Continuous, wearable gastrointestinal monitoring device using a multi-sensor approach
A wearable device with multi-sensor integration and contextual parameter analysis addresses the limitations of current gastrointestinal monitoring by enhancing specificity and sensitivity, enabling personalized and predictive diagnostics with real-time alerting.
Patent Information
- Application Number
- PCT/EP2026/063324
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-07
- Filing Date
- 2026-05-07
- Publication Date
- 2026-11-12
AI Technical Summary
Current gastrointestinal monitoring techniques are invasive, provide inadequate sensitivity and specificity, and lack real-time alerting capabilities, especially for abdominal sounds, and do not account for confounding variables like heart-rate, heart-rate variability, and activity levels.
A wearable device using a combination of acoustic, photoplethysmographic, and inertial sensors, along with a mobile application and cloud backend, for continuous monitoring and analysis of gastrointestinal sounds, incorporating static and dynamic contextual parameters to create personalized diagnostic and predictive tools.
Enhances specificity and sensitivity of gastrointestinal monitoring by reducing inter-individual variability, capturing longitudinal stimulus-response dynamics, and providing clinically interpretable outputs with real-time alerting capabilities.
Smart Images

Figure EP2026063324_12112026_PF_FP_ABST
Abstract
Description
[0001] CONTINUOUS, WEARABLE GASTROINTESTINAL MONITORING DEVICE USING A MULTI-SENSOR APPROACH
[0002] Technical field
[0003] The present invention relates to the technical field of non-invasive gastrointestinal monitoring and more particularly to the field of connected systems and wearable devices for said monitoring.
[0004] Background art
[0005] Gastroenterological conditions affect millions worldwide. However, diagnostics and monitoring techniques for these conditions (for example Endoscopy, capsule endoscopy, electrogastrogram, manometry and imaging techniques such as CT or MRI) tend to be single-point-in-time and are typically invasive.
[0006] Early research has indicated that the monitoring and analysis of abdominal sounds holds significant potential for diagnostic and prognostic value. Abdominal sounds refer to any noises originating from within the abdomen, including stomach sounds and bowel sounds produced during digestive processes. These sounds may be normal or abnormal.
[0007] Abnormal abdominal sounds, changes in sound patterns, or a complete absence of sounds can be indicative of various pathologies, such as bowel obstruction, gastrointestinal bleeding, inflammation, constipation, or ischemic bowel disease. Abdominal sounds can be recorded by placing an acoustic capture device, such as a microphone or stethoscope, on the surface of the abdomen. Herein, the terms bowel sounds, stomach sounds, and abdominal sounds refer to noises originating from within the abdomen, for example originating from digestive processes.
[0008] However, analysis of abdominal sounds alone does not provide adequate sensitivity and specificity for real-world clinical utility. Abdominal sounds are one of several variables linked to gastrointestinal health; many other confounding variables, such as heart-rate,
[0009] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteheart-rate variability, activity levels, and clinical symptoms, have not been considered in gastrointestinal monitoring systems that analyse abdominal sounds alone.
[0010] Furthermore, due to the long periods of time over which abdominal sounds change and respond to stimuli, abdominal sound analysis must be implemented in devices that can be worn continuously by a patient. Typical digital stethoscope solutions do not have a form factor or technical capabilities suitable for continuous use.
[0011] Finally, in response to changes in abdominal sound or bowel sound behaviour indicating potentially life-threatening disease states (such as ileus), there is a need for internet connectivity enabled systems that can provide real-time alerting functionalities.
[0012] The Applicant has developed a wearable device and supporting analysis techniques to address these main technical challenges.
[0013] Disclosure of the invention
[0014] The herein described invention specifies a method of monitoring the gastrointestinal (Gl) tract using a unique combination of sensor types and a supporting hardware / software architecture which enables analysis of these data sources.
[0015] According to a first aspect of the invention, a wearable device for real-time human gastrointestinal monitoring based on acoustic signals is provided, wherein the device is configured to be worn at an abdominal area of a human for non-invasively recording sounds generated by a gastrointestinal system of the human. The device comprises at least a first and a second acoustic sensor configured to capture acoustic signals; at least one photoplethysmographic sensor oriented towards a surface of the abdomen when the device is being worn and configured to measure a photoplethysmographic signal; and at least one inertial measurement unit configured to measure a movement signal of the wearable device. A sensing area of the first acoustic sensor is oriented towards a surface of the abdomen when the device is being worn and is configured to capture a first gastrointestinal acoustic signal. A sensing area of the second acoustic sensor is oriented outwards in a direction opposite a surface of a human abdomen when the device is being
[0016] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteworn and is configured to capture an environmental acoustic signal of ambient noise in an ambient environment around the human. The wearable device further comprises a microcontroller, comprising processing circuitry, connected to measuring outputs of at least the first and second acoustic sensors, of the at least one photoplethysmographic sensor, and of the at least one inertial measurement unit, and configured to perform monitoring and analysis of the gastrointestinal system of the human based on at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal.
[0017] In a second aspect of the invention, a gastrointestinal monitoring system comprises the wearable device according to the invention and further a mobile application for a mobile device such as a smartphone, tablet, or personal computer, wherein the mobile application is configured to accept inputs from a user of the wearable device and a computer server comprising processing circuitry for additional compute resources. The computer server is wirelessly connected to the microcontroller of the device and is configured to receive at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal from the microcontroller, wherein the microprocessor is optionally configured to pre-process these signals before they are received by the computer server; and optionally, a cloud backend configured to provide an alerting functionality.
[0018] The wireless connection between the microcontroller of the device and the computer server may include wired portions. For instance, the microcontroller of the device may exchange data over a wireless connection with a hub or gateway (constituted e. g. by a local computing device such as a smartphone, tablet or notebook computer), which in turn communicates with the server over an available network connection, e. g. over the internet. The communication of the hub / gateway with the server may be wireless, wired or partly wireless and partly wired.
[0019] In a further aspect of the invention, a method for monitoring and analysis of a gastrointestinal system of a human based on acoustic signals using a wearable device or a wearable device and a computer server is provided. The method comprises:
[0020] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltecollecting a first gastrointestinal acoustic signal generated by the internal organ of interest using a sensing area of a first acoustic sensor of the device, wherein the first acoustic sensor is placed at an abdominal area of the human for non-invasively recording sounds generated by the gastrointestinal system of the human and a sensing area of the first acoustic sensor is oriented towards the body surface when the device is being worn;
[0021] collecting an environmental acoustic signal comprising ambient noise using a sensing area of a second acoustic sensor of the device, wherein the sensing area of the second acoustic sensor is oriented outwards in a direction opposite a surface of the human abdomen when the device is being worn;
[0022] measuring a photoplethysmographic (PPG) signal using at least one photoplethysmographic (PPG) sensor of the device, wherein the photoplethysmographic sensor contacts the body surface when the device is being worn;
[0023] measuring a movement signal of the device using at least one inertial measurement unit (IMU) of the device; and
[0024] performing monitoring and analysis of the internal organ of interest based on at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal.
[0025] In a further aspect of the invention, a method of diagnosing a target condition relative to the gastrointestinal system based on acoustic signals using a wearable device or using a wearable device and a computer server (on premises or cloud-based) is provided.
[0026] Preferably, in the context of the gastrointestinal monitoring system, the mobile application is configured to accept inputs from the user of the wearable device which comprise a food intake signal, a toileting event signal, and / or a medication intake data. This information is particularly relevant in the context of monitoring the gastrointestinal system of the human.
[0027] The input data may be used to annotate raw and / or processed sound data for the training of models for identification of specific events (e. g. food intake, predicting toileting). Food
[0028] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteintake information may be used as a future predictor for other events like toileting or symptoms. Correlations between specific food intake and / or medication intake on the one hand and resulting bowel sounds and / or symptoms (provided e. g. as further input data by the user) on the other hand may be identified, providing valuable information on the effect of nutrition and / or medication on the well being of the patient.
[0029] In preferred embodiments, the computer server or the wearable device microcontroller (MCU) is configured to:
[0030] maintain a signal timeline formed by continuously acquired data comprising at least the first gastrointestinal acoustic signal, the photoplethysmographic signal, and the movement signal;
[0031] receive and store at least one static contextual parameter describing a stable characteristic of the human subject and / or at least one dynamic contextual parameter, wherein the at least one dynamic contextual parameter is assigned a timestamp; and
[0032] map the at least one static contextual parameter and / or the at least one timestamped dynamic contextual parameter onto the physiological signal timeline to form a unified multimodal record, and perform monitoring and analysis of the gastrointestinal system based on the unified multimodal record.
[0033] The use of contextual parameters, including static and / or dynamic parameters, alongside the continuously acquired physiological sensor signals provides a number of significant advantages over a system relying exclusively on sensor data in the context of gastrointestinal monitoring.
[0034] A first advantage resides in the reduction of inter-individual variability. The acoustic characteristics of gastrointestinal sounds, the cardiovascular autonomic response reflected in the photoplethysmographic signal, and the movement patterns captured by the inertial measurement units vary substantially between individuals as a function of stable personal characteristics such as age, sex, and body composition. By incorporating static contextual parameters as covariates, personalized reference baselines and personalized anomaly
[0035] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltedetection thresholds for the gastrointestinal monitoring may be generated, creating a genuinely individualized diagnostic and predictive instrument.
[0036] A second advantage resides in the increased specificity of gastrointestinal event detection and condition prediction. Many gastrointestinal acoustic signal patterns that are ambiguous when considered in isolation become unambiguous when interpreted in the context of a concurrent or preceding dynamic contextual event such as a dietary intake, a medication intake, or a reported symptom. The mapping of such dynamic contextual parameters onto the physiological signal timeline in the unified multimodal record substantially resolves such ambiguities and correspondingly increases the specificity of the gastrointestinal monitoring output.
[0037] A third advantage resides in the ability to capture longitudinal stimulus-response dynamics of the gastrointestinal system. Dynamic contextual parameters provide semantically labelled anchors on the continuous physiological signal timeline, enabling the accumulation of individual-level gastrointestinal response profiles — for example in relation to specific dietary intake patterns or pharmacological interventions — over extended monitoring periods, achieving a degree of longitudinal personalization that is fundamentally inaccessible to single-point-in-time gastrointestinal diagnostic methods.
[0038] A fourth advantage resides in the improved clinical interpretability and actionability of the gastrointestinal monitoring outputs. When an anomalous gastrointestinal signal pattern is presented together with its temporally associated contextual events, the output acquires a clinical narrative that is directly interpretable by the healthcare professional and actionable in the context of gastrointestinal patient management.
