CLOSED LOOP WEARABLE SENSOR AND METHOD

Abstract

A method and electronic device for measuring a motion signature and an acoustic signature of a physiological process of a subject, the method comprising: measuring a motion signature and an acoustic signature of a physiological process of the subject, calculating first feature-related data from the motion signature and the acoustic signature, transmitting the first feature-related data to a machine learning service, transmitting the first feature-related data to the machine learning service, and determining a predictive detection of a human scratching behavior by the machine learning service by performing a machine learning operation on the feature-related data.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of and priority to co-pending U.S. patent application Ser. No. 18 / 121,640, filed March 15, 2023, and Provisional Application No. 63 / 321,737, filed March 20, 2022, the contents of each of which are incorporated by reference herein as if set forth in their entirety.

[0002] FIELD OF THE DISCLOSURE

[0002] The embodiments described herein relate generally to systems and methods for detecting physiological parameters, and more particularly, but without limitation, to systems and methods for detecting specific activities based on physiological parameters. [Background technology]

[0003] More than 1 in 10 people suffer from acute or chronic itch. Although itch causes significant morbidity, no technology exists to accurately, objectively, and continuously assess itch by quantifying itch (or other physiological symptoms or behaviors) in a user's natural environment. Thus, treatment options remain limited.

[0004] Existing technologies also rely on transmitting sensor acquired data to remote locations for analysis, which contributes to lags and delays in analytical performance.

[0005] Therefore, a need exists for systems and methods that overcome the shortcomings associated with existing technology. Summary of the Invention [Problem to be solved by the invention]

[0006] This Summary is provided to introduce some concepts in a simplified form that are further described below in the Detailed Description section. This Summary is not intended to identify or exclude key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter. [Means for solving the problem]

[0007] According to one aspect, an embodiment relates to an electronic device for measuring motion and acoustic signatures of a physiological process of a subject, the electronic device comprising: an inertial measurement unit (IMU) for detecting the motion and acoustic signatures; and a microcontroller unit (MCU) communicatively coupled to the IMU, the IMU operable to mechanically communicate directly with the skin of the subject, the MCU for calculating feature-related data from the motion and acoustic signatures, the MCU for transmitting the feature-related data to a machine learning service executing on the MCU, the machine learning service for predicting detection of a human scratching activity by the subject by performing a first machine learning operation on the feature-related data.

[0008] In some embodiments, the electronic device further comprises a vibration motor to alert the subject. In some embodiments, the MCU turns on the vibration motor when a scratching activity of the person is detected.

[0009] In some embodiments, the MCU calculates second feature-related data from the motion signature and the acoustic signature. In some embodiments, the second feature-related data is used to determine an intensity of the detected scratching behavior. In some embodiments, the second feature-related data is used to predict that the subject is asleep. In some embodiments, a second machine learning operation is used to detect the scratching behavior if the subject is predicted to be asleep. In some embodiments, a pattern of vibration motor operation is determined based on whether the subject is predicted to be asleep.

[0010] In some embodiments, detecting the human scratching behavior comprises detecting a start of the scratching behavior, associating the start of the scratching behavior with time 1, detecting an end of the scratching behavior, and associating the end of the scratching behavior with time 2. In some embodiments, the MCU calculates the scratching duration by subtracting time 1 from time 2.

[0011] According to another aspect, an embodiment relates to a method comprising measuring motion and acoustic signatures of a physiological process of a subject, calculating first feature-related data from the motion and acoustic signatures, transmitting the first feature-related data to a machine learning service, and determining a predictive detection of a human scratching behavior by the machine learning service by performing a machine learning operation on the feature-related data.

[0012] In some embodiments, second feature-related data is calculated from the motion signature and the acoustic signature, hi some embodiments, the second feature-related data is used to predict that the subject is asleep.

[0013] In some embodiments, the method further comprises providing haptic feedback to the human upon detecting a scratching behavior of the human.

