Systems and methods for stimulating peripheral nerves

Abstract

Obstructive sleep apnea resulting from upper airway obstruction can cause severe daytime sleepiness, along with many long-term comorbidities associated with this disorder, such as hypertension. Common causes of obstructive sleep apnea include complete concentric collapse of the soft palette or retraction of the genioglossus muscle into the upper airway, which obstructs breathing. Retraction of the genioglossus muscle into the upper airway may be due to muscle fatigue resulting from increased activation during wakefulness, or a larger number of type II muscle fibers contributing to muscle fatigue. It can also be neurological, where the hypoglossal nerve, which controls most of the upper airway muscles, fails to adequately stimulate the genioglossus muscle to prevent posterior movement and pharyngeal obstruction. Described herein are intelligent, personalized closed-loop neuromodulatory systems and methods that inhibit posterior movement of the genioglossus muscle through stimulation of specific branches and muscle motor points of the hypoglossal nerve by providing optimal stimulation using muscle feedback from electromyographic sensors. Alternative embodiments for treating neuropathic pain and bladder dysfunction are also included.

Description

[Technical Field]

[0001] Field of Invention

[0001] The present invention generally relates to a neurostimulation medical device, and more particularly to a device that is worn on the body or implantable for sensing and stimulating nerves and muscles. [Background technology]

[0002] background

[0002] Obstructive sleep apnea (OSA) is a common disorder, estimated to affect one in four people, and is linked to several comorbidities, including hypertension and heart disease. OSA is the result of upper airway obstruction, which leads to temporary cessation of breathing and consequently awakening from sleep. The obstruction can result from complete concentric collapse of the soft palate and lateral walls of the throat, or more commonly, from the genioglossus (GG) muscle being unable to maintain its position during sleep and falling into the airway, causing obstruction. Adequate airway force is required to ensure adequate dilation and patency of the upper airway. The GG is one of four pairs of extrinsic muscles of the tongue that are responsible for tongue movement. While four intrinsic muscles, separated from any bone, work to change the shape of the tongue, four extrinsic muscles control the position of the tongue and attach to the hyoid bone. The GG muscle is a fan-shaped pair responsible for tongue protrusion, forming most of the body of the tongue, originating from the mental spine of the mandible and adjacent to the hyoid bone at the base of the tongue on either side of the midline area. The decreased strength leading to OSA events is thought to be due to either reduced neural drive leading to airway narrowing or collapse of the GG muscle when it falls into the airway, or GG muscle fatigue. This abnormal behavior of the GG in OSA sufferers is indicated by increased activation levels during wakefulness, leading to fatigue during sleep, and in many cases, affected individuals have a higher proportion of type II fibers, which are known to fatigue faster than type I fibers.

[0003]

[0003] The primary treatment for OSA is currently the use of continuous positive airway pressure (CPAP), which is the standard of care. Treatment with CPAP significantly reduces the incidence of apnea events, regardless of severity. Continuous positive pressure is generated by a CPAP machine and delivered from the CPAP machine through a tube to a face mask. The face mask is worn overnight, and positive pressure is applied to either the nasal cavity or the oral cavity, thereby clearing the airway and preventing any kind of obstruction during the inspiratory portion of the respiratory cycle. Treatment with CPAP requires wearing a mask overnight with the hose connected to the machine, which can be cumbersome for the patient or their spouse. Historically, this treatment has had low tolerance due to its cumbersome nature, with patients complaining of shortness of breath and discomfort. As a result, adherence and compliance rates are low. As a result of the resistance to CPAP, more readily tolerable alternative treatments have been developed.

[0004]

[0004] Chin-lingual protraction is one option, in which the surgeon cuts the mandible and moves it forward, and therefore the mandible also moves forward. This reduces the degree of block but may not resolve the condition in many cases. In some cases, this surgery may change the structure and appearance of the patient's face. Oral appliances such as mandibular protraction devices (MADs) also move the mandible forward to reduce the degree of block when the GG falls into the pharynx.

[0005]

[0005] A relatively new therapy employing neuromodulatory techniques is hypoglossal nerve (HPN) stimulation. This newly emerging therapy uses an implantable stimulator (IPG), similar to a pacemaker device implanted in the chest, with leads extending from the stimulator to the neck and then to the hypoglossal nerve branch that stimulates the GG muscle. Additional leads connect respiratory pressure sensors to the intercostal muscle area to monitor respiration and ensure that stimulation occurs only during inspiratory respiration, reducing the likelihood of extreme fatigue of the GG, which may already be fatigued due to excessive daytime activity or may have an excessive number of type II muscle fibers. The first neurostimulator approved by the FDA for the treatment of OSA was developed by Inspire Medical, Inc. and includes a cuff electrode attached to the distal branch of the hypoglossal nerve that specifically stimulates one side of the GG muscle. It is a Class III device due to being a permanent implant and the invasiveness of the surgery required to implant the device. While this therapy has shown effectiveness in treating OSA, it is highly invasive, and patients face surgical risks, including the potential for long-term adverse events commonly associated with surgical implantation, such as infection, lead fragmentation and migration, corrosion, and fibrosis. This technique is prescribed only to patients who cannot tolerate or are unwilling to comply with CPAP.

[0006]

[0006] Non-invasive neuromodulatory techniques for treating OSA have been attempted for many years, mostly in educational settings. None have been able to achieve any level of significance. There are several reasons for the above failures. Fundamentally, it is crucial that the GG muscle is activated in order to achieve any degree of airway patency through stimulation. For this muscle to be successfully activated, the motor point or distal medial nerve of the HGN that controls GG must be successfully activated. Successful activation requires not only that the correct location is precisely targeted, but also that it is stimulated at a stimulation level that can trigger an action potential response, which leads to the firing of motor fibers and thereby causes muscle contraction. This is currently extremely difficult compared to invasive implants that surround the nerve fibers responsible for GG muscle contraction with stimulating electrodes. Stimulating a muscle can only increase blood flow if the wrong location is stimulated. Even with correct electrode placement, it is not sufficient due to the constant movement through snoring and the typical restless behavior seen during sleep, especially in OSA sufferers. This invention describes an improved non-invasive nerve stimulation device that overcomes the shortcomings of the prior art. [Overview of the Initiative] [Means for solving the problem]

[0007] overview

[0007] Embodiments described herein include closed-loop systems and methods for monitoring GG muscle activity and providing feedback to maintain upper airway patency, determining the optimal set of electrode stimulations and the optimal stimulation to deliver in real time.

[0008]

[0008] In a preferred embodiment directed toward the treatment of OSA, EMG signals are collected from multiple electrodes placed in the submental region of the chin to determine the optimal sensing electrode and the optimal stimulating electrode. As the stimulus is delivered to the hypoglossal nerve via the optimal stimulating electrode, feedback from the EMG waveform from the optimal sensing electrode is used to control the stimulus in a closed-loop system.

[0009]

[0009] Additional embodiments include a method for using this EMG signal as a feedback process variable to control the amplitude of stimulation to peripheral nerves. Other embodiments of the present invention include a method for determining the optimal sensing electrode and the optimal stimulating electrode and a method for updating which electrodes are the optimal sensing electrode and stimulating electrode. Other embodiments of the present invention discuss setting the appropriate frequency of stimulation and detecting sleep. In other embodiments, the present invention describes how the inhalation phase is identified. In yet another embodiment, the present invention describes how sleep apnea events can be predicted.

[0010]

[0010] According to one aspect of the present invention, a system for treating obstructive sleep apnea is provided, and this system is An array of multiple electrodes, A memory capable of storing computer executable instructions, and a processor configured to facilitate the execution of executable instructions stored in the memory, Includes, The instruction is sent to the processor, The process involves receiving EMG signals from an array of multiple electrodes placed in the submental region of an individual, filtering the EMG signals to generate a signal envelope, measuring genioglossal muscle activity from the signal envelope to determine the optimal sensing electrode, pulsing each electrode and measuring the response at the optimal sensing electrode to determine the optimal stimulating electrode, identifying the inspiratory and expiratory breathing phases from the signal envelope, delivering stimulation to the hypoglossal nerve via the optimal stimulating electrode at the start of the inspiratory breathing phase, and confirming from the optimal sensing electrode that the stimulation is useful for moving the genioglossal muscle. Have them do it.

[0011]

[0011] Selectively determining the optimal sensing electrode further includes activating the genioglossus muscle and performing an operation to position the electrode in such a way that the response is best.

[0012] Optionally, the command further provides continuous closed-loop feedback from the EMG waveform of the optimal sensing electrode to confirm that the optimal stimulating electrode is stimulating the genioglossus muscle.

[0013]

[0013] Optionally, the optimal stimulating electrode can be updated to a new electrode pair.

[0014]

[0014] Optionally, the optimal sensing electrode can be updated to a new electrode pair.

[0015]

[0015] Optionally, the closed-loop feedback ends and enters a safety setting when a predefined threshold is exceeded.

[0016]

[0016] Optionally, the stimulation has an amplitude and a frequency, and the frequency is adjusted based on the sensation perceived by an individual.

[0017]

[0017] Optionally, the command further includes measuring the M wave and / or H reflex response at the optimal sensing electrode and adjusting the stimulation based on the M wave and / or H reflex response.

[0018]

[0018] Optionally, the start of the inspiration breathing phase is predicted based on the previous cycle time and the average respiratory rate.

[0019]

[0019] Optionally, sleep is detected by measuring fluctuations in the respiratory frequency and tidal volume.

[0020]

[0020] Optionally, apnea events during sleep are predicted based on changes in the EMG signal during inspiration.