[0039] In further preferred embodiments, the computer server or the wearable device microcontroller (MCU) is configured to:
[0040] receive and store at least one static contextual parameter of the human subject, wherein the static contextual parameter comprises a demographic parameter relating to the human subject and / or a clinical health status indicator for the human subject; and
[0041] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltecondition the monitoring and analysis of the gastrointestinal system on the at least one static contextual parameter by applying the static contextual parameter as a covariate.
[0042] The demographic parameter includes in particular at least one of the following: age, sex, height, weight, body mass index (BMI). The clinical health status indicator includes in particular one of the following: a known disease diagnosis, a medical history entry, a record of prior surgical interventions.
[0043] In particular, the conditioning leads to generating a personalized reference baseline for at least one of the first gastrointestinal acoustic signal, the photoplethysmographic signal, and the movement signal that is specific to the individual human subject, wherein deviations from the personalized reference baseline are used as inputs to anomaly detection and alerting logic.
[0044] The conditioning may be advantageous in the context of the monitoring as well as in predictive models.
[0045] In further preferred embodiments, the computer server or the wearable device microcontroller (MCU) is configured to:
[0046] receive at least one dynamically reported contextual event comprising a dietary intake event, a defecation event, a medication intake event, and / or a subjective symptom report;
[0047] assign a timestamp to each dynamically reported contextual event and align each dynamically reported contextual event with the gastrointestinal signal timeline formed by the first gastrointestinal acoustic signal, the photoplethysmographic signal, and the movement signal; and
[0048] extract, for each dynamically reported contextual event, a physiological feature trajectory from the first gastrointestinal acoustic signal, the photoplethysmographic signal, and / or the movement signal within at least one of a configurable pre-event window and a configurable post-event window relative to the timestamp of the
[0049] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltedynamically reported contextual event, and correlate the physiological feature trajectory with the dynamically reported contextual event.
[0050] In particular, the at least one dynamically reported contextual event is received from the mobile application. The subjective symptom report may comprise in particular a pain level, a gastrointestinal discomfort rating, a nausea rating, a bloating rating, and / or a stool consistency rating.
[0051] In further preferred embodiments, the computer server or the wearable device microcontroller (MCU) is configured to:
[0052] receive dietary intake data and / or medication intake data comprising a time of intake; and
[0053] derive, from a correlation between the dietary intake data and / or the medication intake data and a feature trajectory of the first gastrointestinal acoustic signal within a post-event window and optionally within a pre-event window, at least one of:
[0054] (a) an individual food-response model characterizing the effect of a dietary intake event on gastrointestinal motility as reflected in the first gastrointestinal acoustic signal;
[0055] (b) a pharmacodynamic response metric characterizing the effect of a medication intake event on gastrointestinal motility as reflected in the first gastrointestinal acoustic signal, wherein a change in the first gastrointestinal acoustic signal within a pharmacokinetically expected time window following the medication intake event is used as a measure of pharmacological efficacy; and
[0056] (c) a drug-food interaction metric characterizing the combined effect of a concurrent dietary intake event and a medication intake event on the physiological feature trajectory of the first gastrointestinal acoustic signal.
[0057] The dietary intake data and / or medication intake data may be received from the mobile application. The dietary intake data may comprise in particular a food type, a meal
[0058] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltecomposition, a caloric load, a fermentable oligosaccharide, disaccharide, monosaccharide, and polyol (FODMAP) content. The medication intake data may comprise in particular a medication type and / or a dose.
[0059] The computer server or the wearable device microcontroller (MCU) may be configured to predict a target condition such as a target gastrointestinal condition based on an analysis of at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal.
[0060] In particular, the computer server or the wearable device microcontroller (MCU) is configured to:
[0061] receive data relating to at least one target event associated with the gastrointestinal system, wherein the target event comprises at least one of a physiological event, a symptom event, and a behaviorally logged event reported by the human subject or by a healthcare professional;
[0062] retrospectively label, in the physiological signal timeline, at least one time window associated with the target event, and extract from the labelled time window a feature set derived from at least the first gastrointestinal acoustic signal; and
[0063] train a predictive model on the extracted feature set and operate the predictive model in real time to output a probability estimate relating to a future occurrence of the target event, based on the first gastrointestinal acoustic signal and optionally based on at least one of the photoplethysmographic signal, the movement signal, and a dynamically reported contextual parameter.
[0064] In particular, the device is configured to be worn at an abdominal region of the human, wherein the first acoustic sensor is oriented towards a surface of the abdomen and configured to capture a first gastrointestinal acoustic signal which is a gastrointestinal acoustic signal. The internal organ of interest may be one or more organs of the gastrointestinal system, for example, the stomach, the small intestine, the large intestine, and / or the liver.
[0065] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteThe first acoustic sensor may be a microphone to directly capture the sound waves of the bowel sounds. Alternatively, the first acoustic sensor may indirectly capture acoustic signals originating from the peristaltic movement of the digestive system, using another modality, e. g. electrodes to capture electric signals generated in the region of the digestive system.
[0066] The second acoustic sensor may be a microphone.
[0067] In particular, the device includes a microcontroller (MCU), wherein the microcontroller (MCU) is connected to measuring outputs of the at least first and second acoustic sensors, of the photoplethysmographic sensor, and of the at least one inertial measurement unit.
[0068] In a preferred embodiment, the microcontroller (MCU) is configured to perform a preprocessing step comprising removing environmental noise from the first gastrointestinal acoustic signal by subtracting from it a noise signal calculated from the environmental acoustic signal.
[0069] In a preferred embodiment, the microcontroller (MCU) is configured to perform a preprocessing step comprising reducing a quantity of bytes in each of the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and / or the movement signal.
[0070] In a preferred embodiment, the device further comprises at least a third acoustic sensor, wherein a sensing area of the third acoustic sensor is oriented towards the surface of the abdomen when the device is being worn and is configured to capture a second gastrointestinal acoustic signal, wherein the microcontroller (MCU) is connected to a measuring output of the third acoustic sensor and is configured to:
[0071] optionally, remove an environmental noise from the second gastrointestinal acoustic signal by subtracting from it a noise signal calculated from the environmental acoustic signal; and
[0072] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltelocalize an origin within the abdomen of the first and second gastrointestinal acoustic signals based on a phase difference measured between the first and second gastrointestinal acoustic signals.
[0073] Again, the third acoustic sensor may be a microphone to directly capture the sound waves of the bowel sounds. Alternatively, the third acoustic sensor may indirectly capture acoustic signals originating from the peristaltic movement of the digestive system, using another modality, e. g. electrodes to capture electric signals generated in the region of the digestive system.
[0074] In a preferred embodiment, the microcontroller is configured to selectively operate the device in a first operation mode and in a second operation mode, wherein:
[0075] in the first operation mode, the photoplethysmographic signal is not measured and / or processed, and the first gastrointestinal acoustic signal is not measured and / or processed, whereas the movement signal is measured and processed;
[0076] in the second operation mode, the photoplethysmographic signal is measured and processed,
[0077] wherein the device is switched from the first operation mode to the second operation mode if the movement signal indicates movement of the device above a first movement threshold.
[0078] Power consumption in the first operation mode is minimized. Nevertheless, based on the movement signal, it may be detected whether the device is potentially worn. In order to ascertain whether the device is placed on the skin of the human to be monitored, the photoplethysmographic signal obtained in the second operation mode is analyzed.
[0079] Therefore operation of the photoplethysmographic sensor and / or processing of the respective signal is activated only if movement is detected in the first operation mode (low-power monitoring).
[0080] Reduced power consumption increases the battery life of the device and reduces the frequency of recharging, thus ensuring reliable monitoring during long periods and
[0081] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteimproved user comfort. Due to reduced battery capacity requirements, the footprint of the device may be minimized.
[0082] Preferably, the movement signal is obtained by an accelerometer, gyrometer, magnetometer or a combination thereof.
[0083] Preferably, in the second operation mode, contact of the device with the skin of the human is identified based on the photoplethysmographic signal and the device is switched to a third operation mode if skin contact is present, and in the third operation mode, the first gastrointestinal acoustic signal is measured and processed.
[0084] Ascertaining skin contact prior to activating audio capture further reduces power consumption. In addition, useless audio signals relating to time periods when the device is not worn are not stored, thus the required storage space is reduced as well.
[0085] Correspondingly, the device recording time before the internal storage is exhausted is increased. Both the reduced power consumption as well as the reduced storage requirements allow for designing the wearable device more compact and more lightweight. This increases the comfort for the monitored human.
[0086] Using a photoplethysmographic signal allows for reliable assessment of the skin contact between the wearable device and the human and its acquisition requires less energy than alternative modalities.
[0087] Advantageously, the third operation mode comprises a first submode, in which the gastrointestinal acoustic signal is measured and processed, and a second submode, in which the gastrointestinal signal is not measured and / or processed, wherein
[0088] the first submode is activated if the movement signal indicates movement of the device below a second movement threshold, and
[0089] the second submode is activated if the movement signal indicates movement of the device above the second movement threshold.
[0090] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteThis ensures that audio signals are captured and / or processed only in periods when there is a valid chance of obtaining good quality information. Audio capture in time periods with excessive movements are not acquired and / or processed because it is improbable to obtain valid acoustic information in these periods, e. g. due to friction noise, variable skin contact or excessive external noise. This measure further reduces power consumption and storage requirements and improves overall quality of the acquired acoustic information.
[0091] The measures discussed above, in relation to switching the device between different operation modes, based on signals of a movement sensor and a PPG sensor, are particularly advantageous in the context of wearable sensors for long-term monitoring purposes. As such, they are not restricted to real monitoring applications based on acoustic signals. The PPG sensor allows for reliably detecting skin contact, whereas based on the signals of the movement sensor, phases in which acquisition, recordal and processing of a main signal (e. g. a gastrointestinal acoustic signal) yields valuable information may be identified. The measures reduce in particular total power consumption, which greatly improves operation time for a given battery capacity.
[0092] Therefore, less frequent recharging is required and / or the size and weight of a battery may be reduced. Both measures improve user comfort and ensures reliable and comprehensive acquisition of data.
[0093] It is to be noted that the above approach to selectively operate the device in the first operation mode and in the second operation mode, including the described preferable implementations are applicable to wearables that are used to capture acoustic signals originating from other organs than those of the gastrointestinal system. In these cases, the wearable device contacts a body surface in an area of interest of the body which generates sounds, e. g. in the vicinity of an internal organ (e. g. lung, heart, etc.) that generates sound, and the first acoustic sensor is oriented towards a surface of the area of interest, whereas the second acoustic sensor is oriented in a direction opposite the surface. The first acoustic sensor captures physiological acoustic signals originating from the internal organ to be monitored.
[0094] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteIn a preferred embodiment, the microcontroller (MCU) is configured to analyze the environmental acoustic signal to detect speech and to obfuscate speech in the first gastrointestinal acoustic signal if speech is detected.
