Swallowing detection method and operation

The system uses impedance sensors and machine learning to accurately detect swallowing events in infants, addressing the limitations of existing methods and enhancing treatment efficacy.

WO2026241247A1PCT designated stage Publication Date: 2026-11-19RES INST AT NATIONWIDE CHILDRENS HOSPITAL
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/US2026/027774
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-14
Filing Date
2026-05-12
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing methods for diagnosing infant oral feeding difficulties are subjective, invasive, and have predictive limitations, failing to accurately detect swallowing events and patterns, which are crucial for early intervention and improving health outcomes.

Method used

A system utilizing multi-channel impedance sensors positioned along the nasogastric tract to analyze impedance waveforms and identify contemporaneous drops, combined with machine learning, for precise detection and classification of swallowing events.

Benefits of technology

Enables efficient and accurate identification of swallowing abnormalities, improving survival rates and long-term health outcomes by ensuring timely administration of appropriate treatments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure US2026027774_19112026_PF_FP_ABST
    Figure US2026027774_19112026_PF_FP_ABST
Patent Text Reader

Abstract

Systems implementing methods for swallow detection include receiving a plurality of impedance waveforms indicative of swallowing activity along a nasogastric tract over a period of time. Further, the method processes the plurality of impedance waveforms as time-series data to identify a first waveform indicative of swallowing activity at a pharynx along the nasogastric tract and a second waveform indicative of swallowing activity between the pharynx and a proximal end of an esophagus along the nasogastric tract. The method identifies a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform based on an extracted periodic signal feature. Upon detecting contemporaneous drops in impedance based on the first series of candidate impedance drops and the second series of candidate impedance drops, the method classifies the plurality of impedance waveforms as a swallowing event.
Need to check novelty before this filing date? Find Prior Art

Description

SWALLOWING DETECTION METHOD AND OPERATIONRELATED APPLICATIONS

[0001] The present application claims priority to U.S. Provisional Patent Application Serial No. 63 / 805,657 filed May 14, 2025, entitled Swallowing Detection Method And Operation. The above provisional is incorporated by reference in its entirety for all purposes.GOVERNMENT SUPPORT

[0002] This invention was made with government support under R01 DK136762 awarded by the National Institutes of Health. The government has certain rights in the invention.TECHNICAL FIELD

[0003] The present disclosure is directed to swallow detection and oral feeding difficulty diagnostic systems, including but not limited to, identifying swallowing rhythms and patterns in infants based on impedance data for detection and treatment of oral feeding difficulties.BACKGROUND

[0004] Feeding difficulties are a common concern among infants recovering in neonatal intensive care units (NICUs) worldwide. These challenges represent the leading cause of prolonged hospital stays, chronic reliance on tube feeding, and the development of airway, digestive, and neurological morbidities. Early identification and effective management of feeding difficulties can significantly reduce associated complications, improve outcomes and minimize healthcare costs.

[0005] Diagnosing infant oral feeding difficulties involves a combination of clinical assessments, caregiver observations, and specialized diagnostic tools. These methods can be subjective, invasive, and have predictive limitations for long-term outcomes. For accurate diagnosis of feeding difficulties, precise detection and analysis of swallowing are essential. Automated detection methods offer the potential1 of 25NCH -032911 WO ORDto improve accuracy, efficiency, and consistency, reducing reliance on subjective assessments and interpretation.SUMMARY

[0006] According to one aspect, a computer-implemented method includes receiving a plurality of impedance waveforms indicative of swallowing activity along the nasogastric tract over a period of time. Such periods of observations can be during oral feeding or caregiving. The method includes processing the plurality of impedance waveforms as time-series data to identify a first waveform indicative of swallowing activity at a pharynx along the nasogastric tract and a second waveform indicative of swallowing activity between the pharynx and a proximal end of an esophagus along the nasogastric tract. The method includes identifying a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform based on an extracted periodic signal feature. Upon detecting contemporaneous drops in impedance based on the first series of candidate impedance drops and the second series of candidate impedance drops, the method classifies the plurality of impedance waveforms as a swallowing event. Such downward propagation of signals can consitute signature patterns of peristalsis.

[0007] According to another aspect, a system for swallow detection includes a plurality of sensors positioned along a nasogastric tract. A first sensor of the plurality of sensors is located at a pharynx. A second sensor of the plurality of sensors is located between the pharynx and a proximal end of an esophagus. The system includes a processor operatively connected for computer communication to the plurality of sensors. The processor receives a first waveform from the first sensor and a second waveform from the second sensor. The first waveform and the second waveform are obtained simultaneously over a period of time. The processor identifies a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform based on extracted periodic signal features. The processor classifies a swallowing event upon detecting2 of 25NCH -032911 WO ORDcontemporaneous drops in impedance based on the first series of candidate impedance drops and the second series of candidate impedance drops.

