A Wearable Bracelet for Simultaneous Monitoring of Transcutaneous Carbon Dioxide and Pulse Rates

Daily monitoring of psychological parameters, encompassing chemical and physical signals, has become increasingly valuable for the management of chronic diseases. Among various vital signs, carbon dioxide exchange and pulse rate (PR) serve as critical indicators for the comprehensive assessment of the cardiopulmonary system. However, simultaneous monitoring of these chemical and physical signals in a nonintrusive manner is challenging. This study presents a wearable bracelet that enables simultaneous monitoring of transcutaneous carbon dioxide (TcCO2) and PR using a nanomaterials‐modified lensless CMOS imager. The proposed bracelet exhibits exceptional sensitivity, selectivity, and anti‐humidity interference in TcCO2 detection, and achieves reliable tracking of TcCO2 and PR fluctuations induced by daily activities, demonstrating great potential in the comprehensive assessment of cardiopulmonary system. In addition, the proposed compact imaging approach is multiplexed and extensible, providing a promising sensing platform for the future development of multifunctional wearable devices.


Introduction
Cardiopulmonary diseases are the most widespread diseases affecting human well-being.These diseases encompass a wide range of disorders that affect the heart and lungs, including respiratory diseases, cardiovascular diseases, and their associated complications.Patient populations with underlying respiratory diseases, such as chronic obstructive pulmonary disease [1] and severe acute asthma, [2] are prone to experiencing abnormal DOI: 10.1002/aelm.202300760carbon dioxide exchange.These conditions can lead to impaired lung function and compromised gas exchange, resulting in abnormal levels of carbon dioxide in the artery (PaCO 2 ).Therefore, it is imperative to ensure that these critical patients receive continuous monitoring of their PaCO 2 levels during their transition to a home-care environment.[6] To overcome these limitations, noninvasive measurement of the partial pressure of TcCO 2 (PtcCO 2 ), which reflects the diffusion of carbon dioxide through the skin, has emerged as a surrogate method.[12] The requirement for heating poses challenges for long-term continuous monitoring, including patient discomfort and increased power consumption.[15] While significant progress has been made, current wearable TcCO 2 sensors are limited to monitoring a single signal, smart devices that integrate more psychological parameters for comprehensive health monitoring are still to be explored.
Cardiovascular disease, encompassing conditions such as coronary artery disease, myocardial infarction, and cardiac arrhythmia, remains a leading cause of mortality worldwide.Realtime monitoring of cardiac parameters holds immense importance for effective health management of patients with cardiovascular disease. [16,17]Detailed analysis of heart function often involves complex sensory systems and high costs, such as electrocardiography (ECG).Pulse rate, defined as the number of pulse beats observed within a minute, is a commonly used clinical parameter to assess the physiological state of individuals. [18]During daily life, PR serves as a vital sign, and any abnormalities in its values may indicate significant clinical events.Therefore, monitoring and being aware of one's pulse rate is essential for evaluating their overall health condition, which can potentially contribute to the reduction of cardiovascular risks.Among the available methods for pulse rate detection, video-based monitoring approaches have gained attention due to their affordability and suitability for nonclinical purposes. [19,20]Remote photoplethysmography (rPPG) is a technique that exploits the differential light absorption properties of blood cells and surrounding tissues.As a result, any changes in blood flow are manifested in the light that is transmitted or reflected from adjacent areas.Through the analysis of the pulse rate derived from rPPG, it becomes feasible to noninvasively and cost-effectively detect.26][27] This focus has led to the popularity of wearable devices, particularly sports bracelets like Apple Watch, Fitbit, and Garmin, which are widely favored for their affordability, user-friendly interface, and minimal activity restrictions.[30][31] However, despite the advancements in wearable technology, the integration of simultaneous measurement of TcCO 2 and PR within a single smart device has not yet been achieved.
In this work, we propose a wearable bracelet based on a nanomaterials-modified lensless CMOS imager for simultaneous monitoring of TcCO 2 and PR.The concept of lensless imag-ing, previously introduced in our previous work, [32][33][34] is applied in this wearable bracelet.The edge area of the CMOS imager is coated with colorimetric CO 2 sensing units, modified with silica nanoparticles (SiO 2 NPs), for TcCO 2 sensing.Meanwhile, the central part of the imaging area remains unmodified to capture imaging signals for PR analysis.The developed wearable TcCO 2 sensor exhibits a well-behaved response curve to CO 2 , demonstrating a linear response to CO 2 concentration.This wearable device holds significant potential in the healthcare industry as a real-time individual monitoring solution.In the future, the functions of this device can be expanded by incorporating additional sensors on the same image, enabling the monitoring of various other physiological parameters.This expansion would enhance the overall utility and scope of the device, providing comprehensive and valuable information for healthcare professionals.

