NL2033649B1

NL2033649B1 - Device for detecting and treating bruxism and method for doing the same

Info

Publication number: NL2033649B1
Application number: NL2033649A
Authority: NL
Inventors: Johanna Van Der Zee Catharina; Sebastiaan Bergdorff Theodorus
Original assignee: Jawsense Ltd
Priority date: 2022-11-30
Filing date: 2022-11-30
Publication date: 2024-06-04
Also published as: WO2024115682A1; EP4626303A1

Abstract

The present invention relates to a device for detecting and treating bruXism and a method for doing the same. The invention further relates to a user device for displaying bruxism related information, a server for training machine learning models, and a system comprising the device and the user device. The device comprises a first sensing electrode and a reference electrode, wherein the first sensing electrode and the reference electrode are arranged on the device such that, when the device is wom in a using position such that the first sensing electrode is placed on a first temporalis muscle or a first masseter muscle of a user, the reference electrode is placed on a part of the forehead of the user.

Description

DEVICE FOR DETECTING AND TREATING BRUXISM AND METHOD FOR DOING THE

SAME

The invention relates to a device for detecting and treating bruxism and a method for doing the same. The invention further relates to a user device for displaying bruxism related information, a server for training machine learning models, and a system comprising the device and the user device.

Bruxism is a condition in which a person has parafunctional masticatory muscle activities that occur during sleep (characterized as rhythmic or non-rhythmic) or during wakefulness (characterized by repetitive or sustained tooth contact and/or by bracing or thrusting of the mandible). Occlusal forces exerted during sleep bruxism often significantly exceed peak clenching or biting force under consciousness. Excessive mechanical stress during chronic bruxism is a critical risk factor for: tooth decay such as fracture or chippage of teeth and/or molars, periodontal disease, musculoskeletal pain, headaches, migraines, masticatory muscle/temporomandibular joint disorders (TMD) and more.

The primary treatment for sleep bruxism is the use of intra-oral splints, which are generally semi-rigid plastic covers for the upper or lower teeth. However, such occlusal splints have to be produced for a specific individual and primarily aim to protect teeth from damage, rather than to prevent or reduce bruxism. Furthermore, while teeth are protected from wear using a splint, the user may still suffer musculoskeletal pain and possible damage to the temporomandibular joint.

Wearable biofeedback devices exist which aim to reduce bruxism by trying to detect bruxism by various means and to provide biofeedback to the user.

One variation of such a wearable biofeedback device for reducing bruxism that can be commonly found on the market is a so-called smart splint. In this instance, sensing means (e.g. a pressure sensor) are incorporated into an occlusal splint in order to sense the onset of bruxism activity. These approaches are disadvantageous, as they require the presence of electrical devices in the mouth, which users consider invasive. Furthermore, these splints often include batteries, which may contain highly toxic substances and are therefore less suitable to be used in the mouth.

As a result, said devices are associated with electrical and chemical health risks that add to the general drawbacks of intra-oral splints described above. In addition, many of these attempts have resulted in bulky devices which would be even more uncomfortable to wear for the user than traditional occlusal splints.

A second variation of this wearable biofeedback approach includes devices equipping EMG sensors to detect bruxism. The biggest limitation of these devices is that they have been shown to vield inaccurate results when they were compared with the golden standard of polysomnography (PSG) measurements.

Another limitation is that these devices are not able to distinguish different types of bruxism.

Furthermore, in most of these instances, adhesive electrodes were used which significantly impact user experience. In addition, these devices are only designed to be wom at night and don’t offer a solution for awake bruxism.

The device and method according to the invention obviates or at least reduces the abovementioned problems.

To that end, the invention provides a device comprising a first sensing electrode and a reference electrode, wherein the first sensing electrode and the reference electrode are arranged on the device such that, when the device is worn in a using position such that the first sensing electrode is placed on a first temporalis muscle or a first masseter muscle of a user, the reference electrode is placed on a part of the forehead of the user, wherein the device further comprises: — a signal module that is electronically connected to the first sensing electrode and the reference electrode and configured to measure, over time, a first plurality of voltage differential values between the first sensing electrode and the reference electrode, wherein the voltage differential values are indicative of a unilateral muscle activity of the first temporalis muscle or the first masseter muscle on which the first sensing electrode is placed; — a detection module connected to the signal module via a wireless or wired connection and configured to obtain a first action potential signal from the signal module, wherein the first action potential signal comprises and/or is derived from at least some of the first measured voltage differential values, wherein the detection module 1s further configured to select a muscle activity label from a set of predetermined muscle activity labels, the selecting comprising analyzing the first action potential signal; and — a biofeedback module configured to obtain the selected muscle activity label and to provide the user with a biofeedback signal in reaction to the selected muscle activity label being indicative of a bruxism activity.

By having a device, wherein a first action potential signal is obtained from measurement of voltage differential values between the temporalis muscle or the masseter muscle and the forehead of a user, a measurement of the unilateral muscle activity of the temporalis muscle or the masseter muscle on which the first electrode is placed can be obtained for evaluation of the muscle activity.

An advantage of the device according to the invention with which unilateral muscle movements can be measured and analyzed, is that it is possible to distinguish between muscle activities that are bruxism activities such as grinding and gnashing and non-bruxism activities such as talking and eating with a surprisingly higher accuracy compared to conventional devices.

Another advantage of the device according to the invention is that, by being able to perform unilateral evaluation of the jaw, it allows for a surprisingly accurate selection of muscle activity by users with uneven jaw muscle activity. This is relevant due to the existence of dental asymmetries, which lead to less accurate measurements when conventional methods are used. This advantage is especially relevant for users that have suffered from long-standing bruxism, as they more often deal with dental asymmetries caused by malocclusion and/or TMJ disorders due to the long- standing bruxism.

A further advantage is that by being able to assess unilateral temporal muscle activity, one can diagnose accurately with muscle activity recognition whether a user has bruxism when said user has a misaligned bite, which was not possible using the conventional methods.

Another advantage of the device according to the invention is that. by the reference electrode being positioned on the forehead results in an accurate reference signal, as the forchead has relatively low to no muscle activity, which means an even more accurate measurement of the unilateral movement of the jaw muscle can be achieved.

Another advantage of the device according to the invention is that. by using a biofeedback module, the user can be nudged to stop their bruxism activity by the device via the biofeedback module in reaction on the selected muscle activity label being indicative of a bruxism activity.

It will be clear that the first action potential signal may comprise one or more signals that comprise and/or are derived from at least some of the first measured voltage differential values.

In an embodiment of the invention, the analyzing the first action potential signal comprises calculating a bruxism probability of the first action potential signal being indicative of a bruxism activity and wherein the selecting further comprises selecting one of the muscle activity labels that is indicative of a bruxism activity when the bruxism probability is above a predetermined bruxism threshold.

An advantage of this embodiment is that by calculating the bruxism probability and using a predetermined bruxism threshold, a degree of uncertainty with which bruxism is detected is controlled by the predetermined bruxism threshold.

In an example of the previous embodiment, the predetermined bruxism threshold is preferably between 15% and 100%, for example 25%, 50%, 75% or 90%, and more preferably above 90%. Preferably, the predetermined bruxism threshold is between 75% and 100%. It will be clear that the provided numbers for the predetermined bruxism threshold are just provided as an example and do not in any way limit the predetermined bruxism threshold to these numbers. It will further be clear that the predetermined bruxism threshold may be represented in any other suitable way besides using percentages, for example using number between 0 and 1, such that the predetermined bruxism threshold is preferably between 0.15 and 0.95, for example 0.25, 0.50, 0.75. 0.90 etc, and more preferably above 0.8.

It is noted that a higher predetermined bruxism threshold is preferred, as a higher bruxism threshold will result in a reduction in Type 1 errors (false positive), at the cost of an increase of

Type II errors (false negative errors), as missing a bruxism event has fewer negative consequences than inaccurately identifying a bruxism event and unnecessarily providing biofeedback to a user.

In a further embodiment of the invention, the device comprises a threshold control element that is electronically and/or operatively connected to the detection module and wherein the threshold control is configured to enable the user to adjust the predetermined bruxism threshold.

In an example the threshold control element comprises a scroll wheel, wherein the detection module is configured to lower the predetermined bruxism threshold in reaction to the scroll wheel being tumed in a first direction and to increase the predetermined bruxism threshold in reaction to the scroll wheel being turned in a second direction.

In an example the threshold control element comprises a first button and a second button, wherein the detection module is configured to lower the predetermined bruxism threshold in reaction to the first button being pressed and wherein the detection module is further configured to increase the predetermined bruxism threshold in reaction to the second button being pressed.