[0014] In some embodiments, measuring the motion signature and the acoustic signature comprises measuring the signatures by an inertial measurement unit, and determining the predicted detection comprises executing a microcontroller unit, and the inertial measurement unit and the microcontroller are disposed in the sensor.

[0015] In some embodiments, the machine learning operations include at least one of a neural network, a logistic regression classifier, and a random forest classifier. [Brief description of the drawings]

[0016] Non-limiting and non-exhaustive embodiments of the present invention are described with reference to the following figures, in which like reference numbers refer to like parts throughout the various figures unless otherwise specified.

[0017] [Figure 1] FIG. 1 shows a diagram of a sensor according to one embodiment.

[0018] [Diagram 2] FIG. 2 shows a diagram of the electronic board of the sensor of FIG. 1 according to one embodiment.

[0019] [Diagram 3] FIG. 3 illustrates a graphical representation of force versus tear according to one embodiment.

[0020] [Figure 4] FIG. 4 illustrates a graphical representation of various scratching behaviors detected over a period of time according to one embodiment.

[0021] [Diagram 5] FIG. 5 illustrates a workflow for analyzing feature-related data according to one embodiment.

[0022] [Figure 6] FIG. 6 illustrates a graphical representation of a sigmoid function according to one embodiment.

[0023] [Figure 7] FIG. 7 illustrates a random forest classifier according to one embodiment.

[0024] [Figure 8] FIG. 8 illustrates a flow chart of a method for monitoring user behavior according to one embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0025] Various embodiments are described more fully below with reference to the accompanying drawings, which form a part of this specification and show certain exemplary embodiments. However, the concepts of the present disclosure may be embodied in many different forms and should not be construed as being limited to the embodiments set forth herein, but rather, these embodiments are provided as part of the detailed and complete disclosure to fully convey the scope of the concepts, techniques and embodiments of the present disclosure to those skilled in the art. The embodiments may be implemented as a method, a system or an apparatus. Thus, the embodiments may take the form of a hardware implementation, an entirely software implementation or an implementation combining software and hardware aspects. Therefore, the following detailed description is not to be taken in a limiting sense.

[0026] References herein to "one embodiment" or "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one example implementation or technique according to the present disclosure. The appearances of the phrase "in one embodiment" in various places in this specification do not necessarily all refer to the same embodiment. The appearances of the phrase "in some embodiments" in various places in this specification do not necessarily all refer to the same embodiment.

[0027] Some of the descriptions which follow are presented in terms of symbolic representations of operations on non-transient signals stored in a computer memory. These descriptions and representations are used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. Such operations usually require physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical, magnetic or optical signals capable of being stored, transferred, combined, compared and otherwise manipulated. It is sometimes convenient, primarily for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like. Further, it is also sometimes convenient to refer to particular arrangements of steps requiring physical manipulations of physical quantities as modules or code devices, without loss of generality.

[0028] However, all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. As will become apparent from the discussion that follows, unless specifically stated otherwise, throughout this specification, discussions using terms such as "processing" or "computing" or "calculating" or "determining" or "displaying" will be understood to refer to the operations and processing of a computer system or similar electronic computing device that manipulates and transforms data represented as physical (electronic) quantities in the memory or registers of the computer system or other such information storage, transmission or display device. Portions of the present disclosure include processes and instructions that may be implemented in software, firmware or hardware, and when implemented in software, may be downloaded to reside on and be operated from a variety of platforms used by a variety of operating systems.

[0029] The present disclosure also relates to an apparatus for performing the operations herein. The apparatus may be specially constructed for the required purposes or may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such computer programs may be stored on a computer-readable storage medium, including but not limited to a floppy disk, optical disk, CD-ROM, magneto-optical disk, read-only memory (ROM), random access memory (RAM), EPROM, EEPROM, magnetic or optical card, application specific integrated circuit (ASIC), or any type of medium suitable for storing electronic instructions, each of which may be coupled to a computer system bus. Furthermore, the computers referred to herein may have a single processor or may be architectures employing multiple processor designs to increase computing power.

