System, method, and device for scratch tracking

Abstract

Provided are systems, methods, and devices for scratch tracking. The system includes a wearable device configured to be worn on an appendage of an entity, the wearable device comprising at least one accelerometer and at least one motion sensor, and a computing device in communication with the at least one accelerometer and the at least one motion sensor, the computing device configured to receive first sensor data from the at least one accelerometer and second sensor data from the at least one motion sensor, wherein the first sensor data and the second sensor data are different, and determine at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.

Description

SYSTEM, METHOD, AND DEVICE FOR SCRATCH TRACKINGCROSS REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 531,986, filed on August 10, 2023, the disclosure of which is hereby incorporated by reference in its entirety.BACKGROUND OF THE INVENTION1. Field

[0002] This disclosure relates generally to scratch tracking and, in non-limiting embodiments, to a systems, methods, and devices for scratch tracking.2. Technical Considerations

[0003] Eczema is a chronic, incurable skin condition that affects over 31 million people in the United States, about 10% of the total population. The primary symptom in eczema patients is pruritus, commonly known as itch. A study of adults with eczema found that 70.5% reported severe, unbearable itch in the past two weeks, 85.8% reported daily itch, and 62.8% reported itching at least 12 hours per day. Chronic itch also accompanies other medical conditions including psoriasis and liver / kidney disease.

[0004] In clinical settings, itch is primarily assessed qualitatively. Doctors often rely on visual observations of damage to the skin which do not necessarily correlate to the amount of itching a patient experiences and do not convey the effect of this itching on the patient’s sleep, focus, and general well-being. Doctors additionally use the Visual Analog Scale (VAS), a survey used to help a person rate the intensity of itch and pain they face. However, the VAS has been found to be extremely susceptible to recall bias or response shifts. The lack of quantitative measurement of itch specifically presents difficulties in assessing medication and treatment efficacy and identifying itch triggers in patients.

[0005] Objective measures of pruritus focus on binary classification of scratching behaviors, which indicate whether a patient is scratching or not. Tracking of these scratching behaviors provides insight into the effect of itch on an individual over time. Video recording using cameras (sometimes infrared) is the current standard. However, this method is time intensive as it requires manual labeling of scratching events by researchers. This method can additionally suffer from blind spots, has privacy concerns, and is only possible while a patient is sleeping. Wearable devices have shown promise for automatic, objective classification of scratching at night. Yet, advancements are needed to develop devices that perform well in daytime when participants still itch and when there are many confounding movements.

[0006] Accelerometer based wristband devices, such as smartwatches, are currently incapable of accurately capturing finger motions, restricting their applicability. Existing wearable devices are unable to track scratch intensity. Scratch intensity can be related to itch intensity, which is tracked by doctors and patients using scales and questionnaires such as the Peak Pruritus NRS and the 5-D Itch Scale. Scratch intensity is one of the factors that determines damage to skin and holds additional insight into the degree of disruption that itching has on everyday life and sleep.SUMMARY

[0007] According to non-limiting embodiments or aspects, provided is a system comprising: a wearable device configured to be worn on an appendage of an entity, the wearable device comprising at least one accelerometer and at least one motion sensor; and a computing device in communication with the at least one accelerometer and the at least one motion sensor, the computing device configured to: receive first sensor data from the at least one accelerometer and second sensor data from the at least one motion sensor, the first sensor data and the second sensor data are different; and determine at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.

[0008] In non-limiting embodiments or aspects, the at least one scratch metric comprises at least one of the following: a scratch detection event, a scratch intensity, a scratch count, or any combination thereof. In non-limiting embodiments or aspects, wherein determining the at least one scratch metric comprises: detecting a scratch event of the entity based on inputting the first sensor data and the second sensor data into the at least one machine-learning model; and determining an intensity of the scratch event based on inputting the first sensor data and the second sensor data into the at least one machine-learning model. In non-limiting embodiments or aspects, the at least one machine-learning model comprises a first model and a second model, the scratch event is detected based on inputting the first sensor data and the second sensor data into the first model, and the intensity is determined based on inputting the first sensor data and the second sensor data into the second model. In non-limiting embodiments or aspects, the at least one motion sensor comprises at least one of the following: a contact microphone, a capacitive sensor, or any combination thereof.

[0009] In non-limiting embodiments or aspects, the wearable device comprises a ring adapted to be worn on a finger of the entity. In non-limiting embodiments or aspects, the computing device is further configured to: automatically generate a notification based on the at least one scratch metric; and communicate the notification to a device associated with theentity. In non-limiting embodiments or aspects, the computing device is further configured to: receive sensor array data from at least one sensor array from the entity or another entity wearing the wearable device interacting with the sensor array; store the first sensor data and the second sensor data received while the sensor array data is received; and train the at least one machinelearning model based on the sensor array data, the first sensor data received while the sensor array data is received, and the second sensor data received while the sensor array data is received. In non-limiting embodiments or aspects, the system includes: at least one second wearable device configured to be worn on the appendage or at least one other appendage of the entity, the at least one second wearable device comprising at least two sensors, the at least one scratch metric is determined based on inputting the first sensor data, the second sensor data, and third sensor data from the at least two sensors of the at least one second wearable device into the at least one machine-learning model.

[0010] According to non-limiting embodiments or aspects, provided is a method comprising: receiving, with at least one computing device, first sensor data from at least one accelerometer arranged on a wearable device configured to be worn on an appendage of an entity; receiving, with the at least one computing device, second sensor data from at least one motion sensor arranged on the wearable device, the first sensor data and the second sensor data are different; determining, with the at least one computing device, at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machinelearning model.

[0011] In non-limiting embodiments or aspects, the at least one scratch metric comprises at least one of the following: a scratch detection event, a scratch intensity, a scratch count, or any combination thereof. In non-limiting embodiments or aspects, wherein determining the at least one scratch metric comprises: detecting a scratch event of the entity based on inputting the first sensor data and the second sensor data into the at least one machine-learning model; and determining an intensity of the scratch event based on inputting the first sensor data and the second sensor data into the at least one machine-learning model. In non-limiting embodiments or aspects, the at least one machine-learning model comprises a first model and a second model, the scratch event is detected based on inputting the first sensor data and the second sensor data into the first model, and the intensity is determined based on inputting the first sensor data and the second sensor data into the second model. In non-limiting embodiments or aspects, the at least one motion sensor comprises at least one of the following: a contact microphone, a capacitive sensor, or any combination thereof.

[0012] In non-limiting embodiments or aspects, the wearable device comprises a ring adapted to be worn on a finger of the entity. In non- limiting embodiments or aspects, the method includes: automatically generating a notification based on the at least one scratch metric; and communicating the notification to a device associated with the entity. In nonlimiting embodiments or aspects, the method includes: receiving sensor array data from at least one sensor array from the entity or another entity wearing the wearable device interacting with the sensor array; storing the first sensor data and the second sensor data received while the sensor array data is received; and training the at least one machine-learning model based on the sensor array data, the first sensor data received while the sensor array data is received, and the second sensor data received while the sensor array data is received. In non-limiting embodiments or aspects, the method includes: receiving, with the at least one computing device, third sensor data from at least two sensors arranged on at least one second wearable device configured to be worn on the appendage or at least one other appendage of the entity, the at least one scratch metric is determined based on inputting the first sensor data, the second sensor data, and third sensor data from the at least two sensors of the at least one second wearable device into the at least one machine-learning model.