[0021]

[0021] Optionally, the change in the EMG signal during inspiration includes a decrease in genioglossus muscle activity identified from the integrated EMG signal of the genioglossus muscle activity measured by the optimal sensing electrode.

[0022] Optionally, alternatively, stimulation is provided by an optimal stimulation electrode before a predicted inspiration onset occurs.

[0023]

[0023] Optionally, the predicted inspiration onset is given by the formula ti’ p = te k + (ti k - te k-1 ) - e, where ti’ <00​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​

[0026]

[0026] Optionally, the system further includes one or more microphones positioned on either side of the throat to record sounds from the airway.

[0027]

[0027] Optionally, the system further includes an accelerometer positioned in the center of the subchinnic triangle.

[0028]

[0028] Optionally, the system may be further configured to determine from accelerometer data whether the user of the system is asleep or not based on the user's posture and / or sudden movements of the user.

[0029]

[0029] Optionally, the system may be further configured to switch off stimulation from the optimal stimulation electrode in response to the determination of the user's posture from accelerometer data.

[0030]

[0030] Optionally, the accelerometer includes a dynamic component configured to determine one or more of the user's snoring, speaking, breathing, coughing, and choking.

[0031]

[0031] The optimal sensing electrode is determined to be the bipolar electrode pair having the maximum energy in each orientation.

[0032]

[0032] The system is optionally configured to deactivate the stimulation from the optimal stimulation electrode in at least one of the user's sleeping postures, based on user sleep apnea reduction data.

[0033]

[0033] Optionally, the system further includes a temperature sensor and / or an SpO2 sensor.

[0034]

[0034] Optionally, this system may be:

number

[0035]

[0035] Optionally, the system may be further configured to perform a transfer function to filter out stimulus-induced artifacts from the recorded EMG signal.

[0036]

[0036] Optionally, the stimulus waveform is provided with a sufficiently low amplitude so as not to elicit a motor response.

[0037]

[0037] Optionally, the system may be further configured to switch off the stimulation from the stimulation electrodes in response to the detection of the occurrence of stimulation-induced muscle fatigue.

[0038]

[0038] Optionally, stimulus-induced muscle fatigue is determined by comparing the evoked genioglossus muscle EMG center frequencies at the start and end of the stimulus.

[0039]

[0039] According to a further aspect of the present invention, an intraoral device is provided which includes the system described in any one of claims 1 to 29.

[0040]

[0040] The oral device may optionally be a mandibular anterior fixation device, a mouthguard, or a retainer.

[0041]

[0041] Optionally, the oral device is a boil and bite mouthguard.

[0042]

[0042] According to a further aspect of the present invention, a method for treating obstructive sleep apnea is provided, which is To provide an array of multiple electrodes, Receiving EMG signals from an array of multiple electrodes placed in the subchinicular region of an individual, Filtering the EMG signal to generate a signal envelope, The optimal sensing electrode is determined by measuring genioglossal muscle activity from the signal envelope, Each electrode is pulsed, the response at the optimal sensing electrode is measured, and the optimal stimulating electrode is determined. Identifying the inspiratory and expiratory breathing phases from the signal envelope, Delivering stimulation to the hypoglossal nerve via the optimal stimulating electrode at the start of the inspiratory breathing phase, To confirm from the optimal sensing electrode that stimulation is useful in moving the genioglossus muscle, Includes.

[0043]

[0043] The step of optionally determining the optimal sensing electrode further includes activating the genioglossus muscle and performing an operation to position the electrode in such a way that the response is best.

[0044]

[0044] Optionally, the method further includes providing continuous closed-loop feedback from the EMG waveform from the optimal sensing electrode to confirm that the optimal stimulating electrode is stimulating the genioglossus muscle.

[0045]

[0045] Optionally, the method further includes updating the optimal stimulating electrode with a new electrode pair.

[0046]

[0046] Optionally, the method further includes updating the optimal sensing electrode with a new electrode pair.

[0047]

[0047] Optionally, the method further includes terminating the closed-loop feedback and setting it to a safety configuration if a predetermined threshold is exceeded.

[0048]

[0048] The stimulus optionally has amplitude and frequency, and the method further includes adjusting the frequency based on the sensation perceived by the individual.

[0049]

[0049] Optionally, the method further includes measuring the M-wave and / or H-wave reflex response at an optimal sensing electrode and adjusting the stimulus based on the M-wave and / or H-wave reflex response.

[0050]

[0050] Optionally, the onset of the inspiratory breathing phase is predicted based on the previous cycle time and mean respiratory rate.

[0051]

[0051] Optionally, the method further includes detecting sleep by measuring changes in respiratory rate and tidal volume.

[0052]

[0052] Optionally, the method further includes predicting future sleep apnea events based on changes in the EMG signal during inspiration.

[0053]

[0053] Optionally, changes in the EMG signal during inspiration include a decrease in chinoglossal muscle activity, identified from the integrated EMG signal of chinoglossal muscle activity measured by the optimal sensing electrode.

[0054]

[0054] Optionally, as an alternative to the step of providing stimulation at the start of the inspiratory breathing phase, the method includes a step of providing stimulation before the expected start of inspiration occurs.

[0055]

[0055] Optionally, this method is formulated as ti' k+1 =te k +(ti k -te k-1 The further step includes identifying the expected intake onset using )-e, where ti' k+1 This indicates the time when the (k+1)th intake cycle is expected to begin, and te k This indicates the time when the k-th exhalation cycle began, and ti k This indicates the time when the k-th intake cycle began, and te k-1 indicates the time when the (k-1)th exhalation cycle began, and e indicates the margin of error.

[0056]

[0056] The amplitude of the output stimulus signal of the optimal stimulation electrode is given by formula:

number

number

number

[0057]

[0057] Optionally, the method further includes the step of updating the SP value in response to a determination that the user is in a sleep state.

[0058]

[0058] Optionally, the method further includes the step of providing one or more microphones positioned on either side of the throat to record sounds from the airway.

[0059]

[0059] Optionally, the method further includes the step of providing an accelerometer positioned in the center of the subchinnic triangle.

[0060]

[0060] Optionally, the method further includes the step of determining from accelerometer data whether the user of the system is asleep based on the user's posture and / or on the user's sudden movements.

[0061]

[0061] Optionally, the method further includes the step of switching off the stimulation from the optimal stimulation electrode in response to the determination of the user's posture from accelerometer data.

[0062]

[0062] Optionally, the method further includes the step of deactivating the optimal stimulation from the stimulation electrode in at least one of the user's sleeping apnea respiratory reduction data.

[0063]

[0063] Optionally, this method may be:

number

[0064]

[0064] Optionally, the stimulus waveform is provided with a sufficiently low amplitude so as not to elicit a motor response.

[0065]

[0065] Optionally, the method further includes the step of switching off the stimulation from the stimulation electrode in response to the determination that the occurrence of stimulation-induced muscle fatigue has been detected.

[0066]

[0066] Optionally, stimulus-induced muscle fatigue is determined by comparing the evoked genioglossus muscle EMG center frequencies at the start and end of the stimulus.

[0067] Brief explanation of the drawing

[0001] These and other possible embodiments, features, and advantages of the present invention will become clear and obvious from the following description of embodiments of the present invention with reference to the accompanying drawings. [Brief explanation of the drawing]

[0068] [Figure 1]

[0002] One embodiment shows a patch or sublingual oral device to be worn in the subchinnic region, which includes uniformly spaced electrodes for EMG sensing and stimulation of an active site. [Figure 2]

[0003] An electrode patch configuration according to one embodiment is shown. [Figure 3]

[0004] An electronic schematic diagram illustrating various components linked to the patch or oral device of Figure 1 according to one embodiment is shown. [Figure 4]

[0005] A block diagram of a controller unit according to one embodiment is shown. [Figure 5]

[0006] An electronic schematic diagram of a controller unit according to one embodiment is shown. [Figure 6]

[0007] This shows a typical genioglossus muscle EMG waveform during sleep, indicating resting respiration. [Figure 7]

[0008] A flowchart outlining one embodiment is shown, which describes how to initialize the process using calibration data to determine the optimal sensing and stimulating electrodes, and how to calculate the optimal stimulation setpoint. [Figure 8]

[0009] The genioglossus muscle M wave and stimulus-induced artifacts following stimulation according to one embodiment are shown. [Figure 9]

[0010] Shows various stimulation pulse waveforms. [Figure 9a]

[0011] A balanced cathode pulse waveform according to one embodiment is shown. [Figure 9b]

[0012] A sinusoidal stimulation waveform according to one embodiment is shown. [Figure 9c]

[0013] The equilibrium cathode delay waveform according to one embodiment is shown. [Figure 9d]

[0014] The stimulated anode waveform according to one embodiment is shown. [Figure 10]

[0015] A system routine is shown that illustrates the entry procedure to closed-loop control of the upper airway and the exit process from closed-loop control according to one embodiment. [Figure 11]

[0016] The respiratory signal and index derived to calculate the stimulation time according to one embodiment are shown. [Figure 12]

[0017] The derived genioglossal muscle EMG amplitude envelope waveform according to one embodiment is shown. [Figure 13]

[0018] An embodiment of a PI controller is shown, which is adapted by respiratory rate to a setpoint calculated during calibration or sleep with a stable upper airway. [Figure 14]

[0019] One embodiment shows a safety mode operation in which continuous stimulation is applied for the purpose of recalibration. [Figure 15]

[0020] Another embodiment of the pain gate control theory is presented. [Figure 16]

[0021] Another embodiment shows the Hoffmann reflex resulting from stimulation of a mixed nerve. [Figure 17]

[0022] Another embodiment shows the M-wave and H-wave reflex responses to increasing stimulus levels. [Figure 18]