[0008] According to a further aspect, a non-transitory computer-readable medium storing computer-executable instructions that when executed by a computer, which includes at least a processor, cause the computer to receive a plurality of impedance waveforms indicative of swallowing activity along a nasogastric tract over a period of time and process the plurality of impedance waveforms as time-series data to identify a first waveform and a second waveform. The first waveform indicative of swallowing activity at a pharynx along the nasogastric tract and the second waveform indicative of swallowing activity between the pharynx and a proximal end of an esophagus along the nasogastric tract. The computer identifies a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform based on an extracted periodic signal feature. Additionally, the computer detects a swallowing event upon identifying contemporaneous drops in impedance based on the first series of candidate impedance drops and the second series of candidate impedance drops.BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate various systems, methods, devices, and other embodiments of the disclosure. It will be appreciated that the illustrated element boundaries (e.g., boxes, groups of boxes, directional lines, or other shapes) in the figures represent one embodiment of the boundaries. In some embodiments, one element may be designed as multiple elements or that multiple elements may be designed as one element. In some embodiments, an element shown as an internal component of another element may be implemented as an external component and vice versa. Like references numerals refer to like parts throughout the drawings unless described otherwise. Furthermore, elements may not be drawn to scale.

[0010] FIG. 1 is a schematic diagram of a sensor device used to collect impedance and pH data from an infant according to one exemplary embodiment;3 of 25NCH -032911 WO ORD

[0011] FIG. 2 is a block diagram of a system for swallow detection implemented within an exemplary operating environment according to one exemplary embodiment;

[0012] FIG. 3 is a flow diagram of an exemplary method for swallow detection according to an exemplary embodiment;

[0013] FIG. 4 is a flow diagram of an exemplary method for swallow detection according to an exemplary embodiment;

[0014] FIG. 5 is a flow diagram of an exemplary method for swallow detection according to an exemplary embodiment;

[0015] FIG. 6A is a visual representation of an impedance waveform according to one embodiment; and

[0016] FIGS. 6B-1 through 6B-4 collectively illustrate FIG. 6B, which is a visual representation of pH-impedance data waveforms acquired by the sensor device of FIG. 1 over a 24-hour period according to one embodiment.DETAILED DESCRIPTION

[0017] Aspects of the present disclosure are directed to impedance waveform analysis for swallow detection and oral feeding difficulty diagnosis, particularly in infants. In general, and as will be described in detail below, multi-channel impedance sensor data is obtained in a time-series from an infant (i.e. , a target subject). At least one sensor is located at a pharynx of the infant, and another sensor is located between the pharynx and a proximal end of an esophagus. The impedance waveforms from these sensors are analyzed to identify patterns and detect a swallowing event. More specifically, identification of contemporaneous drops in impedance within a particular time period are identified and a swallowing event is classified. The impedance data can be used for oral feeding diagnosis and determination of treatment. In some embodiments, a machine learning model can be trained and applied for the detection of swallowing events.

[0018] As mentioned above, diagnosing infant oral feeding difficulties involves a combination of clinical assessments, caregiver observations, and specialized diagnostic tools. These methods can be subjective, invasive, and have predictive limitations for long-term outcomes. Oral swallowing difficulties and gastroesophageal4 of 25NCH -032911 WO ORDreflux disease (GERD) frequently coexist, presenting a complex clinical scenario in infants. Methods utilizing pH-multi-channel impedance (pH-MII) can be used to evaluate gastroesophageal reflux (GER) events in infants by measuring the amount of gas and / or liquid reflux in the esophagus and backward movement of bolus. While pH-MII can detect reflux events, it does not provide direct insights into other oral feeding difficulties that require accurate detection of swallowing events. Additionally, validation of pH-MII data for forward bolus movement is tedious and labor intensive. Accordingly, with the lack of existing biomarker swallow detection along with the limitations of existing GERD diagnostic systems, there is a clear and unmet need in the art for advanced detection mechanisms of swallowing events and patterns along with oral feeding difficulty diagnosis.

[0019] Thus, the technology disclosed herein provides a technological improvement in detecting swallowing events and diagnosing oral feeding disorders in infants by automatically classifying swallowing events based on analyzing impedance time-series waveforms for contemporaneous drops in impedance within a predetermined time period. Additionally, machine learning components will enable the systems disclosed herein to improve performance for faster and more accurate analysis. This ultimately provides the technological benefit to the functioning of oral feeding disorder detection systems by providing efficient identification of swallowing patterns from impedance data. Early and precise identification of swallowing abnormalities improves survival rates and long-term health outcomes in infants by ensuring the right treatments are administered at the right time.

[0020] Several implementations are discussed herein in more detail with reference to the figures, wherein the showings are for purposes of illustrating one or more exemplary embodiments and not for purposes of limiting same. FIG. 1 is a schematic diagram of a sensor device used to collect impedance and pH data from an infant according to one exemplary embodiment. As shown in FIG. 1, a sensor device 102 includes a probe 104 with a plurality of sensors 106. The probe 104 can be a flexible catheter. The plurality of sensors 106 includes six impedance sensors, Z1-Z6 and a pH sensor 122. It is noted that in FIG. 1 and will be discussed in further detail herein, the sensors 106 are represented schematically within the sensor device 102 and5 of 25NCH -032911 WO ORDdepicted as corresponding geometric shapes along the nasogastric tract 108 to illustrate one example of physical placement within an anatomical environment. More specifically, in FIG. 1, the sensors Z1-Z6 are illustrated as triangular shapes and the pH sensor 122 is illustrated as a square shape along the nasogastric tract 108.

