A miniature magnetic switch unlocking multimodal, chip-free, and batteryless airway sensing

Abstract

Continuously monitoring airway physiological properties in situ is critical for enabling early intervention. Unlike conventional approaches relying on on-board chips and batteries, chip-free sensing based on inductive coupling offers a promising pathway for fully wireless and battery-free monitoring. However, current chip-free strategies are limited in achieving multimodal sensing due to their narrow operational frequency ranges. Here, we introduce a miniature magnetic switch that enables wirelessly controlled channel switching through a cantilever beam mechanism, which anchors, rotates, and toggles between channels in response to external magnetic fields. Integrated with an inductive coil and capacitive sensors, the system generates resonance-based signals that can be selectively and wirelessly detected using a vector network analyzer (VNA). The concept is demonstrated through sensing tissue stiffness by switching between deformation and stress sensing modes, spatiotemporal mapping of tissue pressure, and multimodal sensing of multiple mucus properties. The proposed magnetic switch thus enables next-generation smart airway devices for continuous, minimally invasive, and multimodal sensing in diverse airway diseases.

A miniature magnetic switch allows multiplexing in chip-free sensory devices for sensing airway physiological properties.

INTRODUCTION

Continuous, wireless, and minimally invasive monitoring of deep airway tissue physiological properties outside the hospital is critical for enabling early detection of complications, reducing reliance on invasive bronchoscopies, and improving patient outcomes (1). Airway diseases such as lung cancer (2), chronic obstructive pulmonary diseases (COPDs) (3), and cystic fibrosis (4) often progress silently, and real-time monitoring of tissue stiffness, pressure, mucus accumulation, or temperature can reveal pathological changes before they become life threatening. Traditional method using bronchoscopy or catheters (5, 6) typically requires anesthesia, carries procedural risks, and does not provide longitudinal surveillance in daily life (7). Other methods using computed tomography raise concerns of radiation (8). In addition, continuous, wireless, and minimally invasive monitoring is also critical for monitoring the patency of implantable devices such as airway stents (9, 10). Airway stents play a vital role in maintaining airway openness for patients with lung cancer and other obstructive pulmonary disease (11–13), yet they remain vulnerable to complications that can develop silently between hospital visits. Stent-related problems, including migration, mucus clogging, tissue in-growth, and granulation tissue formation (10), can quickly compromise airway patency, leading to recurrent obstruction, respiratory distress, and even life-threatening emergencies. Similarly, detection relies heavily on periodic bronchoscopies or radiation-based medical imaging, both of which are invasive, costly, and limited to discrete time points, leaving notable gaps in surveillance (14).

Wireless sensing devices may potentially enable continuous, real-time, and minimally invasive monitoring of stent function (15), offering early warning of obstruction or migration before symptoms become severe. Such a system not only reduces the frequency of invasive follow-up procedures and hospitalizations but also provides clinicians with longitudinal data to guide timely interventions and optimize treatment. Both chip-based and chip-free sensing devices have been reported for sensing restenosis (16), flow (17, 18), pressure (19–21), or mucus property (22). On one hand, one approach to monitoring airway stents involves integrating electronics to sense stent conditions in real time (23). Recent works have shown sensing mucus properties in airway stents (22) with on-board sensors, Bluetooth board, and batteries. However, miniaturization is essential yet challenging due to space constraints, the size of the wireless processing and communication unit, as well as the on-board battery or remote charging circuits. These constraints make it difficult to integrate compact, durable sensors with wireless capabilities into airway stents.

On the other hand, chip-free sensing via inductive coupling offers a promising solution for wireless and battery-free biomedical sensing (18, 24, 25). These systems use an externally applied electromagnetic field to excite a passive resonator, which then modulates its resonance characteristics in response to environmental changes (26). Previous studies have demonstrated the use of pickup coils to detect parameters such as pressure (27), temperature (28), and fluid flow (25) by designing resonance circuits that respond to variations in capacitance or inductance. The core components of these devices typically include an inductor, which acts as an antenna for receiving and transmitting signals, and a capacitor-based sensor that adjusts the resonant frequency based on the targeted parameter. Despite their advantages in miniaturization and wireless functionality, conventional chip-free resonance-based sensors have limitations in sensing diversity and multifunctionality (29). Most existing designs of chip-free resonance-based sensors use a single sensor element (30, 31), which restricts them to measuring only one physiological parameter at a time. Some designs rely on different frequency ranges and multiple circuits for multiplexing (24, 32, 33), but they are fundamentally limited by the narrow usable spectrum, the risk of overlap, and extra sizes of multiple coils (see table S1), which constrain their ability to deliver comprehensive airway monitoring. Therefore, despite the advantage of chip-free in-stent sensing in eliminating the need for batteries, improving biocompatibility and resilience, implementing sensing at multiple locations or for multiple biomarkers remains challenging due to the absence of an effective multiplexing mechanism.

To address these challenges, we introduce a fundamental mechanism that enables integrating multiple resonance-based sensors on chip-free sensing devices in the airway by developing a magnetically controlled switch. The switch operates through a cantilever beam mechanism that responds to external magnetic fields by anchoring, rotating, and toggling between channels. When integrated with an induction coil and capacitive sensors, the system produces resonance-based signals that can be selectively and wirelessly detected using a vector network analyzer (VNA), allowing remote monitoring of diverse physiological properties of the airway. We first show the magnetic switch concept with a tissue stiffness sensor that combines capacitive deformation and stress sensing to quantify tissue elasticity. We then demonstrate selective control of two pressure sensors for spatiotemporal mapping of tissue pressure. Last, we show that the magnetic switch enables selective monitoring of mucus accumulation and mucus temperature by capacitive mucus layer thickness and temperature sensors. This chip-free and battery-free multimodal monitoring framework enables continuous, minimally invasive, and early detection of complications, reducing the need for frequent bronchoscopies and improving patient outcomes.

RESULTS

Overview of a magnetic switch enabled sensory ring for monitoring airway physiological properties

We report a miniature magnetic switch which enables a chip-free sensory ring that can be integrated with airway stents to enable real-time monitoring of airway physiological properties and stent patency. The system operates via inductive coupling, providing fully battery-free and wireless sensing while supporting multimodal measurements without any onboard chips. Sensor signals are wirelessly picked up using a readout coil and analyzed through a VNA (Fig. 1A). The sensory ring comprises three core elements including the magnetic switch for selective channel connection, capacitive sensors, and an induction coil as an antenna, together forming a miniaturized, chip-free platform optimized for the airway environment. As illustrated in Fig. 1B, the magnetic switch can connect a specific capacitive sensor to the induction coil and form an inductive/capacitive (LC) circuit. Then, as changes in tissue properties cause a change in the sensor capacitance, the shift in the resonant frequency of the LC circuit can be immediately picked up by the VNA (Fig. 1C). Figure 1D presents a representative VNA readout, where the $S_{11}$-frequency curve exhibits a distinct resonance peak shift corresponding to the changes in the measured physiological property.

Fig. 1. Concept of the magnetic switch enabled sensory ring for wireless, chip-free, battery-free, and multimodal sensing in the airway.