[0023] Another embodiment shows an intelligent, personalized closed-loop neuromodulatory system for treating pain. [Figure 19]

[0024] The stimulation waveform is shown along with the generated EMG signal and the average EMG signal according to another embodiment. [Figure 20]

[0025] Another embodiment shows a PI closed-loop controller for maintaining optimal stimulation. [Figure 21]

[0026] A flowchart illustrating pain stimulus control along with activity measurement according to another embodiment is shown. [Figure 22a]

[0027] This is an example of a dental alginate impression material having a second layer formed beneath the tongue region, according to one embodiment. [Figure 22b]

[0028] This is a yellow stone model base for constructing fixtures coated with acrylic resin, according to one embodiment. [Figure 23a]

[0029] An aerial photograph of an intraoral appliance having a ball-end clasp positioned between teeth to provide retention is provided according to one embodiment. [Figure 23b]

[0030] A side view of an oral appliance fixed to a silicone mouthguard for enhanced retention is provided according to one embodiment. [Figure 24]

[0031] An aerial photograph of a plaster model of mandibular teeth showing the electrode placement positions according to one embodiment is provided. [Figure 25a]

[0032] This is a side view of an intraoral device having an electronic circuit according to one embodiment. [Figure 25b]

[0033] This is an aerial photograph of an oral appliance having an electronic circuit according to one embodiment. [Figure 26]

[0034] An example of EMG waveform measurement for detecting sleep apnea events according to one embodiment is provided. [Figure 27]

[0035] This is from Tkach, D., Huang, H. & Kuiken, TA Study of stability of time-domain features for electromyographic pattern recognition. J NeuroEngineering Rehabil 7, 21 (2010), and provides an example to illustrate how a shift in the center frequency can indicate fatigue. [Modes for carrying out the invention]

[0069] Description of the Embodiment

[0036] Specific embodiments of the present invention will now be described with reference to the accompanying drawings. However, the present invention may be carried out in many different forms and should not be construed as being limited to the embodiments described herein, but rather, these embodiments are provided so as to make this disclosure complete and perfect and to fully convey the scope of the invention to those skilled in the art. The terms used in the detailed description of the embodiments shown in the accompanying drawings are not intended to limit the present invention. In the drawings, similar numbers refer to similar elements.

[0070]

[0037] The present invention describes non-invasive neuromodulatory devices and methods that include a smart system for positioning and adaptively repositioning target points and providing feedback on the success of stimulation and energy delivery. According to a preferred embodiment of the present invention, a device for the treatment of OSA is described. GG muscle activity, in particular among the upper airway muscles, is monitored through an operation that generates an electromyographic (EMG) waveform with contributions mostly from GG muscle fibers. Figure 1 shows electrodes 10 implemented with a patch or oral device 100, which is positioned adhesively under the mandible 5 of an individual or positioned inside the lower part of the oral area through a dental device such as a mouthguard that covers the lower teeth, with the electrodes 10 spaced across the entire device and positioned to cover the mandible 5 of an individual. In either embodiment, the surface or intraoral electrodes 10 are located on the skin surface under the GG in the chin-lingual area or superiorly in the intraoral GG under the tongue. The key is that the patch / device allows the electrodes 10 to cover an area of ​​the upper airway muscles, specifically the GG, the left and right hypoglossal nerve endings, and the GG motor point. In a preferred embodiment, the patch includes a matrix of uniformly spaced electrodes 10 used both to sense EMG signals from the upper airway muscles and to stimulate the motor points of the peripheral branches of the hypoglossal nerve and / or the CG muscle itself. In an alternative embodiment, the electrodes 10 may be arranged in an asymmetrical pattern, close together in certain areas and far apart in other parts of the patch 100.

[0071]

[0038] A total of 19 electrodes 100 are shown in Figure 1, with 9 working electrodes on each side and a reference electrode on the base of one patch or instrument 100, although more or fewer may be used depending on the area below the mandible 5. The electrodes 100 are spaced about 5 mm apart, but the distance between electrodes 100 may be any from 1 mm to 10 mm. Typical EMG electrodes, such as platinum, can be used for both measuring EMG signals and stimulating tissue, i.e., muscle motor points and efferent nerve fibers. In this example, 18 unipolar signals are acquired at a sample rate of 0.25 to 2 kHz. Bipolar signals are derived from the unipolar signals of electrode pairs, as will be described in more detail later. Based on this symmetric matrix configuration with equally spaced electrodes, muscle activity is measured on each side. For example, left and right GG muscle activity is measured to determine which unipolar or bipolar electrode pair is best for measuring GG muscle activity during inspiration.

[0072]

[0039] According to one embodiment, the non-invasive electrode path has a custom arrangement as shown in Figure 2, and the electrodes are iridium, platinum, or a combination of both. Typically, manufacturers offer a wide variety of electrode array designs for different applications, combined with a vast array of packaging options and connector types. In the case of oral appliances, the electrode array design is similar but incorporated into plastic packaging. Embodiments related to oral appliances are best presented in Figures 23–25, 27, and 28. Simultaneous EMG signals can be acquired from each monopolar electrode. All electrodes measure muscle activity or EMG signals from a single electrode relative to a common reference electrode. A bipolar signal is a local signal measured between any two electrodes and can be derived simply by measuring the difference between two captured monopolar signals. This is a more flexible technique and can derive more combinations without additional hardware that shows and provides any possible combination of electrodes. Electrode connectors connect each electrode to a data acquisition amplifier channel, in which the EMG signal can be amplified and digitized.

[0073]

[0040] Figure 3 shows an electrical schematic of additional components connected to patch 100 according to an embodiment of Figure 1. The additional components may be located on patch 100 (or the device) itself or on a separate electrically connected device. Not all of the additional components are required in the basic design of the present invention, and the main component required in all designs is the reference electrode 101 used to identify signals at the working electrode 100. The additional components include microphones 20 paired on each side of the throat to record sounds from the airway, including but not limited to respiration, snoring, speech, choking, chewing, wheezing, and coughing from the inspiratory and expiratory phases of the respiratory cycle. A two-axis or three-axis accelerometer 30 is located at the center of the subchineal triangle. Its static gravity component is used to identify posture, including upright, supine, prone, and lateral. The static component can be derived from the raw accelerometer signal by low-pass filtering. Posture identification is used to switch stimulation settings or turn off treatment if the patient is not experiencing positional OSA (a small number of apnea events when not lying supine). The accelerometer 30 can also be used to identify activity using the dynamic gravity component, which is derived by passing it through high-pass filtering and taking the modulo or square of the residual axis. Activities such as snoring, speech, breathing, coughing, and choking can be identified by the dynamic component. Activities resulting from wakefulness from sleep associated with OSA can also be identified. Additional optional components include a temperature sensor 50, which can measure respiration, and a cable connects the patch 100 to a separate sensor 35, which will be located below the nasal and oral area. Furthermore, an SpO2 sensor 60 may be included, which measures oxygen desaturation from either the carotid or nasal area. Oxygen desaturation and esophageal pressure provide a respiratory waveform. The system may use all sensors or just one to derive a respiratory waveform for the purpose of calculating the optimal stimulation duration and apnea / depressive events.

[0074]

[0041] The device is powered by a coin cell battery 75, but power may also be provided through a hardwired connection to a device such as a laptop, tablet, or primary cell battery. A Bluetooth SoC 65 can communicate with a separate device to send real-time data to a laptop or tablet or update settings. A separate control device can interface directly with an accelerometer 30 that acquires digital data from the x, y, and z axes. Its analog-to-digital (ADC) converter digitizes audio from microphones 20 located on both sides of the trachea, and an integrated algorithm extracts the respiratory waveform through the voice envelope. A dedicated controller 150 is used to record the EMG waveform and apply stimulation. The dedicated controller 150 can be an ASIC, FPGA, SoC, or other dedicated IC. According to one embodiment of the present invention, an 18-channel ASIC is used to control the electrodes 100. A typical block diagram and function of an ASIC embodiment are shown in Figure 4, which includes the main processing block. The main processing block includes a processor, memory, interface, power, and connectivity. An example of a dedicated controller 150 is shown in Figure 5, which is a fully integrated electrophysiological interface chip and has 16 channels of low-noise amplifiers and constant-current stimulators controlled by an industry-standard serial peripheral interface. The array of 16 stimulator / amplifier blocks includes two amplifiers for sensing electrode voltages, having an AC-coupled high-gain amplifier for observing small electrophysiological potentials and a DC-coupled constant-gain amplifier for monitoring electrode potentials in response to stimuli. The high-gain amplifier is called a common shared pin (ref_elec). In many applications, the reference electrode is also used as a stimulus counter (return) electrode and is grounded. Each channel has an independent stimulator module capable of generating biphasic constant-current pulses with amplitudes ranging from 10 nanoamperes to 2.55 milliamperes. These stimulators can maintain a constant-current output over a wide range of electrode voltages and have compliance limits near the stimulus voltage sources VSTIM+ and VSTIM-.

[0075]

[0042] A calibration phase is performed to determine the initial settings of the device. The calibration operation is performed to activate the GG muscle, identify its position relative to the electrodes, and determine the best sensing electrode for measuring GG activity. Since the GG signal is largely present during inspiration when the muscle is stimulated by the internal branch of the hypoglossal nerve that causes the muscle to protrude, calibration is performed during the inspiratory phase. The operation is performed by consciously pressing the tongue against the lower teeth and maintaining this position throughout the respiratory phase to enhance the signal. This operation should be performed at least three times over three respiratory cycles. Waveforms from 18 pairs of unipolar electrodes 100 are memorized and processed to determine the best bipolar GG electrodes for both the left and right GG sides. The best electrodes can be more easily identified by applying greater force to the lower teeth. Once GG muscle activity is identified and can be monitored using the identified electrodes, it is possible to infer the success of muscle activation through stimulation. Closed-loop stimulation can then be performed using the GG muscle EMG signal as feedback.

