WIRELESS MEDICAL SENSORS AND METHODS - Patent application

Abstract

Provided herein are medical sensors and related methods for measuring real-time personal metrics. The medical sensor may include an electronic device having a sensor comprising an accelerometer and a two-way wireless communication system electronically connected to the electronic device for transmitting output signals from the sensor to an external device and for receiving commands from the external controller to the electronic device.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of and priority to U.S. Provisional Patent Application No. 62 / 710,324, filed February 16, 2018, U.S. Provisional Patent Application No. 62 / 631,692, filed February 17, 2018, and U.S. Provisional Patent Application No. 62 / 753,203, filed October 31, 2018, each of which is incorporated by reference herein to the extent not inconsistent.

[0002] Provided herein are medical sensors including mechanical acoustic sensing electronics coupled with an on-board microphone and feedback stimuli, including but not limited to vibration motors, speakers, or LED indicators. Provided are systems and methods for electronics that use a three-axis high-frequency accelerometer to sense mechanical acoustic electrophysiological signals originating from the body. These devices are referred to herein as soft, flexible wearables with advanced power saving features and wireless communication capabilities, including compatibility with Bluetooth®-enabled systems. Within the system are signal processing, signal analysis, and machine learning capabilities that provide a platform for multi-modal sensing of a wide range of physiological and environmental signals, including but not limited to voice, speech time, respiratory rate, heart rate, lung capacity, swallowing function, physical activity, sleep quality, movement, and eating behavior. These systems and methods are compatible with the use of additional sensors, including one or more of an on-board microphone, pulse oximeter, ECG, and EMG (among others). [Background technology]

[0003] Mechanoacoustic signals are known to contain essential information for clinical diagnostics and healthcare applications. In particular, mechanical waves propagating through the body's tissues and fluids resulting from natural physiological activities reveal characteristic signatures of individual events, such as the closure of heart valves, contraction of skeletal muscles, vibration of the vocal cords, the respiratory cycle, scratching movements and sounds, and motility of the gastrointestinal tract.

[0004] The frequencies of these signals may range from a fraction of a Hz (e.g., respiratory rate) to 2000 Hz (e.g., speech) and often have low amplitudes above the hearing threshold. Physiological auscultation is typically performed using an analog or digital stethoscope in a separate procedure performed during a clinical examination.

[0005] An alternative approach relies on accelerometers in traditional rigid electronics packages that are typically physically secured to the body with straps to provide the necessary mechanical coupling. Demonstrations of research results include recording of phonocardiograms (PCG, sounds from the heart), vibratory cardiograms (SCG, chest vibrations induced by the beating heart), ballistocardiograms (BCG, recoil movements accompanying responses to cardiovascular pressure), and sounds accompanying breathing.

[0006] In the context of cardiovascular health, these measurements provide important insights that complement those inferred from electrocardiogram measurements (ECG): for example, structural defects in heart valves manifest as mechanoacoustic responses and do not show up directly in the ECG tracing.

[0007] Previously reported digital measurement methods, while useful for laboratory and clinical studies, have the following disadvantages: (i) their form factor (rigid design and large size, e.g., 150 mm×70 mm×25 mm) limits the choice of mounting locations, compromising their practicality as wearables; (ii) their bulky structure entails physical mass that suppresses subtle movements associated with important physiological events through inertial effects; (iii) their mass density and elastic modulus are different from the skin, thereby creating an acoustic impedance mismatch with the skin; and (iv) they cannot capture, for example, ECG and PCG / SCG / BCG signals simultaneously, resulting in only a single mode of operation; (iv) their method for communication and data transmission to the user interface is via wires tethered to the device and user interface machinery; and (v) their power management is through a wired connection. The devices and methods provided herein address these limitations in the art. [Prior art documents] [Non-patent literature]

[0008] [Non-Patent Document 1] BMJ2018:360:K207 Summary of the Invention [Problem to be solved by the invention]

[0009] Provided herein are methods and devices that provide a telemedicine type platform, where medical sensors attached to or implanted in a user provide useful information on which a caregiver, such as a medical professional, friend, or family member, can act. These devices and methods are useful for diagnostic or treatment applications, but can also be used for training and rehabilitation. This is reflected in devices and systems that provide two-way communication, where information is sent externally to a caregiver requesting action, and commands can be received by the medical sensors, including instructing the user to perform appropriate actions, including swallowing, inhaling, exhaling, and the like.

[0010] The devices and systems can provide real-time output, such as novel clinical metrics, novel clinical markers, and information useful for informative endpoints, thereby improving the overall health and well-being of the user. These devices and systems are particularly suited to utilize off-site cloud storage and analytics that can conveniently, reliably, and easily trigger clinician or caregiver action.

[0011] The special configuration of hardware, software, bidirectional information flow, and remote storage and analysis realizes a fundamentally improved platform for healthcare and welfare in a relatively unobtrusive and mobile manner that is not tied to traditional clinical sites (e.g., limited to hospitals or controlled environments). In particular, software, which may be embedded in a chip or processor, may be on-board or remote to the devices described herein, providing greatly improved sensor performance and clinically actionable information. Machine learning algorithms are particularly useful for further improving device performance. [Means for solving the problem]

[0012] Specifically included herein are the appended claims and any other portions of the specification and drawings.

[0013] In one aspect, provided is a medical sensor comprising: a) an electronic device having a sensor comprising an accelerometer; and b) a two-way wireless communication system electronically connected to the electronic device for transmitting output signals from the sensor to an external device and for receiving commands from an external controller to the electronic device.

[0014] The medical sensor may be wearable, tissue-mounted or implantable, or in mechanical or direct mechanical communication with the subject's tissue. The medical sensor may include a wireless power system for powering the electronic device. The medical sensor may include a processor for providing real-time metrics. The processor may be on-board the electronic device or positioned in an external device located at a distance from the medical sensor and in wireless communication with the wireless communication system. The processor may be part of a portable smart device.

[0015] The medical sensors may continuously monitor and generate real-time metrics, such as social or clinical metrics. For example, clinical metrics may be selected from the group consisting of swallowing parameters, breathing parameters, aspiration parameters, coughing parameters, sneezing parameters, temperature, heart rate, sleep parameters, pulse oximetry, snoring parameters, body movements, scratching parameters, bowel movements parameters, neonatal subject diagnostic parameters, cerebral palsy diagnostic parameters, and any combination thereof. For example, social metrics may be selected from the group consisting of conversation time, word count, vocalization parameters, verbal discourse parameters, dialogue parameters, sleep quality, feeding behavior, physical activity parameters, and any combination thereof.

[0016] The medical sensor may include a processor configured to analyze the output signal. The processor may utilize machine learning to customize the analysis for each individual user of the medical sensor. The machine learning may include one or more supervised and / or unsupervised learning algorithms that are customizable for the user. The machine learning may improve sensor performance parameters and / or personalized user performance parameters used for diagnostic sensing or therapeutic applications.

[0017] The processor described herein may be configured to filter and analyze the measured output from the electronic device to improve the sensor performance parameters. The medical sensor may include a wireless power supply system for wirelessly powering the electronic device. The accelerometer may be a three-axis high frequency accelerometer.

[0018] The electronic device described herein may comprise a stretchable electrical interconnect, a microprocessor, an accelerometer, a stimulator, resistors and capacitors in electronic communication to provide vibration or motion sensing by the accelerometer and stimulation by the stimulator to a user. The sensors may sense multiple or single physiological signals from the subject, and thresholds are used to send triggers for correction, stimulation, biofeedback or reinforcement signals back to the subject.

[0019] The electronic devices described herein may include a network comprising multiple sensors, for example, one sensor may be for sensing the physiological signal from the subject and one sensor may be for providing a feedback signal to the subject.

[0020] The threshold may be personalized for the subject. The stimulator may include one or more of a vibration motor, an electrode, a light emitter, a thermal actuator, or an audio notification.

[0021] The medical sensor described herein may further comprise a flexible encapsulation layer surrounding the flexible substrate and the electronic device. The encapsulation layer may include a lower encapsulation layer and an upper encapsulation layer, and a strain insulating layer, the strain insulating layer being supported by the lower encapsulation layer, and the flexible substrate being supported by the strain insulating layer. There may be an air pocket between the electronic device and the upper encapsulation layer. The medical sensor may be configured such that there is no air pocket between the electronic device and a lower layer of the device that is close to or in contact with a tissue surface of a subject.

[0022] The medical sensor may have a device mass of less than 1 g, less than 500 mg, less than 400 mg, or optionally less than 200 mg, and a device thickness of less than 10 cm, less than 6 mm, less than 5 mm, or optionally less than 3 mm.

[0023] The medical sensor described herein may be configured for therapeutic swallowing applications, social interaction meters, stroke rehabilitation devices, or respiratory therapy devices. The medical sensor may be configured to be worn by a user and used in therapeutic swallowing applications, and the output signal is a signal for one or more swallowing parameters selected from the group consisting of swallow frequency, swallow count, and swallow energy. The medical sensor may further include a stimulator that provides a haptic signal to the user to engage in safe swallowing. Safe swallowing may be determined by sensing the beginning of inspiration and expiration of the user's respiratory cycle. One or more machine learning algorithms may be used in a feedback loop for optimizing the haptic signal timing.

[0024] The medical sensor described herein may be configured to be worn by a user and used as a social interaction meter, with the output signal being a signal for one or more social parameters selected from the group consisting of: conversation time, word count (fluency rate), vocalization parameters, linguistic discourse parameters, or dialogue parameters. The medical sensor may be configured to be worn on the suprasternal notch of a user. The medical sensor may be a sensor for use with one or more additional user well-being parameters selected from the group consisting of sleep quality, eating behavior, and physical activity, with the social and well-being parameters of the medical sensor being combined to provide a social interaction metric.

[0025] The medical sensor may further comprise a stimulator that provides a haptic signal to the user to engage in a social interaction event. The medical sensor may be configured for use with a stroke rehabilitation device to be worn by a user, and the output signal is a signal for a social parameter and / or a swallowing parameter. The medical sensor may be a sensor for use with one or more additional stroke rehabilitation parameters selected from the group consisting of gait, falls, and physical activity. The medical sensor may comprise a stimulator that provides a haptic signal to the user to engage in a safe swallowing event.

[0026] The medical device may be configured for wear by a user and for use in a respiratory treatment device, and the output signals are for inhalation and / or exhalation, i.e., effort, duration, or airflow through the throat. The medical device may include a stimulator that provides tactile signals to the user to engage in breathing exercises.

[0027] The medical devices described herein may include an external sensor operably connected to the electronic device. The external sensor may include a microphone and / or a mouthpiece.

[0028] The medical sensors described herein may be capable of recreating an avatar or animated representation of a subject's body position and movements over time.

[0029] In one aspect, provided is a method of measuring real-time personal metrics using any of the medical sensors described herein.

[0030] In one aspect, provided is a method of measuring real-time personal metrics, the method including the steps of a) wearing any of the above claimed devices on a user's skin surface or implanting subcutaneously; b) detecting a signal generated by the user with a sensor; c) analyzing the filtered signal thereby classifying the filtered signal; and d) providing real-time metrics to the user or a third party based on the classified filtered signal.

[0031] The described method may include filtering the detected signal prior to the analyzing step. The providing step may include one or more of providing a tactile stimulus to the user, storing or displaying clinical metrics, and / or storing or displaying social metrics. The providing step may further include storing the real-time metrics on a remote server for subsequent analysis to generate a clinician or caregiver action. The action may include sending a command to the medical sensor.

[0032] The real-time metrics may be health-related mental, physical, or social metrics. The analyzing step may include use of machine learning algorithms. The machine learning algorithms may include independent supervised learning algorithms, each independently trained to provide personalized real-time metrics specific to an individual user.

[0033] The personalized real-time personal metric may be a metric for a therapeutic or diagnostic application. The therapeutic or diagnostic application may be selected from the group consisting of safe swallowing, respiratory therapy, cerebral palsy diagnosis or treatment, and neonatal diagnosis or treatment.

[0034] The real-time personal metrics may be for medical applications selected from the group consisting of sleep medicine, dermatology, respiratory medicine, social interaction assessment, speech therapy, dysphagia, stroke rehabilitation, nutrition, obesity treatment, fetal monitoring, neonatal monitoring, cerebral palsy diagnosis, prenatal monitoring, bowel function, sleep disorder diagnosis or treatment, sleep therapy, trauma, trauma prevention falls or hyperextension of joints or limbs, trauma prevention during sleep, small arms / ballistic related injuries, and cardiac output monitoring therapy.

[0035] In one aspect, provided is a medical sensor comprising an electronic device having a sensor comprising an accelerometer and a wireless communication system electronically connected to the electronic device.

[0036] The wireless communication system may be a two-way wireless communication system. The wireless communication system may be a system for transmitting output signals from the sensor to an external device. The wireless communication system may be a system for receiving commands from an external controller to the electronic device.

[0037] The medical sensors described herein may be wearable or implantable. The medical sensors may include a wireless power system for powering the electronic device. The medical sensors may include a processor for providing real-time metrics. The processor may be on-board the electronic device or positioned in an external device located at a distance from the medical sensor and in wireless communication with the wireless communication system. The processor may be part of a portable smart device.

[0038] The medical sensors described herein may continuously monitor and generate real-time metrics. The real-time metrics may be social or clinical metrics. The clinical metrics may be selected from the group consisting of swallowing parameters, breathing parameters, aspiration parameters, coughing parameters, sneezing parameters, temperature, heart rate, sleep parameters, pulse oximetry, snoring parameters, body movements, scratching parameters, bowel movements parameters, and any combination thereof.

[0039] The social metric may be selected from the group consisting of conversation time, word count, vocalization parameters, linguistic discourse parameters, dialogue parameters, sleep quality, eating behavior, physical activity parameters, and any combination thereof.

[0040] The medical sensors described herein may include a processor configured to analyze the output signal. The processor may utilize machine learning to customize the analysis for each individual user of the medical sensor. The machine learning may include one or more supervised and / or unsupervised learning algorithms that are customizable for the user. The machine learning may improve sensor performance parameters and / or personalized user performance parameters used for diagnostic sensing or therapeutic applications.

[0041] The described sensors may be disposed on or near the subject's suprasternal notch. The described sensors may be disposed on or near the subject's mastoid process. The described sensors may be disposed on or near the subject's neck. The described sensors may be disposed on or near the subject's lateral neck. The described sensors may be disposed under or near the subject's chin. The described sensors may be disposed at or near the subject's jaw line. The described sensors may be disposed on or near the subject's clavicle. The described sensors may be disposed on or near the subject's bony prominence. The described sensors may be disposed behind the subject's ear.

[0042] The described electronic devices may include one or more three-axis high frequency accelerometers. The described electronic devices may include a mechanical acoustic sensor. The described electronic devices may include one or more of an on-board microphone, an ECG, a pulse oximeter, a vibration motor, a flow sensor, and a pressure sensor.

[0043] The described electronic devices may be flexible and / or stretchable devices. The described electronic devices may have a multi-layer floating device architecture. The described electronic devices may be at least partially supported by an elastomeric substrate, superstrate, or both. The described electronic devices may be at least partially supported by a silicone elastomer that provides strain isolation.

[0044] The described electronic devices may be at least partially encapsulated by a moisture resistant enclosure. The described electronic devices may further include an air pocket.

[0045] The wireless communication system described herein may be a Bluetooth communication module. The wireless communication system described herein may be powered by a wireless rechargeable system. The wireless rechargeable system may include one or more of a rechargeable battery, an induction coil, a full-wave rectifier, a regulator, a charging IC, and a PNP transistor.

[0046] The medical sensors described herein may include a gyroscope, for example a three-axis gyroscope. The medical sensors described herein may include a magnetometer, for example to measure the electric field generated by the patient's breathing. The medical sensors described may be worn near the patient's suprasternal notch.

[0047] In one aspect, provided is a device comprising an electronic device having a sensor comprising an accelerometer and a two-way wireless communication system electronically connected to the electronic device for transmitting output signals from the sensor to an external device and for receiving commands from an external controller to the electronic device, the sensor sensing multiple or single physiological signals from a subject that provide the basis for one or more corrective, stimulating, biofeedback, or reinforcement signals provided to the subject.

[0048] The corrective, stimulating, biofeedback, or reinforcement signals may be provided by one or more actuators. The one or more actuators may be thermal, optical, electrotactile, auditory, visual, tactile, or chemical actuators operably connected to the subject. The device may comprise a processor for feedback control of the one or more corrective, stimulating, biofeedback, or reinforcement signals provided to the subject.