(A) Concept of the sensory ring and self-expandable airway stent implanted inside a human trachea for wireless monitoring of airway conditions, such as tissue stiffness, pressure, temperature, and mucus conditions for disease diagnosis. (B) Illustration of the magnetic switch enabled multichannel LC circuit. (C) Illustration of the conversion from tissue properties to readout resonant frequencies. (D) Illustration of the readout signal from the VNA reader device. Different frequency curves are overlaid in the same figure to show the shift of the peak frequency. (E) Illustration of the chip-free wireless sensing mechanism based on radio frequency. (F) Schematics of the components inside the sensory ring including the magnetic switch, induction coil, and capacitive sensors. (G) (i) Optical image of an example sensory ring with no on-board chips for sensing pressure and tissue stiffness. The dashed box indicates a pressure sensor on the inner surface of the sensory ring. (ii) Optical image of an example sensory ring with no on-board chips for sensing mucus layer thickness and temperature. The dashed box indicates a temperature sensor on the inner surface of the sensory ring. (H) Optical image of a self-expandable metallic airway stent sutured with a sensory ring.

The LC circuit, composed of a capacitive sensor and an induction coil, enables wireless resonant frequency readout (Fig. 1E). Incorporation of the magnetic switch allows selective activation of multiple sensors within the same circuit, achieving multiplexed sensing capability (Fig. 1F). For inductive coupling–based devices, this feature is essential for monitoring multiple locations and diverse physiological parameters simultaneously, whereas previous research incorporating multiple sensors relies on multiple individual circuits and a broad frequency range to accommodate different sensors, which require extra space and result in increased measurement uncertainties.

As an example, Fig. 1G first shows a sensory ring integrating pressure sensors and a tissue stiffness sensor to detect airway pathological changes when there is fibrosis or wall remodeling. Figure 1G also shows that another sensory ring integrating mucus layer thickness and temperature sensors allows monitoring mucus accumulation and potential inflammation for early disease detection (see fig. S1 for detailed illustration). For translational applications, the sensory ring can be bonded to a self-expandable airway stent, ensuring stable anchoring and continuous monitoring of stent patency (Fig. 1H). Collectively, these advances establish a chip-free, multimodal, and wireless sensing platform that enables robust, multiplexed monitoring of airway tissue biomechanics, mucus properties, and stent functionality, offering a transformative approach for long-term management of airway diseases.

Design and control of a magnetic switch for multiplexed sensing

To allow multiplexed chip-free sensing, the key design is a miniature magnetic switch that can connect multiple specific capacitive sensors individually to the same induction coil. Compared with reed switches (34, 35) which can only be controlled to be “on” or “off” by a continuously applied external magnetic field, the proposed magnetic switch allows selecting among multiple channels. It also remains in one channel at a resting state, even when no magnetic field is applied, and only alters to a different channel when a certain external magnetic field is exerted. The magnetic switch with four channels is shown in Fig. 2A, which consists of a conductive base and a magnetic probe. The current enters through one selected channel, flows through the magnetic probe, which is also conductive, and exits at the other end to connect the inlet and outlet (see Materials and Methods and fig. S2 for the fabrication process). On one hand, the conductive base (Fig. 2B) has four conductive pads as outlets, of which each has a 6° angle span and is separated with a 2° gap. This geometry ensures reliable electrical contact between the magnetic bridge and the sensing pads within fabrication tolerances while maintaining sufficient clearance to prevent simultaneous connection of adjacent channels. Further reduction of these angles leads to fabrication and interconnection challenges. In addition to the conductive pad of each channel, a conductive shaft ensures smooth rotation of a magnetic probe and consistent conductivity at different angles. The conductive base also integrates two mechanical stoppers to prevent overspinning of the magnetic probe.

Fig. 2. Mechanism of a magnetically controlled switch for multiplexed sensing.

(A) Optical images of the magnetic switch with a conductive base and a bendable and steerable magnetic probe. (B) Optical images of the conductive base including a top view, a side view, and a zoomed-in view of the conductive pads of the four channels. (C) Optical images of the magnetic probe in the top and side views. (D) Schematics of the magnetic switch in a side view when anchoring and floating (left) and a top view when the magnetic probe is at two different channels (right). Positive ${\mathrm{\alpha }}_{\mathrm{p}}$ corresponds to clockwise rotation of the cantilever plane relative to the x axis as viewed from above, while negative ${\mathrm{\alpha }}_{\mathrm{p}}$ corresponds to counterclockwise rotation. (E) Optical images of the magnetic switch in a side view when anchoring and floating (left) and optical images of the magnetic switch in a top view at different channels (right). Scale bars, 1 mm. (F) Extracted the magnetic probe angles ${\mathrm{\theta }}_{\mathrm{p}}$ (pitch) and ${\mathrm{\alpha }}_{\mathrm{p}}$(yaw) as a function of time when the magnetic switch is operated from channel 1 to channel 4. Green or red areas indicate the circuit is closed or open. (G) Recorded magnetic field during the process in (F). (H) Video frames (movie S1) showing the process of switching from channel 1 to channel 2. Average switching time: 0.5 s. (I) Voltage of the circuit output as a function of the external magnetic field pitch angle ${\mathrm{\theta }}_{\mathrm{B}}$ in five trials. (J) Possibility of the circuit connectivity to each channel as a function of the external magnetic field yaw angle ${\mathrm{\alpha }}_{\mathrm{B}}$.

On the other hand, the magnetic probe (Fig. 2C) has two NdFeB magnets affixed to the front and rear parts of the probe, while a soft joint connects the two components. Upon applying external magnetic field $\mathbf{B}\left(\mathit{t}\right)$, the magnetic probe could either be lifted or rotated on-demand due to the magnetic torque. The conductivity of the probe is established with two pieces of coppers connected by an elastic joint combining a flexible copper sheet (Pyralux) and a soft polymer layer. The soft polymer layer made of polydimethylsiloxane (PDMS) is attached to the copper sheet providing the elastic stress to press the conductive tip onto the pads of the conductive base in the resting state. The tip of the magnetic probe electrically connects the magnetic probe and the conductive pad with a relatively high spatial resolution. The detailed dimensions of the magnetic switch are shown in fig. S3.

The four representative states of the magnetic switch are quantified using the probe angles ${\mathrm{\alpha }}_P$ and ${\mathrm{\theta }}_P$ as illustrated in Fig. 2D. First, at a resting state when the magnetic field applied is relatively small or in a specific direction, the tip is “anchored” onto one of the conductive pads due to the stress of the elastic joint. Meanwhile, connectivity of the channel is established as the magnetic probe maintains a relatively stable electrical connection to the shaft due to the full-copper design. Second, at a switching state, the magnetic field angle ${\mathrm{\theta }}_B$ about the y axis is controlled to lift the magnetic probe tip off from the conductive pad by overcoming the elastic torque via the magnetic torque (36, 37). Third, this lift-off motion switches off the connectivity of the current channel and allows a smooth rotation of the magnetic probe to another channel by eliminating the friction between the magnetic probe tip and the conductive pad. Subsequently, the “floated” probe follows the magnetic field angle ${\mathrm{\alpha }}_B$ about the z axis to reach to another channel. Last, the magnetic probe returns to a resting state when the external magnetic field angle is controlled to reduce the magnetic torque allowing the magnetic probe tip to touch the conductive surface again. These four steps complete a typical switching process. The experimental images of the four states are further shown in Fig. 2E to visualize the process. See note S1 and fig. S4 for the force analysis of the magnetic switch.