[0076]

[0043] During the calibration phase, the optimal stimulation setpoint can also be determined. The setpoint is determined based on calibration data by analyzing the operation that requires approximately 25-50% of the force when the tongue is thrust out and pressed against the lower teeth. The peak of the GGAV envelope waveform generated over three breaths during the inspiratory phase is then calculated. This will vary for everyone and can be manually adjusted by the user or by a healthcare professional during the visit if necessary. Thus, the user will always be titrated at this rate, while prompting the user for feedback on any sensations that can be felt from the skin. Systems employing subchineal transcutaneous stimulation are susceptible to arousal resulting from sensations triggered by stimulating low-threshold (high-sensitivity) encapsulated mechanoreceptors, particularly when turning on and off during the inspiratory phase of breathing. There are four main types of encapsulated mechanoreceptors: Meissner corpuscles, Pacinian corpuscles, Merkel discs, and Ruffini corpuscles, which generate action potential responses that relay sensory information to the central nervous system in response to contact, pressure, vibration, and tension. All low-threshold mechanoreceptors are stimulated by relatively large myelinated Aβ fiber axons. Meissner corpuscles respond to vibrations from contact at approximately 50 Hz. Pacinian corpuscles respond to skin vibrations around 200–300 Hz. Frequencies vary among subjects, and therefore, it is optimal to adjust and personalize the settings for each subject. Since the therapy will be administered only during sleep (discussed in a later section), at this point, any tongue movement or pressure sensation is tolerable. Any skin sensation that may cause discomfort or awakening from sleep may occur at lower frequencies. Therefore, the stimulation frequency is increased until the sensation is sufficiently reduced and tolerable, or until the maximum stimulation frequency threshold is reached.

[0077]

[0044] The optimal stimulating electrode may change due to movement, and therefore feedback is needed to determine the optimal electrode and sufficient stimulation in real time. According to a further embodiment of the present invention, various operations can be performed in each position (supine, left, right, prone) to help identify the optimal electrode for tracking GG activity in each position. If the patient has position-dependent OSA, and thereby the apnea-hypopnea index (AHI) is at least 50% lower when not in the supine position, and the AHI is considered normal in other positions, the stimulation will not be activated in unnecessary positions. AHI is a measure of the hourly mean number of apnea and hypopnea events determined by sleep studies. Apnea is a complete obstruction of the upper airway, while hypopnea is a partial obstruction of the upper airway. Closed-loop control is applied, and a force setpoint of about 25-50% moves the GG slightly forward without disturbing the subject and without risking awakening or tongue abrasion. The operation is repeated, and the setpoint amplitude is captured from the optimal sensing electrode and identified offline (not in real time).

[0078]

[0045] The static breathing signal is shown in the first trace of Figure 6, which is specific to the actual tidal volume waveform derived by integrating the effort sensor waveform captured from the torso or thoracic area or the flow signal from the spirometer. The subsequent waveform GGEMG represents, from one side, the EMG signal captured from a designated sensing electrode or set of electrodes, followed by a moving average of the signal capturing the amplitude envelope GGAV. The signal indicates GG activity during the inspiratory phase of the respiratory cycle, with activity decreasing during the expiratory phase.

[0079]

[0046] When the device is activated, the initialization process begins as shown in Figure 7. The waveform data stored from the calibration phase is loaded in block 700 and identifies the following for each posture and GG side: the optimal sensing electrode in block 701, the optimal stimulating electrode in block 702, the setpoint in block 703, and the pulse waveform stimulation frequency in blocks 704, 705, and 706. The bipolar electrode pair having the maximum energy in each posture, which can be identified by measuring the amplitude of the GGAV waveform, is determined to be optimal for measuring GG activity and is stored in block 707. Alternatively, if there is sufficient energy in other pairs that can be combined and optionally weighted, combinations of electrode pairs can be utilized. Other operations can be performed to identify other muscle groups, for example, swallowing and vocalization operations can be performed to identify GH, moving the tongue backward can identify the styloglossus muscle, or lowering and moving the tongue backward can identify the hyoglossus muscle.

[0080]

[0047] Alternatively, each EMG signal can be identified by applying a pattern recognition routine, and each individual muscle is classified from the signals acquired in each operation during the calibration phase. For example, machine learning using a support vector machine (SVM) can be trained to distinguish each muscle. Alternatively, other routines such as neural networks and dynamic programming techniques may be applied. Furthermore, different muscles have different firing rates and are likely to have different fibers (Type I and Type II). Many OSA sufferers are thought to have Type II muscle fibers that exhibit stronger and faster contractions but fatigue more quickly. As phase tends to have higher frequencies, tonic and phase muscles have different profiles.

[0081]

[0048] To identify GG, other techniques may be employed offline from stored data. For example, the data can be interpolated and each bipolar pair can be cross-correlated to track the direction of conduction along the muscle. Bipolar pairs with a strong correlation (r>0.7) indicate that they are emitted from the same muscle fiber and can be tracked as the EMG waveform propagates. For the GG muscle, the EMG should propagate in anterior to posterior direction. A conduction velocity of approximately 2-3 m / s is identified in GG, and therefore, for electrodes spaced 5 mm-10 mm apart, the M wave should appear at approximately (2*0.005 s-2*0.01 s) 10 ms-20 ms / cos(theta), where theta is the angle the GG traces from the medial to the jaw. The location of GG pairs can be identified by plotting the line for each GG pair based on the maximum correlation (r>0.7) of the bipolar measurements, and CG activity can be measured by measuring the EMG signals from all electrodes closest to the target muscle.

[0082]

[0049] The identified sensing electrodes help determine the most effective stimulating electrode. These electrodes are monitored as one to more pulses are delivered to each set of bipolar pairs, measuring the amplitude of the M-waves generated at the sensing electrodes. This process is referred to herein as pinging the electrodes. An example is shown in Figure 8, where a single stimulating pulse triggers an M-wave recorded at the sensing electrode. A measurement window of 5–20 ms is applied to avoid any stimulating artifacts in the recorded EMG waveform. Once the bipolar electrode pairs receive the ping, the left and right sides of the inner subchineal region that produce the maximum amplitude are selected as the optimal stimulating electrodes.

[0083]

[0050] Several stimulation waveform patterns can be applied to the HGN or GG motion point. Four waveforms are presented in Figure 9, all biphasic, charge-equalizing, and either cathode-type (Figures 9a and 9c) or anode-type (Figures 9b and 9d), with Figure 9b differing in that it is sinusoidal, suggesting that sinusoidal waves can avoid activating sensor receptors that may trigger awakening from sleep. Furthermore, based on current perception threshold (CPT) theory, a 2 kHz sinusoidal wave may activate Aβ fibers for pain management. Equilibrium waveforms are applied to reduce side effects, tissue or electrode damage by approximating zero residual voltage and zero net Faraday charge transfer. Figure 9c shows an interphase delay that is particularly useful when the delivered stimulation energy has just exceeded the action potential threshold, thereby increasing the likelihood of generating the action potential. The delay itself minimizes the threshold while maintaining the anti-corrosion effect of the charge recovery phase between cycles. Furthermore, the waveform in Figure 9d shows a biphasic anode equilibrium waveform with a subthreshold depolarization prepulse, which is shown to enhance the stimulus response. The prepulse is delivered immediately before the stimulus waveform and has the effect of altering the characteristics of the action potential and motion point threshold. The interphase delay further enhances the stimulus response before the charge recovery phase.

[0084] Main process

[0051] The flowchart shown in Figure 10 outlines the main system process for maintaining upper airway patency. The calibration and initialization phases are described in a preceding section on determining the optimal electrode pair for measuring muscle activity such as GG. A processed EMG signal with an amplitude envelope showing the maximum response is determined as the optimal bipolar pair. A dedicated waveform is used to apply bipolar stimulation with a biphasic pulse sequence having a delay after a preceding pulse. The biphasic pulse provides a faster tissue recovery time from stimulation using charge equilibrium, while the interphase delay and prepulse provide a larger stimulation response. The main system process is initiated in signal monitoring phase 800, where signals are acquired and processed from all available sensors to provide a waveform for real-time analysis. Closed-loop stimulation therapy 810 is not initiated until certain conditions are met. Firstly, to ensure patient comfort, it is ideal not to stimulate while the patient is awake. Therefore, the subject cannot be upright, which directly indicates an awake state (801). Secondly, the subject must be in a sleep state—ideally a stage beyond alpha (802). Sleep state is determined through analysis of respiratory waveforms that can be derived from GGAV waveforms, audio processing, nasal / oral breathing temperature, pulse oximetry plethysmogram, esophageal pressure, or any respiratory sensor. Sleep state is determined through analysis of fluctuations in respiratory signals.

