Method and device to treat disordered sleep breathing

Abstract

A medical device is provided that can be used to determine the optimal parameters for the detection of respiratory activity and the best modality for the modulation of the nervous system and / or respiratory muscles in response. Related systems and algorithms for programming the device, as well as the design parameters, are provided.

Description

Method and Device to Treat Disordered Sleep BreathingCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Application No. 63 / 694,669, filed September 13, 2024, the entire contents being hereby incorporated by reference.TECHNICAL FIELD

[0002] The techniques herein relate to the methods for detecting and determining programming parameters for implantable devices that are designed to electrically stimulate the diaphragm, different nerves, and / or efferent and afferent traffic to treat sleep-disordered breathing - such as Obstructive Sleep Apnea (OSA), Central Sleep Apnea (CSA), and Mixed Sleep Apnea (MSA). The techniques may be used to automate the identification parameters that are needed for the recognition and classification of input signals as well as their values. Similarly, techniques disclosed are also used for the optimization of the electrical stimulation therapy delivered by the implantable device.BACKGROUND

[0003] Healthy sleep enhances both physical and mental health. Sleep occurs in stages, including REM and non-REM sleep, allowing the body to rest and restore energy. Restful sleep helps to manage stress, solve problems, and recover from illness. Conversely, insufficient sleep can lead to numerous health issues, impacting cognitive and emotional well-being of an individual.

[0004] During sleep, individuals typically transition through four stages: non-REM N1 , N2, N3, and REM (rapid eye movement). These stages cycle from N1 to REM and then restart with N1 or N2. Healthy children and adults typically spend approximately 50% of their sleep in N2, 20% in REM, and the remaining 30% in the other stages. Sleep Disordered Breathing, including Obstructive Sleep Apnea (OSA), Central Sleep Apnea (CSA), and Mixed Sleep Apnea (MSA), disrupts these normal stage transitions.

[0005] While the neurophysiology of sleep may not be completely understood, it is clear that uninterrupted, cyclic sleep, including REM stages, is essential for health. Sleep apnea syndromes disrupt this continuity, causing daytime sleepiness, fatigue, and other serioushealth issues. Sleep-disordered breathing, including OSA, OSA, and MSA, involves repeated episodes of apnea or hypopnea lasting 10 seconds or more, often occurring hundreds of times per night. These apneas can be obstructive (OSA), central (CSA), or a combination of both (Mixed Sleep Apnea).

[0006] Obstructive Sleep Apnea (OSA) is a well-recognized sleep disorder that affects nearly 1 billion people aged 30-65 years old, accordingto Benjafield et al. OSA is characterized by periodic interruptions of lung ventilation that disrupt sleep due to a momentary collapse and obstruction of the pharyngeal airway.

[0007] Obstruction of the pharyngeal airway can be attributed to decreased upper airway muscle tone and excessive relaxation of the muscles that support the soft tissues in the throat, such as the tongue and / or soft palate, failing to maintain the airway patency. Overrelaxation of these muscles results in narrowing of the pharyngeal airway, causing airflow obstruction that limits respiration, which can ultimately decrease oxygen saturation (hypoxia).

[0008] Central Sleep Apnea (CSA) is a less common sleep disorder characterized by apneas due to a lack of signals from the respiratory center of the brain. With CSA, thoracic neural receptors fail to send the signals to the respiratory center to initiate inspiration. As a result, airflow ceases due to no respiratory muscle activity, and hypoxia results.

[0009] In CSA, the root cause lies in a dysfunction within the central nervous system, particularly in the brainstem's respiratory control centers. Consequently, during sleep, the brain fails to send the appropriate signals to the respiratory muscles, resulting in pauses in breathing or shallow breathing episodes. Since the issue originates centrally, individuals with CSA typically exhibit minimal to no respiratory effort during apneic events. By contrast, OSA is characterized by physical obstruction or collapse of the upper airway despite preserved respiratory drive; individuals with OSA often demonstrate visible respiratory efforts.

[0010] Mixed Sleep Apnea is a form of SDB combining OSA and CSA, where there is both decreased respiratory drive and decreased upper airway muscle tone, resulting in hypoxia. While the methods to open the occluded airway are used for OSA, CSA treatment focuses onrestoring respiratory drive, often by stimulating the phrenic nerve and respiratory muscles such as the diaphragm. In some embodiments, a single stimulation lead positioned on the phrenic nerve can be used to treat both OSA and CSA in a unified therapy system.

[0011] The upper airway is innervated by several key nerves that play crucial roles in breathing, swallowing, and vocalization. Due to their role in airway patency, these nerves have been targeted by electrical stimulation for the treatment of SDB, including both OSA and CSA.

[0012] The hypoglossal nerve is one of the key motor nerves for the movement of the tongue, controlling the genioglossus, hypoglossus, intrinsic, and styloglossus muscles. The genioglossus muscle is typically targeted in treating OSA because it has phasic inspiratory activity and acts as an upper airway dilator. In OSA, decreased genioglossus tone causes the tongue to retract and impede airflow into the trachea. By stimulating the hypoglossal nerve, the tongue root is pushed forward, dislodging the obstruction.

[0013] The vagus nerve is a cranial nerve that provides parasympathetic innervation to much of the upper respiratory tract, includingthe larynx and pharynx. Thevagus nerve is responsible for controlling some of the muscles involved in swallowing and vocalization, and also regulates the constriction of airway smooth muscles, secretions, and the gag reflex.

[0014] The ansa cervicalis is a loop of nerves originating from three cervical nerves (C1 -C3) that innervates the infrahyoid muscles, excluding the thyrohyoid muscle. Stimulation of the ansa cervicalis and the resulting contraction of the sternothyroid muscle has been shown to pull the pharynx caudally, increasing the retropalatal cross-sectional area and airflow. This decreases measures of pharyngeal colla psibility, which can aid in the treatment of OSA and, in combination therapy approaches, CSA.

[0015] The phrenic nerve provides complete motor innervation to the diaphragm. Phrenic afferents project into regions that supply tonic drive to upper-airway dilators (e.g., hypoglossal) and to the rostral ventral respiratory group (rVRG). Activation causes the diaphragm to contract with inspiration, resulting in increased intrapleural space and a flattened diaphragm. A single stimulation lead on the phrenic nerve can provide therapy forboth OSA and CSA while enabling algorithmic control features that are agnostic to apnea type.

[0016] These cranial nerves are critically important for regulating and coordinating upper airway functions, including breathing, swallowing, and vocalization. Dysfunction of these nerves can lead to various upper airway and respiratory issues, including SDB such as OSA and CSA, as well as other health conditions.

[0017] Sleep disorders, particularly obstructive sleep apnea (OSA) and central sleep apnea (CSA), pose significant health risks that extend beyond mere sleep disturbances. These conditions are strongly associated with various cardiovascular diseases, including hypertension, coronary artery disease, and heart failure, as well as increased risks for stroke and metabolic disorders. The impact of sleep apnea also reaches into other areas, such as heightened danger on the roads due to increased likelihood of motor vehicle crashes and the economic burden of untreated sleep apnea is substantial, affecting both individual patients and society at large.

[0018] Studies reported by Knauert et al show a clear association between OSA and hypertension, type II diabetes, stroke, coronary artery disease, and cardiac arrhythmias. Complications in the circulatory system caused by sleep apnea arise from elevated heart rates and surges in blood pressure during apneas and hypopneas. As a result, patients with OSA exhibit higher heart rates, reduced heart rate variability, increased blood pressure, and greater arterial stiffness compared to individuals without OSA, leadingto the development of related cardiovascular diseases. Cardiovascular morbidities and hypertension also present themselves in patients suffering from CSA, where data from heart failure population studies suggest CSA may be present in 30% to 50% of heart failure patients.

[0019] Abbasi et al have also reported that OSA produces a chronic inflammatory state, resultingfrom OSA stimulating white adipose tissue. This chronic inflammatory state leads to atherosclerotic changes in the blood vessels of the patient, causing endothelial and metabolic dysfunction, further contributing to increases in cardiovascular disease.Furthermore, Hirsch et al found that the OSA severity was significantly associated withcancer risk after controlling for relevant covariates, concluding that the OSA severity is an independent risk factor for cancer.

[0020] Accordingto Gottlieb et al., sleep disorders, particularly obstructive sleep apnea, can significantly increase the risk of auto crash risk. Gottlieb et al found that the odds ratio for any motor vehicle crash increased by 15% for every 10-unit increase in AHI in the overall population and by 17% for every 10-unit increase in AHI in participants who did not report excessive sleepiness.

[0021] Furthermore, as the severity of sleep apnea increases the odds ratio for motor vehicle crashes increased as well, showing a positive correlation. Specifically, those without sleep apnea had an odds ratio (aOR) of 0.07, indicating a 7% increase in the odds of any motor vehicle crash, those with mild sleep apnea had an increase of 13%, and those with severe sleep apnea had an increase of 123%, where the AHI of each category is AHI < 5, 5 to < 15, and AHI > or = 30, respectively. This shows that obstructive sleep apnea is associated with motor vehicle crash risk and a dangerto those suffering from this disorder.

[0022] As shown by the prior statistics, sleep disorders, including OSA, CSA, and MSA, can adversely affect apnea patients and also have broader implications for those not suffering from the disorder. Sleep disorders increasingthe risk of motor vehicle crashes is a concern for the entire public, both in terms of safety and economic impact. The American Academy of Sleep Medicine reports findings from Frost & Sullivan that details the economic impact of undiagnosed and untreated obstructive sleep apnea. These findings calculated that the annual economic burden of undiagnosed sleep apnea among U.S. adults is approximately $149.6 billion, consisting of $86.9 billion in lost productivity, $26.2 billion in motor vehicle accidents, and $6.5 billion in workplace accidents. This amount does not include the cost for increase in health complications that result from sleep apnea, which would add an additional $30 billion annually in health care utilization and medication costs. Clearly, sleep disorders affect more than just the patient themselves.SUMMARY

[0023] In certain example embodiments, a device (e.g., a medical device) is provided that can be used to determine the optimal parameters for the detection of respiratory activity andthe best modalityforthe modulation of the nervous system and / or respiratory muscles in response for the treatment of sleep-disordered breathing, including OSA, CSA, and MSA. Related systems and algorithms for programmingthe device, as well as the design parameters, are provided.

[0024] In certain embodiments of the invention, hardware and / or software is used to optimize one or more algorithms that may then be deployed and used in an implantable device. In certain example embodiments, the device is defined such that it can be used online or offline. In certain examples, an ongoing optimization process may be provided to take place once, periodically, or when needed and may include algorithms that are agnostic to apnea type (OSA or CSA), as well as algorithms that are specific to an apnea type.

