The work, led by researchers at Vellore Institute of Technology in Chennai, India, addresses a persistent weakness of existing FBG-based cardiorespiratory sensors. Most reported systems monitor heart rate (HR) and respiratory rate (RR) from chest-wall motion using elastic straps, a configuration that is non-invasive but vulnerable to wear and tear over long-term use, and in which stretching the sensor to maintain skin contact can lead to fiber breakage. Finger plethysmography offers an alternative route: it is a well-established technique that records the arterial pulse waveform by detecting blood-volume changes at the fingertip during each cardiac cycle. Because the pulse waveform is modulated by respiration — through baseline wander, amplitude modulation and frequency modulation — the respiratory signal can be extracted from it with suitable algorithms. Previous finger-based FBG studies, however, relied on bare gratings mounted in silicone diaphragms or 3D-printed holders, focused only on cardiac parameters, and lacked a wearable design adaptable to different individuals. The new sensor, the authors state, is the first to demonstrate simultaneous HR and RR measurement from finger plethysmography using a silicone-embedded wearable FBG.

The sensing element is a 5 mm FBG with a Bragg wavelength of 1540 nm and 80% reflectivity, supplied by RRCAT Indore. An FBG consists of a periodic modulation of the refractive index inside the optical fiber core; when broadband light travels through the grating, a narrow wavelength — the Bragg wavelength — is reflected back. Strain and temperature change both the effective refractive index and the grating period, shifting this wavelength, which forms the basis of the measurement. To protect the inherently fragile grating and make it wearable, the FBG was embedded in Dragon Skin 10 Medium, a soft, biocompatible silicone chosen for its flexibility, ease of fabrication and skin-like texture. The sensing patch was designed as a 20 mm diameter, 2 mm thick disc — circular geometry chosen to match average human fingertip dimensions and to conform to the finger without modification for different users. Fabrication used a customized 3D-printed mold, with a five-hour curing period before the sensor was removed. Data were acquired by a Sentea DM-4120 optical interrogator at 1 kHz for two-minute recordings, stored on a PC via Ethernet.

Because encapsulation alters an FBG’s sensitivity, the team first characterized the embedded sensor metrologically. Strain calibration was performed on a small longitudinal beam instrumented with a reference strain gauge and indicator, with 1 kg weights loaded in increments up to a maximum of 4 kg across three trials, data acquired at 100 Hz and uncertainty evaluated with a t-student distribution at 95% confidence. The strain response was not linear — a quadratic fit yielded an R² of 0.98 versus 0.85 for a linear fit — and the average strain sensitivity was 0.11 ± 0.03 pm/με, roughly an order of magnitude lower than the 1.2 pm/με typical of a bare FBG. The researchers attribute this attenuation to the highly elastic silicone, which limits efficient strain transfer to the embedded grating. Temperature characterization in a mini-tabletop incubator over a 20 °C range produced a linear response with a slope of 10.7 pm/°C, essentially similar to a bare FBG; although the silicone’s thermal expansion coefficient is much higher than silica’s, its low Young’s modulus prevents that deformation from being transmitted, leaving the thermo-optic effect as the dominant contributor. Force testing from 0 to 5 N, applied perpendicular to the patch surface over three loading-unloading cycles, told a different story: sensitivity rose sharply to 30.1 ± 2.2 pm/N with minimal hysteresis of 2.52%, which the team ascribes to the flexible silicone deforming under load and the circular geometry promoting uniform strain transmission. These force results compare favorably with other published sensors based on the same DS-10 matrix.

With the sensor’s behavior established, the team turned to placement. Two volunteers were tested under four conditions — right index and middle fingers, each with palm up and palm down — and the beat-to-beat pulse interval was analyzed. Palm-down conditions showed slightly higher variability because the finger touched the table surface, introducing additional wavelength shifts, while keeping the sensed finger elevated induced involuntary movements that destabilized the output. The most stable, uniform measurement came from the middle finger with the palm facing up. This aligns with prior clinical observations that the middle finger shows higher perfusion, since unlike the index finger it receives blood through both the ulnar and radial arteries. That configuration was adopted for all subsequent trials. Before each recording session, volunteers rested for five minutes, and the sensor was then placed on the skin for three minutes to reach thermal equilibrium with body temperature — a protocol justified by separate experiments showing that the thermal baseline drifts slowly and saturates after roughly 160 seconds, making the temperature component easy to separate from the fast pulsatile signal. The sensor was secured above the distal interphalangeal joint, aligned parallel to the finger, with transparent adhesive tape wrapped around the finger; care was taken to avoid excessive tightness that could restrict blood circulation or add spurious strain.

