Wearable devices and methods for measurement of fetal movements

Abstract

Systems and methods for acquiring and processing data collected by a wearable device for monitoring fetal movements. The wearable device is wearable by the user. A sensor assembly includes at least one of Electrical Resistance Tomography (ERT) or Electrical Impedance Tomography (EIT) electrodes. The sensor assembly detects fetal movements by measuring changes in resistance or impedance. An electronics unit transmits current through the electrodes and collects measurements. A processing unit determines and displays a status of the fetus accordingly. The present disclosure offers high-resolution, continuous tracking in a comfortable design, providing detailed insights into fetal activity and health for comprehensive prenatal monitoring.

Description

WEARABLE DEVICES AND METHODS FOR MEASUREMENT OF FETALMOVEMENTSCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] The present application claims priority to United States provisional patent application no. 63 / 687,192, filed on August 26, 2024, and entitled, “WEARABLE DEVICES AND METHODS FOR MEASUREMENT OF FETAL MOVEMENTS,” the entirety of which is hereby incorporated by reference herein.TECHNICAL FIELD

[0002] The present disclosure is directed to devices and methods designed to detect and measure movements of a fetus in a pregnant individual.BACKGROUND

[0003] Pregnancy is a complex medical journey that brings both joy and anxiety, particularly as expectant individuals face a myriad of health challenges. With the average maternal age rising, the incidence of pregnancy complications has increased. Despite advancements in many areas of medicine, pregnancy health technology has seen little innovation since the introduction of the ultrasound. Traditional practices, such as keeping a paper log of pregnancy details and manually counting fetal movements in the third trimester, remain prevalent. This method, known as “kick counting”, involves tracking fetal movements daily to ensure at least ten movements are felt within two hours, as recommended by the American College of Obstetricians and Gynecologists. However, these guidelines vary internationally and continue to evolve as new research emerges. The Society of Obstetricians and Gynaecologists of Canada recommends that all pregnant individuals regularly monitor fetal movements starting at 26 weeks gestation and if a reduction of fetal movements is identified, present to their care provider or local obstetrical unit immediately for further evaluation. Research from institutions like the Centre of Research Excellence in Stillbirth further advises expectant individuals to monitor their baby’s unique movement patterns in the womb as a baseline for normal activity and seek medical attention if a change is observed.

[0004] Kick counting has proven effective in identifying fetal distress and improving health outcomes for both expectant individuals and newborn babies, serving as an early warning signfor complications such as impaired fetal growth, preterm delivery, or stillbirth. However, the current practice is often subjective, time-consuming, and anxiety -inducing, especially for firsttime or high-risk expectant individuals. Manual tracking methods, whether by memory, chart, handheld clicker, or mobile app, are not only cumbersome but also lack the ability to provide continuous, objective monitoring throughout the day.SUMMARY

[0005] According to a first aspect, there is provided a system; the system comprises: a wearable device that is wearable by a user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Resistance Tomography (ERT) electrodes and a conductive film electrically connected to the plurality of ERT electrodes, wherein the conductive film is positioned over at least a portion of an anterior abdomen of the user when the wearable device is worn by the user and generates one or more electrical signals in response to a movement of the fetus on a surface of the anterior abdomen; and an electronics unit communicatively coupled to the sensor assembly, the electronics unit configured to transmit a current from at least one of the plurality of ERT electrodes, through the conductive film, and to at least another one of the plurality of ERT electrodes such that a measurement associated with a deformation of the conductive film is collected.

[0006] In an embodiment, the system may further comprise a processing unit communicatively coupled to the electronics unit, the processing unit determining a status of the fetus using the measurement in response to receiving the measurement.

[0007] In an embodiment, the processing unit, upon receiving the measurement, may construct a two-dimensional (2D) image of the fetus by processing the measurement using tomography calculations and algorithms.

[0008] In an embodiment, the processing unit may determine the status of the fetus by comparing the measurement with a baseline dataset, the baseline dataset being obtained from historical measurements, predefined standards, or generated by a pre-trained machine learning model.

[0009] In an embodiment, the processing unit may comprise at least one of a Graphics Processing Unit (GPU), a Neural Processing Unit (NPU), an Application-Specific Integrated Circuit (ASIC), or a Field-Programmable Gate Array (FPGA).

[0010] In an embodiment, the electronics unit may comprise a plurality of multiplexers electrically connected to the sensor assembly, a current source electrically connected to at least one of the plurality of multiplexers, and a signal receiver electrically connected to at least another one of the plurality of multiplexers.

[0011] In an embodiment, the measurement may comprise at least one of count, time, strength, frequency, duration, location, or area size of the movement of the fetus.

[0012] In an embodiment, the plurality of ERT electrodes may delimit a sensing area equal to or smaller than the conductive fdm, and the plurality of ERT electrodes are located on a periphery of the sensing area.

[0013] In an embodiment, the conductive fdm may be made of a piezoresistive material.

[0014] In an embodiment, the conductive fdm may have a lattice structure electrically connecting at least some of the plurality of ERT electrodes to each other.

[0015] In an embodiment, the plurality of ERT electrodes may be interconnected by a plurality of connections within the periphery of the lattice structure.

[0016] In an embodiment, the plurality of connections may be strips of conductive material or are cut from a piece of conductive material.

[0017] In an embodiment, the wearable device may be attachable to a garment wearable by the user.

[0018] In an embodiment, the wearable device may be a band wearable by the user, and the sensor assembly and the electronics unit may be embedded into the band.

[0019] In an embodiment, wherein at least one of a size or a position of the sensing area may be calibrated relative to the anterior abdomen of the user by receiving user input through pressing on the sensing area or tracing across the sensing area, such that the processing unit identifies an anatomical landmark or measures an overall size of the anterior abdomen when the wearable device is worn by the user.

[0020] In an embodiment, a sensing area of the sensor assembly may be divided into a plurality of regions of interest (ROIs), and the processing unit is capable of identifying a body part of the fetus using the measurement and the plurality of ROIs.

[0021] In an embodiment, the system may further comprise at least one of an accelerometer for detecting an acceleration value or an inertial measurement unit (IMU) for detecting an absolute orientation value or an angular velocity value, and the processing unit may be configured to determine if the measurement is unrelated to the fetus using the acceleration value or the inertia value.

[0022] In an embodiment, the band may further comprise a peripheral frame surrounding the conductive film and an adjustable side-strap coupled to the peripheral frame such that the conductive film is mechanically isolated from the adjustable side-strap by the peripheral frame.

[0023] In an embodiment, the peripheral frame may be less elastic than the adjustable sidestrap and the conductive film may be more elastic than the adjustable side-strap.

[0024] In an embodiment, the processing unit may be further configured to utilize a machine learning model trained on historical or real-time measurements to predict a position of the fetus within the anterior abdomen, classify a type of the movement of the fetus detected by the sensor assembly, or differentiate between a true fetal movement and an unrelated deformation of the conductive film.

[0025] In an embodiment, the machine learning model may comprise at least one of a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short- Term Memory network (LSTM), a Decision Tree, or a Random Forest.

[0026] According to a second aspect, there is provided a system; the system comprises: a wearable device that is wearable by a user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Impedance Tomography (EIT) electrodes for sensing an impedance change within the abdomen, wherein at least some of the plurality of EIT electrodes contact a skin of the user and surround a circumference of a torso of the user when the wearable device is worn by the user; and an electronics unit communicatively coupled to the sensor assembly, the electronics unit configured to transmit a current from at least one of the plurality of EIT electrodes, through the abdomen of the user, and to at least another one of the plurality of EIT electrodes such that a measurement associated with an impedance change within the abdomen is collected.

[0027] In an embodiment, the system may further comprise a processing unit communicatively coupled to the electronics unit, the processing unit determining a status of the fetus using the measurement in response to receiving the measurement.

[0028] In an embodiment, the processing unit may construct a 2-dimensional (2D) or 3- dimensional (3D) image of the fetus using the measurement in response to receiving the measurement.

[0029] In an embodiment, the processing unit may determine the status of the fetus by comparing the measurement with a baseline dataset, the baseline dataset being obtained from historical measurements, predefined standards, or generated by a pre-trained machine learning model.

[0030] In an embodiment, the processing unit may comprise at least one of a Graphics Processing Unit (GPU), a Neural Processing Unit (NPU), an Application-Specific Integrated Circuit (ASIC), or a Field-Programmable Gate Array (FPGA).

[0031] In an embodiment, the electronics unit may comprise a plurality of multiplexers electrically connected to the sensor assembly, a current source electrically connected to at least one of the plurality of multiplexers, and a signal receiver electrically connected to at least another one of the plurality of multiplexers.

[0032] In an embodiment, at least some of the plurality of EIT electrodes may surround an anterior abdomen of the user when the wearable device is worn by the user.

[0033] In an embodiment, the sensor assembly may further comprise one or more belts for connecting the plurality of EIT electrodes to the electronics unit.

[0034] In an embodiment, the current may be an alternating current (AC).

[0035] In an embodiment, the wearable device may be attachable to a garment wearable by the user.

[0036] In an embodiment, the wearable device may be a band wearable by the user, and the sensor assembly and the electronics unit may be embedded into the band.

[0037] According to a third aspect, there is provided a method for processing data collected by a wearable device, the wearable device comprising: a wearable device that is wearable by auser or attachable to a garment that is wearable by the user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Resistance Tomography (ERT) electrodes and a conductive film electrically connected to the plurality of ERT electrodes, wherein the conductive film is positioned over at least a portion of an anterior abdomen of the user when the wearable device is worn by the user and generates one or more electrical signals in response to a movement of the fetus on a surface of the anterior abdomen. The method comprises: transmitting a current from at least one of the plurality of ERT electrodes, through the conductive film, and to at least another one of the plurality of ERT electrodes to collect a measurement associated with a deformation of the conductive film as at least one variable; and determining a status of the fetus using the at least one variable.

[0038] In an embodiment, the method may further comprise, upon receiving the measurement, constructing a two-dimensional (2D) image of the fetus by processing the measurement using tomography calculations and algorithms.

[0039] In an embodiment, the variable may further comprise a maternal metric associated with physiological data or psychological status of the user.

[0040] In an embodiment, the measurement may comprise at least one of count, time, strength, frequency, duration, location, or area size of the movement of the fetus.

[0041] In an embodiment, determining the status of the fetus may comprise determining the status of the fetus by comparing the variable with a baseline dataset, the baseline dataset being obtained from historical variables or predefined standards, or generated by a pre-trained machine learning model.