[0049] The physiological signals or a single physiological signal may provide an input for said feedback control. The feedback control may include a thresholding step to trigger said one or more corrective, stimulating, biofeedback, or reinforcement signals provided to the subject. The thresholding step may be accomplished by dynamic thresholding.

[0050] In one aspect, provided is a device comprising: an electronic device having a multi-mode sensor system comprising a plurality of sensors, the sensors comprising an accelerometer and at least one sensor that is not an accelerometer; and a two-way wireless communication system electronically connected to the electronic device for transmitting output signals from the sensors to an external device and for receiving commands from an external controller to the electronic device.

[0051] The sensor system may include one or more sensors selected from the group consisting of optical sensors, electronic sensors, thermal sensors, magnetic sensors, optical sensors, chemical sensors, electrochemical sensors, fluid sensors, or any combination thereof. The sensor system may include one or more sensors selected from the group consisting of pressure sensors, electrophysiological sensors, thermocouples, heart rate sensors, pulse oximetry sensors, ultrasonic sensors, or any combination thereof.

[0052] In one aspect, provided is a device comprising an electronic device having a sensor comprising an accelerometer and one or more actuators operatively connected to the sensor, the sensor sensing multiple or single physiological signals from a subject that provide a basis for one or more corrective, stimulating, biofeedback, or reinforcement signals provided to the subject by the one or more actuators.

[0053] The one or more corrective, stimulating, biofeedback, or enhancing signals may be one or more optical signals, electronic signals, thermal signals, magnetic signals, chemical signals, electrochemical signals, fluid signals, visual signals, mechanical signals, or any combination thereof.

[0054] The one or more actuators may be selected from the group consisting of thermal actuators, optical actuators, electrotactile actuators, auditory actuators, visual actuators, haptic actuators, mechanical actuators, or chemical actuators operably connected to the subject. The one or more actuators may be one or more stimulators. The one or more actuators may be heating devices, light emitters, vibration elements, piezoelectric elements, sound generating elements, tactile elements, or any combination thereof.

[0055] A processor may be operatively connected to the electronic device and the one or more actuators, the processor providing feedback control of the one or more corrective, stimulating, biofeedback, or reinforcement signals provided to the subject, and a plurality or a single physiological signal may provide an input for the feedback control.

[0056] The feedback control may include a thresholding step to trigger the one or more corrective, stimulating, biofeedback, or reinforcement signals provided to the subject. The thresholding step may be accomplished by dynamic thresholding.

[0057] The described device may include a two-way wireless communication system electronically connected to the electronic device for transmitting output signals from the sensor to an external device and for receiving commands from an external controller to the electronic device. Corrective, stimulating, biofeedback, or reinforcement signals may be provided to the subject for training or therapy. The training or therapy may be for breathing or swallowing training.

[0058] The described device may continuously monitor and generate real-time metrics. The real-time metrics may be social or clinical metrics. The clinical metrics may be selected from the group consisting of swallowing parameters, breathing parameters, aspiration parameters, coughing parameters, sneezing parameters, temperature, heart rate, sleep parameters, pulse oximetry, snoring parameters, body movement, scratching parameters, bowel movements parameters, neonatal subject diagnostic parameters, cerebral palsy diagnostic parameters, and any combination thereof. The social metrics may be selected from the group consisting of conversation time, word count, vocalization parameters, verbal discourse parameters, dialogue parameters, sleep quality, feeding behavior, physical activity parameters, and any combination thereof.

[0059] The described devices may include a gyroscope, for example a three-axis gyroscope.The described devices may include a magnetometer.

[0060] In one aspect, provided is a method of diagnosis using any of the devices or sensors described herein.

[0061] In one aspect, provided is a method of training a subject using any of the devices or sensors described herein.

[0062] Additionally, the sensor configurations provided may be used in combination to provide more accurate measurements or metrics. For example, an accelerometer may be used in combination with a mechanical acoustic sensor to measure a user's scratching. While scratching motions may be detected by the accelerometer, other common movements (e.g., waving, typing) may be difficult to distinguish from scratching. Incorporating an acoustic sensor close to the skin allows for secondary classification to be performed, improving data collection.

[0063] Differential measurement of separate regions of the patient's body can also help improve data collection and data accuracy. In some cases, a single device can measure two different regions by being positioned on a biological boundary, and in some cases, multiple devices may be used. For example, placing a device on the suprasternal notch allows for both chest and neck acceleration measurements. During breathing, there is a high degree of movement in the chest while the neck is relatively stationary. This allows for more robust measurements and evaluations using the devices described herein.

[0064] Without wishing to be bound by any particular theory, there may be herein a description of a belief or understanding of the underlying principles relating to the devices and methods disclosed herein. Regardless of the ultimate correctness of the mechanical explanations or hypotheses, an embodiment of the present invention may still be operable and useful. [Brief description of the drawings]

[0065] [Figure 1] FIG. 1 is an exemplary schematic exploded view of an epidermal mechanical acoustic electrophysiological measurement device. [Diagram 2] FIG. 2 illustrates an example of a wearable (e.g., epidermally mounted) mechanoacoustic electrophysiological measurement device as presented in FIG. 1. [Diagram 3] FIG. 1 illustrates a device cross-section of a wearable epidermal mechanical acoustic electrophysiological measurement / treatment device including thickness and elastic modulus information. [Figure 4] FIG. 1 is a side view of an exemplary epidermal mechanical acoustic electrophysiological measurement / treatment device illustrating the various layers described herein. A thin film (300 mm) of low and high modulus silicone (Ecoflex Smooth-on, E=60 kPa) surrounds the electronics without physical contact. [Diagram 5] FIG. 1 is an exemplary sensing circuit diagram of an epidermal mechanical acoustic electrophysiological measurement / treatment device. [Figure 6] FIG. 1 is an exemplary charging circuit diagram for an epidermal mechanical acoustic electrophysiological measurement / treatment device. [Figure 7] 1A-1C illustrate examples of adhesive configurations useful for establishing contact between an epidermal mechanical acoustic electrophysiological measurement / treatment device and a surface (e.g., tissue, skin). [Figure 8] FIG. 13 presents an exemplary user interface for a processor that presents data from epidermal mechano-acoustic electrophysiological measurements of the vocal cords (e.g., speaking) and swallowing, including vibration and acceleration, and provides real-time metrics. [Figure 9] FIG. 1 presents data from superficial mechanoacoustic electrophysiological measurements of the vocal folds measuring speech through acceleration. [Figure 10] FIG. 1 presents data from superficial mechanoacoustic electrophysiological measurements of the vocal folds including speaking and swallowing. [Figure 11] FIG. 13 presents a flow chart for epidermal mechano-acoustic electrophysiological function with an external mouthpiece operating as a wireless spirometer with Bluetooth communication with an external device. [Figure 12] FIG. 13 presents a flow chart for epidermal mechanical voice electrophysiological function with external mouthpiece illustrating cloud storage connectivity and machine learning algorithms. [Figure 13] FIG. 1 presents a flow chart for epidermal mechanical voice electrophysiological function with an external mouthpiece utilizing machine learning algorithms. [Figure 14] FIG. 1 presents an example flowchart for supervised machine learning and signal processing that may be used with various devices described herein. [Figure 15] FIG. 1 presents an exemplary flowchart of user (e.g., patient) treatment and analysis that may be used with various devices described herein. [Figure 16] FIG. 1 presents an example flowchart for enhanced machine learning and signal processing that may be used with various devices described herein. [Figure 17]FIG. 13 presents an example flowchart for supervised machine learning and signal processing that may be used with various devices described herein, including the use of thresholding on social interaction scores. [Figure 18] FIG. 1 presents an example flowchart for unsupervised machine learning and signal processing that may be used with various devices described herein. [Figure 19] FIG. 1 illustrates a wireless connection between an epidermal mechano-acoustic electrophysiological measurement / treatment device and a processor (e.g., smartphone, tablet, laptop, etc.). [Figure 20] FIG. 1 illustrates that the devices described herein are not affected by the surrounding environment. [Figure 21] FIG. 1 illustrates the system's ability to identify specific interlocutors and quantify conversation time in a group of three stroke survivors with aphasia and one speech therapist. [Figure 22] FIG. 1 presents an example of a raw data signal collected by an epidermal mechanical acoustic electrophysiological measurement / treatment device. [Diagram 23] FIG. 13 presents exemplary data of on-body cardiac signals collected from the lateral neck. [Figure 24] FIG. 13 presents exemplary data of on-body cardiac signals collected from the lateral neck. [Diagram 25] FIG. 1 presents exemplary data of on-body respiratory signals collected from the lateral neck. [Figure 26] FIG. 1 presents exemplary data of an on-body cardiac signal collected from the suprasternal notch. [Figure 27] FIG. 1 presents exemplary data of an on-body cardiac signal collected from the suprasternal notch. [Figure 28] FIG. 1 illustrates an exemplary configuration for measuring patient scratching using an epidermal mechanical acoustic electrophysiological measurement / treatment device. [Figure 29]FIG. 29 presents exemplary experimental results for measuring patient scratching using the device presented in FIG. 28. [Figure 30A] FIG. 1 is a schematic diagram illustrating potential mounting locations on a subject's body (outlined by superimposed boxes). [Figure 30B] Photographs and schematic diagrams illustrating device placement on a subject's body, including locations near the lateral neck and near the suprasternal notch. [Figure 30C] 1 presents example signals for X, Y, and Z directions corresponding to subject activities including breath-holding, sitting and talking, leaning, walking, and jumping. [Figure 30D] 1 presents example signals for X, Y, and Z directions corresponding to subject activities including breath-holding, sitting and talking, leaning, walking, and jumping. [Figure 31A] 4 is a flow diagram corresponding to a signal processing approach for analysis of a three-axis accelerometer output. [Figure 31B] 1 presents an exemplary signal corresponding to a subject's activity. [Figure 32A] FIG. 1 presents exemplary data of respiratory rate GS versus MA for a range of subjects. [Figure 32B] FIG. 1 presents exemplary data of heart rate GS vs. MA corresponding to subjects. [Figure 32C] FIG. 1 presents exemplary data of conversation time GS vs. MA corresponding to a range of subjects. [Fig. 32D] FIG. 1 presents exemplary data of swallow number GS vs. MA for a range of subjects. [Diagram 33] (A) is a diagram presenting an example signal corresponding to a subject's activity including various configurations of up and down movements of the face and head, and (B) is a diagram presenting an example signal corresponding to a subject's activity including various configurations of up and down movements of the face and head. [Figure 33C] 1 is a plot of rotation angle versus time (min). [Figure 33D]A plot of heart rate (BPM) versus time (min). [Diagram 34] (A) is a schematic diagram illustrating a research-grade wearable sensor of the present invention incorporating a 3-axis accelerometer, gyroscope, and EMG detector in a multi-layer flexible device format, (B) is a schematic diagram showing multiple wearable sensors (5 in total) placed on different areas on a neonatal subject including the limbs and torso. In one embodiment, the sensors are placed on the neonatal subject's body during a 1-hour clinical visit, and (C) presents the accelerometer and gyroscope data obtained from the sensors. [Figure 35A] FIG. 1 is a schematic diagram of the sensor of this embodiment, showing the EMG and accelerometer modules and the Bluetooth communication module. [Figure 35B] FIG. 13 presents examples of data acquired from the sensors including acceleration, reconstructed 3D motion, and EMG. [Figure 35C] FIG. 13 presents examples of data acquired from the sensors including acceleration, reconstructed 3D motion, and EMG. [Figure 35D] FIG. 13 presents examples of data acquired from the sensors including acceleration, reconstructed 3D motion, and EMG. [Diagram 36] 1 is a schematic flow diagram of a method of using the sensors described herein to identify newborn subjects at risk for CP. [Figure 37] 1 is an image of miniaturized flexible accelerometers attached to the limbs and torso of a neonatal subject. [Figure 38] FIG. 13 provides examples of data analytics useful for analyzing the output of sensors of examples of the present invention, for example, for clinical diagnostic applications. [Figure 39] FIG. 13 is a plot showing differences in motor data between infants at risk for CP and typically developing infants using 20 different features extracted from the motor data at 12 weeks of age. [Diagram 40]FIG. 1 presents results for a study of wearable sensors on children with cerebral palsy (ages 24 months and under) compared to age-matched controls, i.e., development of a new early detection tool. [Diagram 41] FIG. 1 presents an example of a sensing system comprising a sensor in communication with a portable electronic device for creating social interaction scores and metrics, including those based on psychometric surveys, based at least in part on validated scales and questionnaires and physical health parameters derived from sensor signals and / or measured characteristics. [Diagram 42] 1 presents exemplary data for the use of the sensor system of the present invention to monitor advanced physical performance metrics for cardiac output. [Diagram 43] FIG. 1A is a diagram of a mechanical acoustic device showing the flexibility of the device, and FIG. 1B is a diagram of a mechanical acoustic device showing an exploded view of the floating device architecture. [Figure 43C] FIG. 1 is a diagram of a mechanical acoustic device and illustrates the operational architecture of a wireless system. [Fig. 43D] FIG. 1 is a diagram of a mechanoacoustic device and simulated deformation of the device at a system level. [Figure 43E] FIG. 1 is a diagram of a mechanical acoustic device, showing variations of the device at a system level. [Figure 44A] FIG. 13 shows sample triaxial accelerometer data acquired from a single MA device worn around the neck of a healthy subject, showing a total of 60 seconds of data capturing a range of biological activities. [Figure 44B] FIG. 1 shows sample triaxial accelerometer data obtained from a single MA device worn around the neck of a healthy subject, with sample time series, spectrograms, and spectral information for heart rate, speech, swallowing, and gait signals. [Figure 45A]FIG. 1 shows signal processing of MA data obtained from a field study of healthy subjects. Block diagram of post-processing analytics for energy intensity (EI), heart rate (HR), respiratory rate (RR), swallow count (SC), and talking time (TT). [Figure 45B] FIG. 13 shows signal processing of MA data obtained from a field study of healthy subjects, illustrating detection of heartbeat peaks as local maxima of the 20-50 Hz bandpass waveform. [Figure 45C] FIG. 13 shows signal processing of MA data acquired from a field study of healthy subjects, illustrating decoupling chest wall motion from triaxial measurements and zero-crossing node counting for RR estimation. [Figure 45D] FIG. 1 shows signal processing of MA data obtained from a field study of healthy subjects, showing that speech signals are characterized by high-quality harmonics of the fundamental frequency in the range of 85 to 255 Hz for a typical adult. [Figure 45E] FIG. 1 shows signal processing of MA data obtained from a field study of healthy subjects. Speech and movement signals are shown after zeroing. A broadband swallowing event is detected when both high- and low-band signals simultaneously exceed the calm limit. [Figure 46A] Bland-Altman analysis for HR. In FIG. 46A, and in FIGS. 46B-D below, the solid and dashed lines represent the mean difference and standard deviation x 1.96, respectively. HR has a mean difference of -3.12 bpm and a standard deviation of 5.43 bpm. RR has a mean difference of 0.25 breaths per minute and a standard deviation of 2.53 bpm. TT has a mean difference of -2.00 s / min and a standard deviation of 2.17 s / min. SC has a mean difference of -0.65 counts per 10 counts and a standard deviation of 2.68 counts per 10 counts. Different colors represent different healthy subjects. [Figure 46B] FIG. 1 shows Bland-Altman analysis for RR. [Figure 46C] FIG. 1 shows Bland-Altman analysis for TT. [Figure 46D] FIG. 1 shows Bland-Altman analysis for SC. [Figure 47A] FIG. 1 illustrates the application of mechanoacoustic sensing in sleep studies, showing an image of the device on the sternum along with the gold standard sleep sensor ensemble including electrocardiogram (ECG), pressure transducer airflow (PTAF), abdominal strain gauge, thoracic strain gauge, thermistor, electroencephalography (EEG), and electro-oculography (EOG). [Figure 47B] FIG. 1 shows an application of mechanoacoustic sensing in sleep studies, demonstrating body orientation detection using 3-axis acceleration data. [Figure 47C] FIG. 1 illustrates an application of mechano-acoustic sensing in sleep studies, comparing heart rate measurements from a mechano-acoustic sensor with electrocardiogram (EKG) measurements during sleep. [Figure 47D] FIG. 1 illustrates the application of mechano-acoustic sensing in sleep studies, comparing respiration rate measurements from mechano-acoustic sensors with nasal pressure transducer airflow (PTAF) and thoracic strain gauge measurements during sleep. [Figure 47E] FIG. 1 illustrates an application of mechano-acoustic sensing in sleep studies, showing a comparison of body orientation measurements from a mechano-acoustic sensor with visual inspection. [Fig.47F] FIG. 1 shows the application of mechano-acoustic sensing in sleep studies. FIG. 2 shows the inference of sleep stages based on HR and RR values ​​from accelerometer compared with clinical test sleep stages. [Figure 47G] FIG. 1 illustrates the application of mechano-acoustic sensing in sleep studies, showing cumulative distribution functions as a function of heart rate and respiration rate. [Fig. 47H] FIG. 1 illustrates an application of machine-acoustic sensing in sleep studies; and FIG. 2 illustrates an exemplary interface for summary statistics. [Figure 48] FIG. 1 presents an exemplary wavelet cross-spectral analysis. [Figure 49A]13A-13C are diagrams of simulations demonstrating wavelet cross-spectral analysis for differential mode signal extraction. [Figure 49B] FIG. 13 is a diagram showing zero-crossing node counts on sample data. [Figure 50A] FIG. 1 shows accelerometer measurements of resting breathing signals compared to electrocardiography (ECG) measurements. [Figure 50B] FIG. 1 shows HR measurements compared to polar monitor measurements. Cardiac amplitude shows a linear correlation with HR measurements. [Figure 51A] FIG. 1 presents exemplary experimental data from a group of 10 subjects. [Figure 51B] FIG. 1 presents exemplary experimental data from a group of 10 subjects. [Figure 52] FIG. 13 presents an example of three-dimensional body orientation detection using a device as described herein. [Figure 53A] FIG. 13 presents exemplary heart rate and respiration rate data correlating with body orientation measurements for subject 1. [Figure 53B] FIG. 13 presents exemplary heart rate and respiration rate data correlated with body orientation measurements for subject 2. [Figure 54A] FIG. 13 shows the optimized mechanical design of the interconnect. Schematic diagram of a double-layer serpentine interconnect with an arc angle of 270°. [Figure 54B] FIG. 13 illustrates an optimized mechanical design of the interconnect, showing the relationship between arc angle and elastic extensibility for a pre-compressed serpentine interconnect and a planar serpentine interconnect. [Figure 55A] FIG. 13 shows a simulation of the deformation of a system-level device, showing 40% compression before yielding. [Figure 55B] FIG. 13 shows a simulation of the deformation of a system-level device, showing a 160° bend before yielding. [Figure 56]1 illustrates the effect of tensile deformation and strained isolation layers in a system level device. Strain isolation minimizes the resistance of silicone substrate deformation from rigid islands of electronics compared to a system without an insulating layer. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0066] Generally, the terms and phrases used herein have their art-recognized meanings, which can be found by reference to standard textbooks, periodical literature, and contexts known to those of ordinary skill in the art. The following definitions are provided to clarify their specific use in the context of the present invention.