[0085]

[0052] Respiratory waveforms while awake can exhibit considerable fluctuations. For example, during voluntary breathing, the inspiratory cycle can be short in duration and high in amplitude to achieve the required tidal volume, or low in amplitude and long in duration to achieve the required oxygen consumption the body needs. Short inspiratory and long expiratory cycles are often observed during speech, which can fluctuate considerably based on, for example, conversation, with deep breaths being taken in preparation for delivering sentence utterances. Conversely, during sleep, a periodic steady pattern resembling a periodic sine wave is observed, as shown in Figure 11. Therefore, the onset of sleep can be identified by measuring the fluctuations in the amplitude of the complete respiratory cycle, inspiratory and expiratory durations, and tidal volume measurements. A significant decrease in amplitude and frequency fluctuations can determine whether the person is in a sleep state or a wakeful state, and potentially determine that they are in a sleep stage. To determine sleep, amplitude and frequency fluctuations are evaluated, and the subject should not be in an upright position, as determined by the accelerometer. Sleep is determined by a decrease in per-breath fluctuations of respiratory rate and continuous ventilation. The amplitudes of both the EMG signal and the audio signal can be applied as unitless measures of tidal volume, and the respiratory rate from each signal can also be applied. The inspiratory phase of the respiratory cycle can be identified from the microphone signal and distinguished from the expiratory phase using the GG EMG signal.

[0086]

[0053] As described above, respiratory waveforms such as pressure and tidal volume are periodically steady during sleep, and the fluctuations between tidal volume and frequency in each cycle are relatively small. The respiratory waveform (RESP) shown in Figure 11 shows this typical periodically steady pattern. Here, the k-th cycle is A k It is represented as and the cycle time is T k It is expressed as follows, and the inspiratory time and expiratory time are ti, respectively. k and te k The sleep state can be determined by monitoring the fluctuations in inter-cycle amplitude (Equation 1), frequency (Equation 2), or a combination of both (Equation 3) over N cycles.

number

[0087]

[0054] If the subject is determined to be asleep, the next step shown in Figure 10 is to determine whether the subject's upper airway (UAW) is unstable (804). As described above, OSA resulting from upper airway instability occurs when the force required to maintain the GG in place is reduced due to either decreased nerve drive or muscle fatigue, which leads to a decrease in GG EMG. To determine this condition, the energy of the GGAV waveform during the respiratory phase is compared. The area under the curve, mean, median, or peak of the inspiratory phase should be at least twice that of the expiratory phase when operating in a stable state. Furthermore, the change in the mean GGAV waveform per breath k is also considered.

number

[0088]

[0055] To avoid muscle fatigue during closed-loop stimulation mode, stimulation is applied only during the inspiratory phase of the respiratory cycle. Continuous stimulation throughout the night is not possible because the aforementioned muscles will fatigue. Therefore, it is very important to allow the muscles to rest during the respiratory cycle by stimulating only during the inspiratory phase when airway obstruction may occur. It is advantageous to stimulate before the start of inspiration, as the goal is not recovery from airway block, but to maintain upper airway patency and avoid obstruction. Furthermore, the processed airway waveform will be delayed due to several factors such as the sensor used, the mechanical effects of breathing, and the delay of the digital filter group. To predict the start of inspiration, the start of the next inspiratory cycle is estimated using knowledge of the previous cycle time and the previously calculated average respiratory rate or cycle time. It is very important that stimulation occurs before the end of expiration. To ensure success, an error margin is calculated or fixed. This error margin can be calculated based on the confidence that the inspiratory time estimate is good. The reliability can be based on, but is not limited to, the variance of intercycle time; low variance provides high reliability of the calculation, while higher variance means lower reliability, and consequently, a larger error margin and offset subtracted from the inspiratory time prediction, which in turn means an earlier stimulus onset time. The predicted inspiratory onset is shown in Figure 11 as ti k+1 As shown by,

number

[0089]

[0056] The amount of stimulation energy to be delivered is determined through closed-loop feedback 805 and can be determined for each respiratory cycle. The stimulation waveform is described above, and the pulse amplitude or pulse width must be large enough to protrude the GG muscle. The average of the GGAV signal provides the best process variables. The ideal threshold for the GGAV signal (bilateral) can be set early in the night (805) if the subject is asleep and no apnea events are occurring (804). If an OSA event occurs (810), stimulation is applied continuously for a short period to return the GGAV signal to a normal level, and then closed-loop stimulation is returned only during the inspiratory cycle. The amplitude or pulse width of the stimulation is increased or decreased for each cycle until the ideal GGAV threshold is reached.

[0090]

[0057] In some embodiments, the GG EMG signal is sampled within a window to ensure that stimulus artifacts, such as those shown in Figure 8, do not affect the derivation of an accurate GGAV signal. The window is a period of time during which EMG activity is recorded and processed to provide one sample of the EMG amplitude envelope shown in Figure 12, and thus has a sample rate that is much lower than the EMG signal and equal to the stimulus frequency. Two typical respiratory sleep cycles are shown in Figure 12. As mentioned above, during sleep, the abdominal muscles are dominant, and the respiratory tidal volume signal appears periodically and steadily with a nearly sinusoidal nature. The upper trace (RESP) is a typical tidal volume signal during sleep, with a frequency of 0.25 Hz (t=4 sec) and a normal respiratory rate of 15 breaths / min. The subsequent trace shows a normal GG EMG signal, which is likely to have contributions from other upper airway muscles such as the GH muscle. The mean EMG signal is then shown, and the traced GGAV can be moved averaged, median filtered, peak filtered, or low-pass filtered to capture the signal envelope. In this case, the signal is time-averaged. Clearly, this trace can be used to identify the inspiratory and expiratory times, showing high levels of EMG activity during inspiration and a significant decrease during expiration. The third trace shows the pulse waveform used for stimulation. The waveform has been described above, and the EMG measurement precedes the next stimulation cycle, following the resolution of stimulation artifacts. In this case, the stimulation frequency Fp = 15 Hz or the cycle length is 67 msec, but can range from at least 2 Hz to over 500 Hz. Thus, GG EMG signal segments can be acquired every 67 msec. The raw GG EMG signal segment samples are processed by averaging or by taking the median or peak M wave. The average is expressed in Equation 4 to derive the signal envelope:

number

[0091] In some embodiments, stimulus-induced artifacts can be filtered from the EMG signal by characterizing the medium between the sensing electrode and the stimulating electrode. This can be achieved using an inverse filter such as a Wiener filter or a real-time adaptive filter. In a preferred embodiment, the medium transfer function is characterized by output / input, where the output is the recorded EMG signal and the input is the stimulus waveform. Artifacts are recorded at the sensing electrode and filtered out by stimulating with a sufficiently low amplitude, not sufficient to elicit a motor response. It will be understood by those skilled in the art that the medium transfer function can be defined in the frequency domain and the time domain.

[0092]

[0058] The closed-loop control process is shown in Figure 13 in the form of a PI controller, where the setpoint is initially set during calibration, updated during sleep periods when there is no upper airway instability, and never lowered, maintaining only the calculated maximum value, as presented in Figure 12. Process Variables

number

number

[0093]

[0059] During closed-loop controlled stimulation, the subject's posture 811, sleep state 814, and controller output 815 are all constantly monitored. If the posture changes to a neutral position 812 and the subject does not have position-dependent OSA but has low AHI on one side, the system terminates closed-loop control 812, stops stimulation, returns to monitoring signal 800, and waits for a change. Alternatively, in the case of position-dependent OSA, a new setting for that posture is loaded and closed-loop control continues. Similarly, if the subject wakes up (814), the closed loop stops and monitoring continues (800). If the controller output exceeds a threshold that could cause tissue damage over time (e.g., >30mA) (815), safety mode 820 is activated. During the closed-loop stimulation phase, monitoring for any interference that could affect closed-loop control is performed. This specification defines interference as apnea or hypopnea events characterized by a cessation of breathing lasting longer than 10 seconds, measured as a significant decrease in tidal volume (a 90% decrease in apnea and a 60% decrease in hypopnea from the previous breath) as measured by any of the above respiratory sensors. Awakening from sleep can be identified by sudden movements recorded as high activity on an accelerometer. Snoring can be detected by audio in the frequency range of 0.2–2 kHz. Choking and other audio-related events are also considered interference. In all cases, the control switches to the safety mode outlined in Figure 14.

[0094]

[0060] A flowchart illustrating the safe mode process is shown in Figure 14. The first step is to determine whether an effective set of stimulating electrodes can be identified (901). This is achieved by pinging each set of electrodes in a bipolar pair with one or more pulses (900) and measuring the resulting M waves with a sensing electrode. If a new set of electrodes cannot be identified, the stimulation process on the side being treated with stimulation is stopped (903). At this point, hopefully, the opposite side is still effectively providing stimulation therapy. If a new set of stimulating electrodes can be identified, continuous stimulation 904 is initiated through both the inspiratory and expiratory phases without interruption to allow the muscles to rest. This is safe for a short period, in this case about 5 minutes (906), and eliminates any disruptive apnea events if the trigger disruption 907 is due to unsuccessful treatment of an awakening, snoring, or apnea / respiratory depression event. During this period of continuous stimulation 904, the controller output is monitored (905) to ensure that the titration level remains within safe limits, and the process is terminated if the controller is unable to achieve the setpoint within reasonable limits due to a system error (903).

[0095]

[0061] The process of identifying the optimal sensing and stimulating electrodes, along with personalized closed-loop controlled titration, can also be applied to other neuromodulatory therapies such as percutaneous tibial nerve stimulation (PTNS) and the treatment of overactive bladder (OAB) using management of neuropathic and nociceptive pain. Neuromodulation is largely a common therapy for treating pain using implanted pulse generators (IPGs). Spinal cord stimulation (SCS) is the most common neuromodulatory marker for which IPGs are employed to treat a variety of pain conditions. One mechanism of action that controls pain based on stimulation is explained by the gate control hypothesis, which blocks pain signals with paresthesia, while the other involves high-frequency stimulation above 10 kHz to excite inhibitory neurons. In the treatment of OSA, we want to avoid stimulating specific sensory mechanoreceptors, but in the treatment of pain, we want to stimulate such fibers.