[0095] Acoustic monitoring devices, including wearable stethoscopes, present considerable potential to identify and monitor acoustic gastrointestinal signals in a longitudinal manner. However, given such devices are listening during large time periods, there is considerable concern regarding the privacy of user's conversations while using the device. By obfuscating speech, such concerns may be addressed.
[0096] Preferably, speech obfuscation includes applying a band pass filter to the first gastrointestinal acoustic signal.
[0097] Further means of obfuscation are available that may be used instead of or in addition to the above, including pure low or high pass filtering. However, especially in the context of monitoring bowel sounds, these means exhibit drawbacks with respect to the quality of bowel sound information remaining in the processed signal and / or with respect to reliable obfuscation of the speech information.
[0098] It is to be noted that the above approach for speech obfuscation is applicable to wearables that are used to capture acoustic signals originating from other organs than those of the gastrointestinal system. In these cases, the wearable device contacts a body surface in an area of interest of the body which generates sounds, e. g. in the vicinity of an internal organ (e. g. lung, heart, etc.) that generates sound, and the first acoustic sensor is oriented towards a surface of the area of interest, whereas the second acoustic sensor is oriented in a direction opposite the surface. The first acoustic sensor captures physiological acoustic signals originating from the internal organ to be monitored.
[0099] Alternatively or in addition, the computer server or the wearable device microcontroller (MCU) may be configured to:
[0100] calculate a heart rate (HR) signal and a heart rate variability (HRV) signal from the photoplethysmographic (PPG) signal; and
[0101] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltedetermine a quality of contact between the wearable device and a skin of the human subject based on a quality of the heart rate (HR) signal and the heart rate variability (H V) signal.
[0102] Alternatively or in addition, the computer server or the wearable device microcontroller (MCU) is configured to
[0103] calculate at least one gastrointestinal factor signal from the heart rate (HR) signal and the heart rate variability (HRV) signal wherein the at least one gastrointestinal factor signal comprises a vagal function signal, a stress signal, an arousal signal, an exercise signal; and
[0104] determine a correlation between the at least one gastrointestinal factor signal and the first gastrointestinal acoustic signal.
[0105] Preferably, the at least one inertial measurement unit (IMU) is configured to measure a 3-axis accelerometer signal, a gyrometer signal, and a magnetometer signal, and the computer server or the wearable device microcontroller (MCU) is configured to calculate a movement of the wearable device in three-dimensional space by calculating a roll signal, a pitch signal, and a yaw signal, and further calculate an orientation in three-dimensional space of the wearable device.
[0106] The computer server or the wearable device microcontroller (MCU) may be configured to:
[0107] interpolate a movement data of the human from the movement signal of the wearable device, wherein the movement data comprises a number of daily steps, a body posture of the human subject, and / or a sleep stage signal; and
[0108] determine a correlation between the movement data and the first gastrointestinal acoustic signal.
[0109] The computer server or the wearable device microcontroller (MCU) may be further configured to calculate a plurality of time-domain features based on a temporal analysis of the first gastrointestinal acoustic signal.
[0110] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteFurther, the computer server or the wearable device microcontroller (MCU) may be configured to calculate a plurality of machine learning features from at least the first gastrointestinal acoustic signal.
[0111] The machine learning features calculated from the first gastrointestinal acoustic signal may be combined with contextual parameters in the analytical model in a number of ways. In a first approach, the static contextual parameters — such as age, sex, body mass index, and known diagnoses — are provided to the machine learning model as additional fixed input features alongside the acoustic-derived feature vector, allowing the model to learn condition-specific and demographic-specific weightings of the acoustic features. In a second approach, the dynamic contextual parameters — such as dietary intake events, medication intake events, defecation event logs, and subjective symptom reports — are encoded as time-stamped input features and concatenated with the acoustic feature vector for the corresponding time window, enabling the model to learn the interaction between situational context and gastrointestinal acoustic characteristics. In a third approach, the contextual parameters are used not as direct model inputs but as conditioning variables that select or weight between a set of context-specific sub-models — for example, a post-prandial sub-model activated in the time window following a logged dietary intake event and a fasting sub-model active during periods without recent food intake — each trained on acoustic features extracted under the corresponding contextual condition. In all of these approaches, the incorporation of contextual parameters into the machine learning framework enhances the model's capability to capture inter-individual variability and temporal dynamics of the gastrointestinal system, thereby improving the granularity, accuracy, and clinical relevance of the generated monitoring outputs.
[0112] In a further aspect of the invention, a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of any of the methods according to embodiments of the invention is provided.
[0113] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteIn a further aspect of the invention, a computer-readable storage medium comprising instructions which, when executed by a computer, cause the computer to carry out the steps of any one of the methods described herein according to various embodiments of the invention is provided.
[0114] Instead of diagnosing the target condition by using a microcontroller (MCU) of the device, the target condition may be diagnosed by using a computer server.
[0115] Similarly, instead of diagnosing the target condition by using a microcontroller (MCU) of the device, the target condition is diagnosed by using a computer server configured to receive at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal from the device, wherein these signals are optionally pre-processed by the device microcontroller (MCU) before they are received by the computer server, and wherein the computer server is configured to diagnose the target condition based on at least the received first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal.
[0116] Further advantageous embodiments and combinations of features of the invention result from the following exemplary embodiments and the set of patent claims as a whole.
[0117] Brief description of the drawings
[0118] Figure 1 illustrates a time-frequency diagram showing both duration and frequency content of recorded abdominal sounds (bowel sounds).
[0119] Figure 2 illustrates an architecture of a wearable device for gastrointestinal monitoring according to an exemplary embodiment of the invention.
[0120] Figure 3 illustrates an architecture of a gastrointestinal monitoring system according to an exemplary embodiment of the invention.
[0121] Figure 4 illustrates a state model for operating the wearable device.
[0122] In general, identical parts are marked with the same reference symbols in the figures.
[0123] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteThese and / or other aspects, features, and / or advantages will become apparent and more readily appreciated from the following description of various example embodiments, taken in conjunction with the accompanying drawings. Thicknesses of layers / elements, and sizes of components / elements, are not necessarily drawn to scale or in actual proportion to one another, but rather are shown as example representations. Each embodiment herein may be used in combination with any other embodiment(s) described herein.
[0124] Ways to carry out the invention
[0125] The following detailed structural and / or functional description(s) is / are provided as examples only, and various alterations and modifications may be made. The example embodiments herein do not limit the disclosure and should be understood to include all changes, equivalents, and replacements within ideas and the technical scope herein.
[0126] Hereinafter, certain examples will be described in detail with reference to the accompanying drawings.
[0127] Hardware Components
[0128] The herein described invention comprises the following core unique components:
[0129] 1. A microphone system that is configured to record abdominal sounds or bowel sounds 2. A Photoplethysmography (PPG) sensor configured to collect data for heart rate and heart rate variability
[0130] 3. One or more inertial measurement units (IMU), individually or collectively comprising accelerometers, gyrometers, magnetometers, wherein the one or more inertial measurement units are individually or collectively configured to collect movement and orientation data
[0131] According to an example embodiment, a wearable device for real-time human gastrointestinal monitoring based on acoustic signals is shown in schematic form in Fig. 2.
[0132] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteThe wearable device may comprise a plurality of audio capture devices, such as microphones. At least a first microphone 79 (Microphone 1 (Body)) may be a bodyfacing microphone when the device is worn and may be configured to record abdominal sounds. The first microphone 79 is a MEMS microphone that is arranged behind a cone used for amplification and isolation. It is sealed with a stethoscope membrane made from a plastic material, which is in direct skin-contact to conduct sounds into the chamber inside the cone. The body microphone is further backed by a foam backer for noise suppression.
[0133] At least a second microphone 80 (Microphone 2 (exterior)) may be an externally-facing (away or outward from the body) configured to record sounds from an ambient environment surrounding a person wearing the device.
[0134] The second microphone is a MEMS microphone. It is placed directly underneath the top surface of the device housing, facing the environment. It is exposed to the environment through a pin-hole and protected from dust ingress by a dust cover.
[0135] The wearable device may in addition comprise one or more inertial measurement units (IMU). In the example shown in Fig. 2, the device includes a first IMU 74 (including a gyro and acceleration sensor) and a second IMU 75 (including a gyro and a magnetometer). The wearable device may in addition comprise a photoplethysmographic (PPG) sensor 77.
[0136] The wearable device may comprise a microcontroller 71 (MCU) configured to carry out methods according to embodiments of the invention. The microcontroller 71 may be connected by an I2S bus 78 to measuring outputs of the plurality of microphones. It is further connected to the at least one inertial measurement unit, and the at least one photoplethysmographic (PPG) sensor, for example via an analog front end 76 (Tl 4404) and I2C Bus 73 as shown in Fig. 2.
[0137] The microcontroller 71 (MCU) may have Bluetooth capability, i. e. it is capable of exchanging data over a Bluetooth connection 84, and include a built-in Bluetooth
[0138] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteantenna as well as a real-time controller (RTC). The microcontroller 71 may read instructions from a micro-SD card 72, for example. The microcontroller 71 may read instructions received wirelessly, for example via a Wi-Fi antenna.
[0139] The device may comprise a power management unit 81 and rechargeable battery 82, for example, and may be recharged for example via a USB-C or micro-USB connection 83 or other suitable connection.
[0140] Additional algorithm data inputs (optional):
[0141] Patient Reported outcomes
[0142] Symptoms:
[0143] o Bristol scale
[0144] o Pain level
[0145] o Nausea
[0146] o Bloating etc.
[0147] Events logging : food intake, medication intake etc.
[0148] Food images extraction to obtain information such as caloric load, ingredients, FODMAPs etc. through analysis of the images, wherein the analysis may be Al based
[0149] Architecture
[0150] According to an example embodiment, a gastrointestinal monitoring system architecture for achieving a functionality according to an embodiment of the invention is shown in schematic form in Fig. 3 and described below:
[0151] a) A wearable platform (body-worn device 10 in Fig. 3) implementing the different sensor types, combined through use of a microcontroller unit with Bluetooth functionality
[0152] b) The performance of pre-processing analysis on the device
[0153] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltec) The transfer of data to a mobile phone (user mobile phone 20 in Fig. 3) using a Bluetooth connection 61
[0154] d) The mobile phone 20 may be used by the patient 5 to provide input regarding events such as food intake, toileting, medications and other relevant contextual data through a suitable mobile application (app Ul) running on the mobile phone 20 e) The transfer of data to a cloud backend 31 providing, for example, alerting functionality
[0155] f) The analysis of the data by the system in a manner that combines all relevant variables
[0156] As shown according to the example embodiment in Fig. 3, the system architecture may additionally comprise a backend server 30, or computer server, on premises, for performing computational resource intensive calculations, and may be connected via for example, an internet connection 64, to the cloud backend 31.