[0030] The processes and displays presented herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform one or more of the method steps. The structure of a variety of these systems is discussed in the description that follows. Furthermore, any specific programming language sufficient to implement the techniques and implementations of the present disclosure may be used. As discussed herein, a variety of programming languages ​​may be used to implement the present disclosure.

[0031] Moreover, the language used herein has been selected primarily for ease of reading and explanation, and may not have been selected to define or encompass the disclosed subject matter. Thus, the present disclosure is intended to be illustrative, not limiting, of the scope of the concepts discussed herein.

[0032] Pruritus, or more informally, "itch," is a common symptom of several medical conditions ranging from skin conditions to systemic diseases such as renal failure, urticaria, diabetes, chronic pruritus of the elderly, and malignancies such as cutaneous T-cell lymphoma (CTCL).

[0033] Acute and chronic itch affect 8% to 14% of the population, respectively, and can adversely affect quality of life. The itch-scratch cycle typically begins with itch, which triggers scratching, resulting in further itch, inflammation, and infection.

[0034] This scratching behavior can become habitual. Habitual scratching behavior is particularly troublesome when it occurs during sleep, which is essential for proper development, physical growth, mental health and general well-being. However, nighttime scratching behavior can interfere with sleep quality and cause or at least contribute to sleep disorders.

[0035] Existing itch assessment techniques administer or prepare patients for subjective surveys with numerical rating scales. However, there is poor correlation between actual and observed scratching behavior. Moreover, the value of these survey results is questionable, since patients may have different tolerances or sensitivity to the degree of itch.

[0036] Another existing approach to assessing itch is direct visualization of scratching via video recording, however this is time consuming, impractical and intrusive to the user.

[0037] The embodiments described herein provide novel systems and methods for assessing itch, detecting scratching behavior, and providing therapeutic biofeedback. Figure 1 illustrates an acousto-mechanical wearable sensor device 100 according to one embodiment. The sensor 100 may be operably placed against a person's skin to collect physiological data, behavioral data, and the like. In particular, the sensor 100 may acquire low frequency data corresponding to movement and high frequency data corresponding to a wide variety of clinical biomarkers.

[0038] The sensor 100 may have a base 102, an electronic board 104, and a top 106. The base 102 may have or be configured with a waterproof silicone portion that may be attached to a user's hand or other body location. Additionally, the base 102 should be soft and flexible so that it can accommodate the movements of the user's hand and remain in operable contact with the user's hand during the movements. The base 102 may be held in operable contact with the user at least in part by an adhesive.

[0039] In some embodiments, the base 102 may be attached to the back of the user's hand. For example, the base 102 may be placed on the metacarpal bones associated with the user's index and middle fingers. In this location, the sensor 100 may detect low and high frequency signals resulting from the user performing a scratching or rubbing action with one or more of the fingers, wrist, hand, etc.

[0040] Upon predicting detection of scratching behavior (e.g., upon determining that the user is engaged in scratching behavior or is at least likely to be engaged in scratching behavior), the sensor may provide haptic feedback to the user, which may alert the patient to the scratching behavior (including involuntary scratching behavior) to encourage the user to cease scratching.

[0041] The electronics board 104 may have or support multiple components including, but not limited to, a microcontroller, a power supply 108, an accelerometer 110, a haptic feedback provider 112, such as a haptic motor, and wireless charging circuitry 114. Thus, the sensor 100 provides multiple hardware components in a compact space to minimize form factor and maximize the utility and practicality of the device with low power consumption and long operating times.

[0042] The upper portion 106 may be formed of or may include a soft, stretchy, waterproof material. The upper portion 106 should be able to be flexible to accommodate the movements of the user's hand or other body parts.