[0096]

[0062] Primary nociceptive afferent fibers are responsible for transmitting fast, intense pain nerve impulses to the central nervous system (CNS) via small myelinated Aδ fibers with moderate conduction velocity, and slow, chronic, throbbing pain nerve impulses via small unmyelinated C fibers with slow conduction velocity. Spinal cord transduction cells relay this information to the brain, and the dorsal horn acts as a gating mechanism. Based on the gate control hypothesis, increasing the excitation of transduction cells increases the throughput of pain stimuli to the brain and intensifies pain perception, while decreasing the activity of transduction cells through inhibition has the effect of reducing pain perception. Inhibition of transduction cells can be achieved through competitive stimulation of medium-sized nociceptive Aβ fibers with moderate conduction velocity, which are responsible for touch and pressure sensation along with other motor functions. Here, Aβ fibers indirectly inhibit the transmission of pain signals from C fibers by closing the gate of transduction cells responsible for relaying pain signals to the brain.

[0097]

[0063] The gate control hypothesis is based on the presence of inhibitory interneuronal connections with Aβ, Aδ, and C fibers that synapse on the same transmitting cell, which can reduce the likelihood that the transmitting cell will fire and transmit pain stimuli, as shown in Figure 15. When inhibitory interneurons fire through excitatory connections with interneurons, the likelihood of transmitting cell firing decreases, while conversely, C fiber firing inhibits the interneuron, increasing the likelihood that the transmitting cell will fire pain signals and send them to the brain. Thus, in this example, the transmitting cell is excited or inhibited depending on the firing rates of Aβ and C fibers. As shown in Figure 15, when Aβ fibers are more active than C fibers, the inhibitory interneuronal input is net positive and produces an inhibitory effect on the transmitting cell. Conversely, when the interneuronal input is net negative.

[0098]

[0064] Larger nerve fibers are more likely to be recruited first, and therefore A fibers, specifically Aα and Aβ fibers, should be recruited before the smaller diameter Aδ and C fibers. As shown in Figure 15, higher Aβ activity generates a net positive input to inhibitory interneurons, thereby resulting in presynaptic inhibition and reduced excitation of the transmission cells. This reduces the therapeutic activity of nociceptive and non-nociceptive neurons. When recruiting afferent Aβ fibers for pain relief, care must be taken to ensure that efferent fiber activation does not occur, which could produce a certain level of continuous discomfort and irritation by generating muscle contractions or spasms. To ensure that any muscle activation is minimal, the disclosed method tracks the amplitude of the M reflex and M wave from the EMG waveform in any muscle innervated by the nerve being treated. The H reflex, or Hoffmann reflex, is a muscular response to stimulation of afferent Ia muscle fibers, which transmit signals from muscle spindles to the spinal cord, producing an efferent response observed in muscles. The tibial and median nerves are commonly used for H reflex analysis. The H reflex has a constant delay regardless of the stimulus amplitude, as it delays the M wave and activates the same neuronal pool, as shown in Figure 16. Therefore, the H reflex can be easily identified based on this delay from the stimulus pulse. Because afferent Ia fibers are larger than motor fibers, the response is more easily elicited, and the M wave should not be induced at levels slightly above the afferent Ia threshold. As mentioned above, the H reflex increases in intensity as the stimulus increases, but the M wave is not induced until just before the H reflex reaches its maximum amplitude, as shown in the recruitment curve in Figure 17.

[0099]

[0065] A system that provides this function is shown in Figure 18 and is based on the same function as described in Figures 1 and 3 to the same extent. The system consists of two patches, one of which has a dedicated sensing electrode and the other has a dedicated stimulating electrode. An accelerometer is present in the sensing patch to measure motion, and the patches may be hardwired or wirelessly connected to each other.

[0100]

[0066] To ensure that muscle spasms do not occur during therapy, it is desirable to titrate below the motor activation threshold and provide motionless paresthesia to mask pain signals. Pain management physicians typically adjust a constant stimulus to a specific level for that purpose. Embodiments disclosed provide closed-loop control feedback using the H reflex, if available. The maximum amplitude of the H reflex is determined by pulse-ping the previously determined optimal stimulating electrode. The amplitude of the excitation pulse is gradually increased until it reaches the maximum H reflex amplitude measured at the optimally estimated sensing electrode. The process variable in this closed-loop system is EMGAV, which is the mean of the EMG signal or its envelope, and the setpoint is fixed just below the maximum H reflex amplitude, as shown in Figure 19. A PI control system diagrams this process in Figure 20, but a P, I, PI, PID, MPC, or on / off controller may perform this function. In Figure 20, the peripheral nervous system (PNS) is stimulated using feedback from the innervated muscles to ensure that the stimulus is maintained below the motor activation threshold. In alternative embodiments, if the H-reflex signal is unavailable, as may apply to specific nerve and muscle groups, the M-wave is monitored, in which case the setpoint is fixed at minimum amplitude to ensure no activation. In this case, the setpoint is similarly determined by pinging the electrodes while increasing the amplitude until the M-wave appears. Accelerometers present on the sensing patch shown in Figure 18, placed on the muscle, can be used to a) ensure no activity when determining the optimal setpoint 1002, b) then determine what amplitude triggers movement, and c) ensure no activity when in closed-loop control mode 1003. This process is shown in the flowchart of Figure 21, where the maximum stimulation level is determined to provide stimulation below the motor activation threshold. During the muscle activation period, the system returns to open-loop mode with a safe setting.

[0101]

[0067] In an alternative embodiment, the system described can be used for the treatment of overactive bladder (OAB) by stimulating the posterior tibial nerve. In one example of this therapy using the patch described in Figure 18, it is performed to achieve detrusor inhibition through stimulation of afferent somatic sacral nerve fibers accessible through the tibial nerve. The exact mechanism of action by which tibial nerve stimulation treats bladder dysfunction is not clear, but it is thought that the sacral plexus can regulate bladder function through the modulation of afferent and efferent fibers. Currently, percutaneous tibial nerve stimulation (PTNS) involves inserting a catheter into the medical ankle, and the patient's response to the stimulation is confirmed by involuntary flexion or extension of the toes. Toe flexion results from stimulation of the S3 nerve root, which is responsible for bladder nerve innervation. In the treatment of pain, the stimulation patch described above is placed on the tibial nerve region above the ankle. The sensing patch can be placed on several muscles responsible for toe flexion, including: abductor hallucis, flexor hallucis brevis, flexor digitorum brevis, quadratus plantae, abductor digiti minimi of the foot, flexor digitorum longus, and flexor hallucis longus of the shin. To treat OAB at home, stimulation therapy is performed at home several times a week for 30 minutes at a time, similar to PTNS in a hospital, using the systems and methods described herein.

[0102]

[0068] As previously stated, in certain embodiments of the present invention, the system for sensing and stimulating the GG muscle can be integrated into an intraoral instrument. This may include an anterior mandibular fixation device, a retainer, a mouthguard, etc. Figures 22a and 22b provide an example embodiment in which a plaster model of the mandibular teeth and the sublingual space of the floor of the mouth can be captured using alginate to form a dental impression material, which can then be set sublingually by a second layer. The model may be fabricated by pouring yellow stone to construct the base formation of the instrument and applying acrylic resin to the model to produce the acrylic piece shown in Figure 22b. A ball-end clasp can be positioned between the teeth to provide the necessary retention shown in the instrument shown in Figure 23a. Alternatively, the acrylic instrument can be fixed to a silicone mouthguard in Figure 23b to provide further stability.

[0103]

[0069] In some embodiments, holes can be drilled in an acrylic piece to place electrodes in order to fabricate the instrument, or preferably, holes can be drilled in the model with the electrodes in place before applying the acrylic to permanently fix the contacts. In this embodiment, the contacts are cylindrical electrodes 5 mm long with a diameter of approximately 1 / 16 inch (approximately 1.6 mm). A plaster model of the mandibular teeth and sublingual space is shown in Figure 24, with ball-end clasps embedded in the acrylic and electrode positions identified for drilling. In the example in Figure 24, the surface electrodes are positioned so as to be in contact with the mucous membrane of the floor of the mouth directly above the genioglossus muscle, with approximately 0.5–1 mm of the electrode exposed. Two linear arrays of electrodes are positioned 3 mm from either side of the midline so as to be recorded above the center of each genioglossus muscle. The anterior electrode is positioned 3 mm posterior to the lingual gingival margin of the mandibular incisor, and further electrodes are placed posteriorly with an inter-electrode distance of 3 mm. An additional array is positioned 3 mm lateral to the first array on each side, and an additional electrode is positioned 3 mm lateral to that array. This electrode is positioned approximately 50% of the way from the mandible and hyoid bone to activate the internal branch of the hypoglossal nerve, which innervates the horizontal compartment of the genioglossus muscle, known to protrude the tongue.