[0157] The backend server 30 may be configured to receive data wirelessly from both the wearable device 10 and the user mobile phone 20, e.g. over a WiFi or LTE connection 62. The backend server 30 may additionally be connected to a connected hospital information system 40 by a local network connection 63 and configured to receive data from such a system, for example data entered by a health care professional such as a doctor 6.
[0158] The use of a computer server in addition to a microcontroller provided on the wearable device 10 in an exemplary embodiment of the gastrointestinal monitoring system provides advantages of offloading computationally intensive calculations from the microcontroller to the computer server (backend server 30), and may allow more data to be stored and taken into account.
[0159] The system may provide an alerting functionality, in which alerts are provided in the patient application or through connection of the backend server 30 to a hospital monitoring app 50, e. g. over a local network connection 65. For example, according to an
[0160] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteexample embodiment of the invention, if analyses performed by the backend server reveal an emergency, a signal may be transmitted to the hospital monitoring app alerting a bedside health care professional 7 to the emergency. The hospital monitoring app 50 may further be adapted to exchange data directly with the cloud backend 31 over an internet connection 66.
[0161] According to an alternative embodiment, a computer server may be used for storage of reports in a manner that allows human-readable summaries of collected data.
[0162] Functionality
[0163] The key differentiator of this invention is that the combination of sensor types provides a higher overall specificity and sensitivity in the analysis of the state of the gastrointestinal (Gl) tract,.
[0164] When analysing sound data, a wide range of features may be extracted, particularly for tasks like bowel sound recognition and disease prediction. These features can be broadly categorized into time-domain features, frequency-domain features, time-frequency features, and statistical / machine learning-derived features.
[0165] Audio Pre-Processing and Sound Localization
[0166] According to an example embodiment, audio data, such as recorded acoustic signals comprising gastrointestinal acoustic signals and environmental acoustic signals may be pre-processed upon the device to achieve outcomes important to the overall system functionality:
[0167] 1. Reduction of environmental noise impact on signal quality
[0168] 2. Localization of abdominal sound origins
[0169] 3. Identification of types of abdominal sounds
[0170] 4. Feature extraction of abdominal sounds
[0171] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteHardware Implementation
[0172] Example implementation: two microphones may be implemented on the device. One external facing microphone may be configured to collect environmental sounds, the other microphone may comprise a body-facing microphone and may be configured to collect abdominal sounds.
[0173] Optionally, the wearable device according to example embodiments may comprise additional body-facing and external microphones with directional characteristics, for example providing additional functionality such as abdominal sound localization.
[0174] Processing of data
[0175] Noise cancelling: The environmental audio data may be used, optionally in addition to deep learning techniques, to remove any environmental noise that "bleeds" into the body facing microphone signal, improving the overall signal to noise ratio for abdominal sounds or bowel sounds.
[0176] According to an example embodiment of the invention, several body-facing microphones, for example 2-6 body-facing microphones, may be used in addition to the at least one external facing microphone.
[0177] An additional functionality of the additional body-facing microphones may be, for example, direction finding, wherein a time delay of arrival may be calculated using phase differences in received audio signals between different body-facing microphones. Using the phase differences, the origin location of sounds can be estimated in the body.
[0178] Relevance to overall functionality
[0179] Noise reduction according to example embodiments of the invention helps eliminate unwanted environmental noise in the abdominal sound signal that could
[0180] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteotherwise impact overall model performance. This improves both the identification, classification and counting of abdominal sounds.
[0181] Direction finding according to example embodiments of the invention improves overall model performance by identifying the origin location in the digestive system. This could be used to isolate the origin location of pathognomonic sounds, improving overall model performance and model explainability, as abdominal sound locations may be clinically relevant.
[0182] Time-Domain Features (Raw Sound Signal)
[0183] According to example embodiments of the invention, one or more of the following time-domain features may be directly derived from the waveform and help in capturing temporal patterns, for example using the on-device microcontroller (MCU) or using a computer server configured for that purpose:
[0184] Feature Description
[0185] Mean Amplitude Average signal value over time.
[0186] Root Mean Square (RMS) Energy Square root of mean squared amplitude — signal energy measure.
[0187] Zero Crossing Rate (ZCR) Rate at which the signal changes sign (positive «-» negative).
[0188] Peak Amplitude Maximum absolute value of the signal. Crest Factor Ratio of peak amplitude to RMS value (indicates transients).
[0189] Skewness Asymmetry of the amplitude distribution curve.
[0190] Kurtosis Sharpness (tailedness) of the amplitude distribution.
[0191] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteSignal Energy Total sum of squared amplitudes.
[0192] Short-Time Energy Localized energy computed over sliding windows.
[0193] Variance Spread of the amplitude values around the mean.
[0194] Autocorrelation Features Measures self-similarity over time lags (periodicity detection).
[0195] Entropy of Energy Measure of energy distribution irregularity across frames.
[0196] Attack Time / Decay Time Time taken for the signal to reach peak amplitude or decay.
[0197] Slope Features Derivatives of the signal (rise / fa 11 rate of the waveform).
[0198] Teager Energy Operator (TEO) Captures instantaneous amplitude and frequency modulations.
[0199] Waveform Shape Features Descriptors based on waveform patterns (rise time, fall time, symmetry, flatness).
[0200] Frequency-Domain Features (Fourier Transform-Based)
[0201] One or more of the following frequency-domain features may be extracted using techniques like Fourier Transform (FFT) from the acoustic signals measured by the wearable device, for example using the on-device microcontroller (MCU) or using a computer server configured for that purpose, according to embodiments of the invention:
[0202] Feature Description
[0203] Spectral Centroid Center of mass of the spectrum (brightness).
[0204] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteSpectral Bandwidth Spread of spectrum around the centroid (frequency spread).
[0205] Spectral Rolloff Frequency below which a given % (e.g.,
[0206] 85%) of total spectral energy is contained. Spectral Flatness Degree to which the spectrum is flat vs. peaky (tone vs noise).
[0207] Spectral Flux Rate of change of the spectrum between frames.
[0208] Spectral Skewness Asymmetry of the spectrum's shape.
[0209] Spectral Kurtosis Peakedness (sharpness) of the frequency spectrum.
[0210] Spectral Slope Trendline slope of the spectrum — rising or falling overall trend.
[0211] Spectral Complexity Number of significant spectral components (frequency richness).
[0212] Harmonic-to-Noise Ratio (HNR) Ratio between harmonic (periodic) and noise (random) parts.
[0213] Fundamental Frequency (FO) Lowest frequency of a periodic waveform (pitch).
[0214] Formants (Fl, F2, F3, ...) Peaks of the resonant frequencies, often in speech and vocal sounds.
[0215] 29703-WO1 PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteLinear Predictive Coding (LPC) Efficient representation based on predicting Coefficients
[0216] future samples.
[0217] Chroma Features Energy distribution across musical pitch classes.
[0218] Spectrogram Statistics Statistical measures (mean, variance, entropy) over spectrogram frames.
[0219] Energy in Different Frequency Bands Split energy into frequency bands (e.g., low, mid, high).
[0220] Modulation Spectral Features Capture modulations of the signal envelope within bands.
[0221] Tonnetz Features Chord and tonal relationships between notes.
[0222] Frequency Peak Counts Number of significant peaks in the frequency spectrum.
[0223] Sub-band Centroids Centroid of smaller sub-bands rather than the whole spectrum.
[0224] Time-Frequency Features (Spectrogram-Based)
[0225] Figure 1 illustrates a time-frequency diagram showing both duration and frequency content of recorded abdominal sounds (bowel sounds). The x-axis denotes time, the y-axis denotes frequency. The amplitude of the signal at a given time and with a given frequency is encoded by intensity (brightness). A number of bowel sounds are recorded in the
[0226] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltedisplayed interval. Characteristic properties are e. g. a sound's frequency range 1, a sound's duration 2 and the interval 3 between subsequent sounds identified.
[0227] According to example embodiments of the invention, one or more of the following features may be extracted from short-time Fourier transforms (STFT), wavelet transforms, or constant-Q transforms from the acoustic signals measured by the wearable device, for example using an on-device microcontroller (MCU) or using a computer server configured forthat purpose, according to embodiments of the invention:
[0228] • Mel-Spectrogram Features - A log-scaled spectrogram for detailed frequency analysis.
[0229] • Wavelet Transform Coefficients - Capture time-varying frequency components.
[0230] • Temporal Evolution of Spectral Features - How spectral characteristics change over time.
[0231] Statistical Features from Signal Processing
[0232] According to example embodiments of the invention, one or more of the following features may be derived from time, frequency, and spectrogram features to enhance machine learning modelling, for example using an on-device microcontroller (MCU) or using a computer server configured for that purpose:
[0233] • Mean, Standard Deviation, Skewness, Kurtosis - Statistical distribution of features.
[0234] • Autoregressive Coefficients - Used to model sound sequences.
[0235] • Entropy of Energy Distribution - Measures disorder in the frequency domain.
[0236] • Fractal Dimension and Recurrence Features - Capture complexity in temporal dynamics.
[0237] Machine Learning-Derived Features
[0238] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte• Deep Learning Embeddings - Feature representations extracted from CNNs, LSTMs, or transformers.
[0239] • Latent Representations from Autoencoders - Learned compact representations of bowel sounds.
[0240] • Graph-Based Features - Analysis of connectivity patterns in sound signals.
[0241] One or more of the above features may be derived from the acoustic signals measured by the wearable device, for example using an on-device microcontroller (MCU) or using a computer server configured for that purpose.
[0242] A multimodal approach may be used to combine one or more of these features with contextual data (e.g., patient symptoms, physiological metrics) to enhance predictive accuracy.
[0243] In addition, deep learning models (Transformers or CNNs) may also be used directly to predict certain diseases, such as target gastrointestinal conditions. In other words, according to embodiments of the invention, every single abdominal sound may be used as an entry to the model, or to give a micro view on the whole abdominal sound signal.
[0244] Photoplethysmogram
[0245] According to embodiments of the invention, a photoplethysmogram (PPG) may be used to measure blood oxygenation in the skin at the site of the wearable device. This is achieved by shining a light into the skin and measuring the absorption through the intensity of reflected light using a photodiode.
[0246] Hardware implementation
[0247] The PPG may, according to an example embodiment, be implemented through use of off-the-shelf sensors.
[0248] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteAccording to an example embodiment, the exact implementation in hardware (mechanical housing and electronics) may be optimized to ensure close contact between the LED and photodiode components with the skin.