[0043] Figure 2 illustrates a diagram 200 of the electronics circuit board 104 of Figure 1 according to one embodiment. Components illustrated in Figure 1 are illustrated in Figure 2 using the same reference numbers. The electronics board 104 of Figure 2 may include a microcontroller unit (MCU) 202, an inertial measurement unit (IMU) 204, a power supply 108, a power management unit (206), storage 210, and a haptic feedback provider 212.

[0044] The MCU 202 may execute or have a memory 214, an interface 216, one or more machine learning models 218, and a general purpose input / output (GPIO) 220. The MCU 202 may be located on-board with the sensor 100 to perform the necessary analysis on the sensor 100.

[0045] The IMU 204 may include or implement one or more accelerometers 110. The accelerometer(s) 110 may measure acceleration along the x-axis and y-axis at 200 Hz. The accelerometer(s) 110 may measure acceleration in the z-axis up to 1,600 Hz. Acceleration in these directions may indicate movement, such as, for example, a scratching behavior.

[0046] The MCU 202 may have various combinations of resistors, transistors and diodes to form any number of logic gates to execute the appropriate program. The MCU 202 may store any combination of types of memory 214, such as a RAM memory configuration. The memory 214 may include non-volatile memory, such as flash memory, EPROM, EEPROM, ROM and PROM, or volatile memory, such as static RAM or dynamic RAM, as discussed above. Of course, the exact configuration / type of the on-board memory 214 may vary, so long as instructions to analyze data from the IMU 204 may be executed by the MCU 202.

[0047] The machine learning model(s) 218 ​​may implement one or more of various machine learning models 218 to analyze the data collected and provided by the IMU 204. The one or more machine learning models 218 may be trained with data from a clinical study monitoring scratching behavior.

[0048] For example, data on scratching behavioral events may be collected over a period of time from various body parts, including, but not limited to, the back of the hand, head, forearm, inner elbow, thigh, and calf. Non-scratching behaviors may also be considered during the training phase, which may include behaviors such as hand waving, texting, tapping, and restlessness-related movements.

[0049] Upon detecting the scratching action, the GPIO 220 may communicate a signal to the haptic feedback provider 212. The haptic feedback provider 212 may then provide haptic feedback to the user. The haptic feedback provider 112 may have an eccentric rotating mass (ERM) motor 222 for generating vibrations.

[0050] In some embodiments, the motor 222 may be configured to rotate at 10,000 rpm to generate vibration feedback of 1.4 G. Regardless of the exact amount of vibration feedback provided, the vibration feedback should be sufficient to alert the user even during sleep.

[0051] The haptic feedback provider 112 may be powered by a voltage regulator 230 controlled by (GPIO) pins 220 from the MCU 202. For example, the voltage regulator 230 may use pulse width modulation to create feedback of variable intensity, duration, or some combination thereof. Pulse width modulation may utilize pulses of various widths to create periods that may modulate the rotational speed of the motor 222. Using this configuration, the haptic feedback provider 112 may generate several types of haptic profiles. For example, the haptic feedback provider 112 may generate profiles having gentle vibrations, hard vibrations, ramp effects, short bursts, etc.

[0052] The power supply 108 may be any type of power source that provides power to the components of the substrate 104. In some embodiments, the power supply 108 may be a rechargeable lithium polymer battery or a lithium ion battery.

[0053] Storage 210 may store data regarding users and signals appropriate for the users. For example, a particular user may require an intense, pulsating haptic signal in order to detect the haptic signal. Thus, the type of haptic profile or haptic feedback provided may vary and may depend on the user, the message intended to be conveyed, etc. For example, the feedback may vary in length, intensity, pattern(s), or combinations thereof.

[0054] Similarly, a user may select the most preferred haptic feedback to minimize aggravation or reduce habituation. For example, some users may not notice or respond to low intensity vibrations from the haptic feedback provider 212. Thus, some users may prefer to receive higher intensity vibrations.