[0104]

[0070] The hypoglossal nerve divides into an internal branch and a lateral branch, with the distal internal branch being responsible for innervating the genioglossus muscle. Previous studies investigating the anatomical structure and function of the genioglossus muscle have revealed that the oblique compartment with vertical fibers is responsible for tongue retraction, while the horizontal compartment with longitudinal muscle fibers is particularly responsible for tongue protrusion. Therefore, the horizontal compartment is most relevant to maintaining upper airway patency. By reviewing anatomical and morphometric studies, the optimal distance for superselective stimulation of the anterior puller branch of the hypoglossal nerve, which innervates the genioglossus horizontal compartment, was determined. Based on a study by Delaey et al., "Specific branches of hypoglossal nerve to genioglossus muscle as a potential target of selective neurostimulation in obstructive sleep apnea: anatomical and morphometric study," Surg Radiol Anat. 2017 May; 39(5):507-515, it was identified that the nerve branches are located approximately midway between the hyoid bone and the mandibular symphysis at a depth of 3–6 mm, and near the anterior edge of the hyoglossus muscle. Therefore, to achieve pharyngeal airway dilation, it is optimal to target the hypoglossal nerve branches feeding into the horizontal compartment, the longitudinal muscle fibers of the horizontal compartment, or both. Highly potential target locations are identified in Figure 24. The first area shown is the beginning of the hypoglossal nerve branches, located approximately 50% of the distance between the hyoid bone level and the mandibular level. The second area approximates the area between the first and second electrode arrays. The first array, located 3 mm from the midline, is generally situated above the longitudinal muscle fibers and is well-positioned to monitor muscle responses. Based on the findings of Mu L, Sanders I (2010) Human tongue neuroanatomy: nerve supply and motor endplates. Clin Anat 23:777-791, it is estimated that 2-3 secondary nerve branches may extend in equilibrium with the muscle fibers and be present between the electrode arrays.The electrode array can be used for both stimulation and sensing, and can be configured to cover a larger anatomical area to stimulate the genioglossus muscle in a localized and selective manner.

[0105]

[0071] Figures 25a and 25b provide a side view and an aerial photograph, respectively, of an intraoral device and electronic circuit according to an embodiment of the present invention. The intraoral device preferably includes the dental device and electrodes described in the above embodiments of the present disclosure, e.g., the arrangement of the dental device and electrodes described in Figures 1 and 2, and optionally additional components referenced in Figure 3. It is clear that the components are small enough to be incorporated into an intraoral device. Each component may have its own substrate and associated electronic circuitry and may be layered, as shown, with the smallest components positioned higher and towards the front of the device. A battery may be the largest component and may be located below the other components. In embodiments, an ASIC interfaces directly with the electrodes and data is passed to a SoC. In certain embodiments, a biocompatible epoxy, such as AA-BOND FDA15 medical-grade epoxy resin, specifically designed for bonding and coating applications in accordance with FDA regulations, may be applied to encase the electronic circuitry. Advantageously, by positioning the largest component on the underside and stacking it forward during forward extension, the tongue slides along the device and does not come into contact with the lower teeth. Tongue abrasions have been reported as a problem associated with the above Inspire device, which requires the tongue to protrude and press against the teeth. Furthermore, positioning the tongue in an upward position and pressing it against the soft palette improves airway patency and subsequently promotes better sleep. In some embodiments, the intraoral device may include a mandibular anterior fixation device or another intraoral device suitable for placement in the user's mouth. In some embodiments, the intraoral device may include a mouthguard. In some embodiments, the intraoral device may include a non-custom device with a boil and bite mouthguard formed from moldable thermoplastic plastic, thereby greatly simplifying the manufacture of exemplary intraoral devices as it does not need to be pre-dimensionalized at the manufacturing stage to complement a given user's mandibular teeth. Instead, the user can apply hot water to the boil and bite mouthguard, bite the material to shape it with the mandibular teeth, and then add electronic circuits, which can be provided together as a kit.

[0106]

[0072] In preferred embodiments, the intraoral instrument is manufactured using a 3D printing method. For example, in some embodiments, an impression material model of the user's mandibular teeth and sublingual space may be captured and used to generate a 3D model to be used for printing the intraoral instrument. Holes may be designed in the model before 3D printing the thermoplastic resin piece for positioning electrodes. Uniformly spaced electrode arrays may be positioned on both sides of the midline so as to be able to record from above the center of each side of the genioglossus muscle. The electrodes preferably contact the mucosa of the floor of the mouth directly above the anterior and posterior genioglossus muscle compartments in each hemisphere. The anterior electrode may be positioned posterior to the lingual gingival margin of the mandibular incisor, for example 3 mm posterior, and further electrodes may be placed posteriorly with an inter-electrode distance of, for example 4 mm. Additional arrays may be positioned 4 mm outward from each other on each side. The matrix of uniformly spaced electrodes is intended to cover the entire instrument so that nerve fibers cross under the bipolar stimulating electrode pairs. The electrodes are incorporated into an intraoral device, with the upper end connected to the instrument amplifier and pulse generator input, and the lower end in direct contact with the tissue. The stimulating electrodes can be configured as anodes, cathodes, or nonpolar, essentially disconnected. By configuring the electrodes in this way, it is possible to manipulate the electric field with current to ensure successful and repeatable muscle stimulation. This may be necessary when nerve branches do not align well with any bipolar pair electrodes.

[0107]

[0073] Figure 26 provides an example embodiment of sleep apnea event detection according to the present invention. Patient airway obstruction can result from a combination of factors. First, the patient must have an airway that can collapse, and in the supine position, the gravitational effect is strongest, and combined with reduced neuromuscular drive, it will fail to stimulate the genioglossus muscle sufficiently to clear or prevent obstruction. The present invention implements a means of predicting potential apnea / respiratory depression based on the fluorescence of the sublingual muscle EMG waveform. An example shown in Figure 22 shows bipolar EMG waveforms recorded using Teflon steel wire electrodes inserted on each side of the genioglossus muscle 4 mm anterior to the frenulum. It will be understood that the specific materials and other parameters of the electrodes are provided as examples only and are not intended to limit. The filtered EMG waveform is highly phased with the airway flow signal and the integrated EMG signal (GG tau ) shows a decrease in muscle activity 2 minutes before obstruction and cessation of airflow. Therefore, apnea events are ensured to be identified by a decrease in oxygen saturation that occurs at approximately 2.75 minutes. EMG waveforms from the upper airway muscles can provide respiratory information and be used to identify sleep and wakefulness periods. The anterior and posterior genioglossal muscle compartments, which retract and protrude the tongue, respectively, are more active, especially in the supine position, and their regulation increases during the inspiratory phase. To the same extent, the genioglossal muscle is a traction muscle and is also known to co-activate during respiration.

[0108]

[0074] In certain embodiments, it may be advantageous to measure stimulus-induced genioglossal muscle fatigue, which can be identified by comparing the evoked sublingual muscle EMG center frequencies at the start and end of stimulation. Figure 27, taken from Tkach, D., Huang, H. & Kuiken, TA Study of stability of time-domain features for electromyographic pattern recognition. J NeuroEngineering Rehabil 7, 21 (2010), provides an example to illustrate how a shift in center frequency indicates fatigue. Peripheral fatigue occurs in the muscle itself, farther from the motor plate. Fluctuations in EMG amplitude can indicate fatigue, but are less certain than frequency shifts. The recorded EMG signal frequency is proportional to the number of motor unit firings. As fatigue begins, higher frequency type II symmetry motor fibers tend to decrease first. In various embodiments, the frequency can be determined through fast Fourier transformation (FFT) analysis or by using a zero-crossing counter and slide window on the filtered EMG signal.

[0109]

[0075] By measuring stimulation-induced muscle fatigue, a “safety mode” can be provided to the system of the present disclosure, which is configured to switch off the stimulation from the stimulation electrode in response to the determination of the occurrence of stimulation-induced muscle fatigue. As just one example, in some embodiments, the safety mode may be configured to switch off the stimulation for a predetermined time interval after stimulation-induced muscle fatigue has been identified, and the stimulation may be reactivated at the end of the predetermined interval.

[0110]

[0076] It should be noted that any measurements, materials, and drawings provided are intended to be illustrative examples for the description of the embodiments described herein and are not intended to expressly limit the embodiments to those literally shown and / or described. Although the embodiments are presented primarily in relation to use in the treatment of OSA, sensing and stimulation can be used in the treatment of a wide variety of diseases.

[0111]

[0077] Although the present invention has been described in relation to specific embodiments and uses, those skilled in the art can, in view of this teaching, generate additional embodiments and modifications without departing from or exceeding the spirit of the claimed invention. Accordingly, it should be understood that the drawings and descriptions herein are proposed as examples to facilitate the understanding of the present invention and should not be construed as limiting the scope of the invention.

Claims

1. A system for treating obstructive sleep apnea, An array of multiple electrodes, A memory capable of storing computer executable instructions, and a processor configured to facilitate the execution of the executable instructions stored in the memory, Includes, The aforementioned instruction is given to the processor, Receiving EMG signals from the array of multiple electrodes positioned in the submental region of an individual; filtering the EMG signals to generate a signal envelope; measuring genioglossal muscle activity from the signal envelope to determine the optimal sensing electrode; pulsing each electrode and measuring the response at the optimal sensing electrode to determine the optimal stimulating electrode; identifying the inspiratory and expiratory breathing phases from the signal envelope; delivering a stimulus to the hypoglossal nerve via the optimal stimulating electrode at the start of the inspiratory breathing phase; and confirming from the optimal sensing electrode that the stimulus is useful for moving the genioglossal muscle. A system that enables this to happen.

2. The system according to claim 1, wherein determining the optimal sensing electrode further includes performing an operation to activate the genioglossus muscle and position the electrode in such a way that the response is best.

3. The system according to claim 1 or 2, wherein the instruction further provides continuous closed-loop feedback from the EMG waveform from the optimal sensing electrode to confirm that the optimal stimulating electrode is stimulating the genioglossus muscle.

4. The system according to claim 3, wherein the optimal stimulating electrode can be replaced with a new electrode pair.

5. The system according to claim 3 or 4, wherein the optimal sensing electrode can be replaced with a new electrode pair.

6. The closed-loop feedback terminates and enters a safety setting when a predetermined threshold is exceeded, according to any one of claims 3 to 5.

7. The system according to any one of claims 1 to 6, wherein the stimulus has amplitude and frequency, and the frequency is adjusted based on a sensation perceived by the individual.