It is obvious that other control elements which enable the user to adjust the predetermined bruxism threshold are also in the scope of the invention.

In an embodiment according to the invention, the device further comprises a preprocessing module configured to connect the signal module and the detection module, wherein the preprocessing module 1s configured to produce the first action potential signals by applying one or more preprocessing steps to the voltage differential values.

It will be clear that in the presence of a preprocessing module, the obtaining of the first action potential signal from the signal module by the detection module may comprise obtaining the first action potential signal from the preprocessing module.

In a further embodiment according to the invention, the one or more preprocessing steps are taken from: - applying a filtration, rectification and/or smoothing function; - applying a bandpass filter function, such as, a Butterworth filter, preferably in the range of 5 to 1000 Hz. - applying an analog to digital conversion function; - applying a statistic function relating to an amplitude and/or a power of the action potential signal, such as, an integrated absolute value function, a root mean square function, a waveform length function: - applying a statistic function relating to a signal frequency and/or a nonlinearity of the action potential signal, such as, a maximum fractal length function, a zero crossing function, a mean frequency function; - applying a statistic function relating to a time series property of the action potential, such as, an autoregressive coefficient function;

- applving a signal decomposition function, such as, a discrete wavelet transform, or a

Fourier transform (for example a fast Fourier transform or a Short-Time Fourier

Transform).

According to the invention the preprocessing function may be comprised in a hardware filter 5 orin a software filter or a combination thereof.

It will be clear that the above list is non-exhaustive and that any other suitable preprocessing step may also be applied according to the invention.

It will be clear that the preprocessing of one voltage differential values may result in one or more action potential signals. It will further be clear that the action potential signals may comprise a combination of unprocessed voltage differential signals and preprocessed voltage differential signals.

In an embodiment according to the invention, the one or more preprocessing steps comprise dividing the action potential signals into multiple, optionally overlapping, smaller action potential signals, for example using a sliding window function applied to the action potential signal, preferably using a window size in the range of 0.1 to 10 seconds.

In a further embodiment, the sliding window function is applied multiple times using different window sizes, for example a first window size of 1-second signal window and a second window size of a ten second window.

In a further embodiment, the dividing of the action potential signals is applied as a first preprocessing step.

In an embodiment according to the invention, analyzing of the first action potential signal comprises applving a predetermined set of rules to the first action potential signal, the predetermined set of rules comprising data characteristics that are indicative of the muscle activity corresponding to the first action potential signal being a bruxism activity.

An advantage of analyzing the first action potential signal by applying a predetermined set of rules to the first action potential signal, 1s that said rules can be applied relatively fast and computational efficient to the first action potential signal, meaning that no heavy processing power is required to detect bruxism in the first action potential.

In a further embodiment according to the invention, the data characteristics comprises one or more action potential intensity threshold. one or more action potential intensity durations, and/or a combination thereof.

In an example an action potential intensity duration can be determined by calculating an area under the curve of the action potential signal, and/or derivatives thereof, for a predetermined duration and/or by applying a rolling average on the action potential signal, and/or derivatives thereof over a predetermined duration period.

In a further or alternative embodiment predetermined set of rules comprises one or more user-agnostic rules and/or one or more user-dependent rules. It is noted that a user-agnostic rule is a rule which is the same for each user. It is noted that a user-dependent rule is determined for a specific user.

In an example the one or more user-dependent rules may be determined for a specific user during an onboarding process were characteristics of the action potential signal of that user are determined.

In an embodiment according to the invention, the selecting of the muscle activity label comprises using the first action potential signal as input to a predetermined machine learning model trained to detect bruxism; and receiving the bruxism probability and/or the muscle activity label from the machine learning model.

An advantage of having a pretrained machine learning model is that, by using machine learning models, it is possible to detect bruxism on unseen/unknown patterns, resulting detection of bruxism in unknown cases. Another advantage of having a pretrained machine learning model is that a higher accuracy can be achieved compared to using only deterministic rule sets.

In a further embodiment according to the invention, the machine learning model is one of the following: a logistic regression model, a multinomial regression model, a support vector model, a learning vector quantization model, a decision tree, a random forest, a XGBoosted tree, a neural network, a convolutional neural network, a deep neural network, a recurrent neural network, any suitable machine learning model and/or classification model, or an ensemble model comprising an ensemble of one or more of the previous mentioned models.

In an example according to the invention. the machine learning module comprises a combination of a convolutional neural network and a recurrent neural network, called recurrent convolutional neural network.

This has as advantage that a convolutional neural network is known to be able to capture spatial information, whereas the recurrent structure captures temporal information, which means that the recurrent convolutional neural network is especially suitable to classify the action potential signals, as these are temporal signals.

In an embodiment according to the invention, the machine learning models are trained using an input dataset or comprise the input dataset, the mput dataset for example comprising a plurality of time series of voltage differential values and/or derivatives thereof. In an example, the plurality of time series of voltage differential values and/or derivatives thereof may comprise, optionally partly overlapping, values voltage differential measurements. In a further or additional example, cach of the time series of voltage differential values and/or derivatives thereof are associated with one or more muscle activity labels.

In an example, the plurality of time series of voltage differential values and/or derivatives thereof are obtained by dividing measurements of voltage differential values into chunks using a predetermined window size and a predetermined increment size, wherein the window size determines a size of the chunk and wherein the increment size determines an overlap of the chunk with a previous chunk, wherein each chunk is a time series of voltage differential values and/or derivatives thereof. For example, a measurement of a certain length is divided using a window size of 250 ms and an increment size of 125 ms results into chunks of 250 ms with a 50% overlap. It is noted that the above window size and increment size are just provided as examples. In another example an appropriate window size and/or an appropriate increment size are determined by performing a parameter optimization sweep during a pretraining phase using the machine learning model.

In an example the training data is preprocessed using a similar method as described for the preprocessing in the preprocessing module.

In an embodiment according to the invention, the device further comprises a second sensing electrode which is arranged on the device such that, when the device is worn in the using position such that the first sensing electrode is placed on the first temporalis muscle or the first masseter muscle of the user, the second sensing electrode is placed on a second temporalis muscle or a second masseter muscle of the user that is located on another side of the forehead of the user compared to the first temporalis muscle or the first masseter muscle, and wherein the signal module is further electronically connected to the second sensing electrode and is further configured to measure, over time, a second plurality of voltage differential values between the second sensing electrode and the reference electrode, wherein the second plurality of voltage differential values are indicative of a unilateral muscle activity of the second temporalis muscle or the second masseter muscle on which the second sensing electrode is placed, and wherein the detection module is further configured to obtain a second action potential signal that comprises and/or is derived from at least some of the second measured voltage differential values and wherein the selecting of the muscle activity label by the detection module further comprises analyzing the second action potential signal.

An advantage of having a second sensing electrode is that this further enables judgment of unilateral temporal muscle activity in the user, which results in a surprisingly high improvement in the accuracy of the bruxism detection of the device.

It is noted that an action potential between the first sensing electrode and the second sensing electrode is not measured according to the invention, measurements are only taken between a sensing electrode and the reference electrode.

In an embodiment according to the invention, the surface electrode sensors are one of: rubber electrodes, paint electrodes, textile electrodes, dry metallic electrodes, silver coated electrodes, or any other suitable electrodes, preferably silver coated electrode.

In an embodiment of the invention, the surface electrode sensors are configured to measure muscle signals in the range of 0 V to 3.3 V or 0.0 V to 5.0 V. and have a high accuracy in the range of 0.0 mV to 50.0 mV m amplitude, preferably have a high accuracy in the range of 0.0 to 20 mV , more preferably in the range 0.01 to 10 mV, wherein a high accuracy is for example an accuracy between 0.001 mV to 0.005 mV.

An advantage of having surface electrode sensors which are configured to measure muscle signals in the range of 0.00 mV to 50 mV in amplitude, is that most muscle signals of interest are typically withing this range, and more typically in the range of 0.001 mV to 10.00 mV in amplitude.

In a further or alternative embodiment, the surface electrode sensors have a sample rate in the range of 10 Hz to 10000 Hz, preferably in the range of 500 Hz to 5000 Hz.

An advantage of having the surface electrode sensors having a sample rate in the range of 500 Hz to 10000 Hz is that, in most cases the sampling rate is at least the Nyquist rate.