[0042] In an embodiment, the method may further comprise identifying a pattern of the variable over a period of time.

[0043] In an embodiment, the variable may comprise at least two variables, at least one of the at least two variables being the measurement, the method further comprising identifying a correlation between the at least two variables.

[0044] In an embodiment, determining the status of the fetus may comprise determining the status of the fetus by comparing at least one of the variable, the pattern, or the correlation witha baseline dataset, and the baseline dataset may be obtained from historical variables or predefined standards, or generated by a pre-trained machine learning model.

[0045] In an embodiment, determining the status of the fetus may comprise determining a rate of change of at least one of the variable, the pattern, or the correlation with the baseline dataset over time, to identify an anomaly in response to the rate of change exceeding a predetermined threshold.

[0046] In an embodiment, the method may further comprise: dividing a sensing area of the sensor assembly into a plurality of regions of interest (ROIs); identifying at least one body part of the fetus using the measurement and the plurality of ROIs; and determining a position of the fetus using the identified body part and the plurality of ROIs.

[0047] In an embodiment, the method may further comprise calibrating at least one of a size or a position of the sensing area relative to the anterior abdomen of the user in response to the user pressing on the sensing area when the wearable device is worn by the user.

[0048] In an embodiment, the method may further comprise determining a change of a size of the anterior abdomen of the user to monitor a growth of the fetus.

[0049] In an embodiment, the method may further comprise: obtaining an acceleration value from an accelerometer or an absolute orientation value or an angular velocity value from an inertial measurement unit (IMU); determining if the measurement is unrelated to the fetus using the acceleration value or the inertial value; and in response to determining the measurement is unrelated to the fetus, removing the measurement.

[0050] In an embodiment, the method may further comprise utilizing a machine learning model trained on historical or real-time measurements to predict a position of the fetus within the anterior abdomen, classify a type of the movement of the fetus detected by the sensor assembly, or differentiate between a true fetal movement and an unrelated deformation of the conductive fdm.

[0051] In an embodiment, the machine learning model may comprise at least one of a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short- Term Memory network (LSTM), a Decision Tree, or a Random Forest.

[0052] In an embodiment, the wearable device may be attachable to a garment wearable by the user.

[0053] In an embodiment, the wearable device may be a band wearable by the user, and the sensor assembly may be embedded into the band.

[0054] According to a fourth aspect, there is provided a method for processing data collected by a wearable device, the wearable device comprising: a wearable device that is wearable by a user or attachable to a garment that is wearable by the user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Impedance Tomography (EIT) electrodes for sensing an impedance change within the abdomen, wherein at least some of the plurality of EIT electrodes contact a skin of the user and surround a circumference of a torso of the user when the wearable device is worn by the user. The method comprises: transmitting a current from at least one of the plurality of EIT electrodes, through the abdomen of the user, and to at least another one of the plurality of EIT electrodes to collect a measurement associated with an impedance change within the abdomen as a variable; and determining a status of the fetus using the variable.

[0055] In an embodiment, the method may further comprise constructing a 3-dimensional (3D) model of the fetus using the measurement.

[0056] In an embodiment, the variable may further comprise a maternal metric associated with physiological data or psychological status of the user.

[0057] In an embodiment, determining the status of the fetus may comprise determining the status of the fetus by comparing the variable with a baseline dataset, the baseline dataset being obtained from historical variables or predefined standards, or generated by a pre-trained machine learning model.

[0058] In an embodiment, the method may further comprise identifying a pattern of the variable over a period of time.

[0059] In an embodiment, the variable may comprise at least two variables, at least one of the at least two variables being the measurement, the method further comprising identifying a correlation between the at least two variables.

[0060] In an embodiment, determining the status of the fetus may comprise determining the status of the fetus by comparing at least one of the variable, the pattern, or the correlation with a baseline dataset, and wherein the baseline dataset is obtained from historical variables or predefined standards, or generated by a pre-trained machine learning model.

[0061] In an embodiment, determining the status of the fetus may comprise determining a rate of change of at least one of the variables, the pattern, or the correlation with the baseline dataset over time, to identify an anomaly in response to the rate of change exceeding a predetermined threshold.

[0062] In an embodiment, the wearable device may be attachable to a garment wearable by the user.

[0063] In an embodiment, the wearable device may be a band wearable by the user, and the sensor assembly may be embedded into the band.

[0064] This summary does not necessarily describe the full scope of all aspects. Other aspects, features and advantages will become apparent to those of ordinary skill in the art upon review of the following description of specific embodiments.BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In the accompanying drawings, which illustrate one or more example embodiments:

[0066] FIG. 1 is an example wearable device based on ERT technology when it is worn by a pregnant individual, according to an embodiment;

[0067] FIGS. 2A and 2B are example configurations of the sensor assembly based on ERT technology, according to an embodiment;

[0068] FIG. 3 is an example configuration of a lattice structure representing the sensor layout used in an example embodiment of the sensor assembly;

[0069] FIGS. 4A to 4D are alternative configurations of the lattice structure, according to various embodiments;

[0070] FIG. 5 is an example of a position map for the sensed measurements acquired by the sensor assembly to be interpreted as positional information of the fetus, according to an embodiment;

[0071] FIG. 6 is another example of a position map for the sensed measurements acquired by the sensor assembly to be interpreted as positional information of the fetus, according to an embodiment;

[0072] FIG. 7 is an example wearable device based on EIT technology when it is worn by a pregnant individual, according to an embodiment;

[0073] FIG. 8 is a schematic view of a system for an ERT wearable device, according to an embodiment.

[0074] FIG. 9 is an example wearable device with a peripheral frame, according to an embodiment.

[0075] FIG. 10 is an example method for processing the collected measurements acquired by a wearable device relying on EIT or ERT technology, according to an embodiment.

[0076] FIG. 11 is an example patchwork understanding of a fetus’s activity over a 24-hour period, according to an embodiment.

[0077] FIG.12 is an example reconstruction image of the conductivity distribution to create a two-dimensional model of the fetus and its position in the womb, according to an embodiment.DETAILED DESCRIPTION

[0078] The present disclosure provides a continuous and high-resolution sensing modality for monitoring fetal movements outside of clinical settings. By non-invasively observing and characterizing fetal movements, it enables user-specific insights into the health of both the expectant individual and their fetus. The data may be collected over extended periods of use, allowing for the creation of comprehensive and individualized health baselines for both the expectant individual and the fetus. Trends and changes in these baselines can be analyzed to provide preemptive warnings and insights about the health of the fetus and the expectant individual.

[0079] The present disclosure is directed at various systems and methods for measuring fetal movements. For example, it may accurately measure various parameters of fetal movements, including count, strength, frequency, duration, location, or area of effect. This information may be used to determine a unique baseline pattern for each user, which can be critical for identifying deviations that may indicate maternal or fetal health issues. Additionally, in at least some embodiments the system can detect the fetus’s position in the womb based on a map of movements, which is for determining whether repositioning should be considered for safe and optimal delivery; however, the primary purpose of distinguishing between different body parts (e.g., a foot versus an arm) is to enable characterization of normal ranges of movement values for those body parts, thereby improving baseline tracking of fetal activity. For instance, the system can characterize fetal activity further, such as by distinguishing between a kick with the foot and a stretch of the arm to determine a reference signal range for each type of movement.

[0080] The disclosed technology offers significant advantages by eliminating the uncertainty and lack of individual user requirements currently present in fetal movement monitoring. By leveraging Electrical Resistance Tomography (ERT) and Electrical Impedance Tomography (EIT) technologies, the system provides a comfortable form factor, simplicity, and feasibility for continuous home monitoring. Tomography, in general, refers to imaging by sections or slices through the reconstruction of internal properties based on boundary measurements. This category of imaging includes techniques that infer internal spatial distributions of physical quantities, such as electrical resistivity or impedance, by applying excitation signals at theperiphery and solving an inverse problem based on measured responses. ERT and EIT both allow for non-invasive, real-time imaging of the fetus and its activities, though they operate differently: ERT is resistivity-based and uses direct current, while EIT is impedance-based and uses alternating current.

[0081] ERT measures resistivity changes using direct current. In ERT, a set of electrodes is arranged around a region of interest, and small direct currents are injected across selected pairs of electrodes while voltage differences are measured across other pairs. These voltage measurements are used to reconstruct a spatial map of resistivity within the enclosed region. ERT is advantageous for monitoring lower-resistance objects and captures real-time images of the fetus and its activities by measuring resistivity changes on a sheet that deforms due to movements of the fetus. ERT becomes an imaging technology in this context by indirectly capturing fetal position through the deformation caused by the fetus’s movements on the abdominal surface.

[0082] EIT provides a non-invasive approach to visualizing and monitoring the internal electrical conductivity distribution of tissues and organs. EIT applies a small electrical current at one or more electrodes on the body's surface and measures the resulting voltage changes at different electrodes. The term “impedance” in this context refers to the combination of resistance and reactance experienced by alternating current. These measurements are used to reconstruct a two-dimensional image or, with multiple sets of electrodes, a three-dimensional image of the internal conductivity distribution can be reconstructed. This capability allows for capturing real-time images of the fetus and its activities by resolving spatial variations in tissue impedance.

[0083] The disclosed system can incorporate ERT, EIT, or both technologies, leveraging their respective strengths. ERT provides indirect imaging by detecting resistivity changes on a measurement plane that lies externally on the anterior abdomen and does not physically intersect the fetus. In contrast, EIT enables direct imaging by applying alternating currents through belts of electrodes placed around the abdomen, producing conductivity maps of the tissues that include the fetus. EIT may combine measurements from multiple electrode belts or positions to approximate internal structures and orientations. A combined ERT and EIT system may also be used to offer a comprehensive solution for continuous fetal and maternal health monitoring.

[0084] FIG. 1 illustrates an example wearable device 100 based on ERT technology when it is worn by a pregnant individual. The wearable device 100 is designed to be worn about the torso of a user, such that a portion of the wearable device 100 covers a portion of the anterior abdomen of the user. It should be understood that the wearable device 100 may be constructed to be directly or indirectly attachable to the user. For example, it may be a band or strap wearable by the user or it is an attachment or insert that is attachable to a garment wearable by the user.

