Wireless knee joint monitoring using biodegradable single-ended pressure sensor in osteoarthritis management

Abstract

Knee joint loading regulates cartilage regeneration and degeneration, influencing both healing and osteoarthritis (OA) progression. Although defining the optimal range of loading is essential, in vivo measurement remains challenging because current nonbiodegradable sensors require removal surgeries, and computational models provide only indirect estimates. Here, we report a fully biodegradable, implantable piezoelectric pressure sensor composed of biocompatible materials and designed with a single-ended configuration, consisting of only one electrode. This configuration maintained stable performance for more than 2 months, which, to our knowledge, is among the longest functional lifetimes reported for biodegradable pressure sensors. Ex vivo studies simulating the in vivo environment demonstrated accurate monitoring of intracranial pressure and knee movement, while in vivo implantation in a rabbit model confirmed both robust biocompatibility and reliable monitoring of joint loading during natural activities. This platform could enable feedback-guided, personalized rehabilitation regimes for OA management by identifying optimal joint loading conditions.

Biodegradable piezoelectric sensor with wireless module enables real-time joint load monitoring without removal surgery.

INTRODUCTION

In medical applications, the real-time monitoring of tissue-level mechanical forces is fundamental for understanding and managing physiological processes, as well as diagnosing/preventing diseases and internal pressure build-ups in organs such as the brain, eyes, blood vessels, and joints (1–4). Within the musculoskeletal system, the knee joint exhibits particularly complex biomechanics and is highly susceptible to injury and degeneration (4).

Knee joint loading plays a critical role in maintaining cartilage health and addressing the challenges of cartilage degeneration and repair. Mechanical forces applied to the knee joint influence cartilage structure and function, substantially affecting both the healing process and the progression of osteoarthritis (OA), a common disease, which is manifested by the cartilage damages and affects millions of Americans every year (5, 6). When applied within a proper range, joint loading promotes cartilage integrity, supports synovial fluid production for lubrication, and enhances nutrient transport (7–10). In contrast, excessive loading can cause cartilage cell death, matrix degeneration, and impaired nutrient supply (4, 11, 12). Thus, identifying an optimal range of mechanical stimulation is essential to promote cartilage regeneration while avoiding damage caused by overload. Measuring joint contact forces can therefore aid in the early detection of OA, optimize rehabilitation protocols, and improve outcomes in procedures such as total knee arthroplasty (13–15). In that regard, implantable sensors that monitor intra-articular pressure or surface loading in real time could provide critical feedback for clinicians and patients during postsurgical recovery, physical rehabilitation, or disease progression monitoring.

Unfortunately, current approaches to measuring joint knee pressure struggle with many problems. Available joint loading measurements primarily rely on indirect estimation methods such as ground reaction forces (GRFs), joint kinematics, or electromyography (EMG) (16–19). While these methods provide general insights into joint mechanics, they are limited by their reliance on controlled laboratory settings, their infeasibility to capture localized loading at the tissue level, and particularly their inherent inaccuracy due to indirect measurement (19–21). GRF-based estimates depend heavily on motion capture accuracy and are sensitive to small measurement errors (20, 22, 23). Joint kinematics suffer from soft tissue artifacts and do not directly reflect internal joint forces or contact pressures (20, 22, 24). EMG signals do not directly correspond to force generation and are influenced by factors such as skin impedance and muscle fatigue (22, 25, 26). Biomechanical modeling integrates GRF, EMG, and kinematic data with anatomical assumptions but often relies on generalized simplifications that often fail to reflect subject-specific joint mechanics (20, 22, 27). The inability to directly measure individual knee joint loading in real time poses a major barrier to understanding the role of mechanical forces in cartilage healing and limits the development of load-based therapeutic interventions.

In response to this limitation, several efforts have been attempted to develop implantable pressure sensors using piezoelectric materials, which convert mechanical force into electrical signals, thereby enabling sensitive and direct force readouts (28–30). However, many of these sensors are constructed from nonbiodegradable piezoelectric materials, and their safety has not been thoroughly demonstrated. Despite their ability to provide direct measurement, their nonbiodegradable nature requires a secondary surgical procedure for retrieval, increasing the risk of infection, tissue damage, and patient discomfort. Furthermore, the long-term presence of nonbiodegradable materials may provoke chronic inflammation, and failure of the sensor’s encapsulation—especially under repetitive mechanical stress in the knee joint—can expose surrounding tissue to materials that may not be fully safe for biological contact (31–33).

Biodegradable organic piezoelectric materials such as poly-l-lactic acid (PLLA), a common medical polymer used extensively in Food and Drug Administration (FDA)–approved erodible implants, have recently emerged as promising alternatives to conventional, nonbiodegradable piezoelectric materials, especially for biomedical applications (34–42). PLLA offers excellent biocompatibility and biodegradability. Our prior studies have shown that when processed into electrospun nanofibers, PLLA exhibits enhanced piezoelectric properties through molecular dipole alignment and crystalline domain formation, resulting in flexible fibrous mats that conform well to curved, moving surfaces (35–40). While we have demonstrated the ability to create biodegradable piezoelectric sensors and transducers (34, 35), several critical challenges remain in adapting them for reliable pressure sensing applications. A primary concern for PLLA sensors, as well as other reported biodegradable pressure sensors (43–46), is ensuring that the sensor’s functional lifetime aligns with the duration of the intended monitoring period, particularly in dynamic biological environments. However, premature degradation or signal drift, frequently due to short circuiting between the pair of two electrodes, can undermine device accuracy before structural failure. To date, the longest reported functional lifetime of a bioresorbable pressure sensor is 25 days, demonstrated in a study targeting intracranial pressure (ICP) monitoring (43). However, the diaphragm-based pressure sensor architecture used in that application may not be suitable for the knee joint monitoring, where substantially higher mechanical loads are present and can easily break down the diaphragm structure.