Muscle signals are generated by electrochemical depolarization and repolarization within muscles and nerves as individual muscle cells fire and contract. Throughout a muscle, individual muscle cells fire at different times in different places. The overall strength of contraction of the muscle at a given moment comes from the number of cells firing at the time. The repetition of firing and contracting of muscle cells in an active muscle result in most of the electrical energy of the muscle signal being concentrated in the range of 20 Hz to 2000 Hz.

In an embodiment according to the invention, the device comprises a headband, wherein the first sensing electrode and the reference electrode are attached to the headband.

An advantage of the device comprising a headband is that the device can be easily worn by a user without interfering with the sleep of the user.

Another advantage of the device comprising a headband is that it can be worn by a user throughout the day without interfering with daily activities, especially activities which involve jaw muscle activity, such as, eating, speaking etc. This is especially relevant when the device is used to reduce awake bruxism.

In a further embodiment according to the invention, the first sensing electrode and the reference electrode are positioned on the headband such that, when the headband is wom in a wearing position, the first sensing electrode is positioned on the skin covering a first temporalis or a first masseter muscle of the user and the reference electrode is positioned on the skin of the forchead of the user.

In an even further or altematve further embodiment, the sensing module, the detecting module, and the biofeedback module are comprised in a housing that is attached to the headband.

In a preferred embodiment, the housing is removably attached to the headband.

In an embodiment according to the invention, the device further comprises a health information module comprising one or more user health information sensors configured to measure, over time, a plurality of health information data measurements relating to the user, the health user sensors comprising one or more of: an optical sensor, an electroencephalography (EEG) sensor, a heart rate sensor, a blood oxygen saturation sensor, a motion sensor, and/or a body temperature sensor; and, wherein the signal module and/or the detection module are further electronically connected to the health information module and wherein the detection module is further configured to obtain, optionally via the signal module and/or the preprocessing module, at least a part of the health information data measurements and wherein the selection of the muscle activity label by the detection module further comprises analyzing the at least part of the health information data measurements and/or derivatives thereof.

An advantage of the device comprising a health information module is that more health information data can be used in the selecting of the muscle activity label, resulting in a higher selection accuracy.

In a further embodiment according to the invention, wherein the obtaining at least a part of the health information data measurements and analyzing at least part of the health information data measurements by the detection module comprising one or more of: — obtaining at least a part of the health data measurements from the EEG sensor, the heart rate sensor, the motion sensor, the optical sensor, the blood oxygen saturation sensor and/or the body temperature sensor; — determining one or more of: a sleep stage, a sleep quality, a sleep time, an oxygenation, a breathing rate, and a stress level, of the user; by analyzing the health information data measurements and/or values derived from the health information data; and wherein the detection module is electronically connected to the sleep stage detection module and wherein the selection of the muscle activity label bv the detection module further comprises observing the determined one or more of: the sleep stage. the sleep quality, the sleep time, and the stress level, of the user.

An advantage of observing one or more of: a sleep stage, a sleep quality, a sleep time, and a stress level of the user is that accuracy can be further improved, as all listed options are factors in and/or triggers of bruxism episodes.

In an additional or alternative further embodiment according to the invention, the detection module is further configured to detect a sleep apnea episode by analyzing at least a part of information data elements and the biofeedback module is further configured to provide biofeedback to the user in response to detection of the sleep apnea episode by the detection module. In an example of this embodiment, the device comprises the blood oxygen saturation sensor and/or the optical sensor, and the detection module is configured to detect a sleep apnea episode and/or breath holding episode by: - detecting a drop in user oxygenation levels using a plurality of measurements and/or derivatives thereof from the blood oxygen saturation sensor; and/or - detecting a drop in breathing rate levels using a plurality of measurements and/or derivatives thereof from the optical sensor.

It is noted that the breathing rate can be deducted from measurements taken by the optical sensor.

An advantage of detecting a sleep apnea episode and providing the user with biofeedback in response to said detection is that the device can additionally be used to reduce sleep apnea episodes of a user, providing the user with additional health benefits.

In a further embodiment according to the invention, wherein the detection module is configured to calculate the bruxism probability, the detection module is configured to increase the bruxism probability and/or decrease the bruxism threshold in reaction to detecting a sleep apnea episode.

An advantage of this embodiment is that, by using the detection of a sleep apnea episode, the accuracy of the detection of bruxism can be improved, because there exists a positive correlation between sleep apnea episodes and bruxism episodes in that a bruxism episode occurs relatively more often after a sleep apnea episode.

In an embodiment according to the invention, wherein the detection module is configured to calculate the bruxism probability, the analyzing the at least part of the health information data elements comprises: — obtaining the heart rate and/or the heart rate variability; — detecting an increase in heart rate and/or a decrease in heart rate variability, wherein the increase in heart rate and/or the decrease in heart rate variability are indicative of a micro-arousal of the user; and — in response to the detection of the increase in heart rate and/or a decrease in heart rate variability, increasing the bruxism probability and/or decreasing the predetermined bruxism threshold.

An advantage of increasing the bruxism probability in response to detection using a detection of an increase in heart rate and/or the decrease of heart variability is that the accuracy with which bruxism is detected is increased.

In an embodiment according to the invention, wherein the biofeedback module comprises a visual feedback module, an audio feedback module, a haptic feedback module and/or an electric feedback module. The visual feedback module may for example comprise a light attached to the device, which will be turned on as biofeedback to the user. The audio feedback module may for example comprise one or more speakers configured to produce one or more audio signals as biofeedback to the user. The haptic feedback module may for example comprise one or more vibrating motors configured to provide vibration to the user as biofeedback. The electric feedback module may comprise one or more electrodes configured to supply electric signal to the user as biofeedback resulting in the user experiencing a small tingling and/or pain sensation.

In an embodiment according to the invention, the biofeedback module comprises one or more biofeedback parameters, the biofeedback parameters comprising a biofeedback intensity, a biofeedback duration, a biofeedback frequency.

In an embodiment according to the invention, the detection module is further configured to determine a muscle activity intensity by observing the first action potential signal and/or (when available) the second action potential, and wherein the biofeedback module is configured to obtain the muscle activity intensity from the detection module and to determine the biofeedback intensity and/or biofeedback duration based on the muscle activity, wherein the biofeedback intensity and/or biofeedback duration positively correlation to the muscle activity intensity.

An advantage of correlating the biofeedback intensity with the muscle activity intensity is that the biofeedback is adapted to the intensity of the bruxism episode.

In a further or alternative embodiment, the biofeedback module is configured to adjust the biofeedback parameters based on one or more of: an age of the user, a sex of the user, previous biofeedback provided to the user, health information data, a detected sleep stage, and other user specific characters.

In an even further embodiment or alternative further embodiment, the biofeedback module comprises a biofeedback prediction model that is configured to observe a change in the action potential signal after biofeedback is provided to the user and to adjust one or more biofeedback parameters based on the observed change in the action potential signal. It is noted that the observed change in the action potential is indicative of the user responding to the biofeedback signal provided to him. It is noted that in the context of the invention, optimizing the biofeedback parameters comprises adjusting the biofeedback parameters such that the biofeedback is as nonintrusive as possible while still having the effect that bruxism activity of the user is stopped or reduced.

In an example the biofeedback model is a machine learning model. In a further example the biofeedback prediction model is a reinforcement learning agent that is trained to adjust the biofeedback parameters in real time and/or depending on characteristics of the user.

An advantage of optimizing the biofeedback parameters is that the biofeedback can be provided to the user, while being as nonintrusive as possible. This is especially advantageous when the device is used during sleep, where intrusive biofeedback might affect the sleep quality of the user.

In an example, wherein the biofeedback module comprises the visual feedback module, an increase in biofeedback intensity results in a change in a color of the visual feedback and/or an increase of a brightness of the visual feedback.

In an example. wherein the biofeedback module comprises the audio feedback module, an increase in biofeedback intensity results in a change in a frequency of the audio feedback and/or an increase of a volume of the audio feedback.

In an example, wherein the biofeedback module comprises the haptic feedback module, an increase in biofeedback intensity results in a change in a vibration pattern, an increase in vibration frequency and/or an increase in vibration strength.

In an example, wherein the biofeedback module comprises the electric feedback module, an increase in biofeedback intensity results in an increase in the tingling and/or pain sensation experienced by the user.

In an embodiment according to the invention the muscle activity labels comprise one or more labels indicative of a bruxism activity and one or more labels indicative of non-bruxism activity. For example, the labels indicative of a bruxism activity are one or more of: bruxism, grinding, clenching, gnashing: and the labels indicative of a non-bruxism activity are one or more of: non-bruxism, eating, yawning, talking.