[0085] As shown in FIG. 1, the wearable device 100 may be or may comprise a band 110 that is wearable by the user around their torso 10, particularly covering the anterior abdomen. The band 110 is sized and is manufactured from materials (e.g. elastic materials) chosen to comfortably and securely hold a sensor assembly against the user's abdomen. The material of the band 110 can be a single-layer or multi-layered fabric, potentially formed from a multilayer laminate, ensuring flexibility and durability for prolonged use.

[0086] Connected to the band 110 is the sensor assembly that includes a conductive film 120 and a plurality of ERT electrodes 130. The sensor assembly may be detachably connected to the band 110 (such that the band 110 can be used as a standalone garment) or fixed to the band 110. The conductive film 120 is positioned over the anterior abdomen when the wearable device 100 is worn by the user. The conductive film 120 is used to generate electrical signals in response to a deformation of the surface of the conductive film 120, such as due to the movement of the fetus resulting in movement on the surface of the anterior abdomen. For example, the conductive film 120 may be a piezoresistive sensor layer, which means it changes its resistivity in response to deformation. The term “conductive film” broadly encompasses various forms of deformable materials suitable for such electrical signal generation, including but not limited to a film, a membrane, a layer of conductive material, or any other substrate that can conduct electricity and respond to physical deformation. This deformation mimics the skin’s movement on the pregnant abdomen during fetal activity, translating deformation and strain changes into resistivity or conductivity changes that the sensor assembly can detect.

[0087] The plurality of ERT electrodes 130 may be placed along the edges of the conductive film 120 to surround the anterior abdomen of the user (for example, the user’s navel). When the wearable device 100 is in use, an electrical current in the form of a direct current (DC) is transmitted from at least one of the ERT electrodes 130, through the conductive film 120, to at least another ERT electrode 130. This current transmission allows the system to measure theresistivity changes caused by fetal movements. The ERT electrodes can be made from any conductive material, such as conductive ink, cut-out fabric, or conductive threads, ensuring adaptability to various designs and user needs. By using the ERT technology, the sensing plane lies externally on the anterior abdomen and does not intersect the fetus, ensuring non-invasive monitoring. The electrical current area is confined to the sensor layer and does not pass through the expectant individual’s abdomen, which enhances safety and comfort for the user. It should be understood that the sensor assembly is driven by an electronics unit 160 (more details described in respect of FIG. 8) that may be integrated or detachably connected to the band 110.

[0088] The wearable device 100 may additionally include a processing unit 170 (more details described in respect of FIG. 8) communicatively coupled to the electronics unit 160. The processing unit 170 can be used to receive the measurements associated with the deformation of the conductive film 120. The processing unit 170 processes these measurements to determine the status of the fetus. This determination can include analyzing various parameters such as the force, duration, time, frequency, location, and the area of effect of the fetal movements to determine a status of the fetus or to establish a baseline pattern unique to each user. By continuously monitoring and analyzing these patterns, the system can provide preemptive warnings about the health of both the fetus and the expectant individual, making it a valuable tool for pregnancy health management outside of clinical settings.

[0089] The processing unit 170 according to various embodiments described herein may be implemented in different forms depending on the design and requirements of the wearable device 100. In some embodiments, the processing unit 170 may be a processor placed directly on or fixed to the band 110. This configuration allows for immediate and localized data processing, providing real-time analysis and feedback to the user without the need for external devices. Alternatively, the processing unit 170 may be any type of suitable processing unit, such as a mobile phone used by the user. In this embodiment, the electronics unit 160 communicates with the mobile phone, either wired or wirelessly, which processes the data collected by the sensor assembly. This setup leverages the computational power and connectivity of modem personal devices to perform complex data analysis and visualization, offering a user-friendly interface for monitoring fetal health. In another embodiment, the processing unit 170 may be part of a remote server. In this configuration, the data collected by the sensor assembly is transmitted to the server for processing. The remote server can utilize algorithms and machine learning models to analyze the data, providing detailed insights andalerts about the fetal movements and health status. This approach enables centralized data processing, which can be particularly useful for aggregating data from multiple users to build comprehensive datasets for research and improved predictive modeling.

[0090] FIG. 2A and FIG. 2B illustrate example sensor assemblies based on ERT technology, highlighting different configurations of the conductive film and electrode placement.

[0091] In FIG. 2A, the sensor assembly comprises a conductive film 120, made of piezoresistive material, with a plurality of ERT electrodes 130 positioned around or within the edge of the conductive film 120. The arrangement of ERT electrodes 130 ensures comprehensive coverage and accurate measurement of resistivity changes across a certain area of the surface of the conductive film 120. The ERT electrodes 130 may be positioned along the periphery of the conductive film 120, for example, in which case the sensing area is substantially the entirety of the conductive film 120. Alternatively, the ERT electrodes 130 may be positioned within the periphery of the conductive film 120, in which case portions of the conductive film 120 may extend outside the area delineated by the ERT electrodes 130. In that case, only the portion of the conductive film 120 bounded by the ERT electrodes 130 is a sensing area. The configuration allows for the detection of fetal movements by measuring the changes in electrical resistance caused by the deformation of the piezoresistive material. This setup effectively captures measurements such as strength, duration, time, frequency, location, and area size of the movements. The measurement can be done by allowing a current to flow from one or more electrodes through the conductive film 120 to the other electrodes. It should be understood that although FIG. 2A shows that the ERT electrodes 130 are disposed around the periphery of the conductive film 120, any number of electrodes may be positioned within the periphery of the conductive film 120.

[0092] In FIG. 2B, a variation of the sensor assembly is illustrated where the conductive film 120 is shaped as an annulus, resembling a doughnut, with an additional second plurality of ERT electrodes 131 located in an inner circle. This annulus configuration of the ERT electrodes on the conductive film 120 provides a uniform distribution of the electrodes, mitigating the decay in resolution near the center by reducing the total distance from some electrodes to others. Additionally, it may be advantageous to achieve more accurate readings toward the center. The plurality of ERT electrodes 130 forming a larger outer circle and the second plurality of ERT electrodes 131 forming a smaller inner circle work together to detect subtle changes inresistivity. More particularly, measurements can similarly be conducted by allowing a current to flow from one or more electrodes through the conductive fdm 120 to the other electrodes.

[0093] FIG. 3 illustrates an example lattice structure of the sensor assembly. The lattice structure is designed to enhance the performance of the ERT system by strategically arranging the conductive paths and electrodes.

[0094] The sensor assembly in FIG. 3 has a plurality of ERT electrodes 130 located on the periphery of the conductive fdm 120. These electrodes are interconnected by a plurality of connections 121 between the ERT electrodes 130. The connections 121 may be strips of conductive material, printed conductive ink, or segments cut from a piece of conductive material, forming a web-like lattice that links the ERT electrodes 130. This lattice structure allows the sensor to be more sensitive to certain movements by controlling the stretchable areas of the conductive fdm. For example, long, thin strips of sensing material between electrodes can maximize the sensitivity in those areas by responding more significantly to deformations along their length rather than their width. The configuration also increases current density by reducing the amount of resistive material between electrodes, leading to a greater potential change when pressure is applied. This higher current density results in a more sensitive sensor system. It should be understood that while the ERT electrodes 130 define the periphery of a sensing area, they do not necessarily constitute the periphery of the conductive film 120. In other words, the ERT electrodes may be positioned within the bounds of the conductive film 120.

[0095] Furthermore, the lattice structure allows for optimized current paths. The connections 121 are designed to facilitate easy current flow in specific directions, enhancing the sensor’s sensitivity in targeted areas. This means that during ERT measurement, the current flows along paths between the electrodes and the lattice manipulates the length of the paths for a better current distribution. In some cases, the paths may be curved, forming arcs connecting the electrodes. These curved paths result from the strategic placement and connection of electrodes within the lattice. The width or thickness of the paths may vary, which impacts the current flow. For example, thin arcs of sensing material connecting each electrode may impact the sensor’s performance by enhancing the flow of electrical current. The lattice configuration can be adjusted to fit the shape of the pregnant individual's growing abdomen and withstand the various stresses that will be applied. By creating more current paths within the lattice oradjusting its shape, the sensor can be optimized for different applications and environments, ensuring accurate and reliable monitoring of fetal movements.

[0096] The ERT electrodes 130 may or may not be flexible, depending on the specific application and design considerations. They can be rigid or semi-rigid, providing stable and consistent electrical connections. For instance, The ERT electrodes 130 can be made of conductive materials such as metals, conductive polymers, or composites. These electrodes do not necessarily need to be stretchable and can be small in size, positioned on the periphery of the sensing area to ensure effective signal detection. The choice of electrode material and flexibility can be adjusted to suit the desired performance and comfort for the user.

[0097] FIGS. 4A to 4D illustrate alternative configurations of the lattice structure, showing different arrangements of the conductive paths and electrodes to adjust sensitivity, particularly in the central portion of the sensing area.

[0098] In FIG. 4A, the sensor assembly comprises a plurality of ERT electrodes 130 arranged in a circular configuration with additional conductive paths forming a square lattice structure. This design increases the density of the conductive paths near the center of the sensing area, thereby improving the sensitivity to movements in this region. FIG. 4B shows a radial configuration where the ERT electrodes 130 are connected by concentric circular paths. This arrangement enhances the sensor's ability to detect movements across different radial distances from the center, making it particularly effective for capturing subtle movements near the navel. In FIG. 4C, the sensor assembly features conductive paths forming a square lattice structure. This configuration is similar to that of FIG. 4A, but more or longer paths are deployed among the ERT electrodes 130 for higher resolution because more areas can be used to sense the fetal movement. In FIG. 4D, the sensor assembly features conductive paths forming a concentric circular lattice structure. This configuration is similar to that of FIG. 4B, but more or longer paths are deployed among the ERT electrodes 130 for higher resolution.

[0099] It should be appreciated that while some of these figures show the electrodes in a circular configuration, the shape of the electrode arrangement is not limited to circles. The ERT electrodes 130 can also be arranged in other shapes, such as rectangles, depending on the specific requirements of the application. In addition, the number of electrodes is not limited to eight as shown in FIGS. 3 and 4A-4D. The sensor assembly may include more or fewer electrodes to meet different sensitivity and resolution requirements. While having more pathslocated in the central region of the sensing area is advantageous due to a more sensitive response closer to the central portion of the user’s anterior abdomen (such as her navel), other distributions are possible. These variations ensure that the sensor assembly can be customized for optimal performance in various use cases, providing reliable and accurate monitoring of fetal movement.