[0025] Additional features further allow the device to work with different patients, different sensing and stimulation configurations, and with different modes of connection to the implantable device while using performance parameters such as heart rate and heart rate variability, in addition to ventilation metrics, to optimize feedback and phase adjustment for OSA and CSA therapy.BRIEF DESCRIPTION OF THE DRAWINGS

[0026] These and other features and advantages will be better and more completely understood by referring to the following detailed description of example non-limiting illustrative embodiments in conjunction with the drawings of which:

[0027] Figure 01 shows a schematic diagram of the overall treatment system, including the implantable pulse generator (IPG), stimulation lead, cuff electrode, a patient programmer and the cloud connection,

[0028] Figure 02 shows the block diagram of the implantable system, including the stimulator, the sensors and the delivery components, such as the stimulation lead and the cuff electrode,

[0029] Figure 03 shows an exemplary timing diagram of electrical stimulation in relationship to the respiratory waveform,

[0030] Figure 04 shows the composition of motor neurons in the phrenic nerve,

[0031] Figure 05 shows the response of sensory nerves to pressure stimulation,

[0032] Figure 06 shows the expected firing pattern of motor neurons,

[0033] Figures 07A and 07B show an exemplary design of a cuff electrode surrounding a nerve, which is designed for the delivery of stimulation and / or detection of nerve activity,

[0034] Figure 08 shows an exemplary configuration of a set of electrical contacts within a cuff electrode,

[0035] Figure 09 shows the strength-duration curves for the motor and sensory nerves,

[0036] Figure 10 shows an electrically evoked compound action potential,

[0037] Figure 11A shows the raw accelerometertrace from a subject during normal breathing,

[0038] Figure 11 B shows the extraction of respiratory cycle information from the raw acceleration signal,

[0039] Figure 12 shows an exemplary configuration for the combination of accelerometer and gyroscope signals,

[0040] Figure 13 shows simultaneously recorded traces of thoracic belt, air flow and transthoracic impedance signals from a subject during normal respiration and during central sleep apnea,

[0041] Figures 14A and 14B show the physical implementation of the tripolar impedance measurement circuit and its electrical equivalent circuit respectively,

[0042] Figure 15 shows the block diagram of an exemplary configuration to extract the air flow signal from transthoracic impedance measurements,

[0043] Figures 16A and 16B show the mono-phasic and biphasic stimulation waveforms respectively, alongwith various simulation parameters,

[0044] Figure 17 shows the opening of the airway during OSA using electrical stimulation that is delivered to the phrenic nerve,

[0045] Figure 18 shows simultaneously recorded traces of thoracic belt, abdominal belt, electrical stimulation, respiratory airflow, subglossal air pressure, peripheral oxygen saturation and tidal volume from a patient who was receiving electrical stimulation where there was entrainment between the stimulation and the respiratory cycle which was keeping the airway open prior to the termination of the therapy,

[0046] Figure 19 shows simultaneously recorded traces of thoracic belt, abdominal belt, electrical stimulation, respiratory airflow, subglossal air pressure, peripheral oxygen saturation and tidal volume from a patient who was having OSA before the onset of electrical stimulation where there was entrainment between the electrical stimulation and the respiration,

[0047] Figure 20 shows simultaneously recorded traces of thoracic belt, abdominal belt, electrical stimulation, respiratory airflow, subglossal air pressure, peripheral oxygen saturation and tidal volume from a patient who was receiving electrical stimulation where there was entrainment between the electrical stimulation and the respiratory cycle and the entrainment was keeping the airway open prior to the loss of entrainment,

[0048] Figure 21 shows simultaneously recorded traces of thoracic belt, abdominal belt, electrical stimulation, respiratory airflow, subglossal air pressure, peripheral oxygen saturation and tidal volume from a patient who was receiving electrical stimulation where stimulation rate was being adjusted to demonstrate effect of entrainment between the stimulation and the respiratory cycle,

[0049] Figure 22A shows simultaneously recorded traces of thoracic belt, abdominal belt, electrical stimulation, respiratory airflow, subglossal air pressure, peripheral oxygen saturation and tidal volume from a patient who was receiving electrical stimulation where phase shift between the stimulation and respiratory cycle was being adjusted to demonstrate effect of the phase shift on ventilation efficacy,

[0050] Figure 22B shows the effect of stimulation timing on the airflow,

[0051] Figure 23 shows simultaneously recorded traces of thoracic belt, abdominal belt, electrical stimulation, respiratory airflow, subglossal air pressure, peripheral oxygen saturation and tidal volume from a patient who was receiving electrical stimulation with entrainment between the stimulation and respiratory cycle prior to the termination of the therapy which in turn allowed the central sleep apnea to surface,

[0052] Figure 24 shows the effect of phase shift between the stimulation cycle and the respiratory cycle on ventilation efficacy, which is represented by the tidal volume,

[0053] Figure 25 shows the flow chart of the “Patient Adjustable Amplitude” subroutine,

[0054] Figure 26 shows the flow chart of the “Upright Stimulation Pause” subroutine,

[0055] Figure 27 shows the flow chart of the “Roll Stimulation Pause” subroutine,

[0056] Figure 28 shows the flow chart of the “Baseline Signal Collection” subroutine,

[0057] Figure 29 shows the flow chart of the “Inspiration-Predictive Stimulation” subroutine,

[0058] Figure 30 shows the flow chart of the “Respiration-Locked Stimulation” subroutine,

[0059] Figure 31 shows the flow chart of the “Adaptive Stimulation Amplitude” subroutine,

[0060] Figure 32 shows the flow chart of the “Adaptive Stimulation Duration” subroutine,

[0061] Figure 33 shows the flow chart of the “Intermittent Stimulation” subroutine,

[0062] Figure 34 shows the flow chart of the “Adaptive Intermittent Stimulation” subroutine,

[0063] Figure 35 shows the flow chart of the “Respiratory Quality Determination” subroutine,

[0064] Figure 36 shows the flow chart of the “Signal Recovery Mode” subroutine,

[0065] Figure 37 shows the orientation of the Cartesian Coordinates used forthe measurement of body acceleration,

[0066] Figure 38 shows a simultaneous recording of chest motion, air flow, tidal volume and transthoracic impedance during normal breathing followed by a central apnea,

[0067] Figure 39 shows the raw transthoracic impedance, filtered impedance as well as the deduced respiratory cycle signals,

[0068] Figure 40 shows the deduced respiratory cycle signal, D(t), along with the stimulation signal, S(t),

[0069] Figure 41 shows a general feedback controller,

[0070] Figure 42 shows a Phase Locked Loop style controller,

[0071] Figure 43 shows the response of the stimulator to changes in the phase shift between the respiratory cycle and the stimulator rate,

[0072] Figure 44 shows an exemplary operation of the optimizer to maximize the respiration efficacy,

[0073] Figure 45A-45D shows different types of stimulation patterns including monophasic, biphasic and triphasic waveforms,

[0074] Figure 46A shows the flowchart of the Subroutine Initialize TTI2D,

[0075] Figure 46B shows the flowchart of the Subroutine Run_TTI2D,

[0076] Figure 47A shows the flowchart of the Subroutine lnitialize_Accel2D,

[0077] Figure 47B shows the flowchart of the Subroutine Run_Accel2D,

[0078] Figure 48 illustrates the classification of obstructive sleep apnea, central sleep apnea, mixed sleep apnea, normal rest, and hypopneas based on transthoracic impedance, accelerometer peak vectors, and predicted inspiration gates,

[0079] Figure 49 shows treatment data from a patient including the abdominal belt, thoracic belt, stimulation pattern, flow, and tidal volume, where a lack of stimulation shows a decrease in tidal volume and an unstable flow rate,

[0080] Figure 51 A illustrates a graph representing a two-factor model using patient pitch and bedtime to determine when stimulation should be turned on based on weighted parameters,

[0081] Figure 51 B illustrates a graph where parameters are weighted based on their variability, with less consistent parameters receiving lower weights,

[0082] Figure 51 C shows a graph where classifiers vary by day, allowing different algorithms to predict therapy onset based on day-specific sleep patterns,

[0083] Figure 52 shows an exemplary implementation of an embedded auto-titration algorithm to be used in the controller,

[0084] Figure 53 shows a logic matrix that categorizes sleep apneas based on accelerometer readings and transthoracic impedance,

[0085] Figures 54A and 54B illustrate an adaptive algorithm that adjusts pulse width to optimize therapy doses based on arousal. Figure 54A shows a graph of the algorithm's performance, while Figure 54B depicts the corresponding flowchart of its implementation,

[0086] Figures 55A and 55B shows an embodiment of a stimulation pattern used during therapy and the resulting respiratory flow, where each stimulation induces a complete respiratory cycle, where Figure 55A is the stimulation pattern and Figure 55B is the respiratory flow, and

[0087] Figure 56A-56B shows an additional embodiment of a stimulation pattern 5602 where stimulation induces an echo of respiratory flow after stimulation is ceased, where Figure 56A is the stimulation pattern and Figure 56B is the respiratory flow 5604.DETAILED DESCRIPTIONDefinitions:

[0088] “Sensory nerves” and “afferents” are terms used interchangeably and they refer to nerves originating at peripheral organs such as the diaphragm and carry information to the central nervous system.

[0089] “Sleep Disordered Breathing (SDB)” is a term used to include Obstructive Sleep Apnea (OSA), Central Sleep Apnea (CSA), and Mixed Sleep Apnea (MSA).

[0090] “Motor nerves” and “efferents” are terms used interchangeably and they refer to nerves originating at the central nervous system and produce excitation to the muscles such as the diaphragm.

[0091] “Cuff electrode” and “electrode” are terms used interchangeably and they refer to the device that is in contact with the target tissue, such as the phrenic nerve, while being connected to the electronics.The term electrically evoked compound action potential (eCAP) represents the synchronous firing of a population of electrically stimulated nerve fibers. It can be directly recorded on a surgically exposed nerve trunk.General Description of the Therapeutic System:

[0092] A system for phrenic nerve stimulation to treat and cure sleep disorder breathing, including OSA, CSA, and MSA, is comprised of five major components, which are:

[0093] One or more sensors,

[0094] A controller,

[0095] A stimulator,

[0096] Implantable hardware, and

[0097] External hardware.

[0098] Figures 1 and 2 show the clinical configuration of the phrenic nerve monitoring and modulation system during its training phase. Figure 1 shows a patient 0110 and the phrenic nerve 0120 of the patient 0110. The phrenic nerve system 0100 includes a stimulation electrode 0130 that is coupled to the phrenic nerve 0120 of the patient 0110. In certainexample embodiments, the stimulation electrode 0130 (which is an example of an implantable device or implantable hardware) can be a cuff-type design.

[0099] Further hardware of Figure 1 includes a controller 0170 which takes input signals 0150 or 0160 from external or internal sensors 0130 or 0140. Internal or external sensors may be attached to devices also providing electrode stimulation to the phrenic nerve 0130 or 0140, to carry an input signal 0150 or 0160 to the controller.

[0100] The phrenic nerve system 0100 can also include one or more (e.g., two or more) sensors 0140 provided to sense physiological characteristics (e.g., different types of physical phenomenon) of the patient 0110. Examples of different types of sensors that may be used in certain example embodiments include transthoracic impedance, accelerometer, pleural pressure, oxygen saturation, and the like. Sensors output signals (either in digital or analog form) via a sensory signal pathway 0180 to electronics 0255 (discussed below). The sensory signal pathway 0180 may be wired or wireless in order to allow signals from the one or more sensors 0140 to be communicated to another device.

[0101] Nerve signal 0180 provides signals regarding sensory nerve firings. Such signals are provided to electronics 0170.

[0102] Phrenic nerve system 0100 also includes electronics 0255. Any or all of the electronics may be implanted within the patient or may be provided externally to the patient. A stimulation lead is also provided that couples the electronics to the stimulation electrode 0130 and, in certain embodiments, a single stimulation lead is configured to deliver therapy that treats both OSA and CSA.

[0103] The set of sensors (e.g., one or more sensors 0140 from Figure 1 ) provide the signals going into the input signal processor. Illustrative examples of different types of sensors that may be used in connection with certain example embodiments include any combination of the following: 1 ) passive electrical sensors that detect the electromyogram (EMG), electrocardiogram (ECG), action potential, compound action potential, and electrically evoked compound action potential (eCAP), 2) active electrical sensors such as the transthoracic impedance, 3) mechanical sensors such as pressure sensors,accelerometers, gyroscopes, and microphones, and 4) optical sensors such as the tissue oxygen sensors and blood oxygen sensors for detecting and managing SDB (OSA and CSA).

[0104] In certain example embodiments, each or any of the sensors produces a signal that is representative of the physical or physiological quantity that is being measured. For example, an acceleration sensor may produce signals that measure acceleration (e.g., proper acceleration) in 2 or 3 axes. The signals that are produced by the sensors may be represented in certain examples as a graph or as waveforms. In certain examples, and as discussed elsewhere herein, the signals from the sensors can be analyzed and one or more values extracted therefrom (e.g., when an acceleration value exceeds a given threshold).

[0105] In certain embodiments, biomarkers of a patient can be used to assess and / or optimize treatment of sleep-disordered breathing (including OSA and CSA) for a patient. Example biomarkers include breath classification types, ventilation efficacy (e.g., as discussed in connection with the ventilation efficacy detector), and / or respiratory cycle data (e.g., (e.g., as discussed in connection with the respiratory cycle detector). Other data that is gathered by sensors that is based on a physical property of the patient may also be used in determining / monitoring a biomarker. Classification types can include normal breathing, obstructive sleep apnea (OSA), and / or central sleep apnea (CSA), mixed sleep apnea (MSA), etc. Ventilation efficacy can be measured through parameters such as inspiratory duty cycle (IDC), tidal volume (TV), peak airflow during inhalation (cbMAX), minute volume (MV), tissue oxygen saturation (Sa(O2)), and the apnea-hypopnea index (AHI). Respiration cycle information can include duration and / or timing of respiratory phases. Biomarker(s) can be used in evaluating ventilation efficiency, detecting disordered breathing events, and adjusting stimulation parameters to enhance therapeutic outcomes. In some examples, a biomarker can be updated based on changes in signals received from one or more sensors.