An advantage of this embodiment is that, depending on, for example, the user preferences, or the method used to determine the bruxism label, the muscle activity can be regarded as a binary model (e.g. bruxism vs non-bruxism) or as a multinomial model with specific type of jaw behavior, making it adaptable to different preferences and/or methods used.

In an embodiment according to the invention, the device further comprises a communication module comprising a wired interface and/or a wireless interface, wherein the communication module is configured to connect to a user device via the wired interface and/or the wireless interface and to enable the device and the user device to exchange user bruxism information data elements relating to one or more selected muscle activity labels and/or the plurality of voltage differential value measurements and/or the plurality of voltage differential value measurements and/or values derived from the plurality of voltage differential value measurements and/or one or more of the health information data measurements and/or biofeedback provided to the user.

An advantage of the device being able to connect to a user device and to exchange user bruxism information data elements with the user device, is that it enables the user to obtain useful insights into their sleep patterns, bruxism episodes etc. via the user device.

In an embodiment according to the invention, the device further comprises a processor and a memory, wherein the processor is electronically connected to the memory, the first sensor, and the reference sensor, and the biofeedback module and wherein at least parts of the signal module, the detection module, the biofeedback module, and/or the preprocessing module are comprised in the processor.

In an embodiment according to the invention, the device further comprises a rechargeable battery configured to supply electric energy to a plurality of electronic components of the device, wherein the device further comprises a charging interface configured to enable a user to charge the battery comprising a wired charging interface and/or an induction charging module.

The invention further relates to the user device. the user device comprising a processor, a memory, a display, and a communication module, wherein the memory has instruction stored thereon that, when executed by the processor, enables the device to: — connect to the device according to the invention; — to receive one or more bruxism information data elements; and — to display at least one of the one or more bruxism information data elements and/or derivates of at least one of the one or more bruxism information data elements on the display.

It is noted that the bruxism information data elements relate to the user bruxism information, for example, data from one or more of: the muscle activity measurements, first action potential signal, second action potential signal, bruxism activity. muscle activity labels, provided biofeedback signal. biofeedback parameters, and optionally other available health data measurements by the device, such as, sleep quality.

An advantage of the user device according to the invention is that it enables the user to obtain useful insights into their sleep patterns, bruxism episodes etc. via the user device.

In a further embodiment according to the invention, the user device is configured to display, to the user, a current bruxism score and/or a bruxism progression score, wherein the current bruxism score is indicative of an amount and/or severity of bruxism activity detected by the device in a first predetermined period and wherein the bruxism progression score is indicative of a change and/or trend in the amount and/or severity of bruxism activity detected by the device in a second predetermined period.

An advantage of the user device indicating to the user their bruxism score and/or a bruxism progression score is that it is straightforward for the user to gain insight in the severity of their bruxism episodes and change and/or trends therein.

In a further or alternative embodiment according to the invention, the user device comprises a gamification module, configured to display, to the user, one or more cues for a range of muscle activity exercises aimed toward exercise of the first and/or second masseter muscle and/or the first and/or second temporalis muscle, and to provide, using the device according to the invention, real- time feedback to the user through the gamification module, wherein the real-time feedback is indicative on whether the user is matching the one or more cues for the range of muscle activity exercises. If the device, during the muscle activity exercises, determines that the muscle activity of the user is a bruxism muscle activity, the bruxism biofeedback module according to the invention may also provide biofeedback signal to the user.

An advantage of the user device comprising a gamification module is that gamification helps to keep the user motivated to engage with their bruxism biofeedback therapy. A further advantage of the gamification engine is that it can help the user to learn how to relax their jaw muscles. A further advantage of the gamification engine is that it can train the user in how to respond to the bruxism biofeedback signal. A further advantage of the gamification engine is that additional user data can be retrieved and used to improve the bruxism detection module according to the invention.

The invention further relates to a server configured to train a machine learning model using a set of predetermined training signals, wherein each of the predetermined training signals in the set of predetermined training signals is labelled with one of the muscle activities from the set of muscle activities and wherein the server further comprises communication means configured to connect to one or more devices that comprises a machine learning model according to the invention, and to upload the trained machine learning module to the device and/or user device.

An advantage of the server according to the invention, is that new machine learning models can be trained on the server and uploaded to device and/or the user device, such that new machine learning models can be used in the detection module.

The invention further relates to a system comprising the device and the user device according to the invention, wherein the device is operatively connectable to the user device, optionally the svstem further comprising a server according to the invention, wherein the server is operatively connectable to the device and/or user device.

The system according to the invention has all the effects and advantages of the device, user device and server according to the invention.

The invention further relates to a method for use of the device, the method comprising: — obtaining the device according to the invention; — placing the device on a head in a wearing position; and — receiving biofeedback from the device in response to the device detecting a bruxism activity.

The method for use of the device has the same effects and advantages as the user device.

The invention further relates to a method for detecting and classifying bruxism, the method comprising: — obtaining a first action potential signal derived from or comprising a plurality of voltage differential values between a first sensing electrode placed on a first temporalis muscle or a first masseter muscle of a user and a reference electrode placed on a forehead of a user, wherein the voltage differential values are indicative of a unilateral muscle activity of the first temporalis muscle or the first masseter muscle on which the first sensing electrode is placed: and — selecting a muscle activity label from a set of predetermined muscle activity labels, the selecting comprising analyzing the first action potential.

An advantage of obtaining and analyzing a first action potential signal derived from or comprising a plurality of voltage differential values between a first sensing electrode placed on a first temporalis muscle or a first masseter muscle of a user and a reference electrode placed on a forehead of a user, is that it is possible to distinguish between muscle activities that are bruxism

I5 activities such as grinding and gnashing and non-bruxism activities such as talking and eating with a surprisingly higher accuracy.

Another advantage of the method according to the invention is that, by being able to perform unilateral evaluation of the jaw, it allows for a surprisingly accurate selection of muscle activity by users with uneven jaw muscle activity. This is relevant due to the existence of dental asymmetries, which lead to less accurate measurements when conventional methods are used. This advantage is especially relevant for users that have suffered from long-standing bruxism, as they more often deal with dental asymmetries caused by malocclusion and/or TMJ disorders due to the long- standing bruxism.

In an embodiment of the invention, the determining of the muscle activity label comprises applying a preselected set of deterministic rules to the first action potential signal, wherein the preselected set of deterministic rules comprises a multiple of signal characteristics with each a corresponding muscle activity from the set of muscle activities.

An advantage of this embodiment is that muscle activity can be determined in a quick and computational inexpensive manner.

In an embodiment according to the invention, the multiple of signal characteristics are predetermined for the user.

An advantage of this embodiment is that, by predetermining the multiple of signal characteristics for the user, the device is more adapted to detect bruxism on a specific user, resulting in a higher accuracy.

In an embodiment according to the invention, the determining of the muscle activity label comprises providing the first action potential as input to a machine learning model, wherein the machine learning model is trained to determine the muscle activity label by classifying the first action potential when it is provided as an input to the machine leaming model.

An advantage of using a machine learning model is that, by using machine learning models, it is possible to detect bruxism on unseen/unknown patterns, resulting detection of bruxism in unknown cases. Another advantage of having a machine learning model is that a higher accuracy can be achieved compared to using only deterministic rule sets.

In an embodiment according to the invention, the machine learning model is a neural network.

An advantage of using a neural network is that more complex unseen/unknown patterns can be captured by the neural network.

In an embodiment according to the invention, the machine learning model is trained on a set of predetermined training signals, wherein each of the predetermined training signals in the set of predetermined training signals is labelled with one of the muscle activities from the set of muscle activities.

The invention further relates to a device comprising a processor, a memory, a first sensing electrode and a reference electrode, wherein the first sensing electrode and the reference electrode are arranged on the device such that, when the device is worn in a using position such that the first sensing electrode is placed on a first temporalis muscle or a first masseter muscle of a user, the reference electrode is placed on a part of the forehead of the user; wherein the memory contains instruction that enables the processor to obtain a first plurality of voltage differential values between the first sensing electrode and the reference electrode and to execute a method according to the invention.

The lónventtion is described in the foregoing as examples. It is understood that those skilled in the art are capable of realizing different variants of the invention without actually departing from the scope of the invention. Further advantages, features and details of the invention are elucidated on the basis of preferred embodiments thereof, wherein reference is made to the accompanying drawings, in which: - Figure 1 shows a top view of an embodiment of the device; - Figure 2 shows a side view of an embodiment of the device: - Figure 3 shows a part of an embodiment of the device; - Figure 4 shows a user wearing an embodiment of the device: - Figure 5 shows an embodiment of the device with removable housing; - Figure 6 shows a schematic overview of an embodiment of the device; - Figure 7 shows a flow diagram of the data processing in an embodiment of the device and/or method; and - Figure 8 shows a schematic overview of an embodiment of the system.