To address these challenges, here, we present a new biodegradable piezoelectric pressure sensor designed with a single-ended configuration, consisting of only one electrode. This single-ended configuration biodegradable piezoelectric pressure sensor provides, to our knowledge, the first demonstration of wireless, continuous, and direct monitoring of joint loading over a lifetime of up to ~8 weeks (2 months), after which it safely degrades. We used the safe and biodegradable piezoelectric PLLA nanofiber mat and integrated a single electrode only on one surface of the film to collect the generated charges. Departing from the conventional differential configuration, which relies on two electrodes to measure charge as a potential difference, this single-ended architecture eliminates the risk of short circuiting arising from direct contact between two top-bottom electrodes in the conventional biodegradable sensors due to mechanical tearing of the piezoelectric material or through conductive pathways formed by ionic body fluids, thus extending the functional lifetime. Furthermore, this structural simplification reduces fabrication complexity, minimizes overall sensor thickness, and enhances mechanical flexibility at the tissue interface.

We fabricated the sensor by using the piezoelectric PLLA nanofiber film in contact with a biodegradable molybdenum (Mo) electrode, encapsulated inside a neutral polylactic acid (PLA) film. We systematically evaluated the sensor’s functional lifetime under both simulated physiological conditions and in vivo animal models. The device reliably operated for up to 9 weeks in a physiologically simulated environment and maintained stable functionality for 7 weeks following implantation in live animals. The sensor effectively measured physiologically relevant internal pressures, including ICP and, most notably, dynamic joint loading in both explanted rabbit knee joints and live rabbits. In live rabbit, it successfully captured real-time joint loading signals corresponding to the animal’s natural movement. We further verified the sensor’s biocompatibility through in vitro analysis and histological examination of the articular cartilage directly interfaced with the implant. This work highlights the piezoelectric PLLA sensor’s potential as a safe, reliable, and biodegradable platform for real-time joint loading monitoring in dynamic musculoskeletal environments.

RESULTS AND DISCUSSION

Materials and design

Conceptual illustration of a biodegradable pressure sensor for monitoring knee joint loading is shown in Fig. 1A. The sensor is implanted in the knee joint of a patient undergoing cartilage repair or anterior cruciate ligament reconstruction. For more than 2 months, this device wirelessly records and transmits pressure data during daily activities such as walking, stair climbing, and rehabilitation exercises. This information helps clinicians track joint loading and adjust rehabilitation as needed. After healing is complete, the sensor biodegrades naturally, eliminating the need for surgical removal and enabling the patient to resume normal activities.

Fig. 1. Overview of the biodegradable PLLA pressure sensor.

(A) Conceptual illustration of the envisioned use case, showing a device implanted in the knee joint to directly monitor joint loading. (B and C) Schematic illustration of the PLLA pressure sensor with a single-ended configuration and a differential configuration.

The schematic illustration in Fig. 1 (B and C) depicts the structure of the single-ended configuration device, in contrast to the conventional differential configuration commonly used in pressure sensors. This sensor consists of a PLLA piezoelectric nanofiber film, a single Mo electrode, and PLA encapsulation layers. Both PLA and PLLA are widely used in FDA-approved medical applications such as tissue scaffolds, bone screws, and drug delivery carriers (41, 42). Their degradation by-products, such as carbon dioxide, water, and lactic acid, are naturally metabolized and eliminated through physiological processes. Mo, an essential element in human nutrition, has also been safely used in a range of medical implants (47).

Traditional two-electrode biodegradable devices generally show functional lifetimes of 4 to 25 days (34, 35, 43). In particular, biodegradable PLLA-based pressure sensors with a differential configuration typically exhibit functional lifetime of less than 2 weeks (34, 35). Although the encapsulation layer properties such as thickness or molecular weight can be tuned to extend performance, device failure in two-electrode systems is ultimately dominated by leakage pathways that form between electrodes. These pathways commonly arise from damage to the piezoelectric film under repeated mechanical loading or from body fluid absorption or infiltration. By contrast, the single-ended configuration offers a simpler architecture, in which the signal generated by the PLLA piezoelectric film is measured relative to a common system ground. This design eliminates critical failure modes of differential sensors, such as short circuits between electrodes or failures at dual signal wire connections, which are especially problematic in biodegradable devices. In the single-ended configuration, the sensor can continue functioning even if localized leakage paths develop in the piezoelectric film, as long as the defects do not expand sufficiently to degrade overall performance. With this single-ended approach, our biodegradable pressure sensor extended its functional lifetime by removing the risk of interelectrode short circuits and simplifying the wiring from two leads to one while also retaining the ability to naturally and safely degrade after completing its intended functional period.

Device characterization and ex vivo performance

In organic materials, piezoelectricity originates from the alignment of molecular dipoles. For PLLA, this alignment can be introduced through electrospinning, a technique that uses electrical forces to draw charged polymer solutions into ultrafine fibers (Fig. 2A). This extent of fiber alignment is determined by the rotational speed of the collecting drum. Figure 2B shows the scanning electron microscopy (SEM) images of the fabricated PLLA nanofibers (see Materials and Methods for details). In the aligned PLLA nanofiber film (Fig. 2B, top), the carbon-oxygen double bonds (C═O) in the helical PLLA backbone are oriented in the same direction, giving rise to a net polarization and a pronounced shear piezoelectric response under applied force. These films, with thicknesses of 22 to 40 μm, display highly uniform alignment. In contrast, the randomly deposited PLLA nanofiber film (Fig. 2B, bottom) shows little-to-no piezoelectricity due to dipole cancellation from disordered orientation. X-ray diffraction (XRD) analysis further reveals that the aligned nanofiber films are enriched in the (200) and (110) crystal phases, corresponding to the β-form crystal structure (Fig. 2C), which is the piezoelectric phase of PLLA (48). Our previous work (35) showed that fiber alignment across the film improves with increasing the rotational speed of the collecting drum. In addition, both the macroscopic orientation of PLLA fibers and the molecular alignment are influenced by the jet speed, which is determined by the applied voltage. By adjusting these parameters, we were able to optimize the piezoelectric performance of the nanofiber films. For comparison, the nanofibers spun at 300 rpm were selected as the control group, showing negligible piezoelectric activity, whereas fibers spun at 4000 rpm exhibited enhanced piezoelectric properties.

Fig. 2. Characterization of piezoelectric PLLA nanofiber and the biodegradable pressure sensor.

