Measurements of redox balance along the gut using a miniaturized ingestible sensor

Nature Electronics | Volume 8 | September 2025 | 856–870 856 nature electronics Article https://doi.org/10.1038/s41928-025-01411-4 Measurements of redox balance along the gut using a miniaturized ingestible sensor Aniek Even 1,2,8 , Roseanne Minderhoud1,3,4,5,8, Tom Torfs 6, Francesca Leonardi 1,2, Arjan van Heusden 1,2, Ria Sijabat 1,2, Dimitrios Firfilionis 1,2, Ivan Dario Castro Miller 6, Ramzy Rammouz 6, Tobias Teichmann 1,2, Ruben van Bergen 1,2, Günter Vermeeren7, Edoardo Capuano 1,4, Rachel Armstrong 1,2, Klaus Mathwig 1,2, Sonja de Vries 1,5, Annelies Goris 1,2, Nick Van Helleputte 6, Guido Hooiveld 1,3 & Chris Van Hoof 1,2,6 Redox balance—the equilibrium between oxidants and reductants—is a key modulator of a healthy gut and consequently overall well-being. Excess reactive species, resulting in oxidative stress, are linked to deleterious processes including inflammation and microbiome dysbiosis. However, a lack of suitable in vivo methods has restricted measurements of redox balance in the human gut. Here we report a miniaturized ingestible sensor that is equipped with an oxidation–reduction potential sensor, an electrochemical reference electrode and pH and temperature sensors. We preclinically validate our wireless gastrointestinal (GI) smart module (GISMO) in GI fluids and an animal model and report in-human measurements in 15 healthy individuals. Our high-temporal-resolution data, measured every 20 s, reveal consistent profiles from an oxidative environment in the stomach to a strongly reducing environment in the large intestine. This non-intrusive method has the potential to advance (GI) disease monitoring and offer insights into the gut microbiome. The gastrointestinal (GI) barrier plays a crucial role in maintaining gut health and, consequently, overall homeostasis. It balances the absorption of essential nutrients and water while preventing entry of harmful foreign substances and pathogens. This balance is facilitated by cooperation between the host, immune system and gut microbiota 1 . A key modulator and important signalling mechanism is redox balance2. Although redox state has not generally received levels of recognition similar to pH, the transfer of electrons in redox reactions is equally vital to living organisms as the transfer of protons in acid–base reactions3. Disturbances in redox biology can lead to oxidative stress—an imbalance between oxidants and antioxidants 4 —which has been associated with several diseases and disorders including inflammatory bowel disease (IBD) 5 , microbiome dysbiosis 6 , GI cancers 7 and even early death in animal models8. Direct measurements of redox balance are feasible with an oxidation–reduction potential (ORP) sensor. When exposed to a solution, the voltage difference between a working electrode and reference electrode (RE) is a direct measurement of redox potential. A positive ORP value indicates an oxidizing environment; a negative potential indicates a reducing environment. Although ORP sensors are commonly used in some disciplines (such as water and soil measurements3), in vivo measurements have only been performed in rodents 9,10 . Attempts to Received: 18 October 2024 Accepted: 8 June 2025 Published online: 16 July 2025 Check for updates 1OnePlanet Research Center, Wageningen, the Netherlands. 2imec, Wageningen, the Netherlands. 3Division of Human Nutrition and Health, Wageningen University & Research, Wageningen, the Netherlands. 4Food Quality and Design Group, Wageningen University & Research, Wageningen, the Netherlands. 5Animal Nutrition Group, Wageningen University & Research, Wageningen, the Netherlands. 6imec, Leuven, Belgium. 7Department of Information Technology, Ghent University, Ghent, Belgium. 8These authors contributed equally: Aniek Even, Roseanne Minderhoud. e-mail: aniek.ev[email protected] Nature Electronics | Volume 8 | September 2025 | 856–870 857 Article https://doi.org/10.1038/s41928-025-01411-4 been developed but lack sensors beyond pH 19,20 , blood detection 21 and fermentation gases22,23, and those with additional sensors have not been applied in a clinical setting24–27. In this Article, we report a miniaturized (length 21 mm, diameter 7.5 mm), low-power, wireless GI smart module (GISMO). The capsule is equipped with an ORP sensor, a custom electrochemical RE and sensors for pH and temperature. Designed to be ingested orally and to wirelessly communicate with a wearable receiver, it can record along the entire GI tract without requiring special bowel preparation. We report in vitro validation and in vivo validation in an animal model as well as results from 15 healthy volunteers. Miniature ingestible sensor capsule design To measure redox balance and pH, the GISMO ingestible system uses a series of electrochemical sensors, as illustrated in Fig. 1. These sensors are integrated into a wireless ingestible capsule together with custom electronics for time-multiplexed sensor readout and wireless measure ORP in humans have been limited to faecal analysis, which were deemed unsuitable for discriminating between patients and healthy controls11, most likely due to the highly reactive nature of oxidants and the effect of oxygen in the environment. Other clinical studies have resorted to indirect indicators of oxidative stress, such as oxidative damage to DNA, lipids and proteins, or antioxidant status. These approaches measure the damage caused by oxidative stress but do not directly capture the dynamic nature of the process12–14. Performing direct measurements in the human GI tract is challenging due to its complexity and inaccessible nature. Common clinical tools (such as upper endoscopy and colonoscopy), which offer visual inspection but are invasive, cannot assess the entire GI tract and lack sensitivity to oxidative stress15. Wireless capsule endoscopies16,17 do facilitate inspection of the small intestine. However, they are equally unsuited for direct redox measurements due to the lack of embedded sensors and the need for bowel preparation, which alters the gut environment 18 . Capsules with embedded biochemical sensing have 1 3 5 7 pH –400 –200 0 200 ORP (mV) 0 10 20 30 40 Time (h) 26 30 34 38 Temperature (°C) Stomach Small intestine Large intestine c d e a b 230 mV Stomach Large intestine Small intestine –26 mV –323 mV Oxidizing OxidizingReducing O x i d a t i v e s t r e s s Fig. 1 | GISMO ingestible system overview and example data of in-human redox balance measurements along the gut. a, The miniaturized ingestible sensor capsule (length 21 mm; diameter 7.5 mm) measures redox balance (ORP), pH and temperature. b, Data are sent wirelessly to a compact wearable receiver worn outside the body. c, Schematic illustration of the full system in operation. d, Example data from our human trial show distinct GI regions. The redox environment (top) is strongly oxidizing in the stomach, slightly reducing in the small intestine and highly reducing in the large intestine. pH values (middle) were measured simultaneously. Temperature (bottom) was used to confirm body exit. e, We acquired gut redox reference values in 15 healthy participants from a total of 66 ingestibles. The redox values in these healthy volunteers rarely exceeded 230 mV in the stomach, −26 mV in the small intestine or −323 mV in the large intestine (95th-percentile values across all measurements). In a disturbed gut environment, for example due to inflammation or microbiome dysbiosis, an imbalance between oxidants and antioxidants can occur, leading to an excess of reactive species and resulting in oxidative stress. This shift to a more oxidative environment is indicated by the arrows, moving from the healthy region in green (baseline values differ per segment) towards the oxidative stress region in red. The size of the illustrated segments corresponds to the average regional transit times: 4.1 h for the stomach, 6.2 h for the small intestine and 57.3 h for the large intestine. Image in c from Pixabay. Nature Electronics | Volume 8 | September 2025 | 856–870 858 Article https://doi.org/10.1038/s41928-025-01411-4 communication, a conformal antenna and two silver oxide batteries, all enclosed in a biocompatible polyetheretherketone (PEEK) housing (Fig. 2a). To reduce the risk of retention and facilitate easy ingestion, the capsule is size ‘0’ (21 mm × 7.5 mm), which is three times smaller in volume than capsule endoscopes 28 , three times smaller than a commonly used pH capsule 19 , 1.7 times smaller than a—not yet clinically validated—inflammation-marker capsule 24 and 2.8 times smaller than a gas-based ingestible sensor29. To achieve such aggressive miniaturization, we optimized the volume occupied by all subcomponents. Although ORP measurements are commonplace in other domains, successfully deploying them in a miniaturized ingestible capsule necessitates development of compact working electrodes and REs. These electrodes need to deliver stable performance throughout the measurement and endure the harsh environment of the GI tract. To this end, we designed a custom chip with a platinum electrode for the ORP sensor. Accurate pH measurement is ensured by two redundant pH sensor chips based on ion-sensitive field-effect transistor technology (ISFET; Sentron). The sensor surfaces are either Pt or chemically inert glass-like oxides and can withstand the acidic environment of the stomach. We custom-designed our RE such that drift remains below 0.06 mV h −1 . The reference occupies the entire dome on one side of the capsule (59 mm3). The dome contains a Ag/AgCl wire immersed in a KCl gel saturated with AgCl. To protect the sensitive electronics inside the capsule from the hostile GI environment, everything is hermetically sealed with epoxy resin such that only the sensor surfaces and RE are exposed. The porous RE frit (thickness 0.4 mm and diameter 1 mm) is designed to ensure an optimal tradeoff between electrical impedance and leakage to the outer environment (Extended Data Fig. 1). The biochemical sensors are integrated on a flexible printed circuit board (PCB), which is stacked on top of two miniature rigid PCBs containing the electronics (Fig. 2b). The first of these rigid PCBs includes the electrochemical sensor readout, a low-power microcontroller, a high-precision temperature sensor and a three-dimensional (3D) magnetometer. The second PCB includes a power conditioning circuit, a magnetic switch for power-on, a 3D accelerometer and a radio transceiver. The use of low-power electronics, hardware reuse for multisensor readout, aggressive duty-cycling and microcontroller software optimization were key to achieving the desired miniaturization. To further save volume inside the capsule, we designed a conformal wrap-around antenna. In a tradeoff between antenna size and efficiency on the one hand and path loss through body tissue on the other hand, 868 MHz (EU license-free band for short range devices) was selected as the communication frequency. An overview of the optimal frequencies for different capsule depths and its comparison to relevant license-free available bands in Europe and Americas is shown in Extended Data Fig. 2a. Because commercial antenna modules are designed for operation in air, they perform poorly in-body due to the characteristic impedance mismatch. Hence, we designed our own custom antenna. It is a custom stepped impedance antenna design 30 adapted to the frequency and the shell material (PEEK) of the pill for efficient operation inside the GI tract. The antenna is designed in a 172-µm-thick flexible circuit board (Fig. 2c) wrapped around the a Biocompatible polymer housing Batteries Miniaturized electronic circuit boards Antenna Reference electrode Biochemical sensors b c d Connector Connector ISFET readout Digital temperature sensor Microcontroller 3D magnetometer Magnetic switch LED Ag/AgCl reference electrode Wireless ingestible capsule ADC Power conditioning µC software pH ISFETs ORP sensor Biochemical sensors (flex PCB) Potentiometric readout Electronics (rigid PCB 1) Conformal antenna Silver oxide primary batteries (2×) Radio transceiver Digital 3D accelerometer Electronics (rigid PCB 2) Fig. 2 | GISMO device and wearable receiver architecture and characterization. a, Exploded view of the ingestible including electrochemical sensors, electronics, conformal antenna, batteries and RE in a PEEK biocompatible housing. b, Block diagram of the main parts of the ingestible subdivided in two thin rigid PCBs, a flexible (flex) PCB with the electrochemical sensors, the RE, batteries and conformal antenna. c, Stacked rigid PCBs and flex PCB together with the battery subassembly and the preshaped conformal antenna, with a 2-euro coin for size comparison. d, Exploded view of the wearable battery-operated receiver providing a simple user interface. Nature Electronics | Volume 8 | September 2025 | 856–870 859 Article https://doi.org/10.1038/s41928-025-01411-4 electronics just inside the encapsulating shell, allowing for optimal use of the volume in the ingestible. This optimization achieved an antenna gain between −33.2 and −20.8 dBi when immersed in a large body phantom of 40 cm diameter. The sensor data are sent in real time wirelessly and encrypted to an unobtrusive wearable receiver (Figs. 1b and 2d). A pair of compact non-rechargeable silver oxide batteries (SR516SW, nominal capacity 12.5 mAh at a voltage of 1.55 V each) are used to power the capsule. This battery chemistry was selected for its safety and proven use in other ingestible capsules 31 . The batteries were assembled using a combination of wire bonding and conductive epoxy. To ensure sufficiently long shelf life, the entire ingestible is activated by a magnetic reed switch such that the power-off consumption is only 0.13 µA, which discharges the batteries only ~7–8% over a period of six months. In line with typical transit times 32,33 , we designed the ingestible to function for at least five days while measuring and transmitting all sensors (temperature, dual pH and ORP) every 20 s, which provides very detailed measurement profiles. The ingestible has an average current consumption of only 28 µA during normal operation (Extended Data Fig. 3). After five days, the ingestible switches to an ultra-low-power mode (9.4 µA average current) where only the temperature and limited housekeeping data are transmitted. The ingestible can operate for at least nine more days (two weeks total) in this mode. This provides an additional safety mechanism to confirm device exit in case of abnormally long transit times. The full functionality, power consumption and battery lifetime were tested in a closed container with physiological saline solution in an oven at 37 °C (Extended Data Fig. 3c). Extensive electronic testing was performed to validate the electronic readout and communication robustness in representative phantoms (Extended Data Fig. 2b,c). In vitro sensor validation To validate the ORP and pH sensors in vitro, we conducted experiments in progressively complex environments, starting with standard solutions of precisely known concentrations and advancing to porcine GI fluids. For the validation in standard solutions, we used commercially available ORP standards of 220 and 600 mV. Due to limited availability of standards in other mV ranges—negative commercial ORP standards do not exist—additional standard solutions were prepared in-house to generate a range of solutions covering the values anticipated in the GI S1 S3 S2 S3 S1 S4 S5 S6 S7 S7 S8 S9 S5 S4 S1 S3 –600 –400 –200 0 200 400 600 800 ORP (mV) a Horiba 1 Horiba 2 Ingestible 1 Ingestible 2 Ingestible 3 0 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 Time (h) 0 2 4 6 8 10 12 14 pH c b d Stomach Small intestine Cecum Proximal colon Middle colon Rectum –600 –400 –200 0 200 400 ORP (mV) Horiba 1 Ingestible 4 Ingestible 5 Ingestible 6 0 2 4 6 8 10 12 14 Time (h) 2 3 4 5 6 7 8 pH Fig. 3 | In vitro validation of the GISMO sensors. a,c, In vitro validation in standard solutions. Staircase measurement of ORP (a) and pH (c) profiles using three ingestibles show good agreement with two commercial sensors (Horiba) in ORP (alternate) standards in the range of −550 to 280 mV with pH range between 0 and 14. The measurements were performed at body temperature (37 °C) within 2 h due to limited stability of ORP standards prepared in-house by dissolving quinhydrone in pH buffers. For a full description of the solution, see Methods. Note that S2 is out of the designed range of the electronics because ORP values above 280 mV are not expected in the human body. It was challenging to create a highly reducing environment outside the body; for S8 and S9, we observed more variability, as well as between the commercial sensors, possibly due to the extremely high pH and the short lifetime of the solutions. b,d, To create a more complex and realistic environment, we performed continuous measurement of ORP (b) and pH (d) in different types of GI fluids collected from a pig post mortem, using three ingestibles versus a commercial sensor (Horiba). Similar trends between the ingestibles and commercial system were observed. The measurements were performed at body temperature (37 °C) and at segmentspecific relevant oxygen levels. The ingestibles were briefly taken out of the anaerobic chamber before entering small intestine conditions, which could have caused the time lag with the commercial system because ORP sensors are sensitive to oxygen concentrations. Nature Electronics | Volume 8 | September 2025 | 856–870 860 Article https://doi.org/10.1038/s41928-025-01411-4 tract of −550 to 280 mV (refs. 9,34,35). Figure 3a,c shows recordings of three ingestibles and two commercial systems (Horiba). The measured ORP values show strong consistency between the ingestibles and the commercial ORP sensors. Some variability is observed in the negative ORP solutions, which is also seen between the two commercial systems. This variability is likely attributable to the short lifetime of the solutions and the extreme pH levels (pH 12 and 14). These extremely basic pH conditions are not expected in the GI tract, but they are needed to create such a strongly reducing environment outside the body. The ingestibles system is designed to measure values up to 280 mV because higher values are not anticipated in the GI tract; hence the clipping for the 600-mV solution. The pH sensors can measure in the full pH range (0–14) and are in good agreement with the commercial systems. To validate the sensors in a more realistic and complex environment, we used GI fluids collected from post mortem pigs placed in a temperature-controlled anaerobic chamber, mimicking gut conditions (Fig. 3b,d). We observed a slight offset between the commercial sensor (Horiba) and our ingestibles and some differences among the three ingestibles in the middle colon and rectum. Nevertheless, the overall trend was similar. Even though these samples are actual GI fluids and the best available method to mimic the gut outside the body, the properties of the GI fluids seem to change over time after thawing. We observed clear phase separation in the sample from the small intestine, which resulted in a lower-than-expected pH. The inhomogeneities in the samples and changing properties could have resulted in some slight variation in measurements. Despite the challenges, we showed that our ingestibles can measure ORP and pH in the relevant range and are in good agreement with a commercial system. In vivo validation in an animal model In vivo functionality was validated in live pigs, a commonly used model for the human digestive system 36–38 . One example data trace of ORP, along with supporting pH and temperature data, is given in Fig. 4a. Three different regions of the GI tract can be distinguished: (1) stomach region with ORP values between −50 and 250 mV, (2) small intestine region with ORP values between −250 and −50 mV and (3) large-intestine region with ORP values lower than −300 mV. Our observations align with findings from the limited number of rodent studies, where direct measurements in the caecum of mice and rats have shown similar reducing environments ranging from −320 to −554 mV (refs. 9,10). Of the nine ingestibles administered, five examples are shown in Fig. 4. Of the other ingestibles, one had only pH sensors, and three stopped prematurely in the stomach (after 22.5 and 25 h, respectively) or in the beginning of the large intestine (after 48.5 h). In four ingestibles, we could not define the entry of the large intestine based on the pH profiles as conventionally used. However, we did observe a notable change in redox values. The example in Fig. 4a shows coinciding starts of the transitions in pH (from 7.9 to 6.0) and ORP (from −100 to −350 mV) going from the small intestine to the large intestine. In contrast, in four ingestibles without a significant pH change at the transition from the small intestine to the large intestine, the ORP dropped from an average 100 mV to −325 mV (Fig. 4b–e). For one of these four sensor capsules, an ultrasound examination was performed 67 h after ingestion (Fig. 4e). The ultrasound scan confirmed that the ingestible was located in the large intestine, as signalled by the low ORP value (−303 mV), which would have been missed based on pH alone. In-human baseline measurements of redox balance For the clinical evaluation, we recruited 15 healthy volunteers and planned for each to receive five ingestibles. Because this was the first time these were used in humans, the initial phase of the study focused on validating safety and reliable operation. Participants received one GISMO capsule in the morning, after an overnight fast. All participants comfortably ingested the capsule, which passed through their bodies without issues or discomfort, and no abnormal effects were observed. All capsules successfully recorded data and transmitted data reliably. After all participants agreed to continue, phase 2 of the clinical evaluation commenced two months later. In this phase, the volunteers received one morning and one evening pill on two