[0249] Firmware configuration of the PPG may be further optimized for implementation of the PPG on stomach skin (as opposed to typical values used for wrist or finger application).
[0250] Processing of data
[0251] PPG data may be retrieved in raw format and then denoised and detrended. Small variations in signal intensity may be used to calculate values for heart rate (HR) and heart rate variability (HRV), and also to calculate oxygen saturation levels (SpO2).
[0252] Signal quality metrics may also be calculated from collected PPG signals according to embodiments of the invention.
[0253] Relevance to overall functionality
[0254] The extraction of HR and HRV may provide the following additional functionality according to example embodiments of the invention:
[0255] 1. Extraction of signal quality metrics allows determination of contact between the device and skin. This may be used to detect periods where the device is not being worn but is still recording audio. This facilitates rejection of potentially irrelevant data.
[0256] 2. Extraction of HR and HRV give indications of vagal function, stress, arousal and exercise.
[0257] This allows the drawing of correlations between environmental and physiological factors on gut sounds. For example, resting heartrate and heartrate variability are known predictive factors for flareups in cases of Irritable Bowel Disorders ("Physiological Data Collected From Wearable Devices Identify and Predict Inflammatory Bowel Disease Flares." Hirten, Robert P. et a\.Gastroenterology, Volume 168, Issue 5, 939 - 951 ,e5). The outcome is improved predictive value and improved explainability of model outputs.
[0258] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte3. HR and HRV values may contribute to predictive value by identifying signs of physiological distress which may indicate onset of a target condition.
[0259] Accelerometer, Gyrometer, Magnetometer
[0260] According to an example embodiment of the invention, one or more inertial measurement units individually or collectively comprising at least a 3-axis Accelerometer, gyrometer and a magnetometer may be used to measure movement of the device.
[0261] Hardware implementation
[0262] These components may be implemented through off-the-shelf components selected for suitability for the application.
[0263] Processing of data
[0264] All three sensors (accelerometer, gyrometer, magnetometer) may be combined to calculate the movement of the device in three-dimensional space (roll, pitch and yaw). This may also be used to calculate the orientation of the device which may be represented in several mathematical formats according to embodiments of the invention (such as quaternion).
[0265] Relevance to overall functionality
[0266] 1. Overall movement of a user: There is evidence to suggest that movement of a patient / user is relevant to the behaviour of the gut and has specific relevance to predicting recovery of a patient in conditions such as bowel obstruction or ileus. Greater movement of the body encourages movement of the bowel, which may support recove ry / in creased motility. Consideration of this variable may further improve predictive value of the overall algorithms according to exemplary embodiments of the invention.
[0267] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte2. Detection of steps: movement data may be used to estimate number of daily steps, which may be used as input to a model according to example embodiments of the invention.
[0268] 3. Detection of body posture: supine lying, side lying, sitting and standing have different impacts on behaviour of the bowel. Detection of these postures may be used as input to the predictive model according to example embodiments of the invention.
[0269] 4. Detection of sleep stages: the relationship between sleep stage and gut behaviour may provide further insights into specific conditions. Relationships between sleep duration and quality and the impact on gastric conditions may be of further relevance to understanding gut behaviour and may improve predictive value of algorithms according to exemplary embodiments of the invention.
[0270] State Model
[0271] Continuous recording in wearable stethoscopes stresses storage and battery, and degrades quality due to motion, friction noise, and loss of skin contact.
[0272] Therefore, in an embodiment of the invention, the wearable device is controlled according to a state machine model, which controls in particular the audio capture using:
[0273] a movement sensor (accelerometer / gyro / magnetometer) for stillness / motion classification;
[0274] the PPG sensor for skin-contact detection
[0275] Recording is OFF by default; it activates only when skin contact is present and motion is below a threshold; quality indexing determines whether to store or discard; hysteresis-based deactivation.
[0276] The state machine model may be implemented by operating the device as follows:
[0277] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte1. Standby mode: audio recording is OFF, PPG sensor is OFF, movement sensor is ON in a low-power MONITORING MODE (here and in the following, OFF relates to a turned off / inoperative sensor and / or no processing of the signals of the respective sensor). 2. If the movement sensor detects a movement over a THRESHOLD value, the device is switched into Contact Detection mode by activating the PPG sensor and analytics (switched to ON state). The resulting signals of the PPG sensor are processed to detect skin contact of the device by an algorithm. This monitoring is active until skin contact is detected or until a MONITORING PERIOD expires, whatever happens first.
[0278] 3. If skin contact is detected, the device is switched into Movement Monitoring mode, i. e.
[0279] signals of the movement sensor are processed by a movement classification algorithm in order to determine if the device is sufficiently still for audio capture (i. e. the movement is below an upper threshold value).
[0280] 4. As soon as the device is sufficiently still, the device is switched into Active Mode and audio recording is turned ON.
[0281] In the Active Mode:
[0282] 5. The recorded audio signal is quality indexed to determine if the respective data should be saved or discarded.
[0283] 6. Movement is periodically re-assessed based on the signals of the movement sensor.
[0284] Audio recording is switched OFF if the movement exceeds the upper threshold value during a certain period (movement hysteresis period), i. e. the device is switched back to Movement Monitoring mode.
[0285] 7. PPG is ACTIVE and periodically re-assesses skin contact. If skin contact is lost during a certain period (skin contact hysteresis period), audio recording is switched OFF and the device goes back into Contact Detection mode. After expiry of a predetermined monitoring period, the device is switched back to Standby mode if no skin contact is detected. If skin contact is detected, the device is switched to Movement Monitoring mode.
[0286] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteThe following table outlines how this state-machine influences sensor activation in common use scenarios:
[0287] Scenario Inertial PPG State Audio State Justification Measurement
[0288] State
[0289] Device not worn, • ON • OFF • OFF Device is stationary • Low-power likely not in monitoring use. Save mode power /
[0290] storage Movement • ON Normal • ON, • OFF Device may detected, skin operation • Skin contact be being contact state assessment donned, unknown check skin activated
[0291] contact before audio recording to save power / storage Device worn, • ON Normal • ON, • ON, Device is in stationary or operation • skin contact • Audio use, minimal • Movement assessment signal conditions movement for data classification active, processing
[0292] active recording • Hear activet rate
[0293] and analysis metric
[0294] analysis
[0295] active
[0296] Device worn, Audio is • ON Normal • ON, • OFF or ON significant unlikely to operation with audio • skin contact
[0297] movement be useable. quality • Movement assessment
[0298] detected Save power / indexed classification active,
[0299] storage. before active • Heart rate
[0300] processing
[0301] metric
[0302] analysis
[0303] activated
[0304] Skin contact lost Device has • ON Normal • ON, • OFF
[0305] been doffed, operation • Skin contact
[0306] cease assessment
[0307] recording to active
[0308]
[0309] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltesave power / storage.
[0310]
[0311] A state model for operating the wearable device in a power-saving manner is illustrated in Fig.
[0312] 4. The input data is obtained from the microphones 79, 80, the PPG sensor 77 and the movement sensors IMUs 74, 75. If the wearable device is not completely deactivated, at least one of the IMUs 74, 75 continuously records movement data in the low power monitoring mode (state 105). Binary data is output, timestamp normalized (step 107) and fed to an N second buffer 108. It is then checked whether PPG is switched on (check 111). In the initial mode, when PPG is not switched on, it is checked whether movement has been detected (check 112). If this is the case, PPG recording is switched on (state 103). PPG data is output, timestamp normalized (step 106) and fed to the buffer 108. If PPG recording is active, it is checked whether skin contact is detected (check 113). If not, PPG and audio recording are switched off (states 101, 104) if no movement is detected (check 112) or PPG recording is continued (state 103) if movement is detected.
[0313] If skin contact is detected (check 113), the IMU data is classified to obtain a movement classification (classification 115). If the movement is below a threshold and audio recording is not yet active (check 116), audio recording is activated (state 102). This leads to feeding the recorded audio data to the buffer 108. If the movement is above the threshold, audio recording is deactivated (state 101).
[0314] Speech obfuscation
[0315] In order to preserve the privacy of the user and other people, speech is detected in the acquired environmental audio signal and the audio data is processed to obfuscate the speech information in the audio data acquired by the two microphones. In the context of the described embodiment, speech obfuscation is achieved by use of two microphones on the wearable stethoscope:
[0316] 1. Microphone for monitoring environmental sounds (ENV Mic)
[0317] 29703-WO1 PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte2. Microphone for recording gastrointestinal sounds (Body Mic)
[0318] Audio data acquired by the ENV mic is processed by an algorithm to detect speech. If speech is detected, a band pass filter blocking a frequency band that is important for speech recognition is applied on the body and environmental channels. Speech detection is regularly repeated and the band pass filter is switched off if no speech is detected with in a predetermined time period.
[0319] In particular, this specific band pass method preserves spectral features needed to identify bowel sounds.
[0320] Operation and Use Cases
[0321] Gastroenterological monitoring
[0322] An example use case of the gastrointestinal monitoring system and wearable device according to embodiments is provided in the following. This example use case is provided as an example only. The device may be worn by the patient or user continuously minus periods for charging or showering, for example. The data may be continuously streamed to a Bluetooth-connected smartphone. The system may provide analysis outputs to the user based on the unique combination of sensor types of the wearable device according to embodiments of the invention in order to provide insights into, for example, Gl tract behaviour.
[0323] This invention may be applied in a range of areas:
[0324] 1. Achieving a digital therapeutic effect by providing insights into correlations between specific events and bowel behaviour
[0325] 2. Predict defecation events
[0326] 3. Measure gastric emptying
[0327] 4. Monitoring for colic in humans and animals
[0328] 5. Monitoring of neonatal bowel behaviour
[0329] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte6. Influence of sports on bowel behaviour
[0330] 7. Monitoring of nutrition and food intake
[0331] 8. Recommendations of when to eat (how much)
[0332] 9. Constipation + diarrhea prevention
[0333] 10. Influence of sleeping on digestion
[0334] 11. Influence of activity on digestion
[0335] 12. Diet tracking and weight loss
[0336] 13. Monitoring of intolerances and IBS
[0337] 14. Measurement of reaction to medications
[0338] 15. Detection of colorectal cancer
[0339] 16. Detection of neurologic diseases
[0340] 17. Stomach tumour detection
[0341] Integration of Contextual and Patient- Reported Data with Sensor Signals
[0342] Architecture of the Contextual Data Layer
[0343] The physiological signals acquired by the wearable device 10 — namely the first gastrointestinal acoustic signal from the body-facing microphone 79, the photoplethysmographic (PPG) signal from the PPG sensor 77, and the movement signal from the inertial measurement units 74, 75 — constitute the primary, continuously acquired data streams. These streams are time-stamped and stored in a common data buffer on the computer server 30, where they form the physiological signal timeline of the individual subject 5.