[0055] Additionally or alternatively, some users may find high intensity vibrations annoying or intrusive, and therefore, some users may prefer to receive lower intensity vibrations.

[0056] Data regarding sleep patterns, scratching behavior, and other user behavior may be collected in advance. As part of the training phase, the user wearing the sensor may be asked to perform scratching behaviors with a force sensor (not shown in FIG. 1).

[0057] For example, a user wears a sensor and performs a scratching action on a force resistance resistor (“FSR”). The surface of the force sensor may be coated with aluminum foil to make the coefficient of friction (“COF”) of the surface similar to the back of a hand (“DH”). The FSR may detect the force applied to its surface during the test scratching action (i.e., while the user scratches the surface of the FSR). As the user wears the sensor during the test scratching action, an accelerometer in the sensor may simultaneously collect acceleration data associated with the user's movements.

[0058] The applied force may represent an intensity level of the scratching behavior, i.e., the greater the force detected as a result of the scratching behavior, the more intense the scratching behavior.

[0059] Figure 3 shows a graph 300 showing the relationship between frequency and force. The graph 300 also shows a Spearman correlation showing the relationship between scratching strength (measured in force (N)) and the sensor signal. In each trial, the scratching speed and the angle between the finger and the FSR surface were kept constant. The power of the scratching signal collected by the sensor can be calculated. As the scratching strength increases, the power of the signal between 40Hz and 60Hz increases.

[0060] 4 shows a graphical representation 400 of various scratching behaviors detected over a period of time. As can be seen in the graphical representation 400, acceleration data of scratching behaviors of various intensities are detected over a period of time. As the scratching intensity increases, the power of the signal between 40 Hz and 60 Hz increases.

[0061] During operation, the IMU 204 may collect data regarding the user's actions. For example, as described above, the accelerometer(s) 110 may collect acceleration data in various axes to identify actions that may be indicative of scratching behavior.

[0062] The machine learning models 218 may implement one or more of various machine learning models 218 to analyze the data from the IMU 204. The one or more machine learning models 218 may be trained with data from a clinical study monitoring scratching behavior.

[0063] For example, to develop machine learning algorithms, scratching training events may be collected from healthy individuals who are at least 18 years of age and have no skin allergies to adhesives or current skin irritation near the expected sensor site. To detect these scratching events, the sensor 100 may be placed on the back of the participant's hand along the second finger metacarpal bone using skin adhesive. Each participant may perform 35 seconds of scratching and 65 seconds of non-scratching. Participants may perform multiple scratching actions over clothing (e.g., abdomen, thighs, calves, knees, and shoulders) and multiple scratching actions directly on the skin (e.g., cheeks, head, inner elbows, outer elbows, forearms, backs of hands, and palms).

[0064] The machine learning model 218 may include one or more of a random forest classifier, logistic regression, neural network, etc. The exact type(s) of machine learning model 218 used may vary so long as it is capable of analyzing data from the IMU 204 to identify scratching behavior.

[0065] 5 illustrates a workflow 500 for analyzing data from an IMU 204 according to one embodiment. One or more machine learning models 218 may execute the workflow 500, or at least a portion of the workflow 500. The workflow 500, or at least a portion of the workflow 500, may be used to train one or more aspects of the machine learning model 218. This particular model may include a signal processing block 502, a convolutional recurrent neural network (CRNN) 504, a classification module 506, and a clustering module 508.

[0066] Training data acquired from subjects performing scratching and non-scratching behaviors is first sent to processing block 502. This may be z-axis acceleration data, and a sliding window module 510 may divide the acceleration data into groups. For example, in some embodiments, the raw z-axis acceleration time series data is divided into frames of 320 ms (512 data points) using a window with 75% overlap between adjacent frames. A fast Fourier transform module 512 may then calculate short-time Fourier transform (STFT) features.