separate days spread over two weeks. The morning ingestibles were given after an overnight fast; the evening ingestibles were given just before the evening meal. In total, 69 ingestibles were administered over both phases combined, of which 66 (Methods) were used in the subsequent analyses. The ingestible sensors were calibrated pre-ingestion, for interpretation of the GI data, and post-excretion, for sensor performance and drift characterization. Throughout the trial, we observed robust wireless communication. During nighttime use, participants placed the wearable receiver and an additional non-wearable receiver next to their bed. Participants were allowed to remove the wearable receiver for short durations (for example, during showering or light-intensity sports). Despite this, we still achieved overall data coverage (that is, percentage of correctly received data packets) of 90.3% (range 59.1% to 98.9% across all ingestibles). Excluding the period when one of the wearable devices malfunctioned due to being dropped by the participant, the overall data coverage increased to 90.8%, with a range of 71.5–98.9% across all ingestibles. Remaining coverage variability is expected to be caused by factors such as personal habits (for example, relative placement of recording devices at night, removing the device for showering or performing sports), because an evaluation of body mass index (BMI) against coverage did not show a significant correlation. It is worth noting that the data from the ingestible with the receiver malfunction remain relevant and representative of the entire GI tract, as the time when there was no wearable receiver available is sparsely covered by data received from the non-wearable receiver and data recovered from the memory flash of the device. In general, data gaps were rather evenly distributed over the GI tract. The most typical gap size (mode as well as median of the distribution) in our data was a single missed measurement, whereas the 95th and 99th percentile gap sizes were 2.9 and 8.5 min, respectively. A total of 14 gaps were longer than 1 h, with a maximum of 2.5 h. Three ingestibles showed premature battery failure: subsequent investigation showed that this was due to a suboptimal conductive epoxy assembly step, and this assembly step was further optimized to avoid similar premature battery failures in future devices. We obtained high-temporal resolution sensor data, measured every 20 s. Figure 5 shows resulting pH and ORP traces. The different GI segments were annotated based on significant changes in pH recordings in accordance with generally accepted thresholds (Methods). Temperature readings were used to confirm pill exit (Fig. 1d). Combined data from all analysed ingestibles are shown in Fig. 5a. To account for vastly varying transit times between individuals (Extended Data Table 1), we rescaled the time-series data to physiological regions, making data visualization and interpretation across ingestibles more intuitive (Methods). We found mean (± s.d.) pH values of 2.6 (±1.6) in the stomach, 7.4 (±0.3) in the small intestine and 6.5 (±0.6) in the large intestine. ORP values decreased over the course of the GI tract, starting with an oxidative environment in the stomach (162 ± 70 mV), a slightly reducing state in the small intestine (−126 ± 60 mV) and a strongly reducing environment in the large intestine (−360 ± 16 mV). This is to be expected because there is a strong gradient of decreasing oxygen concentration and increasing microbial activity along the gut39, both of which influence redox potential. Our unique biochemical characterization of the healthy human gut can serve as a reference for future patient studies, with redox values rarely exceeding 230 mV in the stomach, −26 mV in the small intestine or −323 mV in the large intestine (95th-percentile values across all measurements), as shown in Fig. 1e. In a disrupted gut, we anticipate that these values will increase, shifting towards a more oxidative state. Although the average pH and ORP signals across the 66 ingestibles display a characteristic profile in which different GI regions can be Nature Electronics | Volume 8 | September 2025 | 856–870 861 Article https://doi.org/10.1038/s41928-025-01411-4 clearly distinguished, individual pH and ORP signals displayed substantial variability; examples are shown in Fig. 5c–f (see also Extended Data Figs. 4, 5 and 6). Most often, the ORP signal was stable across most of the small intestine before dropping steeply upon entering the large intestine (Fig. 5c). In other cases, the ORP signal decreased in two distinct steps within the small intestine, potentially corresponding to the jejunum and ileum section (Fig. 4d). In the remaining cases, the ORP decreased more gradually (Fig. 4e) or the decline occurred much earlier in the small intestine (Fig. 5f; see also Extended Data Fig. 5 for a more extensive characterization of the range of ORP profiles observed in the small intestine and Extended Data Fig. 6 for the pH profiles). This variability in the ORP signal appeared to be consistent within participants; there was a significant difference among participants in the relative timing of the pH and ORP drops around the transition into the large intestine (Kruskal–Wallis test; H(12) = 26.8, P = 0.008, η 2 = 0.31). This suggests that the timing of these events may reflect some stable physiological or anatomical properties that vary between individuals. A comparison between the ingestibles administered in the morning on an empty stomach and those taken in the evening shortly after a meal is depicted in Fig. 5b. The pH signal in the stomach fluctuates strongly depending on stomach contents. The pH dips significantly lower for the evening ingestibles (P = 9.999 × 10−5; cluster-based permutation test), which has been described previously40. pH signals also diverge significantly in the small intestine, rising to a higher 0 5 10 15 20 25 30 35 40 Time (h) 0 1 2 3 4 5 6 7 8 9 pH b –500 –400 –300 –200 –100 0 100 200 300 ORP (mV) pH ORP 0 5 10 15 20 25 30 35 40 45 50 Time (h) 0 1 2 3 4 5 6 7 8 9 pH c ORP (mV) 0 10 20 30 40 50 60 70 80 90 100 Time (h) 0 1 2 3 4 5 6 7 8 9 pH d –500 –400 –300 –200 –100 0 100 200 300 –500 –400 –300 –200 –100 0 100 200 300 –500 –400 –300 –200 –100 0 100 200 300 ORP (mV) 0 10 20 30 40 50 60 70 80 90 100 Time (h) 0 1 2 3 4 5 6 7 8 9 pH e Large intestine confirmed by ultrasound ORP (mV) 0 10 20 30 40 50 Time (h) 0 1 2 3 4 5 6 7 8 pH a –500 –400 –300 –200 –100 0 100 200 300 ORP (mV) –400 –300 –200 –100 0 100 200 300 ORP (mV) 17.5 18.5 19.5 20.5 21.5 22.5 Time (h) 1 2 3 4 5 6 7 8 pH pH ORP Fig. 4 | In vivo validation of the GISMO ingestible sensor in pigs. a, Measured ORP (mV) and pH profiles along the pig’s gut. A zoom-in is provided of the small intestine–large intestine transition. The vertical lines indicate the stomach–small intestine and small intestine–large intestine transition based on combined pH and ORP data. b–e, Sensor readings of an ingestible showing no distinct small intestine–large intestine transition in pH data, but it was evident in the ORP data. The vertical line indicates entry into the large intestine based on the ORP data. The ingestible was rapidly emptied from the stomach; hence, there is only one transition line. c, An ingestible with a very comparable profile to b. d, Another example where ORP could aid localization. e, Similar