[0013] According to non-limiting embodiments or aspects, provided is a computer program product comprising at least one non-transitory computer-readable medium including program instructions that, when executed by at least one computing device, cause the at least one computing device to: receive first sensor data from at least one accelerometer arranged on a wearable device configured to be worn on an appendage of an entity; receive second sensor data from at least one motion sensor arranged on the wearable device, the first sensor data and the second sensor data are different; and determine at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.

[0014] Other non-limiting embodiments or aspects will be set forth in the following numbered clauses:

[0015] Clause 1: A system comprising: a wearable device configured to be worn on an appendage of an entity, the wearable device comprising at least one accelerometer and at least one motion sensor; and a computing device in communication with the at least one accelerometer and the at least one motion sensor, the computing device configured to: receive first sensor data from the at least one accelerometer and second sensor data from the at least one motion sensor, wherein the first sensor data and the second sensor data are different; anddetermine at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.

[0016] Clause 2: The system of clause 1, wherein the at least one scratch metric comprises at least one of the following: a scratch detection event, a scratch intensity, a scratch count, or any combination thereof.

[0017] Clause 3: The system of clause 1 or 2, wherein determining the at least one scratch metric comprises: detecting a scratch event of the entity based on inputting the first sensor data and the second sensor data into the at least one machine-learning model; and determining an intensity of the scratch event based on inputting the first sensor data and the second sensor data into the at least one machine-learning model.

[0018] Clause 4: The system of any of clauses 1-3, wherein the at least one machinelearning model comprises a first model and a second model, wherein the scratch event is detected based on inputting the first sensor data and the second sensor data into the first model, and wherein the intensity is determined based on inputting the first sensor data and the second sensor data into the second model.

[0019] Clause 5: The system of any of clauses 1-4, wherein the at least one motion sensor comprises at least one of the following: a contact microphone, a capacitive sensor, or any combination thereof.

[0020] Clause 6: The system of any of clauses 1-5, wherein the wearable device comprises a ring adapted to be worn on a finger of the entity.

[0021] Clause 7: The system of any of clauses 1-6, wherein the computing device is further configured to: automatically generate a notification based on the at least one scratch metric; and communicate the notification to a device associated with the entity.

[0022] Clause 8: The system of any of clauses 1-7, wherein the computing device is further configured to: receive sensor array data from at least one sensor array from the entity or another entity wearing the wearable device interacting with the sensor array; store the first sensor data and the second sensor data received while the sensor array data is received; and train the at least one machine-learning model based on the sensor array data, the first sensor data received while the sensor array data is received, and the second sensor data received while the sensor array data is received.

[0023] Clause 9: The system of any of clauses 1-8, further comprising: at least one second wearable device configured to be worn on the appendage or at least one other appendage of the entity, the at least one second wearable device comprising at least two sensors, wherein the at least one scratch metric is determined based on inputting the first sensor data, the second sensordata, and third sensor data from the at least two sensors of the at least one second wearable device into the at least one machine-learning model.

[0024] Clause 10: A method comprising: receiving, with at least one computing device, first sensor data from at least one accelerometer arranged on a wearable device configured to be worn on an appendage of an entity; receiving, with the at least one computing device, second sensor data from at least one motion sensor arranged on the wearable device, wherein the first sensor data and the second sensor data are different; determining, with the at least one computing device, at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.

[0025] Clause 11: The method of clause 10, wherein the at least one scratch metric comprises at least one of the following: a scratch detection event, a scratch intensity, a scratch count, or any combination thereof.

[0026] Clause 12: The method of clause 10 or 11, wherein determining the at least one scratch metric comprises: detecting a scratch event of the entity based on inputting the first sensor data and the second sensor data into the at least one machine-learning model; and determining an intensity of the scratch event based on inputting the first sensor data and the second sensor data into the at least one machine-learning model.

[0027] Clause 13: The method of any of clauses 10-12, wherein the at least one machinelearning model comprises a first model and a second model, wherein the scratch event is detected based on inputting the first sensor data and the second sensor data into the first model, and wherein the intensity is determined based on inputting the first sensor data and the second sensor data into the second model.

[0028] Clause 14: The method of any of clauses 10-13, wherein the at least one motion sensor comprises at least one of the following: a contact microphone, a capacitive sensor, or any combination thereof.

[0029] Clause 15: The method of any of clauses 10-14, wherein the wearable device comprises a ring adapted to be worn on a finger of the entity.

[0030] Clause 16: The method of any of clauses 10-15, further comprising: automatically generating a notification based on the at least one scratch metric; and communicating the notification to a device associated with the entity.

[0031] Clause 17: The method of any of clauses 10-16, further comprising: receiving sensor array data from at least one sensor array from the entity or another entity wearing the wearable device interacting with the sensor array; storing the first sensor data and the second sensor data received while the sensor array data is received; and training the at least onemachine-learning model based on the sensor array data, the first sensor data received while the sensor array data is received, and the second sensor data received while the sensor array data is received.

[0032] Clause 18: The method of any of clauses 10-17, further comprising: receiving, with the at least one computing device, third sensor data from at least two sensors arranged on at least one second wearable device configured to be worn on the appendage or at least one other appendage of the entity, wherein the at least one scratch metric is determined based on inputting the first sensor data, the second sensor data, and third sensor data from the at least two sensors of the at least one second wearable device into the at least one machine-learning model.

[0033] Clause 19: A computer program product comprising at least one non-transitory computer-readable medium including program instructions that, when executed by at least one computing device, cause the at least one computing device to: receive first sensor data from at least one accelerometer arranged on a wearable device configured to be worn on an appendage of an entity; receive second sensor data from at least one motion sensor arranged on the wearable device, wherein the first sensor data and the second sensor data are different; and determine at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Additional advantages and details are explained in greater detail below with reference to the exemplary embodiments that are illustrated in the accompanying schematic figures, in which:

[0035] FIG. 1 is a schematic diagram of a system for scratch tracking according to nonlimiting embodiments;

[0036] FIG. 2 is another schematic diagram of a system for scratch tracking according to non-limiting embodiments;

[0037] FIG. 3 is another schematic diagram of a system for scratch tracking according to non-limiting embodiments;

[0038] FIG. 4 shows example sensor data for a system for scratch tracking according to non-limiting embodiments;

[0039] FIG. 5 is a flow diagram of a method for scratch tracking according to non-limiting embodiments;

[0040] FIG. 6 is another flow diagram of a method for scratch tracking according to nonlimiting embodiments;

[0041] FIG. 7 is another schematic diagram of a system for scratch tracking according to non-limiting embodiments;

[0042] FIG. 8 shows graphical user interfaces of a system for scratch tracking according to non-limiting embodiments; and

[0043] FIG. 9 illustrates example components of a computing device used in connection with non-limiting embodiments.DESCRIPTION

[0044] For purposes of the description hereinafter, the terms “end,” “upper,” “lower,” “right,” “left,” “vertical,” “horizontal,” “top,” “bottom,” “lateral,” “longitudinal,” and derivatives thereof shall relate to the invention as it is oriented in the drawing figures. However, it is to be understood that the invention may assume various alternative variations and step sequences, except where expressly specified to the contrary. It is also to be understood that the specific devices and processes illustrated in the attached drawings, and described in the following specification, are simply exemplary embodiments or aspects of the invention. Hence, specific dimensions and other physical characteristics related to the embodiments or aspects disclosed herein are not to be considered as limiting.