[0021] As shown in FIG. 1, the probe 104 is placed through the nasogastric tract 108 of an infant (i.e. , a target subject) 110. The nasogastric tract 108 is a pathway that runs from the nose (i.e., nasal cavity) 112, down the pharynx 114, bypasses the trachea 116, into the esophagus 118 and then ultimately reaches the stomach 120. The plurality of sensors 106 are located at a plurality of different locations spaced apart from each other along a length of the nasogastric tract 108. In some embodiments, the plurality of sensors 106 are spaced evenly along the length of the nasogastric tract 108. For example, the plurality of sensors 106 can be positioned along a length of the nasogastric tract 108, where each sensor is spaced apart from adjacent sensors by a predetermined uniform distance.

[0022] In FIG. 1 , the Z1 sensor is at the level of the pharynx 114, the Z2 sensor is located between the pharynx 114 and a proximal end 118a of the esophagus 118, and the Z6 sensor is at the level of a distal end 118b of the esophagus 118. The Z3-Z5 sensors are spaced apart between the Z2 sensor and the Z6 sensor. The pH sensor 122 is located at the distal end 118b of the esophagus 118 or nasogastric tract 108, for example, about one-third the distance above the gastro-esophageal junction. In FIG. 1, the pH sensor 122 is at the level of the T7-T8 vertebrae. In other embodiments, the pH sensor 122 can be located between the Z5 sensor and the Z6 sensor.

[0023] In operation, for example, during an infant oral feeding challenge, the impedance sensors Z1-Z6 measures antegrade movement of bolus from the pharynx 114, through the esophagus 118, and to the stomach 120. The result of which is a plurality of impedance versus time waveforms obtained simultaneously in a timeseries. Typically, the antegrade movement of the bolus causes a sequential decrease in the impedance as it moves from the upper sensors (e.g., Z1) to the lower sensors (e.g., the pH sensor 122). In the embodiments described herein, a swallowing event can be detected based on impedance drops that occur contemporaneously in the region of the pharynx 114 to the proximal end of the esophagus 118a.6 of 25NCH -032911 WO ORD

[0024] FIG. 2 is a block diagram of an exemplary operating environment 200 in which the swallow detection system can operate. In particular, the exemplary operating environment 200 includes a swallow detection system 202, the sensor device 102, other sensors 204, and a clinic device 206, each operatively connected for computer communication via, for example, a network 208. The swallow detection system 202 facilitates swallow detection and oral feeding disorder diagnosis and can include provisions for processing, communicating, and interacting with various components of the swallow detection system 202 and other components in the environment 200. The swallow detection system 202 generally includes processor(s) 210, memory 212, storage 214, I / O interface 216, and communication interface 218. It is understood that one or more of the components of the swallow detection system 202 and / or one or more of the functions of the components of the swallow detection system 202 can be distributed across multiple computing devices in the environment 200 (e.g., the sensor device 102, the clinic device 206 or other external devices).

[0025] The processor(s) 210 can include logic circuitry with hardware, firmware, and software architecture frameworks for facilitating control of the swallow detection system 202 and communication between the swallow detection system 202 and other components of the environment 200. Thus, in some embodiments, the processor(s) 210 can store application frameworks, kernels, libraries, drivers, application program interfaces, among others, to execute and control hardware and functions discussed herein. In some embodiments, the processor(s) 210 can include machine learning engines to facilitate training and processing of machine learning models. In one embodiment, the processor(s) 210 can include a data acquisition module 220 and a swallow detection module 224. The data acquisition module 220 measures and / or obtains sensor data, from for example, the sensor device 102, and will be described in more detail with FIG. 3. The swallow detection module 224 identifies impedance patterns and swallowing events and will be described in more detail with FIGS. 3-5.

[0026] The memory 212 can include volatile or non-volatile storage and can include both read-only and writable memory. The memory 212 can store programs and software including, but not limited to, operating systems and programs that control hardware and functions discussed herein. The storage 214 can include one or more7 of 25NCH -032911 WO ORDlocal devices and / or interfaces to remote storage devices. The storage 214 can also include databases or other data stores. In FIG. 2, the storage 214 can include ML models 226. The ML models 226 can be stored locally and / or on a remote server or storage device. The ML models 226 are machine learning models that can be used for training and / or detecting swallowing events as described herein.

[0027] The storage 214 can also include biomarker data 228 that is associated with the target subject. The biomarker data 228 can include physiological biomarkers that are monitored in neonatal intensive care unit (NICU) or other clinical settings. For example, physiological biomarkers can include, but are not limited to, cardiac biomarkers and respiratory biomarkers. Cardiac biomarkers can be derived from electrocardiogram (ECG) signals, such as heart rate, heart rate variability, other heart rhythm characteristics. Respiratory biomarkers can include respiratory rate, respiratory rhythm, breathing patterns, among others. Other types of biomarkers can include oxygen-related biomarkers such as pulse oximetry waveforms.

[0028] In one embodiment, the biomarker data 228 is measured during collection of the impedance data, for example, during an oral feeding challenge. In some embodiments, the biomarker data 228 can be measured via the other sensors 204 and / or the clinic device 206, as detailed below. As will be discussed in further detail herein, the biomarker data 228 can be used in conjunction (e.g., amalgamated) with the impedance and pH measurements from the sensor device 102 to identify and / or classify physiological events (e.g., changes in specific biomarkers) associated with the swallowing event.