[0100] FIG. 5 illustrates an example of a position map for the sensed measurements by the sensor assembly to be interpreted as positional information of the fetus, particularly when the fetus is upright and facing forward. The sensing area of the sensor assembly is divided into a plurality of regions of interest (ROIs), which helps in identifying the position and specific body parts of the fetus, such as arms and head. In this context, a “map of movements” refers to a structured representation of detected activity across the sensing area, where each ROI may be assigned a spatial meaning. By correlating repeated patterns of movement in particular ROIs with known fetal anatomy, the system can distinguish between different body parts and thus estimate the overall position of the fetus. In addition, such distinctions enable the system to establish characteristic ranges of fetal activity for different body parts, which can be tracked as baselines for health monitoring.

[0101] In the context of ERT, the sensing plane does not intersect the fetus, making it an indirect measurement method. However, the technology can interpret the data to determine the probability that the fetus is in a certain position based on current and past available fetal activity data. For instance, a small activation area with a strong signal and frequent activations may indicate the locations of smaller body parts like hands and feet. Conversely, a large activation area with a weak signal and infrequent activations, especially towards the edge of the sensor, may suggest the presence of the head or buttocks. A full sensor activation with consistent strength and frequency may indicate that the fetus is inverted with its back against the anterior abdomen. Thus, in the context herein, the map of movements does not refer to a literal anatomical image, but rather a probabilistic mapping of how different signal strengths and frequencies correspond to expected fetal anatomy. This probabilistic map can therefore be used not only to infer likely positions, but also to quantify movement patterns for particular body parts as part of baseline characterization.

[0102] Over an extended period of time with consistent readings, the system can become more confident in estimating the fetal position. This interpretation is strengthened by understanding a baseline or previously known state of the fetus. For example, if the fetus iscurrently upright and facing the anterior abdomen, it would be expected to take some time to move to an upside-down and internally -facing position. This baseline can be used as a weight in an algorithm to determine the accuracy of the current data suggesting a specific fetal position. In effect, the baseline may act as a stabilizing reference for the map of movements, reducing ambiguity when transient or overlapping signals are detected across ROIs. The dual use of baselines, both for position inference and for expected ranges of fetal movement, provides a consistent foundation for interpreting new activations.

[0103] Knowing which part of the body creates the activation may be beneficial for characterizing a normal range of force values for that specific body part. As the fetus grows bigger and stronger with gestational aging, this information may be used to further monitor its development. To quantify this algorithm, the sensing area may be divided into octets, as shown in FIG. 5, or other multiplicities to create locations that can be compared to each other. These ROIs can be created through software mapping with specific reference points set up to orient the regions, independent of the physical sensor itself, although resolution may be improved based on the sensor structure. Accordingly, ROIs may serve as logical units of analysis that convert raw sensing signals into interpretable features, which can then be compared across time to build a reliable picture of fetal activity and position. In particular, identifying body-part- specific activations can support defining baseline movement ranges and growth-related changes, with positional interpretation serving as a complementary outcome.

[0104] In an upright fetal position, as shown in FIG. 5, left and right octets 520 are more likely to correspond to the arms or legs, while upper and lower octets 510 are more likely to correspond to the head or feet. The orientation of the fetus may vary, creating a rotation of these octets. The frequency and strength of certain stimuli occurring at specific locations can add weight to a certain prediction. For example, frequent and strong signals in the lower regions may confirm the foot’s position, while similar signals in the lateral regions may indicate arm movements. This weighting and decision-making can be facilitated by machine learning models or by a trained clinician. In this way, the map of movements can become a dynamic interpretation tool, where ROI-based activations can be continuously updated and rotated to reflect the fetus’s changing orientation. At the same time, consistent ROI activations corresponding to specific body parts can be used to refine normal ranges of movement, enabling personalized baseline development across gestation.

[0105] FIG. 6 illustrates another example of how ROIs can be divided in the software to analyze fetal movements. In this configuration, the sensing area is divided into a grid pattern, with rows labeled A, B, C, and D representing regions of less activity, and numbered regions 1 through 20 representing areas of high activity. This division allows for detailed mapping and analysis of fetal movements within specific sections of the sensor surface. While FIG. 5 demonstrates an anatomical-style mapping, FIG. 6 demonstrates a coordinate-style mapping, which can be useful for algorithmic processing. Both approaches represent alternative ways of defining ROIs for constructing the map of movements.

[0106] This method of dividing the sensing area into ROIs enables the processing unit 170 to identify specific body parts of the fetus based on the measured data and the designated regions. For example, frequent activations in certain numbered regions can indicate specific fetal movements, such as kicks or stretches, while the less active lettered regions help in establishing a reference point for normal activity levels. By continuously monitoring and analyzing the data from these ROIs, the system may provide comprehensive insights into the fetus’s position, movement patterns, and overall health, for facilitating the non-invasive prenatal monitoring. The distinction between active and less active ROIs may thus affect how the map of movements is interpreted, with high-activity regions pointing to localized fetal actions and low-activity regions serving as comparative baselines.

[0107] The user can calibrate sensor placement or alignment through various activations of the sensing area and known reference points. For example, the user may indicate the location of the belly button (from the side of the sensor facing away from the skin) by pressing on it when prompted by their interface. The user may also use their finger to trace across a middle cross section (or a grid of multiple cross sections) of the belly to ensure the intersection point is at specific points (such as a high point) of the belly. Guidelines for these tracings may be shown on the external side of the wearable to assist the user. Similarly, this calibration mechanism can be used to automatically measure the overall size of the pregnant belly (e.g., comparing the measured or calculated circumferences of the hemisphere between usages) which would directly imply fetal growth (i.e., the size of the belly will change as the fetus grows bigger). This data can be useful for further characterization of fetal movement, baseline determination, and image reconstruction. In order to create an accurate measurement of belly size changes, the user can ensure that the band is placed appropriately or properly aligned to then conduct the measurement. Calibration can therefore link the physical placementof the sensor to the logical ROIs used in the maps of FIGS. 5 and 6, so that detected activations correspond to consistent anatomical or coordinate references across different monitoring sessions.

[0108] FIG. 7 illustrates another example of wearable device 100 based on EIT technology when it is worn by a pregnant individual. The wearable device 100 is designed to be worn about the torso of a user, such that a portion of the wearable device 100 covers a portion of the abdomen of the user. It should be understood that the wearable device 100 may also be constructed such that it is attachable to a normal garment wearable by the user.

[0109] As shown in FIG. 7, the wearable device 100 may be or may comprise a sensor assembly connected to the band 110. The sensor assembly may include a first belt 141 and a second belt 151, on each of which multiple EIT electrodes are placed. A first plurality of EIT electrodes 142 may be placed onto the first belt 141, positioned adjacent to each other without contact, and the belt is positioned to surround the user’s torso, forming a ring around it. Similarly, a second plurality of EIT electrodes 152 are placed on the second belt 151 and are positioned to surround the user’s anterior abdomen (the expanded portion due to pregnancy), forming another ring. This arrangement ensures that at least some of the EIT electrodes are in direct contact with the skin of the user, enabling the measurement of impedance changes within the abdomen.

[0110] It should be understood that the sensor assembly is driven by the electronics unit160 that may be integrated or detachably connected to the band 110. The electronics unit 160 can be used to transmit a current in the form of an alternating current (AC) from at least one of the plurality of EIT electrodes, through the abdomen of the user, to at least another one of the plurality of EIT electrodes. This process allows for the collection of measurements associated with the impedance changes, which can then be used to construct an image of the fetus by the processing unit 170, such as the one used for ERT electrodes described in respect of FIG. 1. The measurement typically comprises a voltage change across a plane formed by the plurality of EIT electrodes, which is indicative of the internal conductivity distribution of the tissues. In simpler terms, the electrodes define sensing “paths” across the abdomen, and the differences in impedance along these paths may be used to reconstruct patterns of conductivity that correlate with the fetus’s shape and movement.

[0111] The use of multiple belts of EIT electrodes, such as the first plurality of EIT electrodes 142 and the second plurality of EIT electrodes 152, can enhance the capability of the system to generate detailed images of the fetus. Rather than relying on a single set of measurements, the presence of two or more belts can provide complementary perspectives, similar to taking multiple cross-sectional views of the abdomen. These views can be combined by the processing unit 170 to approximate either a two-dimensional (2D) map of conductivity or, when multiple belts are used, a volumetric or three-dimensional (3D) representation. This approach avoids the need for literal “intersecting planes” and instead relies on combining multiple measurement bands to improve resolution and anatomical interpretation.

[0112] The wearable device 100 is designed to be flexible and comfortable for continuous wear. The sensor assembly, the electronics unit 160, and the processing unit 170 may be detachably connected to or embedded into the band 110, allowing for unobtrusive monitoring. As discussed above in respect of the ERT system, the processing unit 170 may be a mobile phone used by the user, so that the electronics unit 160 communicates with the mobile phone (can be via a cable or wirelessly), which processes the data collected by the sensor assembly. Like the ERT electrodes as described above, the EIT electrodes may or may not be flexible.

[0113] EIT works by applying a small, AC current through electrodes placed around the body region of interest and measuring the resulting voltage changes. These measurements are then used to reconstruct images of the internal conductivity distribution. In practice, the result may not be a photographic image but a conductivity map that highlights differences in tissue composition and fetal boundaries. FIG. 7 is intended to illustrate how arranging multiple belts of electrodes improves the accuracy and interpretability of these conductivity maps. In this embodiment, the sensing plane intersects the fetus, allowing for direct imaging of the fetus and its activities. The electrical current travels through the pregnant abdomen, capturing realtime data that can be processed to monitor the fetus’s well-being continuously.

[0114] Although the example of FIG. 7 shows that two belts, the first belt 141 and the second belt 151, are used to position the plurality of EIT electrodes 142 and 152 respectively, it should be understood that the configuration is not limited to this specific arrangement. There can be only one belt or more than two belts depending on the design requirements and the desired resolution of the imaging. Additionally, although the example shows that the EIT electrodes are placed on either the first belt 141 or the second belt 151, the belt can be omittedin other embodiments. In such cases, the EIT electrodes can be directly attached or fixed to the band, as long as the EIT electrodes can be electrically connected to the electronics unit 160. Furthermore, while the belts are depicted in the form of rings surrounding the torso and the anterior abdomen, the belts can also be designed in a meandering or serpentine pattern to cover a larger surface area and adapt to the contour of the body. This flexibility in design ensures that the wearable device can be tailored to various user needs and anatomical differences, maintaining the functionality of continuous and accurate fetal monitoring.