Principle of the Wearable Bracelet for TcCO 2 and PR Monitoring
Figure 1 provides a conceptual visualization of the wearable bracelet for TcCO 2 and PR monitoring.The wearable bracelet, as shown in Figure 1a and Figure S1 (Supporting Information), consists of a customized metal body that accommodates the lensless CMOS sensor (Figure 1b).The metal body is designed to be in contact with the skin, creating a miniaturized chamber with a volume of 0.5 mL for volatile gas to diffuse in.The working area of the lensless CMOS imaging chip can be divided into two independent regions, enabling the simultaneous incorporation of For TcCO 2 detection, the edge region of the imaging area was coated with colorimetric CO 2 sensing units modified with SiO 2 NPs.These CO 2 sensing units consist of a pH-sensitive dye (mcresol purple) for CO 2 sensing (the color change mechanism is shown in Figure S3, Supporting Information), glycine for CO 2 molecule adsorption, and porous SiO 2 NPs to enhance the specific surface area of the contact interface.As depicted in Figure 1c, TcCO 2 molecules are emitted from the skin and directly diffuse into the miniaturized gas chamber.To ensure air-tight sealing and prevent gas leakage, the top of the gas chamber is coated with a layer of silicone.The diffusion of CO 2 molecules in the sensing site, combined with the subsequent colorimetric chemical reaction, triggers a noticeable color transition in the porous sensing unit.(Figure S4, Supporting Information) Once equilibrium is established, the grey value of the sensing unit becomes correlated with the concentration of TcCO 2 .Since the colorimetric sensing unit is in direct contact with the image pixels, the color responses can be accurately resolved by the lensless CMOS imagers.
In addition, to ensure accurate gas detection for real-time TcCO 2 monitoring, it is crucial to address the potential interference caused by diffused water molecules through the skin.These molecules can participate in colorimetric chemical reactions through electrolysis, thereby affecting the accuracy of the measurements.In this study, the interference from skin moisture is eliminated by utilizing a hydrophobic PDMS membrane.(Figure 1c) PDMS membranes are well known for their low permeability to water vapor in chemical sensors. [35,36]By applying this protective barrier on the surface of the modified CMOS imager, the interference caused by skin moisture is effectively eliminated.
Inspired by the imaging-based heart rate measurement technology, [37,38] we employed imaging-based pulse rate detection in our wearable bracelet.This approach is based on the fundamental principle that heart contractions induce volume pulsations in the microvessels present on the skin surface, thereby influencing the penetration and reflection of the light source.To capture the pulse waveform, we focused on monitoring changes in the optical properties of a specific area of the wrist skin that is influenced by the pulsating blood content.As illustrated in Figure 1d, we utilized the central unmodified imaging area of a CMOS imager to capture image information of the wrist arteries.Subsequently, changes in pixel intensity within the red channel were analyzed to obtain the pulse waveform.This imaging-based approach enables us to track variations in blood flow and derive accurate PR measurements.