To demonstrate the ability of switching between multiple channels, we track and plot the controlled probe angles ${\mathrm{\alpha }}_{\mathrm{P}}$ and ${\mathrm{\theta }}_{\mathrm{P}}$ in Fig. 2F and the corresponding magnetic fields in Fig. 2G. The green and red areas represent whether the switch is connected to a channel or not, respectively. At the beginning, the magnetic probe connects to channel 1. The probe tip is then lifted with ${\mathrm{\theta }}_{\mathrm{p}}$ changing from 10° to around 18° when applying the external magnetic field to break the connection. To connect to channel 2, the external magnetic field is applied to rotate the magnetic probe about the z axis and then the x axis such that the probe tip lands on channel 2 pad, with ${\mathrm{\alpha }}_{\mathrm{p}}$ reaching to about 5° and ${\mathrm{\theta }}_{\mathrm{p}}$ back to 10°. The process of floating, switching, and anchoring is sequentially shown in Fig. 2H with a similar process when switching to channel 3 or channel 4.

To quantify the robustness and precision of the magnetic switch, we investigate the repeatability and control precision. We test the channel connectivity with ${\mathrm{\theta }}_B$ ranging from −180° to 180°, at a magnitude of 15 mT using a simple voltage measurement circuit (fig. S5). Within five trials of controlling ${\mathrm{\theta }}_B$ from −107° to 21°, the probe is anchored on the conductive pad with a nonzero voltage output as shown in Fig. 2I. When ${\mathrm{\theta }}_B$ decreases to approximately −100°, the negative component of the magnetic field induces a sudden rotation of the switch, causing it to collide with the stopper and disconnect (fig. S6). To further investigate the selectivity of channels when the probe is floating, we control ${\mathrm{\alpha }}_B$ from −22° to 22° to allow repeated anchoring and lifting of the probe. Figure 2J shows the selectivity of a specific channel when varying ${\mathrm{\alpha }}_B$, where the peaks indicate the optimal control magnetic field angle to land in a specific channel, ensuring a repeatable channel selection mechanism. We further conducted 400 switch connection tests resulting in a 96% success rate (fig. S7). Under the prescribed magnetic field angle (${\mathrm{\alpha }}_B$), the probe maintained a stable orientation (${\mathrm{\alpha }}_{\mathrm{p}}$) with no observable drift, demonstrating highly reproducible operation. The magnetic switch and sensors are further encapsulated in epoxy housings (figs. S8 and S9). The effectiveness of the encapsulation is validated through accelerated aging tests on the fully assembled device following 7 days of soaking in phosphate-buffered saline (PBS) (fig. S10). The magnetic field required for switching remains comparable before and after aging, indicating unchanged magnetic torques. Our proposed magnetic switch thus enables repeatable and controllable multichannel selection and retains its state. Despite only four channels are shown, the magnetic switch can allow more channels to be integrated.

We further demonstrate the control of the magnetic switch using an external magnetic actuation unit, as shown in Fig. 3 and fig. S11. First, we examine the design of the elastic joint in the magnetic probe to ensure stable anchoring under various magnetic fields applied along the z axis (Fig. 3A). The maximum z-axis magnetic field that each magnetic switch design can withstand is experimentally characterized. The key design parameter is the thickness of the polymer sheets, which determines the stiffness of the elastic joint. As shown in Fig. 3B, $B_z$ serves as the primary component governing the vertical motion of the switch, where the upper limit of $B_z$ defines both the maximum field a switch can withstand during anchoring and the minimum field required to initiate lifting. A negative field threshold also exists, beyond which excessive twisting may lead to loss of connection, when the contact resistance between the conductive probe and the sensor pad exceeds 5 k$\mathrm{\Omega }$, which results in failure to trigger the corresponding resonance channel.

Fig. 3. Characterization and optimization of the magnetic switch.

(A) Schematic illustration of the magnetic switch actuated by an external magnetic field in the z axis. (B) Magnetic field ranges along the z axis, at which the switch remains connected, for magnetic switch designs with different polymer layer thicknesses. (C) Illustration of the magnetic actuation distance using the magnetic actuation system. $d_{\mathbf{s}}$ is the distance measured from the center of the external permanent magnet to the base plate of the implant. (D and E) Optical images of the magnetic actuation system with three actuation units including (i) two linear dc motors for distance control, (ii) a step motor for pitch angle control, and (iii) a servo motor for yaw angle control. Scale bars, 2 cm. (F) Simulated time-varying magnetic field for lifting and anchoring the magnetic probe at $d_{\mathrm{s}}= 5 \text{cm}$. (G) Simulated magnetic field magnitude when varying the actuation distance $d_{\mathrm{s}}$ from 4 to 8 cm. (H) Repeatability test of the magnetic switch over 500 actuation cycles and the zoomed-in plot of the channel voltage outputs at the beginning and the end of the test. (I) Video (movie S2) frames of light-emitting diodes (LEDs) being sequentially lighted up using the magnetic switch and the magnetic actuation unit. (J) The recorded magnet angles of the magnetic actuation unit during the LED switching process shown in (I).

Among different designs, the switch without a PDMS layer in its elastic joint can barely resist any positive $B_z$ field, whereas the switch incorporating a 0.12-mm-thick PDMS layer can withstand fields up to ~15 mT. Increasing the PDMS layer thickness further enhances anchoring stability but simultaneously increases the difficulty of lifting the probe. Magnetic fields applied along other directions also influence the switch’s connectivity, as illustrated in fig. S12, which quantifies the $B_x$ and $B_y$ components required to rotate the probe about the z axis. Fields along the x axis have a relatively minor effect on lifting performance, as magnetic switches with polymer layers can maintain connectivity even above 20 mT. This behavior arises because, during anchoring, resilience is primarily provided by the friction between the probe tip and the contact pads. A stiffer elastic joint generates greater normal force at the contact interface, thereby enhancing connection robustness.

Furthermore, for practical applications, we developed a magnetic actuation unit to control the operation of the switch, as shown in Fig. 3 (C to E). The relative positioning between the control unit and the switch is illustrated in Fig. 3C, with $d_s$ defined as the distance measured from the center of the external permanent magnet to the base plate of the implant. As shown in Fig. 3D, a servo motor drives the rotation of the permanent magnet about the z axis, while a stepper motor governs its spinning motion to lift and lower the probe following the programmed waveforms. In addition, a pair of dc motors enables linear translation of the magnet along the z axis to modulate the magnetic field strength. An example magnetic field distribution is presented in Fig. 3E, demonstrating that the system can generate magnetic fields up to 20 mT (see figs. S13 and S14) when positioned on a human chest phantom, where the distance between the top surface of the actuation unit and the measurement point is ~2.5 cm (fig. S15). Figure 3F illustrates simulated magnetic field distributions under different actuation commands, while Fig. 3G plots the measured magnetic field strength as a function of distance $d_s$, defining the effective operational range for switches with different joint stiffness. For instance, the 0.12-mm-thick joint design requires a field strength of ~15 mT to lift the probe, corresponding to an $d_s$ of 6 cm with the current unit (25 mm by 25 mm by 25 mm NdFeB magnet).