[0067] "Mechanical sound" refers to any sound, vibration, or movement by a user that can be detected by an accelerometer. Thus, the accelerometer is preferably a high-frequency three-axis accelerometer that can detect a wide range of mechanical sound signals. Examples include breathing, swallowing, organ (lung, heart) movement, movement (scratching, exercise, moving), speaking, bowel activity, coughing, sneezing, and the like.

[0068] "Bidirectional wireless communication system" refers to an on-board component of the sensor that provides the ability to send and receive signals. In this manner, an output can be provided to an external device, including a cloud-based device, a personal portable device, or a caregiver computer system. Similarly, commands can be sent to the sensor by an external controller or the like, which may or may not be compatible with the external device. Machine learning algorithms can be used to improve the signal analysis, and then a command signal can be sent to the medical sensor, including the stimulator of the medical sensor to provide a tactile signal to the user of the medical device to aid in treatment. More generally, these systems can be incorporated into a processor, such as a microprocessor, located on-board the electronic device of the medical sensor or physically separate from the electronic device.

[0069] "Real-time metrics" is used broadly herein to refer to any output that is useful in medical welfare. It may refer to social metrics that are useful in understanding the social well-being of a user. It may refer to clinical metrics that are useful in understanding or training biological functions, such as breathing and / or swallowing.

[0070] "Customized machine learning" refers to analysis of sensor output tailored to an individual user. Such a system recognizes personal variability between users, with criteria including medical condition (stroke vs. dementia), weight, baseline fluency, resting respiratory rate, basal heart rate, etc. Specifically tailoring the analysis to the individual user significantly improves the sensor output and what is done downstream by the caregiver. This is generally referred to herein as improving "sensor performance parameters." Exemplary parameters include, for example, precision, repeatability, fidelity, and classification accuracy.

[0071] "Near" refers to a location in close proximity to another element and / or a location on a subject, such as a human subject. For example, in one embodiment, near is within 10 cm, optionally for some applications within 5 cm, and optionally for some applications within 1 cm, of another element and / or a location on a subject.

[0072] In some embodiments, the inventors' sensor systems are wearable and tissue-mounted or implantable, or in mechanical or direct mechanical communication with a subject's tissue. As used herein, mechanical communication refers to the ability of the sensors of the present invention to directly or indirectly interface with the skin or other tissue in a conformable, flexible, direct manner (e.g., no air gaps) that allows for greater insight and enhanced sensing capabilities while reducing motion artifacts compared to accelerometers that are strapped to the body (wrist or chest) in some embodiments.

[0073] Various embodiments of the present technology generally relate to sensing and physical feedback interfaces, including "mechano-acoustic" sensing. More specifically, some embodiments of the present technology relate to systems and methods for mechano-acoustic sensing electronics configured for use in respiratory diagnostics, digestive diagnostics, social interaction diagnostics, skin inflammation diagnostics, cardiovascular diagnostics, and human machine interfaces (HMIs).

[0074] Physiological mechanoacoustic signals, often with frequencies and intensities above values ​​associated with the audible range, can provide information of high clinical utility. Although conventional packaged stethoscopes and digital accelerometers can capture some relevant data, neither is suitable for use in a continuous wearable mode, typical of non-stationary environments, and both have drawbacks associated with mechanical transduction or signals through the skin.

[0075] Various embodiments of the present technology include a soft, conformable, extensible class of devices that maximize detectable signals, enable multi-modal operation, such as electrophysiological recording, and neurocognitive interaction, and can be used in almost any part of the body, and are specifically configured for mechano-acoustic recording from the skin.

[0076] Experimental and computational studies highlight the critical role of low effective elastic modulus and low areal mass density for effective operation in this type of measurement mode on the skin. Demonstrations with vibratory cardiac measurements and heart murmur detection in a series of cardiac patients illustrate the utility in advanced clinical diagnostics. Monitoring pump thrombosis in ventricular assist devices provides an example of mechanical implant characterization. Tracking the swallowing behavior of healthy subjects with respect to the respiratory cycle provides a new understanding of natural physical behavior. Measuring motion and listening to sounds of the respiratory, circulatory and digestive systems, and even typical motions such as scratching simultaneously with a single device, provide a whole new dimension of pathological diagnosis. Voice recognition and human-machine interfaces represent additional proven applications. These and other possibilities suggest widespread use for soft, skin-integrated digital technologies capable of capturing acoustics from the human body. Physical feedback systems integrated with sensors add additional therapeutic functions to the device.

[0077] In the following description, for purposes of explanation, numerous specific details are set forth to provide a thorough understanding of embodiments of the present technology. However, it will be apparent to those skilled in the art that embodiments of the present technology may be practiced without some of these specific details. For convenience, embodiments of the present technology are described with respect to cardiovascular diagnosis, respiration and swallowing correlation, and scratch strength detection, however, the present technology also provides numerous other applications in a wide variety of potential technical fields.

[0078] The technology introduced herein may be embodied as dedicated hardware (e.g., circuitry), as a programmable circuit appropriately programmed by software and / or firmware, or as a combination of dedicated and programmable circuitry. Thus, the embodiments may include a machine-readable medium storing instructions that can be used to program a computer (or other electronic device) to execute a process. The machine-readable medium may include, but is not limited to, a floppy diskette, an optical disk, a compact disk read-only memory (CD-ROM), a magneto-optical disk, a ROM, a random access memory (RAM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic or optical card, a flash memory, or other types of media / machine-readable media suitable for storing electronic instructions.

[0079] The phrases "in some embodiments," "according to some embodiments," "in the illustrated embodiment," "in another embodiment," and similar phrases generally mean that the particular feature, structure, or characteristic that follows the phrase is included in at least one implementation of the technology of the present invention, and may be included in multiple implementations. Additionally, such phrases do not necessarily refer to the same or different embodiments.

[0080] FIG. 1 illustrates an exploded view of one example of a medical device 10, such as an epidermal mechanoacoustic electrophysiological measurement device, in accordance with some embodiments of the present technology.

[0081] In this exemplary embodiment, the epidermal mechanical acoustic electrophysiological measurement device comprises a lower elastomeric shell 20, a silicone strain insulating layer 30, a stretchable interconnect 40, electronic devices 50 such as microprocessors, vibration motors, resistors, capacitors, and the like, and an upper elastomeric shell 60.

[0082] 2 illustrates an example of a wearable (e.g., epidermally attached) mechanical acoustic electrophysiological measurement device according to some embodiments of the present technology. This example assembly includes the example epidermal mechanical acoustic electrophysiological measurement device of FIG. 1 together with a stimulator, such as a vibration motor.

[0083] The technology of the present invention provides a different type of mechano-acoustic electrophysiological sensing platform that utilizes the most advanced concepts in flexible and stretchable electronics to enable soft conformable integration with the skin without the need for wired connections to the device. This technology allows for accurate recording of physiological vital signals in a manner that avoids many of the limitations of previous technologies (e.g., heavy and bulky packaging) with the freedom of application environment. The mechano-acoustic modality includes a miniaturized low-power accelerometer with high sensitivity (16384 LSB / g) and wide frequency band (1600 Hz), which may have enhanced functional limitations. A soft strain-insulating packaging assembly, along with electronics for electrophysiological recording and active feedback systems, represents another exemplary feature of these stretchable systems. An exemplary embodiment of the present technology has a mass of 300 mg (or less than 600 mg, or between 100 mg and 500 mg), a thickness of 4 mm (or between about 3 mm and 5 mm), and an effective modulus of 100 kPa (in x and y directions) (or between about 50 kPa and 200 kPa), which correspond to values ​​orders of magnitude lower than those previously reported. In this manner, any of the medical devices provided herein may be described as conformable, including conformable to the skin of a user. Such physical device parameters allow for wear over extended periods of time without being overly conformal.

[0084] Exemplary embodiments of the present technology provide qualitative improvements in measurement capabilities and wearability with a fully wireless form factor capable of wirelessly transmitting, communicating, and powering, in a format that can interface with nearly any area of ​​the body, including the curvilinear portion of the neck to capture signals associated with breathing, swallowing, and speech. The following description and figures illustrate the properties of this technology and highlight its usefulness in a wide range of examples, from human body studies on patients to personal health monitoring / training devices with customizable applications.

[0085] Specific data shows simultaneous recording of gait, breathing, cardiac activity, respiratory cycle, and swallowing. Also, the vibroacoustics of ventricular assist devices (VADs) (i.e., devices used to augment failing myocardial function, which is often exacerbated by in-device clot formation) are captured and can be used to detect pump thrombosis or device malfunction.

[0086] In addition, there are applications of speech recognition and classification for human-machine interfaces in a mode that captures laryngeal vibrations without interference from noise in the surrounding environment. Fundamental studies on the biocompatibility of the skin interface as well as the mechanical properties and fundamental aspects of the interface coupling provide additional insight into the operation of the technology of the present invention.

[0087] Also, the device's ability to interact with the patient through stimulation capabilities integrated into the sensors allows it to become a therapeutic device. The wireless form factor and availability of the device for personal and even clinical use allows for the collection of large amounts of data. With machine learning, the device uses stimulation not only as an output based on scheduled time points, but also as an input for the study of mechano-acoustic signals related to physiological responses.

[0088] FIG. 3 shows a device cross-section of an example medical device including a wearable epidermal mechanical acoustic electrophysiological measurement / treatment device according to some embodiments of the present technology including thickness and elastic modulus information.

[0089] The layers of an exemplary epidermal mechanical acoustic electrophysiological measurement / treatment device are described in more detail with reference to FIG. 4. The lower elastomeric shell includes a 100 μm layer of Sylvion with a modulus of elasticity of 100 kPa. The silicone gel layer between the device and the shell includes a 50 μm layer of Sylvion gel with a modulus of elasticity of 5 kPa. The stretchable interconnect includes a bilayer of 18 μm thick serpentine copper traces, encapsulated between two 12 μm layers of polyimide (PI), each with a modulus of elasticity of 2.5 GPa. The electronic device is bonded to the stretchable interconnect and then covered with an upper elastomeric shell, including a 100 μm layer of Sylvion that includes an air pocket between the electronics and the upper elastomeric shell.

[0090] The fabrication process involves five parts: (i) producing a flexible PCB (fPCB) device platform, (ii) chip bonding onto the fPCB device platform, (iii) casting the top and bottom elastomeric shells from a mold, (iv) layering Silvion gel, and (v) bonding the top and bottom elastomeric shells together.

[0091] The fabrication process is now described in more detail: (i) A photolithography and metal etching process, or a laser cutting process, defines the pattern of the interconnects in the copper. A spin-coating and curing process produces a uniform layer of PI over the resulting pattern. Photolithography and reactive ion etching (RIE, Nordson MARCH) define the top, middle, and bottom layers of PI with geometries that match the interconnect geometries. (ii) A chip bonding process assembles the electronic components required to operate the device. (iii) A pair of concave and convex dies for each of the top and bottom elastomeric shells defines the shape of the device's outer structure. (iv) The recessed area in the bottom shell contains a layer of Silvion gel for both adhesion and strain isolation purposes of the device platform. (v) A curved thin top elastomeric membrane shell is bonded to a flat bottom elastomeric shell to package the electronic components with an air pocket.

[0092] FIG. 5 illustrates a sensing circuit diagram of an example epidermal mechano-acoustic electrophysiological measurement / treatment device, in accordance with some embodiments of the present technology.

[0093] The sensing circuitry includes a mechanical acoustic sensor (BMI160, Bosch), a coin motor, and a Bluetooth-enabled microcontroller (nRF52, Nordic Semiconductor). The sensor has a frequency range (1600 Hz) that lies between various target frequencies of respiration, heart, scratching, and vocal cord movement and sound. Additional sensors in the platform may include, but are not limited to, an on-board microphone, ECG, pulse oximeter, vibration motor, flow sensor, and pressure sensor.

[0094] FIG. 6 illustrates a charging circuit diagram of an example epidermal mechanical acoustic electrophysiological measurement / treatment device, in accordance with some embodiments of the present technology.

[0095] The wireless charging circuit includes an induction coil, a full-wave rectifier (HSMS-2818, Broadcom), a regulator (LP2985-N, Texas Instruments), a charging IC (BQ2057, Texas Instruments), and a PNP transistor (BF550, SIEMENS).

[0096] The device can also be coupled to an external component, such as an external mouthpiece for measuring lung volume. The mouthpiece houses a diaphragm, the deflection of which is associated with a particular pressure. The amount of deflection of the membrane using the device defines the amount of air volume transferred during exhalation.

[0097] For a healthy adult, the first sound (S1) and the second sound (S2) of the heart have acoustic frequencies of 10 to 180 Hz and 50 to 250 Hz, respectively. The vibration frequency of the human vocal cords is in the range of 90 to 2000 Hz. The average fundamental frequency is about 116 Hz (males, average age 19.5 years), about 217 Hz (females, average age 19.5 years), and about 226 Hz (children, 9 to 11 years) during dialogue. To allow detection of cardiac motion and speech, the cutoff frequency of the low-pass filter is 500 Hz. The high-pass filter (cutoff frequency, 15 Hz) removes motion artifacts.

[0098] Low frequency respiratory cycles (0.1-0.5 Hz), cardiac cycles (0.5-3 Hz), and snoring signals (3-500 Hz) each have their own unique frequency bands. By passing these specific frequency bands for each of these biomarkers, the filter removes high frequency noise and low frequency motion artifacts.