[0067] The CRNN 504 may then receive and process these features through one or more 1D convolutional layers 514, a linear layer 516, and a gated recurrent unit (GRU) 518 layer that analyzes the temporal patterns of the signal.

[0068] The output of the CRNN 504 is provided to a fully connected neural network 520 for classification. For example, the predicted outputs are clustered using a Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm. For example, a sigmoid function 522 may classify the behavior as a scratching behavior or a non-scratching behavior depending on the associated feature data.

[0069] 6 illustrates a graphical representation of a sigmoid function 600 according to one embodiment. The sigmoid function 600 may represent the probability that a particular behavior indicates a scratching behavior based on feature-related data on the x-axis. The x-axis may correspond to acceleration data obtained from the IMU 204, for example.

[0070] A user or administrator (e.g., medical personnel) may specify the decision boundary 602 such that the feature(s) that place the behavior to the right of the boundary 602 are more likely to be scratching behavior than non-scratching behavior. Thus, the classification module 506 may classify the behavior as a scratching behavior.

[0071] In other words, the sigmoid function 600 represents the probability that a particular behavior represents a scratching behavior as determined by data from the IMU 204. Behaviors characterized to the right of the boundary line 602 may be classified as scratching behaviors, and behaviors characterized to the left of the boundary line 602 may be classified as non-scratching behaviors.

[0072] The workflow 500 may be used to classify actions in at least near real-time. A sigmoid function 522 may classify actions as scratching or non-scratching based on the output from the neural network. The clustering module 506 clusters the feature-related data, for example, by implementing a DBSCAN algorithm.

[0073] In one aspect, the embodiments described herein enable in-sensor computation of the CRNN model using edge computing. Based on the developed CRNN model, the sensor 100 may use a fixed size queue of 512 data points (320 ms at 1600 Hz sampling rate) defined to replicate the frames of the CRNN model. When the queue is full, the data contained in the queue is passed to a signal processing block to compute the STFT. The STFT is then passed to the CRNN model. Each time a new data point is input, the oldest data point is removed and the new data point is added to the queue in a first-in-first-out (FIFO) manner. The scratch detection computation is repeated for every 128 new data points.

[0074] As another example, embodiments herein may use a random forest-based machine learning model to classify an activity as a scratching activity. For example, FIG. 7 illustrates a random forest classifier 700 that receives feature-related data from the IMU 204.

[0075] The random forest classifier 700 may have multiple trees, each configured to detect whether some feature (one or more) is present or associated with an identified behavior. For example, tree 1 may first consider whether a behavior such as z-axis acceleration above a certain threshold has been detected (e.g., whether the user is likely asleep, whether the acceleration has occurred a threshold number of times, etc.). If so, the classifier 700 may traverse a certain branch of the tree.

[0076] Certain nodes in the trees are darkened to indicate the presence or absence of certain features. Although three trees are shown in Figure 7, any number of trees may be used to consider any number of features.

[0077] Each tree may output a classification decision as to whether the behavior is predictive of scratching behavior or not. These predictions are based on the presence or absence of certain features.

[0078] The classification decisions of one or more trees may be combined in step 704. A classification decision is then provided in step 706.

[0079] The examples and illustrations of machine learning models provided above are exemplary. Other types of machine learning models, such as support vector machines, logistic regression-based models, etc., may also be used.

[0080] 2, upon detecting a scratching action, the GPIO(s) 220 may emit a signal to the haptic feedback provider 212 to provide haptic feedback to the user. For example, the motor 222 may cause the sensor 100 to vibrate in some form, intensity, duration, etc.

[0081] The user may detect or sense this vibration and may associate the vibration with the scratching behavior, for example, the vibration may remind the user that they are scratching themselves and prompt the user to stop scratching.

[0082] As mentioned above, the user may be asleep while wearing the sensor 100. The user may unconsciously scratch themselves while asleep. The sensor may nevertheless provide this haptic feedback, thereby prompting the user to cease the scratching behavior, even if at least partially unconscious.