Signal processing proceeded in MATLAB. The raw wavelength-shift signal was min-max normalized to account for physiological differences in pulse amplitude between volunteers, then filtered with a fourth-order Chebyshev type-II filter. A 0.1–5 Hz band preserved both cardiac and respiratory components, while a narrower 0.5–5 Hz band retrieved the arterial waveform alone, and interquartile-range calculations removed motion-artifact outliers. To test robustness, a volunteer performed deliberate hand movements during a trial; the raw signal showed baseline variations and abrupt shifts from motion artifacts, but the filter suppressed these while preserving the pulse. The processed waveforms displayed the expected arterial morphology: a well-defined systolic peak, a smaller diastolic peak, and an elevated dicrotic notch. Repeatability was assessed by recording the same volunteer on three consecutive days; the waveforms remained uniform with a consistent morphology, differing only in pulse duration as heart rate varied between days, while morphological differences between two volunteers further supported the sensor’s ability to capture individually distinct pulse shapes.

Beyond heart rate, the sensor supports second-derivative analysis of the pulse waveform, known as the acceleration plethysmograph (APG), which represents the acceleration of blood flow and is used to evaluate vascular aging through wave-amplitude ratios such as b/a, c/a, d/a and e/a, plus the aging index (b−c−d−e)/a. For two volunteers aged 18 and 26, the computed ratios fell within the healthy range with small standard deviations; the older volunteer showed increased magnitude of b/a and the aging index with decreases in the remaining ratios, consistent with the increased arterial stiffness expected with age and matching values reported in earlier FBG studies. Signal quality was quantified by signal-to-noise ratio, averaging 39.3 ± 5.9 dB across volunteers with minimal deviation between individuals, indicating clean, repeatable recordings.

Validation against a commercial pulse oximeter — worn on the right index finger while the FBG sensor sat on the right middle finger — showed strong agreement. Peak detection of the processed waveform yielded heart rates within about ±1.4 beats per minute of the reference across volunteers (measured values including 84, 86, 80 and 79 beats/min). For respiratory rate, the team computed the fast Fourier transform of the pulse waveform, which revealed two distinct spectral peaks: a small one in the 0.1–0.5 Hz respiratory band and a larger one around 1–2 Hz corresponding to the cardiac signal. A 0.5 Hz low-pass Butterworth filter extracted the baseline-modulated respiratory waveform, from which peak intervals gave RR values (15–17 breaths/min) averaging within ±1.6 breaths/min of the reference. Bland-Altman analysis found a mean bias of −0.65 beats/min for HR, with 95% limits of agreement of 3.77 and −5.06 beats/min, mean absolute error of 1.9 beats/min and RMSE of 2.2 beats/min. For RR, the bias was 0.39 breaths/min with limits of agreement of 5.79 and −5.02 breaths/min, MAE of 2.2 breaths/min and RMSE of 2.8 breaths/min. Both mean biases sat close to zero, with most data points inside the limits of agreement, although the RR limits were notably wider than the HR limits.

The study’s limitations are acknowledged by the authors themselves. The reference pulse oximeter did not offer continuous monitoring, which would have enabled stronger validation, and the adhesive tape securing the patch, while simple, requires frequent replacement and can wear off with prolonged use. The volunteer sample was small, and the thermal contribution to the wavelength shift, though slow and separable, was neglected on the basis of controlled conditions and short experiment durations. The team plans to fabricate a 3D-printed sensor holder wearable across different anatomies, to validate against a gold-standard reference with a larger sample, and to explore cuffless, continuous blood pressure estimation using the same finger-plethysmography platform.

Even so, the results mark a meaningful step for photonic wearables. Compared with prior FBG pulse studies that measured only cardiac parameters such as pulse transit time, augmentation index or heart rate at the fingertip, radial or carotid arteries, the silicone-embedded circular patch delivers HR, RR and APG simultaneously from a single, comfortable fingertip site. Because FBGs are immune to electromagnetic interference, chemically stable and multiplexable, such sensors could eventually find use in clinical settings — including MRI environments, where conventional electronic sensors are restricted — while the soft silicone construction points toward the comfort and wearability that personalized, continuous cardiorespiratory monitoring will demand.