[0115] The EIT electrodes 142, 152 may or may not be flexible, depending on the specific application and design considerations. They can be rigid or semi-rigid, providing stable and consistent electrical connections. For instance, the EIT electrodes 142, 152 can be made of conductive materials such as metals, conductive polymers, or composites. These electrodes do not necessarily need to be stretchable and can be small in size to ensure effective signal detection. The choice of electrode material and flexibility can be adjusted to suit the desired performance and comfort for the user.

[0116] FIG. 8 illustrates a system 800 for an ERT wearable device. The system 800 is designed to generate and process the electrical signals required to create images based on resistivity changes within the conductive film.

[0117] The system 800 includes a current source 820, which generates a direct current. The current source 820 may include a battery that provides the needed current. In one example, the current is transmitted by the current multiplexers (MUX) 831 and MUX 832 to the first set of electrodes such as 811 and 812. Then, the multiplexers can transmit the current to a second set of electrodes such as 813 and 814. This pattern can continue across all electrodes.

[0118] As the current travels through the conductive film, it encounters varying resistivities resulting from fetal movement. This results in voltage changes that are collected by a third electrode 813 and a fourth electrode 814. These electrodes measure the differential voltage changes caused by the resistivity changes within the conductive film.

[0119] The voltage multiplexers select the voltage signals measured at the electrodes to transmit to the signal receiver. In this example, the voltage signal from a third electrode 813 is sent by a first voltage MUX 833, and the voltage signal from a fourth electrode 814 is sent by a second voltage MUX 834.

[0120] A signal receiver 840 is used to receive the voltage signals from the first voltage MUX 833 and the second voltage MUX 834. This receiver processes the analog signals and converts them into a digital format suitable for further analysis. The current source 820, the signal receiver 840, and the multiplexers can be collectively referred to as the electronics unit 160. In some embodiments, the electronics unit 160 may be implemented as a microcontroller, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or a field- programmable gate array (FPGA), any of which may integrate current-driving and signalprocessing functionality into a single package.

[0121] The processed signal is then transmitted to the processing unit 170, which may be responsible for analyzing the data and constructing a 3D image of the fetus based on the measurements, for example. The processing unit 170 may include algorithms and machine learning models to enhance the accuracy and detail of the generated images. In some embodiments, the processing unit 170 may comprise at least one processor and a non-transitory computer-readable medium storing program code, the program code comprising instructions which, when executed by the processor, cause the processing unit to perform one or more of the measurement, analysis, classification, and image-reconstruction functions described herein.

[0122] Although FIG. 8 specifically illustrates a system based on an ERT sensor assembly, the system has a similar structure for an EIT sensor assembly. The primary difference lies in the configuration of the electrodes, where ERT electrodes encircle a conductive film as explained in respect of FIG. 1 and EIT electrodes encircle the abdomen as explained in respect of FIG. 7. In the EIT system, the current source generates an alternating current instead of a direct current, and the impedance in the body tissues are measured rather than resistivity changes in the conductive film. The fundamental components such as current multiplexers, voltage multiplexers, signal receivers, and processing unit remain similar, ensuring the system’s adaptability for both ERT and EIT technologies. In addition, if the system utilizes both the ERT and EIT sensor assemblies, the structure can be modified accordingly to accommodate both technologies. In all such implementations, at least one processor and computer-readable medium may be present to execute software routines for controlling current injection sequences, switching multiplexer states, acquiring and digitizing measurements, and processing or transmitting results.

[0123] The present disclosure may be constructed of user-friendly and comfortable materials to encourage continuous use outside of the hospital. This is beneficial for expectant individuals who need to monitor fetal movements continuously and in a non-intrusive manner.

[0124] Conductive fabrics or polymers can be used in the ERT and EIT systems. These materials can be layered into a fully textile or soft material sensor stack that integrates seamlessly with conductive or non-conductive textiles. This integration creates form-fitting, breathable, and comfortable wearables that are suitable for extended use. The lattice structures described above can be created using printed conductive ink, cut-out fabric, or conductive thread, ensuring flexibility and adaptability in the design. The lattice can be printed on, placed on top of, or embedded in non-conductive or less conductive materials. If the lattice is placed on a conductive material, it can establish connections to all or a subset of the ERT electrodes that are on the periphery of the sensing area.

[0125] The ERT and EIT systems may be in close, form-fitted contact with the user’s abdomen. To achieve this, the systems may be constructed of stretchable textiles or soft materials, ensuring that they conform to the natural shape and movements of the body. One method to secure the system to the abdomen is via an adjustable, stretchable band, similar to maternal belly bands. This ensures that the sensor assembly remains in the correct position to provide accurate measurements. Additionally, other types of sensors can be integrated into the soft sensor stack or the garment itself at any point within the surface area, enhancing the functionality and versatility of the wearable device.

[0126] FIG. 9 illustrates the wearable device 100 including a peripheral frame 910 for isolating the conductive film 120 from an adjustable side-strap 920, according to an embodiment. The wearable device 100 may be provided with an elastic material to maintain adjustability for different waist sizes and to change the tightness and fit. Since the conductive film 120 is also elastic, sewing it directly to or in series with another elastic material would cause the tension used to adjust the fit to be transferred to the conductive film 120, thereby preventing or limiting its operation. To prevent this, a fabric interface is provided as the peripheral frame 910 surrounding the conductive film 120 and connected to the adjustable sidestrap 920, so that the conductive film 120 is isolated from the adjustable side-strap 920 by the peripheral frame 910. The sensing area of the conductive film 120 may be defined by the placement of peripheral electrodes, which may be located within the film or at the interface of the frame 910 and the film 120. The frame therefore has both a mechanical role (isolation andperipheral structural support for the conductive film) and, in some designs, an indirect relationship to the sensing area through electrode placement, although the sensing area itself may be further refined in software mapping.

[0127] The term “peripheral” as used herein refers to the surrounding relationship between the frame 910 and the sensing film 120, without limiting the shape of the opening. For example, the peripheral frame is annular, as shown in FIG. 9. While FIG. 9 does not depict the electrodes, this is for clarity since the figure emphasizes the material relationships rather than electrical components. The shape associated with the frame 910 is not restricted to a literal “opening”, but rather refers to the shape defined by the interface between the frame 910 and the sensing film 120, which may be circular, oval, or irregular depending on construction. This fabric interface has a high spring constant relative to all other materials in the wearable device 100, making it effectively rigid. This rigid fabric interface allows the tension from the adjustment to bypass the conductive film 120 and be balanced throughout the adjustable sidestrap 920.

[0128] The wearable device 100 may thus comprise three parts - the adjustable sidestrap 920, the peripheral frame 910 and the conductive film 120, the adjustable side-strap being connected to the peripheral frame 910 and the peripheral frame 910 surrounding the conductive film 120. Each part has a spring constant K representing the elasticity of the material of the adjustable side-strap 920 (Ki), the peripheral frame 910 (K2), and the conductive film 120 (K3). The value of K3 for the conductive film 120 is lower than the value of Ki for the adjustable side-strap920 in order to detect fetal movement, which may be slight. On the other hand, the spring constant K2 of the material used for the peripheral frame 910 is significantly higher than Ki, being either inelastic or at least 10 times higher, making it relatively less elastic. This configuration is not limited to any particular construction method and may include sewn, bonded, or integrated textile interfaces, provided the mechanical decoupling between the sensing region and the adjustment tension is maintained. This configuration ensures that the adjustment tension (T) does not deform the conductive film 120 when the adjustable side-strap 920 is adjusted.

[0129] The electronics unit 160 may be housed in a plastic case that can be either free from or stitched to the band. In some embodiments, the electronics unit 160 can comprise a flexible printed circuit board (PCB) that does not require hard housing, further improving comfort and flexibility. This case may contain most or all of the essential components, such asthe computing core, multiplexers, wireless communicator, analog-to-digital converters (ADCs), and other related hardware. The system may also include an isolated power supply, such as a battery, or a wired connection to the user’s phone, providing multiple options for power management and data communication.

[0130] By using these materials and construction techniques, the present disclosure ensures that the wearable device is both effective in monitoring fetal movements and comfortable for continuous wear. The modular nature of the construction further enables tailoring of the geometry, materials, and layout to suit various body types and use scenarios without compromising the sensing performance. This combination of advanced technology and user-centric design makes it a valuable tool for prenatal care, allowing for continuous and accurate monitoring of fetal health in a non-clinical setting.

[0131] FIG. 10 illustrates a method 1000 for processing the collected measurements from a wearable device, such as those incorporating ERT or EIT sensor assemblies, to determine the status of a fetus. The method 1000 involves several operations to ensure comprehensive and accurate monitoring.

[0132] The initial operation 1001 involves collecting measurements from the wearable device, which include signals from the sensor assembly comprising ERT or EIT electrodes. These measurements, referred to as intrinsic variables, are associated with the movements and activities of the fetus. The measurements can include data such as count, time, strength, frequency, duration, location, or area of effect of the fetal movements. These intrinsic variables provide the primary data needed to monitor fetal activity.

[0133] Next, at 1002, the system may additionally receive maternal metrics from other integrated sensors or direct user inputs, which are classified as extrinsic variables. These extrinsic variables can include physiological data or psychological status of the user, such as maternal heart rate, mood, eating habits, or physical activities. Collecting both intrinsic and extrinsic variables allows for a comprehensive understanding of the factors influencing fetal movements and overall pregnancy health. However, it should be understood that the collection of extrinsic variables is optional.

[0134] At 1003, the system then identifies patterns (or referred to as trends) from at least one variable of the collected extrinsic or intrinsic variables. A pattern refers to the repetitive observations of a variable over a period of time. For example, the system mayidentify a consistent frequency of fetal movements during a specific time period each day or a regular strength range of fetal movements at a particular gestational age. Recognizing these patterns helps in understanding the normal behavior and activity levels of the fetus.

[0135] Once patterns are identified, at 1004, the system looks for correlations (also referred to as insights) between at least two variables, which can be from either the extrinsic variables or intrinsic variables, such as one intrinsic variable and one extrinsic variable. A correlation is an interpretation found through the relationship between two or more variables. For instance, a correlation might be found between the area size of the activation (an intrinsic variable) and the maternal heart rate (an extrinsic variable), indicating how the mother’s physiological state affects fetal movements. These correlations provide deeper insights into the interactions between maternal and fetal health.