[0083] 8 shows a flow chart of a method 800 for monitoring user activity according to one embodiment. The sensor 100 of FIG.

[0084] Step 802 measures motor and acoustic signatures of a subject's physiological processes. The subject (or simply "user") may have been diagnosed with a medical condition for which itch is a symptom. Thus, the user may have a tendency or desire to scratch one or more locations on their body. As discussed above, scratching may exacerbate the symptom or condition.

[0085] A diagnosis need not be received, rather, a user may wish to inhibit scratching behavior on their own without a diagnosis from medical personnel.

[0086] A user may place a sensor, such as sensor 100, on a body location, such as on the back of the hand between the second and third metacarpal bones. Alternatively, a medical professional may place the sensor on the user.

[0087] The sensors may continuously measure motor and acoustic signatures of the subject's physiological processes, for example, while sleeping, during a medical examination, etc. For example, a user may wish to inhibit scratching behavior while sleeping.

[0088] Step 804 includes calculating first feature-related data from the motion and acoustic signatures. An MCU, such as MCU 202 of FIG. 2, may analyze the received raw data relating to motion and acoustics associated with a user.

[0089] Step 806 includes transmitting the first feature-related data to a machine learning service, which may include models of the types described above and other types of models currently available or hereafter developed.

[0090] Step 808 includes determining detection of a human scratching activity by the machine learning service by performing a machine learning operation on the feature-associated data. Step 808 may include running one or more machine learning models to determine whether the feature-associated data is indicative of a user's scratching activity or whether the feature-associated data is at least indicative of a user's scratching activity. For example, the machine learning model(s) may rely on previously collected data associated with known scratching activity.

[0091] Upon detecting scratching behavior, step 810 includes providing haptic feedback to the user to encourage cessation of scratching behavior regardless of whether the user is asleep or awake. Thus, embodiments herein provide a biofeedback tool for the treatment of itch.

[0092] Applying low power, high fidelity, Al-based analytics integrated into epidermal sensors with skin-friendly adhesives such as sensor 100 for extended wear maximizes clinical potential and relevance. The integration of machine learning analytics into the haptic feedback of sensor 100 provides a closed-loop manner to quantify scratching events and provide real-time accurate feedback to improve patient outcomes. This technology can serve as a platform to classify and provide feedback to patients with various dermatological diseases and comorbidities, and to help guide and evaluate the effectiveness of treatment plans.

[0093] The methods, systems, and devices described above are examples. Various configurations may omit, substitute, or add various procedures or components as appropriate. For example, in alternative configurations, the methods may be performed in a different order than the steps described, and various steps may be added, omitted, or combined. Also, features described with respect to a particular configuration may be combined in various other configurations. Different aspects and elements of the configurations may be combined in a similar manner. Also, because technology evolves, many of the elements are illustrative and do not limit the scope of the disclosure or the claims.

[0094] The embodiments of the present disclosure have been described above with reference to block diagrams and / or operational diagrams of methods, systems, and computer program products according to the embodiments of the present disclosure. The functions / acts described in the blocks may occur out of the order shown in any flowchart. For example, two blocks shown in succession may in fact be executed substantially simultaneously, or the blocks may sometimes be executed in the reverse order, depending on the functions / acts involved. Additionally or alternatively, not all of the blocks shown in the flowchart need to be executed. For example, if a flowchart has five blocks including functions / acts, only three of the five blocks may be executed. In this example, any of the three blocks of the five blocks may be executed.

[0095] A statement that a value exceeds (or is above) a first threshold is equivalent to a statement that the value meets or exceeds a second threshold that is slightly greater than the first threshold, e.g., the second threshold is a value greater than the first threshold at the resolution of the relevant system. A statement that a value is less than (or within) a first threshold is equivalent to a statement that the value is less than or equal to a second threshold that is slightly less than the first threshold, e.g., the second threshold is a value less than the first threshold at the resolution of the relevant system.