Traditional organic field-effect transistors have shown promising capabilities for biosensing applications; however, they often suffer from signal distortions due to environmental factors such as bending, stretching, moisture, and temperature fluctuations. These issues can lead to unexpected signal artefacts and drifts, complicating the tasks of data interpretation and reliability in biosensing applications. Researchers have addressed these challenges head-on to develop a new breed of biosensors that maintain signal integrity even under varied physical stresses and environmental conditions.

The novel biosensors utilize capacitive coupling combined with interference signal subtraction techniques. This method employs two extended gates that are functionally differentiated through the use of specific target and reference bioreceptors. Such a setup is invaluable, as it not only enhances the sensitivity of the sensors but also drastically reduces signal distortion—by an impressive two orders of magnitude compared to standard organic field-effect transistors that lack this configuration. This stark improvement in performance is particularly notable when faced with various stressors such as bias stress instability, uniaxial strain (up to 100%), and compression (up to 50 mN), as well as temperature changes ranging from 25 to 40 degrees Celsius.

Applications of these advanced biosensors are vast, with initial testing focusing on aptamer-based sensing for cortisol, enzyme-based sensing for glucose, and potentiometric sensing for sodium ions utilizing ion-selective membranes. Each of these applications presents unique challenges, including the need for specificity, sensitivity, and rapid response times. The ability of these biosensors to deliver reliable data under varying conditions enhances their suitability for personal health applications, particularly in monitoring stress-related biomarkers and metabolic parameters.

Moreover, this research culminates in the integration of a hybrid wearable system that consolidates soft sensors and a flexible printed circuit board. This innovative design allows for seamless wireless communication with smartphone applications, bridging the gap between complex biosensing technology and user accessibility. As these devices communicate with smartphones, users can easily access real-time health metrics, supporting informed decision-making regarding personal wellness and health management.

In practical terms, one of the applications of this cutting-edge technology has been demonstrated through cortisol sensing from human sweat during acute stress events. Stress hormones like cortisol play a critical role in numerous physiological processes, and the ability to monitor their levels non-invasively offers tremendous potential benefits for both clinical and everyday health assessments. This capability not only paves the way for timely detection of stress responses but also fosters greater mindfulness and well-being practices among users.

The implications of this research extend beyond simple monitoring; they touch upon the broader objectives of preventative healthcare. Accessing physiological data continuously allows for the early identification of health issues, potentially leading to preemptive measures before conditions escalate. This aligns perfectly with the growing emphasis placed on health management and the continuous monitoring of physiological states as a pathway to improved outcomes.

As healthcare technology continues to progress, integrating biosensor technology seamlessly into our daily lives will increasingly become a reality. The incorporation of skin-like, stretchable biosensors that remain functionally robust under mechanical stress represents a critical step in this ongoing revolution, setting the stage for future innovations that could redefine personal and remote healthcare.

In conclusion, the development of skin-like drift-free biosensors reliant on stretchable diode-connected organic field-effect transistors marks a significant milestone in wearable technology. These devices have been meticulously designed to withstand environmental challenges while providing accurate and reliable biosensing capabilities. As the world continues to embrace digital health and personalized medicine, such advancements will crucially shape the landscape of future health monitoring solutions. The potential to transform how individuals interact with their health data and respond to physiological cues underscores the transformative power of modern biosensor technology.

As this research gains momentum, it is anticipated that these biosensors will inspire further innovations within the field, laying the groundwork for advanced, multifunctional wearable devices that seamlessly integrate into everyday life. Ultimately, as technology merges with biology, the future of health monitoring will be characterized by enhanced precision, proactive healthcare management, and improved quality of life for individuals across the globe.

Subject of Research: Development of skin-like drift-free biosensors utilizing stretchable diode-connected organic field-effect transistors.

Article Title: Skin-like drift-free biosensors with stretchable diode-connected organic field-effect transistors.

Article References:

Zhao, C., Park, J., Maulà, D. et al. Skin-like drift-free biosensors with stretchable diode-connected organic field-effect transistors.
Nat Electron (2025). https://doi.org/10.1038/s41928-025-01465-4

Image Credits: AI Generated

DOI:

Keywords: Wearable technology, biosensors, organic field-effect transistors, health monitoring, cortisol sensing, glucose sensing, sodium ion sensing, personal health applications, real-time monitoring, digital health.