Last, we demonstrate the repeatability of the magnetic switch and the actuation unit, as shown in Fig. 3H, using the same testing circuit illustrated in fig. S5. Connectivity is defined as a binary variable: 1 when the conductive bridge establishes a stable electrical connection with an output voltage larger than 0.9 V, and 0 when the output voltage is less than 0.1 V. The actuation unit was programmed to automatically cycle through four channels for 500 consecutive cycles. The results show that the switching behavior remains consistent, with identical response patterns observed at both the initial and final cycles, confirming reliable long-term operation. To further illustrate the functionality of the magnetic switch, we integrated it into a four–light-emitting diode (LED) circuit, where each LED represents one active channel. As shown in Fig. 3I, the LEDs illuminate sequentially as the corresponding magnetic fields are applied, following the programmed actuation pattern depicted in Fig. 3J.

Mechanism and demonstration of sensing tissue elastic modulus by multiplexed sensing

With the integration of the magnetic switch, multiple sensors can communicate with an external pickup unit, enabling measurements at different locations, monitoring multiple physiological parameters, or capturing properties that require more than a single sensing modality. One such property is tissue stiffness, which requires both stress and strain information for accurate characterization (38, 39). Airway tissue stiffness is a critical biomarker in a variety of lung diseases (40, 41), including cancer, fibrosis, asthma, and COPD. To elucidate the sensing mechanism, the fundamental principle of tissue stiffness sensing is illustrated in Fig. 4 and movie S3. The tissue stiffness sensor, shown in Fig. 4A, consists of two conductive plates separated by a dielectric liquid, forming a parallel-plate capacitor. An integrated magnet enables remote actuation of the top plate and the attached palpation probe for tissue indentation. The sensor is fully encapsulated within a resin protective housing, as depicted in Fig. 4B. The detailed fabrication process and dimensions of the sensor are provided in figs. S16 and S17. When a magnetic field is applied along the z axis, a torque is induced on the top plate and probe, causing the probe to press into the soft material beneath, as illustrated in Fig. 4C.

Fig. 4. Induction coupling–based capacitive and magnetic sensor for wireless sensing tissue stiffness.

(A) Schematics of the working principle of the tissue stiffness sensor. (B) Optical images of the tissue stiffness sensor in a side view. (C) Sequential optical images of probing process at different magnetic field strength. (D) Sensor probe displacement $d$ as a function of the sensor capacitance. (E) Sensor capacitance and the corresponding applied magnetic field as a function of time. (F) Capacitance of the tissue stiffness sensor on synthetic materials of different elastic moduli when varying external magnetic fields. (G) Applied magnetic field as a function of the displacement of the sensor probe on synthetic materials of different elastic moduli. (H) $∆B_z/∆d$ as a function of material elastic modulus. (I) Optical image of the experimental setup showing the distance between the readout coil of the VNA and the sensor inductor (2.5 cm). (J) $S_{11}$ value of the VNA reading when different magnetic fields are applied. Sample elastic modulus: 30.33 kPa. (K) Resonant frequencies as a function of external magnetic fields when testing the stiffness sensor on two synthetic materials (7.17 and 30.33 kPa).

To explain the sensing mechanism, this palpation process can be modeled using the Hertzian contact theory (42). The equivalent elastic modulus, assuming a small deflection, can be simplified as follows: $E^{\ast }=B_z/\left(\mathit{kd}\right)-E_1$, where $E^{\ast }=\frac{E}{1-\mathrm{\nu }}$ is the equivalent material Young’s modulus ($\mathrm{\nu }$: poison ratio), $B_z$ represents the applied magnetic field strength in the z direction, $d$ is the deflection, $k$ is a constant related to the probe’s geometry and magnetization, and $E_1$ is a constant (see note S2 and fig. S18 for the details of this model). Since the distance between the two plates of the capacitor determines the sensor’s capacitance ($C$), a direct correlation is established between the measured capacitance and tissue stiffness. The sensor was first calibrated by measuring $C$ as a function of $d$ (Fig. 4D), revealing a linear relationship within the tested range. During the experiments, both capacitance and magnetic field were recorded simultaneously (Fig. 4E). The sensor was then tested on samples of varying stiffness under progressively increasing magnetic fields (Fig. 4F). Using the calibration model in Fig. 4D, the displacement $d$ was computed from the measured capacitance (Fig. 4G). These results were subsequently used as ground truth to determine the unknown constant for model calibration (Fig. 4H).

To validate the wireless sensing capability, a pickup coil is positioned 2.5 cm above the sensor induction coil (Fig. 4I) which connects to a VNA machine (fig. S19). Figure 4J shows a sequence of $S_{11}$ curves picked up by the VNA machine when the sensor is tested on a sample of 30.33 kPa under different magnetic fields. The resonant frequency responses also clearly show that the frequency-$B_z$ curve for a softer sample decreases more rapidly than on a stiffer one in Fig. 4K. Using a curved induction coil, we conduct experiments at a 4.5-cm separation to represent clinically relevant chest wall thickness exceeding the average stomal distance of 2.69 cm (43), defined as the distance from the chest skin to the posterior tracheal wall. Although the peak resonance signal amplitude decreases at 4.5 cm, the resonance peaks remain clearly identifiable, and channel switching remains robust with only a modest increase in magnetic field strength (fig. S20).

The tissue stiffness sensor is further characterized, optimized, and verified as shown in Fig. 5. One key design parameter is the probe length which decides the sensitivity and sensing range of the sensor. As shown in Fig. 5A, the length of the probe ($L_{\mathrm{p}}$) is critical to the sensor’s performance as it relates to the tissue deformation. A smaller $L_{\mathrm{p}}$ leads to a smaller initial gap, thus a larger capacitance value but a smaller range of allowable deformation. In Fig. 5 (B to D), we test the sensor capacitance when varying external magnetic field for sensors of different $L_{\mathrm{p}}$ values including 0.25, 0.5, and 0.75 mm. First, when $L_{\mathrm{p}}$ is 0.25 mm, there is no clear distinction between samples of different stiffnesses because of the saturation of deformation (Fig. 5B). Second, when $L_{\mathrm{p}}$ is 0.5 mm, distinguishable linear trends are observed for palpation on different elastic materials (Fig. 5C) as the two conductive plates are relatively close to each other soon after the magnetic field increases to a given value. Last, when $L_{\mathrm{p}}$ is 0.75 mm, a plateau region is shown when the sample is hard, which indicates a lack of sensitivity as the capacitance value is relatively small due to the large distance between the conductive plates when the external magnetic field is weak (Fig. 5D). Therefore, 0.5 mm is selected for the probe length to allow distinguishing different elastic materials with elastic modulus from 7 to 30 kPa.