[0106] Ventilation efficacy can be calculated using one or more of the inspiration time, expiration time, tidal volume, heart rate, heart rate variability, and the number of apnea hypopnea events of various durations.

[0107] In some examples, the ventilation efficacy is calculated usingthe inspiration time:VE01 = C01 — (InspirationTimey1

[0108] where VE01 is the ventilation efficacy, InspirationTime is the duration of the inspiratory period (e.g., the length of time that the air flows into the lungs of the patient) in seconds, n is a number larger than zero, such as 0.2, or 1 .0 or 2.3, and C01 is a constant, such as 5.0.

[0109] In some examples, the ventilation efficacy is calculated using both the inspiration time and the expiration time:(Inspiration! ime)‘lVE02 = C02 - -;- - — . . „ - —(InspirationTime + Expiration / ime)m

[0110] where VE02 is the ventilation efficacy, InspirationTime is the duration of the inspiratory period (e.g., the length of time that the air flows into the lungs of the patient) in seconds, InspirationTime is the duration of the expiratory period (e.g., the length of time that the air flows out of the lungs of the patient) in seconds, n is a number larger than zero, such as 0.2, 1 .0, 2.3, or the like, and m is a number larger than zero, such as 0.5, or 1 .0 or 2.5, and C02 is a constant, such as 1.0.

[0111] In some examples, the ventilation efficacy is calculated usingthe tidal volume: VE03k= f TV^TV^ TV^ . TVk-N)

[0112] where VE03k is the value of the ventilation efficacy following the k-th breath, TVk is the tidal volume forthe i-th breath (e.g., measured in Liters), f is a function, such as the mean, median, minimum, or maximum, or root mean square, and N is a positive integer.

[0113] In some examples, the ventilation efficacy is calculated usingthe heart rate:

[0114] where VE04k is the value of the ventilation efficacy following the k-th cardiac cycle, HRk is the heart rate for the i-th cardiac cycle, measured in beats per minute, f is a function, such as the mean, median, minimum, or maximum, or root mean square, N is a positive integer, and C04 is constant, such as 175.0.

[0115] In some examples, the ventilation efficacy is calculated usingthe heart rate variability:VE05k= f(HRk,HRk-1,HRk-2HRk-N)

[0116] where VE05kis the value of the ventilation efficacy following the k-th cardiac cycle, HRkis the heart rate for the i-th cardiac cycle, measured in beats per minute, f is a function, such as the standard deviation and range (max-min), and N is a positive integer.

[0117] In some examples, the ventilation efficacy is calculated usingthe apnea and hypopnea counts:7£06(t) = Number of events during the time interval of t — AT to t

[0118] where VE06(t) is the value of the ventilation efficacy at time t, AT is the time interval, Events are apneas, hypopneas or combination of apneas and hyopneas.

[0119] As shown in Figure 2, information is obtained from the cloud 0245 (one or more computer systems) after implanting an implantable device 0150 into the patient. Following the implant, an initial set of parameters are downloaded into the implant 0150 via a cloud storage device 0255 (a computer system that includes one or processors). The cloud storage device 0255 then can act as a bridge for collecting and sending data from the sensors 0130 to the cloud 0245. For example, the cloud storage device 0255 could be an Arduino, where the Arduino connects to the cloud 0245 via wired or wireless connectivity. For example, cloud storage device 0255 may include ESP8266 and / or ESP32 modules that enable wireless and / or wired connectivity- which may then connectto Arduino loT Cloud (or other third-party cloud platforms). Other cloud storage devices can include mobile (e.g., smart) phones, tablets, laptops, desktop computers, network-attached storage (NAS) devices, and loT (internet of things) devices, and the like.

[0120] Data that is generated from the sensors 0130 and the treatment parameters are stored in the cloud storage device 0255 and periodically uploaded 0265 into the cloud 0245 where additional optimization may take place. Subsequently, the treatment parameters for the algorithms are downloaded and used by the implant 0230. This embodiment offers two advantages: First, control / adjustment of the treatment parameters may be performed in an ongoing manner. This allows, for example, the treatment parameters to be optimized over a period of time (e.g., based on data from a given patient and / or other patients). Second, as one or more sets of treatment parameters reside in the cloud, then can then be used as the initial set of parameters for other patients (e.g., futurepatients). Algorithms controlling this system will be discussed herein, including algorithms that are agnostic or specific to apnea type.

[0121] Figure 3 shows an exemplary timing diagram of the stimulation pattern 302 in reference to the respiration waveform 304 representing the tidal volume of the patient. In this example, the stimulation is delivered in such a way that the onset of the burst train occurs before the onset of the inhalation. The time interval between the onset of the stimulation and the onset of the inhalation is defined as "APPROACH" 310. Each stimulation pulse is characterized by an amplitude 306 and a pulse duration 308. The pulses are repeated with a defined period 312, representing the interval between successive pulses, and all events are shown with respect to time.

[0122] Sensory nerves, also known as afferents, carry information from the peripheral organs to the central nervous system. They respond to the sensory stimulation by changing their firing rate, as illustrated in Figure 05. In the example that is shown in Figure 05 the firing frequency of the sensory nerve begins to increase, in this case from zero, as the input stimulus is increased beyond PMIN 502. The nerve firing rate saturates at fMAX 504 once the stimulus level reaches and exceeds PMAX 506. It should be noted that some sensory nerves maintain a non-zero firing rate even in the absence of any physical stimuli.

[0123] In addition to having a non-linear input-output relationship as shown in Figure 05, sensory nerves also have a time dependent response to the physical stimulus that they receive. Although they produce a rapid response to the initial stimulus, their response to the same level of stimulus does decrease overtime.

[0124] Motor nerves, also known as efferents, carry information from the central nervous system to peripheral organs. They provide the excitation to the muscles, such as the diaphragm of the respiratory system. A typical firing pattern of a motor neuron is shown in Figure 6. A normal firing duration of the motor neuron is usually 300 milli-seconds, as shown by the train of firings 610 from t = 100 milli-seconds to t = 400 milli-seconds in Figure 6. It should be noted that the firing frequency of the motor neuron 640 is not fixed, but it increases as a function of time. In the example that is shown in Figure 6, the firing frequency 640 increases from 40 Hertz to 60 Hertz, which may or may not be in a linear fashion. Most ofthe time, the recorded signal is too noisy to be analyzed in detail, hence its time integral 620 is constructed. Furthermore, a smoothed version 630 of the time integral is used for signal processing purposes.

[0125] Both the motor nerves and the sensory nerves can be stimulated with externally applied electrical signals. This can be done with various types of electrodes, such as the cuff electrode 0850 on the nerve 0810 as shown in Figure 8, with electrodes 0820 embedded in the cuff. Typical strength-duration curves for the capture of sensory and motor nerves are shown in Figure 9. Due to the reduced cross-sectional area of the sensory nerves, larger pulse-widths are required to capture them compared to the motor nerves. Following the delivery of the electrical stimulation to the nerve, a resulting action potential 1000 travels through the nerve fiber, as shown in Figure 10. The waveform includes a stimulus artifact 1010, followed by an initial negative deflection 1020, a subsequent positive peak 1030, and a later negative trough 1040. The compound action potential amplitude can be quantified as Vmax 1030 minus the average of Vminl 1020 and Vmin2 1040. It should be noted that even though the duration of the action potential waveform shown in Figure 10 is rather short, muscle contractions that last much longer can be achieved by repeated application of the stimulation train to generate a sustained tetanic contraction.Components

[0126] An implantable device is disclosed which delivers electrical stimulation to a nerve for the treatment of disordered breathing including OSA and CSA, the device can include:- An electrical pulse generator;- A set of sensors;- A respiratory cycle detector;- A ventilation efficacy detector;- A processor configured to perform operations of an algorithm to adjust the amplitude, frequency, and / or the time offset of the stimulation with respect to the respiratory cycle;- A single stimulation lead;A cuff electrode;A power supply; andA telemetry unit.

[0127] The following sections describe these components.Electrical pulse generator:

[0128] The electrical pulse generator includes output circuitry that generates stimulation pulses and is housed within the implantable pulse generator (IPG). The pulses generated can be monophasic or multiphasic, such as biphasic ortriphasic, as illustrated in Figures 16 and 45. Monophasic pulses 4505, shown in Figures 16A and 45A, can be easier to generate and can have short durations, which can decrease battery consumption. However, they may result in unbalanced charge delivery to the tissue. Biphasic pulses 4510 and 4515, shown in Figures 16B, 45B, and 45C, have both positive (anodic) and negative (cathodic) phases, providing charge-balanced stimuli to the tissue. The biphasic waveform 4510 in Figure 45B features an interphase delay between the anodic and cathodic pulses, whereas the waveform in Figure 45C eliminates this interphase delay. Figure 45D illustrates a triphasic stimulation waveform 4520, which has the added benefit of reducing electrode polarization, thereby improving the detection of electrically evoked potentials, such as electrically evoked compound action potentials (eCAP).

[0129] The waveforms used for electrical stimulation of nerves are detailed in Figures 16A and 16B. Figure 16A depicts a monopolar stimulation waveform, while Figure 16B shows a bipolar stimulation waveform. Several parameters of these stimulation waveforms are programmable, including amplitude (1120a and 1120b), pulse-width (1130a, 1130b, and 1130c), pulse period (1110a and 1110b), train duration (1140a and 1140b), and silence duration (1145a and 1145b). Additionally, the stimulation can be delivered as either a voltage or a current waveform, and may remain constant or vary during the stimulation period, particularly in terms of pulse-width.

[0130] The stimulation waveforms delivered to the nervous tissue can be in the form of either constant voltage or constant current, meaning that the waveforms shown in Figures 16A / 16B and 45 could represent voltages or currents. These parameters provide flexibility intailoring the stimulation to specific physiological needs, with additional waveform properties and control mechanisms illustrated in Figures 16A and 16B. Stimulation rate is the rate that the pulse trained are delivered at and is given as the reciprocal of sum of pulse train duration (1140a and 1140b) and silent duration (1145a and 1145b):

[0131] Stimulation Rate = 1 / (TTRAINS + TSILENT).Sensors:

[0132] Sensors may include: 1) linear accelerometers, 2) rotational gyroscopes, 3) transthoracic impedance sensors, 4) microphones, 5) oxygen saturation sensors, 6) intrathoracic pressure sensors, 7) electrocardiogram sensors, 8) electromyogram sensors, 9) evoked potential sensors, and. or optical sensors.

[0133] Sensors may be implanted. In other words, they may be completely inside the body of the patient, such as an accelerometer that is inside the casing of the pulse generator, or a pressure sensor that is in the thoracic cavity. It is also possible to have sensors that are mounted on the body of the patient, either on direct contact with the skin, orwearable type, such as a microphone or a transthoracic impedance sensor acting as an impedance plethysmography device. It is also possible to have sensors that are placed completely remotely, such as an accelerometer or a millimeter wave motion sensor that is placed on or underthe mattress orthe pillowto detect the motion and or the respiration of the patient. In all cases, the sensors provide the measurements that they make to the processor via wired or wireless connection.Respiratory cycle detector:

[0134] Respiratory cycle detector can include hardware and, in certain examples, software in combination with the hardware. Respiratory cycle detector uses one or more Sensors to detect a respiratory cycle that includes one or more segments of inhalations, one or more segments of exhalations as well as one or more segments where there is no respiratory activity, aka quiet period. Usually, a respiratory cycle includes an inhalation, and exhalation and may have zero, one or more quiet periods. A respiratory cycle detector may identify the various phases of the respiration including the inhalation, exhalation, holdingthe breath, and so on.

[0135] An example of a respiratory cycle detector which uses the transthoracic impedance sensor is illustrated in Figures 14A, 14B, and 39. As shown in Figure 14A, a low amplitude current source 1005 is applied to the tip (“T”) electrode 1010 of the transthoracic impedance lead, where the return electrode is the case (“C”) electrode 1020. The voltage drop between the ring electrode (“R”) 1030 and the case (“C”) electrode 1020 is measured by sensing circuitry and used as an indicator of the transthoracic impedance. Figure 14B provides an electrical equivalent circuit model of the impedance measurement circuitry.