[0029] Referring again to FIG. 2, the I / O interface 216 controls input to the processor(s) 210 and / or output to other components of the swallow detection system 202. In particular, the I / O interface 216 can facilitate transfer of data between the system 202 and input / output devices (not shown). Input / output devices can include, but are not limited to, a mouse, keyboard, a touchscreen, a display, a touchpad, a wearable input device, a camera, a microphone, wireless remote sensor and / or data acquisition devices, among others.

[0030] The communication interface 218 can include software and hardware to facilitate data input and output between the components of the swallow detection8 of 25NCH -032911 WO ORDsystem 202 and other components of the environment 200. Specifically, the communication l / F 218 can include network interface controllers (not shown) and other hardware and software that manages and / or monitors connections and controls bidirectional data transfer between the communication l / F 218 and other components of the environment 200 using, for example, a network 208. As another example, the communication l / F 218 can facilitate communication (e.g., exchange data and / or transmit messages) with the sensor device 102, other sensors 204, and / or the clinic device 206.

[0031] As mentioned with FIG. 1, the sensor device 102 includes the plurality of sensors 106. Although not shown in FIG. 2, the sensor device 102 can include processor(s), memory, storage, I / O interfaces, and communication interfaces providing similar functionality as those same named components described with the swallow detection system 202. Other sensors 204 can include any other sensors that measure biometric and / or physiological information about the infant 110. As will be discussed herein, the data sensed by the other sensors 204 can be used with the impedance data for swallowing event detection and / or diagnosis of oral feeding difficulties. The sensors 204 can be part of one or more monitoring systems, wearable devices, or other portable devices. For example, the sensors 204 can measure heart rate, blood pressure, oxygen content, electrical activity in the brain, among other biometric and / or physiological information. In some embodiments, the other sensors 204 can include imaging sensors that sense gestures, head movement, body movement, eye and / or facial movement, among others. The sensors 204 can be contact sensors and / or contactless sensors and can include electric current / potential sensors (e.g., proximity, inductive, capacitive, electrostatic, electromagnetic, radio frequency), ultrasonic sensors, vibration sensors, visual sensors, photoelectric sensors, among others.

[0032] The clinic device 206 can be one or more computing devices implemented within a clinical setting external from the body of the infant 110. In some embodiments, the clinic device 206 is operated by a clinician or a remote user or entity. For example, the clinic device 206 can exchange and process data between the clinic device 206 and the environment 200 for oral feeding diagnosis and / or treatment. Although not9 of 25NCH -032911 WO ORDshown in FIG. 2, the clinic device 206 can include processor(s), memory, storage, I / O interfaces, and communication interfaces providing similar functionality as those same named components described with the swallow detection system 202. In some embodiments, the clinic device 206 measures and / or stores biomarker data 228.

[0033] In some implementations, one or more of the components described above can execute one or more of the processes described below. Accordingly, exemplary methods for swallow detection will now be described. FIG. 3 is a flow diagram illustrating a method 300 used in some implementations of the swallow detection technology discussed herein. FIG. 3 will be discussed with reference to the components described above with FIGS. 1 and 2. Additionally, the method 300 will be described in more detail with FIGS. 4-6.

[0034] At block 302, the method 300 includes receiving and processing sensor data, for example, from the sensor device 102. The sensor data 304 includes data indicative of swallowing activity received from sensors simultaneously over a period of time to form time-series data. For example, the sensor data 304 can include a plurality of impedance waveforms indicative of swallowing activity along a nasogastric tract 108. More specifically, this sensor data 304 includes impedance measurements associated with bolus movement along the nasogastric tract 108. The sensor data can also include pH measurements from the pH sensor 122 and / or additional biomarker data 228. Further, at block 302, the plurality of impedance waveforms can be processed to organize the time-series data and identify waveforms associated with sensors and / or regions of the nasogastric tract 108. Exemplary waveform processing and identification techniques will be discussed in further detail with respect to FIGS.4-6.

[0035] At block 306, the method 300 includes detecting swallow patterns based on impedance measurements in the sensor data. For example, as will be discussed in further detail with FIGS. 4-6, swallow patterns are detected based on identifying candidate impedance drops across multiple impedance waveforms. At block 308, the method 300 includes detecting a swallowing event based on the swallow patterns (i.e. , identified at block 306). For example, as will be discussed in more detail with respect to FIGS. 4 and 5, upon detecting contemporaneous impedance drops within a10 of 25NCH -032911 WO ORDpredetermined time window, the plurality of impedance waveforms can be classified as indicating a swallowing event. It is understood that in some embodiments, “swallow pattern,” as used herein, can refer to one or more detected characteristics, relationships, trends, or temporal arrangements within the impedance waveform data that are indicative of potential swallowing activity. A “swallowing event,” as used herein, can refer to a classified or physiological swallowing occurrence determined based on the swallow patterns that satisfy predetermined criteria discussed herein.

[0036] At block 310, the method 300 includes characterizing the swallowing event and / or generating output. In some embodiments, after a swallowing event is identified at block 308, the swallowing event can be further classified, characterized, annotated, and / or associated with one or more generated outputs. For example, the swallowing event can be further classified as normal and / or abnormal based on one or more swallow patterns identified in the impedance waveform data. For example, abnormal swallow patterns can include delayed swallow responses, incomplete bolus propagation, swallowing events associated with reflux events and / or associated with other physiological changes (e.g., detected based on biomarker data).