Figures 1 — 3 show examples of device 2 having first sensing electrode 4a, second sensing electrode 4b and reference electrode 6 which are all attached to inside surface 14 of headband 8.

Headband 8 has first end 8a and second end 8b, wherein first end 8a and second end 8b are connectable via connector 10 to enable a user to easily wear device 2 on its head (H). Connector 10 is for example magnetic connector or a Velcro connector. Altematively, first end 8a and second end 8b are connectable via a clip (not shown). Alternatively, parts 8a and 8b are not separable but form one component together, fabricated with elastic material to provide the required flexibility to fit the headband on a wide range of head shapes (not shown). For example, the elastic material might allow the material to stretch from a first circumference to a second, larger circumference.

The first circumference is for example between 52 and 57cm while the second circumference is between 58 and 63cm. Device 2 further has additional sensors and/or biofeedback modules 18, 20, 22, and 24, for example optical sensor 18, vibrator and/or buzzer 20, movement sensor 22 and body temperature sensor 24. Optical sensor 18 is used to obtain measurements of the user relating to heart rate, heart rate variability, oxygenation, breathing rate and more. It will be evident that the placing and order of sensors and/or biofeedback modules 18, 20, 22, and 24 is just one example of possible placings and orders according to the invention and the skilled person will understand that other placings and others are possible within the scope of the invention.

Device 2 further has housing 12 attached to outside surface 16 of headband 8. Housing 12 may house internal component of device 2, such as, (not shown) processor, memory, motion sensor, signal module, preprocessing module, detection module, biofeedback module and/or communication module. Figure 4 shows device 2 being worn by the user such that headband 8 surrounds head H of the user, such that reference sensor 6 is position on the forehead of the user and first sensing electrode 4a and second sensing electrode (not shown) are positioned near the temporal muscle of the user. Housing 12 extends away from the user, such that it does not bother the user during sleep.

Figure 5 shows housing 12 being detachable from headband 8, wherein headband 8 comprises edge 9 which defines a housing receiving space configured to receive housing 12, wherein, when housing place 12 is placed in the house receiving space, edge 9 of headband 8 tightly surrounds housing 12 such that housing 12 remains in place.

Figure 6 shows device 102 with first sensing electrode 104a, second sensing electrode 104b, reference electrode 106 and additional sensors 118 all electronically connected to processor 130.

Processor 130 is further electronically connected to memory 140, biofeedback module 150 and communication module 160.

When device 102 is in use, processor 130 measures voltage differential values between first sensing electrode 104a and reference electrode 106 and second sensing electrode 104b and reference electrode 106. Processor 130 stores voltage differential values in memory 140. Processor

130 further derives action potential signals from voltage differential signals and retrieves a trained machine learning model from memory 140 to determine muscle activity labels by feeding the action potential signals to the machine learning model and saves muscle activity labels in memory 140 with the corresponding action potential signals. In response to determined muscle activity labels being bruxism labels, processor 130 further sends signal (not shown) to biofeedback module 150 which provides biofeedback to the user. Periodically, processor 130 uses communication module 160 and connects to user device (see figure 7) to upload voltage differential signals action potential signals, and/or muscle activity labels from memory 140 to the user device.

Figure 7 shows a flow diagram of the data processing in an embodiment of the device and/or an embodiment of the method. First part 200 relates to steps for detecting bruxism, while second part 300 relates to other health related steps. However, one or more steps from first part 200 and one or more steps from second part 300, may be combined according to the invention without taking all steps from the corresponding part. Parts 200 and/or 300 may be repeated during use at the same of different frequencies. For example, steps from part 200 might be repeated in a continuous loop, while steps from part 300 are performed every 5 minutes. Furthermore. steps from the parts may be executed simultaneously, i.e. not all steps from part 200 have to be performed, before the steps from part 200 are executed again.

In step S202 a first action potential signal is obtained. for example by sampling voltage differential signals between first sensing electrode 4a and reference electrode 6. In an example, a sampling rate of 4000 Hz is used. and a sampling is taken in a duration of 0.1 seconds to 2 minutes, for example 10 seconds, or 1 minute, however, other sampling rates and durations are also possible. It is also possible for consecutive samples to overlap for a duration smaller than the entire duration of the sample.

Optionally, in step S204 a second action potential is obtained, for example by sampling voltage differential signals between second sensing electrode 4b and reference electrode 6. The second action potential signal may have the same sampling rate and duration as the first action potential.

It will be clear that, when the term first action potential and/or second action potential is used. this may refer to either the direct signal and/or one or more derivatives thereof, i.e. the singular use of signal may be used to refer to a plurality of signals which all have the same source signal. An example of a derivative of the direct signal is the area under the curve.

In describing the next steps, the term action potential signal is used to include both the first action potential signal and, when optional step S204 is performed, also the second action potential signal.

Next, the action potential signal is preprocessed in preprocessing step S206, for example by applying one or more of the following functions: filtration, rectification, smoothing, RMS,

bandpass filtering, A/D conversion, Fourier Transformation, discrete wavelet transform. It will be clear that said list is not exhaustive and that other filtermg or preprocessing steps may be applied.

Step S206 may result in one or more preprocessed variants of the action potential signal. It will be clear that step S206 is optional and may be skipped.

After step S206 is performed, or after step S202 (and optionally step S204) when no preprocessing is applied, a muscle activity label is selected by performing steps S208 and S212, steps S210 and S212, or steps S208. S210 and S212. It is noted that step S212 may be incorporated into step S208 and/or S210. In step S208 one or more deterministic rules are applied to the (preprocessed) action potential signal to determine whether or not the muscle activity represented by the action potential signal is a bruxism activity. For example, if the action potential signal has a segment of ten seconds or longer during which the average action potential value is above 10mV, then it is determined that the action potential signal corresponds to a bruxism activity. In another example. if the action potential signal has a segment of 1 minute or longer for which the area under the curve is 600 mV*S, then it is determined that the action potential signal corresponds to a bruxism activity. The one or more deterministic rules may further comprise user-dependent rules.

For example, if the action potential signal has a segment of ten seconds or longer during which the average action potential value is above five times an action potential baseline predetermined for the user, then it is determined that the action potential signal corresponds to a bruxism activity. In another example, if the action potential signal has a segment of ten seconds or longer during which the average action potential value is above a maximum voluntary contraction level predetermined for the user, then it is determined that the action potential signal corresponds to a bruxism activity.

Alternatively or additionally to determining that the action potential signal corresponds to a bruxism activity, a bruxism probability may be calculated based on the observed action potential signal.

In step S210 a muscle activity label is determined using a machine learning model that is trained on a set of predetermined training signals. wherein each of the predetermined training signals in the set of predetermined training signals is labelled with one or more of the muscle activities from the set of muscle activities. In an example, the machine learning model is a recurrent convolutional neural network. In a further example, the recurrent convolutional neural network has the following lavers:2D convolution, batch normalization, bidirectional LSTM, fully connected layer with ReLU activation functions, fully connected layer with output size equal to the number of muscle activity labels. It will be clear that the configuration of the recurrent convolutional neural network can comprise additional layers, layers with a different configuration, and/or a repetition of the example layers as well.

It will be clear that the above is just one example of a machine learning model, and that many more machine learning models, not restricted to neural networks, are possible to be used.

In step S212 the muscle activity label is determined, for example by calculating a bruxism probability based on the outputs of steps S208 and S210 and comparing said probability with a predetermined bruxism threshold.

Next in step S214 biofeedback is provided to the user based on the muscle activity label determined in step S212. In an example, the intensity of the biofeedback is adjusted depending on the calculated probability, for example, the intensity of the biofeedback is strong when the bruxism probability is high, and the intensity of the biofeedback is low when the bruxism probability is low.

Step S214 may be skipped if the bruxism probability is below the predetermined bruxism threshold and/or if the muscle activity label is not associated bruxism activity.

In optional part 300, additional health measurement of the user is measured in step S302. For example, in S302 the heart rate and heart rate variability are measured over a period of 30 seconds every 3 minutes, the oxygenation is measured every 5 minutes, the temperature is measured every 10 minutes. and movement of the user is measured every 10 seconds. It will be clear that the above measuring intervals are provided as an example and are not limiting,

In step S306, the additional health measurements taken in step S302 are preprocessed similar to the preprocessing described in step S206. The preprocessed measurements may be combined with the data from steps S202 and/or S204 in preprocessing step S206 to be used as additional health data in step S208 and/or S210, the measurements may for example be an extra input vector in the machine learning model used in step S210.