[0136] Figure 13 shows the ability of transthoracic impedance 1310 to detect normal breathing as well as an apnea. Thoracic belt 1320 and signal 1302 indicates the effort made by the subject to breath, where the airflow trace 1330 shows the results of that effort. As can be seen from the first 1340 and the third quarters 1350 of the traces of Figure 13, the patient is making an effort to breathe - e.g., successfully inhaling and exhaling - and the impedance trace 1310 indicating the results. However, during the second 1350 and the fourth quarter 1370 of the traces, the patient is having a central apnea, indicated by lack of effort and lack of airflow, and the flat line of the impedance signal agreeing with the apnea conclusion.Hence, the transthoracic impedance signal 1310 can be used to detect the parameters of the respiratory function in real time according to certain example embodiments.

[0137] Figure 38 also shows recordings that were obtained as a patient transitioned from natural breathingto a central apnea in the absence of any therapy. As it can be seen from this trace, the patient took 4-5 normal breaths, and then the breaths got shallower, and eventually stopped breathing, as indicated by the diminishing amplitude of the Chest Motion Y(t) 3805, where increasing amplitude is the chest expanding 2850 and decreasing amplitude is the chest contracting 3855, Air Flow <t>(t) 3810 and Tidal Volume V(t) 3815 traces. Transthoracic Impedance Trace 3820 closely follows the respiration signal indicating that transthoracic impedance 3820 can be used to determine the phases of the respiration, such as inhalation 3825 and exhalation 2830 as well as no airflow condition.

[0138] Figure 39 shows the process that can be followed to construct the respiratory cycle from the transthoracic impedance signal (TTI) 3905. Since the transthoracic impedance signal 3905 contains a drifting baseline, the baseline drift 3908 (shown by thedashed Line in the upper trace of Figure 39) must be removed, which can be accomplished using an equation as shown below:

[0139] In this equation, TTI(n) represents the raw transthoracic impedance 3905, Z(n) is the filtered impedance signal 3910, and M is a positive integer, e.g., 50 or 750, chosen such that the duration of M samples is longer than the expected period of the baseline drift. After baseline removal, the filtered signal Z(t) 3910 exhibits a cyclic nature centered at zero, as shown in the middle trace 3925 of Figure 39.

[0140] An impedance threshold 3915, such as ±0.1 Q, is applied to the filtered impedance 3910 to obtain a digital signal D(t) 3930, shown in the lower trace of Figure 39. The digital representation D(t) 3930 corresponds to the deduced respiratory cycle, with binary transitions marking inspiration (“1 ”) 3935, expiration (“Ex”) 3940, and complete respiratory cycles (“Cycle”) 3945. The peak-to-peak amplitude 3950 of the impedance waveform can also be measured as part of this process, providing a direct indicator of tidal volume.

[0141] Below is a pseudo code of an example subroutine for the implementation of the process for conversion of TTI(t) first to Z(t) and then to D(t):

[0142] Figure 40 illustrates the deduced respiration signal D(t) 4930, derived from the process described with respect to Figure 39, together with a corresponding stimulation signal S(t) 4005. The deduced respiration signal D(t) 4930 provides a binary representation of the patient’s respiratory cycle and serves as the input for determining the timing of stimulation delivery. In the example shown, the stimulation signal S(t) 4005 includes discrete bursts of stimulation pulses 4010, 4020, 4030, 4040, and 4050, each of which is scheduled to occur shortly prior to the onset of inspiration 4015, 4025, 4035, 4045, and 4055, respectively.

[0143] In this example, the interval between consecutive respiratory cycles is constant, allowing the control algorithm to operate predictively with high accuracy. Because the onset of each inspiration can be anticipated reliably, stimulation is delivered at a consistent phase relative to the inspiratory transition. This timing strategy ensures that stimulation consistently precedes the rise of the inspiratory effort, which is advantageous for augmenting tidal volume and maintaining synchrony with minimal algorithmic adjustment.

[0144] Furthermore, the example of Figure 40 serves as a baseline condition in which the algorithm’s predictive model operates without requiring correction. As described in subsequent figures, when variability in respiratory timing arises, additional control mechanisms, such as the phase-locked loop algorithm, are employed.

[0145] Figure 43 illustrates an example of the deduced respiration signal, D(t), 3930, and a corresponding stimulation signal, S(t) 4305 in a case where the patient’s breathing pattern is irregular, with variable intervals between inspirations. The deduced respiration signal 3930 is shown as a binary waveform, with respiratory cycles 4310, 4320, 4330, 4340, and 4350 marked by elevated portions of the signal. The stimulation signal 4305 consists ofdiscrete bursts of stimulation pulses 4315, 4325, 4335, 4345, and 4355 delivered in association with the predicted onset of inspiration.

[0146] In the example shown, the predictive algorithm generates stimulation bursts 4315, 4325, 4335, 4345, and 4355 based on the assumption that the patient’s respiratory rate will remain constant, such that the interval between consecutive inspirations is approximately fixed. Additionally, the algorithm schedules each stimulation burst to occur slightly before the predicted onset of inspiration.

[0147] As shown, the patient’s respiration 3930 deviates from the predicted periodicity. Specifically, inspiration 4330 occurs earlier than anticipated, resulting in a phase shift relative to stimulation burst 4335. In such cases, the algorithm realigns stimulation timing usingthe control strategy outlined in Figure 42, which employs a phase-locked loop (PLL) to compare the deduced respiration signal D(t) with the oscillator driving the stimulator. When misalignment occurs, the PLL detects the phase shift and adjusts the stimulator rate so that subsequent bursts (e.g., 4345 and 4355) are shifted toward the actual inspiration timing. This adaptive response enables the algorithm to re-establish synchrony with the patient’s respiration within one or two cycles.

[0148] Subroutine I nitia lize_TTI2D is called once for the initialization of the buffer, andFigure 46A provides a flow chart of this process. Subroutine Run_TTI2D is called each time a new value of TTI(n) is read, and it returns a digital signal, D(t), in the form of a zero or a one, representing the deduced respiratory cycle signal. The routine begins at start block 4602, followed by the initialization of the buffer size M = 1 ,5004604. Next, an index variable is initialized to i = 1 4606. A comparison step 4608 checks whether the current index i has exceeded the buffer size M. If i is not greater than M, the routine assigns the buffer value stor(i) = 0.04610 and then increments the index with i = i + 1 4612. The process loops back to step 4608 until the index exceeds the buffer size. Once the condition i > M 4608 evaluates true, the initialization is completed by setting D(n) = 04614. The routine then terminates at end block 4616.

[0149] Figures 46B and 46C illustrate a flowchart of Subroutine Run TTI2D, which processes thoracic impedance values TTI(n) to generate a digital respiratory cycle signalD(n). The subroutine begins with initialization of a variable Sum 4620 setto zero, and an index i 4622 set to one. A decision block 4624 checks whether i is greater than M, the buffer length. If not, the current value of stor(i) 4628 is added to Sum 4620, and i 4630 is incremented by one. This loop continues until i exceeds M 4624. Once the loop ends, the subroutine calculates Z(n) 4632, defined as TTI(n) minus the average of the stored values, obtained by dividing Sum 4620 by M 4626.

[0150] Following the calculation of Z(n) 4632, the index i is reset to two 4634. A decision block 4636 checks whether i is greater than M 4626. If not, the stored value at position i-1 is set equal to the stored value at position i 4638, and the index i 4640 is incremented. This loop shifts the stored values for use in subsequent iterations. When the loop completes, the final value stor(M) is set equal to the current TTI(n) 4642.

[0151] Using this updated information, Z(n) 4632 is evaluated. If Z(n) is greaterthan a defined threshold of 0.1 4644, D(n) is set to one 4646. If Z(n) is less than 0.1 4648, D(n) is set to zero 4650. If neither condition is met, D(n) retains the value from the previous cycle, D(n- 1) 4652. The subroutine then ends, providing an updated digital respiratory cycle signal D(n).

[0152] Another implementation of the Respiratory Cycle Generator relies on the use of the accelerometer data to generate the digital representation of the respiratory cycle, D(t). Pseudo code for Subroutine lnitialize_Accel2D as well as Subroutine Run_Accel2D are shown below:

[0153] Subroutine Initialize Accel2D is called once for the initialization of the buffer, and Figure 47A provides a flow chart of Subroutine lnitialize_Accel2D. SubroutineRun Accel2D is called each time a new value of accel(n) is read, and it returns a digital signal, D(t), in the form of a zero or a one, representing the deduced respiratory cycle signal. The process begins at start block 4702. A buffer length M is initialized to 1 ,500 as shown in block 4704, and the index i is initialized to one in block 4706. A decision is made at block4708 to determine whether the index i is greater than M. If i is not greater than M, the storage location stor(i) is set to zero in block 4710, after which the index i is incremented by one in block 4712, and the process loops back to decision block 4708. Once the index i exceeds M, the subroutine proceeds to block 4714, where the digital output D(n) is set to zero. The process then ends at block 4716, completing the initialization of the buffer for subsequent use by the Run_Accel2D subroutine

[0154] Figure 47B contains a flow chart of the Subroutine Run_Accel2D. The process begins at start block 4720. At block 4722, the maximum value of the stored buffer is determined and assigned to MAX. At block 4724, the minimum value of the buffer is determined and assigned to MIN. The range is then calculated as MAX minus MIN in block 4726, and at block4728 the midpoint MID is determined as MIN plus one-half of the range.

[0155] At decision block 4730, the current acceleration value accel(n) is compared to the calculated midpoint MID. If accel(n) is greaterthan MID, the output D(i) is setto zero at block4732. If accel(n) is not greaterthan MID, the output D(i) is setto one at block4734.

[0156] Following this assignment, the process initializes the index variable i to two at block 4736. At decision block 4738, the algorithm checks whether i is greater than M. If i is not greater than M, the contents of the buffer are shifted such that stor(i— 1 ) is set equal tostor(i) at block 4740, and then the index is incremented by one at block 4742, after which the process loops back to block 4738.

[0157] Once the index i exceeds M, the final storage location stor(M) is updated with the new acceleration value accel(n) at block 4744. The subroutine then terminates at end block 4746.

[0158] Acceleration signal 1105 that is going into the Subroutine Run_Accel2D as well as the resultingvalues of MAX 1110, MIN 1120, MID 1130 and D(n) 1140 are shown in Figures 11 A and 11 B for an acceleration trace that was recorded from a human subject.

[0159] Subroutine Run_Accel2D as shown above uses the threshold value that happens to be at the middle of the range from MIN to MAX, i.e., MID = (MAX-MIN) / 2.0. However, it is possible to use different thresholds, by changing the threshold level from exact middle of the MIN to MAX range to a different value, such as THRESHOLD = MIN + K*(MAX-MIN), where K is a real number in the range of 0.0 to 1 .0, such as 0.56. It is also possible to use a threshold value determined by non-linear means, such asTHRESHOLD = median { stor }.Ventilation efficacy detector:

[0160] The ventilation efficacy detector can be embodied, in some examples, as a combination of hardware and software. In some examples, the functionality that is provided by the software may be instead provided by firmware or arranged hardware (e.g., an FPGA or ASIC). Ventilation efficacy can be defined as one or more of the following, including their combinations:

[0161] - Inspiratory duty cycle (IDC),

[0162] - Tidal volume (TV),

[0163] - Peak airflow during inhalation (<PMAX),

[0164] - Minute volume (MV),

[0165] - Tissue oxygen saturation, Sa(O2), and

[0166] - Apnea-Hypopnea Index (AHI).

[0167] Aventilation efficiency value is used as a measure of the quality of breathing of the patient, and the algorithms running on the processor can use this value to adjust one ormore stimulation parameters. In some embodiments, additional performance parameters used for feedback and phase adjustment include cardiac metrics such as heart rate and heart rate variability (HRV), enabling improved synchronization and therapy titration for both OSA and CSA. For example, reductions in HRV during apneic burden may increase stimulation amplitude or advance stimulation phase to stabilize ventilation.

[0168] There are multiple ways that the Ventilation Efficacy can be calculated, and few of them will be described below:

[0169] Inspiratory duty cycle (IDC) can be used for the estimation of ventilation efficacy. IDC is the ratio of inspiration duration to the total respiratory cycle duration, or the respiratory period. It can be estimated usingthe Equation 02 as shown below:

[0170] where TD(t>=i is the duration when the digital signal representing the deduced respiratory cycle signal is one, and

[0171] where TDm=o is the duration when the digital signal representing the deduced respiratory cycle signal is zero.