[0045] No aspect, component, element, structure, act, step, function, instruction, and / or the like used herein should be construed as critical or essential unless explicitly described as such. Also, as used herein, the articles “a” and “an” are intended to include one or more items and may be used interchangeably with “one or more” and “at least one.” Where only one item is intended, the term “one” or similar language is used. Also, as used herein, the terms “has,” “have,” “having,” or the like are intended to be open-ended terms. Further, the phrase “based on” is intended to mean “based at least partially on” unless explicitly stated otherwise.

[0046] As used herein, the terms “communication” and “communicate” refer to the receipt or transfer of one or more signals, messages, commands, or other type of data. For one unit (e.g., any device, system, or component thereof) to be in communication with another unit means that the one unit is able to directly or indirectly receive data from and / or transmit data to the other unit. This may refer to a direct or indirect connection that is wired and / or wireless in nature. Additionally, two units may be in communication with each other even though the data transmitted may be modified, processed, relayed, and / or routed between the first and second unit. For example, a first unit may be in communication with a second unit even though the first unit passively receives data and does not actively transmit data to the second unit. As another example, a first unit may be in communication with a second unit if an intermediaryunit processes data from one unit and transmits processed data to the second unit. It will be appreciated that numerous other arrangements are possible.

[0047] As used herein, the term “computing device” may refer to one or more devices configured to process data. A computing device may, in some examples, include the necessary components to receive, process, and output data, such as a processor, a display, a memory, an input device, a network interface, and / or the like. A computing device may be a processor, such as a central processing unit (CPU) or graphics processing unit (GPU), a mobile device, and / or other like devices. A computing device may also be a desktop computer, a server computer or other form of non-mobile computer. Reference to “a processor,” as used herein, may refer to a previously -recited processor that is recited as performing a previous step or function, a different processor, and / or a combination of processors. For example, as used in the specification and the claims, a first processor that is recited as performing a first step or function may refer to the same or different processor recited as performing a second step or function.

[0048] Non-limiting embodiments of the systems, methods, and devices for scratch tracking that facilitates the capturing and classification of scratch movements by individuals through a wearable device with multiple types of sensors. Through the use of multimodal sensing using multiple types of sensors, and a sensor array for training, non-limiting embodiments improve upon existing systems by allowing for intensity scratch metrics and for statistically significant differentiation among various scratch intensity levels. As used herein, the term “scratch tracking” may refer to identifying, determining, detecting, and / or classifying scratch movements by entities, including but not limited to detecting scratch events, determining scratch metrics, and / or the like.

[0049] Referring now to FIG. 1, a system 1000 for scratch tracking is shown according to non-limiting embodiments. A wearable device 120 includes a first sensor 102 and a second sensor 104. The wearable device 120 may include a ring adapted to be arranged on an appendage (e.g., finger, wrist, palm, top of hand, arm, and / or the like) of an entity (e.g., a human being or other type of animal). In non-limiting embodiments the wearable device 101 may include one or more wristbands, armbands, stickers, cloths, gloves, and / or the like. It will be appreciated that various types of wearable devices may be used to capture movement of an arm, wrist, hand, and / or finger. In non-limiting embodiments the wearable device 101 may be worn on an index finger of the dominant hand of an entity, although other arrangements are possible.

[0050] With continued reference to FIG. 1, the first sensor 102 may include one or more accelerometers configured to output acceleration data in one or more axes. For example, thefirst sensor may be a three (3)-axis accelerometer. The second sensor 104 may include one or more sensors that detect and / or measure motion (e.g., such as vibration) or parameters from which motion can be derived and / or predicted, and are not accelerometers (e.g., are a different mode than an accelerometer). For example, the second sensor 104 may include one or more of a contact microphone, a capacitive sensor, a magnetometer, a gyroscope, an IMU (Inertial Measurement Unit), a force-torque sensor, a standard microphone, a force sensing resistor, an infrared sensor, a time-of-flight sensor, a distance sensor, a proximity sensor, and / or the like. The first sensor 102 and / or second sensor 104 may be contained within, embedded in, and / or supported by the wearable device 101. The second sensor 104 may be configured to measure a different metric and / or parameter than the first sensor 102, such that the second sensor 104 provides sensor data that represents a different modality than the first sensor 102.

[0051] With continued reference to FIG. 1, the first sensor 102 and second sensor 104 may be in communication with a computing device 100. The computing device 100 may be local to the wearable device 101 (e.g., may be contained within, embedded in, and / or supported by the wearable device 101) in non-limiting embodiments. In non-limiting embodiments, the computing device 100 may be located a distance away from the wearable device 101 and communicate with the first sensor 102 and second sensor 104 via wireless signals, such as but not limited to radio frequency signals. In non-limiting embodiments, the computing device 100 may be a remote server that is in communication with the first sensor 102 and second sensor 104 via a network connection. The network connection may be provided by the wearable device 101 (e.g., a network adapter not shown in FIG. 1) or by a separate device (e.g., a mobile device in communication with the first sensor 102 and second sensor 104). In non-limiting embodiments, there may be multiple computing devices that perform one or more steps, such as but not limited to an on-board computing device (e.g., a processor integrated with the wearable device 101), a separate computing device (e.g., a companion device, such as a mobile device or personal computer), and / or a remote server computer.

[0052] Still referring to FIG. 1, the computing device 100 may process the raw sensor data received from the first sensor 102 (e.g., first sensor data) and the second sensor 104 (e.g., second sensor data). The first sensor data and second sensor data may be processed to determine at least one scratch metric. A scratch metric may include, for example, a detection of a scratch, a scratch force, a scratch velocity, a scratch power, a scratch intensity (e.g., based on force and velocity, such as a product of force and velocity), and / or the like.

[0053] In non-limiting embodiments in which a contact microphone is used as a sensor, it provides a piezo-electric element that senses vibrations through solid objects and is unaffectedby acoustic waves traveling through the air, hence eliminating audio and speech privacy concerns. In non-limiting embodiments, the wearable device 120 or a connected device (e.g., such as a connected circuit board) may include a signal amplifier (not shown in FIG. 1), such as but not limited to a non-inverting operational amplifier, to amplify the signal(s) from one or more of the sensors 102, 104. In non-limiting embodiments, a resistor may also be provided to remove drift from the sensor signal (e.g., a contact microphone signal). In non-limiting embodiments using a microcontroller, a protection diode may be provided. It will be appreciated that different and / or additional components may be used to construct the wearable device 120.