After preprocessing, in step S308 additional health data is determined based on the additional health measurement. such as, a breathing rate. a sleep stage. a sleep quality, a sleep time, a breathing rate, and a stress level. The additional health data may also be used in steps S208 (arrow not shown) and/or in step S210, for example as additional input vector.

In step S312 it is determined if the additional health measurements and/or additional health data is indicative of a sleep apnea episode and/or a breath-holding episode. Output of step S312 may optionally be used as additional health data input in step S208 and/or S210 (arrows not shown). For example, the sleep apnea indicator may be an extra input vector in the machine learning model used in step S210. Furthermore, output of step S312 may optionally be used to adapt the bruxism probability. For example, the bruxism probability is increased when a sleep apnea episode is detected. Furthermore. if a sleep apnea episode and/or breath-holding episode is detected in step S312, additional biofeedback information may be provided to the user in step

S214.

It will be clear that the above-described steps may be executed by device 2, 102. For example, memory 140 may store instructions to perform the steps from parts 200 and 300 and processor 130 of device 102 may be configured to perform one or more steps from parts 200 and 300 by executed the corresponding instructions stored in memory 140.

It will be clear that the results of one of more of the intermediate results of the steps in parts 200 and 300 may be stored in the device 2, 102, for example in memory 140.

Figure 8 shows system 400 including device 402, user device 404 and server 406. In system 400, device 402 is wirelessly connected with user device 404 and server 406. Device 402 periodically connects to user device 404 to upload various data relating to bruxism to user device 404, such that user device 404 may present said data to the user. Device 402 further connects to user device 404 to retrieve user preferences relating to, for example, biofeedback setting, bruxism threshold settings and other settings. Device 402 further periodically connects to server 406 to retrieve updated machine leaming models and/or firmware updates. Device 402 or user device 404 may further connect to server 406 to (anonymously) exchange user data relating to bruxism.

The functions of the various elements shown in the figures, including any functional blocks labelled as “processors”, may be provided through the use of dedicated hardware as well as hardware capable of executing software in association with appropriate software. When provided by a processor, the functions may be provided by a single dedicated processor, by a single shared processor, or by a plurality of individual processors, some of which may be shared. Moreover. explicit use of the term “processor” or “controller” should not be construed to refer exclusively to hardware capable of executing software, and may implicitly include, without limitation, digital signal processor (DSP) hardware, network processor, application specific integrated circuit (ASIC), field programmable gate array (FPGA), read only memory (ROM) for storing software, random access memory (RAM). and non volatile storage. Other hardware, conventional and/or custom, may also be included. Similarly, any switches shown in the figures are conceptual only.

Their function may be carried out through the operation of program logic, through dedicated logic, through the interaction of program control and dedicated logic, or even manually, the particular technique being selectable by the implementer as more specifically understood from the context.

It should be appreciated by those skilled in the art that any block diagrams herein represent conceptual views of illustrative circuitry embodying the principles of the present invention.

Similarly, it will be appreciated that any flowcharts, flow diagrams. state transition diagrams, pseudo code, and the like represent various processes which may be substantially represented in computer readable medium and so executed by a computer.

The present invention is by no means limited to the above described preferred embodiments thereof. It will be clear that one or more features from an embodiment be combined with one or more features from one or more other embodiments according to the invention. It will further be clear that terms like “received”, “retrieved”. “send”, or any other term which suggest any form of direction of communication, are used as being non limited and should merely be interpreted to communication being present or possible. E g., received may be interpreted as meaning retrieved and vice versa. It will also be clear that terms like “measurement”, “signal”, “value”, “data element”. “time series” or any other term which suggest any form of describing a piece of information are used as being non limited and should merely be interpreted as referring to a value or a series of values or information that can be measured, processed. calculated. displayed. and otherwise be handled.

The rights sought are defined by the following claims within the scope of which many modifications can be envisaged.

CLAUSES

I. Device comprising a first sensing electrode and a reference electrode, wherein the first sensing electrode and the reference electrode are arranged on the device such that, when the device is wom in a using position such that the first sensing electrode is placed on a first temporalis muscle or a first masseter muscle of a user, the reference electrode is placed on a part of the forehead of the user, wherein the device further comprises: — a signal module that is electronically connected to the first sensing electrode and the reference electrode and configured to measure, over time, a first plurality of voltage differential values between the first sensing electrode and the reference electrode, wherein the voltage differential values are indicative of a unilateral muscle activity of the first temporalis muscle or the first masseter muscle on which the first sensing electrode is placed; — a detection module that is connected to the signal module via a wireless or wired connection and that is configured to obtain a first action potential signal from the signal module, wherein the first action potential signal comprises and/or is derived from at least some of the first measured voltage differential values, wherein the detection module is further configured to select a muscle activity label from a set of predetermined muscle activity labels, the selecting comprising analyzing the first action potential signal; and — a biofeedback module configured to obtain the selected muscle activity label and to provide the user with a biofeedback signal in reaction to the selected muscle activity label being indicative of a bruxism activity. 2. Device according to clause 1, wherein the analyzing the first action potential signal comprises calculating a bruxism probability of the first action potential signal being indicative of a bruxism activity and wherein the selecting further comprises selecting one of the muscle activity labels that is indicative of a bruxism activity when the bruxism probability is above a predetermined bruxism threshold. 3. Device according to clause 1 or 2, wherein analyzing of the first action potential signal comprises applving a predetermined set of rules to the first action potential signal, the predetermined set of rules comprising data characteristics that are indicative of the muscle activity corresponding to the first action potential signal being a bruxism activity.

4. Device according to clause 3, wherein the data characteristics comprise one or more action potential intensity threshold, one or more action potential intensity durations, and/or a combination thereof and wherein applying the predetermined set of rules comprises determining one or more equivalences between the data characteristics and the first action potential and selecting the predetermined muscle activity label based on the equivalences.

5. Device according to clause 1 or 2, wherein selecting of the muscle activity label comprises using the first action potential signal as input to a predetermined machine learning model trained to detect bruxism; and receiving the bruxism probability and/or the muscle activity label from the machine leaming model.

6. Device according to any of the previous clauses, wherein the device further comprises a second sensing electrode which is arranged on the device such that. when the device is worn in the using position such that the first sensing electrode is placed on the first temporalis muscle or the first masseter muscle of the user, the second sensing electrode is placed on a second temporalis muscle or a second masseter muscle of the user that is located on another side of the forehead of the user compared to the first temporalis muscle or the first masseter muscle. and wherein the signal module is further electronically connected to the second sensing electrode and is further configured to measure, over time, a second plurality of voltage differential values between the second sensing electrode and the reference electrode, wherein the second plurality of voltage differential values are indicative of a unilateral muscle activity of the second temporalis muscle or the second masseter muscle on which the second sensing electrode 1s placed, and wherein the detection module is further configured to obtain a second action potential signal that comprises and/or is derived from at least some of the second measured voltage differential values and wherein the selecting of the muscle activity label by the detection module further comprises analyzing the second action potential signal.

7. Device according to any of the previous clauses, wherein the surface electrode sensors are one of: rubber electrodes, paint electrodes, textile electrodes, dry metallic electrodes, silver coated electrodes, or any other suitable electrodes, preferably silver coated electrodes.

8. Device according to any of the previous clauses, wherein the device further comprises a headband, wherein the first sensing electrode and the reference electrode are attached to the headband.