(A) Photograph of electrospun PLLA nanofiber film. (B) SEM images of aligned piezoelectric PLLA nanofibers (top) and randomly oriented nonpiezoelectric PLLA nanofibers (bottom). Scale bars, 10 μm. (C) XRD spectra of aligned and random PLLA nanofiber films. (D) Representative image of a fabricated biodegradable pressure sensor. Scale bar, 5 mm. (E) Piezoelectric responses of aligned and random PLLA nanofiber films under varying applied forces. Dashed lines indicate linear fits based on measurements (n = 5). (F) Relative resistance changes of Mo electrodes with and without PLA encapsulation during immersion in PBS at 37°C (n = 3). (G) Accelerated degradation of the biodegradable pressure sensor in PBS at 60°C (n = 3). (H) Ex vivo knee joint loading test in a rabbit cadaver using the PLLA pressure sensor and a commercial pressure sensor. (I) Comparison of knee joint loading signals recorded by the PLLA sensor and the commercial sensor. a.u., arbitrary units.

We fabricated the biodegradable pressure sensor with a single-ended configuration (Fig. 1B), as illustrated in Fig. 2D. The device consisted of a piezoelectric PLLA nanofiber film (5 mm by 10 mm) sandwiched with a single Mo electrode between PLA encapsulation layers. To characterize device performance, the sensor response was quantified under a series of controlled mechanical forces (Fig. 2E). Sensors fabricated with the film spun at 300 rpm, which exhibit negligible piezoelectric response due to limited molecular alignment, served as nonpiezoelectric controls. In contrast, sensors with the film spun at 4000 rpm showed clear responses to varying applied mechanical forces, producing output charges that scaled proportionally with the magnitude of the applied forces. The film spun at 4000 rpm exhibited an estimated piezoelectric coefficient (d₁₄) of ~14.5 pC/N, determined from the sensitivity (slope) of the linear fit (R² = 99.06%), while the 300 rpm samples could not be reliably estimated. In addition to sensitivity, the sensor demonstrated a pressure detection range of 1 to ~696 kPa, minimal hysteresis (2.01%), and stable output across the physiological frequency range (0.8 to 3 Hz) (fig. S1). Collectively, these results also confirm that the recorded signals originate from the intrinsic piezoelectric properties of the aligned nanofibers, rather than triboelectric artifacts or motion-induced noise.

For implantation applications, the long-term stability of biodegradable device is critical, as their performance naturally deteriorates over time through hydrolysis of the polymer matrix or oxidation of the metallic electrode. To assess these degradations, we first monitored the resistance of Mo electrodes under simulated physiological conditions (Fig. 2F). When encapsulated within PLA, a Mo electrode immersed in phosphate-buffered saline (PBS) at 37°C maintained stable resistance for nearly 10 weeks, whereas a bare Mo electrode exhibited a gradual increase in resistance, reflecting progressive oxidation. These findings demonstrate that PLA encapsulation effectively delays electrode degradation, thereby mitigating one of the major contributors to functional decline in biodegradable sensors.

The sensor is intentionally designed to degrade after fulfilling its functional role, minimizing the need for secondary surgical intervention. Accelerated degradation studies performed by immersing the sensors in 1× PBS at 60°C further illustrate this process (Fig. 2G). Under these conditions, structural degradation became evident after around 14 weeks, with complete disintegration into small fragments by week 32 (fig. S2). Fatigue testing (fig. S3) demonstrated that the encapsulated sensor encapsulation maintained structural integrity and mechanically stability after 60,000 cycles at 3 Hz under physiological pressures of 400 kPa. Beyond 44,000 cycles, a gradual decline in piezoelectric output was observed, due to cumulative mechanical stress within the PLLA film. However, ≥90% signal retention was maintained until 52,000 cycles, with no evidence of encapsulation failure. Together, these results validate both the functional performance and the controlled biodegradability of the single-electrode pressure sensor, highlighting its potential for transient biomedical applications.

For a proof-of-concept demonstration of potential biomedical applications, we used the pressure sensor fabricated with 4000 rpm nanofiber to monitor physiological pressure in an ex vivo cadaveric model mimicking the in vivo environment. As illustrated in Fig. 2H, both the biodegradable sensor and a commercial pressure sensor were overlapped and implanted in the knee joint of a rabbit cadaver (see Materials and Methods for details). The commercial sensor served as the reference, demonstrating strong linear agreement with the biodegradable sensor (fig. S1D). The leg was manually flexed and extended, while output signals from both sensors were recorded simultaneously. Unlike the leg of a live rabbit during sitting or hopping, the cadaveric leg was fully relaxed and did not experience GRFs, resulting in lower overall magnitudes of joint loading during movement. Specifically, when the knee was flexed, the posterior femoral condyle experienced a modest pressure increase (~1.2 kPa) as the femoral and tibial surfaces came closer together (Fig. 2I). In contrast, when the knee was extended, the pressure sharply decreased. This trend was consistently observed in both the biodegradable sensor and the commercial sensor, with closely matching pressure signals. These findings validate the ability of the biodegradable sensor to detect physiologically relevant joint pressure changes and highlight its potential application for monitoring of knee joint loading, which is particularly relevant in the context of cartilage repair and rehabilitation. Beyond the primary application of joint loading applications, we further validated the sensor’s utility in a rat cadaver model. Specifically, both the biodegradable sensor and a commercial pressure sensor were sealed onto bilateral parietal defects with sealant in a rat cadaver. This setup allowed for the direct benchmarking of the biodegradable sensor against the commercial sensor as a reference standard for ICP monitoring (fig. S4; see Materials and Methods for details). To simulate variations in ICP, PBS was manually injected through a channel into the cranial cavity. In clinical practice, the normal ICP range for adults lies between 5 and 15 mmHg, with transient fluctuations occurring throughout the day due to routine activities (49). However, ICP values exceeding 20 mmHg are considered pathological and may indicate conditions such as brain tumors, stroke, or traumatic brain injury (1, 49). The biodegradable sensor successfully detected simulated ICP values ranging from 3.0 to 39.5 mmHg, closely matching the readings of the calibrated commercial pressure sensor. These results demonstrate that the sensing performance of the biodegradable sensor is maintained across different anatomical environments, confirming its broad potential for real-time monitoring in various postoperative scenarios beyond orthopedic care.