Using SiO 2 NPs to Enhance Sensitivity
The primary challenge for transdermal gas sensing is the low concentration of analytes and a limited number of molecules released from the skin within a tiny confined space, which puts a high demand on the sensitivity of the sensor.In our study, we enhance the sensitivity of CO 2 detection by incorporating SiO 2 NPs into the sensing solutions.Initially, the coating of mcresol purple on the CMOS imager resulted in low sensitivity and a weak response, which can be attributed to the limited surface-to-volume ratio of the nonporous sensing material layer on microlenses (Figure 2a).To address this issue, we introduced SiO 2 NPs (133 mg mL −1 ) into the sensing unit, as depicted in Figure 2b.This modification resulted in a substantial increase in the response to CO 2 (500 ppm) by a factor of five, as shown in Figure 2c.The significant enhancement can be attributed to the porous structure of the SiO 2 NPs, which provide a larger specific surface area for increased interaction with CO 2 molecules.Furthermore, the presence of SiO 2 NPs introduces a light scattering effect, prolonging the optical path within the sensing unit matrix.This leads to increased light absorbance, as per the Beer-Lambert law, resulting in a more pronounced color change.
The gas response signal is significantly influenced by both the thickness and size of the sensing unit, as they directly impact the light absorption properties.The thickness of the sensing unit was controlled by adjusting the inkjet time, while the size was manipulated by varying the inkjet voltage.As shown in Figure S5a (Supporting Information), it was observed that a moderate size (120 μm) of the sensing unit is advantageous for light transmission, resulting in a larger response signal.Additionally, the sensing response value increased with an increase of the inkjet voltage (Figure S5b, Supporting Information), this can be attributed to the fact that thicker sensing material layers lead to increased light absorption, according to the Beer-Lambert law.

CO 2 Concentration Calibration
The CO 2 response curve was shown in Figure S6 (Supporting Information), the CO 2 response curves showed an initial sharp increase followed by reaching plateaus.It was observed that the plateau response was directly proportional to the concentrations of CO 2 diffusing into the sensor within the concentration range of 1000-3000 ppm, as shown in Figure 2d.The calculated limit of detection (LOD) was 118 ppm.This indicates that the sensing unit possesses the required sensitivity and dynamic range to effectively support human TcCO 2 monitoring, which fluctuates between 1200 and 2500 ppm in normal physiological states. [15]

Using PDMS Membrane to Mitigate the Humidity Interference
Another significant challenge in TcCO 2 detection is the interference caused by skin humidity.It is crucial to implement antihumidity measurements to ensure reliable results.PDMS membrane is applied as an ideal material for CO 2 sensing due to its low permeability to water and high permeability to CO 2 .To evaluate the effectiveness of PDMS in CO 2 sensing, we compared the response of a PDMS membrane to CO 2 before and after coating.The response signal and response speed remained unaffected, indicating that PDMS can effectively and rapidly permeate CO 2 gas molecules (Figure 3a).Furthermore, as depicted in Figure 3b, the response of the colorimetric sensing unit to CO 2 gas diluted by dry air and wet air (humidity = 100%) was comparable.This suggests that the PDMS membrane effectively mitigates the interference caused by humidity in CO 2 detection.

Selectivity of CO 2 Detection
It is essential for the sensor to exhibit selectivity in order to specifically detect CO 2 molecules amidst other potential interfering substances.This selectivity ensures that the measurements obtained are accurate and not influenced by other tran-scutaneous volatile molecules.In our study, the fabricated sensor demonstrated remarkable selectivity for CO 2 sensing.The response signal of the colorimetric sensing unit in the presence of 0.3% CO 2 was significantly higher compared to the responses observed in the presence of pure water vapor, 1% ammonia, and 1% acetone (Figure 3c).These volatile molecules are commonly released from human skin and can potentially interfere with the CO 2 measurements.Moreover, Figure S7 (Supporting Information) presents a comparison of sensor response to pure CO 2 (2000 ppm) and mixed CO 2 (2000 ppm CO 2 , 1%  ammonia, 1% acetone with 100% humidity), effectively simulating the complex gas mixtures encountered in real gas exchange scenarios with body tissues.The results reveal that the presence of mixed interfering gases has no significant impact on the CO 2 detection.The high selectivity of the sensor indicates its ability to discriminate and specifically detect TcCO 2 molecules, thereby ensuring the accuracy and reliability of the obtained measurements.