9. Device according to any one of the previous clauses, wherein the device further comprises a health information module comprising one or more user health information sensors configured to measure, over time, a plurality of health information data measurements relating to the user, the health user sensors comprising one or more of: an optical sensor, an electroencephalography (EEG) sensor, a heart rate sensor, a blood oxygen saturation sensor, a motion sensor, and/or a body temperature sensor: and wherein the signal module and/or the detection module are further electronically connected to the health information module and wherein the detection module is further configured to obtain, optionally via the signal module. at least a part of the health information data measurements and wherein the selection of the muscle activity label by the detection module further comprises analyzing the at least part of the health information data measurements and/or derivatives thereof. 10. Device according to clause 9, wherein the obtaining at least a part of the health information data measurements and analyzing at least part of the health information data measurements by the detection module comprising one or more of: — obtaining at least a part of the health data measurements from the optical sensor, the

EEG sensor, the heart rate sensor, the motion sensor and/or the body temperature sensor; — determining one or more of: a sleep stage, a sleep quality, a sleep time, and a stress level, of the user by analyzing the health information data measurements and/or values derived from the health information data; and wherein the detection module is electronically connected to the sleep stage detection module and wherein the selection of the muscle activity label by the detection module further comprises observing the determined one or more of: a sleep stage, a sleep quality, a sleep time, and astress level, of the user. 11. Device according to any one of the clauses 9 or 10, wherein the detection module is further configured to detect a sleep apnea episode and/or a breath-holding episode by analyzing at least a part of information data elements and wherein the biofeedback module is further configured to provide biofeedback to the user in response to detection of the sleep apnea episode by the detection module. 12. Device according to any one of the clauses 9, 10, or 11, in combination with clause 2, wherein the analyzing the at least part of the health information data elements comprises: — obtaining a heart rate and/or a heart rate variability;

— detecting an increase in heart rate and/or a decrease in heart rate variability, wherein the increase in heart rate and/or the decrease in heart rate variability are indicative of a micro-arousal of the user; and, in response to the detection of the increase in heart rate and/or decrease in heart rate variability, increasing the bruxism probability and/or decreasing the predetermined bruxism threshold.

13. Device according to any one of the previous clauses, wherein the biofeedback module comprises a visual feedback module, an audio feedback module, a haptic feedback module and/or an electric feedback module.

14. Device according to any one of the previous clauses, wherein the biofeedback module is configured to dynamically determine one or more biofeedback parameters, the biofeedback parameters comprising a biofeedback intensity, a biofeedback duration, a biofeedback frequency,

wherein the determination of the biofeedback parameters comprises:

i. observing on one or more of the following: an age of the user, a sex of the user, previous biofeedback provided to the user, one or more health information data measurements, a detected sleep stage, and other user specific characters; and/or il. obtaining a muscle activity intensity from the detection module and determining the biofeedback intensity by positively correlating the biofeedback intensity with the muscle activity intensity, wherein the muscle activity intensity is determined by the detection module by observing the first action potential signal and/or, when in combination with clause 6, the second action potential.

15. Device according to any one of the previous clauses, wherein the muscle activity labels comprise one or more labels indicative of a bruxism activity and one or more labels indicative of non-bruxism activity.

16. Device according to clause 15, wherein the labels indicative of a bruxism activity are one or more of: bruxism, grinding, clenching, gnashing; and wherein the labels indicative of a non- bruxism activity are one or more of: non-bruxism, chewing, yawning, talking, swallowing, blowing, whistling, playing a musical instrument. and other non-bruxism activities of the mouth.

17. Device according to any one of the previous clauses, the device further comprismg a communication module comprising a wired interface and/or a wireless interface, wherein the communication module is configured to connect to a user device via the wired interface and/or the wireless interface and to enable the device and the user device to exchange bruxism information data elements relating to one or more selected muscle activity labels and/or the plurality of action potential values and/or the plurality of voltage differential value measurements and/or values derived from the plurality of voltage differential value measurements and/or one or more of the health information data measurements and/or biofeedback provided to the user.

18. User device comprising a processor, a memory, a display, and a communication module, wherein the memory has instruction stored thereon that, when executed by the processor, enables the user device to:

— connect to the device according to clause 17; — to receive one or more bruxism information data elements; and — to display at least one of the one or more bruxism information data elements and/or derivates of at least one of the one or more bruxism information data elements on the display.

19. Server configured to train a machine learning model using a set of predetermined training signals, wherein each of the predetermined training signals in the set of predetermined training signals is labelled with one of the muscle activities from the set of muscle activities and wherein the server further comprises communication means configured to connect to one or more devices according to clause 5 or clauses 6 — 17 in combination with clause 5, and to upload the trained machine leaming module to the user device.

20. System comprising the device according to any one of the clauses 1 — 17 and the user device according to clause 18, wherein the device is operatively connectable to the user device,

optionally the system further comprising a server according to clause 19, wherein the server is operatively connectable to the device.

21. Method for use of the device, the method comprising:

— obtaining the device according to any of the clauses 1 — 17: — placing the device on a head in a wearing position; and, — receiving biofeedback from the device in response to the device detecting a bruxism activity.

22. Method for detecting and classifying bruxism, the method comprising:

— obtaining a first action potential signal derived from or comprising a plurality of voltage differential values between a first sensing electrode placed on a first temporalis muscle or a first masseter muscle of a user and a reference electrode placed on a forehead of the user, wherein the voltage differential values are indicative of a unilateral muscle activity of the first temporalis muscle or the first masseter muscle on which the first sensing electrode is placed: and — selecting a muscle activity label from a set of predetermined muscle activity labels, the selecting comprising analyzing the first action potential signal. 23. Method according to clause 22, wherein the determining of the muscle activity label comprises applying a preselected set of deterministic rules to the first action potential signal, wherein the preselected set of deterministic rules comprises a multiple of signal characteristics with each a corresponding muscle activity label from the set of muscle activities labels. 24. Method according to clause 23, wherein the multiple of signal characteristics are predetermined for the user.

25. Method accordmg to clause 22, wherein the determining of the muscle activity label comprises providing the first action potential as input to a machine learning model, wherein the machine learning model is trained to determine the muscle activity label by classifving the first action potential when it is provided as an input to the machine learning model.

26. Method according to clause 25, wherein the machine learning model further uses one or more of: the second action potential signal, a sleep stage, a user’s sex, a user’s age, a user's weight, a time of day, a time of year, and/or one or more historical observations relating to bruxism activities of the user. as the input to the machine learning model.

27. Method according to clauses 25 or 26, wherein the machine learning model is a neural network, preferably a recurrent convolutional neural network. 28. Method according to clauses 25, 26 or 27 wherein the machine learning model is trained on a set of predetermined training signals, wherein each of the predetermined training signals in the set of predetermined training signals is labelled with one of the muscle activities from the set of muscle activities. 29. Device comprising a processor. a memory, a first sensing electrode and a reference electrode, wherein the first sensing electrode and the reference electrode are arranged on the device such that, when the device is worn in a using position such that the first sensing electrode is placed on a first temporalis muscle or a first masseter muscle of a user, the reference electrode is placed on a part of the forehead of the user; wherein the memory contains instruction that enables the processor to obtain a first plurality of voltage differential values between the first sensing electrode and the reference electrode and to execute the method according to any of the clauses 22 — 28.

Claims

CONCLUSIONS

1. An apparatus comprising a first measuring electrode and a reference electrode, the first measuring electrode and the reference electrode being arranged on the apparatus such that, when the apparatus is worn in a position of use such that the first measuring electrode is positioned on a first jaw muscle or a first masseter muscle of a user, the reference electrode is positioned on a portion of the forehead of the user, further comprising: — a signal module electronically connected to the first measuring electrode and the reference electrode and adapted to measure, over time, a first plurality of differential voltage values between the first measuring electrode and the reference electrode, the differential voltage values being indicative of unilateral muscle activity of the first jaw muscle or first masseter muscle on which the first measuring electrode is positioned; — a detection module connected to the signal module via a wireless or wired connection and adapted to obtain a first action potential signal from the signal module. wherein the first action potential signal comprises and/or is derived from, at least a portion of the first differential voltage values, the detection module further configured to select a muscle activity label from a set of predetermined muscle activity labels, the selecting comprising analyzing the first action potential signal; — a biofeedback module configured to obtain the selected muscle activity label and to provide the user with a biofeedback signal in response to the selected muscle activity label being indicative of bruxism activity.

2. The apparatus of claim 1, wherein analyzing the first action potential signal comprises calculating a bruxism probability that the first action potential signal is indicative of a bruxism activity and wherein selecting further comprises selecting one of the muscle activity labels indicative of a bruxism activity when the bruxism probability is above a predetermined bruxism threshold.

The apparatus of claim 1 or 2, wherein analyzing the first action potential signal comprises applying a predetermined set of rules to the first action potential signal, the predetermined set of rules comprising data characteristics indicative of the muscle activity corresponding to the first action potential signal being a bruxism activity.

The apparatus of claim 3, wherein the data characteristics comprise one or more action potential intensity thresholds, one or more action potential intensity durations, and/or a combination thereof, and wherein applying the predetermined set of rules comprises determining one or more correspondences between the data characteristics and the first action potential signal and selecting muscle activity labels based on the correspondences.