In vitro biocompatibility assessment

We evaluated the biocompatibility of the PLLA pressure sensor through in vitro assays. First, a hemolysis test was performed using rabbit blood. The major components of the sensor, PLLA and Mo, tested individually and in combination, exhibited hemolysis rates below 2%, thereby meeting the safety criteria for medical devices specified in ISO 10993-4 by the International Organization for Standardization (ISO) (Fig. 3A).

Fig. 3. In vitro biocompatibility assessment of the biodegradable pressure sensor and in vitro chondrogenesis study.

(A) Hemolysis testing of PLLA, Mo, and the combination of PLLA and Mo, with normal saline (negative control) and deionized water (DI; positive control) (n = 5). Dashed line indicates the ISO threshold for implantable devices (< 2%). (B) Viability of BMSCs seeded with the biodegradable sensor compared with control (cells only) at 1, 3, 7, and 14 days (n = 5; data expressed as mean ± SD). (C) Live/dead cell imaging assay of BMSCs incubated with the biodegradable sensor and without the sensor (control). Live cells are stained green, and dead cells are stained red. Statistical comparisons were performed using an unpaired t test (n = 5). Scale bars, 500 μm. (D) Relative gene expression of chondrogenic markers COL2A1, ACAN, and SOX9 in BMSCs cultured with the biodegradable sensor and without the sensor (control), along with glycosaminoglycans (GAGs) content. Data are expressed as mean ± SD. Statistical comparisons were performed using an unpaired t test (n = 5). *P < 0.05 and **P < 0.01. ns, not significant.

Second, we assessed cytocompatibility by culturing rabbit bone marrow–derived mesenchymal stem cells (BMSCs) in well plates with the sensor or without the sensor (control) for 14 days (see Materials and Methods for details). As shown in Fig. 3B, no significant differences in cell viability were observed between the groups. To further visualize the cellular response, we performed live/dead cell imaging assays. In Fig. 3C, live cells stained green with Calcein acetoxymethyl ester (Calcein AM), indicating intracellular esterase activity, while dead cells stained red with BOBO-3 iodide, indicating nucleic acid exposure through damaged membranes. Comparable ratios of live to dead cells were observed in the sensor and control (only cells) groups, indicating no cytotoxic effect from the sensor.

Last, to assess potential effects of the sensor on cartilage health, we investigated the chondrogenic differentiation of BMSCs seeded with the biodegradable sensor samples and without the sensor (control). We analyzed the gene expression analysis of key chondrogenic markers—COL2A1, ACAN, and SOX9—in both the sensor and control groups (Fig. 3D). The results showed that the presence of the sensor promoted up-regulation of these markers, suggesting enhanced chondrogenic activity. PLLA degradation generates lactic acid, which can dissociate into lactate under physiological conditions (41, 50). Lactate has been reported to influence cellular metabolism and may participate in signaling pathways related to HIF-1α, a transcription factor involved in cartilage homeostasis (51, 52). The observed transcriptional changes are consistent with these previously reported biological responses, although the precise underlying mechanisms require further investigation. At the protein level, glycosaminoglycans (GAGs) expression did not show significant differences between groups. These findings indicate that the presence of the sensor in the joint does not negatively affect cartilage health and may even promote chondrogenic differentiation at the gene expression level, supporting its potential for safe application in cartilage-related therapies.

In vivo rat studies for evaluating device functionality and stability

To evaluate functional lifetime, we fabricated both single-ended and differential configurations, immersed them in PBS at 37°C, and monitored their outputs under mechanical loading. Testing continued until each sensor ceased proper function. The single-ended configuration was grounded to the system and initially exhibited a noise level comparable to that of the differential configuration (fig. S5). As shown in Fig. 4A, the differential configuration failed within 2 weeks, consistent with the 1-week lifetimes reported in a previous study (35). In contrast, the single-ended configuration demonstrated stable performance for up to 9 weeks, exceeding 2 months of functional lifetime. The relative performance (%) of the single-ended sensors over time (n = 3) was normalized to their initial values at week 0 (Fig. 4B). Although the average output decreased by about 10% at week 9, the change was not statistically significant, and the best-performing sample retained 93.9% of its initial performance. These results confirm sustained functionality for more than 2 months. To the best of our knowledge, this represents the longest reported functional lifetime among biodegradable pressure sensors.

Fig. 4. Functional lifetime of the biodegradable pressure sensor in vitro and in vivo.

(A) Comparative pressure-sensing response of differential and single-ended configurations during immersion in PBS at 37°C, monitored for 11 weeks. W, week. (B) Relative performance change over time. Differences among time points were analyzed using a one-way repeated-measures analysis of variance (ANOVA) followed by Tukey’s post hoc test for multiple comparisons (n = 3). (C) Schematic overview of sensor implantation for the in vivo functional lifetime evaluation. Mild pressure with fingertip (orange arrow) was applied at the implantation site for testing. (D) Relative functional degradation of implanted biodegradable sensors over time, compared with external commercial sensor. Differences among time points were analyzed using a one-way repeated-measures ANOVA followed by Tukey’s post hoc test for multiple comparisons (n = 3). (E) Performance of implanted sensors with piezoelectric PLLA films up to 8 weeks. Orange arrows indicate when pressures were applied.