profile as b–d. For this ingestible, an ultrasound was performed after 67 h confirming that the ingestible was in the large intestine, as signalled by the ORP data. Nature Electronics | Volume 8 | September 2025 | 856–870 862 Article https://doi.org/10.1038/s41928-025-01411-4 level in the evening ingestibles (P = 0.023). Similarly, the ORP signal differs significantly between morning and evening ingestibles in the distal small intestine (P = 0.043) as well as for a brief period in the stomach (P = 0.047). Redox measurements to improve ingestible localization Knowing the GI segment in which the ingestible is located is key to assessing regional transit times in patients with motility disorders, to correlate biomarker measurements to specific gut segments (for example, to locate inflammation) and to guide drug-release or sampling capsules. Although pH measurements are generally accepted for segmentation, the transition between small and large intestine can be difficult to pinpoint33. This is corroborated in both our animal study and our clinical trial, in which we observed cases of challenging GI segment annotation based on pH alone, but when combined with ORP, distinct gut regions could be discriminated. Figure 6a shows an example in which clear segmentation based on either pH or ORP is possible, because they drop at the same time. In other cases, the ORP measurement showed a significant change in redox environment, although a clear transition in pH was absent (Fig. 6b,c). We note that in the majority of ingestibles, the ORP drop was observed before the pH drop, and in most other cases they occurred simultaneously. This is in line with a scintigraphy study following the location of a radiolabeled wireless motility capsule while recording pH values, which noted that ingestibles pass the ileal–caecal junction before the pH signal starts Time (relative to segments) 0 1 2 3 4 5 6 7 8 pH –400 –300 –200 –100 0 100 200 ORP (mV) a pH ORP Stomach Small intestine Large intestine Stomach Small intestine Large intestine Time (relative to segments) 0 1 2 3 4 5 6 7 8 pH *** * –400 –300 –200 –100 0 100 200 ORP (mV) ** b pH, morning pH, evening ORP, morning ORP, evening 0 4 8 12 16 20 24 Time (h) 1 2 3 4 5 6 7 8 pH c –400 –300 –200 –100 0 100 200 ORP (mV) 0 4 8 12 16 20 24 Time (h) 1 2 3 4 5 6 7 8 pH d –400 –300 –200 –100 0 100 200 ORP (mV) 0 4 8 12 16 20 24 Time (h) 1 2 3 4 5 6 7 8 pH e –400 –300 –200 –100 0 100 200 ORP (mV) 0 4 8 12 16 20 24 Time (h) 1 2 3 4 5 6 7 8 pH f –400 –300 –200 –100 0 100 200 ORP (mV) Fig. 5 | Characterization of pH and redox (ORP) measurements along the human GI tract. a, Average pH and ORP recordings of all 66 pills analysed, ingested by 15 participants, linearly rescaled to equal-length compartments (stomach, small intestine and large intestine). Shaded error margins indicate the mean ± 1 s.d. across ingestibles. b, As in a, but with data from phase 2 of the experiment only, split into ingestibles swallowed in the morning (on an empty stomach; n = 26) compared to ingestibles swallowed in the evening (shortly before a meal; n = 25). Whiskered horizontal lines at the bottom indicate clusters of contiguous time points that were found to be significantly different between morning and evening pills in a cluster-based permutation test (N = 10,000 permutations; *P < 0.05, ***P < 0.0001; cluster P values, from left to right: P = 0.047, P = 9.999 × 10−5, P = 0.023, P = 0.043; all P values are two-tailed). c–f, Illustrative examples of different ORP profiles (along with pH signals from the same ingestibles) observed across the experiment without rescaling to physiological regions. For clarity of exposition, only the first 24 h of data are shown. Vertical lines indicate the (manually annotated) times of smalland large-intestine entry. c, The most common profile, consisting of two sharp drops coinciding with the ingestible’s entry into the small and large intestine, as identified based on pH. d, An example in which the ORP profile shows an additional, intermediate step drop in the small intestine. e, An example in which the ORP profile shows a gradual decrease in the small intestine instead of a sharp drop. f, ORP profile where the drop in the small intestine is initiated substantially earlier (about halfway through the small intestine) than the drop in pH. Nature Electronics | Volume 8 | September 2025 | 856–870 863 Article https://doi.org/10.1038/s41928-025-01411-4 dropping 41 . This suggests that ORP measurements could be a better indication of the anatomical location of large-intestine entry. The combined ORP and pH measurements, transmitted to the receiver in real time every 20 s, offer superior localization compared to pH data alone. Clinical impact and applications The GISMO capsule offers easy, affordable and non-intrusive measurements of GI redox balance without requiring any unpleasant bowel preparation, providing a potential substantial clinical impact. Redox sensing gives a combined measure of various physiological processes essential for maintaining gut health and, consequently, overall well-being. Redox balance is not a specific measure of one process but rather a measure of a complex interplay between the host, gut microbiome and immune system that could be used to monitor a variety of gut disturbances. In diseases and disorders of the GI tract, such as IBD and GI cancers, inflammation and dysregulation of the immune response combined with changes in the microbial community lead to an imbalance of oxidants and antioxidants5,7,42. For example, indirect markers of redox state in mucosal biopsies of IBD patients have shown a 10-fold to 100-fold increase in reactive oxygen metabolites43,44. Lacking tools to measure ORP in vivo, the direct impact of inflammation on redox potential can only be estimated from limited cell culture experiments. Research in neutrophils has shown that an induced inflammatory event leads to a significant, dose-dependent rise in ORP, providing a proof of concept for the use of an ingestible ORP sensor to detect inflammation 25 . An ingestible device that measures inflammation-related markers in the GI tract could significantly impact the patient journey, particularly in screening, stratification and monitoring. For instance, in IBD patients, about one-third do not respond to initial treatments, and half of the patients experience diminishing treatment effectiveness over time45. Inadequate treatment directly affects quality of life, and prolonged inflammation increases the risk of irreversible long-term damage to the intestinal wall46. There is a clear need for objective biomarkers that can help guide decisions on changing treatment, adjusting dosage or transitioning to surgery, to ensure that patients receive the right treatment as quickly as possible47. In follow-up trials with the GISMO ingestible, the potential of an ingestible ORP sensor to quantify both the severity of inflammation and its approximate location will be evaluated. These capabilities would enable more frequent monitoring compared to invasive procedures like colonoscopy or endoscopy and offer a more objective and timely assessment compared to symptom-based evaluations. Redox balance in the gut plays a vital regulatory role not only in GI diseases but also in shaping overall health through its influence on the gut microbiome. Redox balance regulates how effectively