Fig. 5. Characterization and validation of the wireless stiffness sensor in phantoms and on ex vivo tissues.

(A) Illustration of the sensor probe length and the sensor capacitance as a function of the probe displacement $d$. (B to D) $\mathrm{\Delta }C/C_0$ as a function of magnetic field magnitude for sensors with $L_{\mathrm{p}}$ = 0.25, 0.5, and 0.75 mm. (E) $\mathrm{\Delta }C/C_0$ as a function of magnetic field magnitude $B_z$ for stiffness sensors with different dielectric materials. (F) $\mathrm{\Delta }C/C_0$ as a function of magnetic field magnitude $B_z$ for stiffness sensors with different numbers of magnets. (G) $\mathrm{\Delta }C/C_0$ when the sensor is deformed by external magnetic fields for 300 cycles to demonstrate repeatability. (H) $\mathrm{\Delta }C/C_0$ as a function of magnetic field magnitude B for a stiffness sensor actuated with different out-of-axis magnetic fields. (I) Elastic modulus of the synthetic softer materials sensed by the stiffness sensor and indentation as a comparison. (J) Comparison of the elastic modulus of porcine trachea tissue by the stiffness sensor and indentation. Error bars represent standard deviations (SD) for $n$ = 5 trials.

Other key design parameters affecting the sensor performance include the dielectric constant of the liquid material between the copper plates and the magnetization of the top plate. As shown in Fig. 5E, sensors filled with air, deionized (DI) water, glycerol, and oil are individually tested for capacitance response. Compared to air, liquids such as water, glycerol, and oil all yield higher capacitance values; however, oil provides only limited improvement due to its low dielectric constant. DI water, while having a high relative permittivity (value: 78 to 80), exhibits abnormal capacitance drops as the plates approach each other, and evaporation causes notable fluctuations even without deformation. Among the tested liquids, glycerol demonstrates the most stable electrical behavior along with a high relative electrical permittivity (value: 41 to 43), making it the optimal choice as the dielectric material for the sensor. In addition, we investigate how the magnetization of the top plate influences sensor performance. As shown in Fig. 5F, reducing the number of magnets decreases the sensitivity due to insufficient magnetic torque to drive palpation, resulting in smaller deformations. Although increasing magnet size could enhance sensitivity, this approach is constrained by the overall size requirements of the sensor.

Furthermore, repeatability is critical for long-term sensor operation. As shown in Fig. 5G, the sensor demonstrates consistent performance over 300 cycles, with a shift of 3%. In practical applications, misalignment of the actuation magnetic field can introduce measurement errors. Figure 5H shows that when the field is offset by less than 45°, the sensing curves remain similar to the control case (0° offset). However, larger angular deviations result in noticeable discrepancies, suggesting that although with robustness, alignment of the magnetic actuation unit is required.

Last, to evaluate the sensing performance, we validate the sensor’s accuracy using both synthetic materials and ex vivo tracheal tissue samples, comparing the results with those obtained from a standard indentation method. The same samples are characterized using our stiffness sensor and a commercial indenter (Hysitron BioSoft In-Situ Indenter, Bruker AG). As shown in Fig. 5I, the sensor achieves accurate stiffness predictions for samples below 70 kPa, exhibiting an average relative error of less than 5%. For stiffer samples, deviations increase slightly, likely due to minor twisting of the probe during measurement. To further demonstrate the capability for detecting pathological tissue stiffening, the sensor is tested on porcine tracheal tissue (Fig. 5J). Measurements are conducted at two distinct sites with one representing normal tissue and the other exhibiting heat-induced fibrosis. The sensor successfully differentiates the stiffness between these regions, consistent with results obtained from the indenter. Tracking changes in tissue stiffness during disease progression, particularly in large-animal models, is critical for understanding pathophysiological remodeling and enabling long-term, continuous monitoring in translational studies. Moreover, real-time stiffness assessment provides valuable feedback during pharmacological treatment, allowing quantitative evaluation of therapeutic efficacy over extended timeframes.

Demonstrating multiplexed sensing of tissue pressure and stiffness in a sensory ring

A magnetic field sensor using similar capacitive measurement mechanism is developed to obtain the magnetic field information as shown in fig. S21. The sensing mechanism is based on two capacitive plates with an elastic joint (fig. S21A). When a magnetic field of different magnitudes in the z axis is applied, the bending angle of the sensor varies, inducing the capacitance change (fig. S21, B and C). The sensor shows a relatively small hysteresis curve (fig. S21D). Further repeatability test confirms the ability of long-term usage of the sensor as shown in fig. S22. A comparison with commercial magnetic field sensor validates the effectiveness of the magnetic field sensor with a root mean square error of less than 0.45 mT in fig. S23.

With the magnetic field sensor and the deformation sensor, the functionality of the magnetic switch and the tissue stiffness sensor is demonstrated with a soft sensory ring, as shown in Fig. 6 and movie S4. With the sensors ready, the circuit is connected to an induction coil to work with an outside pickup coil and a VNA (Fig. 6A). Another two pressure sensors are added to enrich the sensory ring’s functionality, as well as fully exploit the four channels of the switch. The whole experiment setup is shown in fig. S24. Figure 6C shows the switching process from channel 1 to channel 4, with each channel equipped with one sensor.

Fig. 6. Demonstration of a sensory ring for multimodal and multisensor chip-free and battery-free sensing of airway tissue stiffness and pressure.

(A) Schematic of the experiment setup for the sensory ring, together with an expanded view of the inside circuit integrating a stiffness sensor, a magnetic sensor, two pressure sensors, and a magnetic switch. (B) Optical image of the sensory ring integrated with a magnetic switch, a stiffness sensor, and two pressure sensors. (C) Optical images showing the switch landing on four channels. Scale bar, 1 mm. (D) Resonant frequency change of the sensory ring, picked up by a VNA, as a function of sensor capacitance when the ring is placed inside a trachea phantom or a sheep trachea. (E) Time-varying resonant frequency change from the stiffness sensor placed on stiff and soft tissues. (F) Calculated time-varying deflection of the stiffness sensor probe on stiff and soft tissues. (G) Time-varying resonant frequency change from a magnetic field sensor on stiff and soft tissues. (H) Calculated time-varying magnetic field ($B_z$) detected by the magnetic field sensor on stiff and soft tissues. (I) Estimated tissue stiffness at two locations of sheep trachea tissue with different stiffness (19 and 62 kPa). (J) Optical image of the sensory ring placed inside a sheep trachea under applied external pressure. (K) Time-varying resonant frequency change when selectively connected to two pressure sensors. (L) Comparison of estimated pressure and load cell measurements for the pressure sensor 1. (M) Comparison of estimated pressure and load cell measurements for the pressure sensor 2.