[0099] Apart from the frequency band of the present invention, many other mechanical and acoustic biosignals are measured from the raw data (eg scratching movements (1-10 Hz), scratching sounds (15-150 Hz)).

[0100] 8-10 show examples of superficial mechanoacoustic electrophysiological measurements of the vocal cords (eg, speech) and swallowing with several processing algorithms of the present technique, including signal filtering and automated analysis.

[0101] Signal processing algorithms including, but not limited to, Shannon energy transform, moving average smoothing, Savitsky-Golay smoothing, and auto-thresholding set up faster analysis of large amounts of data.

[0102] General signal processing involves seven parts: (i) collecting raw data, (ii) filtering the data, (iii) normalizing the filtered data, (iv) energy transformation of the data, (v) smoothing and enveloping the data, (vi) thresholding, and (vii) masking the data.

[0103] Signal processing is now described in more detail: (i) capturing the raw acceleration signal without the use of analog filters provides multiple signals that are superimposed on each other; (ii) filtering of the data in various bands of the frequency spectrum separates the raw signal into multiple layers of signals specific to different biomarkers; (iii) normalization of each filtered data allows a reasonable comparison of each signal; (iv) transforming the normalized filtered signal simplifies the signal to all positive values. When observing signals above the DC frequency range, the signal fluctuates across the zero baseline. For information related to, but not limited to, the duration of speech, coughing, or swallowing, a true measurement is possible in the energy interpretation of the signal; (v) smoothing the data includes the normalized filtered signal to represent the measured signal in a simpler way; (vi) using a histogram, or an automatic thresholding algorithm, specific activities can be determined and classified; (vii) using the picked threshold, the number of samples associated with the activity by the mask is determined.

[0104] The wavelet transform method simply extracts signals related to specific activities such as speaking, laughing, coughing, or swallowing. The scale and time information from the transform is used to classify the specific characteristics of swallowing for specific types of food intake, as well as types of communication and interaction.

[0105] Supervised machine learning of labeled signals involves two parts: (i) labeling the signals with activities by time-stamping the data with the time when the event occurred, and (ii) multi-class classification methods, including but not limited to random forest methods.

[0106] Such classification produces classifications for specific incidents of breathing patterns (inhalation, exhalation), swallowing specific types of food (liquids, solids), and human-machine interfaces for vocal cord vibration recognition.

[0107] We now describe the human interface in more detail. The coin cell learns trends in breathing cycles and swallowing incidents in healthy individuals, and activates to cue appropriate swallowing times based on breathing cycles for people with swallowing difficulties. It also measures vocal cord motion and frequency, and learns which letters and words are associated with specific vibrations.

[0108] Subjective studies involving social meters utilize unsupervised learning, which involves dimensionality reduction methods such as Latent Dirichlet to obtain predictors, and then clustering methods, including but not limited to k-modes and DBSCAN, to classify specific groups of people who share similar behavior of the signals.

[0109] The reinforcement learning seeks to correlate the clinical outcomes of the treatments given by the device's user interface. The implementation of the reinforcement learning is carried out towards the end of the classification and pilot study set.

[0110] FIG. 5 illustrates a sensing circuit diagram of an example epidermal mechano-acoustic electrophysiological measurement / treatment device, in accordance with some embodiments of the present technology.

[0111] The system may employ any of a variety of two-way communication systems, including those that support the Bluetooth® standard, to connect to any standard smartphone (Figure 19), tablet, or laptop. This is a secure user interface for both the consumer and the researcher. The data measurements are compatible with HIPAA-compliant data transfer and cloud storage--we already use Box® as a HIPAA-compliant storage platform for our wireless sensors. Further signal analysis work allows for classification of other relevant behaviors for individuals suffering from AD, such as personal hygiene activities (brushing teeth), housework, or driving. This signal processing and further machine learning based on the output of the sensor can be deployed either on the device itself, on the smartphone, or in a cloud-based system.

[0112] On-board memory allows maximum flexibility in a wireless environment even without having a user interface machine linked to the device for data streaming or storage.

[0113] Going beyond the use of traditional adhesives, we propose a novel skin-device interface that incorporates an adhesive that can withstand up to two weeks of continuous wear. Rather than requiring the user to adhere and completely remove the sensor to the skin, especially in the sensitive neck area, our device can be attached and detached using magnets. Other attachment mechanisms can involve buttons, clasps, hooks, and loop connections. The adhesive that adheres to the skin can vary in type (e.g., acrylic, hydrogel, etc.) and can be optimized for the desired length of skin adhesion (Figure 7).

[0114] Wireless Features:

[0115] Communication with a user interface machine that displays, stores, and analyzes data is commonly known. Here, in contrast, we present a sensor technology that has an on-board processor, data storage, and communicates with a user interface via a wireless protocol such as Bluetooth®, or an ultra-wideband or narrowband communication protocol, which in some embodiments can optionally provide secure transmission. In this way, the device can be utilized in a natural setting without the need for an external power source.

[0116] The device is powered by inductive coupling and can also communicate and / or transfer data via Near Field Communication (NFC) protocol. When a user utilizes the device in an enclosed environment, such as a bed while sleeping, or in a hospital environment, power and data transmission can occur via an inductive coil resonating at 13.56 MHz. This allows for continuous measurements without the need for an on-board battery or external power source.

[0117] The wireless battery charging platform enables a fully encapsulated device that isolates the electronics from the surroundings, preventing substances that would otherwise damage the sensor. The encapsulation layer is made from a thin film of a polymer or elastomer, such as a silicone elastomer (Silbione RTV 4420). Such an encapsulation layer is significantly less permeable than the polydimethylsiloxane and ecoflex described in the prior art.

[0118] Advanced Signal Processing

[0119] Digital Filtering: (Finite Impulse Response) FIR and Infinite Impulse Response (IIR) types of digital filters are used as appropriate. Specific frequency bands are selected to reduce the effects of artifacts and noise and maximize the signal of interest, while specific time windows are automatically selected within regions with high signal-to-noise ratios.

[0120] Algorithms for signal-specific analysis: One method involves processing the filtered signal in the time domain. A particular event of interest (e.g., speaking vs. coughing vs. scratching) is better revealed from the raw output of an acousto-mechanical sensor when the signal of interest is filtered in the appropriate frequency band. Using energy information generated from the acceleration of the sensor, information such as the duration of a discrete event or the number or frequency of events is better calculated. Another processing technique in our system uses power frequency spectrum analysis to evaluate the power distribution of each frequency component. This allows for the derivation of additional information from the raw signal (e.g., pitch from audio).

[0121] Machine Learning

[0122] Supervised Learning: Supervised machine learning of labeled signals involves two parts: (i) labeling the signals with activities by time stamping the data with the time when the event occurred, and (ii) multi-class classification methods, including but not limited to random forests. Such classification produces classifications for specific incidents of breathing patterns (inhalation, exhalation), swallowing specific types of food (liquids, solids), and human-machine interfaces for vocal cord vibration recognition.

[0123] For example, using the scale and time information from the transform, we can classify specific swallow characteristics related to the food content eaten (e.g., thin liquids like water, thick liquids, soft foods, or regular foods) through supervised machine learning, a process that does not require the same degree of time or frequency ambiguity as the Fast Fourier Transform.

[0124] The human interface will now be described in more detail. The coin cell learns trends in breathing cycles and swallowing incidents in healthy individuals and activates to cue appropriate swallowing times based on breathing cycles for people who have lost the ability to time their swallows with their breath. Also, sensors measure vocal cord movement and frequency and learn the letters and words associated with each particular signal.

[0125] Unsupervised learning: This is accomplished without labeled signal input. In the case of the wearable social interaction meter, we employ unsupervised learning. This includes dimensionality reduction methods such as Latent Dirichlet to obtain features relevant to quantifying social interactions. This includes features of voice (tone, pitch), physical activity, sleep quality, and conversation time. Clustering methods (e.g., k-mode and DBSCAN) then classify certain groups of signals into categories.

[0126] Reinforcement learning: This involves the sensor system learning the effect of tactile stimuli on swallowing and then measuring actual swallowing events along with breathing. This allows the system to self-tune and calibrate to ensure that the measured swallowing events correspond to ideal timing within the respiratory cycle.

[0127] Personalized "physical" biomarkers

[0128] Combining high fidelity sensing, signal processing and machine learning allows the creation of novel metrics that can serve as physical biomarkers of health and well-being.For example, quantifying spontaneous swallowing during the day has already been shown to be an independent measure of swallowing dysfunction.Therefore, the sensor provided herein can be used to calculate a swallowing function score that is sensitive to small but clinically meaningful changes within the patient's natural environment.

[0129] The timing of swallowing with respect to the respiratory cycle (inhale, exhale) is important in avoiding problems such as aspiration, which can lead to choking or pneumonia. The ability to time swallowing is largely under involuntary control leading to coordinated efforts between breathing and swallowing. However, in diseases such as stroke or head / neck cancer where radiation is administered, this coordination is lost. Our sensor could then quantify the swallowing event in the context of the respiratory cycle and provide a measure of a "safe swallow". A social interaction score could also be created via signal processing and machine learning to create a total score of social activity. This could be used as a threshold to engage the caregiver, or loved one, in increasing daily social interactions when baseline thresholds are not met. These are examples of how novel metrics can be derived from the sensor system to enable patient behavioral changes or clinician and caregiver interventions.

[0130] Therapeutic Wearable Sensors

[0131] In this disclosure, advanced capabilities are presented for sensor systems that are useful for therapeutic purposes, whereas previous work has focused only on diagnostic applications.

[0132] Two example therapeutic applications are described herein. First, safe swallow timing allows for the prevention of dangerous events such as aspiration, which may lead to choking, pneumonia, or even death. Our sensor can be transformed into a therapeutic swallow primer that triggers a user swallow based on sensing the onset of inspiration and expiration of the respiratory cycle. This allows the sensor to trigger swallowing in a safer part of the respiratory cycle (typically in mid-to-late expiration). Furthermore, machine learning algorithms can be used to optimize the timing of the trigger in a feedback loop. For example, the sensor can track both respiratory rate and swallowing behavior. A trigger is issued that is timed to lead to a swallowing event within an ideal respiratory timing window. To trigger swallowing in this embodiment, we propose a vibration motor that provides direct tactile feedback. Other trigger mechanisms can include visual notification (e.g., light-emitting diode), electrical impulse (e.g., electrodes), temperature notification (e.g., thermistor). In some embodiments, for example, the system is configured to provide sensors that detect one or more parameters used as the basis of input for a feedback loop involving a signal generating device component that provides one or more signals to a subject (e.g., a patient), such as a vibration signal (e.g., an electromechanical motor), an electrical signal, a thermal signal (e.g., a heating device), a visual signal (e.g., an LED or a full graphic user interface), an audio signal (e.g., an audible tone), and / or a chemical signal (elution of a skin-perceptible compound such as menthol or capsaicin). In such embodiments, the feedback loop is executed for a specified time interval based on measurements by the sensor, and one or more signals are provided to the subject periodically or repeatedly based on the sensed parameters. The feedback approach may be implemented using machine learning, for example, to provide an individualized response based on the measured parameters specific to a given subject.

[0133] In one embodiment, on-body sensing is accomplished by enclosed sensing / stimulation circuitry enabled through real-time processing, and feedback loops may be tactile, electrotactile, thermal, visual, audio, chemical, etc. In one embodiment, sensors may also operate in a network, where anatomically separate sensing functions allow for more information, allowing one sensor to measure (e.g., on the suprasternal notch) but trigger feedback on a sensor located elsewhere (e.g., on the chest) that is relatively hidden.

[0134] The second treatment modality is for the sensor to operate as a wearable respiratory therapy system. In conditions such as chronic obstructive pulmonary disease (COPD), dyspnea or shortness of breath is a common symptom that significantly impacts quality of life. Respiratory therapy is a commonly deployed method in which subjects train to control their breathing (both timing and respiratory effort) to increase lung aeration and improve respiratory muscle recovery. Our sensor can be used to track inhalation and exhalation effort as well as duration. Based on these measurements, haptic feedback (or visual feedback via LEDs) could potentially train the user to prolong or shorten inhalation or exhalation to maximize airflow. Respiratory inhalation efforts could also be triggered in a similar way. For example, if a certain inhalation effort is achieved, a threshold is passed, triggering a haptic vibration. This haptic feedback could also be triggered after a certain length of time has been reached for the inhalation effort. Thus, the sensor tracks airflow through the throat, which could be used as a way to perform physical breathing exercises. In another embodiment, the sensor itself can be fitted with an external mouthpiece (FIGS. 11-13) that can operate as a wireless spirometer for a training session and then be returned to the throat for normal sensing.

[0135] Another treatment modality involves the use of the sensor system of the present invention to evaluate and optionally treat a patient with respect to positioning the subject's body, or a portion thereof, to prevent injury and / or support a given treatment outcome. Physical injury may occur when a limb moves and displaces to a point of significant deformation. This may occur, for example, when a limb (e.g., shoulder) is injured and must be placed in a relatively immobile state or restricted to a safe range of movement, for example to support healing or treatment. When sleeping or engaging in daily activities, the subject may inadvertently move this limb into a position of deformation that would cause injury. In these embodiments, the sensor of the present invention is used as a sentinel system to evaluate the limb's position in space and provide notifications (either tactile, sound, visual, thermal, chemical, electrical, etc.) to alert the user and / or caregiver.

[0136] Medical Use Cases

[0137] Sleep Medicine: A wireless sleep tracker that can measure time to sleep, wake time after sleep onset, sleep duration, respiration rate, heart rate, pulse oximetry, inhalation time, exhalation time, snoring time, respiratory effort, and body movement. Skin contact over the suprasternal notch allows for capturing respiration rate and heart rate given the proximity to the carotid artery and trachea. As an example, sleep medicine applications can go beyond simply measuring vital signs of sleep or provide metrics of sleep quality. The sensor system of the present invention also supports applications for improving sleep. Examples of applications of this aspect include: 1. During sleep, sensors can detect that the subject has altered vital signs (abnormal vital signs), which may include a combination of increased or decreased heart rate, cessation of breathing rate, decreased pulse oximetry, or snoring (abnormal breathing sounds). This then triggers a feedback mechanism such as vibration, audio, visual, electrical, or thermal that causes the individual to shift position or become aware / wake up. 2. In the case of injury or post-surgical situations, excessive movement or range of motion can lead to aggravation of the injury, especially during periods of unconsciousness such as sleep. Acousto-mechanical sensors, alone or in a network of multiple spatially separated sensors, can detect the limb in space and trigger a feedback mechanism (e.g., vibration, audio, visual, etc.) that notifies the user to return to a safe position or avoid aggravation of the injury. 3. When symptoms (e.g., pain and itch) are difficult to quantify, sleep quality is a surrogate marker of the severity of these symptoms. Sensors can thereby be used to indirectly assess symptoms (e.g., pain or discomfort) by measuring sleep quality. Another novel feature of this aspect of the invention is the use of the sensor to cycle through sleep positions and / or movements over time. This allows for the use of accelerometers on the sensor to reconstruct movement and body position. This allows for direct video feedback to the user, which may visually associate body positions with vital signs or breathing sounds (e.g., snoring). FIG. 41 presents example sensor data for using the inventor's multi-modal sensor in relation to sleep therapy, for example, for the determination of body position and correlation of body position with vital signs and / or breathing sounds.

[0138] In one embodiment, the sensor can assess the position in space for a particular limb or body configuration prone to injury (e.g., post-surgery rotator cuff), and if a dangerous range of motion or position is sensed, this triggers a biofeedback signal to alert the user or have the user change their position to avoid sleeping on the side of the injured arm. The sensor system of the present invention is also useful for monitoring medical treatments related to snoring, for example, where the detection of snoring provides vibration biofeedback that triggers a position change.

[0139] In one embodiment, the sensors are used to repeat a video and / or visual representation of the subject's position in space. Advantages of this aspect of the invention included that it also reduces privacy issues, data storage.

[0140] Dermatology: The ability to capture scratching behavior and distinguish it from other limb movements through combined mechanical and acoustic signal processing.

[0141] Respiratory Medicine: Chronic obstructive pulmonary disease (COPD) is a chronic condition characterized by recurrent pulmonary symptoms. Our sensor can quantify key markers indicative of COPD exacerbations, including coughing, throat clearing, wheezing, altered air volume in forced lung expiration, respiratory rate, heart rate, and pulse oximetry. Asthma and idiopathic pulmonary fibrosis can be similarly assessed with the same scale.