8. The system according to any one of claims 1 to 7, wherein the command further comprises measuring the M-wave and / or H-wave reflex response at the optimal sensing electrode and adjusting the stimulus based on the M-wave and / or H-wave reflex response.

9. The system according to any one of claims 1 to 8, wherein the start of the inspiratory breathing phase is predicted based on the previous cycle time and average respiratory rate.

10. The system according to any one of claims 1 to 9, wherein sleep is detected by measuring changes in respiratory rate and tidal volume.

11. The system according to any one of claims 1 to 10, wherein sleep apnea events are predicted based on changes in the EMG signal during inspiration.

12. The system according to claim 11, wherein the change in the EMG signal during inhalation includes a decrease in chinoglossal muscle activity identified from the integrated EMG signal of chinoglossal muscle activity measured by the optimal sensing electrode.

13. Alternatively, the system according to any one of claims 1 to 12, wherein stimulation is provided by the optimal stimulation electrode before the predicted initiation of inspiration occurs.

14. The predicted intake start is given by formula ti' ｋ＋１ =te ｋ + (ti ｋ -te ｋ－１ ) - e is used to calculate, in the formula, ti' ｋ＋１ This indicates the time when the (k+1)th intake cycle is expected to begin. te ｋ This indicates the time when the k-th exhalation cycle began. ti ｋ This indicates the time when the k-th intake cycle began. te ｋ－１ This indicates the time when the (k-1)th exhalation cycle began. The system according to claim 13, wherein e represents the margin of error.

15. The amplitude of the output stimulation signal of the aforementioned optimal stimulation electrode is given by: [Math 1] It is characterized by, in the formula, [Math 2] This is the average of the GGAV signals for each breath k, where, [Math 3] where m = F ｓ / F ｐ where F ｓ is the sampling rate of the EMG signal, and F ｐ is the stimulation frequency K ｐ and K Ｉ These are the proportional and integral control gains, respectively. The system according to any one of claims 1 to 14, wherein SP is a predefined threshold value of the EMG moving average amplitude.

16. The system according to claim 15, wherein the SP value is updated in response to a determination that the user is in a sleep state.

17. The system according to any one of claims 1 to 16, further comprising one or more microphones positioned on either side of the throat to record sound from the airway.

18. The system according to any one of claims 1 to 17, further comprising an accelerometer positioned at the center of the subchinicular triangle.

19. The system according to claim 18, further configured to determine from accelerometer data whether the user is in a sleeping state based on the user's posture and / or on the user's sudden movements.

20. The system according to claim 18 or 19, further configured to switch off stimulation from the optimal stimulation electrode in response to a determination of the user's posture from accelerometer data.

21. The system according to any one of claims 17 to 20, wherein the accelerometer includes a dynamic component configured to determine one or more of the user's snoring, speech, breathing, coughing, and choking.

22. The system according to any one of claims 17 to 21, wherein the optimal sensing electrode is determined to be a pair of bipolar electrodes having the maximum energy in each orientation.

23. The system according to any one of claims 17 to 22, configured to deactivate stimulation from the optimal stimulation electrode in at least one of the user's postures based on user sleep apnea reduction data.

24. Temperature sensor and / or SpO ２ The system according to any one of claims 1 to 23, further comprising a sensor. [Request Item 25] [Number 4] The system according to any one of claims 1 to 24, further configured to identify a transfer function characterized by, wherein the output is a recorded EMG signal and the input is a stimulus waveform.

26. The system according to claim 25, further configured to perform the transfer function to filter out stimulus-induced artifacts from the recorded EMG signal.

27. The system according to claim 25 or 26, wherein the stimulus waveform is provided with a sufficiently low amplitude so as not to elicit a motor response.

28. The system according to any one of claims 1 to 27, further configured to switch off the stimulation from the stimulating electrode in response to the detection of the occurrence of stimulation-induced muscle fatigue.

29. The system according to claim 28, wherein stimulus-induced muscle fatigue is determined by comparing the evoked genioglossus muscle EMG center frequencies at the start and end of the stimulus.

30. An intraoral device comprising the system described in any one of claims 1 to 29.

31. The oral device according to claim 30, wherein the oral device is a mandibular anterior fixation device, a mouthguard, or a retainer.

32. The oral device according to claim 31, wherein the oral device is a boil and bite mouthguard.

33. A method for treating obstructive sleep apnea, To provide an array of multiple electrodes, Receiving EMG signals from the array of multiple electrodes arranged in the subchinicular region of an individual, The EMG signal is filtered to generate a signal envelope, The optimal sensing electrode is determined by measuring the genioglossal muscle activity from the aforementioned signal envelope. Each electrode is pulsed, and the response at the optimal sensing electrode is measured to determine the optimal stimulating electrode. Identifying the inspiratory and expiratory breathing phases from the aforementioned signal envelope, To deliver stimulation to the hypoglossal nerve via the optimal stimulating electrode at the start of the inspiratory breathing phase, Confirmation from the optimal sensing electrode that the aforementioned stimulus is useful in moving the genioglossus muscle, A method that includes this.

34. The method according to claim 33, wherein the step of determining the optimal sensing electrode further includes activating the genioglossus muscle and performing an operation to position the electrode in such a way as to provide the best response.

35. The method according to claim 33 or 34, further comprising providing continuous closed-loop feedback from the EMG waveform from the optimal sensing electrode to confirm that the optimal stimulating electrode is stimulating the genioglossus muscle.

36. The method according to any one of claims 33 to 35, further comprising replacing the optimal stimulating electrode with a new electrode pair.

37. The method according to any one of claims 33 to 36, further comprising replacing the optimal sensing electrode with a new electrode pair.

38. The method according to claim 37, further comprising terminating the closed-loop feedback and setting a safety configuration when a predefined threshold is exceeded.

39. The method according to any one of claims 33 to 38, wherein the stimulus has amplitude and frequency, and the method further comprises adjusting the frequency based on a sensation perceived by the individual.

40. The method according to any one of claims 33 to 39, further comprising measuring the M-wave and / or H-wave reflex response at the optimal sensing electrode, and adjusting the stimulus based on the M-wave and / or H-wave reflex response.

41. The method according to any one of claims 33 to 40, wherein the start of the inspiratory breathing phase is predicted based on the previous cycle time and average respiratory rate.

42. The method according to any one of claims 33 to 35, further comprising detecting sleep by measuring changes in respiratory rate and stroke volume.

43. The method according to any one of claims 33 to 36, further comprising predicting future sleep apnea events based on changes in the EMG signal during inhalation.

44. The method according to claim 43, wherein the change in the EMG signal during inhalation includes a decrease in chinoglossal muscle activity identified from the integrated EMG signal of chinoglossal muscle activity measured by the optimal sensing electrode.

45. The method according to any one of claims 33 to 44, wherein, as an alternative to the step of providing stimulation at the start of the inspiratory breathing phase, the method includes the step of providing stimulation before the expected initiation of inspiration occurs.

46. Formula ti' ｋ＋１ =te ｋ + (ti ｋ -te ｋ－１ ) -e further includes the step of identifying the predicted intake onset using the formula, ti' ｋ＋１ This indicates the time when the (k+1)th intake cycle is expected to begin. te ｋ This indicates the time when the k-th exhalation cycle began. ti ｋ This indicates the time when the k-th intake cycle began. te ｋ－１ This indicates the time when the (k-1)th exhalation cycle began. The method according to claim 45, wherein e indicates a margin of error.

47. The amplitude of the output stimulation signal of the aforementioned optimal stimulation electrode is given by: [Math 5] It is characterized by, in the formula, [Math 6] This is the average of the GGAV signals for each breath k, where, [Number 7] Therefore, m = F ｓ / F ｐ F ｓ F is the sample rate of the EMG signal. ｐ This is the aforementioned stimulation frequency, K ｐ and K Ｉ These are the proportional and integral control gains, respectively. The system according to any one of claims 33 to 46, wherein SP is a predefined threshold of the EMG moving average amplitude.

48. The method according to claim 47, further comprising the step of updating the value of SP in response to a determination that the user is in a sleep state.

49. The method according to any one of claims 33 to 48, further comprising the step of providing one or more microphones positioned on either side of the throat to record sounds from the airway.

50. The method according to any one of claims 33 to 49, further comprising the step of providing an accelerometer positioned in the center of the subchinnic triangle.

51. The method according to claim 50, further comprising the step of determining from accelerometer data whether the system is in a sleeping state based on the user's posture and / or on the user's sudden movements.

52. The method according to claim 50 or 51, further comprising the step of switching off the stimulation from the optimal stimulation electrode in response to a determination of the user's posture from accelerometer data.

53. The method according to any one of claims 33 to 52, further comprising the step of deactivating the stimulation from the optimal stimulation electrode in at least one of the user's positions based on user sleep apnea reduction data. [Request Item 54] [Number 8] The method according to any one of claims 33 to 53, further comprising the step of identifying a transfer function characterized by, wherein the output is a recorded EMG signal and the input is a stimulus waveform.

55. The method according to claim 54, further comprising the step of performing the transfer function to filter out stimulus-induced artifacts from the recorded EMG signal.

56. The method according to claim 54 or 55, wherein the stimulus waveform is provided with a sufficiently low amplitude so as not to elicit a motor response.

57. The method according to any one of claims 33 to 56, further comprising the step of switching off the stimulation from the stimulating electrode in response to the determination that the occurrence of stimulation-induced muscle fatigue has occurred.

58. The method according to claim 57, wherein stimulation-induced muscle fatigue is determined by comparing the evoked genioglossus muscle EMG center frequencies at the start and end of stimulation.

Images