[0054] Referring now to FIG. 2, a system 2000 for scratch tracking is shown according to non-limiting embodiments. A wearable device 202 is shown worn on an index finger of a hand. It will be appreciated that the wearable device 202 may be worn on other fingers and / or other parts of the hand and / or arm. The wearable device 202 may also be work on the leg, paw, and / or the like of an animal. In the example shown in FIG. 2, a wristband 204 is worn that supports a computing device 206, such as a microcontroller, circuit board, and / or other hardware for communicating with the wearable device 202. Additionally or alternatively, the wearable device 202 may communicate with a companion device, such as a mobile device (e.g., a smartphone or the like) or other type of non- wearable computing device via a wireless adapter (not shown in FIG. 2). In some non-limiting examples, the wearable device 202 may include a computing device 206 integrated and / or embedded therein. It will be appreciated that other arrangements are possible. The communication between the wearable device 202 (e.g., the sensors of the wearable device) and the computing device 206 may be wireless or wired.

[0055] Referring now to FIG. 3, a system 3000 for scratch tracking is shown according to non-limiting embodiments. FIG. 3 shows a processing pipeline that starts with a first sensor 301, such as an accelerometer, and a second sensor 303, such as a contact microphone, capacitive sensor, and / or the like. Each sensor 301, 303 outputs raw sensor data from which respective features 302, 304 (e.g., feature vectors) are extracted to represent the raw data. The raw sensor data may be processed with a Discrete Fourier Transform (DFT) process, as an example, to extract frequency features from the raw sensor data. The extracted features 302, 304 may then be combined (e.g., concatenated) to form an input feature 306 to input into one or more models. The features may also be normalized to form the input feature 306. The input feature 306 may be a feature vector, as an example, that represents both features 302, 304. The input feature 306 may then be input into separate models 308, 310. For example, the inputfeature 306 may be separately input into a first neural network 308 and into a second neural network 310.

[0056] With continued reference to FIG. 3, the first neural network 308 and second neural network 310 may be separate neural networks or other machine-learning models that process the same input feature 306. It will be appreciated that, in some non-limiting embodiments, different input features may be input into the first neural network 308 and second neural network 310. The first neural network 308 may be trained to output a first scratch metric 312, such as scratch intensity, scratch force, scratch velocity, scratch power, and / or the like. The first scratch metric may include a score from a range of values (e.g., 0 to 1, 1 to 10, and / or the like) and / or may include an energy metric (e.g., a value in watts). The second network 310 may be a scratch detection network trained to output a scratch probability 314. The scratch probability may be a binary value that indicates the presence or absence of a scratch. The scratch probability may be a value from a range of values (e.g., 0 to 1, 1 to 10, and / or the like) to represent a likelihood score. It will be appreciated that various types of output formats may be used in different non-limiting embodiments.

[0057] Non-limiting embodiments may use various types of neural network architectures and algorithms, including but not limited to Multi-Layer Perceptrons, Recurrent Neural Networks (RNN), Long Short-Term Memory Networks (LSTN), Graph Neural Networks, Pretrained Models, and / or Foundation Models. In non-limiting embodiments, RNNs with LSTMs and transformer neural networks may be used to model time- series sensor data. Additionally or alternatively, in non-limiting embodiments, supervised machine-learning algorithms and models may be used to classify the sensor data based on a scratch metric, such as but not limited to Naive Bayes, Random Forest, and / or Support Vector Machines, as examples.

[0058] Referring to FIG. 4, sensor data over time is shown for a first sensor (a contact microphone in this example) and a second sensor (an accelerometer in this example) according to non-limiting embodiments. FIG. 4 shows raw sensor data over four (4) time steps over five (5) tenths of a second. The sensor data is converted to features for inputting into one or more machine-learning models as discussed herein.

[0059] In non-limiting embodiments, ground truth measurements of the power at which an entity, such as a person, scratches a surface are used to train one or more machine-learning models via supervised learning. In non-limiting embodiments, force measurements are obtained through a sensor array, such as a pressure-sensitive tablet with a plurality of pressure sensors arranged in two dimensions. As an example, a Sensei Morph tablet may be used. It willbe appreciated that various sensor-based systems may be used to measure interaction with an entity. In non-limiting embodiments, a sensor array (e.g., such as a tablet device) may be provided with a material (e.g., a cotton cloth or other like material) to provide a texture. The material may be fastened to the tablet to prevent creasing. The material allows for realistic scratching that replicates patients with pruritus scratching over their clothing to satisfy an itch.

[0060] Referring now to FIG. 5, a method for scratch tracking is shown according to nonlimiting embodiments. It will be appreciated that the order of the steps shown in FIG. 5 is for illustrative purposes only and that non-limiting embodiments may involve more steps, fewer steps, different steps, and / or a different order of steps. In non-limiting embodiments, one or more steps may be automatically performed in response to performance and / or completion of a preceding step. The steps shown in FIG. 5 may be performed by one or more computing devices (e.g., a computing device of a wearable device, a computing device of a companion device, a computing device of a remote server computer, and / or the like). At step 500, first sensor data is received from a first sensor (e.g., such as an accelerometer). At step 502, second sensor data is received from a second sensor (e.g., such as a contact microphone, a capacitive sensor, and / or other like sensors as described herein). It will be appreciated that one or more additional sensors and / or types of sensors may also be used in non-limiting embodiments. Steps 500 and 502 may be performed continuously as a stream of data and / or in time intervals, repeating at step 504 until a timestep is reached (e.g., 0.1 seconds, 1 second, 10 seconds, and / or the like). When a timestep is reached at step 504, the method proceeds to step 506.

[0061] At step 506 of FIG. 5, features are generated for each set of sensor data from each sensor and each timestep. For example, one or more feature vectors may be generated based on the raw sensor data for a timestep to represent features of that sensor data. In non-limiting embodiments, a feature may be generated for each set of sensor data. At step 508, the features generated at step 506 may be combined to form one or more features that represent sensor data from two or more sensors (and representing two or more modalities of sensing). For example, at step 508 the features generated at step 506 may be concatenated together to form a new feature vector (a combined feature vector). At step 510, the combined feature generated at step 508 may be input into one or more machine-learning models. For example, the feature generated at step 508 may be separately input into a first model configured to output a first scratch metric and into a second model configured to output a second scratch metric. The first scratch metric may be a detection of a scratch (e.g., a binary metric representing that a scratch event likely occurred or not). The second scratch metric may be a scratch force, scratch velocity, a scratch power, a scratch intensity (e.g., based on force and velocity, such as aproduct of force and velocity), and / or the like. It will be appreciated that, in some non-limiting embodiments, one machine-learning model may be used and a second scratch metric (e.g., such as scratch detection) may be derived from a first scratch metric (e.g., such as scratch intensity).