[0142] Quantifying social interaction metrics, acoustic and linguistic features of single-speaker and multi-speaker tasks: Measuring spoken discourse and speech signals as components of social interactions is complex and requires sensors that can capture a wide range of acoustic and linguistic parameters, as well as the acoustic features of the speaking environment. The sensors can quantify key parameters of social interactions related to the incoming acoustic signals, including conversation duration and word count. The recorded signals can be used to extract additional data, including speech features (e.g., F0, spectral peaks, voice onset time, temporal features of speech), and even linguistic discourse markers (e.g., pauses, hesitations). When worn by individual interlocutors, the sensors can capture linguistic features across multiple interlocutors from separately recorded signals, facilitating the analysis of dialogical social interactions. Coupling to the skin along the suprasternal notch allows for accurate quantification of true user conversation duration regardless of ambient conditions. Furthermore, social interactions are complex multifactorial complexes. The present disclosure makes it possible to quantify important physical parameters (e.g., sleep quality, feeding behavior, physical activity) that can potentially be combined into novel metrics for social interactions.

[0143] The sensor system of the present invention is also useful for creating and monitoring social interaction scores and metrics, for example, using approaches based on sensor signals, feedback analysis, and / or signaling to the subject. FIG. 40 presents an example of a sensing system with sensors in communication with a portable electronic device (e.g., smartphone) for creating social interaction scores and metrics, including those based on psychometric surveys, based at least in part on validated scales and questionnaires (using a smartphone in combination with psychosocial health parameters (e.g., conversation time (min / day), voice biomarkers (tone, pitch), dialogue partners (#), GPS location information (from smartphone))) and physical health parameters derived from sensor signals and / or measured characteristics (e.g., steps, sleep quality, eating behavior, etc.). In some embodiments, the sensor output and surveys on the smartphone app are weighted to generate a social interaction score representative of the subject.

[0144] The ability to monitor a wide range of acoustic and linguistic features in an ecologically valid environment is important in identifying individuals at increased risk for mood disorders, at risk for social isolation that can cause increased risk of cognitive decline, and at risk for other diseases marked by early changes in speech, voice, and language quantity / quality (e.g., Alzheimer's dementia, Huntington's disease prodrome, fluency changes in multiple sclerosis, early language changes in Parkinson's disease, among others).

[0145] Acquired neurocognitive and neurolinguistic disorders (e.g., aphasia, cognitive-communicative disorders associated with neurodegenerative disorders with or without dementia, traumatic brain injury, right brain injury), acquired motor speech and fluency disorders, neurodevelopmental disorders, and infant language disorders. The device can also be used in clinical applications in recording speech quantity and quality in hearing loss treatment / hearing rehabilitation applications. The device can also be used to monitor vocal usage patterns in professional users and people with vocal pathologies.

[0146] The sensor systems and methods of the present invention are also useful in the treatment of diseases associated with loss of muscle or nerve function, such as amyotrophic lateral sclerosis, Lambert-Eaton myasthenic syndrome, myasthenia gravis, Duchenne muscular dystrophy, and the sensors can be used to assess the functional performance of a subject, for example, by assessing physical activity, respiratory performance, or swallowing performance in these conditions.

[0147] As noted above, the ability to quantify speech recovery in a wearable format that is not affected by ambient noise conditions will hold high value in assessing the nature and treatment outcomes of numerous disorders associated with dysphonia, speech, language, non-verbal, and cognitive-communicative disorders. Further applications include quantifying the frequency and severity of stuttering in individuals suffering from fluency and fluency-related disorders. Coupling to the skin along the suprasternal notch allows this functionality while minimizing the stigma associated with wearing the device. Recording large amounts of data from an ecologically valid environment is important in improving clinical assessment, monitoring, and intervention options for numerous disorders.

[0148] Dysphagia and Swallowing Problems: Difficulty in swallowing (dysphagia) remains a problem across many medical conditions, including but not limited to head / neck cancer, stroke, scleroderma, and dementia. Previous findings have shown that the frequency of spontaneous swallowing is an independent marker of dysphagia severity. Furthermore, in hospitalized patients, being able to determine the safety and efficiency of swallowing function is crucial to identifying patients at risk of aspiration, facilitating timely discharge and avoiding re-admissions related to aspiration pneumonia by improving dietary habits to optimize nutrition and prevent aspiration. The sensor could potentially act as a screening tool to detect abnormal movements associated with dysphagia and / or potentially guide dietary recommendations. Improvement of dysphagia with therapeutic interventions could also be tracked with the sensor. The application may be applicable to a wide range of age groups, from neonates to the elderly.

[0149] Stroke rehabilitation: As mentioned, the sensor offers the unique ability to assess speech and swallowing function, both of which are important parameters in stroke recovery. Beyond this, the sensor can also measure gait, falls, and physical activity as a comprehensive stroke rehabilitation sensor.

[0150] Nutrition / Obesity: The preferred deployment of the sensor is via skin intimate attachment to the suprasternal notch. This allows for quantification of swallows and swallow counts. As food passes, a unique sensor signature is generated that allows us to make predictions about mealtime and eating behavior. Swallowing mechanics differ based on the density of the food or liquid bolus ingested. Thus, our sensor can detect the ingestion of liquids versus solids. Additionally, our sensor can evaluate swallowing signals that can differentiate between the ingestion of solid foods, denser semi-liquid foods (e.g., peanut butter), or thin liquids (e.g., water). This can be useful for food intake tracking for weight loss. Other applications include assessing food intake in individuals with eating disorders (e.g., anorexia or bulimia). A further application is evaluating mealtime behavior in individuals who have undergone gastric bypass surgery, and the sensor can provide an alert if too much food or liquid is ingested after surgery.

[0151] Maternal / Fetal Monitoring: Currently, ECHO Doppler is the most common modality for capturing fetal heart rate in pregnant women. However, this modality is limited in the sense that capturing fetal heart rate from obese patients can be difficult. Furthermore, the Doppler signal is frequently lost as the fetus descends during labor. Previous work has demonstrated the potential value of mechanoacoustic sensing for fetal heart rate monitoring. Our wearable sensor system is highly advantageous for this application.

[0152] Post-operative surgical monitoring of bowel function: Stethoscopes are commonly used to assess recovery of bowel function after abdominal surgery. Ileus, or failure to recover bowel function, is a common cause of hospitalization or delayed discharge. Sensors that can quantify recovery of bowel function through acoustic signal measurements are useful in this context.

[0153] Cardiology: The stethoscope is the standard of care for diagnosis and disease monitoring. The sensors presented here show the ability to continuously capture data and information derived from the stethoscope. This includes continuous assessment of abnormal heart murmurs. In some cases, such as congenital heart defects, the presence of a heart murmur is critical to the health of the subject. The sensor system of the present invention may provide continuous acoustic measurements of cardiac function. Abnormal sounds also reflect cardiac valve disease. Thus, the sensors herein may be used to track the stability or worsening of valve diseases such as aortic stenosis, mitral stenosis, mitral regurgitation, tricuspid stenosis or regurgitation, or pulmonary stenosis or regurgitation.

[0154] Specific to cardiology, non-invasive methods for assessing cardiac output and left ventricular function are still elusive. Cardiac echocardiography is non-invasive, but requires specialized training and is not conducive to continuous wearable use. A non-invasive method for continuously tracking cardiac output would have high clinical value for many conditions, including congestive heart failure. An embodiment of the sensor system of the present invention can provide a measure of both heart rate and stroke volume (volume of blood pumped per beat). Cardiac output is the product of heart rate and cardiac output. This can be accomplished, for example, by evaluating the peak-to-peak delay time relative to heart rate. The decay of the accelerometer amplitude then represents the strength of each heart beat by measuring the skin displacement at each beat.

[0155] Figure 42 presents example data for the use of the sensor system of the present invention to monitor advanced physical performance metrics for cardiac output. As shown, after intense physical activity, the sensor picks up an elevated heart rate, but also an elevated excursion. When the user returns to baseline, the heart rate and amplitude normalize. This is an example of how amplitude can be used to assess and correlate the amount of blood pumped with each heart beat.

[0156] Another embodiment is in the military. Injuries from firearms or explosions result in the propagation of mechanical waves from the point of impact. Sensors can be used to assess the severity of such impacts as a means of non-invasively assessing a user's bullet impact or proximity to the blast wave. Sensors can also be used to assess the likelihood of damage to vital organs (e.g., placed over the heart or lungs). Sensors may be deployed directly on the user (e.g., police, soldiers), or on clothing, or on bulletproof vests.

[0157] External modifications: Any of the medical devices provided herein may have one or more external modifications, including providing access to new diagnostic and therapeutic functions. For example, the addition of an external mouthpiece allows for a controlled release of airflow from the user that can be measured by sensing elements (e.g., accelerometers or microphones) in the sensor system. This allows for quantification of airflow (volume over time) without the need for expensive equipment such as spirometers. Critical parameters such as forced expiratory volume (FEV1) within one second can then be collected at home, and the data can be wirelessly transmitted and stored. Changes in airflow parameters such as FEV1 can then be combined with other parameters such as wheezing, cough frequency, throat clearing, etc., to create novel metrics of disease that can act as an early warning system of deterioration.

[0158] Therapeutic Applications: In respiratory diseases such as chronic obstructive pulmonary disease (COPD) or asthma, respiratory training is a key component to reduce shortness of breath (dyspnea). This involves teaching breathing techniques such as pursed lip breathing (PLB). This involves breathing out through tightly closed lips and breathing in through the nose with the mouth closed. The length of inhalation and exhalation is also adjusted to suit the patient's unique breathing condition. The length of exhalation and inhalation can be adjusted according to the user's comfort. The sensor can then be deployed in a therapeutic regime where the sensor differentiates between mouth breathing and nose breathing by throat vibration or airflow changes. The sensor can also time the length of inhalation and exhalation. The respiratory therapist can also set an ideal time length, for example, and the sensor can provide haptic feedback to the patient or user when the ideal inhalation or exhalation time length is reached. Overall, the sensor can act as a "wearable" respiratory therapist that reinforces effective breathing patterns and techniques that improve breathing and patient symptoms, preventing worsening of respiratory diseases. Further research could also combine this with continuous pulse oximetry.

[0159] Alzheimer's disease:

[0160] Alzheimer's dementia (AD) affects 5.4 million Americans, costs $236 billion in annual expenditures, and collectively costs 18.1 billion hours of caregiving by loved ones. First, reduced social interaction or loneliness is a key driver of cognitive decline and directly increases the risk of depression in patients with AD. Second, high-quality social interaction is associated with reduced risk of dementia in late life, providing a non-pharmacological strategy to reduce morbidity and mortality in AD. Third, changes in social interaction and dialogue are potential biomarkers for early identification of AD and disease progression. A major barrier in advancing the use of social interaction in AD patients has been the lack of tools that can comprehensively assess the quantity and quality of social interactions in real-world environments. Rating measures of social interaction (self-report / proxy-report) are subject to reporting bias and lack sensitivity. Smartphones have limited sensing accuracy, variability in sensor performance across manufacturers, are unable to measure critical parameters (e.g., mealtime behavior), and suffer from reduced audio fidelity in noisy ambient environments. Although devices for measuring social interactions have been reported in the literature, these systems are bulky and heavy, do not allow for continuous use, and lack the comprehensive sensing capabilities required to adequately capture the entire spectrum of parameters in social interactions. Moreover, these systems have not been rigorously validated in older generations who have low technological literacy.

[0161] To advance care of patients suffering from AD, noninvasive remote monitoring technologies that are acceptable to the wearer and capable of tracking a wide range of parameters related to social interactions across mental, social, and physical health domains are needed. To address this, we propose to develop the first integrated wearable sensor capable of continuous measurement of critical parameters of social interactions within a networked environment that minimizes user stigma through an optimized wearable form factor. The current prototype incorporates a high-frequency triaxial accelerometer that can measure speech, physiological parameters (e.g., heart rate, heart rate variability), sleep quality, mealtime activity, and physical activity (e.g., steps) in an ecologically valid environment through additional signal analytics. The sensor is fully enclosed in medical-grade silicone less than 4 mm thick, with bending and elastic modulus parameters several orders of magnitude lower compared to previously reported technologies. The sensor, adhered to the suprasternal notch with a hypoallergenic adhesive, enables an unobtrusive skin-tight connection, allowing our technology to collect mechano-acoustic signals that are invisible from wristband-based sensors and smartphones. This includes the ability to measure respiratory rate, heart rate, swallowing rate, and conversation time with accuracy unattainable by other technologies. We propose the development of a fully integrated social interaction sensor with additional functionality, rationally designed with input from AD patients and their caregivers, with more advanced signal processing capabilities, and clinically validated against standard equipment. The estimated cost of each sensor is less than US$25, with a total addressable annual market size of $288 million annually. Goal 1 is to add an integrated microphone to our existing wearable flexible sensor platform, which already includes a high frequency 3-axis accelerometer with continuous communication via Bluetooth. Success criteria are successful bench testing showing high fidelity audio capture from a full range of inputs from 38 dB (whisper) to 128 dB (concert), and successful wireless data transfer to a HIPAA secure database. A user interface is available to researchers to allow for more advanced analytics. Pitch, tone Additional parameters such as tone, under-talking, over-talking time, and dialogue turn-taking can be extracted.

[0162] The development of the first truly wearable social interaction sensor capable of continuous, multimodal, and real-world sensing will be a significant innovation for the AD research community as an observational tool, as well as for patients and their caregivers as an interventional tool. By accurately, reliably, and separately capturing multiple parameters related to social interaction, we believe our sensor can detect social isolation in individuals with AD and provide subtle feedback that encourages greater participation and reduces loneliness.

[0163] Alzheimer's dementia (AD) affects 5.4 million Americans and is the sixth most common cause of death, increasing 71% from 2000 to 2013, with annual expenditures reaching $236 billion and a total of 18.1 billion hours of caregiving by loved ones. There are limited therapies (behavioral and pharmaceutical) for AD, with numerous candidates failing in late-stage clinical trials. Advances in the next generation of AD treatments will depend on high-quality clinical measurement tools to detect novel, ecologically valid, and sensitive biophysical markers of cognitive decline. As the search for new treatments continues, there is an immediate need for alternative strategies to bend the disease course by addressing contributing factors and consequences of social interactions that accompany AD. Central to these strategies is the recognition that loneliness and social isolation are serious threats to the health of older adults, which leads to self-harm, self-neglect, cognitive impairment, physical disability, and increased mortality. Addressing modifiable risk factors, particularly social isolation, is a major policy goal of public health agencies and governments to reduce the enormous burden of AD. Numerous rigorous studies support the protective effect of high quality social interaction in mitigating the deleterious effects of AD and optimizing healthy aging (mental, physical, and social). Increased dialogue difficulties, such as breakdowns in message exchange between interlocutors or slower delivery and understanding of messages, emerge early in AD, resulting in heightened social isolation, which accelerates cognitive decline and significantly increases caregiver burden in AD. In addition, as the natural course of AD is characterized by periods of disease stability and punctuated by periods of rapid functional decline, longitudinal measurements of changes in social interaction facilitate a deeper understanding of the natural progression of AD. Dialogue and social interaction behaviors extracted from real-world communication are promising next-generation biophysical markers of cognitive change and treatment outcome measures. Despite their significant clinical importance, dialogue ability and changes in social interaction in real-world contexts are not easily assessed in clinical visits. Clinicians must rely on patient and proxy reports, which are subject to inaccuracies and reporting biases.It would be of great help to the field to develop reliable, non-invasive, user-accepted wearable technology for collecting dialogue and social interaction data.Currently, there is no commercially available technology that can measure a wide range of relevant parameters for social interaction in a form factor that allows long-term real-world use by individuals suffering from AD.Therefore, any of the devices and methods provided herein can be used in AD assessment, diagnosis and treatment.

[0164] Parameters of importance for social interactions (physical, mental, and social): Social interactions are a complex construct. Previous studies have linked social interactions to cognitive function, mental health, sleep quality, physical activity, social activity, feeding behavior, and language use in dementia. Therefore, the assessment of social interactions requires tools that can collect a multitude of behaviors within a natural environment.

[0165] (1) Physical functioning: The domains of physical activity, sleep quality, and mobility are all related to social interactions.

[0166] (2) Vocal features: speech rate, duration, vocal pitch, tone, pauses, intensity, intelligibility, and prosody that reflect aspects of mood and also the source of the dialogue breakdown.