[0344] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteAlongside this continuous physiological timeline, the system maintains a contextual data layer, which comprises two broad categories: (i) static parameters, which are entered once or updated infrequently and describe the stable characteristics of the individual; and (ii) dynamic parameters, which are reported in real time or near-real time by the user 5 via the mobile application running on the mobile phone 20, entered by the healthcare professional 6 via the hospital information system 40 or retrieved automatically from other sources. Both categories are stored on the computer server 30 and made available to the analytical model as additional input channels.
[0345] The computer server 30 is configured to associate each dynamic contextual event with a precise timestamp and to map it onto the physiological signal timeline, thereby creating a unified, temporally aligned multimodal record. All analytical operations described in the following subsections are performed on this unified record.
[0346] Static Demographic and Clinical Parameters & environmental data
[0347] Static contextual parameters include, but are not limited to:
[0348] • Demographic parameters: age, biological sex, height, weight, and body mass index (BMI), entered by the user 5 or a healthcare professional 6 at the time of device setup and updated as needed;
[0349] • Clinical health status indicators: known diagnoses (e.g., Crohn's disease, ulcerative colitis, irritable bowel syndrome, diabetes mellitus, hypothyroidism, previous abdominal surgery), current chronic medications, and relevant medical history;
[0350] • Anatomical and morphological parameters: abdominal circumference, which influences acoustic transmission characteristics between internal organs and the body-facing microphone 79; alternatively the height and weight of the human is obtained and the transmission characteristics are determined by taking into account these quantities and / or the BMI derived therefrom.
[0351] Environmental parameters: time of day, time of year, weather, location data
[0352] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteThese static parameters are embedded into the analytical model on the computer server 30 as fixed covariates. Rather than applying population-derived reference ranges to the acoustic, PPG, and movement feature values, the model conditions its predictions on the individual's static profile, thereby generating personalized reference ranges and personalized anomaly detection thresholds.
[0353] Example: BMI and Acoustic Signal Normalization
[0354] A subject 5 with a high BMI presents a thicker abdominal wall, which attenuates the transmission of gastrointestinal acoustic signals from the intestinal lumen to the bodyfacing microphone 79 and shifts the dominant frequency band of detected bowel sounds downward. Without BMI-conditioned normalization, the feature extractor may systematically underestimate bowel sound event rate or energy in high-BMI subjects relative to lean subjects, introducing a confound that degrades model performance across a heterogeneous patient population.
[0355] The computer server 30 is configured to apply a BMI-dependent spectral correction to the acoustic feature vector: specifically, the spectral centroid and the energy distribution across frequency bands of the first gastrointestinal acoustic signal are adjusted by a BMI-derived correction function, which may be a learned regression model trained on a population of subjects spanning a range of BMI values and validated against ground-truth intestinal motility measures. After correction, the normalized acoustic features are comparable across subjects with differing body compositions.
[0356] Additionally, the weight and BMI of the subject are relevant to the interpretation of the movement signal from the inertial measurement units 74, 75: the accelerometric energy associated with a given physical activity is body-mass-dependent. The computer server 30 may therefore apply a mass-normalized step count estimator and a BMI-conditioned posture classification model, improving the accuracy of activity quantification and posture detection described herein.
[0357] Example: Medical History as a Prior for Anomaly Detection
[0358] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteConsider a subject 5 with a documented prior diagnosis of Crohn's disease in remission. The computer server 30 incorporates this diagnosis as a binary covariate in the predictive model. During a monitoring period, the model detects a gradual increase in bowel sound event rate, an increase in spectral energy in the mid-frequency band (200-600 Hz) of the first gastrointestinal acoustic signal, and a concurrent rise in resting HR and reduction in HRV as derived from the PPG signal. In a subject without a prior gastrointestinal diagnosis, this pattern might remain below the alert threshold. However, because the model is conditioned on the subject's Crohn's disease diagnosis, it assigns elevated prior probability to a disease flare event; the combination of acoustic and autonomic signals crosses the personalized alert threshold, and an alert is transmitted via the cloud backend 31 to the hospital monitoring application 50, notifying the healthcare professional 6.
[0359] This example illustrates that the incorporation of medical history as a static covariate does not merely improve average population-level model performance but specifically enhances sensitivity in high-risk individuals while avoiding excessive false-positive alerts in low-risk individuals — a clinically meaningful distinction.
[0360] Example: Surgical History and Baseline Calibration
[0361] A subject 5 who has undergone a right hemicolectomy (surgical removal of the right colon) presents a structurally altered gastrointestinal tract. The spatial distribution of bowel sounds detectable at the abdominal surface is correspondingly altered: sounds originating from the anastomosis site and the remaining colon may be detected at different locations and with different acoustic characteristics compared to an anatomically intact gastrointestinal tract. If the wearable device 10 further comprises at least a third microphone oriented towards the abdominal surface, the sound localization function described herein — based on phase differences between the first and the second acoustic signals from two body-facing microphones — may be adapted using the surgical history to constrain the anatomical region of interest in which sound origins are expected. This reduces localization errors and improves the interpretability of direction-finding outputs.
[0362] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteFurthermore, the pre-operative acoustic baseline is by definition unavailable for comparison in post-surgical monitoring. The computer server 30 is therefore configured to establish, for subjects with documented abdominal surgery, a new individual acoustic baseline during an initial wearing period of configurable duration (e.g., the first seven days of use), rather than comparing to a pre-operative or population reference.
[0363] Dietary Intake Data
[0364] Dietary intake events are logged by the user 5 via the mobile application running on the mobile phone 20. For each intake event, the mobile application may accept:
[0365] • the type of food or meal consumed, entered as free text, selected from a structured food database, or derived from image-based analysis (wherein the user 5 photographs the meal and an Al-based image analysis algorithm extracts estimated caloric load, macronutrient composition — carbohydrates, proteins, fats — and the content of fermentable oligosaccharides, disaccharides, monosaccharides, and polyols (FODMAPs), which are known triggers of gastrointestinal symptoms in susceptible individuals);
[0366] • the approximate quantity consumed;
[0367] • the time of consumption, automatically provided by the mobile application or manually adjusted by the user.
[0368] These structured dietary intake events are transmitted to the computer server 30 and stored with a precise timestamp in the unified multimodal record of the subject 5.
[0369] Example: Pre- / Post-Prandial Acoustic and Autonomic Response Profiling
[0370] For each logged dietary intake event at time t0, the computer server 30 is configured to extract, from the unified multimodal record, the following temporal trajectories in a configurable post-prandial analysis window [t0+ t_ onset, t0+ t_offset], where t_onset (e.g., 15 minutes, accounting for gastric emptying onset latency) and t_offset (e.g., 4 hours, covering the transit time through the small intestine) are configurable parameters:
[0371] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltethe bowel sound event rate, spectral energy, and inter-event interval distribution derived from the first gastrointestinal acoustic signal;
[0372] • the HR and HRV trajectory derived from the PPG signal, reflecting the postprandial autonomic shift (typically a transient increase in HR and reduction in HRV associated with the cephalic phase and the gastric phase of digestion);
[0373] • the movement signal, used to exclude physical activity artifacts from the postprandial acoustic and PPG analysis.
[0374] In an analogous way, pre-prandial profiles may be obtained. By accumulating such pre-and / or post-prandial response profiles across multiple dietary intake events, the computer server 30 builds an individualized food-gut response model for the subject 5. This model maps the nutritional content of a meal (caloric load, FODMAP content, fat content — a potent stimulator of the gastrocolic reflex — macronutrient composition) to the expected post-prandial acoustic response trajectory and the expected autonomic response trajectory.
[0375] Example: Circadian Bowel Habit Pattern Characterization
[0376] Over a monitoring period of several weeks, the computer server 30 accumulates the timestamps of all logged defecation events for the subject 5. By analyzing the distribution of these timestamps relative to the time of day, the first logged dietary intake of the day (breakfast), and the movement signal (wake-up time estimated from the transition from sleep to wakefulness as detected by the inertial measurement units 74, 75), the computer server 30 constructs a circadian bowel habit profile for the individual.
[0377] For example, the system may determine that a specific subject reliably defecates within 30-90 minutes of waking and within 20^45 minutes of consuming breakfast, consistent with an active gastrocolic reflex. For this subject, the pre-defecation acoustic and autonomic feature pattern is incorporated into the personalized defecation prediction model with these circadian and post-prandial temporal covariates as additional input features, improving prediction specificity. Conversely, if the circadian regularity of the defecation pattern decreases (e.g., the inter-event timing becomes more variable or the
[0378] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteevents cluster at unusual times), the computer server 30 may flag this change in bowel habit pattern as a clinically relevant finding and include it in the periodic summary report provided to the healthcare professional 6.
[0379] Medication Intake Data
[0380] Medication intake events are logged by the user 5 via the mobile application, including the name or category of the medication, the dose, and the time of intake. Alternatively, medication schedules may be entered by the healthcare professional 6 via the hospital information system 40 and stored on the computer server 30 as a scheduled medication timeline, against which the user's logged intake events are compared to assess adherence. Medication categories of particular relevance to gastrointestinal monitoring include, but are not limited to:
[0381] • Prokinetic agents (e.g., metoclopramide, domperidone, erythromycin): expected to increase gastrointestinal motility and hence bowel sound event rate;
[0382] • Opioid analgesics (e.g., morphine, oxycodone): expected to decrease gastrointestinal motility (opioid-induced constipation), reducing bowel sound event rate and shifting the acoustic spectral features toward lower energies;
[0383] • Antibiotics (e.g., rifaximin, metronidazole): expected to alter the intestinal microbiome and potentially the fermentation-driven acoustic profile over days to weeks;
[0384] • Laxatives and prokinetics (e.g., polyethylene glycol, bisacodyl, linaclotide):
[0385] expected to increase motility and stool frequency, detectable in both acoustic and defecation event log data;
[0386] • Antispasmodics (e.g., hyoscine butylbromide, mebeverine): expected to reduce bowel sound event rate and to attenuate the acoustic response to dietary intake. Example: Opioid-Induced Constipation Detection and Monitoring
[0387] A subject 5 undergoing post-operative pain management is prescribed an opioid analgesic. Opioid administration is logged by the healthcare professional 6 via the hospital
[0388] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteinformation system 40. The computer server 30 is configured to monitor, from the time of the first opioid administration, the trajectory of bowel sound event rate and spectral energy in the first gastrointestinal acoustic signal. A progressive reduction in bowel sound event rate below an individually calibrated baseline level — corrected for posture and activity using the movement signal — occurring within the expected pharmacokinetic timeframe of the opioid (e.g., within 2-6 hours of the first dose) provides objective, realtime evidence of opioid-induced reduction of gastrointestinal motility, a precursor of opioid-induced constipation.