[0037] As mentioned above, and as discussed with FIG. 4, block 410 below, biomarker data can be integrated with the impedance waveform data. Accordingly, based on this integration, at block 310 the method 300 can also include identifying metadata and / or a potential cause associated with the swallowing event based on additional biomarker data and / or pH values. In other words, autonomic changes can be identified based on biomarker data occurring within or during a classified swallowing event. For example, in one embodiment, the plurality of sensors 106 includes a pH sensor 122 that generates a time-series pH waveform. In some embodiments, the processor 210 can determine a pH value associated with the swallowing event from the time-series pH waveform. Said differently, a pH value corresponding to the swallowing event is determined by identifying a pH value from the time-series pH waveform at a time associated with the swallowing event (i.e., within a time window associated with the detected contemporaneous drops in impedance). Based on the pH value, the swallowing event can be correlated with a reflux event and / or a potential cause of the swallowing event. For example, if the pH11 of 25NCH -032911 WO ORDvalue indicates an acidic condition during or near the swallowing event, the swallowing event can be correlated with, temporally associated with, and / or indicative of an acid gastroesophageal reflux (GER) event.

[0038] Additionally, in some embodiments, block 310 can include generating an oral feeding assessment, recommendation, graphical report, and / or other types of output to, for example, the clinic device 206. In some embodiments, machine learning training and / or application can be performed at blocks 306, 308, and / or 310 to identify, classify, and / or characterize swallowing events. As an illustrative example, a swallowing impedance pattern can be identified and / or extracted based in part on the detected contemporaneous drops. The swallowing impedance pattern can then be input into the ML models to generate an output indicative of whether a swallowing event has occurred. Thus, impedance waveform data, extracted features and patterns, timing information, biomarker correlations, detected impedance swallow patterns, detected swallowing events, and other classifications can be used to train ML models 226. Further, the ML models 226 can be continuously updated and refined using newly acquired waveform data and / or subsequently classified swallowing events.

[0039] The method 300 will now be described in more detail with reference to FIG.4. As mentioned above with block 302, the methods described herein can include processing the plurality of impedance waveforms. For example, at block 402, the method 400 can include identifying waveforms and at block 404, the method 400 can include extracting periodic features. In particular, block 402 can include processing the plurality of impedance waveforms as time-series data to identify a first waveform indicative of swallowing activity at a pharynx along the nasogastric tract and a second waveform indicative of swallowing activity between the pharynx and a proximal end of an esophagus along the nasogastric tract. For example, the plurality of impedance waveforms can be received from the sensor device 102 and the plurality of sensors 106 positioned along the nasogastric tract. The first waveform is received from a first sensor Z1 and the second waveform is received from a second sensor Z2 in the plurality of sensors 106. As shown in FIG. 1, the first sensor Z1 is located at the pharynx 114 and the second sensor Z2 is located between the pharynx 114 and the12 of 25NCH -032911 WO ORDproximal end 118a of the esophagus 118. Exemplary waveforms and sensor output from the sensor device 102 will be described with reference to FIGS. 6A and 6B.

[0040] FIG. 6A illustrates an exemplary general waveform having periodic signal features. In FIG. 6A the periodic signal features can include any characteristic of the waveform that reoccurs in the signal over a period of time. For example, as shown in FIG. 6A, the periodic signal features can include, but are not limited to, crests, troughs, amplitudes, and wavelengths. In some embodiments, a waveform cycle can extend between successive recurring periodic signal features, such as between adjacent crests or adjacent troughs. For example, in FIG. 6A, a wave A and a wave B can each be defined between adjacent amplitude peaks. As will be discussed in more detail herein, drops in impedance are detected by analyzing the trough of each waveform.

[0041] FIGS. 6B-1 through 6B-4 collectively illustrate FIG. 6B, which is presented across multiple sheets for clarity and readability. FIGS. 6B-1 through 6B-4 together depict a visual representation of pH-impedance waveform data acquired by the sensor device of FIG. 1 over a 24-hour period according to one embodiment. During the 24-hour period, a probe (e.g., the sensor device 102) with 6 impedance sensors (Z1-Z6) and 1 pH sensor 122 sensor are placed through the nasogastric tract 108 with the pH sensor 122 at the level of the T7-T8 vertebrae. Z1 is at the level of the pharynx while Z6 is at the level of the distal esophagus. The data acquired during the study by each sensor is shown in the left panel 602. In the left panel 602, the rectangles 606a-606h denote the eight oral feeding sessions during the study. The results / analysis of the data acquired by each sensor according to the swallowing detection method and systems discussed herein during a 20 second time window associated with each of the sensors is displayed in the right panel 604. More specifically, the results associated with each sensor are displayed in the right panel 604, wherein section 608a corresponds to results associated with sensor Z1, section 608b corresponds to results associated with sensor Z2, section 608c corresponds to results associated with sensor Z3, section 608d corresponds to results associated with sensor Z4, section 608e corresponds to results associated with sensor Z5, section 608f corresponds to results associated with sensor Z6, section 608g corresponds to results associated with sensor pH.13 of 25NCH -032911 WO ORD

[0042] In the right panel 604, dashed lines shown with respect to sensor Z1 in section 608a represent a manually determined swallow used as reference annotations (i.e., ground truth swallowing events), while the smaller dashed lines with respect to sensor Z1 in section 608a represent the detected swallowing events based on the detected valleys. The inverted triangles with respect to sensors Z1 and Z2 in sections 608a and 608b denote automatically detected signal valleys (i.e., troughs) in Z1 and Z2. All predicted swallows shown are within ± 250ms of the ground truth swallow. The data shown in FIG. 6B will be described in further detail herein with the methods for swallow detection.