We first demonstrate tissue stiffness sensing using the sensory ring in porcine trachea tissue. The sensory ring is calibrated about the capacitance and the corresponding readout resonant frequency. With the same ring material, dimension, and coil, we plug in various capacitors into the circuit. Previous works have shown that surrounding tissues will reduce the magnitude of the signal and induce peak frequency shifts (24). We compare the acquired resonant frequencies of the signal when the sensory ring is placed inside a sheep trachea ex vivo and in a phantom, respectively. As shown in Fig. 6D, although the two curves do not overlap, they both exhibit a linear trend with a similar slope angle value of about −0.372 MHz/pF. Therefore, with a capacitance change of less than 30 pF, the capacitance change is linearly dependent on the resonance frequency change regardless of the surrounding tissue. Figure 6E shows the captured resonant frequency change of the sensory ring when connecting to the stiffness sensor during its palpation on a relatively stiff part and a relatively soft area. With the slope angles shown in Fig. 6D and the calibration data shown in Fig. 4, the resonant frequencies are further converted to the tissue deformation as shown in Fig. 6F. Once the channel is switched to connect the magnetic field sensor, the resonant frequency change is measured using the magnetic field sensor as shown in Fig. 6G. Subsequently, the corresponding magnetic field values in the z axis are calculated using the precalibrated model as shown in Fig. 6H. With the known deformations and magnetic fields, the tissue stiffnesses at the specific locations are further estimated in Fig. 6I, where the device clearly distinguishes the difference between the two spots on the trachea tissues.

With the same sensory ring, the magnetic switch enables not only stiffness sensing but also pressure sensing functions at different locations. To showcase the multimodal sensing ability, two calibrated pressure sensors (fig. S25) mounted on the surface of the outer surface of the sensory ring are connected to channel 3 and channel 4 of the magnetic switch, respectively. As shown in Fig. 6J, a load cell with a 5-mm-diameter probe sequentially presses the trachea tissue where the pressure sensors are located. The load cell readings are recorded as well as the readout resonant frequencies of the pick-up coil of the VNA at the same time, as shown in Fig. 6 (K to M). The calculated pressure by the pressure sensors on the sensor ring agrees with the pressure measurement given by the load cell, suggesting the ability of the sensory ring in detecting pressure changes for disease monitoring. The root mean square errors of sensor 1 and sensor 2 are 8.7 and 10.3%, respectively. These errors stem mainly from fabrication variability and can be reduced through improved fabrication and individual sensor calibration, particularly to compensate for slight differences in dielectric material thickness.

Demonstration of multimodal sensing of mucus layer thickness and temperature in a sensory ring

In Fig. 7 and movie S5, we demonstrate the multimodal sensing with a second sensory ring. Severe signal attenuation is observed in the high-frequency range (150 to 200 MHz) when biological materials are placed between the sensory ring and the pickup coil. We therefore optimize the induction coil (figs. S26 and S27) to enhance the resonance peak in the low-frequency regime (<30 MHz), achieving a sensing distance exceeding 2.6 cm, which corresponds to the average stomal distance defined as the distance from the posterior tracheal wall to the skin (43). With the same switch and induction coil, the sensory ring is now equipped with three mucus layer thickness sensors and one temperature sensor, as shown in Fig. 7A. As a validation of the foldability, the device is folded to test its mechanical resilience when the flexibility of the sensory ring is checked by squeezing it hard to mimic the stent’s deployment process as shown in Fig. 7B and movie S6. No cracks are observed, and the picked-up signals of the sensory ring remain unchanged after the sensory ring recovers from being folded (Fig. 7C).

Fig. 7. Demonstration of a sensory ring for multimodal and multisensor chip-free and battery-free sensing of airway mucus thickness and temperature.

(A) Optical images of the sensory ring. Scale bars, 5 mm. UV, ultraviolet. (B) Optical images when the sensory ring is folded. Scale bar, 5 mm. (C) S₁₁ spectrum of the sensory ring before and after folding. (D) Experiment setup of the sensory ring inside tissue. (E) Optical image of the sensory ring with mucus added. Scale bar, 1 cm. (F) Calibration of the mucus layer thickness sensor. (G) Frequency signals of the three mucus thickness sensors connected to the circuit sequentially by the magnetic switch, as well as the measured mucus thickness at three different locations by the three sensors. (H) Optical image of the temperature sensor and its validation experimental setup and a zoomed-in optical image of the temperature sensor. Scale bar, 2 mm. (I) Calibrated model of the temperature sensor. (J) Frequency signal as a function of time during temperature sensing. (K) Validation of temperature sensing. (L) X-ray image of the sensory ring inside a sheep trachea. Scale bar, 10 mm. (M) Bronchoscopic view of the sensory ring with mucus thickness sensors and a temperature sensor. Scale bar, 5 mm. (N) Fluorescence images of the cells in the control and sensory ring groups. Scale bar, 100 μm.

We further set up an ex vivo experiment using porcine and ovine tissues (fig. S28) to demonstrate the sensing function of the proposed device (Fig. 7D). The optical images of measuring a mucus layer thickness are shown in Fig. 7E. The mucus layer thickness sensor is composed of two parallel plates that form into a capacitor which is calibrated to measure the mucus thickness. With a calibration model shown in Fig. 7F, when mucus is added to the sensor that is connected to the circuit sequentially, the capacitance of the corresponding mucus layer thickness sensor increases as demonstrated in Fig. 7G. In each phase, the resonant frequency experiences a drop, indicating a rise in the circuit’s capacitance.

In addition, a capacitive temperature sensor is produced by two face-to-face copper plates clamping a thin layer of temperature-sensitive material (44). The temperature sensor is positioned at the anterior ring, where airflow-induced temperature gradients are greatest, providing the most physiologically representative measurements for detecting airway inflammation. The temperature sensor and validation setup are shown in Fig. 7H, where a heating pad surrounds a segment of sheep trachea tissue to increase the internal temperature, and a thermal couple is inserted between the ring and the tissue to collect the ground truth data. With a calibrated model given in Fig. 7I, Fig. 7J shows the picked up $S_{11}$-frequency signals in this process using the temperature sensor. Subsequently, the measured temperature by the sensory ring is compared with the data collected by the thermal couple, which shows a relatively good accuracy as shown in Fig. 7K.

Furthermore, the sensory ring is tested in a sheep trachea and remotely actuated by the portable magnetic actuation system to connect to different channels. The components of the sensory ring can be fully visualized using medical imaging such as x-ray imaging (Fig. 7L) for patency monitoring in case of a problem. The specific channel type for each channel corresponds to a unique magnetic actuation angle, allowing unambiguous assignment. A sensor ring is first deployed in a sheep trachea and visualized under x-ray as shown in Fig. 7L and under a bronchoscope in Fig. 7M. Then, when integrated with an airway stent, the sensor ring is still clear to see under x-ray imaging with the magnetic switch status clearly shown in the sequential images (fig. S29), confirming the ability of monitoring the device conditions when necessary.

Last, as the sensory ring is used inside a humid environment, it needs to be well encapsulated for electrical insulation and biocompatibility. The magnetic switch is encapsulated in a biocompatible case (fig. S30), while the circuit part is covered by polyimide (PI) tapes with their edges sealed with PDMS. To validate the insulation, the sensory ring is filled with mucus inside. The pickup signals are similar before and after the mucus are added, indicating a relatively good insulation performance. To validate the biocompatibility, the encapsulated sensory ring is submerged inside cell solutions for cell viability test. As shown in Fig. 7N, the encapsulation and sealing ensure cell viability. A 7-day soaking in PBS and follow-up biocompatibility test further validate the encapsulation and biocompatibility of the device (fig. S30).