[0389] The computer server 30 tracks this acoustic motility suppression continuously throughout the post-operative monitoring period. If the acoustic motility indices fail to recover toward baseline within a configurable timeframe (e.g., 48-72 hours post-operatively), or if the acoustic signal characteristics of ileus (absent bowel sounds for a sustained period) are detected, an alert is transmitted via the cloud backend 31 to the hospital monitoring application 50, notifying the bedside healthcare professional 7 of the risk of prolonged opioid-induced constipation or post-operative ileus, prompting earlier clinical intervention.
[0390] Conversely, when a laxative or prokinetic is administered in response to the constipation, the computer server 30 monitors the acoustic response: recovery of bowel sound event rate and spectral energy toward baseline within the expected pharmacodynamic window of the administered agent (e.g., 6-12 hours for an oral laxative) constitutes objective evidence of pharmacological efficacy, which may be documented in the clinical record via the hospital information system 40.
[0391] Example: Drug-Food Interaction Detection
[0392] A subject 5 taking a prokinetic agent for gastroparesis logs both medication intake events and dietary intake events via the mobile application. The computer server 30 is configured to analyze the interaction between the timing of medication intake relative to meal intake and the resulting post-prandial acoustic response. Specifically, for meals preceded by prokinetic intake within a configurable time window (e.g., 30 minutes before the meal), the
[0393] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltepost-prandial acoustic response — as measured by the onset timing, magnitude, and duration of increased bowel sound activity in the first gastrointestinal acoustic signal — is compared with the response to equivalent meals not preceded by prokinetic intake.
[0394] This analysis quantifies the pharmacodynamic effect of the prokinetic on gastric emptying and small intestinal transit at the individual level, providing objective evidence of drug efficacy that is personalized to the subject's individual dietary habits and meal composition. Over a treatment period, the computer server 30 tracks the evolution of the prokinetic effect over time, potentially detecting pharmacological tolerance (diminishing acoustic response to the same dose) and generating a trend report for the healthcare professional 6. The combination of medication intake logs, dietary intake logs, and continuous acoustic monitoring thus creates a comprehensive, longitudinal pharmacodynamic monitoring framework that would be impractical with conventional single-point-in-time clinical assessments.
[0395] Joint Multi-Contextual Analysis and Personalization
[0396] The full clinical value of the contextual data layer is realized when all categories of contextual data — static demographic and clinical parameters, dietary intake, symptom reports, defecation event logs, and medication intake — are analyzed jointly with the continuous physiological sensor signals in a unified analytical model. The computer server 30 is configured to maintain, for each subject 5, a continuously updated individual baseline defined by the distribution of acoustic, PPG, and movement features conditioned on the subject's static demographic profile and current health status. Deviations from this individual-specific, contextually conditioned baseline — rather than from a populationlevel mean — form the primary inputs to anomaly detection and alerting logic.
[0397] As a comprehensive worked example integrating all categories of contextual data simultaneously: consider a subject 5 with a known diagnosis of Crohn's disease (static clinical parameter), BMI of 27 (static demographic parameter), currently in a maintenance phase on a biologic therapy (medication intake), who logs a high-FODMAP meal (dietary intake) and subsequently reports increasing abdominal pain and bloating (symptom report) over the following two hours. The computer server 30 detects, in the first
[0398] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltegastrointestinal acoustic signal, a bowel sound pattern exceeding the individual's Crohn's-disease-conditioned and BMI-normalized baseline by a statistically significant margin. The PPG-derived HRV simultaneously declines below the individual's resting baseline. The movement signal indicates that the subject 5 is in a stationary seated posture (excluding motion artifact). The defecation event log shows no defecation in the preceding 36 hours (indicating constipation, cross-referenced with a Bristol Stool Scale report of type 2 from the previous morning). The medication intake log confirms that the biologic therapy dose was administered on schedule three weeks prior and is within its expected therapeutic window.
[0399] The computer server 30 integrates all of these data streams in the predictive model and outputs: (i) an elevated probability of a Crohn's disease flare, triggering an alert to the healthcare professional 6 via the hospital information system 40; (ii) a personalized dietary recommendation transmitted to the subject 5 via the mobile application on the mobile phone 20, advising avoidance of high-FODMAP foods in the near term; and (iii) an entry in the structured clinical report noting the temporal association between the high-FODMAP dietary event, the acoustic motility increase, the autonomic response, and the symptom report, which may inform the healthcare professional's decision regarding dose escalation or dietary counseling.
[0400] This example illustrates that the contextual data layer does not merely augment individual predictions but fundamentally transforms the system into a personalized, longitudinal clinical decision support tool capable of capturing the complex, multi-factorial interactions between diet, medication, autonomic state, motility, and subjective experience that characterize chronic gastrointestinal disease.
[0401] Use outside of gastroenterology
[0402] The invention with modification could potentially have uses in:
[0403] 1. Pregnancy monitoring (baby heartrate monitoring)
[0404] 2. Identification and monitoring of conditions of the lung
[0405] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte3. Identification and monitoring of conditions of the heart
[0406] 4. Identification and monitoring of conditions of the stomach
[0407] The gastrointestinal monitoring system and wearable device according to embodiments of the invention comprises a unique combination of sensors implemented in the wearable device and additionally in preferred embodiments, the use of Al to analyse multiple different data types collected by the plurality of sensors to draw conclusions regarding gastrointestinal health.
[0408] The selection of these sensor types and the combination thereof is based on prior experience of gastroenterologists.
[0409] According to example embodiments of the invention, the device may include additional sensors configured for the following measurements:
[0410] 1. Skin conductance measurement
[0411] 2. Electrogastrogram
[0412] 3. Skin temperature measurement
[0413] 4. Strain gauge
[0414] The inclusion of additional sensors configured for the above measurements according to an exemplary embodiment of the invention may provide a small further improvement in the diagnostic potential when used in combination with acoustic signals, PPG signals, and movement data measured by the wearable device.
[0415] Each "microcontroller" herein includes processing circuitry, and / or may include multiple microcontrollers. For example, as used herein, including the claims, the term "microcontroller" may include various processing circuitry, including at least one processor, wherein one or more of at least one processor, individually and / or collectively in a distributed manner, may be configured to perform various functions described herein.
[0416] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteAs used herein, when "a microcontroller", "at least one microcontroller", and "one or more microcontroller" are described as being configured to perform numerous functions, these terms cover situations, for example and without limitation, in which one microcontroller performs some of recited functions and another microcontroller(s) performs other of recited functions, and also situations in which a single microcontroller may perform all recited functions. Additionally, the at least one microcontroller may include a combination of processors performing various of the recited / disclosed functions, e.g., in a distributed manner. At least one processor of the at least one microcontroller may execute program instructions to achieve or perform various functions.
[0417] Each embodiment herein may be used in combination with any other embodiment(s) described herein. While the disclosure has been illustrated and described with reference to various example embodiments, it will be understood that the various embodiments are intended to be illustrative, not limiting. It will further be understood by those skilled in the art that various changes in form and detail may be made without departing from the true spirit and full scope of the disclosure, including the appended claims and their equivalents. It will also be understood that any of the embodiment(s) described herein may be used in combination with any other embodiment(s) described herein.
[0418] 29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und MarkenanwalteList of reference numbers
[0419] 1 Frequency range
[0420] 2 bowel sound duration
[0421] 3 interval
[0422] 5 patient
[0423] 6 doctor
[0424] 7 nurse / doctor (at bedside)
[0425] 10 body-worn device
[0426] 20 user mobile phone
[0427] 30 backend server (on premises) 31 cloud backend
[0428] 40 hospital information system 50 hospital monitor app
[0429] 61 Bluetooth connection
[0430] 62 Wifi or LTE connection
[0431] 63 local network connection 64 internet connection
[0432] 65 local network connection 66 internet connection
[0433] 71 MCU
[0434] 72 micro-SD-Card
[0435] 73 I2C Bus
[0436] 74 IMU 1
[0437] 75 IMU 2
[0438] 76 Analog Front End
[0439] 77 PPG sensor
[0440] 78 I2S Bus
[0441] 79 Microphone 1 (Body)
[0442] 29703-WO1 - PR / jes Keller Schneider
[0443] 7 May 2026 Patent- und Markenanwalte80 Microphone 2 (exterior) 81 Power Management 82 Battery
[0444] 83 Micro-USB
[0445] 84 Bluetooth
[0446] 101 state
[0447] 102 state
[0448] 103 state
[0449] 104 state
[0450] 105 state
[0451] 106 step
[0452] 107 step
[0453] 108 buffer
[0454] 111 check
[0455] 112 check
[0456] 113 check
[0457] 115 classification
[0458] 116 check
[0459] 29703-WO1 PR / jes Keller Schneider
[0460] 7 May 2026 Patent- und Markenanwalte
Claims
Claims1. A wearable device for real-time human gastrointestinal monitoring based on acoustic signals, wherein the device is configured to be worn at an abdominal area of a human for non-invasively recording sounds generated by a gastrointestinal system of the human, the device comprising:at least a first and a second acoustic sensor configured to capture acoustic signals; at least one photoplethysmographic sensor oriented towards a surface of the abdomen when the device is being worn and configured to measure a photoplethysmographic signal;at least one inertial measurement unit configured to measure a movement signal of the wearable device;wherein a sensing area of the first acoustic sensor is oriented towards a surface of the abdomen when the device is being worn and is configured to capture a first gastrointestinal acoustic signal, and wherein a sensing area of the second acoustic sensor is oriented outwards in a direction opposite a surface of the human abdomen when the device is being worn and is configured to capture an environmental acoustic signal of ambient noise in an ambient environment around the human; and a microcontroller, comprising processing circuitry, connected to measuring outputs of at least the first and second acoustic sensors, of the at least one photoplethysmographic sensor, and of the at least one inertial measurement unit (IMU), and configured to perform monitoring and analysis of the gastrointestinal system of the human based on at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal.
2. The wearable device according to claim 1, wherein the microcontroller (MCU) is configured to perform a preprocessing step comprising:29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteremoving environmental noise from the first gastrointestinal acoustic signal by subtracting from it a noise signal calculated from the environmental acoustic signal.
3. The wearable device according to any of the previous claims, wherein the microcontroller (MCU) is configured to perform a preprocessing step comprising: reducing a quantity of bytes in each of the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and / or the movement signal.