[0167] (3) Mealtime behavior: number of meals eaten using swallow frequency counts, bulimia, or anorexia.

[0168] (4) Dialogue and language behaviour from the person with dementia and their interlocutors: number of turns, duration of turns, overtalking (when one partner talks more than the other), dialogue breakdowns and repairs, staying on topic, word-finding difficulties.

[0169] Assessing adult social interaction collection typically involves self-report and proxy-report psychometric surveys (e.g., Friendship Scale, Yale Physical Activity Scale, SF-36). However, this method of data collection is prone to bias, lacks sensitivity, and is frequently difficult to utilize in individuals with cognitive and language impairments. Furthermore, psychometric survey tools alone do not reflect changes in interaction abilities that often underlie social interaction changes in aging and dementia. Thus, survey tools are best considered in conjunction with objective measures of interaction changes in real-world settings. Smartphones with custom mobile apps have previously been considered for this purpose. Older adults are the least likely to use smarts, indicating low technological literacy. However, smartphones have several advantages, including widespread availability, on-board sensors (e.g., accelerometers, microphones), and wireless communication capabilities. Previous studies have shown that smartphone-collected data (text messaging and phone use) correlate with traditional psychometric mood assessments, but the overall accuracy of these smartphone-based assessments remains poor (<66%). Although there is strong evidence that voice, dialogue, and language features are sensitive markers of changes in mood, cognitive language, and social interaction, smartphone recordings are of insufficient quality for clinical monitoring of these behaviors, especially in real-world situations with high ambient noise. Furthermore, smartphone-based accelerometers for monitoring physical activity and sleep have accuracy issues. The many available smartphone platforms have different hardware specifications, making it impossible to normalize data input. Commercially available wrist-worn systems (e.g., FitBit®) are largely limited to tracking steps and therefore do not capture range data. Remote data recording systems such as LENA offer more advanced signal processing, but they have only been tested in parent-child social interactions, are limited to audio collection only, and have not demonstrated the ability to capture important speech features in AD. For example, measuring “overtalking” times is a valuable tool in determining whether a person is overtalking, as demonstrated by dialogue breakdowns and healthy dialogue partners. Mealtime behavior is a major contributor to negative attitudes toward eating and is important in the context of AD. The most advanced systems reported in the literature for social interaction include both an accelerometer and a microphone in the device attached by a strap. However, the systems are bulky and cannot be worn daily, raise issues of user stigma, and require quiet ambient conditions for operation. Furthermore, these systems are unable to collect relevant physiological parameters for social interaction (e.g., heart rate, heart rate variability, respiratory rate). As mealtime behavior is related to changes in mental health and social interaction, numerous groups have reported wrist- and neck-mounted sensors to measure hand movements and chewing / swallowing behavior, but only with moderate accuracy. These eating behavior sensors lack the ability to collect other relevant parameters such as speech, physical activity, or physiological metrics. Currently, there is an urgent need for technology that can provide objective, comprehensive, and unobtrusive measurements that capture a wide range of parameters important for social interaction for individuals with AD.

[0170] Recent advances in material science and mechanical construction principles have enabled a new class of stretchable, bendable, and soft electronics. These systems match the elastic modulus of skin and can be coupled to any curved surface of the body, allowing mechanically invisible use for up to two weeks. Similar to temporary tattoos, the close coupling to the skin allows physiological measurements with data fidelity comparable to FDA-approved medical devices. In particular, mechano-acoustic signals have high clinical relevance. The propagation of mechanical waves through the body, measurable through the skin, reflects a variety of physiological processes including the opening and closing of heart valves in the chest, the vibration of vocal cords in the neck, and swallowing. Wearable sensors tightly connected to the skin are therefore key to sensing these biosignals and realizing the potential for widespread sensing. This is in contrast to external accelerometers, which are embedded in smartphones and wrist-based versus traditional "wearables", which are limited to measuring only basic physical activity metrics (e.g., number of steps). Described is the use of high-frequency accelerometers coupled to the skin to sense a wide range of parameters relevant to the assessment of social interactions.

[0171] We present a novel mechano-acoustic sensing platform (Figure 19) that incorporates state-of-the-art concepts in stretchable electronics glued to the suprasternal notch that can provide continuous measurement, storage, and analysis of key parameters of social interactions in a distributed network. The mechano-acoustic system incorporates filamentary serpentine copper traces (3 μm) placed between polyimide encapsulation layers that connect miniature chip components. The central sensing unit can be a high-frequency 3-axis accelerometer that can capture low-frequency signals at fractions of a Hz (e.g., step count, breathing rate) to high-frequency signals up to 1600 Hz (e.g., speech), all while operating at ultra-low power consumption. This ability to sample high-frequency signals is in stark contrast to the majority of commercially available accelerometer-based sensors (e.g., Actigraphy / FitBit®), which only operate in the low-frequency band. The resulting device has a mass of 213.6 g, a thickness of 4 mm, an effective elastic modulus of 31.8 kPa (x-axis) and 31.1 kPa (y-axis), and a bending stiffness of 1.02 μNm (x-axis) and 0.94 μNm (y-axis), corresponding to values ​​several orders of magnitude lower than previously reported values ​​and allowing for extended wear. The entire system floats within an ultra-low modulus elastomer core (Silbione RT Gel). Another thin layer of ultra-low modulus silicone (Ecoflex) acts as a shell that reduces contact stresses on the skin surface, maximizing user comfort and water protection.

[0172] This platform provides a system that employs a skin-fitted, high-frequency accelerometer enabled by a low modulus structure and robust adhesion capable of multi-mode operation. The system may communicate with a smartphone using Bluetooth®, with the smartphone acting primarily as a visual display and additional data storage unit. The current system can also optionally interface with the smartphone's sensors, including a microphone, in an additional manner.

[0173] Software and signal analytics for novel data collection related to social interactions: Provided is a set of signal processing functions with bandpass filters of the raw acoustic-mechanical signals within a selected range within the bandwidth of the accelerometer, which allows multimodal sensing of numerous biomarkers, from steps and breathing (low band of the spectrum) to swallowing (mid band of the spectrum) and speech (high band of the spectrum). The close skin coupling allows for highly sensitive measurements with a high signal-to-noise ratio. This allows the sensor to measure both subtle mechanical activities and acoustic biosignals below the threshold for audibility by conventional microphones. We demonstrate that our acoustic-mechanical sensor can be used to detect words (left, right, up, down) by differentiating its time-frequency characteristics from the vocal cord vibrations associated with the creation of each word. This ability can then be used by the sensor to control computer games (e.g., Pac-Man). For the calculation of speech time, the raw acoustic mechanical signal is filtered with an 8th order Butterworth filter. The filtered signal is then passed through a root-mean-square threshold. The energy of the signal is then interrogated in a 50 ms window, allowing for the determination of conversation duration and word count. A short-time Fourier transform defines a spectrogram of the data. The results are averaged and dimensionally reduced using principal component analysis to form feature vectors. Finally, the feature vectors are classified using linear discriminant analysis. We demonstrate the system's ability to identify specific interlocutors and quantify conversation duration in a group of three stroke survivors with aphasia and one speech therapist (Figure 21).

[0174] Another important advantage is that the collection of acoustic and mechanical signals can be synchronously combined to enable capture of wearer-specific talk time in both noisy and quiet ambient conditions. Compared to a smartphone microphone (iPhone 6, Apple, Cupertino), we demonstrate minimal performance difference of our sensor in quiet and noisy ambient conditions. This overcomes a fundamental limitation of other technologies that struggle to capture true user talk time in noisy ambient conditions. Also, a unique ID applied to each sensor allows to identify the number of dialogue partners.

[0175] Beyond acoustic signals, the sensor has the ability to leverage additional analytics to measure other parameters related to social interactions through intimate skin connections. As previously reported in studies employing signal processing strategies from acoustic signals derived from electrocardiograms and stethoscopes, we use Shannon energy calculations to induce higher contrast to prominent mechano-acoustic signatures in the time domain from signal noise. A Savitzky-Golay smoothing function is then applied to form an envelope on the transient energy data. Examples of benefits of the system include measurements of respiratory rate transmitted through the neck and arterial blood pulsation through the external carotid artery, with measurements such as heart rate, heart rate variability, and respiratory rate being relevant to assessing sleep quality (Figures 26-27). The sensor also has the ability to measure simpler sleep quality metrics such as duration, restlessness, and sleep onset. Additionally, our system demonstrates the ability to calculate swallow counts, which can provide direct insight into mealtime behavior and provide surrogate markers for binge eating or anorexia (Figure 10) and even mealtimes. Finally, the sensor can determine the number of steps taken each day as a measure of physical activity that is comparable to existing commercial systems.

[0176] Form factor - reduced burden and stigma for caregivers and wearers: The sensor's flexible platform maximizes user comfort with neck movements, speech, and swallowing. High visibility neck-worn based sensors (necklace and circumferential neck sensors) are another limitation to other published solutions. 79% of respondents expressed significant reluctance and concerns regarding wearing neck-worn based sensors daily. Therefore, a highly wearable sensor capable of capturing the required parameters must minimize potential stigma for the person with AD and their interlocutors. Previous qualitative studies on user acceptance of wearables in AD highlight the importance of low device maintenance, data security, and individuality of wearing. The deployment of the sensor in the suprasternal notch with medical grade adhesive is an important advantage in user acceptability in that it allows the device to be largely covered by a collared shirt while still capturing relevant signals transmitted from the speech production system. The sensor is also encapsulated by silicone that matches the user's skin tone. Finally, the sensor supports full wireless charging and waterproof use, allowing the device to be left in place while taking a bath. With regard to adhesive selection to maximize wearer comfort, we have extensive experience in identifying optimal adhesives that can be adjusted based on the desired duration of use (1 day to 2 weeks). In cases of high fragility of mature skin, we currently employ a mild acrylic polymer matrix adhesive (STRATGEL®, Nitto Denko) that works without causing significant skin irritation or redness even with prolonged daily use (>2 weeks) in healthy adults. In summary, the key advantages of the wearable acousto-mechanical sensor for social interaction compared to existing systems and previous reported studies are:

[0177] Multi-modal capabilities: The sensor has already demonstrated the ability to collect a maximum number of valuable parameters for assessing social interactions in one technology platform made available through skin-tight coupling. Parameters include conversation time, number of dialogue partners, swallow count, respiratory rate, heart rate, sleep quality, and physical activity. Additional parameters are compatible with the devices and methods provided herein.

[0178] Real-world continuous sensing capability: The sensor can measure sound only when mechanical vibrations are sensed by the user's throat, allowing for highly specific recording of true user talk time regardless of noisy or quiet ambient environments. This allows for real-world deployment outside of controlled clinical environments.

[0179] Low-intrusive, unobtrusive form factor: The sensor passively collects data without requiring user adjustments. Wireless charging limits user discomfort and facilitates adherence. Deployment over the suprasternal notch allows for high-fidelity signal capture without the stigmata of highly visible cervical deployment systems.

[0180] Advanced Signal Analytics: Various signal processing techniques can be employed to derive additional meaningful metrics for social interactions.

[0181] The hardware may be employed within a flexible wearable platform. Currently, the central microprocessor has up to eight analog channel inputs with a 2.4GHz 32-bit CPU and 64kB RAM. A commercially available microphone may be used to determine the ideal specifications. Specifically, the MP23AB01DH (STMicroelectronics) series offers a low-profile microphone MEMS system (3.6mm x 2.5mm x 1mm) that does not further increase the wearable form factor. Furthermore, the system consumes low power (250μA), exhibits a sensitivity as low as 38dB, and exhibits a high signal-to-noise ratio (65dB). The microphone can operate in sync with a 3-axis accelerometer to collect external audio signals. The current lithium-ion battery has a capacity of 12mAh. Therefore, we expect that adding an external microphone will not significantly impact battery life. To determine success, microphone performance and audible intelligibility are tested with standardized blocks (60 seconds) of spoken text at increasing decibel (10) levels, ranging from 38 dB (whisper) to 128 dB (concert).

[0182] Software and signal analysis extensions - Bluetooth can be used to connect to a standard smartphone, tablet, or laptop. A user interface can display the raw signal and data storage. The sensor may be used as an observational tool for social interactions, including using a secure user interface focused on the researcher. This includes software protocols that allow for HIPAA-compliant data transfer and cloud storage, but we already use Box as a HIPAA-compliant storage platform for our wireless sensors. While signal processing (Savitzky Golay filtering, Butterworth filtering, Shannon energy envelope techniques) allows for the derivation of many important metrics of social interactions, additional signal processing functions will derive additional, more advanced metrics. For example, paralinguistic features such as pitch, tone, and vocal reaction time of users during interactions are all correlated with depression, including in people with dementia. Turning and overtalking are additional metrics of interest. We propose a multi-pronged approach that includes using a hidden Markov model approach, open access speech processing algorithms (e.g., COVAREP), and wavelet analysis. In particular, we believe that wavelet analysis is the most promising strategy given the established theory of previous studies where mother wavelets for specific metrics of interest are classified from the raw input acoustic-mechanical signals. The user interface allows researchers to freely manipulate the raw data in various ways and deploy various signal processing strategies and toolboxes of interest. Furthermore, signal analysis allows for the classification of other relevant behaviors for individuals suffering from AD, such as personal hygiene activities (brushing teeth), housework, or driving.

[0183] While global wearable medical devices will grow 20% over the next decade to exceed $3 billion, the elderly population is critically underserved despite greater need. The platform provided herein is applicable to a wide range of dementia indications, as well as additional sensing applications (e.g., sleep or dysphagia sensors). Dementia, including AD, is a devastating condition. Increasing meaningful social interactions is an immediate strategy to reduce cognitive decline and morbidity to AD while simultaneously offering a potential prevention strategy in older adults. The wearable medical sensors provided herein have the opportunity to become an important clinical outcome tool for AD researchers by providing the first technology that can comprehensively assess social interactions in natural environments. Additionally, the sensors can directly help individuals and their caregivers, and on days when a person with AD is not being spoken to or meaningfully engaged, the sensors provided herein can notify the appropriate people and reduce feelings of loneliness on that day.

[0184] Example 1: An exemplary epidermal device using mechanical acoustic sensing and actuation capabilities

[0185] An exemplary device employing mechanical acoustic sensing and actuation capabilities has been fabricated and tested for overall functionality and mechanical properties.

[0186] FIG. 43B presents an exploded view of the inventive mechanical acoustic device for epidermal sensing and actuation functions. As shown, the mechanical acoustic sensor includes a silicone gel layer encapsulated within a silicone elastomer substrate and a superstrate (e.g., overlayer) component to provide an overall multi-layer floating device architecture. As shown, the multi-layer device includes an IC component, a power source (e.g., a battery), traces including contact and interconnect components (e.g., flexible serpentine interconnects and contact pads), and an intermediate layer (e.g., a polyimide layer). The multi-layer architecture and device components are arranged to allow effective integration with the subject's tissue (e.g., epidermis) and the ability to undergo deformation without delamination and / or failure. FIG. 43A presents the deployment of the device on a subject's body adjacent to the lateral neck, for example, for speech and / or swallowing monitoring applications. FIG. 43E presents an image showing the ability of the device to deform without failure, for example, via stretching and twisting deformation. Fig. 43D presents a series of schematic diagrams illustrating the ability of a device to incorporate serpentine interconnects to accommodate extensional and torsional deformations without inducing high enough levels of strain that would result in significant device degradation or failure. Fig. 43C presents a schematic diagram showing an embodiment of bidirectional wireless communication, for example, to transmit output signals from a sensor to an external device and to receive commands from an external controller to an electronic device. The schematic diagram also illustrates an embodiment of power provided by wireless charging of a battery, such as a lithium-ion battery, for example, to provide power to a 2.4 GHz Bluetooth wireless communication component.

[0187] Fig. 30A is a schematic diagram illustrating potential mounting locations on a subject's body (schematically shown with superimposed boxes). Fig. 30B is a photograph and schematic diagram illustrating device placement on a subject's body, including a location near the lateral neck and a location near the suprasternal notch. Figs. 30C and 30D present example signals for the X, Y, and Z dimensions corresponding to subject activities including breath-holding, sitting and talking, leaning, walking, and jumping.

[0188] Figure 31A is a flow diagram corresponding to a signal processing approach to the analysis of a three-axis accelerometer output. Figure 31B presents an example signal corresponding to a subject's activity.