[0062] In non-limiting embodiments, one or more actions may be automatically performed in response to determining a scratch metric. For example, in response to a scratch intensity satisfying a threshold (e.g., meeting or exceeding a threshold), an automatic notification may be generated and communicated to a computing device associated with the patient / user, a caregiver, a healthcare provider, and / or the like. It will be appreciated that one or more different thresholds may be used with different automatic actions, such as but not limited to notifying the patient that scratching should stop, incrementing a scratch counter, notifying a third-party or caregiver, and / or the like. In some non-limiting embodiments, an automatic diagnosis may be made and / or recommended based on the data, such as a level of eczema or other condition.

[0063] Referring now to FIG. 6, a method for scratch tracking is shown according to nonlimiting embodiments. It will be appreciated that the order of the steps shown in FIG. 6 is for illustrative purposes only and that non-limiting embodiments may involve more steps, fewer steps, different steps, and / or a different order of steps. In non-limiting embodiments, one or more steps may be automatically performed in response to performance and / or completion of a preceding step. The steps shown in FIG. 6 may be performed by one or more computing devices. FIG. 6 depicts the collection and processing of training data to train one or more machine-learning models for determining one or more scratch metrics. A sensor array 600 includes x and y dimensions and is configured to measure force applied to different portions of the array 600 over time. As an example, the sensor array 600 may include a pres sure- sensitive tablet such as but not limited to a Sensei Morph tablet. Various other sensor arrays 600 may be used.

[0064] At step 602, a force map is generated. A force map may be generated as output by the sensor array 600 and / or may be separately generated based on raw sensor data from the sensor array 600. The force map may include an indication of force for locations with x-y coordinates on the sensor array 600. A force map may be generated at each time step t by the sensor array 600. The force map may be in the form of an image or a matrix of measurements over the surface area of the tablet. From this force map, individual contact(s) may be identified. This processing may be performed by an on-board computing device of the sensor array (e.g., firmware) and / or an external computing device (such as the wearable device, companion device, server computer, and / or the like). For each contact point in a time step / , a force measurement ftand centroid location of the contact (xy, yf) may be determined. An individualmay be instructed to interact with the sensor array 600 while wearing the wearable device to produce data for collection. For example, an individual may be instructed to keep their palm and fingers from touching the sensor array 600 during data collection to ensure there is only a single point of contact at each time step.

[0065] With continued reference to FIG. 6, at step 604 an average force is determined based on the force map. The average force may be determined as the average force over a period of time (e.g., 1 second, 10 seconds, 1 minute, a period of interaction with the array, and / or the like). In non-limiting embodiments, a force algorithm may be applied to determine an estimated mean force for each sample (e.g., for each contact point of each timestep). The force algorithm may be as follows:where F e Fsis the force value for a single timestep from the array of all force values P for a sample of data, and n is the number of values in the sample.

[0066] Steps 606, 608, 610 represent a velocity determination algorithm to determine velocity for each timestep of a period of time. At step 606, the raw position data (e.g., based on x-y coordinates) representing locations of interaction on the sensor array 600 may be interpolated. For each z (e.g., 1-second) sample of data with n number of values, there is an array x = |x / , X2,...,xn] including the x-axis values of the finger contact over time, array y = [ y / , y2,---,yn\ including the y-axis values of the finger contact over time, and array t = \h, t2,---,tn\ including the timesteps corresponding to arrays x and y. The values of x and y may be interpolated in order to fill in any gaps where contact is broken (e.g., values that are Not a Number (NaN)) between the finger / appendage and tablet. Subsequently, this process may yield xland yl.

[0067] At step 608, the interpolated sensor data is processed with one or more filters and / or algorithms. For example, a Savitzky-Golay filter and a peak / valley finder may be applied to the data at step 608. It will be appreciated that various filters and / or algorithms may be applied to the sensor data to smooth the distribution and / or to identify critical points in the sensor data. The critical points (peaks and valleys) along the y-axis may be representative of an appendage moving up and / or down. For example, a fifth order polynomial Savitzky-Golay filter with a filter window length of 0.21 seconds (31 points), or of other sizes, may be applied to smooth yl, which results in ys, with parameters chosen through visual observation of the sensed contacts and generated peaks and valleys from a randomly selected subset of samples fromeach participant (e.g., individual participating in training). For the jth point in ys, the formula for the Savitzky-Golay filtering may be as follows:where the window length m = 31 and the convolution coefficients Ltmay be looked up in the Savitzky-Golay tables. A peak and valley finding algorithm may subsequently be applied to ys. This algorithm may identify peaks and valleys, which are critical points, and represent incidents at which the velocity of the finger / appendage changes sign along the trajectory of the scratching motion. A vector (e.g., feature vector) of critical points may be defined as yce Ikm, where m is the number of critical points in a given second of data. The corresponding x positions for each critical point are signified as xce Ik”1and the corresponding timesteps for each critical point are signified as tce Ikm.

[0068] At step 610, an average velocity for each timestep is determined based on the processed data from step 608. Scratching behavior occurs in one of two ways: (1) either the entity maintains contact with the skin through the entire scratching motion or (2) the entity lifts their finger off of their skin in between each scratching motion. An individual may be instructed to scratch on the sensor array for a predetermined time z (e.g., 10 seconds). The data received from that time may then be segmented into discrete scratching motions characterized by the appendage with the wearable device moving up or down along the y-axis, as identified by detecting peaks and valleys (critical points) in the y-position signal output by the sensor array. By calculating the change in distance and the change in time during each of these individual scratching motions, the average velocity across each time step (e.g., 1-second sample or the like) may be determined.

[0069] To calculate the velocity estimate for each 1-second sample, a subset of the critical points yc,s6 IRq, where q is the number of critical points which occur in the 1-second window, may be selected, along with the corresponding x positions,and timesteps, tc,se IRq. The Euclidian distance, in millimeters, between each adjacent pair of yc,sand xc,scoordinates, may be calculated to form an array of distances dc,s, wherein:

[0070] The difference in the timestamps may also be defined as:rlC,s I

[0071] An array of velocities may also be calculated using v =u / ^c,s and the mean velocity for a 1 -second sample may be calculated using Vs= a=iva(9 ~ 1)is the velocity estimate between two adjacent critical points from the array of all velocities Vsfor a sample of data.

[0072] At step 612, the average force determined at step 604 may be combined with the average velocity determined at step 610 to determine a power or intensity scratch metric (which may be measured in watts, as an example). The power label in milliwatts may be calculated using ps= Fsvswherein Fs(N) is the calculated average force and Vs(mm / s) is the average velocity for the sample.

[0073] In non-limiting embodiments, various metrics may be collected for ground truth labels of intensity, including but not limited to the distance moved by the finger, force, velocity, acceleration, angular velocity, angular acceleration, jerk, energy, power, and / or the like, and any higher-order derivatives of such parameters.

[0074] In non-limiting embodiments, outliers in the velocity algorithm may be removed to mitigate errors in the power labeling procedure. Filtering may be performed under three conditions: (1) if there are less than 2 peaks / valleys in a 1-second window of data, (2) if there are consecutive peaks or consecutive valleys detected in a window of data, and / or (3) if the change in contact x or y positions in consecutive timesteps is greater than 5 mm (or some other predetermined threshold). Any outliers greater than 600 mW (or some other predetermined threshold) may be removed, as power values higher than this limit are often infeasible.