bacteria can grow in a certain environment3. Our bodies’ immune system uses this mechanism by modulating the chemical environment to fight off invaders2. Changes in redox potential affect not only pathogenic bacteria but also the large community of commensal gut microbes, which are a crucial element of the gut barrier48,49. The presented data show that in healthy people with an anticipated healthy microbiome, the large intestine is strongly reducing. Microbiome disturbances are expected to shift this balance to an unfavourable, more oxidative environment. Limited faecal and post mortem analyses in mice have shown that microbiome-disrupting antibiotics can induce ORP increases as large as 200–400 mV (refs. 49,50), far exceeding the 72-mV-wide range (5th–95th percentile) that we observed in the large intestine of our healthy population (see also the interand intra-participant variability displayed in Extended Data Fig. 4b). The GISMO capsule has the potential to offer insights into microbiome function, moving beyond current methodologies that mainly rely on faecal samples. In cases of suspected dysbiosis, ingestible redox measurements could offer a comprehensive view of gut health across the entire GI tract. ORP measurements may also be easier to interpret compared to microbiome composition data, which can be challenging to analyse due to functional redundancy and high dimensionality. Further clinical studies are needed to provide deeper insights into how ORP changes are influenced by microbiome alterations: for example, involving individuals undergoing antibiotic treatment, given the drastic impact of antibiotics on the gut environment. Insights into this antibiotics-eradicated microbiome on the one hand and the microbiome of healthy people on the other hand could show the two extremes of the spectrum and create a foundation for the interpretation of gut redox studies in other contexts, such as the gut–brain axis, aging and (nutritional) interventions. In our small cohort of healthy volunteers, the data suggest that redox signals in the large intestine are relatively comparable between participants and remain stable over time, when comparing the five ingestibles administered per participant over a 0 3 6 9 12 15 18 21 24 Time (h) 0 1 2 3 4 5 6 7 8 9 pH apH↓ t = 6.5 h –500 –400 –300 –200 –100 0 100 200 300 ORP (mV) ORP↓ t = 6.5 h pH↓ region pH ORP 0 3 6 9 12 15 18 21 24 Time (h) 0 1 2 3 4 5 6 7 8 9 pH b –500 –400 –300 –200 –100 0 100 200 300 ORP (mV) ORP↓ t = 5.8 h 0 3 6 9 12 15 18 21 24 Time (h) 0 1 2 3 4 5 6 7 8 9 pH c –500 –400 –300 –200 –100 0 100 200 300 ORP (mV) ORP↓ t = 2.2 h Fig. 6 | ORP measurements to support localization. a, Clear small intestine– large intestine transitions are visible in both pH and ORP, indicated by the two arrows. b, Example of a case in which it is difficult to pinpoint entry into the large intestine based on pH alone and ORP could be of added value. c, Another example, showing a big jump in ORP and only a gradual drop in pH. The grey shaded area shows the uncertainty region for pH. The ORP drop, indicated by the arrow, could help in these cases. For clarity of exposition, only the first 24 h of data are shown. t, time. Nature Electronics | Volume 8 | September 2025 | 856–870 864 Article https://doi.org/10.1038/s41928-025-01411-4 four-month period. The redox signal was not significantly affected by our dietary intervention (Extended Data Fig. 7), although the diets were not specifically designed for that purpose. The diets were designed to create a contrast in protein fermentation. Besides the effect of dietary pro-oxidants and antioxidants that may survive absorption in the small intestine (for example, phenolic compounds), metabolites produced by microbiota during fermentation, of proteins but also carbohydrates, have a large effect on the intestinal environment and gut health 51,52 . The interplay between nutrition, microbiome and redox balance impacts fermentation and consequently the production of metabolites. We foresee that if these relationships are better understood in the future, the GISMO ingestible could play a role in personalized nutrition. Individuals with disrupted gut health could compare their data to (personal) reference values and assess how dietary and lifestyle changes might shift their data closer to their own previous healthy baseline or to a healthy reference population. The in-human redox values reported in this study vary substantially along the GI tract. This is to be expected because there is a strong gradient of decreasing oxygen concentration and increasing microbial activity along the gut39, both influencing redox potential. We observed that this variation of ORP along the tract could enhance localization of ingestibles. Incorporating ORP data alongside pH information could boost confidence in assessing regional transit times (for example, in studying motility disorders), guiding sampling and drug-release capsules and linking measurements to specific regions of the gut. Follow-up studies with a reference standard (such as ultrasound) should validate the use of ORP for localization. This variation along the tract also means that assessment of redox-based gut-health measures should be done per segment, based on group averages or, preferably, personalized predictions. In this work, we provide a reference dataset of gut redox values of healthy volunteers for the different GI segments. A limitation of using ingestibles to assess gut health is the non-specificity of redox measurements. Although this can be an advantage as a general indicator of gut health, it may require supplementation with other clinical evaluations to achieve a specific diagnosis or provide tailored advice. Despite this limitation, given the scarcity of tools capable of directly objectively measuring inside the GI tract, even non-specific measurements can offer significant value. Follow-up studies are expected to provide greater insight into specific clinical applications. Another limitation of the GISMO ingestible is its inability to measure at a specific location, as it relies on gut motility. Additionally, long-term measurements with the same device are not feasible; wearable sensors are more suitable for that purpose. However, the ease of swallowing and minimal impact on the user make repeated measurements with the GISMO ingestible practical. A final limitation is the risk of device retention, especially in patients with strictures: for example, in patients with Crohn’s disease. The small size of our capsule (three times smaller than camera capsules and a commonly used pH capsule) should reduce this risk, but those patients should be closely monitored. Conclusions We have reported the development, validation and clinical testing of ingestible sensors that can measure in vivo redox balance along the human gut. The sensors are designed to work without any unpleasant bowel preparation and provide real-life measurements. Further clinical investigations will be required to evaluate our GISMO ingestible sensor in patient populations. We anticipate that in vivo ORP measurements can offer functional insights into the gut microbiome and could enable easy localization and objective quantification of inflammation, facilitating improved disease diagnosis and monitoring. Given the central role of the gut in maintaining overall health, our ingestible sensors could