[0189] Figure 32A presents exemplary data for respiratory rate GS vs. MA for a range of subjects. Figure 32B presents exemplary data for heart rate GS vs. MA for a range of subjects. Figure 32C presents exemplary data for conversation time GS vs. MA for a range of subjects. Figure 32D presents exemplary data for swallow count GS vs. MA for a range of subjects.

[0190] 33A and 33B present exemplary signals corresponding to subject activity including various configurations of face and head up and down movements. Fig. 33C is a plot of rotation angle versus time (min). Fig. 33D is a plot of heart rate (BPM) versus time (min).

[0191] Example 2: Wearable sensors for early triage of high-risk newborns for CP

[0192] This example demonstrates the utility of the flexible wearable sensor device of the present invention for diagnostic applications including early triage of high-risk neonatal subjects for cerebral palsy (CP). Prediction of final neurological function in high-risk neonates is difficult, and studies have demonstrated that lack of restless movement predicts the onset of CP (see, e.g., BMJ 2018:360:K207). Assessment of CP in neonatal subjects is typically performed by General Motor Assessment (GMA), which corresponds, for example, to a 5-minute video assessment of the infant in a supine position using a standardized rubric.

[0193] In some embodiments, networked sensors provide added value. If limb movements can be assessed - synchronized in time through a network of sensors attached to the body - it is possible to gain deeper insight into abnormal movements. As with sleep, this allows for a visual reconstruction of the movements, which can provide GMA-like video data for future analysis. Advantages here include reduced data storage required, anonymization of subjects, and the ability to operate in low light conditions (e.g., at night and during sleep).

[0194] GMA is the current gold standard with the best available evidence of positive and negative predictive value, but performing GMA requires specialized training that is not always feasible for more widespread screening. 3D computer vision and motion trackers are also potentially useful for GMA, but have the drawbacks of being very expensive, requiring vast amounts of computational power, and requiring large training sets.

[0195] The sensor of the present invention offers an alternative approach that can accurately monitor and analyze the movements of neonatal subjects in real time, thus supporting applications for providing clinically relevant predictive information for the diagnosis of CP.

[0196] Fig. 34A is a schematic diagram illustrating a research-grade wearable sensor of the present invention incorporating a 3-axis accelerometer, gyroscope, and EMG detector in a multi-layer flexible device format. Fig. 34B is a schematic diagram showing multiple wearable sensors (5 in total) placed on different areas on a neonatal subject including the limbs and torso. In one embodiment, the sensors are placed on the neonatal subject's body during a one-hour clinical visit. Fig. 34C presents accelerometer and gyroscope data obtained from the sensors.

[0197] Figure 35A is a schematic diagram of the sensor of this embodiment showing the EMG and accelerometer modules and the Bluetooth communication module. Figures 35B, 35C, and 35D provide examples of data acquired from the sensor including acceleration, reconstructed 3D motion, and EMG.

[0198] FIG. 36 is a schematic flow diagram of a method of using the sensors described herein to identify neonatal subjects at risk for CP. As shown, a miniaturized flexible accelerometer records locomotor activity. A neurologist annotates periods of locomotor activity and whether they are normal from the video recording. The data is uploaded to a server via Bluetooth and a machine learning classifier is trained to detect the presence of abnormal movements based on ground truth labels provided by the clinician. The model is periodically tested and updated / re-refined.

[0199] FIG. 37 is an image of miniaturized flexible accelerometers attached to the limbs and torso of a neonatal subject.

[0200] FIG. 38 provides an example of data analytics useful for analyzing the output of sensors of examples of the present invention, for example, for clinical diagnostic applications.

[0201] FIG. 39 is a plot showing differences in movement data between infants at risk for CP and typically developing infants using 20 different features extracted from the movement data at 12 weeks of age.

[0202] FIG. 40 presents the results for a study of wearable sensors on children (up to 24 months) with cerebral palsy compared to age-matched controls; i.e., the development of a new early detection tool.

[0203] Example 3: Machine-acoustic sensing summary

[0204] Traditional multimodal biosensing requires multiple rigid sensors to be worn at designated locations and scheduled times at multiple measurement sites. A soft conformable device utilizing MEMS acceleration sensors is a game changer for this tradition. It is suitable for use in a continuous wearable operating mode in recording mechanoacoustic signals derived from human physiological activity. The advantage of the device including multiplexed sensing capability is that it can continuously detect approximately 5×10 -3 m·s -2 from minute vibrations of the skin on the order of 20 m s -2This establishes a new area of ​​opportunity for continuous recording of high fidelity signals on the epidermis ranging from large inertial amplitudes of the body and static gravity up to the 800Hz audio band. The minimal spatial and temporal constraints of the device operating beyond the clinical environment will extend the benefits of unconventional mechanisms of electronics. Here, we develop a system-level wireless flexible mechano-acoustic device to record multiple physiological information from a single location, the suprasternal notch. From this unique arrangement, a 3-axis accelerometer simultaneously acquires gait movement, anatomical orientation, swallowing, respiration, cardiac activity, vocal cord vibration, and other mechano-acoustic signals that fit within the bandwidth of the sensor capacity, which are superimposed into a single stream of data. Multiple streamlining algorithms analyze this dense information into meaningful physiological information. Recordings continue for 48 hours. We have also demonstrated the device's ability to measure essential vital signals (heart rate, respiration rate, energy intensity) as well as non-traditional biomarkers (speech duration, swallow count, etc.) from healthy normal subjects in multiple field studies. We validate these results against the gold standard to demonstrate clinical agreement and application in clinical sleep research.

[0205] Introduction The human body continuously generates numerous mechanoacoustic (MA) signals that attenuate at the skin-air interface (1-5). These signals contain important information about the physiological activity of the body and often have intensities and frequencies beyond the range associated with the audible range. These signals include, but are not limited to, vocal cord vibration (about 100 Hz), cardiac activity (about 10 Hz), gait (about 1 Hz), respiration (about 0.1 Hz), and anatomical orientation (about 0 Hz). Traditional health monitoring tools are limited to clinical environments, and therefore the mode of recording continuous physiological activities is rather discrete. In addition, physical conditions in clinical settings may have causal effects from the unnatural environment and output distorted physiological information that does not reflect the subject's natural state (5). Long-term continuous recording of physiological events in the daily environment would provide more true information of the subject. However, it is difficult to have both continuous measurements and high-fidelity signal recording with traditional electronic devices, such as stethoscopes and acceleration sensors (6). The good mechanical coupling of conventional electronics to the skin usually breaks down during natural body movements, resulting in distorted signals. Recent advances in flexible electronics (1,7-11) have enabled high-fidelity measurement of physiological data from the epidermis. Similarly, epidermal mechano-acoustic sensors have a flexible structure with low mass density, allowing for high-fidelity physiological information to be obtained (1). This epidermal mechano-acoustic sensor utilizes an accelerometer that is seamlessly coupled to the skin by a flexible substrate (1). This results in high sensitivity to skin- and body-related motion, but low sensitivity to ambient noise. The soft conformable form factor allows the device to achieve a continuous wearable mode without burdening the skin from mechanical mismatching and its induced stresses. However, the wires that conduct power and communication to the device still reduce these benefits of mechanical isolation from the surroundings. A continuous wearable mode of operation free from spatial and motion constraints is not possible with a wired configuration.

[0206] Near Field Communication (NFC) offers a solution for wireless data and power transmission to wearable sensors through inductive coupling of 13.56 MHz device and transmitter antennas (12, 13). The system has the advantage of operating without batteries, but the problem of limited operating range that depends on antenna geometry and power persists. Bluetooth is another wireless communication mode that allows communication in meter-scale ranges using batteries (14). Thus, by maintaining a connection with a portable hosting device such as a mobile phone, the device operates without space limitations. However, the Bluetooth platform requires relatively large electronic components compared to other ICs and passive components. As a result, the entire device becomes rigid after conventional solid elastomer encapsulation.

[0207] Described herein is a soft, stretchable, wireless mechanoacoustic sensing platform that provides a solution to these challenges and allows continuous monitoring of multimodal physiological information with high fidelity through Bluetooth Low Energy protocol, rechargeable lithium-ion batteries, and air pocket encapsulation that bypasses the effects of rigid and relatively large electronic components. The result is a system-level continuous diagnostic soft electronics with enhanced robustness and no spatial and temporal constraints, which is, as a result, impermeable to water or other foreign bodies. Careful consideration of the measurement site provides a single stream of data with a wealth of physiological information. The suprasternal notch is the location of the notch between the clavicles. Because the neck bridges the circulatory and respiratory systems between the head and torso, there are signals with various intensities and frequencies that are coupled to those physiological systems. The algorithm considers the specific features of each signal and associated events to analyze the single data into multiple physiological information.

[0208] result Device design and circuit considerations The ultra-thin, soft form factor of the wireless mechanical acoustic sensor enables the measurement of mechanical signals from the suprasternal notch with high signal fidelity. Figure 43 (A) highlights the conformable structure of the device, which allows it to deform naturally with large neck movements. The design incorporates stretchable flexible interconnects, strain insulating layers, and soft encapsulation to accommodate large mechanical deformations of the circuitry featuring wireless communication and robust power delivery.

[0209] Figure 43B shows the overall architecture of the system. The electronic platform is a double-sided copper flexible PCB (fPCB) with polyimide as the insulating layer sandwiching the copper layers. The fPCB utilizes annealed rolled copper, which has a fatigue limit 6.5 times higher than conventional electrodeposited copper films (16).

[0210] We designed the electronics around three main components for MA signal acquisition and wireless operation: a 3-axis digital accelerometer sensor (BMI160, Bosch) that measures vibrations over a wide range (±2g), with high resolution (16-bit) and a sampling frequency of 1600Hz, a microcontroller (nRF52832, Nordic Semiconductor) that acquires the data and wirelessly communicates with the user interface via Bluetooth Low Energy (BLE), and a wireless charging circuit that inductively charges a 45mAh Li-ion battery (Figure 43E). The BLE communication protocol works over a distance range of about 2m.

[0211] Although the adoption of commercially available IC components and batteries offers benefits in terms of robustness and production yield, their stiff and bulky structure may inhibit the overall flexibility of the device. To solve this problem, we use serpentine-shaped interconnects to mechanically decouple a small flexible PCB island (1 cm × 1 cm) densely populated with electronic components related to the microcontroller and charging circuitry as shown in Figure 43B. By densely allocating the IC components, the sensor reserves 71% of the total area for flexible interconnects and device edges for deformation absorption. The interconnects are compressed by 10% of their original length in the inactive state (Figure 43C). The pre-buckled structure increases the deformation capability of the device by nullifying the initial 10% tensile elongation (Figure 48). Simulation results show that the pre-buckled serpentine structure with an arc angle of 270° recovers 42% tensile strain, 40% higher than the previously reported design (12), and twists with a twist angle of 90° (Figure 43D). From the simulation, the device may be subjected to 40% compression and 160° bending before yielding (Figure 55).

[0212] We apply a 0.4 mm thick viscoelastic silicone gel with an ultra-low modulus of 6 kPa under the flexible PCB for strain isolation. This insulating layer decouples the rigid non-accelerometer electronics islands from the large in-plane deformations of the substrate up to 40% strain (Figure 56). Figure 56 shows the relationship between strain isolation and skin stress with various gel thicknesses.

[0213] Wireless devices are encapsulated by a silicone elastomer membrane for use in daily activities. As a result, the devices are impermeable to water or other foreign matter. A thin membrane (300 mm) made from silicone (Ecoflex Smooth-on) with low modulus of 60 kPa and high modulus of 500 MPa encapsulates the electronics without physical contact with them (see SI for details). This design aims to minimize hardening effects from the encapsulation. Thin-film encapsulation with hollow cores is achieved with a thickness of 450 mm. 3 Solid silicone encapsulation compared to 68mm 3 The hollow core also allows the serpentine interconnects to deform in a self-supporting manner, providing greater extensibility compared to serpentines that are restricted to in-plane deformation. The hollow encapsulation also contributes to the low mass density of the device and the high sensitivity of the acceleration sensor. Due to the mechanical and material engineering described above, the device is mechanically robust and functions under large deformations, as shown in FIG. 43E.

[0214] In-situ biosignal measurement The soft, conformal, unconstrained contact of the device with the skin is g=9.8m / s 2 is the gravitational acceleration of about 5×10 -4 It allows the measurement of large inertial amplitudes of the body, approximately 2 g, from subtle vibrations of the skin on the order of g / √Hz (Figure 48) to low to high frequencies (0-800 Hz). When worn at the suprasternal notch, which bridges the circulatory and respiratory systems between the head and torso, a single device simultaneously captures not only gravity, but also mechanical motions and acoustic vibrations resulting from the subject's core body motions, heart murmurs, breathing, speaking, and swallowing. Figure 44A presents sample 60-second triaxial acceleration data acquired from a healthy subject, illustrating a series of biological activities such as sitting, talking, drinking water, leaning, walking, and jumping.

[0215] Acceleration signals originating from different physiologies exhibit distinct features in both the time and frequency domains, conveying a rich set of information about the associated biological activity. We focus on z-axis acceleration data, which highlights motion and vibration normal to the surface of the skin. Respiratory activity, manifested as low-frequency chest wall motion, induces changes in the magnitude of gravity projection in all axes. The subject held his breath at approximately the 10-second mark, producing a plateau in the acceleration signal. Quasi-static 3D acceleration provides a gravity vector measure that indicates body orientation. Figure 44B shows detailed features of individual physiological events. The top, middle, and bottom panels show, respectively, a zoomed-in time series, a time-frequency spectrogram, and a sample spectrum of a representative high-frequency (>10 Hz) event. For frequency analysis, we apply a 0.1-second Hanning window with 0.98-second overlap. Cardiac activity--systole and diastole (6)--produces paired pulses with peak amplitudes of approximately 0.05 g, with power concentrated in the 20-50 Hz band. Speech signals are characterized by high-quality harmonics of fundamental frequencies in the range of 85 to 255 Hz for a typical adult. The swallowing event begins with slow (approximately 0.1 s) vocal cord and laryngeal mechanics during the pharyngeal phase and ends with a high-frequency ring-down of water during the esophageal phase [see references]. Walking or jumping movements induce large amplitude (approximately 1 g) impulse forces over a wide frequency range up to approximately 100 Hz.

[0216] Single-device MA measurements stream superimposed information from multiple signal sources. We set up an offline data processing flow utilizing characteristic features, as shown in Fig. 44, to extract biomarkers that may play important roles in clinical and healthcare applications, namely, energy expenditure (EE), heart rate (HR), respiratory rate (RR), swallow count (SC), and speaking time (TT) (Fig. 45A).

[0217] For all filtering processes, we used a fourth-order Butterworth infinite impulse response (IIR) discrete-time filter followed by a non-causal zero-phase filtering approach. We estimate the EE in a 2-s, 50% overlapping time window as the sum of band-limited root-mean-square (BLRMS) of all axial low frequencies (1–10 Hz) [Liu2011]. We estimate the EE in a 0.05g 2 A threshold of s = s + 5 δs is used to classify routinely active versus inactive states, where s ≈ 0.012g and δs ≈ 0.008g are the characteristic mean and standard deviation of the EE measures for the subjects.

[0218] Heart rate analysis begins by bandpass filtering the z-axis acceleration data (f1 = 20Hz, f2 = 50Hz) to suppress noise outside the frequency range of interest. We determined that excessive exertion was detected (EE>0.05g) given a minimum peak height of 0.005g and a minimum peak distance of 0.33 seconds (approximately 180 BPM). 2 ) The algorithm zeros out the signal within a time window and identifies the cardiac pulse as a maxima in the time series of the bandpass signal (Figure 45B). The algorithm filters out peak-to-peak intervals longer than 1.2 seconds (approximately 50 BPM). Averaging is applied to the peak-to-peak intervals over a 5 second time window to obtain a running HR estimate.

[0219] Respiration measurements are sensitive to motion artifacts due to overlap in their frequency domain (0.1-1 Hz). We develop a noise subtraction algorithm that utilizes time-synchronized 3-axis acceleration measurements. That is, for a given device position and orientation (Figures 43A-43B), both z-axis and x-axis measurements are sensitive to chest wall motion, while y-axis acceleration is primarily associated with core body motion. We apply continuous and cross-wavelet transforms to isolate common modes between z-axis and x-axis measurements. xz Extract and then s xz Differential mode between s and y-axis measurement (xz)y'The number N of zero-crossing nodes of the band-pass (f1 = 0.1 Hz, f2 = 1 Hz) signal is used to count the number of inspirations and expirations per minute and estimate RR as N / 2 breaths per minute (BPM) (Fig. 45C, see SI for details).