Analytical Performance of the Wearable Bracelet for PR Detection
When a light source illuminates the skin area of the wrist artery, the CMOS imager effectively captures the variations in blood volume resulting from the pulse beat, leading to corresponding alterations in the color of reflected light.To validate this fundamental principle, we employed an unmodified CMOS imager chip that was placed in direct contact with the skin of the wrist.By analyzing the continuously recorded images, we successfully obtained a continuous pulse waveform.
As depicted in Figure 4a, through the analysis of the intensity of the red channel in the captured images, we can precisely examine the real-time pulse fluctuations of the subject.The high resolution provided by the CMOS imager is crucial for capturing subtle changes in blood volume associated with the pulsebeat, thereby ensuring the accuracy and fidelity of the recorded pulse waveform.This discovery highlights the potential of the CMOS imager as a dependable and non-invasive tool for pulse monitoring.To calculate the PR value, we employed fast Fourier transforms (FFT), a widely used technique for analyzing periodic signals.By applying FFT to the recorded pulse waveform, we were able to determine the frequency of the pulse wave.As shown in Figure 4b, the frequency was determined to be 1.3 when the subject was in a calm state, indicating a pulse rate of 78 beats per minute (bpm).This calculated value aligns closely with the results obtained from a commercial heart rate meter.The agreement between our calculated pulse rate and the results from the commercial device further validates the accuracy and reliability of our wearable bracelet.

Use the Wearable Bracelet to Track TcCO 2 and PR During Daily Activities
To confirm the practicality and effectiveness of our innovative wearable bracelet, we conducted a prospective observational study involving a carefully chosen healthy volunteer.The primary aim of this study was to evaluate the wearable bracelet's capability to accurately monitor TcCO 2 and PR data during the volunteer's diverse daily activities in real-life conditions.The experiment duration was set at 1 h to capture a comprehensive snapshot of the volunteer's activities.To validate the accuracy of the TcCO 2 measurements obtained by the wearable bracelet, we compared them against end-tidal CO 2 (EtCO 2 ) measurement, which is considered a reliable indicator of arterial CO 2 levels. [39,40]The calibration curve shown in Figure 5a illustrates the directly proportional correlation between the TcCO 2 readings obtained from the wearable bracelet and the simultaneously obtained ETCO 2 readings.This calibration curve enables the conversion of the wearable bracelet's signal into PtcCO 2 readings in partial pressure (mmHg).Figure 5b and Figure S8 (Supporting Information) visually illustrate the dynamic behavior of TcCO 2 levels in response to the subject's daily activities.Initially, the sensing signal exhibited a noticeable upward trend as CO 2 molecules diffused from the skin to the porous sensing units, driven by the existing concentration gradient.Subsequently, the signal reached a steady plateau (≈ 1300 ppm), indicating the establishment of concentration equilibrium.Significantly, the TcCO 2 level exhibited a rapid and remarkable increase following the subject's consumption of snacks.This observation aligns with our expectations, as the ingestion of food stimulates metabolic activity, leading to the release of CO 2 .In addition, the consumption of black coffee resulted in a further rapid rise in TcCO 2 levels.During the test, TcCO 2 and ETCO 2 levels exhibited similar trends, further validating the wearable bracelet's ability to accurately capture and monitor changes in TcCO 2 levels.
Simultaneously, we captured real-time pulse waveforms to gain additional insight into the subject's physiological dynamics.Figure 5c shows the spectrum analysis results of the pulse waveform recorded by the wearable bracelet before black coffee consumption and after black coffee consumption.Prior to black coffee intake, the calculated pulse rate remained stable at ≈75 bpm.However, after the subject promptly consumed black coffee, the pulse rate significantly increased to 91 bpm.This calculation result aligns with the measurements obtained from the heart rate meter, further highlighting the wearable bracelet's ability to provide accurate and timely information regarding the subject's cardiovascular status.
The results of this experiment demonstrate the feasibility and efficacy of our wearable bracelet for non-invasive tracking of TcCO 2 and PR in real-world, uncontrolled environmental conditions.The trial conducted on the volunteer boldly substantiates the potential of our wearable bracelet as a valuable tool for qualitatively assessing cardiopulmonary status in personal health management.

Conclusion
In conclusion, our study presents a wearable bracelet that enables simultaneous monitoring of biophysical and biochemical signals of the human body.The compact and lightweight design of the wearable bracelet ensures its portability, while its ergonomic and comfortable fit allows for a seamless and intimate interface with the human epidermis.A nanomaterials-modified lensless CMOS imager was employed to enable the integration of TcCO 2 and PR monitoring in a single wearable device.The colorimetric CO 2 sensing units demonstrate remarkable sensitivity, selectivity, and anti-humidity interference in tracking TcCO 2 levels.Furthermore, the simultaneous monitoring of pulse rates enables a comprehensive assessment of the cardiopulmonary system.This portable, multifunctional, and noninvasive wearable bracelet holds great promise for daily healthcare applications and exhibits significant potential for scalability.By empowering individuals to monitor their biophysical and biochemical signals in real time, this wearable bracelet opens up new possibilities for proactive health monitoring and early detection of potential health issues.Its introduction into the healthcare domain paves the way for new horizons in personalized health management.