[0136] In the next operation 1005, at least one of the identified variables, patterns, or correlations are compared with a baseline dataset to determine the rate of change. The baseline dataset can be obtained from historical variables, predefined standards, or generated by a pretrained machine learning model. This comparison helps in detecting anomalies by observing the rate of change of the variables over time. For example, a sudden decrease in the strength of fetal movements compared to the baseline may indicate potential fetal distress. By continuously monitoring these changes, the system can provide preemptive warnings about the health of both the fetus and the expectant individual.

[0137] This means that contrary to the current method of kick counting, users can access a vast data pool of fully unique “performance and growth” metrics. This data can also be sent to clinicians for evaluation if any concerns are raised. By using these operations, the present disclosure helps to ensure that the collected data may be thoroughly analyzed to provide accurate and actionable insights into fetal and maternal health. The method involves transmitting a current from at least one of the plurality of ERT or EIT electrodes, through the respective conductive film or abdomen, and collecting the measurements associated with a deformation of the conductive film or impedance change within the abdomen. The data is then used to construct two-dimensional or three-dimensional models of the fetus, providing detailed visualizations of fetal activity and position.

[0138] The system may also allow for dividing the sensing area into ROIs and identifying specific body parts of the fetus based on the measurements and ROIs, as described in respect of FIGS. 5 and 6.

[0139] Additionally, the method may differentiate between true fetal movements and unrelated activities, ensuring the accuracy of the data. The shape and pressure associated with the deformation of the ERT conductive film can be further classified, such as by a pre-trained classifier. For example, a finger or hand touching the user’s anterior abdomen can cause localized depressions or stretches in the sensing plane. These actions typically produce signals of smaller activation areas and shorter duration compared to fetal movements, which involve larger and more sustained deformations of the anterior abdomen. By analyzing the pattern and characteristics of the signal, such as the specific areas activated and the duration of the signal, the system can identify and exclude these false positives. The integration of additional sensors like accelerometers or inertial measurement units (IMUs) may further enhance this capability by providing additional context to the detected movements. If the accelerometer detects a hand movement in sync with the sensed deformation, the system may recognize this as a non-fetal movement and remove it from the dataset. By implementing these advanced data processing techniques, the present disclosure offers a robust solution for non-invasive prenatal monitoring, providing valuable insights and enhancing prenatal care.

[0140] In further embodiments, other secondary sensors may be integrated to provide additional datapoints. Examples include wearable ultrasound, optical sensors, or other physiological sensors that can capture motion, tissue displacement, or body-surface characteristics complementary to ERT / EIT measurements. These secondary sensors may be advantageous in that they can additionally enhance robustness.

[0141] Additionally, the present disclosure may leverage machine learning (ML) models to enhance the accuracy and functionality of the fetal monitoring system. ML models can be instrumental in identifying the position of the fetus, determining the likelihood of a movement being contributed by a particular part of the fetus (e.g., the left arm), classifying types of deformations, and more.

[0142] The use of ML models begins with the collection of large datasets comprising various measurements and metrics. These datasets include at least one of historical data, predefined standards, or real-time measurements. The processing unit can utilize these datasetsto train ML models, enabling the system to make accurate predictions and classifications based on new data inputs. Training the ML models involves using historical or ongoing measurements and user inputs to create a comprehensive dataset that the models can learn from. This training process helps the models recognize patterns and correlations in the data, improving their accuracy over time. In some embodiments, ML models may also incorporate datapoints from secondary sensors, such as IMUs or wearable ultrasound, thereby strengthening the confidence of positional classification.

[0143] One application of ML in this system is identifying the position of the fetus from the measurements collected by the ERT sensor assembly. By analyzing patterns in the collected measurements, the ML models can determine the probable position of the fetus within the abdomen. For instance, a model can learn from the relationship between the areas of activation on the sensor surface and the corresponding movements (such as a kick) to predict whether the fetus is in an upright position, facing the anterior abdomen, or in another orientation.

[0144] ML models can also assess the likelihood of a movement being contributed by a specific part of the fetus. For example, the system can be trained to recognize the characteristic signals produced by movements of the left arm, right leg, or head considering the identified position or orientation of the fetus as well as the division of the sensing area of the ERT conductive film. This is achieved by feeding the model with labeled data where specific movements have been previously identified and categorized. Over time, the model improves its accuracy in associating certain signal patterns with particular fetal body parts.

[0145] Classifying the types of deformations detected by the sensor assembly may be another area where ML models prove beneficial. The system can distinguish between different kinds of deformations, such as those caused by fetal kicks, stretches, or shifts, or differentiate between true fetal movements and unrelated activities. Secondary sensors may again be used in these classifications to disambiguate between external, maternal-origin activities and true fetal-origin events. By classifying these movements, the system provides more detailed insights into fetal activity, allowing for better monitoring of fetal health and development.

[0146] Several ML models may be effective in these applications. Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Long Short-Term Memorynetworks (LSTMs), Decision Trees, Random Forests are some of the potential models that can be utilized.

[0147] Furthermore, unsupervised learning models, such as clustering algorithms, may be utilized to identify previously unknown patterns in the data. These models can group similar signals together, helping to uncover new insights into fetal movements that may not be immediately apparent through traditional analysis methods. For example, the use of unsupervised learning may be advantageous when applied to multimodal data collected from ERT / EIT in combination with optional secondary sensors, as the clustering can reveal new correlations between sensor modalities.

[0148] By integrating these ML techniques, the present disclosure provides a powerful tool for non-invasive prenatal monitoring. The use of ML models enhances the system’s ability to deliver accurate, detailed, and actionable insights into fetal health and activity. This, in turn, empowers expectant individuals and clinicians with the information needed to ensure optimal prenatal care and timely intervention when necessary.

[0149] For example, there are a few different models that can be utilized in the disclosure for the purpose of fetal detection or anomaly identification:• 3D Reconstruction Imaging Model: The goal of this model is to generate an understanding of where the fetus is in 3D space within the uterus using ERT or EIT data (location, area, size, frequency, etc.) as input. This can take the form of a supervised convolutional neural network with ERT or EIT reconstructions as data. The data may be labeled in collaboration with clinicians who compare ERT or EIT reconstruction images with ultrasound data to begin labeling sensor data with observed fetal movements of hands, feet, head, back, etc. The outputs of the model may be the predicted body part that caused the current frame. These frames may create weighted reference points for various other parts of the body. These may then be attached to a 3D model of a fetus, and since there are only so many orientations the fetus can organically be in, this may be further used to extrapolate the fetus’s current position or even its overall motion.• Clustering Model: This model aims to compare and contrast the data collected by the sensors. It may be implemented as an unsupervised model that compares different sensor outputs and attempts to find patterns or correlations between these datasets. The model may use the relative timing, frequency, and amplitude of the compared signals to try to find significance between certain variables. It may group these similar data sets into clusters according to these parameters. These clusters may then be analyzed and tested independently of the model. If the model is found to have correctly identified related data, further research and manual data collection may be done to create a labeled data set for data correlation. This allows to replace parts of the unsupervised model with supervised correlation models, where the change in one variable can be used to predict a change in another. These can take the form of simple linear regression models with one set of linked data as the input and the other linked variable as the output.• Health Interpretation Model: This model is designed to automate tasks normally performed by a clinician. Initially, the system will present clinicians with data from the above models, as well as raw location, intensity, frequency, or other data from the sensors. As they work to diagnose specific problems or highlight areas of concern, these large frames will be labeled accordingly. Models such as random forests could be used, or depending on the data used in the diagnosis, CNNs may be employed. Depending on how much of the data is used to make a particular diagnosis, the models can be simplified and labeled to reduce computational complexity. These simplified models may be combined and fed into the random forest to weigh into the larger overall diagnosis, or to interrupt a slower process with an immediate danger sign diagnosis. The simplified models may include linear regression, CNN, Naive Bayes classifiers, etc. and would take the selected portion of data as input and output a state diagnosis in the simplest case: “abnormal” or “normal”. The ensemble of data in the random forest model may create a decision tree that determines whether the fetus’s current state is “normal”, “abnormal”, or trending toward one or the other. Deviations from normal may tend toward abnormal and vice versa. This “normal” or “abnormal” output and trends may then label the current data. The health interpretation model can be trained on its own output or on data manually labeled by clinicians to account for shifting baselines over time.• Baseline Generation: All labeled data used in these models may be compared to various “baselines” to determine if there are significant deviations from what is considered normal for the fetus. These baselines can then be compared to nationalaverages of fetal movements at specific stages of pregnancy to inform clinical guidance. However, every pregnancy is different, so each user will have their own baseline that represents a healthy, growing fetus. The user’s baseline may be an aggregate of all labeled data from previous uses of the device, which can be used in conjunction with data from the health interpretation model or manual input from the user or clinicians. This aggregate data can indicate change by subtracting the current state of the fetus from the aggregate past state. This can be done with an ensemble of data, including ERT, EIT, external state, and other sensor data, or individually. This technique can augment the data for the health interpretation model, but does not require a separate ML model. The baseline change is an indicator to retrain the models on more recent user data. The models should then be able to adjust their state predictions according to the updated data and produce a more individualized prediction profde. The magnitude of the change in this baseline at any given time may also be used as an indicator of “normal” or “abnormal” health. This simple subtraction may serve as a simpler alternative to the machine learning algorithm, with potentially less granularity and accuracy.

[0150] The processing unit 170 for carrying out the processes described above, according to various embodiments described herein, may be a processor placed on or fixed to the band 110, or a computing device such as a mobile phone used by the user, or a remote server accessed via wireless communication. The processing unit 170 is responsible for receiving the measurements and applying the ML models to analyze the data. For executing machine learning models, the processing unit may be a GPU (Graphics Processing Unit), as GPUs are suitable for the parallel processing required for ML tasks. Alternatively, a neural processing unit (NPU) specifically designed for accelerating ML models may be used.

[0151] Alternatively or additionally, the processing unit may include other specialized processors such as ASICs and FPGAs. ASICs are customized for specific tasks and can offer superior performance and efficiency for particular machine learning operations. FPGAs can provide flexibility by allowing the hardware to be reconfigured for different tasks, making them suitable for various stages of data processing and ML model execution. These specialized processors enhance the efficiency and speed of data analysis, enabling real-time monitoring and interpretation of fetal movements. The integration of these processing technologies ensuresthat the system can handle the complex computations required for accurate and timely fetal health assessments.