[0075] Linear interpolation may be used to fill in any missing values in the contact microphone (or other type of sensor) and accelerometer z-axis signals, c;eIR8000and a eIR400respectively. Power information may be captured in the frequency domain of the sensor data (e.g., the contact microphone and accelerometer data). The Discrete Fourier Transform (DFT), J7, may be applied to a discrete signal in the time domain to convert it to the frequency domain.For the Zdh point in the DFT result, d. the DFT formula is declared as dk=where g is the time domain signal and N is the length of g. DFT may be applied to the contact microphone signal in order to extract the single-sided frequency amplitude, cA=2 where B = 8000. For the accelerometer signal, aA= -E(a / ), where E = 400.

[0076] The first 400 amplitudes from cAare selected to obtain contact microphone (or other sensor) features that correspond to frequencies from 0 to 399 Hz (or another predetermined range). The first 175 amplitudes (or some other sample size) from aAare selected to obtainaccelerometer features that correspond to frequencies from 0 to 174 Hz (or another predetermined range). The amplitudes from each may be concatenated to form a feature vector, G elR575, and min-max normalization, fitted on the training data, may be applied to G in order to scale all the features to be between 0 and 1.

[0077] In non-limiting embodiments, a neural network for regression of scratch intensity may be trained using normalized wearable features and power labels from a sensor array. In non-limiting embodiments, the best model (e.g., a most desirable model) may be found from a grid search over model hyperparameters, including the number of hidden layers, the number of nodes in the hidden layers, the dropout probability, and / or the number of accelerometer and contact microphone (or other sensor) frequency features. In non-limiting embodiments, cross validation (e.g., Leave-One-Subject-Out Cross Validation (LOSO-CV)) may be used to calculate the Mean Absolute Error (MAE) across all folds, allowing for the selection of the model with the lowest MAE. A dropout of 0.1 or the like may be applied to each hidden layer to reduce overfitting during training. The training may utilize the Adam optimizer with a learning rate of 5 * 10-6, MAE loss, batch sizes of 64, and 150 epochs, as example parameters. It will be appreciated that other parameters may be used in non-limiting embodiments.

[0078] The sensor data may be divided into 1-second samples using a stride of 0.25 seconds and features may be extracted using the DFT. The first 275 amplitudes from cAare selected to obtain contact microphone (or other sensor) features that correspond to frequencies from 0 to 274 Hz (or another predetermined range). The first 200 amplitudes from aAare selected to obtain accelerometer features that correspond to frequencies from 0 to 199 Hz (or another predetermined range). The amplitudes from each may be concatenated to form a feature vector, H elR475, and min-max normalization, fitted on the training data, may be applied to H in order to scale all the features to be between 0 and 1.

[0079] In non-limiting embodiments, these normalized frequency features may be used to train a fully connected neural network with three (3) hidden layers with ReLU activation and a hidden layer size of 1200 nodes, as an example. A sigmoid activation may be applied to the output layer, and during training an applied dropout of 0.2 may be applied to each hidden layer. Binary cross-entropy loss and the Adam optimizer with a learning rate of 1 * 10-5may be utilized with the training of the neural network. The network may additionally be trained for 150 epochs with batch size of 64.

[0080] In non-limiting embodiments, the training data may be augmented by generating additional data points by applying various transformations, such as time-shifting, scaling,and / or noise addition, to the original dataset. This may help increase the diversity of the dataset and help the model(s) generalize better to new, unseen data.

[0081] In non-limiting embodiments, the system and / or device may automatically upload processed data (e.g., outputs) to patient records stored remotely by a server computer, where the data may be aggregated, analyzed, and / or visualize in an application. The data may additionally be synchronized with other information such as sleep data, temperature / humidity, and / or patient-reported details such as diet, allowing individuals to identify correlations between potential itch triggers and subsequent itch symptoms.

[0082] In non-limiting embodiments, the wearable device may include an integrated capacitive and accelerometer sensing wristband. The principle of self-capacitance may be used with copper electrodes on the inside wrist to measure capacitance changes as the tendons in the wrist move when the fingers extend and retract. The capacitive sensing pay be paired with the accelerometer sensor data to measure arm motions during scratching and to capture high frequency vibrations propagating from the fingernail or fingertip through the hand and to the device. Data may be obtained from the device in various ways and in various intervals. As an example, the data may be received at 1000 Hz and 20 Hz for the accelerometer and capacitive sensing electrodes, respectively.

[0083] In non-limiting embodiments, a wearable device may include a computing device such as, for example, a Particle Photon 2 microcontroller and antenna which has Wi-Fi® connectivity. Twelve capacitive sensing electrodes, fabricated from copper tape, may be glued or otherwise attached to a stretchable wristband and overlaid with clear plastic tape. The electrodes, out of which only 8 may be used in non-limiting embodiments, are wired to a circuit board. The device may also include a battery. It will be appreciated that various other arrangements of computing devices, wireless communication devices, sensors, and / or the like may be used in different non-limiting embodiments.

[0084] Referring now to FIG. 7, a system 7000 and method for scratch tracking is shown according to non-limiting embodiments. A wearable device 702 is in communication with a companion device 704, such as a mobile device, personal computer, other computing device on a local area network, and / or the like. It will be appreciated that, in some non-limiting embodiments, a computing device of the wearable device and / or a remote server computer may perform one or more steps described herein as being performed by the companion device 704.

[0085] In non-limiting embodiments, the computing device of the wearable device 702 may execute one or more scripts to collect data. For example, a script may be executed by the wearable device 702 that establishes a wireless network connection with the companion device704. The capacitive sensor may be configured to sample at 20 Hz and the accelerometer to sample at 1000 Hz using software timers, as examples. The sampling frequency of the sensors may be set by the computing device of the wearable device 702. It will be appreciated that various other sample rates may be used.

[0086] In non-limiting embodiments, a second script may be executed on the companion device 704 to set up two files 706, 708 (e.g., such as CSV files, XML, and / or other structured data formats) to log and store data from both the capacitance sensor and the accelerometer. As shown in FIG. 7, the second script may use multithreading to manage the collection process, with the first thread instructing the user to perform a series of scratching and non- scratching activities in a random sequence (e.g., randomly shuffling over activities to perform and recording data for each activity), with durations varying randomly between three (3) and nine (9) seconds. The second processing thread focuses on processing incoming data from the accelerometer and capacitance sensor and appending the sensor data to the files 706, 708 in real time (e.g., as the data is received during operation and collection). Each file 706, 708 may log the timestamp and the activity name. Additionally, the capacitance file 706 may record values from electrodes, while the accelerometer file 708 may record the x, y, and z axis values.