Experimental Section
Materials: m-cresol purple and glycine were purchased from Innochem.Ethanol, ethylene glycol, hexadecyltrimethylammonium bro-mide, and tetrabutylammonium hydroxide (25% in water) were purchased from Aladdin.PDMS prepolymer and the curing agent (Sylgard 184) were purchased from Dow Corning.Silica nanoparticles (5-15 nm) were purchased from Sigma-Aldrich.All chemical reagents were used directly without further purification.Ultra-pure water (18 mΩ) was produced by Millipore Direct-Q.CO 2 and ultrapure air were purchased from Jingong Gas Co., Ltd.
Preparation of CO 2 Sensing Solutions: To prepare the CO 2 sensing solution, m-cresol purple (36 mg), glycine (45 mg), and silica nanoparticles (800 mg) were first dissolved in 6 mL solvent (the volume ratio of water, ethylene glycol, and ethanol was 5:5:2) and shaken for 30 min.Then, 2 mL of tetrabutylammonium hydroxide (25% in water) and 2 mL of hexadecyltrimethylammonium bromide (0.1 mol L −1 in ethanol) were added and stirred evenly.
Wearable Bracelet Fabrication: For the sensor fabrication, 5-megapixel CMOS imagers (model ov5647, OmniVision Technologies) with a pixel size of 1.4 μm and an imaging area of 3.7 × 2.7 mm 2 were utilized.To enable the deposition of CO 2 sensing elements onto the pixels of the CMOS imager, the lens of the CMOS imager was carefully disassembled prior to the printing process.The sensing solutions were then precisely printed onto the CMOS imager using an inkjet deposition system (Sonoplot, Microplotter Proto) with an inkjet voltage of 7.0 V and a resonance time of 100 ms.To provide a protective coating, a thin layer of polydimethylsiloxane (PDMS) was applied onto the surface of the modified CMOS imager.This was achieved by spin coating a mixture of PDMS prepolymer with a curing agent, with a mass ratio of 10:1.The mixing of PDMS prepolymer with the curing agent activated the polymer chains and transformed the liquid materials into a solid elastomer.The curing time of PDMS typically relies on the temperature.In this study, the coated CMOS imager was heated at 60 °C for 12 h to facilitate the PDMS curing process.To provide illumination during the measurements, four white LEDs (model LEDtronics) were affixed to the chamber to serve as a light source.Additionally, a silicone gasket was applied to the circuit board to ensure gas tightness and maintain a controlled environment within the chamber.
CO 2 Concentration Calibration: To calibrate the CO 2 concentration, samples of CO 2 gas were prepared by diluting a 4% CO 2 gas with ultrapure air, and these samples were collected in airbags made of aluminum foil.The concentrations of the diluted CO 2 samples were verified using an NDIR CO 2 sensor (Jingxun Sensor Co., 1-5000 ppm).The CO 2 detection experiments were conducted within a custom-made resinous chamber that was fabricated using a 3D printer (Formlabs, Form 3).Delivery of the CO 2 samples into the chamber was accomplished using a diaphragm gas pump (Pengpu) at a flow rate of 500 mL min −1 .Prior to introducing each CO 2 sample, pure air was delivered for 1 min to establish a stable baseline.The CMOS imager was connected to a Raspberry Pi 4B for configuration and image recording purposes.The gain and white balance settings of the CMOS imager were adjusted to optimize imaging performance.Full-resolution images (2592 × 1944) were recorded at a frame rate of 60 frames per minute.The open-source software ImageJ was employed to identify the CO 2 sensing elements and measure their intensities, specifically utilizing the red channel for analysis.The response of a sensing element was defined as: where Intensity (t 0 ) and Intensity (t) represent the mean gray value(red channel)of a sensing element before and after CO 2 exposure, respectively.The sensor response was calculated by the average value of all sensing units.According to Beer-Lambert's law, the linear fitting curve of CO 2 concentration and sensor response was obtained by the testing data between 1000 and 3000 ppm.The limit of detection (LOD) was calculated by: where SD air represents the standard deviation of the sensor response in clean air and S represents the sensitivity, which is the slope of the linear fitting regression line.TcCO 2 and PR Monitoring: To conduct the experiments, a healthy volunteer who willingly participated in the data collection process was recruited.The custom-made wearable bracelet, designed for TcCO 2 and pulse rate monitoring, was securely worn on the volunteer's wrist.During the test, the subject was instructed to sit down, rest, and breathe normally until the readings from the wearable bracelet reached a stable plateau.From this point onward, the sensor data were analyzed to assess the performance of the wearable bracelet.Real-time images were recorded at a rate of 60 frames per minute, while the real-time video was captured at a rate of 15 frames per second by the CMOS imager.These recordings were conducted indoors, under ambient lighting conditions.The real-time TcCO 2 concentration was calculated utilizing the colorimetric sensor response and the calibration curve.The raw video was converted into individual images using the "Free Video to JPG Converter" software.Subsequently, spectrum analysis was performed using Origin Software to analyze the pulse rate.
To evaluate the performance of the wearable bracelet, a comparison was made through EtCO 2 measurements, which serve as a reliable representation of arterial CO 2 levels.EtCO 2 was measured using a breath-based CO 2 monitor by an end-tidal CO 2 analyzer (Kingst, KMI605C).The inlet of the CO 2 monitor was connected to a respiratory gas pipeline, which the subject wore to facilitate normal breathing under testing conditions.The carbon dioxide analyzer was also equipped with a finger clip pulse oximeter, which was utilized to verify the accuracy of PR test data.