4. The wearable device according to any of the previous claims, wherein the device further comprises at least a third acoustic sensor, wherein a sensing area of the third acoustic sensor is oriented towards a surface of the abdomen when the device is being worn and is configured to capture a second gastrointestinal acoustic signal, wherein the microcontroller (MCU) is connected to a measuring output of the third acoustic sensor and is configured to:optionally, remove an environmental noise from the second gastrointestinal acoustic signal by subtracting from it a noise signal calculated from the environmental acoustic signal; andlocalize an origin within the human gastrointestinal system of the first and second gastrointestinal acoustic signals based on a phase difference measured between the first and second gastrointestinal acoustic signals.
5. The wearable device according to any of the previous claims, wherein the microcontroller is configured to selectively operate the device in a first operation mode and in a second operation mode, wherein:in the first operation mode, the photoplethysmographic signal is not measured and / or processed, and the first gastrointestinal acoustic signal is not measured and / or processed, whereas the movement signal is measured and processed;in the second operation mode, the photoplethysmographic signal is measured and processed,29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltewherein the device is switched from the first operation mode to the second operation mode if the movement signal indicates movement of the device above a first movement threshold.
6. The wearable device according to the previous claim, wherein in the second operation mode, contact of the device with the skin of the human is identified based on the photoplethysmographic signal and wherein the device is switched to a third operation mode if skin contact is present,wherein in the third operation mode, the first gastrointestinal acoustic signal is measured and processed.
7. The wearable device according to the previous claim, wherein the third operation mode comprises a first submode, in which the gastrointestinal acoustic signal is measured and processed, and a second submode, in which the gastrointestinal signal is not measured and / or processed, whereinthe first submode is activated if the movement signal indicates movement of the device below a second movement threshold, andthe second submode is activated if the movement signal indicates movement of the device above the second movement threshold.
8. The wearable device according to any of the previous claims, wherein the microcontroller (MCU) is configured to analyze the environmental acoustic signal to detect speech and to obfuscate speech in the first gastrointestinal acoustic signal if speech is detected.
9. A gastrointestinal monitoring system comprising the wearable device according to any of the previous claims and further comprising:a mobile application for a mobile device such as a smartphone, tablet, or personal computer, wherein the mobile application is configured to accept inputs from a user of the wearable device;29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltea computer server comprising processing circuitry for additional compute resources, and wherein the computer server is wirelessly connected to the microcontroller of the device and is configured to receive at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal from the microcontroller, wherein the microcontroller is optionally configured to pre-process these signals before they are received by the computer server; andoptionally, a cloud backend configured to provide an alerting functionality.
10. The gastrointestinal monitoring system according to the previous claim, wherein the mobile application is configured to accept inputs from the user of the wearable device which comprise in particular a food intake signal, a toileting event signal, and / or a medication intake data.
11. The gastrointestinal monitoring system according to claim 9 or 10, wherein the computer server or the wearable device microcontroller (MCU) is configured to:maintain a signal timeline formed by continuously acquired data comprising at least the first gastrointestinal acoustic signal, the photoplethysmographic signal, and the movement signal;receive and store at least one static contextual parameter describing a stable characteristic of the human subject and / or at least one dynamic contextual parameter, wherein the at least one dynamic contextual parameter is assigned a timestamp; andmap the at least one static contextual parameter and / or the at least one timestamped dynamic contextual parameter onto the physiological signal timeline to form a unified multimodal record, and perform monitoring and analysis of the gastrointestinal system based on the unified multimodal record.
12. The gastrointestinal monitoring system according to any of claims 9 to 11, wherein the computer server or the wearable device microcontroller (MCU) is configured to:29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltereceive and store at least one static contextual parameter of the human subject, wherein the static contextual parameter comprises a demographic parameter relating to the human subject and / or a clinical health status indicator for the human subject; andcondition the monitoring and analysis of the gastrointestinal system on the at least one static contextual parameter by applying the static contextual parameter as a covariate.
13. The gastrointestinal monitoring system according to any of claims 9 to 12, wherein the computer server or the wearable device microcontroller (MCU) is configured to:receive at least one dynamically reported contextual event comprising a dietary intake event, a defecation event, a medication intake event, and / or a subjective symptom report;assign a timestamp to each dynamically reported contextual event and align each dynamically reported contextual event with the gastrointestinal signal timeline formed by the first gastrointestinal acoustic signal, the photoplethysmographic signal, and the movement signal; andextract, for each dynamically reported contextual event, a physiological feature trajectory from the first gastrointestinal acoustic signal, the photoplethysmographic signal, and / or the movement signal within at least one of a configurable pre-event window and a configurable post-event window relative to the timestamp of the dynamically reported contextual event, and correlate the physiological feature trajectory with the dynamically reported contextual event.
14. The gastrointestinal monitoring system according to any of claims 9 to 13, wherein the computer server or the wearable device microcontroller (MCU) is configured to:receive dietary intake data and / or medication intake data comprising a time of intake; and29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltederive, from a correlation between the dietary intake data and / or the medication intake data and a feature trajectory of the first gastrointestinal acoustic signal within a post-event window and optionally within a pre-event window, at least one of:(a) an individual food-response model characterizing the effect of a dietary intake event on gastrointestinal motility as reflected in the first gastrointestinal acoustic signal;(b) a pharmacodynamic response metric characterizing the effect of a medication intake event on gastrointestinal motility as reflected in the first gastrointestinal acoustic signal, wherein a change in the first gastrointestinal acoustic signal within a pharmacokinetically expected time window following the medication intake event is used as a measure of pharmacological efficacy; and(c) a drug-food interaction metric characterizing the combined effect of a concurrent dietary intake event and a medication intake event on the physiological feature trajectory of the first gastrointestinal acoustic signal.
15. The gastrointestinal monitoring system according to any of claims 9 to 14, wherein the computer server or the wearable device microcontroller (MCU) is configured to: predict a target condition such as a target gastrointestinal condition based on an analysis of at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal.
16. The gastrointestinal monitoring system according to claim 15, wherein the computer server or the wearable device microcontroller (MCU) is configured to:receive data relating to at least one target event associated with the gastrointestinal system, wherein the target event comprises at least one of a physiological event, a symptom event, and a behaviorally logged event reported by the human subject or by a healthcare professional;29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalteretrospectively label, in the physiological signal timeline, at least one time window associated with the target event, and extract from the labelled time window a feature set derived from at least the first gastrointestinal acoustic signal; andtrain a predictive model on the extracted feature set and operate the predictive model in real time to output a probability estimate relating to a future occurrence of the target event, based on the first gastrointestinal acoustic signal and optionally based on at least one of the photoplethysmographic signal, the movement signal, and a dynamically reported contextual parameter.
17. The gastrointestinal monitoring system according to any of claims 9 to 16, wherein the computer server or the wearable device microcontroller (MCU) is configured to: calculate a plurality of machine learning features from at least the first gastrointestinal acoustic signal.
18. The gastrointestinal monitoring system according to any of claims 9 to 17, wherein the computer server or the wearable device microcontroller (MCU) is configured to: calculate a heart rate (HR) signal and a heart rate variability (HRV) signal from the photoplethysmographic (PPG) signal; anddetermine a quality of contact between the wearable device and a skin of the human subject based on a quality of the heart rate (HR) signal and the heart rate variability (HRV) signal.
19. The gastrointestinal monitoring system according to any of claims 9 to 18, wherein the computer server or the wearable device microcontroller (MCU) is configured to: calculate at least one gastrointestinal factor signal from the heart rate (HR) signal and the heart rate variability (HRV) signal wherein the at least one gastrointestinal factor signal comprises a vagal function signal, a stress signal, an arousal signal, an exercise signal; anddetermine a correlation between the at least one gastrointestinal factor signal and the first gastrointestinal acoustic signal.29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte20. The gastrointestinal monitoring system according to any of claims 9 to 19, wherein the at least one inertial measurement unit (IMU) is configured to measure a 3-axis accelerometer signal, a gyrometer signal, and a magnetometer signal, and the computer server or the wearable device microcontroller (MCU) is configured to: calculate a movement of the wearable device in three-dimensional space by calculating a roll signal, a pitch signal, and a yaw signal, and further calculate an orientation in three-dimensional space of the wearable device.
21. The gastrointestinal monitoring system according to any of claims 9 to 20, wherein the computer server or the wearable device microcontroller (MCU) is configured to: interpolate a movement data of the human from the movement signal of the wearable device, wherein the movement data comprises a number of daily steps, a body posture of the human subject, and / or a sleep stage signal; anddetermine a correlation between the movement data and the first gastrointestinal acoustic signal.
22. The gastrointestinal monitoring system according to any of claims 9 to 21, wherein the computer server or the wearable device microcontroller (MCU) is configured to: calculate a plurality of time-domain features based on a temporal analysis of the first gastrointestinal acoustic signal.
23. A method for monitoring and analysis of a gastrointestinal system of a human based on acoustic signals using a wearable device, comprising:collecting a first gastrointestinal acoustic signal generated by the gastrointestinal system using a sensing area of a first acoustic sensor of the device, wherein the first acoustic sensor is placed at an abdominal area of the human for non-invasively recording sounds generated by the gastrointestinal system of the human and a sensing area of the first acoustic sensor is oriented towards the body surface when the device is being worn;29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwaltecollecting an environmental acoustic signal comprising ambient noise using a sensing area of a second acoustic sensor of the device, wherein the sensing area of the second acoustic sensor is oriented outwards in a direction opposite a surface of the human abdomen when the device is being worn;measuring a photoplethysmographic (PPG) signal using at least one photoplethysmographic (PPG) sensor of the device, wherein the photoplethysmographic sensor contacts the body surface when the device is being worn;measuring a movement signal of the device using at least one inertial measurement unit (IMU) of the device; andperforming monitoring and analysis of the gastrointestinal system based on at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal.
24. The method according to the previous claim, wherein the method comprises performing monitoring and analysis of the gastrointestinal system by using a microcontroller (MCU) of the device, wherein the microcontroller (MCU) is connected to measuring outputs of the at least first and second acoustic sensors, of the photoplethysmographic sensor, and of the at least one inertial measurement unit.
25. The method according to claim 24, wherein the method comprises performing monitoring and analysis of the gastrointestinal system by using a computer server configured to receive at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal from the device, wherein the computer server is configured to monitor and analyze the gastrointestinal system based on at least the received first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal, wherein at least the first gastrointestinal acoustic signal, the environmental acoustic signal, the photoplethysmographic signal, and the movement signal are optionally pre-processed by a microcontroller (MCU) of the device before they are received by the computer server.29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte6026. The method according to any of claims 23 to 25, wherein the computer server is configured to monitor and analyze the gastrointestinal system in addition based on user inputs received in a mobile application and transmitted to the computer server, wherein the user inputs comprise at least a food intake signal, a toileting event signal, and / or a medication intake data.29703-WO1 - PR / jes Keller Schneider 7 May 2026 Patent- und Markenanwalte