[0087] In non-limiting embodiments, a classification pipeline may be used to classify the raw sensor data. This pipeline may begin with a feature extraction stage and segments the data into 1-second windows or other predetermined windows. For each window, a Fast Fourier Transform (FFT) function is performed on the three axes of the accelerometer and each of the capacitive electrodes (e.g., eight (8) electrodes). The extracted features from each sensor may then be concatenated to form a comprehensive feature vector. This feature vector is used to train the machine-learning classification model (such as a Random Forest (RF) model). In nonlimiting embodiments, the computing device may process the raw sensor data, extract features, scale the features, and generate a prediction every 0.1 seconds (or at other predetermined timesteps). Such functions may be performed by multithreading, with multiple computing devices, and / or the like. The second script may read the sensor data over a time period (e.g., over the last five (5) seconds) and the latest prediction from a file outputted by the first script. The second script may then generate and update a plot every 0.1 seconds or at other intervals.

[0088] In non-limiting embodiments, various performance metrics, such as sensitivity, specificity, overall accuracy, and / or computational complexity, may be used to compare the algorithms and determine the most suitable types of models and algorithms for scratch detection and intensity inference. These evaluation metrics include, for example, sensitivity (the proportion of actual scratching events that the model correctly identifies as scratching),specificity (the proportion of non-scratching events that the model correctly identifies as not scratching), precision (the proportion of predicted scratching events that are actually scratching events), recall (the proportion of actual scratching events that are correctly predicted by the model, Fl Score (the harmonic mean of precision and recall, providing a balanced evaluation of the model’s performance, Area Under the Receiver Operating Characteristic (ROC) Curve (AUC-ROC) (a graphical representation of the trade-off between sensitivity and specificity, with a higher AUC-ROC value indicating better model performance), coefficient of determination, R2 (a statistical measure that represents the proportion of variance in the dependent variable (scratching events) that is explained by the independent variables (model predictions), with a higher value indicating better model performance), computational complexity (the measure of CPU / GPU resources and wall-clock time required for the model to make predictions), and / or mean squared error of scratching intensity (a statistical measure of accuracy using the average squared difference between the predicted and actual scratch intensity values). In non-limiting embodiments, users (e.g., such as patients) may be asked to perform a series of scripted scratching and non-scratching tasks when they first use the wearable device to generate calibration data. This calibration data may then be utilized to personalize and fine-tune one or more pre-trained models for a new user.

[0089] Referring now to FIG. 8, shown are graphical user interfaces (GUIs) 802, 804 for visualizing sensor data according to non-limiting embodiments. Various data visualization techniques may be used, including but not limited to line charts, bar charts, pie charts, and / or heat maps, to display scratching statistics and trends. The GUIs 802, 804 may include interactive elements, allowing users to manipulate and explore the data. For example, users may select a specific time range to view data and / or compare measured scratch behavior to other variables, such as temperature and / or geolocation, that are recorded in association with the scratch metrics. In non-limiting embodiments, a mobile application may include a customizable dashboard for each user configured to allow the users to choose the types of visualizations and statistics they find most informative and spatially arrange them according to their preferences. This customizable approach will support users to explore scratch data in a manner that best suits their needs and enables them to derive meaningful insights. In addition to a dashboard, the GUIs 802, 804 may contain a section to display daily, weekly, and monthly trends to track if scratching behavior is improving or worsening over time.

[0090] The GUIs 802, 804 may include one or more selectable options to set personalized thresholds and / or triggers for notifications based on specific needs and preferences. For example, a user can choose to receive alerts when their scratching frequency or intensityexceeds a certain level or if a new trend emerges in their data. Notifications may be designed to provide users with contextual information and insights to help them understand the significance of the changes and / or trends detected. This may include information on potential environmental triggers, correlations with other health data, and / or comparisons to historical patterns. In non-limiting embodiments, a notification system may be seamlessly integrated with mobile devices operated by the users, allowing the users to receive alerts through preferred channels, such as push notifications, text messages, and / or emails, as examples.

[0091] In non-limiting embodiments, the GUIs 802, 804 may include one or more selectable options for users to share their scratch data and / or scratch behavior statistics with their dermatologist or other caregiver or healthcare professional. When sharing such data, a consent management system may be used to enable patients to control which aspects of their data are shared with their dermatologists or other providers. Users may be given the option to grant access to specific data elements, with suggestions on which data features to share according to what is reported to be most useful to dermatologists. Data sharing may be facilitated through a secure portal, ensuring the safe transfer of sensitive health information. All data collected, transmitted, and stored on a user’s device may be encrypted to protect from unauthorized access.

[0092] In non-limiting embodiments, contextual data, such as local temperature, humidity, activity (e.g., physical activity), self-reported diet, and / or sleep tracking data, may be correlated with the scratch data to identify potential patterns or trends in eczema flare-ups and scratching behavior. For example, data may be collected including temperature, humidity, air quality, dust, allergen levels, and / or the like from public and / or private weather sources or other data sources based on a geolocation of a user and the time the data was collected. Sleep quality and stress data may also be collected from the wearable device or other wearable device as two other factors that can influence eczema symptoms. For example, user-reported sleep and stress measures may be used along with sleep statistics reported from a smartwatch or other wearable device if available. Data preprocessing may involve cleaning, filtering, and / or normalizing the collected environmental data to ensure its quality, relevance, and comparability across different sources and geographic locations. With the preprocessed data, statistical methods and machine learning techniques may be used to analyze the relationships between environmental variables and scratching behaviors in individuals with eczema or other conditions. The significance of these correlations may be defined according to a set of eczema triggers compiled by medical professionals. By analyzing the correlations between environmental variables and scratching behavior, patterns and trends that may contribute to eczema flare-ups and symptom severitymay be identified. The output may be used in non-limiting embodiments to develop personalized eczema management strategies and recommendations for patients and clinicians.

[0093] Referring now to FIG. 9, shown is a diagram of example components of a computing device 900 for implementing and performing the systems and methods described herein according to non-limiting embodiments. For example, the computing device 900 may correspond to the computing device 100 shown in FIG. 1. In some non-limiting embodiments, device 900 may include additional components, fewer components, different components, or differently arranged components than those shown in FIG. 9. Device 900 may include a bus 902, a processor 904, memory 906, a storage component 908, an input component 910, an output component 912, and a communication interface 914. Bus 902 may include a component that permits communication among the components of device 900. In some non-limiting embodiments, processor 904 may be implemented in hardware, firmware, or a combination of hardware and software. For example, processor 904 may include a processor (e.g., a central processing unit (CPU), a graphics processing unit (GPU), an accelerated processing unit (APU), etc.), a microprocessor, a digital signal processor (DSP), and / or any processing component (e.g., a field-programmable gate array (FPGA), an application- specific integrated circuit (ASIC), etc.) that can be programmed to perform a function. Memory 906 may include random access memory (RAM), read only memory (ROM), and / or another type of dynamic or static storage device (e.g., flash memory, magnetic memory, optical memory, etc.) that stores information and / or instructions for use by processor 904.