[0043] Referring again to FIG. 4, at block 404, the method 400 includes extracting periodic features from the waveforms. As discussed above with FIG. 6A, periodic signal features can include any characteristic of the waveform that reoccurs in the signal over a period of time. In one embodiment, the extracted periodic signal features includes trough amplitude. The troughs correspond to the local minima within he waveform and the trough amplitude refers to the vertical distance (i.e., magnitude) between the baseline and the lowest point (i.e., the trough) in the wave. Said differently, it represents the magnitude of the impedance decrease relative to the baseline.

[0044] At block 406, the method 400 includes identifying impedance drops based on the periodic features. More specifically, a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform are identified based on extracted periodic signal features. In one embodiment, the first series of candidate impedance drops are identified based on a trough amplitude of each wave in the first waveform and the second series of candidate impedance drops are identified based on a trough amplitude of each wave in the second waveform. In some embodiments, a candidate impedance drop is determined if the trough amplitude is greater than or equal to a predetermined value. With reference to FIG. 6B-2, the inverted triangles with respect to sensors Z1 and Z2 in sections 608a and 608b represent extracted periodic features, namely, troughs, that are identified as candidate impedance drops. Thus, the inverted triangles shown in section 608a can be a first series of candidate impedance drops and the inverted14 of 25NCH -032911 WO ORDtriangles shown in section 608b can represent a second series of candidate impedance drops.

[0045] In some embodiments, a candidate impedance drop is determined if the trough amplitude illustrates a decrease in impedance from baseline within a predetermined amount thereby indicating an abrupt impedance drop. In some embodiments, a 50% drop in impedance from baseline indicates a candidate impedance drop. In other embodiments, a 20% decrease in impedance from baseline indicates a candidate impedance drop. As an illustrative example with reference to FIG. 6B, a 20% decrease in impedance from baseline in Z1 and Z2, followed by an antegrade drop in impedance reaching at least the Z3 channel or below and a return to the baseline in each channel may indicate a candidate impedance drop.

[0046] At block 408, the method 400 includes detecting contemporaneous impedance drops between the first waveform and the second waveform based on the candidate impedance drops. More specifically, it is determined for each candidate impedance drop in the first series of candidate impedance drops whether a corresponding candidate impedance drop in the second series of candidate impedance drops occurs within a predetermined time window. In one embodiment, the predetermined time window is 250ms. As shown in FIG. 6B-2, the smaller dashed lines shown in section 608a represent the detected swallowing events based on the candidate impedance drops. Detecting contemporaneous impedance drops will now be described in more detail.

[0047] FIG. 5 a flow diagram illustrating a method 500 used in some implementations of the swallow detection technology discussed herein. Throughout method 500, variable i represents a candidate impedance drop in the first series of candidate impedance drops. At block 502, the method 500 includes determining if i is less than the length of the array represented by Z1 (i.e. , i is less than the number of candidate impedance drops in the first series). If YES, the method 500 proceeds to block 504. If NO, the method 500 ends. Said differently, blocks 504, 506, 508 and 510 are performed for each candidate impedance drop in the first series of candidate impedance drops.15 of 25NCH -032911 WO ORD

[0048] At block 504, an impedance drop time interval is determined between Z1(i) and Z2(i). The impedance drop time interval is the time difference between the occurrence of Z1(i) and the occurrence of Z2(i). At block 506, it is determined if the drop time interval is less than or equal to a predetermined time window. In one embodiment, the predetermined time window is 250ms. If the determination at block 506 is YES, the method proceeds to block 508, where a swallowing event is detected as TRUE. If the determination at block 506 is NO, the method 500 proceeds to block 510, where a swallowing event is detected as FALSE.

[0049] Referring again to FIG. 4, the method 400 can include at block 410 integrating biomarker data. As mentioned above, the biomarker data 228 can be used in conjunction (e.g., amalgamated) with the impedance and pH measurements from the sensor device 102 to identify and / or classify physiological events (e.g., changes in specific biomarkers) associated with the swallowing event. Thus, in one embodiment, the biomarker data 228 includes biomarker time-series data contemporaneous with the impedance waveforms. Thus, at block 410, the method 400 can include integrating the biomarker time-series data with the plurality of impedance waveforms. Additionally, a change in the biomarker time-series data occurring within the detected contemporaneous drops in impedance can be identified. Based on the integration of the biomarker data, characterization of swallowing events and / or generated output can perform as discussed herein and at block 310 of FIG. 3. Thus, the systems and methods described herein can ultimately provide the technological benefit to the functioning of oral feeding disorder detection systems by providing efficient identification of swallowing patterns from impedance data.

[0050] DEFINITIONS

[0051] The following includes definitions of selected terms employed herein. The definitions include various examples and / or forms of components that fall within the scope of a term and that may be used for implementation. The examples are not intended to be limiting. Further, the components discussed herein, may be combined, omitted or organized with other components or into different architectures.