5. The apparatus of claim 1 or 2, wherein selecting the muscle activity label comprises using the first action potential signal as input to a predetermined machine learning model trained to recognize bruxism; and receiving the bruxism probability and/or the muscle activity labels from the machine learning model.

6. The apparatus of any preceding claim, wherein the apparatus further comprises a second measurement electrode arranged on the apparatus such that, when the apparatus is worn in the user position such that the first measurement electrode is positioned on the user's first jaw muscle or first masseter muscle, the second electrode is positioned on a second jaw muscle or second masseter muscle of the user that is located on a different side of the user's forehead than the first jaw muscle or masseter muscle, and wherein the signal module is further electronically connected to the second measurement electrode and is further configured to measure over time a second plurality of differential voltage values indicative of unilateral muscle activity of the second jaw muscle or second masseter muscle on which the second measurement electrode is positioned, and wherein the detection module is further configured to obtain a second action potential signal comprising and/or derived from at least a portion of the second differential voltage values, and wherein selecting the muscle activity label by the detection module further comprises analyzing the second action potential signal.

7. Apparatus according to any one of the preceding claims, wherein the measuring electrode(s) and the reference electrode are surface electrode sensors which are one of the following types: rubber electrode, paint electrode, textile electrode, dry metal electrode, silver coating electrode, or any other suitable electrode, preferably silver coating electrode.

8. The apparatus of any preceding claim, wherein the apparatus further comprises a headband, the first measurement electrode and the reference electrode being disposed on the headband.

9. The apparatus of any preceding claim, wherein the apparatus further comprises a health information module comprising one or more user health information sensors configured to measure, over time, a plurality of health information data measurements related to the user, the user health information sensors comprising one or more of: an optical sensor, an electroencephalography sensor, an EEG sensor, a heart rate sensor, an oxygen saturation sensor, a motion sensor, and/or a body temperature sensor; and wherein the signal module and/or the detection module are further electronically connected to the health information module and wherein the detection module is further configured to obtain, optionally via the signal module, at least a portion of the health information data measurements and wherein selecting the muscle activity label by the detection module further comprises analyzing at least a portion of the health information data measurements and/or derivatives thereof.

10. The apparatus of claim 9, wherein obtaining at least a portion of the health information data measurements and analyzing at least a portion of the health information data measurements by the detection module comprises one or more of: — obtaining at least a portion of the health information data measurements from the optical sensor, the EEG sensor, the heart rate sensor, the motion sensor and/or the body temperature sensor; — determining one or more of: a sleep stage, a sleep quality, a sleep time, a stress level, of the user by analyzing the health information data measurements and/or derivatives thereof; and — wherein selecting the muscle activity label further comprises sensing the determined one or more of: a sleep stage, a sleep quality, a sleep time. a stress level, of the user.

The apparatus of any of claims 9 or 10, wherein the detection module is further configured to detect a sleep apnea episode and/or a breath hold episode by analyzing at least a portion of the health information data measurements and wherein the biofeedback module is further configured to provide biofeedback to the user in response to the detection module detecting the sleep apnea episode and/or the breath hold episode.

12. The apparatus of any of claims 9, 10 or 11, in combination with claim 2, wherein analyzing at least some of the health information data measurements comprises: — obtaining a heart rate and/or a heart rate variability; — detecting an increase in heart rate and/or a decrease in heart rate variability, wherein the increase in heart rate and/or the decrease in heart rate variability is indicative of a microarousal of the user, and — in response to detecting the increase in heart rate and/or the decrease in heart rate variability, increasing the bruxism probability and/or decreasing the predetermined bruxism threshold.

13. The apparatus of any preceding claim, wherein the biofeedback module comprises a visual feedback module, an audio feedback module, a haptic feedback module, and/or an electrical feedback module.

14. The apparatus of any preceding claim, wherein the biofeedback module is configured to dynamically determine one or more biofeedback parameters, the biofeedback parameters including a biofeedback intensity, a biofeedback duration, a biofeedback frequency, the determining of the biofeedback parameters comprising: — sensing one or more of the following: an age of the user, a gender of the user, previous biofeedback provided to the user, one or more of the health information data measurements, an established sleep stage, and other user specific characteristics: and/or — obtaining a muscle activity intensity from the sensing module and determining the biofeedback intensity by positively correlating the biofeedback intensity with the muscle activity intensity, the muscle activity intensity being determined by the sensing module by sensing the first action potential signal and/or, when in combination with claim 6, the second action potential signal.

15. The apparatus of any preceding claim, wherein the muscle activity labels comprise one or more labels indicative of a bruxism activity and one or more labels indicative of a non-bruxism activity.

16. The apparatus of claim 15, wherein the labels indicative of bruxism activity include one or more of: bruxism, teeth grinding, clenching, teeth gnashing; and wherein the labels indicative of non-bruxism activity include one or more of: no-bruxism, chewing, yawning, talking, swallowing, blowing, whistling, playing an instrument, and other non-bruxism activities of the mouth.

17. The apparatus of any preceding claim, wherein the apparatus further comprises a communications module comprising a wired interface and/or a wireless interface, the communications module being adapted to connect to a user device via the wired interface and to enable the apparatus and the user device to exchange bruxism information relating to one or more selected muscle activity labels and/or the plurality of action potential signals and/or the plurality of differential voltage values and/or values derived from the plurality of differential voltage values and/or one or more of the health information data measurements and/or biofeedback provided to the user.

18. A user device comprising a processor, a memory, a display, and a communications module, the memory having instructions stored thereon which, when executed by the processor, enable the user device 1n to: — connect to the device according to claim 17; — receive one or more bruxism information data elements; and — display the at least one of the one or more bruxism information data elements and/or derivatives of the bruxism information data elements on the display.

19. A server configured to train a machine learning model using a set of predetermined training signals, each of the predetermined training signals in the set being labeled with one of the muscle activities from the set of muscle activities and wherein the server further comprises communication means configured to connect to one or more devices according to claim 5 or claims 6 to 17 in combination with claim 5, and to upload the trained machine learning model to the user's device.

20. A system comprising the apparatus of any one of claims 1 to 17 and the user device of claim 18, the apparatus being operatively connectable to the user device, the system optionally further comprising the server of claim 19, the server being operatively connectable to the apparatus.

21. A method of using the apparatus, the method comprising: — obtaining an apparatus according to any one of claims 1 to 17; — placing the apparatus on a head in a wearing position; and, — receiving biofeedback from the apparatus in response to the apparatus detecting a bruxism activity.

22. A method for detecting and classifying bruxism, the method comprising: — obtaining a first action potential signal derived from or composed of a plurality of differential voltage values between a first measuring electrode positioned on a first jaw muscle or a first masseter muscle of a user and a reference electrode positioned on a forehead of the user, the differential voltage values being indicative of unilateral muscle activity of the first jaw muscle or first masseter muscle on which the first measuring electrode is positioned; and, — selecting a muscle activity label from a set of predetermined muscle activity labels, the selecting comprising analyzing the first action potential signal.

23. The method of claim 22, wherein determining the muscle activity labels comprises applying a preselected set of deterministic rules to the first action potential signal, the selected set of deterministic rules comprising a plurality of signal characteristics each having a corresponding muscle activity label from the set of muscle activity labels.

24. The method of claim 23, wherein the plurality of signal characteristics are predetermined for the user.

25. The method of claim 22, wherein determining the muscle activity label comprises providing the first action potential signal as input to a machine learning model, the machine learning model being trained to determine a muscle activity label by classifying the first action potential signal when provided as input to the machine learning model.

26. The method of claim 25, wherein the machine learning model further accepts as input to the machine learning model one or more of the second action potential signal, a sleep stage, a gender of the user, an age of the user, a weight of the user, a time of day, a time of year, and/or one or more historical observations relating to bruxism activities of the user.

27. Method according to claim 25 or 26, wherein the machine learning model is a neural network, preferably a recurrent convolutional neural network.

28. The method of any of claims 25, 26 or 27, wherein the machine learning model is trained on a set of predetermined training signals, wherein each of the predetermined training signals in the set of predetermined training signals is labeled with one of the muscle activities from the set of muscle activities.

29. An apparatus comprising a processor, a memory, a first measuring electrode and a reference electrode, the first measuring electrode and the reference electrode being disposed on the apparatus such that, when the apparatus is worn in a position of use such that the first electrode is disposed on a first jaw muscle or a first masseter muscle of a user, the reference electrode is disposed on a portion of the forehead of the user; the memory containing instructions enabling the processor to measure a plurality of first differential voltage values between the first measuring electrode and the reference electrode and to perform the method of any of claims 22 to 28.