[0094] With continued reference to FIG. 9, storage component 908 may store information and / or software related to the operation and use of device 900. For example, storage component 908 may include a hard disk (e.g., a magnetic disk, an optical disk, a magneto-optic disk, a solid state disk, etc.) and / or another type of computer-readable medium. Input component 910 may include a component that permits device 900 to receive information, such as via user input (e.g., a touch screen display, a keyboard, a keypad, a mouse, a button, a switch, a microphone, etc.). Additionally, or alternatively, input component 910 may include a sensor for sensing information (e.g., a photo-sensor, a thermal sensor, an electromagnetic field sensor, a global positioning system (GPS) component, an accelerometer, a gyroscope, an actuator, etc.). Output component 912 may include a component that provides output information from device 900 (e.g., a display, a speaker, one or more light-emitting diodes (LEDs), etc.). Communication interface 914 may include a transceiver-like component (e.g., a transceiver, a separate receiver and transmitter, etc.) that enables device 900 to communicate with other devices, such as via a wired connection, a wireless connection, or a combination of wired and wireless connections.Communication interface 914 may permit device 900 to receive information from another device and / or provide information to another device. For example, communication interface 914 may include an Ethernet interface, an optical interface, a coaxial interface, an infrared interface, a radio frequency (RF) interface, a universal serial bus (USB) interface, a Wi-Fi® interface, a cellular network interface, and / or the like.

[0095] Device 900 may perform one or more processes described herein. Device 900 may perform these processes based on processor 904 executing software instructions stored by a computer-readable medium, such as memory 906 and / or storage component 908. A computer- readable medium may include any non-transitory memory device. A memory device includes memory space located inside of a single physical storage device or memory space spread across multiple physical storage devices. Software instructions may be read into memory 906 and / or storage component 908 from another computer-readable medium or from another device via communication interface 914. When executed, software instructions stored in memory 906 and / or storage component 908 may cause processor 904 to perform one or more processes described herein. Additionally, or alternatively, hardwired circuitry may be used in place of or in combination with software instructions to perform one or more processes described herein. Thus, embodiments described herein are not limited to any specific combination of hardware circuitry and software. The term “programmed or configured,” as used herein, refers to an arrangement of software, hardware circuitry, or any combination thereof on one or more devices.

[0096] Although embodiments have been described in detail for the purpose of illustration, it is to be understood that such detail is solely for that purpose and that the disclosure is not limited to the disclosed embodiments, but, on the contrary, is intended to cover modifications and equivalent arrangements that are within the spirit and scope of the appended claims. For example, it is to be understood that the present disclosure contemplates that, to the extent possible, one or more features of any embodiment can be combined with one or more features of any other embodiment.

Claims

WHAT IS CLAIMED IS:

1. A system comprising: a wearable device configured to be worn on an appendage of an entity, the wearable device comprising at least one accelerometer and at least one motion sensor; and a computing device in communication with the at least one accelerometer and the at least one motion sensor, the computing device configured to: receive first sensor data from the at least one accelerometer and second sensor data from the at least one motion sensor, wherein the first sensor data and the second sensor data are different; and determine at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.

2. The system of claim 1, wherein the at least one scratch metric comprises at least one of the following: a scratch detection event, a scratch intensity, a scratch count, or any combination thereof.

3. The system of claim 1, wherein determining the at least one scratch metric comprises: detecting a scratch event of the entity based on inputting the first sensor data and the second sensor data into the at least one machine-learning model; and determining an intensity of the scratch event based on inputting the first sensor data and the second sensor data into the at least one machine-learning model.

4. The system of claim 3, wherein the at least one machine-learning model comprises a first model and a second model, wherein the scratch event is detected based on inputting the first sensor data and the second sensor data into the first model, and wherein the intensity is determined based on inputting the first sensor data and the second sensor data into the second model.

5. The system of claim 1, wherein the at least one motion sensor comprises at least one of the following: a contact microphone, a capacitive sensor, or any combination thereof.

6. The system of claim 1, wherein the wearable device comprises a ring adapted to be worn on a finger of the entity.

7. The system of claim 1, wherein the computing device is further configured to: automatically generate a notification based on the at least one scratch metric; and communicate the notification to a device associated with the entity.

8. The system of claim 1, wherein the computing device is further configured to: receive sensor array data from at least one sensor array from the entity or another entity wearing the wearable device interacting with the sensor array; store the first sensor data and the second sensor data received while the sensor array data is received; and train the at least one machine-learning model based on the sensor array data, the first sensor data received while the sensor array data is received, and the second sensor data received while the sensor array data is received.

9. The system of claim 1, further comprising: at least one second wearable device configured to be worn on the appendage or at least one other appendage of the entity, the at least one second wearable device comprising at least two sensors, wherein the at least one scratch metric is determined based on inputting the first sensor data, the second sensor data, and third sensor data from the at least two sensors of the at least one second wearable device into the at least one machine-learning model.

10. A method comprising: receiving, with at least one computing device, first sensor data from at least one accelerometer arranged on a wearable device configured to be worn on an appendage of an entity; receiving, with the at least one computing device, second sensor data from at least one motion sensor arranged on the wearable device, wherein the first sensor data and the second sensor data are different;determining, with the at least one computing device, at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machinelearning model.

11. The method of claim 10, wherein the at least one scratch metric comprises at least one of the following: a scratch detection event, a scratch intensity, a scratch count, or any combination thereof.

12. The method of claim 10, wherein determining the at least one scratch metric comprises: detecting a scratch event of the entity based on inputting the first sensor data and the second sensor data into the at least one machine-learning model; and determining an intensity of the scratch event based on inputting the first sensor data and the second sensor data into the at least one machine-learning model.

13. The method of claim 12, wherein the at least one machine-learning model comprises a first model and a second model, wherein the scratch event is detected based on inputting the first sensor data and the second sensor data into the first model, and wherein the intensity is determined based on inputting the first sensor data and the second sensor data into the second model.

14. The method of claim 10, wherein the at least one motion sensor comprises at least one of the following: a contact microphone, a capacitive sensor, or any combination thereof.

15. The method of claim 10, wherein the wearable device comprises a ring adapted to be worn on a finger of the entity.

16. The method of claim 10, further comprising: automatically generating a notification based on the at least one scratch metric; and communicating the notification to a device associated with the entity.

17. The method of claim 10, further comprising: receiving sensor array data from at least one sensor array from the entity or another entity wearing the wearable device interacting with the sensor array; storing the first sensor data and the second sensor data received while the sensor array data is received; and training the at least one machine-learning model based on the sensor array data, the first sensor data received while the sensor array data is received, and the second sensor data received while the sensor array data is received.

18. The method of claim 10, further comprising: receiving, with the at least one computing device, third sensor data from at least two sensors arranged on at least one second wearable device configured to be worn on the appendage or at least one other appendage of the entity, wherein the at least one scratch metric is determined based on inputting the first sensor data, the second sensor data, and third sensor data from the at least two sensors of the at least one second wearable device into the at least one machine-learning model.

19. A computer program product comprising at least one non-transitory computer-readable medium including program instructions that, when executed by at least one computing device, cause the at least one computing device to: receive first sensor data from at least one accelerometer arranged on a wearable device configured to be worn on an appendage of an entity; receive second sensor data from at least one motion sensor arranged on the wearable device, wherein the first sensor data and the second sensor data are different; and determine at least one scratch metric based on inputting the first sensor data and the second sensor data into at least one machine-learning model.

Images