DISCUSSION

In summary, we have reported a remotely controllable miniature magnetic switch that enables selective channel switching which is generic for achieving multimodal sensing in chip-free and battery-free sensing devices. The magnetic switch functions through a cantilever beam mechanism that responds to external magnetic fields by anchoring, rotating, and toggling between channels. When integrated with an induction coil and capacitive sensors, the system produces resonance-based signals that can be selectively and wirelessly detected using a VNA, allowing remote monitoring of multiple physiological properties. With the magnetic switch, we have demonstrated a chip-free wireless sensing device integrating a tissue stiffness sensor for monitoring airway tissue stiffness conditions, which requires both active deformation and force sensing. Consistent sensing performance has been demonstrated across samples with a range of elastic modulus. In addition, the device has also been shown to have multilocation sensing of tissue pressure and multimodal sensing of mucus properties including layer thickness and temperature, capabilities that are not achievable with conventional radio frequency–based sensors. Therefore, the proposed magnetic switch and the resulting chip-free sensing framework have the potential to enable next-generation miniaturized, robust implantable devices for continuous and multimodal monitoring of airway disease conditions.

The potential limitations of the proposed device include sensing distance, sensing accuracy within animal models, and deployment strategy. With the current VNA providing an output power of −10 to 0 dBm, the device can be detected at ~2.5 cm inside tracheal tissue. However, this range may not be sufficient for human applications, as the distance between the trachea and the chest surface can exceed 3 cm. One possible solution is to use a higher-power and more sensitive VNA capable of transmitting stronger signals. For instance, increasing the VNA power to be −10 from −30 dBm, a reliable resonance detection up to 3.3 cm has been shown to be achieved (figs. S31 and S32), covering the expected human chest wall thickness for adult airway implants. Increasing the VNA power allows a larger signal-to-noise ratio (figs. S33 and S34), allowing a larger detection distance. Simulation of the induction coupling–based sensing also shows that increasing the VNA sensitivity of the sweeping frequency and $S_{11}$ value could allow increasing the sensing distance (fig. S35). Regarding sensing accuracy, a notable signal shift was observed when transitioning from phantom models to actual tissue models. Although the $S_{11}$ response as a function of sweeping frequency maintains a similar slope, an offset is present. In more complex body environments, the pickup signals may exhibit varying slope angles due to heterogeneous tissue composition and dielectric properties. In addition, the device demonstrates favorable foldability, suggesting strong potential for minimally invasive deployment using a customized delivery tool (22).

Toward clinical applications, several key directions will be pursued in future work. First, we will address challenges in wireless communication by mitigating signal transmission disturbances caused by tissue interference and environmental noises. Efforts will also focus on extending the detection range (45) through systematic optimization of the VNA power and sensitivity, pickup coil, and onboard inductor coil, including refinements in geometry, material selection, and resonance tuning. Second, systematic in vivo studies will be conducted to assess the device’s biocompatibility, safety, and long-term durability under physiological conditions, ensuring reliable operation over extended implantation periods. Last, comprehensive animal studies will be undertaken to validate both the sensing functions and the performance of the magnetic switch under physiologically relevant conditions. These investigations will not only benchmark device reliability but also provide critical translational data to guide future clinical trials.

The proposed magnetic switch and multimodal sensing framework allow broader clinical applications with the potential to integrate the sensory system in other implantable stents. Beyond respiratory care, the magnetic switch and capacitive sensing framework also hold promise for integration into cardiovascular and orthopedic implants, paving the way for more personalized and adaptive biomedical devices. For example, coupling cardiovascular stents could enable continuous monitoring of vascular pressure for the management of hypertension and related cardiovascular diseases (46). Similarly, integration with esophageal stents (47, 48) could provide real-time feedback on esophageal tissue mechanics, supporting early diagnosis and treatment of disorders such as strictures or reflux disease (49, 50). Therefore, the proposed sensory ring has the potential to provide a versatile and robust platform technology for next-generation implantable medical devices.

MATERIALS AND METHODS

Fabrication of the magnetic switch

The magnetic switch consists of a conductive base and a probe. The substrate of the base was made of a single-layer Flame Retardant 4 (FR4), which was replaced with Pyralux (DuPont) on the sensory ring. The substrate was patterned with four conductive pads, each representing a channel, and a conductive middle pad. The channel pads and the middle pads were separated but were to be connected when a specific channel was selected. The center of the middle hole is left with a hole of 0.3 mm in diameter to install a steel pole. Two round copper pieces, 0.9 and 0.5 mm in diameter and both 0.3 mm in thickness, were stacked on top of the middle pad, which were electrically connected by solder paste. Those copper pieces helped maintain a smooth rotation of the probe. After the probe was installed, two additional copper pieces were stacked to prevent it from falling out. These pieces were 0.1-mm thick and 0.5 mm in diameter and 0.3-mm thick and 0.5 mm in diameter, respectively. Last, two stoppers were bonded to one side of the conductive base to constrain the probe’s rotation.

To fabricate the probe, three copper pieces were first cut out from a copper board (0.3 mm in thickness). One has a large hole, 0.6 mm in diameter, to be placed onto the middle pad of the base. The other two were soldered together to contact the channel pads. These two parts were then connected by a layer of Pyralux for electrical conductivity and covered with another layer of PDMS with a thickness of 0.12 mm. Two permanent magnets (NdFeB, 1 mm in diameter, 1 mm in length) were glued to the front and rear sides of the probe and magnetized in the axial direction for magnetic actuation. In addition to this four-channel switch design, more than four channels could be integrated on a sensory ring to allow integrating more sensors as shown in fig. S36.

Fabrication of the capacitive stiffness sensor

The tissue stiffness sensor is composed of a conductive base and a magnetic layer. The conductive base was fabricated from a flexible copper film (Pyralux) and was patterned using an LPKF U4 machine (LPKF Laser & Electronics North America). This layer was then spin-coated with PDMS to provide electrical insulation (3000 rpm, 3 min). For the magnetic layer, a layer of magnetic composite (0.8 mm in thickness), was affixed to the nonconductive side of a piece of Pyralux to enhance stiffness and mitigate undesired bending. Four permanent magnets with the same magnetization, each measuring 1 mm in height and 1 mm in diameter, were glued onto the magnetic composite. A cylindrical steel probe (0.15 mm in diameter) was attached to the free end of the sensor. The probe penetrated through a hole in the base to contact with the tissue. The conductive side of the magnetic layer was then soldered to the conductive base for capacitance measurement.

Fabrication of the capacitive magnetic field sensor

The fabrication process for the magnetic sensor mirrors that of the tissue stiffness sensor but does not incorporate permanent magnets or the probe. The body of the magnetic sensor is made of magnetic composite (PDMS:NdFeB = 1:2, by weight) with a thickness of 0.25 mm. A piece of magnetic composite was added to the tip to enhance the sensitivity with the same thickness. Both sensors were affixed to a flexible circuit made from Pyralux. The copper pads that connect to the sensors were spin-coated with PDMS to ensure electrical insulation. Liquid dielectrics were introduced into the gap between the capacitive layer of the sensor and the pads of the circuit.