We further validated the sensor lifetime in vivo. A biodegradable sensor was subcutaneously implanted in the dorsal region of rats, while a commercial pressure sensor was placed externally on the overlaying skin for comparison (Fig. 4C). Mild manual pressure was applied with a fingertip (~10 to 30 kPa), and signals from both sensors were recorded. We evaluated the functional degradation of the implanted biodegradable sensors, using the commercial sensor as reference. Under in vivo conditions, the biodegradable sensor’s performance differed by less than 9.6% from the reference, showing no statistically significant difference (Fig. 4D). At week 8, however, the performance declined sharply, exhibiting an average difference of 94.1%. As shown in Fig. 4E, the signals recorded from the implanted sensor closely matched those of the commercial sensor throughout the monitoring period. The biodegradable sensor maintained reliable performance for 7 weeks. By week 8, its signals showed only small peaks relative to the commercial sensor, marking the defined end of its functional lifetime (~2 months). In addition, we implanted a biodegradable sensor fabricated with nonpiezoelectric film to serve as a control (fig. S6). Unlike the commercial sensor, this device showed no measurable response to applied pressure. Additional noise characterization was performed to quantify baseline stability under in vivo conditions (fig. S7). The charge amplifier for the single-ended sensor was referenced to the animal body, which effectively reduced floating potentials. By this way, external electromagnetic interference and motion artifacts are introduced as common-mode signals, which are attenuated to maintain a stable baseline during movement. Together, these result further verified that the recorded signals from the functional PLLA sensors were not influenced by external artifacts such as motion or triboelectric effects. Histological analysis of hematoxylin and eosin (H&E)–stained skin tissue at the implantation site revealed only a mild immune response, without evidence of persistent or severe inflammatory changes (fig. S8). Such transient inflammatory responses are commonly observed following device implantation. Furthermore, no morphologic evidence of cell apoptosis was observed, further supporting the strong biocompatibility of the sensor after 2 months of implantation. Overall, these results demonstrate that the biodegradable pressure sensor maintains stable functionality throughout its intended operational lifetime and exhibits favorable in vivo biocompatibility.

In vivo rabbit studies of device performance in knee joint loading measurement

We conducted an in vivo knee joint monitoring study to evaluate the functionality of the biodegradable pressure sensor in the joint environment and to assess its effects on surrounding tissues. The overall monitoring system is shown in Fig. 5A. The sensor was implanted onto the medial femoral condyle of a rabbit knee. The implantation was performed carefully to ensure seamless integration, with no visible protrusion of the sensor on the skin surface following closure and suturing (Fig. 5B). To enhance conformal contact, the sensor was precontoured to match the local curvature of femoral condyle, thereby ensuring effective mechanical engagement of the active sensing region (fig. S9). A custom-built printed circuit board (PCB) for signal processing and Bluetooth communication was enclosed in a leak-proof package and subcutaneously implanted in the rabbit’s thigh to enable wireless data transmission (fig. S10; see Materials and Methods for details). After 1 month of implantation, the sensor maintained its structural integrity at the implantation site. Tissue collection confirmed that the sensor retained its native form, and x-ray imaging further verified its presence and stable positioning within the joint (Fig. 5C).

Fig. 5. Biodegradable pressure sensor for in vivo joint loading monitoring.

(A) Overview of the joint loading monitoring system. (B) Images of sensor implantation in rabbit knee during surgery. (C) Collected tissue and retrieved implant 1-month postsurgery, along with an x-ray image confirming the presence of the sensor at the implantation site. (D and E) Representative recordings from implanted sensor in rabbit: (D) during brief stance transitions and (E) during continuous weight-bearing stances—(1) resting, (2) body partially supported by forelimbs, (3) brief standing, and (4) hopping and standing. (F) Evaluation of knee joint loading pressure under stationary and moving conditions using data from implanted sensor. Statistical comparisons were performed using a paired t test (n = 5). (G) Histological evaluation of femoral and tibial cartilage using H&E and Safranin O/Fast Green staining in the implantation group and the control (no implantation) group at 1 month. The asterisk (*) marks the location of the implanted device. Scale bars, 200 μm. (H) Quantitative assessment was performed using the International Cartilage Repair Society (ICRS) histological scoring system. Statistical comparisons between two groups were performed using Mann-Whitney U test (n = 3). Scores represent the averaged results from four independent, blinded evaluators. Data are presented as mean ± SD.

For functional validation in vivo, we monitored knee joint activity by recording signals during the rabbit’s free movement after surgery. As shown in movie S1, the PLLA pressure sensor produced dynamic responses corresponding to the animal’s movements, such as hopping, sitting, and brief standing. During hopping, the sensor recorded peak pressures exceeding 400 kPa, corresponding to ~53.6% of the rabbit’s body weight, calculated by dividing the measured force by the active area of the sensor (1 N/cm² = 10 kPa) (Fig. 5D). In contrast, when the rabbit stopped moving and sat, the knee joint experienced relatively lower pressures ranging from 90 to 211 kPa, equivalent to ~11.8 to 27.5% of its body weight. These results demonstrate that the sensor can reliably detect and distinguish between different stance conditions. Figure 5E shows continuous weight-bearing conditions: (1) resting, (2) body partially supported by forelimbs with slightly increased hindlimb pressure, (3) brief standing with 43.1% of body weight, and (4) hopping and standing, showing peak loads of 58.8% of body weight (movie S2). The results validated the system’s ability to monitor continuous and dynamic knee joint loading in real time. Quantitative analysis shows pressures of 51 to 211 kPa (6.9 to 27.5% body weight) at rest and 211 to 532 kPa (27.5 to 69.2% body weight) during movement (Fig. 5F). These results align well with previously reported direct measurements in New Zealand white rabbits, where the knee joint bears ~6 to 25% of body weight during sitting and ~21 to 60% during hopping (53). The agreement between our measurements and literature values further validates the functionality and accuracy of the biodegradable pressure sensor for monitoring physiological knee joint loading. In addition, the implanted sensor produced clear pressure signals when the knee was manually flexed (movie S3), demonstrating its responsiveness beyond the rabbit’s natural movements. Notably, at ~5 to 6 s, increased flexion produced a distinct peak in the recorded signal, indicating sensitivity to varying loading amplitudes. This suggests potential utility in therapeutic rehabilitation settings, where controlled joint loading could be applied and quantitatively monitored to support cartilage regeneration.