[0152] The above data flow and variables can be combined with fetal heart rate monitors, such as Dopplers or electrocardiograms (ECGs), and uterine contraction sensors, such as transducers, to perform an at-home fetal non-stress test. This integration allows for comprehensive monitoring of both fetal movements and heart rate, providing a more complete picture of fetal well-being.

[0153] If a fetal movement or uterine contraction is detected, there is typically an expected acceleration in the fetal heart rate. This measurement, which is usually only performed in hospital settings, can be effectively replicated at home using the disclosed system. By combining the data collected from the ERT or EIT sensors with heart rate data from an ECG or Doppler, the system can provide insights comparable to those obtained from a traditional fetal non-stress test. By leveraging machine learning models, the system can accurately correlate fetal movements with heart rate changes, ensuring that the insights provided are both precise and reliable.

[0154] In various embodiments, additional features may be implemented to enhance the functionality and accuracy of the fetal monitoring system. To achieve efficiency and speed of measurement, pre-processing and filtering of the input signal can be applied to remove noise caused by breathing, external touches, and general body movements. Fetal movements are relatively slow, typically occurring in the range of tens of Hertz (Hz). By utilizing Nyquist sampling, the system can sample at about half this rate before a fetal movement is detected, thereby minimizing power usage and computational load. Upon detecting a fetal movement, the system can switch to a maximum operating sample frequency to collect the most detailed data possible.

[0155] Increasing the resolution with more electrodes and a faster scanning rate can enhance the accuracy of the reconstructed fetal position within the womb. This process involves creating a heat map of the location and amplitude of pressure applied by the fetus, which can be visualized on a two-dimensional surface or as a three-dimensional model of the fetus’s body in real-time. By integrating an inertial measurement unit, the system can reconstruct the geometry of the anterior abdomen and the orientation of the sensor on the body, providing a comprehensive visualization of fetal movements.

[0156] The system can analyze data over a specified period (e.g., two weeks) to determine the mode and mean for each user, setting thresholds to link specific or similar groups of data. Machine learning algorithms can classify and note changes in calibration. A calibration test before each session or use can generate a consistent baseline regardless of how the sensor is worn. This involves taking a burst of data when the garment is first donned and subtracting it from all following frames, ensuring unique auto-calibration for every user. Established patterns and noted deviations can be recorded, with baselines comprising both raw data and interpreted patterns from the ML system. For example, if the system identifies a fetus generally kicking at a frequency of 3Hz and then detects a reduction to 0.5Hz, it can note this deviation.

[0157] The system can track notable growth and baseline trends as pregnancy advances, expecting predictable changes in movement strength and overall sensor displacement as the fetus grows. Tracking these changes and their rates provides insights into fetal growth during the third trimester. If the current rate of change deviates significantly from the expected rate, the system may flag and warn the user.

[0158] The system allows users to select the time period for which the baseline is determined. Data can be presented to users via a mobile or web application using charts, graphs, written insights, and other visual aids. The application can include features such as a chat room or contact form for direct communication with a clinic, and an algorithm-interpreted insight section to provide detailed feedback.

[0159] By using measured locations and the cross-sectional area of a kick, the system can compute the approximate location and orientation of the fetus in the womb. This involves classifying which body part was used to perform a fetal movement via neural networks or other classification methods. High frame rate and live tracking over extended periods will generate and identify unique patterns for users in terms of location, strength, duration, and area. These patterns and deviations can alert users to potential health risks.

[0160] The rate at which a user’s patterns change can be used to quantify risk or perceived danger to a pregnancy. Rapid changes in patterns are cause for alarm, while gradual changes can be used to monitor fetal growth. Patterns may include the frequency of movements, the strength of certain movements, the location drift of movements over time, the strength of movements in specific locations, and the duration of movements. The system can classify data into types of movements and associated risk levels.

[0161] FIG. 11 illustrates a patchwork understanding of a fetus’s activity over a 24- hour period, spanning multiple days of the week. This figure exemplifies how the system collects and integrates data from various sessions to create a comprehensive view of fetal activity. Each shaded block represents a session during which data was collected, showing the duration and timing of monitoring sessions across different days.

[0162] For instance, on Monday, data is collected for a total of three hours, while on Tuesday, data collection occurs for a cumulative duration of four hours, spread across two distinct sessions. Similarly, on Thursday and Friday, there are sessions that span three and five hours respectively, while on Saturday and Sunday, the data collection is limited to one-hour sessions. The aggregate row at the bottom illustrates how these individual sessions are combined to form a continuous understanding of fetal activity throughout the week.

[0163] This patchwork approach allows for the identification of patterns (trends) in fetal activity over time. By comparing data from the same time slots on different days, the system can detect areas of increased or decreased activity, providing insights into the fetus’s energy levels and movement patterns. This data can also be cross-referenced with manually inputted parental activities to understand how different activities may influence fetal movements.

[0164] The system collects variables such as location, area size, and strength of movements, along with frequency and duration-based variables. These variables are grouped based on the type of movement performed by the fetus, allowing the system to generate an adaptable database that evolves with the fetus. For example, today’s data can be compared with data from previous weeks to establish a rate of change, helping to monitor the fetus’s growth and alerting the expectant individual to any potential issues.

[0165] By slotting session times into a 24-hour period and aggregating data over time, the system creates a comprehensive patchwork of a full day’s worth of activity for the fetus. This helps in setting baseline calibration metrics for each individual and establishing a baseline quality for the data collected. Furthermore, this method allows for daily, weekly, and even hourly analysis of fetal movements, capturing slow trends in fetal growth and activity. It can monitor important milestones such as cephalic presentation and provide valuable insights into the fetus’s development.

[0166] FIG.12 illustrates an example reconstruction image of the conductivity distribution to create a two-dimensional model of the fetus and its position in the womb. As shown, the horizontal axis represents the distance from the center of the sensing fdm in the -X and +X directions, while the vertical axis represents the distance from the center of the sensing fdm in the -Y and +Y directions, thereby indicating the location of fetal movement on the sensing fdm. The gray-scale gradient represents the change in conductivity resulting from fetal movement.

[0167] It should be understood that various modifications, alterations, and adaptations may be made to the specific elements and configurations disclosed, including but not limited to dimensions, materials, positions, and operational mechanisms, without departing from the essence and scope of the disclosure.

[0168] The terminology used herein is only for the purpose of describing particular embodiments and is not intended to be limiting. Accordingly, as used herein, the singular forms “a”, “an”, and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises” and “comprising”, when used in this specification, specify the presence of one or more stated features, integers, steps, operations, elements, and components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and groups. Directional terms such as “top”, “bottom”, “upwards”, “downwards”, “vertically”, and “laterally” are used in the following description for the purpose of providing relative reference only, and are not intended to suggest any limitations on how any article is to be positioned during use, or to be mounted in an assembly or relative to an environment. Additionally, the term “connect” and variants of it such as “connected”, “connects”, and “connecting” as used in this description are intended to include indirect and direct connections unless otherwise indicated. For example, if a first device is connected to a second device, that coupling may be through a direct connection or through an indirect connection via other devices and connections. Similarly, if the first device is communicatively connected to the second device, communication may be through a direct connection or through an indirect connection via other devices and connections.

[0169] Use of language such as “at least one of X, Y, and Z,” “at least one of X, Y, or Z,” “at least one or more of X, Y, and Z,” “at least one or more of X, Y, and / or Z,” or “at least one of X, Y, and / or Z,” is intended to be inclusive of both a single item (e.g., just X, or just Y,or just Z) and multiple items (e.g., {X and Y}, {X and Z}, {Y and Z}, or {X, Y, and Z}). The phrase “at least one of’ and similar phrases are not intended to convey a requirement that each possible item must be present, although each possible item may be present.

[0170] It is contemplated that any part of any aspect or embodiment discussed in this specification can be implemented or combined with any part of any other aspect or embodiment discussed in this specification, so long as such those parts are not mutually exclusive with each other.

[0171] While every effort has been made to provide a detailed and accurate description of the disclosure herein, it should be noted that the scope of the disclosure is not limited to the exact configurations and embodiments described. The description provided is intended to illustrate the principles of the disclosure and not to limit the disclosure to the specific embodiments illustrated. It is intended that the scope of the disclosure be defined by the appended claims, their equivalents, and their potential applications in other fields.

Claims

CLAIMS1. A system, comprising: a wearable device that is wearable by a user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Resistance Tomography (ERT) electrodes and a conductive fdm electrically connected to the plurality of ERT electrodes, wherein the conductive fdm is positioned over at least a portion of an anterior abdomen of the user when the wearable device is worn by the user and generates one or more electrical signals in response to a movement of the fetus on a surface of the anterior abdomen; and an electronics unit communicatively coupled to the sensor assembly, the electronics unit configured to transmit a current from at least one of the plurality of ERT electrodes, through the conductive film, and to at least another one of the plurality of ERT electrodes such that a measurement associated with a deformation of the conductive film is collected.

2. The system of claim 1, further comprising a processing unit communicatively coupled to the electronics unit, the processing unit determining a status of the fetus using the measurement in response to receiving the measurement.

3. The system of claim 2, wherein the processing unit, upon receiving the measurement, constructs a two-dimensional (2D) image of the fetus by processing the measurement using tomography.

4. The system of claim 2 or 3, wherein the processing unit determines the status of the fetus by comparing the measurement with a baseline dataset, the baseline dataset being obtained from historical measurements, predefined standards, or generated by a pretrained machine learning model.

5. The system of any one of claims 2-4, wherein the processing unit comprises at least one of a Graphics Processing Unit (GPU), a Neural Processing Unit (NPU), an Application- Specific Integrated Circuit (ASIC), or a Field-Programmable Gate Array (FPGA).

6. The system of any of claims 1-5, wherein the electronics unit comprises a plurality of multiplexers electrically connected to the sensor assembly, a current source electricallyconnected to at least one of the plurality of multiplexers, and a signal receiver electrically connected to at least another one of the plurality of multiplexers.

7. The system of any one of claims 1-6, wherein the measurement comprises at least one of count, time, strength, frequency, duration, location, or area size of the movement of the fetus.

8. The system of any one of claims 2-7, wherein the plurality of ERT electrodes delimit a sensing area equal to or smaller than the conductive fdm, and the plurality of ERT electrodes are located on a periphery of the sensing area.