[0220] Speech signals are distinguishable by the presence of the second harmonic of the fundamental frequency F0 as a maxima of the power spectral density within the range of the human voice (Figure 45D). Swallowing events, on the other hand, are characterized by both low-frequency mechanical motion (0.1-5 Hz) and high-frequency (>100 Hz) acoustic ring-down. After zeroing out speech and loud motion signals, the algorithm simultaneously detects high- and low-frequency signals above its silence-time threshold as swallowing events (Figure 45E, GMMHMM model).

[0221] We test the process flow in two field study schemes: exercise and eating. In the exercise scheme, each subject cycles or rests on an elliptical trainer aiming to span a range of heart rates from 50 BPM to 180 BPM. The algorithm outputs heart rate and compares it to the polar monitor recording every 5 seconds (Figure 50). Subjects manually count the number of respiratory cycles per minute while active. In the eating scheme, each subject speaks and swallows periodically for 5 minutes according to a given speaking time and swallow count. In this scheme, each subject performs five 5-minute tests. Within each minute of the nth test, the subject speaks for n×10 seconds and then swallows for (n+k)×10 seconds, where k=1,...,6-n.

[0222] FIG. 46 shows the Bland-Altman analysis for HR, RR, TT, and SC. The solid and dashed lines mark the mean and 1.96 times the standard deviation of the difference between the mechanoacoustic measurements and the reference standard, respectively. HR has a mean difference of -3.12 BPM and a standard deviation of 5.43 BPM. RR has a mean difference of 0.25 BPM and a standard deviation of 2.53 BPM. TT has a mean difference of -2.00 s / min and a standard deviation of 2.17 s / min. SC has a mean difference of -0.65 counts / 5 min and a standard deviation of 2.68 counts / 5 min. The one-sample Kolmogorov-Smirnov test fails to reject the null hypothesis that the difference data are from a standard normal distribution against the alternative hypothesis that the difference data are not from a standard normal distribution at the 5% significance level for all test parameters.

[0223] sleep research Application to sleep studies demonstrates the usefulness of the device and adaptive algorithms in advanced clinical diagnostics. Figure 47A shows a subject wearing one mechano-acoustic device on the suprasternal notch along with a gold standard polysomnography ensemble including electrocardiogram (EKG), pressure transducer airflow (PTAF), abdominal strain gauge, thoracic strain gauge, thermistor, electroencephalography (EEG), and electro-oculography (EOG). In addition to HR, RR detection, taking advantage of the absence of excessive movement during sleep, the mechano-acoustic sensor monitors body orientation by measuring gravity alone during quiet periods. We demonstrate body orientation detection using 3-axis acceleration data as shown in Figure 52.

[0224] Figure 47C-E compare HR, RR, and body orientation measurements from the golden rule device and the mechanical acoustic device throughout a ~7-hour sleep study in a male subject. Figure 47C compares HR analyzed from a 60-s, 50% overlapping time window of band-pass (1-50 Hz) EKG signal vs. band-pass (20-50 Hz) mechanical acoustic z-axis signal. Figure 47D shows RR analyzed from a 120-s, 50% overlapping time window of PTAF signal and device z-axis signal with band-pass filters (f1 = 0.1 Hz, f2 = 0.8 Hz). Golden rule body orientation is investigated by visual inspection. The device captures body orientation by measuring a quasi-static gravity projection in the device frame associated with the core body frame (see SI for details). Figure 47E shows that the device captures the general tendency of body orientation as a rotation angle φ around the vertical axis, where zero degrees is defined as supine and positive orientation is defined as rightward. In addition to supine, prone, left, and right decubitus, the MA signal reconstructs additional details associated with the relative rotation of the head with respect to the core body. Figure 47F shows sleep stage inference from machine learning accelerometer data compared to clinical test sleep stages. We apply a Gaussian Mixture Hidden Markov Model (GMMHMM) to Mel-Frequency Cepstral Coefficient (MFCC) features to cluster five sleep stages from wakefulness to rapid eye movement (REM).

[0225] In addition to traditional sleep studies, we analyze the correlation of HR, RR, and body orientation. Figure 47G shows the cumulative distribution functions of HR and RR statistics in four classes of body orientation (supine: -45°<φ<45°, left: -135°<φ<-45°, right: 45°<φ<135°, prone: φ>135° or φ<-135°). The data are taken from 7 nights of MA measurements of one male subject. We utilize an in-house use case for large-scale statistics on 10 subjects (Figure 51). The results show that HR and RR are significantly higher when subjects sleep in a position close to prone.

[0226] Essay Materials and Methods Flexible Electronics Platform: Cut the board outline and circuit design along with the serpentine interconnects with a UV laser cutter (LPKF U4). The cut circuit board is made from a double-sided copper sheet with a thin copper clad laminate (12 μm) and copper film (12 μm) bonded to a polyimide (PI) film (25 μm) (FLP 7421).

[0227] A CO2 laser cutter (VLS3.50) is used to cut an FR-4 (0.381 mm, McMaster Carr 1331T37) board (FIG 43B) with the shape of two islands (FIG 43B) as reinforcements for added robustness. The board is glued to the backside of the circuit board, which is bent along the designated bend lines and glued to the other side of the FR-4 board using adhesive (Loctite Tak Pak 444). This forms a small area, double-layer component island. The components are fixed to the circuit board with solder paste (Chip Quik TS391LT).

[0228] Strain isolation: Cut an FR-4 board shadow mask with a CO2 laser cutter (VLS3.50) and screen print a layer of silicone gel (Silbione RT Gel 4717 A / B, Bluestar Silicone, E=5 kPa) on the bottom encapsulating elastomer layer. The gel is cured on a hotplate at 100 °C for 5 min.

[0229] Encapsulation: A 3-axis milling machine (Roland MDX 540) cuts an aluminum mold according to the 3D encapsulation mold design from CAD software (ProE Creo 3.0). From two pairs of aluminum molds, a base silicone elastomer film and a capping silicone elastomer film (Ecoflex, 00-30, Smooth-on) are cast separately. Each pair of molds has a concave mold design and a convex mold design to form a hollow space in the encapsulant. The Ecoflex poured into the mold is cured in an oven at 70°C for 7 minutes. After depositing silicone gel (strain insulating layer) on the casted lower elastomer, the electrons are bonded to the substrate by the silicone gel, which is a strain insulating layer. Then, the uncured Ecoflex is used as an adhesive to bond the capping film to the substrate.

[0230] Supplementary Information Suppression of motion artifacts in wavelet coherence-based respiratory analysis: Given two time series x, where n=1,2,...,N, n and y n The wavelet cross spectrum of C xy (s,n)=C * x (s,n)C y (s,n), (S.1) Here, C y (s,n) and C y (s,n) represents the continuous wavelet transform (CWT) of x and y at scale s and position n. * represents the complex conjugate.

[0231]

number

[0232] For its particular application in suppressing motion artifacts occurring in the frequency range of the respiratory cycle, we use the Morlet wavelet in our calculations. We choose a sampling period Δt = 20 s to cover all time scales of interest. The smallest scale for the Morlet wavelet is s0 = 2Δt. The CWT discretizes the scales with 16 voices per octave. The number of octaves is the nearest integer less than or equal to log2N-1, which is 10 in this case. We perform a moving average filter to smooth the CWT coefficients over the 16 scales. We use the built-in MATLAB functions "cwt" and "smoothCFS" to perform the continuous wavelet transform as well as the smoothing operation.

[0233] Gaussian Mixture Hidden Markov Model (HMM): For robust and flexible classification problems on time series signals, an effective method is to utilize a probabilistic approach that can infer random patterns using probability. In this study of mechanoacoustic biosignals, we introduce the Gaussian Mixture Hidden Markov Model. This model is constructed to describe unobserved states associated with an event of interest with discrete probabilities linked by a Markov chain. We apply this algorithm to swallowing detection and sleep stage identification.

[0234] To consider the hyperparameters of the probabilistic model, we manually select the number of states as n=5. For the feature extraction approach, we use Mel-frequency cepstral coefficients (MFCCs). MFCCs integrate the power spectral density of low frequencies in a narrow band, but for high frequencies in a wide band (a bandwidth of about f needs to be specified). MFCC coefficients take the form of the power of each band. We choose to deal with a total of 15 bands, which has been shown to give a good balance between the system complexity and the feature representation ability for signals sampled at frequencies of about 1-2 kHz. In swallowing detection, significant features due to swallowing activity appear in the lower order MFCCs and decay with increasing order. In contrast, speech consists of harmonic components and shows a unique pattern in the higher order MFCCs.

[0235] INCORPORATION BY REFERENCE AND MODIFICATIONS All references throughout this application, e.g., patent documents, including issued or granted patents or equivalent documents, patent application publications, and non-patent literature or other materials, are incorporated by reference in their entirety herein, to the extent that each reference is not at least partially inconsistent with the disclosure of this application, as if each were individually incorporated by reference (e.g., a partially inconsistent reference is incorporated by reference except for the partially inconsistent portion of the reference).

[0236] The terms and expressions used in this specification are used as terms of description and not of limitation, and there is no intention to use such terms and expressions to exclude the equivalents of the features shown and described or portions thereof, and it is understood that various modifications are possible within the scope of the claimed invention. Thus, although the present invention has been specifically disclosed by preferred embodiments, exemplary embodiments, and optional features, it should be understood that those skilled in the art may resort to modifications and variations of the concepts disclosed herein as a last resort, and such modifications and variations are considered to be within the scope of the present invention as defined by the appended claims. The specific embodiments presented herein are examples of useful embodiments of the present invention, and it will be apparent to those skilled in the art that the present invention can be implemented using numerous variations of the devices, device components, and method steps described herein. As will be apparent to those skilled in the art, the methods and devices useful for the methods of the present invention may include numerous optional compositions and processing elements and steps.

[0237] Where groups of alternatives are disclosed herein, it is understood that all individual members of the group and subgroups are separately disclosed. Where Markush or other groups are used herein, all individual members of the group and all possible combinations and subcombinations of the group are intended to be included individually in the disclosure.

[0238] All permutations or combinations of components described or exemplified herein may be used to practice the present invention unless otherwise indicated.

[0239] Whenever a range is given herein, e.g., thickness, size, modulus, mass, temperature range, time range, or composition or concentration range, all intermediate ranges and subranges, as well as all individual values ​​contained within the given range, are intended to be included in the disclosure, It will be understood that any subrange or individual value within a range or subrange contained in the description herein may be excluded from the claims of the invention.

[0240] All patents and publications mentioned in this specification indicate the level of the person skilled in the art to which this invention pertains.The references cited in this specification are incorporated herein by reference in their entirety to indicate the state of the art as of the publication or filing date, and it is intended that this information may be adopted herein to exclude certain embodiments that are in the prior art, if necessary.For example, when a composition is claimed, it should be understood that compounds that are known and available in the art before the applicant's invention, including compounds whose enabling disclosures are presented in the references cited herein, are not intended to be included in the composition claims herein.

[0241] The term "comprising" as used herein is synonymous with "including," "containing," or "characterized by," and is inclusive or open-ended, and does not exclude other unrecited elements or method steps. "Consisting of" as used in the specification excludes elements, steps, or ingredients not specified in the claim element. "Consisting essentially of" as used herein does not exclude materials or steps that do not materially affect the basic, novel characteristics of the claim. In each instance herein, "comprising," "consisting essentially of," and "consisting of" may be replaced with either of the other two terms. The invention as suitably illustrated herein may be practiced in the absence of one or more elements, one or more limitations, not specifically disclosed herein.

[0242] Those skilled in the art will understand that starting materials, biological materials, reagents, synthesis methods, purification methods, analytical techniques, assay methods, and biological methods other than those specifically exemplified may be used in the practice of the present invention without resorting to undue experimentation. All functional equivalents known in the art of such materials and methods are intended to be included in the present invention. The terms and expressions used are used as expressions of description rather than of limitation, and there is no intention in the use of such terms and expressions to exclude the equivalents of the features shown and described or portions thereof, and it is understood that various modifications are possible within the scope of the claimed invention. Thus, although the present invention has been specifically disclosed by preferred embodiments and optional features, it should be understood that those skilled in the art may resort to modifications and variations of the concepts disclosed herein as a last resort, and such modifications and variations are considered to be within the scope of the present invention as defined by the appended claims. [Explanation of symbols]

[0243] 20 Lower elastomer shell 30 Silicone Strain Isolation Layer 40 Expandable Interconnect 50 Electronic Devices 60 Upper elastomer shell

Claims

1. a. an electronic device having a sensor comprising an accelerometer; b. a two-way wireless communication system electronically connected to the electronic device for transmitting output signals from the sensor to an external device and for receiving commands from an external controller to the electronic device; c. a processor configured to analyze the output signal; Equipped with the sensor senses multiple or a single physiological signal from a subject, the physiological signal providing a basis for a stimulus provided to the subject, and a threshold is used as a trigger for delivering the stimulus to the subject; The medical sensor, wherein the processor analyzing the output signal includes personalizing the threshold value for the subject.

2. 13. The medical sensor of claim 1, which is wearable and attached to or implantable in the subject's tissue, or which mechanically obtains information from the subject's tissue, or which mechanically obtains information directly from the subject.

3. The medical sensor of claim 1 or 2, further comprising a wireless power supply system for wirelessly powering the electronic device.

4. 4. The medical sensor of claim 1, wherein the processor is further configured to provide real-time metrics and / or filter and analyze measured output from the electronic device to improve sensor performance parameters.

5. 5. The medical sensor of claim 3 or 4, wherein the processor utilizes machine learning to customize the analysis for each individual user of the medical sensor, the machine learning including one or more supervised and / or unsupervised learning algorithms customizable for the user.

6. 6. The medical sensor of claim 5, wherein the machine learning improves sensor performance parameters and / or personalized user performance parameters for use in diagnostic sensing or therapeutic applications.

7. 7. The medical sensor of any one of claims 1 to 6, which continuously monitors and generates real-time metrics.

8. 8. The medical sensor of claim 1, further comprising a stretchable electrical interconnect, a microprocessor, an accelerometer, a stimulator, a resistor and a capacitor in electronic communication for sensing vibration or motion by the accelerometer and providing stimulation to a user by a stimulator.

9. 10. The medical sensor of claim 8, wherein the electronic device comprises a network comprising a plurality of sensors, at least one of the sensors for sensing the physiological signal from the subject, and at least one of the sensors for providing a feedback signal to the subject.

10. 10. The medical sensor of claim 8, wherein the stimulator includes one or more of a vibration motor, an electrode, a light emitter, a thermal actuator, or an audio notification.

11. 11. The medical sensor of claim 1, further comprising a flexible encapsulation layer surrounding a flexible substrate and the electronic device, the flexible encapsulation layer including a lower encapsulation layer and an upper encapsulation layer, and a strain insulating layer, the strain insulating layer being supported by the lower encapsulation layer, and the flexible substrate being supported by the strain insulating layer.

12. 12. The medical sensor of any one of claims 1 to 11, having a device mass of less than 400 mg and a device thickness of less than 6 mm.

13. configured to be worn by a user; for use in a therapeutic swallowing application, wherein the output signal is a signal for one or more swallowing parameters selected from the group consisting of swallow frequency, swallow count, and swallow energy; and / or for use as a social interaction meter, said output signal being a signal for one or more social parameters selected from the group consisting of: conversation time, word count (fluency rate), vocalization parameters, linguistic discourse parameters, or dialogue parameters; and / or for use in a stroke rehabilitation device, said output signal being a signal for a social parameter and / or a swallowing parameter, 13. The medical sensor of claim 1 for use with one or more additional stroke rehabilitation parameters selected from the group consisting of gait, falls and physical activity.

14. 14. The medical sensor of claim 13, further comprising a stimulator that provides a tactile signal to a user to engage in at least one of a social interaction event, a safe swallowing event, and a breathing exercise.

15. 15. The medical sensor of claim 14, wherein safe swallowing is determined by sensing the onset of inspiration and expiration of a user's respiratory cycle.

16. The medical sensor of claim 14 or 15, wherein one or more machine learning algorithms are used in a feedback loop for optimization of the timing of the tactile signal.

17. 17. The medical sensor of any one of claims 1 to 16, further comprising an external sensor operatively connected to the electronic device.

Images