[0052] “Component,” as used herein, refers to a computer-related entity (e.g., hardware, firmware, instructions in execution, combinations thereof). Computer16 of 25NCH -032911 WO ORDcomponents may include, for example, a process running on a processor, a processor, an object, an executable, a thread of execution, and a computer. A computer component(s) may reside within a process and / or thread. A computer component may be localized on one computer and / or may be distributed between multiple computers.

[0053] “Computer communication,” as used herein, refers to a communication between two or more computing devices (e.g., computer, personal digital assistant, cellular telephone, network device, vehicle, vehicle computing device, infrastructure device, roadside device) and may be, for example, a network transfer, a data transfer, a file transfer, an applet transfer, an email, a hypertext transfer protocol (HTTP) transfer, and so on. A computer communication may occur across any type of wired or wireless system and / or network having any type of configuration, for example, a local area network (LAN), a personal area network (PAN), a wireless personal area network (WPAN), a wireless area network (WAN), a wide area network (WAN), a metropolitan area network (MAN), a virtual private network (VPN), a cellular network, a token ring network, a point-to-point network, an ad hoc network, a mobile ad hoc network, among others. Computer communication may utilize any type of wired, wireless, or network communication protocol including, but not limited to, Ethernet (e.g., IEEE 802.3), WiFi (e.g., IEEE 802.11), communications access for land mobiles (CALM), WiMax, Bluetooth, Zigbee, ultra-wideband (UWAB), multiple-input and multiple-output (MIMO), telecommunications and / or cellular network communication (e.g., SMS, MMS, 3G, 4G, LTE, 5G, GSM, CDMA, WAVE), satellite, dedicated short range communication (DSRC), among others.

[0054] “Database,” as used herein, is used to refer to a table. In other examples, “database” may be used to refer to a set of tables. In still other examples, “database” may refer to a set of data stores and methods for accessing and / or manipulating those data stores. A database may be stored, for example, at a disk and / or a memory.

[0055] “Memory,” as used herein may include volatile memory and / or nonvolatile memory. Non-volatile memory may include, for example, ROM (read only memory), PROM (programmable read only memory), EPROM (erasable PROM), and EEPROM (electrically erasable PROM). Volatile memory may include, for example, RAM (random access memory), synchronous RAM (SRAM), dynamic RAM (DRAM),17 of 25NCH -032911 WO ORDsynchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), and direct RAM bus RAM (DRRAM). The memory may store an operating system that controls or allocates resources of a computing device.

[0056] “Module,” as used herein, includes, but is not limited to, non-transitory computer readable medium that stores instructions, instructions in execution on a machine, hardware, firmware, software in execution on a machine, and / or combinations of each to perform a function(s) or an action(s), and / or to cause a function or action from another module, method, and / or system. A module can also include logic, a software-controlled microprocessor, a discrete logic circuit, an analog circuit, a digital circuit, a programmed logic device, a memory device containing executing instructions, logic gates, a combination of gates, and / or other circuit components. Multiple modules can be combined into one module and single modules can be distributed among multiple modules.

[0057] “Operable connection,” or a connection by which entities are “operably connected,” is one in which signals, physical communications, and / or logical communications may be sent and / or received. An operable connection may include a wireless interface, a physical interface, a data interface, and / or an electrical interface.

[0058] “Processor,” as used herein, processes signals and performs general computing and arithmetic functions. Signals processed by the processor may include digital signals, data signals, computer instructions, processor instructions, messages, a bit, a bit stream, that may be received, transmitted and / or detected. Generally, the processor may be a variety of various processors including multiple single and multicore processors and co-processors and other multiple single and multicore processor and co-processor architectures. The processor may include logic circuitry to execute actions and / or algorithms.

[0059] The embodiments discussed herein can also be described and implemented in the context of “computer-readable medium” or “computer storage medium.” As used herein, “computer-readable medium” or “computer storage medium refers to a non-transitory medium that stores instructions, algorithms, and / or data configured to perform one or more of the disclosed functions when executed. Computer-readable medium can be non-volatile, volatile, removable, and non-18 of 25NCH -032911 WO ORDremovable, media implemented in any method or technology for storage of information such as computer readable instructions, data structures, modules or other data. Computer-readable medium can include, but is not limited to, a floppy disk, a flexible disk, a hard disk, a magnetic tape, other magnetic medium, an application specific integrated circuit (ASIC), a programmable logic device, a compact disk (CD), other optical medium, a random access memory (RAM), a read only memory (ROM), a memory chip or card, a memory stick, solid state storage device (SSD), flash drive, and other media from which a computer, a processor or other electronic device can interface with. Computer-readable medium excludes non-transitory tangible media and propagated data signals.

[0060] It will be appreciated that various embodiments of the above-disclosed and other features and functions, or alternatives or varieties thereof, may be desirably combined into many other different systems or applications. Also, that various presently unforeseen or unanticipated alternatives, modifications, variations or improvements therein may be subsequently made by those skilled in the art which are also intended to be encompassed by the following claims.19 of 25NCH -032911 WO ORD

Claims

CLAIMS:

1. A computer-implemented method for swallowing detection, comprising:receiving a plurality of impedance waveforms indicative of swallowing activity along a nasogastric tract over a period of time;processing the plurality of impedance waveforms as time-series data to identify a first waveform indicative of swallowing activity at a pharynx along the nasogastric tract and a second waveform indicative of swallowing activity between the pharynx and a proximal end of an esophagus along the nasogastric tract;identifying a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform based on an extracted periodic signal feature; andupon detecting contemporaneous drops in impedance based on the first series of candidate impedance drops and the second series of candidate impedance drops, classifying the plurality of impedance waveforms as a swallowing event.