Figure 1 .
Figure 1.Principle of the wearable bracelet.a) Schematic illustration of the wearable bracelet; b) Schematic illustration of the nanomaterials-modified lensless CMOS imager, the edge area of the CMOS imager is modified by colorimetric CO 2 sensing units for TcCO 2 detection, while the central area remains unmodified to capture images of wrist artery for PR detection; c) Sensing principle of the wearable bracelet for TcCO 2 , the colorimetric CO 2 sensing units display color transition when reacting with diffused TcCO 2 molecules, a hydrophobic PDMS membrane is applied to repel water molecules; d) Sensing principle of the wearable bracelet for PR, the CMOS imager captures the fluctuation of pulsatile blood volume through the central unmodified imaging area, and the pulse waveform is derived by analyzing the reflected light that is captured.

TcCO 2 and
PR information within a single multimodal CMOS image.(Figure S2, Supporting Information)

Figure 2 .
Figure 2. a) SEM image of CO 2 sensing units without SiO 2 NPs; b) SEM image of CO 2 sensing units with SiO 2 NPs (133 mg mL −1 ); c) Comparison of the CO 2 (500 ppm) sensing response between the sensing unit without (blue) and with (red) SiO 2 NPs (133 mg mL −1 ), the error bars represent standard deviations of six sensing units; d) Calibration curve of the CO 2 sensing units, the error bars represent standard deviations of six sensing units.

Figure 3 .
Figure 3. a) Comparison of the CO 2 sensing response before and after the application of PDMS coating; b) Comparison of the sensing response by 500 ppm CO 2 diluted by dry air and humid air; c) Comparison of the sensing response by 0.3% CO 2 , pure water vapor, 1% ammonia, and 1% acetone, the error bars represent standard deviations of three sensing units.

Figure 4 .
Figure 4. a) Real-time pulse waveform recorded by the wearable bracelet; b) Spectrum analysis using FFT of the pulse waveform.

Figure 5 .
Figure 5. a) Proportional correlation between the TcCO 2 readings obtained from the wearable bracelet and the concurrently obtained ETCO 2 readings, the error bars of TcCO 2 represent standard deviations of 3 sensing units, the error bars of ETCO 2 represent standard deviations of three test results; b) Real-time tracking of TcCO 2 and ETCO 2 level during different scenarios, including the resting status, consumption of snacks, and consumption of black coffee, the error bars of ETCO 2 represent standard deviations of three test results; c) Spectrum analysis of the pulse waveform recorded by the wearable bracelet before (black line) black coffee consumption and after (red line) black coffee consumption.