[0172] For the evaluation of equation 02 above, one can use the D(t) that is produced by the subroutines of the Respiratory Cycle Generator, such as the Run_TTI2D or Run_Accel2D.

[0173] Tidalvolume (TV) can be used forthe estimation of ventilation efficacy. Tidal volume can be estimated as the peak-to-value of the respiration waveform that can be derived from the transthoracic impedance or accelerometer signals. For example, the Subroutine Run_Accel2D captures the peak-to-peak value of the accelerometer signal in the variable RANGE, which in turn can be used as a surrogate forthe tidalvolume. Alternatively, one can use the time integral or the area under the curve to estimate the ventilation efficacy.

[0174] An alternative method for estimating tidal volume using chest position is shown in Figure 12. In this approach, one or more accelerometer signals, such as AccelX 1201 , AccelY 1202, and AccelZ 1203, along with gyroscope signals, such as GyroX 1204, GyroY 1205, and GyroZ 1206 are combined using appropriate scaling coefficients to obtainchest acceleration 1202— CAX 1205, CAY 1210, CAZ 1215, CGX 1220, CGY 1225, and CGZ 1230 — and then integrated 1235 once to derive chest motion 1240 and integrated again 1242 to derive position information 1245. The resulting chest position data 1245 serves as an indicator of ventilation efficacy.

[0175] Figure 37 illustrates the six movements of the human body that can be measured by an implantable 6-axis accelerometer. The orientation of the body relative to the accelerometer axes is shown, where the X axis corresponds to the lateral direction 3705, the Y axis to the anterior-posterior direction 3710, and the Z axis to the inferior-superior direction 3715. Linear acceleration may be detected along each of these axes: (1 ) ACCELX in the lateral direction; (2) ACCELY in the anterior-posterior direction; and (3) ACCELZ in the inferior-superior direction. Rotational acceleration may be detected around these same axes: (4) GYROX (QX) around the lateral axis; (5) GYROY (QY) around the anterior-posterior axis; and (6) GYROZ (QZ) around the inferior-superior axis. Representative signal outputs 3720 from these axes are depicted at the top of Figure 37. These acceleration signals can be combined to produce a single chest acceleration signal, as described in Figure 12.

[0176] In addition to estimating respiratory flow, Figure 15 illustrates a block diagram of the flow detection system 1500 that can estimate peak airflow during inhalation (<t>MAX) using the impedance signalZ(t) 1510. Z(t) 1510 is a processed version of the raw transthoracic impedance signal TTI(t), where the baseline drift has been removed. As the lungs fill with air, Z(t) 1510 increases, and as the patient exhales, Z(t) 1510 decreases, approximating lung volume. The signal Z(t) 1510 is passed through a low-pass filter block 1520, and the resulting signal is differentiated by derivative block 1530, producing an airflow signal <t»(t) 1540. By differentiating Z(t) 1510, changes in lung volume can be captured, which corresponds to airflow <D(t) 1540. A peak detector, not shown in Figure 15, would estimate peak airflow during inhalation (cpMAX), which can be used as a measure of ventilation efficacy. This method of calculating peak airflow and respiratory flow provides valuable data for monitoring respiratory patterns and detecting conditions like apnea, as described further in Figure 53.

[0177] Minute volume (MV) can be used for the estimation of ventilation efficacy.Minute volume is the product of respiration rate (RR) and the tidal volume (TV), as shown in Equation 03 below:Equation 3: MV = RR x TV

[0178] Respiration rate can be determined from the digital signal, D(t) and the methods that can be used for the estimation of tidal volume were listed above.

[0179] Tissue oxygen saturation, Sa(O2), can be used forthe estimation of ventilation efficacy. Oxygen saturation can be determined using an optical oxygen saturation measurement using the absorption or reflectance of the tissue in red and infrared wavelengths of 660 nanometers (red) and 940 nanometers (infrared). Alternatively, an amperometric oxygen sensor using an oxygen permeable membrane that enables a chemical reduction reaction, which in turn produces an electrical signal.

[0180] Apnea-Hypopnea Index (AHI) can also be used forthe estimation of ventilation efficacy. An estimate of AHI can be constructed usingthe digital signal D(t) to determine periods of no breathing, indicated as lack of ones in D(t), or poor breathing where the tidal volume or minute volume are low.Processor:

[0181] A hardware processor in certain examples may be located in the IPG, in the patient programmer, physician programmer or remotely in the cloud. Accordingly, the processing aspects discussed herein may be distributed between different processors — such as a first processor and a second processor, which communicate with one other by using, for example, one or more transceivers. Illustrative algorithms may run on a processor in any one or more of those locations. An example of a processor is a digital computer or a microprocessor with access to volatile and / or non-volatile memory for the storage of the Algorithms as well as the stimulation parameters and the data sets. In some examples, a processor also includes additional features such as watchdog timers and battery status monitors that are used to, for example, govern the overall operation of the system.Algorithms:

[0182] Algorithms that are running on the processor can be used to govern the operation of the device, direct the output circuit to deliver the stimulation at the specific times with the required parameters, and manage the interactions with the external devices, such as the patient programmer, the physician programmer and the cloud devices.Algorithms include those to implement the Method of Treatment, Method of Monitoring, and the Method of Control, examples of which are provided below and include both apnea-type- agnostic algorithms (usable across OSA and CSA) and apnea-type-specific algorithms.

[0183] Methods of treatment algorithms may include: 1 ) Mode Switching (e.g., different stimuli for OSA vs CSA), 2) Synchrony only (e.g., stimulate and sense to keep synchrony), and / or 3) Agnostic (stimulate all the time / some of the time, monitor retrospectively), where the Agnostic mode, along with “therapy pause” and “amplitude adjust,” operate for both OSA, CSA, and MSA without a need to distinguish which sleep disordered breathing is beingtreated.

[0184] Methods of monitoring algorithms may include: 1 ) Beat to beat monitoring and immediate action, 2) Ensemble average of last N beats and react afterwards, and / or 3) Periodic monitoring and periodic response with Periodic functioning as an apnea-type- specific strategy.

[0185] Methods of control algorithms may include: 1 ) Feedback (proportional, nonlinear, PID), 2) Phase locked loop, and / or 3) Phase / delay optimizer, which can incorporate heart rate and HRV as auxiliary feedback signals for phase adjust and gain scheduling to improve performance in OSA, CSA and MSA.

[0186] Before starting the description of the above techniques, an illustrative breath classification subroutine is introduced, as illustrated in Figure 53. Figure 53 shows a 2x2 matrix allowing the classification of breaths, which can be implemented as a subroutine which is described below:

[0187] As shown above in Table 3, classification of OSA and / or CSA may be performed using signals from one or more sensors (e.g., an accelerometer and a transthoracic impedance sensor). Classification of CSA may be performed when the detected acceleration fails to satisfy a first threshold (e.g., it is Less thanAccelTH RESHOLD) and the detected transthoracic impedance fails to satisfy a second threshold (e.g., it is less than TTITHRESHOLD). Classification of OSA may be performed when the detected acceleration satisfies a first threshold (e.g., it is greater than AccelTHRESHOLD) and detected transthoracic impedance fails to satisfy a second threshold (e.g., it is less than TTITHRESHOLD).

[0188] Figure 53 illustrates a logic matrix for determining if there is a disordered breathing event, and if so, which kind. The logic matrix includes whether or not there is an obstruction of airflow 5302 (e.g., the patient is having difficulty breathing), by using transthoracic impedance. An accelerometer also helps to determine if there is respiratory effort 5304, which indicates whether or not the patient is experiencing obstructive sleep apnea or central sleep apnea when there is a lack of airflow. For example, if a patient has no air flow 5302 but does exhibit respiratory effort, the patient is experiencing obstructive sleep apnea 5305. In another example, if a patient has no air flow 5302 but does not exhibit any respiratory effort, the patient is experiencing central sleep apnea 5310.

[0189] When transthoracic impedance shows there is air flow 5302, no type of sleep apnea is occurring, and the patient is breathing normally (NB) 5315. In the case of an accelerometer indicating respiratory effort 5304 while there is air flow, a sensor's integrity has most likely been compromised 5320.

[0190] Now a detailed description of algorithmic methods and embodiments fortheir implementation are provided below:Method of Treatment: Mode Switching:

[0191] In certain example embodiments, different modes of stimulation can be used for the treatment of different forms of sleep apnea, e.g., central sleep apnea (CSA), obstructive sleep apnea (OSA). For example, an illustrative example subroutine may be as follows:

[0192] Efferent stimulation to excite the motor nerves is applied to the target nerve at a fixed rate.

[0193] Afferent stimulation is applied to the target nerve with the intention to excite the sensory fibers, although the stimulation may also capture the motor fibers. Afferent stimulation can be applied in different modes, including asynchronous mode (e.g., ApplyAsynchronousAfferentStim) and synchronous mode (e.g., (ApplySynchronousStim) .

[0194] It should be noted that the Subroutine ApplyAsynchronousAfferentStim may be similar to the Subroutine ApplyEfferentStim, except for the fact that the start time of the first stimulation may be adjusted to bring it into synchrony with the native breathing cycle of the patient. However, asynchronous stimulation mode does not attempt to form asynchrony between the stimulation and the natural respiratory rate. This mode has the advantage of decreasing the amount of sensory data that is used, thus resulting in power savings for an implantable device.

[0195] The can algorithm acts to maintain a synchrony between the stimulation and the natural respiratory cycle. The synchrony requires the stimulation rate to be the same as the respiratory rate. For additional enhancement of the respiratory efficacy, the time delay or the phase shift between the stimulation cycle and the respiratory cycle may be adjusted usingfeedbackthat may include HR and HRV in addition to respiratory metrics.

[0196] Synchronous stimulation will be established in two steps, first by determining the native respiratory rate, and then stimulating at the proper rate, as outlined in the following two subroutines:

[0197] Once the subroutines DetermineNaturalRate and DetermineStimRate are executed, the system would be stimulatingthe patient in synchrony. At this point, it is possible to leave the system alone, and let it stimulate the patient, which can be done using the Subroutine ApplyAsynchronousAfferentStim. It is also possible to monitor the ventilationefficacy, and make adjustments to the stimulation timing or the stimulation amplitude to keep the ventilation efficacy high, which can be done using the Subroutine ApplySynchronousAfferentStim as listed below:

[0198] Adjustment of the stimulation can be done by alteration of the stimulation energy, including adjustment of the stimulation amplitude, adjustment of the stimulation timing, adjustment of the pulse width, adjustment of the frequency, adjustment of the stimulation duration, or any combination thereof. In other words, for example, the parameter (P) to be adjusted to improve the ventilation efficacy could be the stimulation amplitude and / or the time difference (P = AT) / phase shift (P = Acfj) between the stimulation cycle and the respiration cycle with adjustment logic optionally gated by HR / HRV trends.

[0199] Search for the newvalue of the parameter P can be done by any suitable method, including:

[0200] - Exhaustive search: AP e [APMIN , APMAX],

[0201] - Binary search, or

[0202] - Perturbation search, which can be implemented as: PNEW1 «— POLD - 6,Measure VE1 «— VentilationEfficacy

[0203] PNEW2 «— POLD + 6, Measure VE2 «— VentilationEfficacy

[0204] where if VE1 > VE2 , then PNEW ^ PNEWI else PNEW <— PNEW2.Method of Treatment: Synchrony:

[0205] Another option is to run in the synchrony mode, where the device generates and delivers the therapy while sensing to maintain synchrony. In this case, the Algorithm would treat the OSA and CSA without the need to distinguish the breath type, thus being an agnostic-type stimulation, as outlined in the Subroutine SynchronyOnly.Method of Treatment: Agnostic:

[0206] Yet another option is to run in the agnostic mode, where the device generates and delivers a set of therapy while sensing and recordingthe ventilation efficacy. In this case, the Algorithm would adjust the stimulation parameters only periodically and for an ensemble average of breaths, rather than doing it breath by breath, as outlined in the Subroutine Agnostic:_Method of Monitoring: Beatto beat monitoring and immediate action:

[0207] This method of monitoring includes monitoring of the respiratory cycle (e.g., for each breath). The controller may then react to detected decreases in ventilation efficacy in the next respiration cycle. Subroutine ApplySynchronousStim which was provided earlier constitutes an example for this type of monitoring and can be used for OSA and CSA.Method of Monitoring: Ensemble average of last N beats and react afterwards:

[0208] This method of monitoring is implemented such that successive respiratory cycles are monitored, and the controller reacts to the overall value of ventilation efficacy. Subroutine Agnostic, which was provided earlier, constitutes an example for this type of monitoring and another embodiment of an agnostic algorithm.Method of Monitoring: Periodic monitoring and periodic response:

[0209] This method of monitoring is implemented such that only some of the respiratory cycles are monitored and the controller reacts to the changes in ventilation efficacy of the monitored respiratory cycles. This particular implementation has the advantage of reducingthe power consumption associated with the usage of sensors. Subroutine Periodic which is shown below constitutes an example for this type of monitoring and may be configured as an apnea-type-specific control (e.g., CSA-oriented “rate lock”):

[0210] Subroutines outlined above call another subroutine titled AdjustStim. As mentioned before, the adjustment of the stimulation can be done by alteration of the stimulation parameter (P), which could be the stimulation energy, including the stimulation amplitude, pulse width, frequency, stimulation duration, stimulation timing, or any combination thereof. These adjustments can be made using different methods of control, as described herein[and may use HR / HRV as additional feedback to refine phase and amplitude selections for OSA and CSA.