In vivo biocompatibility assessment

We further investigated the biocompatibility of the pressure sensor within the knee joint environment by analyzing tissue samples collected 1-month postsurgery. H&E staining was performed to assess tissue morphology, inflammatory cell infiltration, and apoptosis, while Safranin O/Fast Green staining was used to evaluate hyaline cartilage integrity (Fig. 5G). Histological analysis of femoral and tibial sections from both the implanted group and the nonimplanted control group revealed no substantial inflammation, cartilage degeneration, structural damage, or abnormal tissue remodeling. Chondrocyte distribution and cartilage matrix staining remained comparable between groups, with no evidence of fibrosis, necrosis, or proteoglycan loss. Synovial inflammation was quantified using the Krenn synovitis scoring system (fig. S11) (54). The implanted group exhibited a mean score of 3.5, corresponding to low-grade, nonspecific synovitis. This mild inflammatory response is consistent with a typical foreign body reaction to implanted biomaterials (55). Occasional multinucleated giant cells were identified. However, no evidence of apoptosis or tissue necrosis was observed. As shown in Fig. 5H, quantitative assessment using the International Cartilage Repair Society (ICRS) system (56) showed no significant differences between the implantation and control groups in either femoral or tibial samples, indicating that the sensor implantation did not adversely affect cartilage quality or overall joint health. The results suggest that the biodegradable sensor maintains good local biocompatibility in the knee joint environment, supporting its suitability for monitoring applications without compromising cartilage health.

This study demonstrates that a biodegradable, implantable PLLA pressure sensor can be engineered to monitor mechanical loading in the knee joint as well as other physiological pressures. The single-ended configuration extends the functional lifetime beyond 2 months, addressing a major limitation of conventional differential configuration devices. Moreover, the sensor’s inherent biodegradability eliminates the need for secondary removal surgery once its clinical function has ended. For knee joint monitoring applications, the device enables direct and accurate measurement of dynamic joint loading, providing critical data to guide rehabilitation protocols, evaluate cartilage repair outcomes, and manage conditions such as OA or postsurgical recovery. Overall, these advances establish a new paradigm for biodegradable bioelectronic implants: devices designed to operate over clinically relevant timeframes and then safely degrade within the body. Despite these advancements, several challenges remain. While the sensor is fully biodegradable and eliminates the need for high-risk secondary joint surgery, the current system relies on a nonbiodegradable subcutaneous PCB module for wireless data transmission. This configuration necessitates a minor surgical procedure for the module retrieval, representing a limitation in the overall biodegradability of the platform. Future work should investigate encapsulation materials and polymers that better resist fluid penetration while maintaining biocompatibility and controlled degradability. Validation in larger animal models with expanded sample sizes will be necessary to further support translational development toward clinical applications. In parallel, the development of biocompatible semiconductor materials and fabrication strategies will be important to ensure safety for long-term implantation. For certain long-term monitoring applications, functional lifetime needs to be extended beyond the current 2-month window under physiologically relevant dynamic mechanical conditions. Despite these remaining hurdles, the proposed biodegradable, biocompatible, and implantable pressure sensor with wireless communication represents a paradigm shift in in vivo joint loading measurement for OA management, prevention, and treatment. More broadly, it offers substantial potential for wireless monitoring of diverse internal and intra-organ pressures, expanding the impact of biodegradable electronics in health care.

MATERIALS AND METHODS

Fabrication of piezoelectric PLLA nanofibers and biodegradable pressure sensor

Piezoelectric and nonpiezoelectric PLLA nanofibers were fabricated using the electrospinning method (35). PLLA (0.8 g; Corbion Purac Amsterdam, the Netherlands) was dissolved in a 1:4 (v/v) mixture of N,N-dimethylformamide (Sigma-Aldrich, anhydrous, ≥99.9%) and dichloromethane (Sigma-Aldrich, anhydrous, ≥99.8%). The solution was delivered at flow rate of 2 ml/hour through a flat-tipped 22-gauge needle (Jensen Global, Santa Barbara, CA) with 14 kV applied and was then electrospun onto a grounded aluminum drum wrapped in aluminum foil. To obtain aligned (piezoelectric) fibers, the drum was rotated at 4000 rpm; for randomly oriented (nonpiezoelectric) fibers, the rotation speed was set to 300 rpm. The electrospinning was conducted under ambient conditions at a relative humidity of 30 ± 10%. The resulting PLLA mats were annealed at 105°C overnight and allowed to cool slowly to room temperature before repeating the process at 160.1°C.

The electrospun nanofiber mats were cut into 5 mm–by–10 mm films. Encapsulating PLA films were prepared via compression molding using a Carver Press (3950-1011) at 200°C, followed by quenching with dry ice. One Mo electrode, attached to a 1-mm-wide wire, was cut from a 25-μm-thick Mo sheet (ESPI Metals). The electrode was hot-embossed into PLA using the Carver Press at 110° to 130°C. The resulting films were then sandwiched between the encapsulating PLA layers. All edges of the encapsulation were sealed using a heat press. For implantation in the rabbit knee joint, the sensor was contoured during encapsulation and sealing process to conform to the anatomical curvature of the femoral condyle.

SEM and XRD

For SEM, samples were mounted on a standard SEM stub (Ted Pella, Redding, CA) using carbon conductive tape (Ted Pella, Redding, CA). The samples were sputter-coated with Au/Pd using a sputter coater (CCU-010, Safematic) and then imaged with a Verios 460 L SEM at 15 kV and 6500 magnifications. For XRD, sample measurements were carried out at room temperature using a Bruker D2 Phaser, from 5° to 40° (2θ) with a scan speed of 5°/min.

Piezoelectric characterization under impact system

The electrospun PLLA film was cut into a 10 mm–by–10 mm square film and was compactly sandwiched between layers of polyimide tape (Dupont, Wilmington, DE) with an aluminum (Al) electrode. The assembled sample was securely mounted on the impact system. Applied force [newton (N)] was measured using a quartz dynamic force sensor (208C02, PCB Piezotronics, NY) attached to an electrodynamic transducer (ET-132-2, Labworks Inc., CA), which was driven by a 1-V peak-to-peak signal from a function generator (4054B, BK precision, CA). An electrometer (Keithley 6514, Tektronix Inc., OR) was connected to the sample to record the output charge [coulomb (C)] during impact, and the resulting signal was monitored using an oscilloscope (4828, Pico Technology, TX).

Sensor calibration under impact system

A sensor to be calibrated was securely mounted onto the impact testing platform. Applied force was measured using a factory-calibrated quartz dynamic force sensor (208C01 and 208C02, PCB Piezotronics, NY) attached to the electrodynamic transducer. The output of the sensor under test was connected to a charge amplifier circuit to measure generated charge. The resulting signal was monitored using an oscilloscope (4828, Pico Technology, TX). By systematically varying the applied force amplitude, a calibration curve was obtained to determine sensor sensitivity. The commercial pressure reference sensor (ESS201, Tekscan, MA) was calibrated according to the manufacturer’s protocol using controlled applied loads.

