven, straps are tight, and pressure conditions shift. By preserving the fine structure of pulse waves, it moves fatigue and cardiovascular monitoring closer to the moment when an alert can still make a difference.” Beyond fatigue detection, the researchers showed that their sensing platform could monitor several additional driver-related signals, including eye blinks, yawning, pedal use, seat occupancy, and whether a seat belt was fastened. This suggests a potential path toward a more comprehensive wearable safety system that combines physiological data with behavioral indicators. Driver fatigue and sudden cardiovascular events are among the leading causes of traffic fatalities. Current monitoring approaches often rely on camera-based systems or vehicle dynamics, which detect drowsiness only after driving performance has already deteriorated. A wrist-worn sensor that reads physiological markers directly could provide earlier warnings. The IETS design addresses a key practical barrier by showing that nanoengineered interface structures can preserve signal quality under the variable contact conditions that real-world wearable devices inevitably face.