[0211] Figure 52 illustrates a block diagram of a system for respiratory control. A physiologic respiratory drive 5205 interacts with a patient or respiratory pump 5210. The respiratory activity is monitored by a signal monitoring module 5215, which forwards data to a moving window FFT and power calculation block 5220. The resulting calculation is compared 5225 against a target value 5230, which in turn informs a control algorithm 5235.

[0212] The control algorithm 5235 generates commands for a stimulation pattern generator 5240, which produces stimulation signals to drive phrenic nerve stimulation (PNS) 5245. The stimulation then influences the respiratory pump 5210, thereby closingthe feedback loop.Method of Control: Feedback:

[0213] This method of control is outlined in Figure 41 . Briefly, an error signal (E) 4105 that is the difference 4110 between the desired ventilation efficacy 4112 and measured ventilation efficacy is calculated. Plant 4115 converts this value to the stimulation parameter (P) 4120 which is fed into the Stimulator 4125 to excite the nerve tissue 4126 via a cuff electrode 4127. Signals from the Sensors 4130 are used to derive the Measured Ventilation Efficacy 4135, which constitutes the feedback. Plant function can be implemented as any one or more of any feedback type controller including the proportional controller, non-linear controller and proportional integral differential (PID) controller.Method of Control: Phase locked loop:

[0214] This method of control is outlined in Figure 42. Briefly, the digital signal D(t) 4202 is fed into a T flip-flop 4204 to generate a square wave with half the frequency of the respiratory rate. Similarly, the output of the voltage-controlled oscillator (VCO) 4212 is also fed into another T flip-flop 4206 to generate a secondary square wave. The two square waves are compared using an exclusive-or (XOR) gate 4208, which produces a logic 1 signal only when the two inputs (the square wave derived from D(t) and the square wave derived from the VCO output) are not the same. Integration of the output of the XOR is performed by the integrator 4210, producing a signal that is proportional to the phase shift between the D(t) signal 4202 and the VCO 4212. This phase shift signal is then compared with the desired phase shift 4215 at the comparator / summation block 4214, and the resulting error signaldrives the VCO 4212, which in turn generates the input to the Stimulator 4216 to maintain phase alignment in both OSA and CSA.Method of Control: Phase optimization:

[0215] This method of control was described earlier for a generic parameter (P) to be adjusted to improve the ventilation efficacy, which in this case is the time difference (AT) or the phase shift (Ac|)) that exists between the stimulation cycle and the respiration cycle. Search for the optimum phase shift can be done by any suitable method, including: Exhaustive search: Ac[) e [A MIN , Ac|)MAX], Binary search, or Perturbation search, optionally biased by HR / HRV-derived arousal risk.

[0216] An exemplary list of the subroutines that can be used for the implementation of methods of treatment include Mode switching (Example: Subroutine ModeSwitching), Synchrony only (Example: Subroutine SynchronyOnly), and / or Agnostic (Example: Subroutine Agnostic).

[0217] An exemplary list of the subroutines that can be used for the implementation of methods of monitoring include 1 ) Beat to beat monitoring and immediate action (Subroutine ApplySynchronousStim), 2) Ensemble average of last N beats and react afterwards (Example: Subroutine Agnostic), and / or 3) Periodic monitoring and periodic response (Example: Subroutine Periodic).

[0218] Determination of when treatment of sleep apnea is turned on or off can be controlled by different methods, with each method given a specific tier of weight to prevent methods from overriding each other. In this embodiment, the on-off algorithm featuring a tiered on / off mechanism has the following tiers:

[0219] Manual on / off, where the patient can turn the stimulation off via a manual input.

[0220] A location-specific transmitter with Bluetooth / NFC / RF to turn on or off the device when in a set proximity for a set amount of time. For example, if the patient were to get up in the middle of the night to use the bathroom after X number of minutes has passed in the bathroom or out of the bed, stimulation ceases.

[0221] 3. A state-machine algorithm including multiple weighted parameters such as the time of day, pitch of the patient, activity level of the patient, apnea detection, and any device-specific safety checks.

[0222] Since the manual on / off switch is in the first tier, if the patient is in a location where the location-specific transmitter transmits 'ON' for stimulation, but the patient has manually input for stimulation to be 'OFF,' stimulation would not occur. For example, if a patient were in bed reading for an X number of minutes where the location-specific transmitter reads ’ON,1but the patient is not ready to sleep, the patient may manually set stimulation to be ’OFF.’

[0223] A location-specific transmitter, such as a near-field communication (NFC) mat which the patient places where they typically lie, may be used as an input into the on / off algorithm. With specific weighting, this or any subset of variables may become the sole defining variable to decide on / off behavior.

[0224] State-machine algorithms can have multiple weighted parameters that may be patient controllable. For example, the patient may be able to control the weighted parameters via a sensitivity slider, where they can lower or raise the overall threshold for when the device will be on. The patient may also have a time-delay slider for the duration that that threshold must be maintained / surpassed before the device turns on. In addition to being changed by the patient manually, the weighted parameters may be programmed manually or sensed on how a parameter can predict that a patient is sleeping. For example, the patient’s bedtime parameter could be programmed manually where a set time of day begins stimulation by the usual bedtime taken over each day of the week for the past month, while also sensing the variability of the patient's bedtime to raise and lower the timing weight.

[0225] Figure 51 A illustrates a two-factor graph 5100 representing a patient’s pitch 5105 and bedtime 5110 that can be used to weight parameters. Figure 51 A shows data taken from a patient’s sleep pattern, where circles 5115 represent when the patient is awake and X’s 5120 represent when the patient is asleep. A weighted linear classifier 5125 is shown as a dashed line, which separates the two categories. In this example, a weighted linear classifieris used for n predictors (where n is 2), and the classifier function may be represented as an (n-1) dimensional hyperplane. This hyperplane 5125 indicates a threshold in which the parameters would need to be met for stimulation to occur. That is, all variations of the pitch and bedtime parameters falling on the right side of the hyperplane 5130 would cause stimulation, and all variations on the left side of the hyperplane 5135 would result in stimulation being off.

[0226] Figure 51 B illustrates another embodiment of the state-machine algorithm to weigh different identified parameters. The graph represents patient pitch 5105 versus bedtime 5110, with circles 5115 showing when the patient is awake and X’s 5120 showing when the patient is asleep. A classifier hyperplane 5125 separates the two categories, and in this embodiment, the weighting of parameters shifts the classifier. Weights may be determined by the variability of a given parameter, with a higher variability indicating a lower weight due to less reliability of that predictor, and a lower variability indicating a higher weight. For example, if a patient has a highly variable bedtime, the weighting of the bedtime parameter decreases, shifting the classifier line 5125 to the right, as shown by the adjusted hyperplane 5145. The shift direction is indicated by arrow 5150.

[0227] A higherweight moves the classifier further to the right, reducingthe likelihood of stimulation. In this embodiment, if the patient adjusts a sensitivity slider, the hyperplane 5125, 5145 may move in all n dimensions, resulting in stimulation being less likely to be on. Furthermore, if a few factors account for the majority of variability (as per the Pareto principle), parameters with higher variability may be removed from sensing and / or calculations to save battery life.

[0228] Figure 51 C illustrates a different embodiment of the weighted state-machine algorithm, where the algorithm organizes parameters by different classifier equations. The graphs again show pitch 5105 plotted against bedtime 5110, with circles 5115 representing awake states and X’s 5120 representing asleep states. Due to the variability of parameters, data can be taken and organized under different classifiers to more accurately predict when a patient is sleeping. For example, a classifier hyperplane 5125 may be organized by day, such that each day has a different classifier equation 5175 due to different bedtimes,different sleeping spots or positions, or other variability. For instance, as shown in the Thursday graph 5165, the classifier hyperplane 5125 separates the awake and asleep states at an earlier bedtime. In contrast, as shown in the Friday graph 5170, the classifier hyperplane 5125 has shifted left to reflect later bedtimes, which may occur due to patients staying up later or having more variable sleep schedules. Flexible start and stop times, as well as suspension of therapy, may be applied to patients with insomnia or poor sleep hygiene to improve treatment outcomes.

[0229] The classifier values may be manually set and updated, or may be calculated via statistical or machine learning methods in a cloud or external hardware setting. Such methods may include weighted linear discriminant analysis, Adaboost, logistic regression, support vector machines, or Bayesian estimators. Once the hyperplane equations 5125, are defined off-device, they can be uploaded to the implanted device for local computation of that night’s stimulation turn-on time.

[0230] Atime-delay slider (sleep latency) may be implemented, which delays device turn-on until a set time after the classifier algorithm determines appropriate conditions for sleep onset and thus stimulation onset. Another metric for sensitivity may replace a manual delay with an automatically adjusting time delay based on a consistency metric, in which the appropriate “stim-on” condition must be detected a certain proportion of the time over a defined period. For example, the stim-on condition may need to be satisfied 80% of the time over a 10-minute window before stimulation is initiated.Stimulation lead:

[0231] Stimulation lead of the system provides the electrical connection between the IPG and the cuff electrode that is placed around the target nerve. Lead would have one or more conductors, electrically isolated from each other and from the surrounding tissue. In the case where there is more than one conductor, the conductors within the lead could be laid in a multi-lumen configuration, with a concentric placement, or multi-conductors in parallel spiral pattern. In certain embodiments, a single implanted lead delivers therapy effective for both OSA and CSA, with algorithmic control selecting apnea-agnostic or apneaspecific modes as needed.Cuff electrode:

[0232] The cuff electrode is designed to be placed around the target nerve and confine electrical stimulation to a narrow region of interest. As shown in Figures 7A and 7B, the cuff electrode includes a cuff body 710, which includes a top piece 720 and a bottom piece 730, hinged together to form an oblate ellipsoidal lumen 740 that accommodates the nerve, ensuring proper placement even in cases of nerve swelling. The top and bottom pieces are shaped such that the side parallel to the major axis of the lumen is flat, while the side perpendicular is curved, forming an oblate ellipsoid when joined. The bottom piece 730’s longer curved side overlaps the top piece 720 to create a tunnel for the nerve to pass through, which can decrease nerve compression and corresponding patient discomfort.

[0233] The cuff may contain one or more electrical contacts to deliver stimulation in either a unipolar or bipolar configuration. In the unipolar setup, stimulation occurs between the electrode and the implanted device's case, while in the bipolar configuration, stimulation is delivered between the cuff’s contacts. As illustrated in Figure 07, the cuff electrode can deliver the stimulation to the target nerve through these configurations.Electrodes 750 are embedded into the bottom piece 730 to orient them along the bottom of the nerve, where the phrenic nerve naturally rests, reducing the nerve capture threshold. This configuration prevents the top piece 720 from obstructing nerve stimulation if it folds inside the cuff. The flat external sides 760 of the cuff body 710 ensure it fits securely within the surrounding anatomy, lying against the anterior scalene muscle and the posterior cervical fat pad and enabling a single-lead therapy approach for OSA and CSA.