enhance understanding of gut-linked diseases, the gut–brain axis, aging53, (nutritional) interventions5,6,54,55 and drug–gut interactions5,56,57. Methods Ingestible architecture, sensors, electronic design and electrical testing The ingestible architecture is shown in Fig. 2b. At the centre of that architecture is a low-power ARM Cortex M4F-based microcontroller (STM32L432, ST Microelectronics). It is used to control the system (sensing, communication, turn-on/off and so on) and to encrypt the data using AES-128. The system electronics were implemented on a stack of two 400-µm thin, four-layer high-density interconnect 5.8 mm × 10.8 mm PCBs, as shown in Fig. 2c. A third layer implements the sensor functionality on a 150-µm thin two-layer polyimide flexible PCB. This combination of thin rigid and flexible boards allowed optimizing the volume taken up by the electronics in a tradeoff between interconnect density and fine-pitch component assembly on the one hand and substrate thickness on the other hand. A simplified schematic diagram of the ingestible system electronics is shown in Supplementary Fig. 1. A sub-gigahertz radio transceiver from ST Microelectronics (S2LP) is used to communicate with the receiver, to receive device configuration and send measurement data. We selected the conformal stepped impedance antenna design of ref. 30 as the basis for the GISMO pill antenna. Reference 30 showed that this type of conformal microstrip antenna radiates efficiently in lossy tissue, is robust against changes in dielectric loading caused by the surrounding tissue, occupies minimal volume in the pill (hence allowing maximum space for electronic circuitry and sensors) and aims for a communication range up to 15 m for deep (>5 cm depth) implants 58 . We optimized this conformal stepped impedance antenna for operation in the selected 868-MHz frequency band considering the dimensions, the dielectric properties of the filling material and the encapsulation selected for the GISMO pill. By adjusting the dimensions of the ground plane, the patch, the slots on the ground plane and the insets on the patch, we realized an efficient conformal antenna for the GISMO pill. The antenna is manufactured as a flexible circuit board etched out of copper on a polyimide substrate and measures 10 mm × 17 mm with a thickness of 172 µm. The antenna is attached to the inner wall of the PEEK tube that contains the electronics and is filled with epoxy. The simulated radiation efficiency (η) in a spherical phantom with a radius of 5 cm and filled with muscle tissue equalled 0.5%. This outperforms commercial small in-body antennas below 1 GHz, which show radiation efficiencies ranging between 0.02% and 0.3% (ref. 59). Immersed in a large body phantom of 40-cm diameter filled with muscle tissue, the optimized conformal stepped impedance antenna achieved a realized gain between −33.2 and −20.8 dBi (Extended Data Fig. 2). The negative value for the realized gain is a consequence of the lossy nature of human body tissue. The entire communication was extensively validated in vitro in a large barrel (120 l, 40-cm diameter) filled with stomach-simulating liquid as described by ref. 60 (measured relative permittivity of 65.2 and conductivity of 1.17 S m −1 at 868 MHz). The −10-dB impedance bandwidth covered the targeted communication band at 868 MHz (ref. 39). Radio communication was verified at various distances between capsule and receiver. A reception rate of 95–100% was obtained at up to 20 cm (the centre of the 40-cm phantom, in line with the 95th-percentile waist circumference, females and males combined, as reported in the US National Health and Nutrition Examination Survey 61 ). We found that at the chosen frequency, for the S2LP radio coupled to our custom antenna, +6 dBm (4 mW) of transmission power was sufficient for reliable communication. For the ORP sensor, we used a platinum electrode on a Si/SiOx substrate (sputtering 40 nm Pt + 5 nm Cr as adhesion layer, 700 µm × 850 µm; MicruX). As shown in Supplementary Fig. 1, the ORP sensor voltage is measured with a high-impedance amplifier circuit (AD8505ACBZ, Analog Devices) with input current <1 pA and digitized by the microcontroller analog-to-digital converter. For the ISFETs, a readout circuit similar to the circuit described by ref. 62 was implemented. This circuit biases the Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Table 1 | Regional and total transit times as measured by the GISMO ingestible Parameter Group NMedian (h:min) Min (h:min) Max (h:min) 5th percentile (h:min) 95th percentile (h:min) Gastric emptying time All 66 01:11 00:03 13:12 00:18 11:12 Baseline 15 00:39 00:18 01:16 00:20 01:13 Diet WPI 24 04:31 00:03 13:12 00:33 12:06 Diet BPP 27 04:29 00:06 11:13 00:13 10:58 Morning 41 00:46 00:03 05:35 00:11 05:00 Evening 25 09:48 01:08 13:12 01:49 12:04 Small intestinal transit time All 65 06:17 01:33 10:38 03:00 09:05 Baseline 15 06:15 02:21 09:39 03:27 09:16 Diet WPI 24 06:13 01:33 10:04 02:49 08:35 Diet BPP 26 06:52 02:48 10:38 03:53 08:55 Morning 41 06:18 02:21 10:38 03:45 09:39 Evening 24 06:07 01:33 08:45 02:50 08:24 Large intestinal transit time All 65 45:58 06:07 168:48 13:54 121:37 Baseline 15 41:00 13:34 97:32 16:18 90:39 Diet WPI 24 45:13 13:03 138:37 16:26 114:07 Diet BPP 26 54:42 06:07 168:48 08:31 140:47 Morning 41 45:07 06:17 168:48 15:13 121:57 Evening 24 51:17 06:07 147:04 13:58 117:47 Whole gut transit time All 66 61:46 14:13 178:03 22:36 133:18 Baseline 15 49:01 22:26 104:28 23:01 97:41 Diet WPI 24 52:36 15:45 146:28 23:27 121:34 Diet BPP 27 71:40 14:13 178:03 17:53 151:50 Morning 41 51:51 15:34 178:03 23:04 132:55 Evening 25 73:03 14:13 159:45 17:47 130:18 The ingestibles are grouped in the different experimental conditions. All: summarizing the data of the 66 ingestibles of the entire study; Baseline: data of phase 1 of the study, participants are on their habitual diet, ingestibles are given in the morning; Diet WPI: Whey Protein Isolate diet intervention, including the morning and evening pills; Diet BPP: Bovine Plasma Protein, including the morning and evening pills; Morning: all ingestibles given in the morning, including baseline WPI and BPP diet; Evening: all ingestibles given in the evening, including WPI and BPP diet. Note one of the ingestibles had a premature shutdown in the small intestine, hence missing small and large intestinal transit times, retrieval of the pill gave us the whole gut transit time. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 1 | Reference electrode (RE) design and optimization. (a) Optical image of the RE dome of the ingestible; the white circular area represents the porous frit material. (b) 3D drawing of the ingestible case where the red dome represents the reference electrode volume. (c) Drift and (d) electrochemical impedance study of a series of RE having three different frit diameters, that is 0.8, 1 and 1.5 mm. The diameter of the porous frit was evaluated to optimize the RE performance. The RE with a 1 mm frit diameter shows the best compromise in terms of drift and electrochemical impedance. (e) Potential profile of 10 ingestible reference electrode versus a commercial Ag/AgCl RE recorded in 0.1 M KCl at 37 °C. The potential of the REs was monitored for more than 325 h showing highly reproducibility and an overall drift < 0.06 V/h. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 2 | Optimization and performance of the custom-designed ingestible antenna. (a) Overview of simulated optimal communication frequencies from an in-body capsule for different capsule depths. Simulation of an ingestible communicating from the center of spherical phantoms of different radius show the selected 868 MHz band as the most optimal license-free band that will work well over a various range of capsule depths. (b) Measured realized gain of the antenna in a large phantom of 40 cm diameter filled with muscle tissue, with values between −33.2 dBi and −20.8 dBi. Negative value is a consequence of the lossy nature of the human body tissue. (c) Measured S11 reflection of used antenna in phantom. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 3 | GISMO ingestible device power consumption. (a) Power profile in full functionality mode (all capsule sensor signals measured). (b) Power profile in reduced functionality mode (limited subset of signals measured, aiming to detect ingestible excretion or confirm its in-body presence for longer time than expected). (c) Operation time obtained from tests in the oven, confirming the 5 days in full functionality mode and well over the required 14 days including reduced functionality mode. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 4 | Betweenand within-subject variability in pH and ORP measurements. Variability was calculated over linearly rescaled pH (left panel) and ORP (right panel) time courses. This analysis only included subjects for whom data from all five experimental conditions were available. Total variability was calculated as the standard deviation across all ingestibles (for each rescaled time point – this is equivalent to the shaded error margins presented in Fig. 5a). To calculate between-subject variability, time courses were first averaged across all conditions (diets and morning and evening pills) for each subject, and then the standard deviation was computed across these subject-average time courses. Strict within-subject, repeated-measurements variability is somewhat challenging to estimate on our data, since no experimental condition was precisely repeated within the experiment. However, while we saw clear differences between morning and evening ingestibles (Fig. 5b), no significant differences could be detected between ingestibles swallowed the WPI and BPP diets (Extended Data Fig. 7). For the purposes of this analysis, we therefore take these as our best available proxy of repeated measurements (two measurements each, of the morning and evening conditions) and calculated the standard deviation across these data points. Note that there is high variability in the stomach for both sensor modalities, due the ingestion of food and drink, which can cause dramatic fluctuations in the chemical composition of the stomach contents. In the small intestine, a clear difference can be seen between pH and ORP, with the former being far more consistent. ORP is especially variable towards the end of the small intestine, as we noted previously, but appears to be more consistent within, than between subjects there. This is consistent with our observation that the relative timing of the ORP and pH drops towards the end of the SI was also somewhat consistent within subjects. Interestingly, ORP is very stable in the large intestine of these healthy subjects, which lends confidence that these measurements represent a reliable healthy baseline against which deviations, linked to disturbances in gut health, might be detected. Finally, it is worth bearing in mind that the variability shown in these plots can be caused by differences in sensor measurements at a given location in the gut, or by differences in the rates of transit of different capsules through different stages of the GI tract. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 5 | Overview of all observed ORP profiles. ORP signals from all ingestibles, linearly rescaled to equal-duration segments (stomach, small intestine and large intestine), and plotted in separate panels for each of the 15 participants, with up to 5 different conditions for each participant plotted in different colors. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 6 | Overview of all observed pH profiles. pH signals from all ingestibles, linearly rescaled to equal-duration segments (stomach, small intestine and large intestine), and plotted in separate panels for each of the 15 participants, with up to 5 different conditions for each participant plotted in different colors. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 7 | No significant effect of dietary intervention on pH, ORP, or transit times. (a) Lines and shaded regions indicate the mean ± 1 s.d. of the (linearly rescaled) pH and ORP signals in the two different controlled diet conditions. Cluster-based permutation tests revealed no clusters of time points where a significant difference between these two diets could be observed: for the ORP data, the smallest (two-tailed) cluster Pvalue was 0.33; for the pH data, no candidate clusters (contiguous sequences of time points which each exceed an individual, uncorrected t-test threshold of p < 0.05, two-tailed) were observed to begin with, nor did any individual time points from either sensor meet the Bonferroni-adjusted significance threshold (adjusted for the number of time points being compared). In short, we observed no significant effect of the dietary intervention on the pH or ORP measurements in any part of the GI tract. (b) Scatter plots of WPI-diet vs. BPP-diet transit times of the ingestible capsules through the stomach, small intestine, large intestine, and the whole gut, separated out for ingestibles taken in the morning and in the evening. Equality lines are plotted along the diagonal of each panel, such that points above the diagonal indicate capsules whose transit times were longer for the BPP diet than the WPI diet. Wilcoxon signed-rank tests revealed no significant effect of dietary intervention on any of these transit times (two-tailed Pvalues reported in panels; smallest Pvalue: p = 0.13, for the evening pills’ whole-gut transit time). Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 8 | In vitro validation of the GISMO sensors post-ingestion. Staircase measurement of ORP (a) and pH (b) profiles using three ingestibles post-ingestion retrieved from the participant of the clinical trial versus two commercial sensors (Horiba, Kyoto, Japan) in ORP (alternate) standards in the range of −550 mV to 600 mV with pH range between pH 0 and pH 14. The measurements were performed at body temperature (37 °C) within 2 h due to limited stability of ORP standard prepared in house by dissolving quinhydrone in pH buffers. The total transit times of these ingestibles were 46 h, 23 h, and 25 h. Nature Electronics Article https://doi.org/10.1038/s41928-025-01411-4 Extended Data Fig. 9 | Pre-ingestion calibration versus post-excretion calibration profiles of in total six pH ISFETs of three retrieved ingestibles with different transit times. (a) and (d) show ISFET1 and ISFET2 of a pill with a median drift profile (15.1 mV and 13.0 mV). (b) and (e) show the calibration of the pill with the shortest transit time (~0.5 day) with a drift of 5.1 mV and 3.0 mV; (c) and (f) show the same for an ingestible with the longest transit time of more than 7 days (12.9 mV and 17.6 mV drift).