Unit conversion

In this work, “pressure (kPa or mmHg)” was calculated by normalizing the measured force to the sensor’s effective surface area (1 N/cm² = 10 kPa ≈ 75 mmHg). To maintain technical clarity, we use “force” exclusively to quantify the external mechanical input (N) as measured by the commercial sensor. The term “loading (e.g., knee joint loading)” is used to describe the broader mechanical environment and physiological stress within the joint. The percentage of body weight (%BW) was calculated by normalizing the derived force to the animal’s body weight. Force was obtained by multiplying the measured pressure by the defined active sensing area. The geometric sensing area was considered the effective load-bearing area. The sensor output was normalized to body weight according to

$$\%\text{BW}=\frac{F}{W}\times 100$$

where F is the calculated force over the sensing area, and W is the animal’s body weight expressed in newtons.

Degradation study and functional lifetime test

Biodegradable sensor samples were individually immersed in 50 ml of PBS (pH 7.4; Thermo Fisher Scientific) in separate beakers and placed on a hot plate at 60°C. The PBS was refreshed weekly. Every 3 to 4 weeks, the weight of each sample was recorded. Before weighing, each sample was removed from the beaker, rinsed with deionized (DI) water, and dried. For the Mo resistance degradation test, 10 mm–by–10 mm square Mo pieces with or without PLA encapsulation (n = 3) were immersed in PBS at 37°C. On a weekly basis, the samples were cleaned with DI water, dried, and their resistance change was recorded. To evaluate its functional lifetime, fabricated biodegradable pressure sensor samples were stored in a beaker containing PBS, and the PBS was refreshed weekly. The sensor samples were temporarily transferred to a well of a six-well plate filled with fresh PBS. An electrodynamic transducer was used to apply a precise mechanical load to the sample. The applied force was controlled to be a 5 N compressive force, which was verified by a quartz dynamic force sensor (208C02, PCB Piezotronics, NY). An electrometer was connected to the sample’s electrode to record its electrical output. The real-time response of the sensor sample to the applied force was captured using an oscilloscope. After the measurement, the samples were returned to the beaker for continued incubation.

Ex vivo experiments

Biodegradable pressure sensor and a commercial pressure sensor (ESS102, Tekscan, MA) were prepared for ex vivo testing with comparable sensing areas. After implantation, each sensor was independently connected to amplifiers, and output signals were recorded using a digital oscilloscope (PicoScope 400, Pico Technology, TX). The biodegradable sensor was evaluated alongside the precalibrated commercial reference sensor. The commercial sensor was calibrated according to the manufacturer’s specifications before the experiments. For ICP experiment, adult Sprague-Dawley rats were euthanized by CO₂ asphyxiation followed by cervical dislocation. The skull was excised and cleaned of soft tissues. Two circular defects (2.7 mm in diameter) were created on the central parietal bone using the Ideal Micro-Drill (Braintree Scientific, MA). An additional hole (0.8 mm) was drilled in the frontal bone for PBS injection using a butterfly needle to simulate increased ICP. The biodegradable and commercial pressure sensors were positioned over the parietal defects and secured with silicone adhesive (KWIK-SIL, World Precision Instruments, FL). Experiments were performed in a cadaveric model to eliminate the influence of active physiological pulsatility. A zero-pressure baseline was established before injection. For knee joint pressure experiment, a New Zealand White rabbit carcass (~3 kg; Animal Technology Inc., TX) was thawed at room temperature. The patella was palpated, and a medial skin incision was made. A medial parapatellar arthrotomy was performed to release soft tissue and access the joint space, and the patella was displaced laterally. A previously assembled device—comprising both the biodegradable and commercial pressure sensors sealed together with Parafilm—was implanted into the joint cavity. The skin was then closed using intradermal sutures. The knee joint was manually and gradually flexed and extended to simulate dynamic changes in intra-articular pressure.

Cell culture

Rabbit BMSCs (RBXMX-01001, Cyagen) were cultured following the vendor’s guidance. Briefly, cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum for proliferation. For chondrogenic differentiation, cells were cultured in DMEM containing 0.2 mM ascorbic acid, insulin-transferrin-selenite (50 μg/ml) premix, freshly added 0.1 μM dexamethasone, and transforming growth factor–β3 (10 ng/ml). Cells at passages 2 to 5 were used for subsequent experiments.

Cell viability test

Cells were seeded onto 0.1% gelatin-coated culture plates and biodegradable pressure sensor samples at a density of 1 × 10⁵ cells/ml. After incubation at 37°C and 5% CO₂ until reaching 80 to 90% confluency, cell viability was assessed using the LIVE/DEAD cell imaging kit (R37601, Invitrogen) following a 30-min staining protocol. Fluorescence images were acquired using a fluorescence microscope (RVL2-K4, ECHO, CA) equipped with fluorescein isothiocyanate (excitation: 470/40 nm, emission: 525/50 nm) and TxRED (excitation: 560/40 nm, emission: 630/75 nm) filters. Live and dead cells were quantified using the software, ImageJ (National Institutes of Health, Bethesda, MD).

Quantitative polymerase chain reaction

After 14 days of culture in chondrogenic medium, cells were collected and pelleted by centrifugation. Total RNA was extracted using the PureLink RNA Mini Kit (Invitrogen), following the vendor’s guidance. cDNA was synthesized from the isolated RNA using the iScript cDNA synthesis Kit (1708897, Bio-Rad). Quantitative real-time polymerase chain reaction (PCR) was performed using the synthesized cDNA, Universal SYBR Green Supermix (Bio-Rad), and a real-time PCR system (CFX Connect Optics Module, Bio-Rad, CA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the housekeeping gene, while SOX9, aggrecan (ACAN), and type II collagen (COL2A1) were analyzed as chondrogenic markers. In the experiments, SOX9 expression was detectable as early as day 3, while ACAN expression appeared after day 7 and COL2A1 after day 14. The primer sequences used are as follows: GAPDH, 5′ GTCGGAGTGAACGGATTTG 3′ (forward) and 5′ GTAGAGCATGTAGTGGAGGT 3′ (reverse); SOX9, 5′ CTCCGACACCGAGAATAC 3′ (forward) and 5′ CCTCTTGGCTCTCCTTCTT 3′ (reverse); ACAN, 5′ CAGCCGGACAAGTTCTTT 3′ (forward) and 5′ GTGAAGGGTAGGTGGTAATTG 3′ (reverse); COL2A1, 5′ GGCTTCCACTTCAGCTATG 3′ (forward) and 5′ CAGTGGTAGGTGATGTTCTG 3′ (reverse). Relative gene expression was calculated using the 2^–ΔΔCT method.