[0234] In addition to delivering electrical stimuli, the cuff electrode can be used for sensing evoked potentials from the nerve resulting from electrical stimulation. The cuff may also feature passive or active fixation methods for immobilization, such as spring-loaded barbs or suture holes 770. The top and bottom pieces may house sensors 785 and 790 for monitoring nerve activity and other physiological parameters, such as electrocardiogram (ECG), electromyogram (EMG), inertial sensors, temperature, pressure, and ultrasonic sensors. These sensors allow the cuff to detect afferent traffic, monitor muscle activity, and ensure proper nerve function without interference. The lead connector of the cuff can beperformed in various shapes to stabilize it around the nerve's anatomy. While this example specifically targets the phrenic nerve, the cuff electrode may also be used to stimulate other nerves, such as the vagus or hypoglossal nerves, for applications like treating sleep disorders.Power supply:

[0235] Power supply of the system is contained with the IPG, and it can be a primary battery which is not rechargeable, or chargeable type or a combination of either. It can also be a type that generates electrical power from the motion of the body of the patient and stores it on a rechargeable battery or capacitor, such as a super capacitor.Telemetry:

[0236] Telemetry allows the implementation of the actual communications between the IPG, the patient programmer and physician programmer. Forms of telemetry can be magnetic only, electromagnetic such as Bluetooth, ultrasonic or optical, such as infrared. Illustrative Examples:

[0237] To demonstrate the effectiveness of the above-described methods, a set of experiments were carried out with human subjects, which are described below:

[0238] Experimental Setup: All patients signed informed consent prior to any study specific activity. Sensors recording physiologic signals were connected to patients undergoing routine drug induced sleep endoscopy (DISE). Physiological signals monitored included respiratory effort as measured via an abdominal and thoracic belt, airflow as measured by a pneumotach, intrathoracic pressure measured by a Millar pressure catheter located proximal to the epiglottis, oxygenation data from an oximeter, and the delivered stimulation. Parameters such as tidal volume were calculated as the integral of the flow channel. Additionally, data was collected from an accelerometer including motion in the x, y, and z direction as well as gyroscopic measurements (rotational speed around an axis) for the x, y, and z axis to further measure respiratory effort and transthoracic impedance to measure tidalvolume.

[0239] Figure 17 shows simultaneously recorded traces of the thoracic belt 1705, abdominal belt 1710, electrical stimulation 1715, air flow 1720, and pulmonic pressure 1725from a patientwho is sufferingfrom obstructive sleep apnea. When there is no stimulation 1730, chest and abdominal movement show reduced amplitude, correlating with limited airflow and low pulmonic pressure, indicative of partial airway obstruction commonly seen during an OSA event. It can be seen that once stimulation is turned on again, there is a marked increase in thoracic and abdominal movement, accompanied by a corresponding rise in airflow and pulmonic pressure, indicating that the stimulation effectively enhances respiratory effort and restores more efficient ventilation, likely by overcoming airway obstruction. This indicates that the stimulation is effectively reducing airway obstruction, possibly by activating muscles to maintain airway patency, such as through phrenic nerve stimulation. The adaptive stimulation method is designed to prevent apneic episodes by enhancing respiratory effort, thereby improving airflow and ventilation in OSA patients. This approach can be integrated into advanced therapeutic devices aimed at reducing OSA severity by dynamically adjusting stimulation based on respiratory signals.

[0240] Figure 18 shows simultaneously recorded traces of thoracic belt 1805, abdominal belt 1810, electrical stimulation 1815, respiratory air flow 1820, intrathoracic pressure 1825, peripheral oxygen saturation 1830 and tidal volume 1835 from a patientwho was receiving electrical stimulation where there was entrainment between the stimulation and the respiratory cycle which was keeping the airway open prior to the termination of the therapy.

[0241] On the left-hand side 1840, the patient is being stimulated with a fixed rate stimulation train. There is one stimulation per breath and the stimulation is delivered in the optimal respiratory phase. Because the stimulation rate is fixed, and the patient is breathing with the stimulation, the patient is said to be entrained. The patient’s tidal volumes 1835 (between 4-500mL) are associated with normal breaths. Once stimulation ceases, tidal volume 1835 drops to < 100mL despite the subject still exerting effort to breath as seen on the Abdominal 1805 and Intrathoracic traces 1810. While oxygenation is stable, shortly after the patient goes into the apnea, oxygenation begins to decrease. This demonstrates that stimulation delivered at a patient specific rate and specific respiratory phase maintainsairway patency thus allowing normal breathing and avoiding a respiratory event as shown when stimulation is stopped.

[0242] Figure 19 shows simultaneously recorded traces of thoracic belt 1905, abdominal belt 1910, electrical stimulation 1915, respiratory air flow 1920, intrathoracic air pressure 1925, peripheral oxygen saturation 1930 and tidal volume 1935 from a patient who was having OSA before the onset of electrical stimulation where there was entrainment between the electrical stimulation and the respiration.

[0243] On the left-hand side 1940, the patient is having an obstructive apnea as shown by the lack of airflow despite effort as shown in the abdominal 1905 and thoracic belts 1910 as well as intrathoracic pressure traces. Stimulation is initiated at a fixed rate specific to this patient. In response to stimulation the apnea is halted, and the patient’s tidal volumes 1935 and oxygenation 1930 increases. While the patient is still flow limited the tidal volumes 1935 have increased to 400mL per breath and are approaching normal tidal volumes 1935. There is one stimulation per breath and the stimulation is delivered in the optimal respiratory phase. Because the stimulation rate is fixed, and the patient is breathing with the stimulation, the patient is said to be entrained. This demonstrates that stimulation delivered at a patient specific rate and a specific respiratory phase can halt an apnea and both restore and maintain airway patency thus restoring flow.

[0244] Figure 20 shows simultaneously recorded traces of thoracic belt 2005, abdominal belt 2010, electrical stimulation 2015, respiratory air flow 2020, pulmonary pressure 2025, peripheral oxygen saturation 2030 and tidalvolume 2035 from a patient who was receiving electrical stimulation where there was entrainment which was keeping the airway open prior to the loss of entrainment.

[0245] On the left-hand side 2040, the patient is being stimulated with a fixed rate stimulation train. There is one stimulation per breath and the stimulation is delivered in the optimal respiratory phase. Because the stimulation rate is fixed, and the patient is breathing with the stimulation, the patient is said to be entrained. The patient’s tidalvolumes are associated with normal breaths.

[0246] After the 11thbreath 1945 there is a respiratory disturbance. Following this disturbance the morphology of the breath changes and becomes more variable and the timing of the stimulation relative to the breath becomes more variable. This respiratory variability indicates that entrainment was lost due to the respiratory disturbance. Once entrainment is lost, airway patency decreases as evidenced by the increase in depth of the intrathoracic pressure trace and the decrease in tidal volume 2035 post which occurs post entrainment loss. This demonstrates that entrainment promoted respiratory stability, airway patency, and increased tidal volumes 2035 and once entrainment was lost, respiratory instability occurred with decreased airway patency and tidal volumes 2035.

[0247] Figure 20 shows simultaneously recorded traces of thoracic belt 2005, abdominal belt 2010, electrical stimulation 2015, respiratory air flow 2020, pulmonary pressure 2025, peripheral oxygen saturation 2030 and tidalvolume 2035 from a patient who was receiving electrical stimulation where there was entrainment between the electrical stimulation and the respiratory cycle and the entrainment was keepingthe airway open prior to the loss of entrainment.

[0248] Figure 21 shows simultaneously recorded traces of thoracic belt 2105, abdominal belt 2110, electrical stimulation 2115, respiratory air flow 2120, pulmonary pressure 2125, peripheral oxygen saturation 2130 and tidalvolume 2135 from a patient who was receiving electrical stimulation where stimulation rate was being adjusted to demonstrate effect of entrainment between the stimulation 2115 and the respiratory cycle.

[0249] Figure 21 can be divided into two main parts. The first third 2140, left-hand side, includes delivery stimulation that is not at the patient specific rate for entertainment. There is no 1 :1 correlation between stimulation and breathing. Additionally, the onset of stimulation can occur at different respiratory phases as shown by the wide dispersion of phase duringthe period where the patient is not entrained. Duringthis period, a few breaths exhibit an increase in tidal volume 2135, and airway patency as shown by the decreased depth of the intrathoracic pressure efforts thus indicating that airway patency has been temporarily restored. For these breaths, the onset of stimulation occurs within the optimal period. For other breaths there is decreased flow despite increased respiratory effort. Thephase trace shows that the onset of stimulation occurred after the onset of stimulation, defined as phase = 0 and was not duringthe optimal time period.

[0250] When the patient specific rate of 17 breaths per minute was reached, and the stimulation occurred at the optimal respiratory phase (onset of inspiration) four physiological responses are seen in the traces. First, the flow 2120 waveform becomes much less flow limited and more organized. Second, a rise in the thoracic belt 2110 indicates that lung volume has increased. Third, the intrathoracic pressure 2125 trace shows that the patient is generating less negative pressure to create larger amounts of airflow thus indicating that patency has been restored to the airway. Finally, all the above results in increased tidal volumes 2135. In the phase trace, one can see that the onset of stimulation occurs immediately priorto the onset of inspiration.

[0251] Finally, the right third 2145 of the graph shows there was respiratory disturbance which caused the patient’s entrainment to move to a different, less optimal respiratory phase. The stimulation train is still 1 :1 and at the patient specific rate however onset of stimulation now occurs during mid to late inspiration ratherthan immediately prior to onset of inspiration. This non-optimal phase results in misentrainment and is evidenced by the decrease in flow, decrease in lung volume (as shown by the thoracic belt 2105, the dips on the intrathoracic pressure trace 2125 become deeper despite a lower tidal volume 2135 this indicating that the airway tone has decreased. In the phase trace, one can see that the onset of stimulation occurs in mid to late inspiration. The comparison of entrainment and misentrainment demonstrates that not only is rate important in triggering the optimal response from the body but that respiratory phase is also a necessary component.

[0252] Figure 22A shows simultaneously recorded traces of thoracic belt 2205, abdominal belt 2210, electrical stimulation 2215, respiratory air flow 2220, intrathoracic pressure 2225, peripheral oxygen saturation 2230 and tidalvolume 2235 from a patient who was receiving electrical stimulation where phase shift between the stimulation 2215 and respiratory cycle 2220 was being adjusted to demonstrate effect of the phase shift on ventilation efficacy.

[0253] Figure 22B illustrates the effect of shifting the onset of stimulation 2215 to correspond with different respiratory phases. For the first 42240 breaths the onset of stimulation occurs either too early or too late relative to the onset of inspiration. The following three breaths 2245 show an increase in tidal volume and flow 2220 due to more optimal stimulation timing, however the respiratory phase is still not optimal as it is slightly early. The final four breaths 2250 demonstrate increased tidal volume 2220 and normal flow as the respiratory phase was further adjusted to increase the physiological response. This demonstrates that adjusting the respiratory phase to a specific point in the respiratory cycle, usually immediately precedingthe onset of inspiration, can increase the physiological response resulting in increased airway patency, air flow, and tidal volume.

[0254] Figure 23 shows simultaneously recorded traces of abdominal belt 2305, thoracic belt 2310, electrical stimulation 2315, respiratory airflow 2320, peripheral oxygen saturation 2325 and tidal volume 2330 from a patient who was receiving electrical stimulation 2325 with entrainment between the stimulation and respiratory cycle prior to the termination of the therapy which in turn allowed the central sleep apnea to surface.

[0255] Figure 23 demonstrates the ability of phrenic nerve stimulation to stabilize respiration. While stimulation 2325 is active, the patient is breathing and has stable oxygenation 2315. Once stimulation 2325 is stopped 2335, the patient immediately goes into a series of central apneas where there is little respiratory effort, no flow, thus no tidal volume 2330, and the patient experiences oxygen desaturation, which is an undesirable medical condition and should be avoided. This demonstrates that in addition to treating obstructive sleep apnea, as shown in the prior figures, phrenic nerve stimulation can treat central sleep apnea and stabilize both oxygenation and respiration.

[0256] Figure 48 illustrates a system for detecting and classifying respiratory events through the use of a combination of sensors, includingTransThoracic Impedance (TTI) 4802 and an accelerometer peak vectors 4804. The system processes waveform data from both sensors to generate a predicted inspiration gate signal 4806 that classifies the respiratory cycle into inspiration and expiration phases. Inspiration is identified as phase 4820 of the waveform, and expiration is identified as phase 4822 of the waveform.