Fabrication of the capacitive pressure sensor

The pressure sensor was fabricated by stacking two layers of Pyralux with their conductive sides facing each other. PDMS is spin-coated to one side of the copper as the dielectric material, with 2000 rpm for 1 min. The PDMS was cured on a hot plate with 100°C for 20 min. The resulting sandwich structure measured 4 mm by 2 mm by 0.5 mm. The sensor was encapsulated along the sides to ensure electrical insulation and biocompatibility and subsequently bonded to the surface of the sensory ring. See fig. S37 for the illustration of the working principle of the pressure sensor.

Fabrication of the capacitive mucus layer thickness sensor

The electrodes of a capacitive mucus layer thickness sensor were laser patterned by LPKF on a Pyralux sheet. Each electrode is 5 mm in length and 1 mm in width, separated by a 0.1-mm gap. The electrodes were then spin-coated with PDMS at 2000 rpm for 5 min. The PDMS layer was cured at 100°C for 20 min. The electrodes were connected to the substrate circuit with silver paste and then covered by liquid tape for electrical insulation. See fig. S37 for the illustration of the working principle of the mucus layer thickness sensor.

Fabrication of the capacitive mucus temperature sensor

DI water (10 g) and glycerol (5 g) were mixed under magnetic stirring for 10 min to prepare the solvent. Polyvinyl alcohol (PVA; 2.6 g) and sodium chloride (NaCl; 50 mg) were then gradually added. The mixture was heated at 75°C under magnetic stirring for 3 hours to ensure complete dissolution of PVA and yield a transparent solution. To fabricate thin films for the dielectric layer, the solution was cast onto an acrylic board with two polyethylene terephthalate tape layers serving as spacers on both sides and shaved into a film. The film was subsequently heated on a hot plate at 75°C for 1 hour. Last, 3 mm–by–3 mm square dielectric layers were cut from the film using an ultraviolet laser machine (LPKF U4, LPKF Laser & Electronics North America). See fig. S37 for the illustration of the working principle of the mucus temperature sensor.

Fabrication of the inductor for the sensory ring

The sensory ring consists of a flexible circuit to connect with the switch and sensors, an inductor, and a flexible ring-shaped back layer. The flexible circuit was made of Pyralux with 45 μm in thickness. The pattern of the circuit came from an etching process, where the mask was made of PI tapes and patterned by LPKF U4. The inductor for wireless communication was also made of Pyralux with an etching process. The inductor that has five turns is 1.5 mm in width. The flexible back layer was made by PDMS. A sacrificial mold was three-dimensional (3D) printed from PVA. Then, liquid PDMS (10:1) was degassed and poured into the mold. The PDMS was then cured on a hot plate at 100°C for 24 hours. The PVA mold was then dissolved by water to release the PDMS ring. The flexible circuit and the inductor were connected by copper wires through holes on the back layer and were bonded to the back layer with PDMS. The ring was wrapped in PI tapes on the outside for electrical insulation.

Validation of the elastic modulus of synthetic material and tissues

To validate the stiffness sensing capabilities, testing samples of varying stiffness levels were prepared. These samples were made from Ecoflex 00-30 and Slacker in different weight ratios, with Slacker used to soften the polymer for a lower stiffness. The indentation of the prepared synthetic soft materials was performed using a benchtop test tool (Hysitron BioSoft In-Situ Indenter, Bruker AG). During the experiments, a sapphire spherical probe with a diameter of 400 μm was used. The tip approached and indented the sample at a constant rate of 2 μm/s over a 20-s loading period, followed by a 240-s hold at the maximum indentation depth. Subsequently, the indenter was retracted at the same rate of 2 μm/s during the unloading phase. A Hertzian contact model was used for estimating stiffness from the loading force and displacement curves.

Procedure of the biocompatibility test

Devices were fully submerged in PBS and incubated in a 37°C water bath for 7 days to simulate physiological preconditioning. Human fibroblast cells (CRL-2522), obtained from the American Type Culture Collection, were cultured in minimum essential medium (Gibco) supplemented with 10% fetal bovine serum (Gibco) under standard conditions (37°C, 5% CO₂, humidified atmosphere). Fibroblasts were seeded into six-well plates at a density of 5 × 10⁵ cells per well. After 24 hours, the preconditioned devices were transferred into the wells and cocultured with the cells for an additional 72 hours. Cell viability was assessed using the LIVE/DEAD Viability/Cytotoxicity Kit (Invitrogen, Thermo Fisher Scientific), and fluorescence signals were visualized using a fluorescence microscope (Eclipse Ti-E, Nikon, Japan).

Preparation of porcine tissues for validation

Porcine and ovine trachea tissues were prepared for testing the tissue stiffness sensing function of the sensor. The porcine tissues were obtained from a local slaughterhouse, and the ovine trachea tissues were obtained from Vanderbilt University Medical Center. The tissues were kept clean and cool in a freezer before experiments. During the experiments, the tissues were placed on a petri dish after being warmed to room temperature.

VNA-based signal pick-up

A VNA (LibreVNA, Seesii) was used to communicate with the sensor wirelessly. The LibreVNA antenna analyzer provides an ultrawide frequency range of 100 kHz to 6 GHz. Equipped with three Analog-Digital-Converters (ADC) for simultaneous data sampling and field programmable gate arrays for signal processing, it can perform full dual-port measurements at more than 10,000 points in less than 1 s. A readout coil was positioned ~2.5 cm to 4 cm above the sensor inductor to transmit and receive signals. The VNA was connected to a personal computer via USB, and the acquired signals were logged and further analyzed in MATLAB 2025a (MathWorks Inc.).

Portable magnetic actuation system

The system comprises an upper housing, a base, motor mounting board, dual linear motors, shaft stabilizer, and motor mount for the magnetic actuation module. The system housing was fabricated by 3D printing polylactic acid (PLA). For the magnetic actuation unit, a single NdFeB cube magnet with a side length of 2.5 cm (N45 grade) was housed within a PLA box, designed to connect via flange to a motor. The rotation of the permanent magnet was achieved using a NEMA 8 stepper motor controlled by a dc motor driver (L298N). Meanwhile, a servo motor (HobbyPark 20KG high torque servo) was used to regulate the xy planar orientation of the magnet by rotating the motor mounting board from 0° to 180°. Two linear 10-mm dc motors connect the upper housing to the base and are actuated to adjust the distance of the magnet in the z axis. To manage the operation of the stepper motor, servo (yaw) motor and dual linear motors, a wirelessly controlled embedded controller (Arduino MEGA) was used. Power was supplied by a dc power supply (Drok 24V). The magnetic field at the sensor location was controlled by rotating and translating the magnet inside the magnetic actuation system.

Supplementary Materials

The PDF file includes:

Tables S1 and S2

Supplementary Notes S1 to S4

Figs. S1 to S37

Legends for movies S1 to S6

References

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S6