[0096] In this specification, specific details are provided to provide a thorough understanding of example configurations (including implementations). However, configurations can be practiced without these specific details. For example, well-known circuits, processes, algorithms, structures and techniques are shown without unnecessary detail to avoid obscuring the configurations. This description only provides an example configuration and does not limit the scope, applicability or configuration of the claims. Rather, the above description of the configurations provides one of ordinary skill in the art with an enabling description for implementing the described techniques. Various changes can be made in the function and arrangement of elements without departing from the spirit or scope of the present disclosure.

[0097] While several example configurations have been described, various modifications, alternative configurations, and equivalents may be used without departing from the spirit of the disclosure. For example, the above elements may be components of a larger system, other rules may take precedence over or modify the application of various embodiments or techniques of the disclosure, and multiple steps may be performed before, during, or after the consideration of the above elements.

[0098] Given the present description and illustrations, one skilled in the art will be able to envision variations, modifications, and alternative embodiments that fall within the scope of the general inventive concept discussed in this application without departing from the scope of the following claims.

Claims

1. 1. An electronic device for measuring a motion signature and an acoustic signature of a physiological process of a subject, comprising: an inertial measurement unit (IMU) for detecting the motion signature and the acoustic signature; a microcontroller unit (MCU) communicatively coupled to the IMU; the IMU operable to be in direct mechanical communication with the subject's skin; the MCU computes feature related data from the motion signature and the acoustic signature; the MCU sending the feature-related data to a machine learning service running on the MCU; The machine learning service predicts detection of a human scratching behavior by a subject by performing a first machine learning operation on the feature-associated data.

2. The electronic device of claim 1 , further comprising a vibration motor for alerting the subject.

3. The electronic device of claim 2 , wherein the MCU turns on the vibration motor when a human scratching action is detected.

4. The electronic device of claim 1 , wherein the MCU calculates second feature-related data from the motion signature and the acoustic signature.

5. The electronic device of claim 4 , wherein the second feature-related data is used to predict whether a subject is asleep.

6. The electronic device of claim 5 , wherein a second machine learning operation is used to detect scratching behavior when the subject is predicted to be asleep.

7. The electronic device of claim 5 , wherein the pattern of operation of the vibration motor is determined based on whether the subject is predicted to be asleep or not.

8. 2. The electronic device of claim 1, wherein detecting the scratching behavior of a human comprises detecting a start of the scratching behavior, associating the start of the scratching behavior with a time 1, detecting an end of the scratching behavior, and associating the end of the scratching behavior with a time 2.

9. The electronic device of claim 8 , wherein the MCU calculates the scratch duration by subtracting the time 1 from the time 2.

10. The electronic device of claim 4 , wherein the second feature-related data is used to determine an intensity of the detected scratching behavior.

11. Measuring a motion signature and an acoustic signature of a physiological process of the subject; calculating first feature related data from the motion signature and the acoustic signature; sending the first feature-related data to a machine learning service; determining a predictive detection of a human scratching behavior by the machine learning service by performing a machine learning operation on the feature-related data; A method for providing the above.

12. The method of claim 11 , further comprising calculating second feature related data from the motion signature and the acoustic signature.

13. The method of claim 12 , wherein the second feature-related data is used to predict whether the subject is asleep.

14. The method of claim 11 , further comprising providing haptic feedback to the human upon detecting the scratching behavior of the human.

15. 12. The method of claim 11, wherein measuring the motion signature and the acoustic signature comprises measuring the signatures with an inertial measurement unit, and determining the predicted detection comprises executing a microcontroller unit, the inertial measurement unit and the microcontroller being disposed in a sensor.

16. The method of claim 11 , wherein the machine learning operations include at least one of a neural network, a logistic regression classifier, and a random forest classifier.

Images