2. The computer-implemented method of claim 1, wherein the plurality of impedance waveforms is received from a plurality of sensors positioned along the nasogastric tract, wherein the first waveform is received from a first sensor in the plurality of sensors and the second waveform is received from a second sensor in the plurality of sensors.

3. The computer-implemented method of claim 2, wherein the first sensor is located at the pharynx and the second sensor is located between the pharynx and the proximal end of the esophagus.

4. The computer-implemented method of claim 1, wherein the extracted periodic signal feature is trough amplitude.

5. The computer-implemented method of claim 1 , wherein the first series of candidate 20 of 25NCH -032911 WO ORDimpedance drops are identified based on a trough amplitude of each wave in the first waveform and the second series of candidate impedance drops are identified based on a trough amplitude of each wave in the second waveform.

6. The computer-implemented method of claim 5, wherein detecting contemporaneous drops in impedance includes determining the trough amplitude of each wave in the first series of candidate impedance drops whether the trough amplitude of each wave in the second series of candidate impedance drops occurs within a predetermined time window.

7. The computer-implemented method of claim 1, wherein detecting contemporaneous drops in impedance includes determining for each candidate impedance drop in the first series of candidate impedance drops whether a corresponding candidate impedance drop in the second series of candidate impedance drops occurs within a predetermined time window.

8. The computer-implemented method of claim 7, wherein the predetermined time window is 250ms.

9. The computer-implemented method of claim 1, wherein detecting contemporaneous drops in impedance is performed by applying a machine learning model.

10. The computer-implemented method of claim 1, further including integrating biomarker time-series data with the plurality of impedance waveforms, and identifying a change in the biomarker time-series data occurring within the detected contemporaneous drops in impedance.

11. A system for swallow detection, comprising:a plurality of sensors positioned along a nasogastric tract, wherein a first sensor of the plurality of sensors is located at a pharynx and a second sensor is located21 of 25NCH -032911 WO ORDbetween the pharynx and a proximal end of an esophagus; anda processor operatively connected for computer communication to the plurality of sensors, wherein the processor:receives a first waveform from the first sensor and a second waveform from the second sensor, wherein the first waveform and the second waveform are obtained simultaneously over a period of time;identifying a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform based on extracted periodic signal features; andclassifying a swallowing event upon detecting contemporaneous drops in impedance based on the first series of candidate impedance drops and the second series of candidate impedance drops.

12. The system of claim 11, wherein detecting contemporaneous drops in impedance includes determining for a trough amplitude of each wave in the first series of candidate impedance drops whether a trough amplitude of each wave in the second series of candidate impedance drops occurs within a predetermined time window.

13. The system of claim 12, wherein the predetermined time window is 250ms.

14. The system of claim 11, wherein the processor further extracts a swallowing impedance pattern based on the detected contemporaneous drops and inputting the swallowing impedance pattern into a trained machine learning model, and generates, using the trained machine learning model, an output indicative of whether a swallowing event has occurred.

15. The system of claim 11 , wherein the plurality of sensors includes a pH sensor that generates a time-series pH waveform, wherein the processor further extracts a pH value associated with the swallowing event from time-series pH waveform.

16. The system of claim 15, wherein the processor further determines whether the22 of 25NCH -032911 WO ORDswallowing event is associated with a reflux event based on the pH value.

17. A non-transitory computer-readable medium storing computer-executable instructions that when executed by a computer, which includes at least a processor, cause the computer to:receive a plurality of impedance waveforms indicative of swallowing activity along a nasogastric tract over a period of time;process the plurality of impedance waveforms as time-series data to identify a first waveform indicative of swallowing activity at a pharynx along the nasogastric tract and a second waveform indicative of swallowing activity between the pharynx and a proximal end of an esophagus along the nasogastric tract;identify a first series of candidate impedance drops in the first waveform and a second series of candidate impedance drops in the second waveform based on an extracted periodic signal feature; anddetect a swallowing event upon identifying contemporaneous drops in impedance based on the first series of candidate impedance drops and the second series of candidate impedance drops.

18. The non-transitory computer-readable medium of claim 17, wherein the plurality of impedance waveforms is received from a plurality of sensors positioned along the nasogastric tract, wherein the first waveform is received from a first sensor positioned at the pharynx, and the second waveform is received from a second sensor positioned between the pharynx and the proximal end of the esophagus.

19. The non-transitory computer-readable medium of claim 17, wherein the first series of candidate impedance drops are identified based on a trough amplitude of each wave in the first waveform and the second series of candidate impedance drops are identified based on a trough amplitude of each wave in the second waveform.

20. The non-transitory computer-readable medium of claim 19, wherein detecting contemporaneous drops in impedance includes determining the trough amplitude of23 of 25NCH -032911 WO ORDeach wave in the first series of candidate impedance drops whether the trough amplitude of each wave in the second series of candidate impedance drops occurs within a predetermined time window.24 of 25NCH -032911 WO ORD