GAG assay

Cells were collected after 14 days of culture and digested in 250 μl of tris-HCl buffer containing proteinase K (1 mg/ml) at 65°C for 16 hours. Following digestion, 100 μl of the lysate was added to each well of an enzyme-linked immunosorbent assay plate and mixed with 100 μl of 1,9-dimethymethylene blue dye solution, following the vendor’s instructions for the GAG assay kit (#6022, Chondrex). Absorbance was measured at 525 nm using a microplate reader.

Hemolysis test

Healthy rabbit whole blood (anticoagulated with sodium citrate; Innovative Research Inc., MI) was diluted with normal saline at a 4:5 volume ratio. Square films (10 mm by 10 mm) of PLLA nanofiber and Mo were incubated in normal saline at 37°C for 1 hour. Subsequently, each film was incubated in 10 ml of normal saline containing 200 μl of the diluted blood at 37°C for 1 hour. Saline and DI water without films served as negative and positive controls, respectively. After incubation, the samples were centrifuged at 1000g for 5 min, and the supernatants were transferred to a 96-well plate. Absorbance was measured at 545 nm to determine hemolysis. The hemolysis percentage was calculated as follows

$$\text{Hemolysis rate} (\%)=\frac{({\text{Absorbance}}_{\text{test sample}}-{\text{Absorbance}}_{\text{negative}})}{({\text{Absorbance}}_{\text{positive}}-{\text{Absorbance}}_{\text{negative}})}\times 100\%$$

Surgical procedure for in vivo experiments in rat models

All animal use procedures were approved by the University of Connecticut Institutional Animal Care (protocol A23-028) and Use Committee and conducted in full accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and the guidelines of the National Society for Medical Research. Male Sprague-Dawley rats (Charles River) received subcutaneous injections of buprenorphine (0.65 mg/kg) and meloxicam (1 mg/kg, 5 mg/ml). Anesthesia was induced and maintained with isoflurane gas. A dorsal incision was made, and the biodegradable pressure sensor was implanted and secured using preformed suture holes on the device. For performance testing of the implanted sensor, the rat was anesthetized with isoflurane gas, and the sensor was connected via leads to an amplifier and oscilloscope. A commercial pressure sensor was placed over the implantation site, and gentle pressure was applied to evaluate the device response, which was then compared to the commercial sensor output as a reference.

Surgical procedure for in vivo experiments in rabbit models

All animal use procedures were approved by the University of Connecticut Institutional Animal Care (protocol A25-054) and Use Committee and conducted in full accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and the guidelines of the National Society for Medical Research. Male New Zealand White rabbits (Charles River) were used for the in vivo experiments. Biodegradable pressure sensor implantation surgery was performed on the knee joint of each animal. Specifically, each rabbit received buprenorphine (0.02 to 0.05 mg/kg) as a preoperative analgesic. Anesthesia was then induced using a mixture of ketamine, xylazine, and glycopyrrolate and maintained with 2.5 to 3.5% isoflurane. The knee joint was accessed via a medial parapatellar arthrotomy, followed by lateral displacement of the patella. The infrapatellar fat pad was excised to expose the joint space, and the knee was gently flexed to facilitate sensor placement. Before implantation, the sensor was precontoured such that the active sensing area conformally covered the femoral surface. A transcondylar bone tunnel was created from the superior aspect of the medial femoral condyle to the lateral condyle, following an approach similar to that of anterior cruciate ligament reconstruction, to allow passage of the lead wire from the sensor to a custom-built Bluetooth module. The Bluetooth module was securely encapsulated and implanted subcutaneously in the lateral thigh region.

In vivo experimental setup

Bluetooth transmission was programmed to operate for 20 min/day using a timer function integrated into the electronic module. During each transmission period, rabbits were allowed to move freely, and simultaneous video recordings were obtained. Representative data were selected from clearly defined and behaviorally relevant movement phases.

Micro–computed tomography

After 1 month of implantation, rabbit’s knees were collected. The knee was then placed on the station of micro–computed tomography equipment (IVIS SpectrumCT, PerkinElmer). The scanned knee bones were exported using RadiAnt DICOM software (Medixant, Poland)

Histology

Harvested tissues were fixed in 10% neutral-buffered formalin, dehydrated, and embedded in paraffin. Rabbit femur and tibia samples were decalcified before processing. Tissue sections were stained with H&E for morphological assessment and evaluation of inflammatory infiltrate and with Safranin-O/Fast Green for visualization of articular cartilage. For the rat skin samples, a histomorphology assessment was performed by a pathologist blinded to the experimental groups (n = 3 animals, with two samples analyzed per animal).

Scoring system for cartilage evaluation

The morphological assessment of cartilage was independently performed by four professionals with expertise in cartilage histology. The evaluation followed the ICRS visual histological assessment scale (56).

Statistical analysis

Data with a sample size of n = 5 (unless otherwise indicated) are presented as means ± SD with individual data points shown. Normality was assessed using the Shapiro-Wilk test. For comparisons among multiple groups, one-way analysis of variance (ANOVA) was performed followed by Tukey’s post hoc test for multiple comparisons. Comparisons between two groups were conducted using a t test, as appropriate. Statistical analysis was performed using GraphPad Prism. Experiments shown in representative photographs were independently repeated three times (unless otherwise noted) with consistent results.

Supplementary Materials

The PDF file includes:

Figs. S1 to S11

Legends for movies S1 to S3

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S3