9. The system of any one of claims 1-8, wherein the conductive fdm is made of a piezoresistive material.

10. The system of any one of claims 1-9, wherein the conductive fdm has a lattice structure electrically connecting at least some of the plurality of ERT electrodes to each other.

11. The system of claim 10, wherein the plurality of ERT electrodes are interconnected by a plurality of connections within the periphery of the lattice structure.

12. The system of claim 11, wherein the plurality of connections are strips of conductive material, printed conductive ink, or are cut from a piece of conductive material.

13. The system of any one of claims 1-12, wherein the wearable device is attachable to a garment wearable by the user.

14. The system of any one of claims 1-12, wherein the wearable device is a band wearable by the user, and wherein the sensor assembly and the electronics unit are embedded into the band.

15. The system of claim 8, wherein at least one of a size or a position of the sensing area is calibrated relative to the anterior abdomen of the user by receiving user input through pressing on the sensing area or tracing across the sensing area, such that the processing unit identifies an anatomical landmark or measures an overall size of the anterior abdomen when the wearable device is worn by the user.

16. The system of any one of claims 1-15, wherein a sensing area of the sensor assembly is divided into a plurality of regions of interest (ROIs), and the processing unit is capable of identifying a body part of the fetus using the measurement and the plurality of ROIs.

17. The system of any one of claims 1-16, further comprising at least one of an accelerometer for detecting an acceleration value or an inertial measurement unit (IMU) for detecting an absolute orientation value or an angular velocity value, wherein the processing unit is configured to determine if the measurement is unrelated to the fetus using the acceleration value or the inertia value.

18. The system of claim 14, wherein the band further comprises a peripheral frame surrounding the conductive film and an adjustable side-strap coupled to the peripheral frame such that the conductive film is mechanically isolated from the adjustable sidestrap by the peripheral frame.

19. The system of claim 18, wherein the peripheral frame is less elastic than the adjustable side-strap and the conductive film is more elastic than the adjustable side-strap.

20. The system of any one of claims 2-5, wherein the processing unit is further configured to utilize a machine learning model trained on historical or real-time measurements to predict a position of the fetus within the anterior abdomen, classify a type of the movement of the fetus detected by the sensor assembly, or differentiate between a true fetal movement and an unrelated deformation of the conductive film.

21. The system of claim 20, wherein the machine learning model comprises at least one of a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short-Term Memory network (LSTM), a Decision Tree, or a Random Forest.

22. A system, comprising: a wearable device that is wearable by a user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Impedance Tomography (EIT) electrodes for sensing an impedance change within the abdomen, wherein at least some of the plurality of EIT electrodes contact a skin of the user and surround a circumference of a torso of the user when the wearable device is worn by the user; andan electronics unit communicatively coupled to the sensor assembly, the electronics unit configured to transmit a current from at least one of the plurality of EIT electrodes, through the abdomen of the user, and to at least another one of the plurality of EIT electrodes such that a measurement associated with an impedance change within the abdomen is collected.

23. The system of claim 22, further comprising a processing unit communicatively coupled to the electronics unit, the processing unit determining a status of the fetus using the measurement in response to receiving the measurement.

24. The wearable device of claim 23, wherein the processing unit constructs a 2- dimensional (2D) or 3-dimensional (3D) image of the fetus using the measurement in response to receiving the measurement.

25. The system of claim 23 or 24, wherein the processing unit determines the status of the fetus by comparing the measurement with a baseline dataset, the baseline dataset being obtained from historical measurements, predefined standards, or generated by a pretrained machine learning model.

26. The system of any one of claims 23-25, wherein the processing unit comprises at least one of a Graphics Processing Unit (GPU), a Neural Processing Unit (NPU), an Application-Specific Integrated Circuit (ASIC), or a Field-Programmable Gate Array (FPGA).

27. The system of any of claims 22-26, wherein the electronics unit comprises a plurality of multiplexers electrically connected to the sensor assembly, a current source electrically connected to at least one of the plurality of multiplexers, and a signal receiver electrically connected to at least another one of the plurality of multiplexers.

28. The system of any one of claims 22-27, wherein at least some of the plurality of EIT electrodes surround an anterior abdomen of the user when the wearable device is worn by the user.

29. The system of any one of claims 22-28, wherein the sensor assembly further comprises one or more belts for connecting the plurality of EIT electrodes to the electronics unit.

30. The system of any one of claims 22-29, wherein the current is an alternating current (AC).

31. The system of any one of claims 22-30, wherein the wearable device is attachable to a garment wearable by the user.

32. The system of any one of claims 22-30, wherein the wearable device is a band wearable by the user, and wherein the sensor assembly and the electronics unit are embedded into the band.

33. A method for processing data collected by a wearable device, the wearable device comprising: a wearable device that is wearable by a user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Resistance Tomography (ERT) electrodes and a conductive fdm electrically connected to the plurality of ERT electrodes, wherein the conductive fdm is positioned over at least a portion of an anterior abdomen of the user when the wearable device is worn by the user and generates one or more electrical signals in response to a movement of the fetus on a surface of the anterior abdomen, the method comprising: transmitting a current from at least one of the plurality of ERT electrodes, through the conductive fdm, and to at least another one of the plurality of ERT electrodes to collect a measurement associated with a deformation of the conductive fdm as at least one variable; and determining a status of the fetus using the at least one variable.

34. The method of claim 33, further comprising, upon receiving the measurement, constructing a two-dimensional (2D) image of the fetus by processing the measurement using tomography.

35. The method of claim 33 or 34, wherein the variable further comprises a maternal metric associated with physiological data or psychological status of the user.

36. The method of any one of claims 33-35, wherein the measurement comprises at least one of count, time, strength, frequency, duration, location, or area size of the movement of the fetus.

37. The method of any one of claim 33-36, wherein determining the status of the fetus comprises determining the status of the fetus by comparing the variable with a baseline dataset, the baseline dataset being obtained from historical variables or predefined standards, or generated by a pre-trained machine learning model.

38. The method of any one of claim 33-36, further comprising identifying a pattern of the variable over a period of time.

39. The method of claim 38, wherein the variable comprises at least two variables, at least one of the at least two variables being the measurement, the method further comprising identifying a correlation between the at least two variables.

40. The method of claim 39, wherein determining the status of the fetus comprises determining the status of the fetus by comparing at least one of the variable, the pattern, or the correlation with a baseline dataset, and wherein the baseline dataset is obtained from historical variables or predefined standards, or generated by a pre-trained machine learning model.

41. The method of claim 40, wherein determining the status of the fetus comprises determining a rate of change of at least one of the variable, the pattern, or the correlation with the baseline dataset over time, to identify an anomaly in response to the rate of change exceeding a predetermined threshold.

42. The method of any one of claims 33-41, further comprising: dividing a sensing area of the sensor assembly into a plurality of regions of interest (ROIs); identifying at least one body part of the fetus using the measurement and the plurality of ROIs; and determining a position of the fetus using the identified body part and the plurality of ROIs.

43. The method of claim 42, further comprising calibrating at least one of a size or a position of the sensing area relative to the anterior abdomen of the user in response to the user pressing on the sensing area when the wearable device is worn by the user.

44. The method of claim 43, further comprising determining a change of a size of the anterior abdomen of the user to monitor a growth of the fetus.

45. The method of any one of claims 33-44, further comprising: obtaining an acceleration value from an accelerometer or an absolute orientation value or an angular velocity value from an inertial measurement unit (IMU); determining if the measurement is unrelated to the fetus using the acceleration value or the inertial value; and in response to determining the measurement is unrelated to the fetus, removing the measurement.

46. The method of any one of claims 33-45, further comprising utilizing a machine learning model trained on historical or real-time measurements to predict a position of the fetus within the anterior abdomen, classify a type of the movement of the fetus detected by the sensor assembly, or differentiate between a true fetal movement and an unrelated deformation of the conductive fdm.

47. The method of claim 46, wherein the machine learning model comprises at least one of a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short-Term Memory network (LSTM), a Decision Tree, or a Random Forest.

48. The method of any one of claims 33-47, wherein the wearable device is attachable to a garment wearable by the user.

49. The system of any one of claims 33-47, wherein the wearable device is a band wearable by the user, and wherein the sensor assembly is embedded into the band.

50. A method for processing data collected by a wearable device, the wearable device comprising:a wearable device that is wearable by a user, wherein the wearable device comprises a sensor assembly, the sensor assembly comprising a plurality of Electrical Impedance Tomography (EIT) electrodes for sensing an impedance change within the abdomen, wherein at least some of the plurality of EIT electrodes contact a skin of the user and surround a circumference of a torso of the user when the wearable device is worn by the user, the method comprising: transmitting a current from at least one of the plurality of EIT electrodes, through the abdomen of the user, and to at least another one of the plurality of EIT electrodes to collect a measurement associated with an impedance change within the abdomen as a variable; and determining a status of the fetus using the variable.

51. The method of claim 50, further comprising constructing a 2-dimensional (2D) or 3- dimensional (3D) model of the fetus using the measurement.

52. The method of claim 50 or 51 , wherein the variable further comprises a maternal metric associated with physiological data or psychological status of the user.

53. The method of any one of claims 50-52, wherein determining the status of the fetus comprises determining the status of the fetus by comparing the variable with a baseline dataset, the baseline dataset being obtained from historical variables or predefined standards, or generated by a pre-trained machine learning model.

54. The method of any one of claims 50-53, further comprising identifying a pattern of the variable over a period of time.

55. The method of claim 54, wherein the variable comprises at least two variables, at least one of the at least two variables being the measurement, the method further comprising identifying a correlation between the at least two variables.

56. The method of claim 55, wherein determining the status of the fetus comprises determining the status of the fetus by comparing at least one of the variable, the pattern, or the correlation with a baseline dataset, and wherein the baseline dataset is obtainedfrom historical variables or predefined standards, or generated by a pre-trained machine learning model.

57. The method of claim 56, wherein determining the status of the fetus comprises determining a rate of change of at least one of the variable, the pattern, or the correlation with the baseline dataset over time, to identify an anomaly in response to the rate of change exceeding a predetermined threshold.

58. The method of any one of claims 50-57, wherein the wearable device is attachable to a garment wearable by the user.

59. The system of any one of claims 50-57, wherein the wearable device is a band wearable by the user, and wherein the sensor assembly is embedded into the band.

Images