[0257] At the top of the figure, the TTI waveform 4802 shows the variation in thoracic impedance over time, with distinct patterns corresponding to a regular breath 4808, obstructive apnea 4810, central apnea 4812, mixed apnea 4814, and hypopnea 4816. The waveform amplitude changes represent the transition between inspiration and expiration, where the start of expiration is marked by the end of the TTI waveform.

[0258] Beneath the TTI waveform, the accelerometer peak vector waveform 4804 represents chest or body movements related to breathing. The accelerometer signal provides additional data to determine the start of expiration, represented by the middle of the waveform's descent. This waveform is used in combination with the TTI waveform 4802 to ensure accurate classification of each respiratory event, correlating body movement with thoracic impedance.

[0259] Below the accelerometer waveform, the predicted inspiration gate 4806 is shown as a binary signal that simplifies the waveforms into discrete phases of inspiration (a high gate signal) and expiration (a low gate signal). The gate signal 4806 is derived by processing the waveform data from both sensors, allowing for breath-to-breath comparison and minimizing discrepancies between TTI and accelerometer data.

[0260] The system employs a slidingwindow method 4818, analyzing a period of 50 seconds to classify each breath as either normal, OSA, CSA, mixed apnea, or hypopnea. Central events are identified by periods of no airflow exceeding 10 seconds, combined with subsequent obstructive apnea events, where CSA accounts for less than 80% of the window. These criteria allow for reclassification of events, such as distinguishing mixed apnea from pure obstructive or central events.

[0261] Classification of OSA or CSA occurs at the end of expiration by comparingthe TTI waveform 4802 and accelerometer waveform 4804. If both signals are below their respective thresholds, the system classifies the event as apnea. Normal breathing is identified when the TTI amplitude falls within a typical range, while deviations above this range indicate hypopnea 4816, and deviations belowthis range indicate apnea or flow limitation.

[0262] This dual-sensor system provides enhanced accuracy in detecting and classifying respiratory events during sleep. The combination of TTI 4802 and accelerometer 4804 signals ensures subtle differences in breathing patterns are captured, enabling detailed classification of respiratory events in real time. This capability makes the system suitable for diagnostic and therapeutic devices aimed attreating sleep disorders, including obstructive sleep apnea and central sleep apnea.

[0263] Figure 49 shows the relationships between the abdominal belt 4902, thoracic belt 4904, stimulation 4906, flow 4908, and tidal volume 4910. Abdominal and thoracic belts are commonly utilized in respiratory muscle training (RMT) to target specific muscle groups involved in breathing. The abdominal belt focuses on the diaphragm and abdominal muscles, while the thoracic belt targets the intercostal muscles and chest expansion. Stimulation methods, such as respiratory muscle trainers or electrical stimulation, play a crucial role in activating and training respiratory muscles. These devices provide resistance or assistance to breathing, depending on the desired training effect. Through training, respiration flow, or the rate of air movement in and out of the lungs, can be influenced, potentially leading to smoother and more efficient breathing patterns. Additionally, tidal volume, the volume of air inspired or expired with each breath during normal breathing, can be impacted by respiratory muscle training. By enhancin the strength, endurance, and coordination of respiratory muscles, individuals may experience improvements in breathing efficiency, respiratory function, and overall respiratory performance.

[0264] The circled region 4912 highlights a portion of the trace where stimulation is ceased. In this period, the respiration cycle becomes less stable, with irregular flow and reduced tidal volume, demonstrating how stimulation withdrawal negatively impacts breathing quality.

[0265] Figure 54A-54B shows the flowchart of the adaptive increase of therapy dose at the onset of sleep. Dose is increased stepwise, and if it causes arousal, then the step size is progressively reduced to allow reaching to the therapeutic dose without arousing the patient. Pseudo code for Subroutine AdaptivelncreaseTherapyDose is shown below:

[0266] Figures 54A and 54B illustrate an adaptive therapy dosing algorithm. Figure 54A shows an exemplary stimulation amplitude profile over time. At therapy onset 5402, the amplitude is initially increased in relatively large steps 5404 to quickly achieve a therapeutic effect. Once an arousal is detected 5406, the algorithm transitions to smaller incremental steps 5408, allowingfiner control and reducingthe likelihood of overstimulation. This adaptive increase provides a balance between rapid onset of efficacy and patient comfort.

[0267] Figure 54B provides a flowchart representation of the algorithm shown in Figure 54A. The process begins by defining a target dose 5410 and calculating a step value 5412 based on the target dose. The amplitude is initialized 5414 relative to the start dose and incrementally increased by the step amount 5416. A counter variable / 5418 is set, and the algorithm proceeds in cycles. If the value of counter i is not greater than 105420, therapy is delivered 5424, followed by a delay period 5426, after which the counter is incremented 5428. If an arousal is detected 5430, the step size is reduced 5432, ensuring more gradual increases in stimulation. If there is no arousal detect, the algorithm loops back to checking if i is greater than 105420. If i is greater than 10, the amplitude is checked to see if the amplitude is greater than or equal to the target dose 5422. If the amplitude is larger than or equal to the target dose, the algorithm is ended. If the amplitude is not larger than or equal to the target dose, the amplitude is increased by a step size 5416.

[0268] Figures 55A-55B show an embodiment of stimulation therapy where stimulus evokes respiratory flow. When the phrenic nerve is stimulated, it sends signals to the diaphragm muscle, instructing it to contract. This contraction causes the diaphragm to flatten and move downward, increasingthe volume of the thoracic cavity. As the diaphragm contracts and moves downward, it expands the thoracic cavity, which lowers the pressure within the lungs. This decrease in pressure relative to the atmospheric pressure outside the body causes air to flow into the lungs, leading to inhalation or inspiration. After inhalation, the phrenic nerve activity decreases, allowing the diaphragm to relax. As the diaphragm relaxes, it moves back up into its dome shape, reducing the volume of the thoracic cavity. This increase in pressure within the lungs relative to the atmosphere causes air to flow out of the lungs, resulting in exhalation or expiration, which is shown in Figure 55B.

[0269] Figure 55A shows the stimulation pattern that causes a patient to inhale and exhale, thus producing respiratory flow. In this embodiment, stimulation causes one oscillation of the respiratory cycle (e.g., one inhale and one exhale). The amplitude of the stimulation may be increased depending on the patient’s response, as well as the stimulation’s pulse width and frequency, where pulse width may increase the duration of the patient’s respiratory cycle and frequency increase would result in more breath being taken by the patient.

[0270] Figures 55A-55B show an embodiment of stimulation therapy where stimulation pulses 5502 evoke respiratory flow 5504. When the phrenic nerve is stimulated by pulses 5506, it sends signals to the diaphragm muscle, instructing it to contract. This contraction causes the diaphragm to flatten and move downward, increasing the volume of the thoracic cavity. As the diaphragm contracts and moves downward, it expands the thoracic cavity, which lowers the pressure within the lungs. This decrease in pressure relative to the atmospheric pressure outside the body causes air to flow into the lungs, leading to inhalation or inspiration. After inhalation, the phrenic nerve activity decreases, allowing the diaphragm to relax. As the diaphragm relaxes, it moves back up into its dome shape, reducingthe volume of the thoracic cavity. This increase in pressure within the lungsrelative to the atmosphere causes air to flow out of the lungs, resulting in exhalation or expiration, which is shown in Figure 55B by the respiratory flow waveform 5504.

[0271] Figure 55A shows the stimulation pattern 5502 defined by pulses 5506, which causes a patient to inhale and exhale, thus producing respiratory flow 5504. In this embodiment, stimulation pulses 5506 cause one oscillation of the respiratory cycle (e.g., one inhale and one exhale) 5508. The amplitude of the stimulation may be increased depending on the patient’s response, as well as the stimulation’s pulse width and frequency, where pulse width may increase the duration of the patient’s respiratory cycle and frequency increase would result in more breaths being taken by the patient.

[0272] Figures 56A-56B shows another embodiment of stimulation, which causes an echo of respiratory flow 5603 after stimulation is ceased 5605. In some cases, especially if the stimulation is strong or sustained, it can trigger a series of respiratory cycles 5603 rather than just a single breath. This happens because the reflex arc involving the phrenic nerve can become self-sustaining, leading to repetitive contractions of the diaphragm and subsequent breathing cycles without the need for additional stimulation.

[0273] This phenomenon is similar to how certain reflexes, like the knee-jerk reflex, can cause repetitive muscle contractions in response to a stimulus. In the case of the phrenic nerve, the repetitive respiratory cycles can continue until the stimulus is removed, other inhibitory mechanisms come into play, or the respiratory centers regain control over the breathing pattern.

[0274] Figure 56A shows the stimulation pattern 5602 applied to the phrenic nerve. Figure 56B shows the respiratory flow 5604, where after five respiratory cycles 5601 stimulation is ceased 5605. However, due to the reflex arc involving the phrenic nerve becoming self-sustaining, two additional breaths 5607 are taken despite no stimulation occurring. After the stimulus has been removed for an amount of time and the respiratory center regains control, the respiration cycles cease 5605 where there is no stimulation. Once stimulation is then applied again to the phrenic nerve, respiration begins again 5608.

[0275] The stimulation pattern 5602 from Figure 56A shows multiple burst stimulations in the beginning of treatment, thereby producing respiratory cycles with eachstimulation. When stimulation ceases, it can be seen that respiratory cycles still continue without stimulation for a duration. Once respiratory cycles missing 2608 occur again stimulation continues and the process repeats until treatment is terminated.Additional Subroutines Used for System Operation:

[0276] Figure 25 illustrates a flowchart of an algorithm used to adjust nerve stimulation parameters in a device aimed at treating sleep disorders, such as sleep apnea, through nerve stimulation. The process begins by initializing stimulation parameters from the Clinician Programmer 2505, which sets the maximum amplitude (MaxAmp), minimum amplitude (MinAmp), and the current amplitude (Amp) of the implanted pulse generator (IPG). These values are then transferred to the Patient Programmer 2510, where the patient can view and adjust the stimulation amplitude within the predefined range.

[0277] The Patient Programmer Display 2515 shows the current amplitude value, allowing the patient to adjust the stimulation level. The algorithm continuously checks whether the patient is adjusting the amplitude 2520. If no adjustments are made, then the process stops 2522. However, if the patient decides to adjust the amplitude, the algorithm checks whether the patient is attemptingto increase or decrease the stimulation.

[0278] If the patient is increasing the amplitude 2525, the algorithm verifies whether the current amplitude is less than the maximum allowed value (MaxAmp) 2530. If this condition is true, the amplitude is incremented by a small value (e.g., 0.05 units) 2535 to increase the nerve stimulation intensity. The new amplitude is then updated in both the IPG and the patient programmer 2540.

[0279] On the other hand, if the patient is decreasingthe amplitude, the system checks if the current amplitude is above the minimum allowed value (MinAmp) 2545. If so, the algorithm reduces the amplitude by the same small value (0.05 units) 2550, and the updated value is saved in both the IPG and the patient programmer.

[0280] This process allows for fine-tuned adjustments of the nerve stimulation amplitude, ensuring that the patient remains within the safe therapeutic range set by the clinician. The continuous feedback loop between the patient and the device ensures optimal and personalized nerve stimulation therapy for managing sleep disorders.

[0281] Figure 26 includes an Upright Stimulation Pause feature designed to provide automatic cessation of nerve stimulation therapy when a patient temporarily rises during the therapy period and automatic resumption once the patient lies down again. This feature ensures seamless operation by detectingthe patient's position and adjusting stimulation accordingly, without requiring manual intervention. The system works by automatically pausing stimulation when the patient’s position is detected as Upright and resuming stimulation when the patient returns to a Recumbent position.

[0282] The algorithm for this feature defines two key positional states: Upright and Recumbent 2605, with stimulation enabled only when the patient is in the Recumbent state. A Recumbent Delay Time parameter is also introduced 2610, which specifies a brief delay between when the patient returns to a recumbent position and when the therapy resumes. During the therapy session, the system continuously monitors the patient's position and therapy begins 2620 once the patient’s position is recumbent 2615. If the position shifts to Upright 2625, the system automatically pauses the therapy 2630. Once the patient lies down again 2635, entering the Recumbent position, the Recumbent Delay Time is initiated 2640. When this delay time expires 2645, the stimulation automatically resumes 2620, ensuring that therapy is not interrupted for brief periods of patient movement during the night.

[0283] The system assumes that the patient will start therapy when beginning their sleep period. In addition to the basic operation, potential modifications to this feature include an automatic enabling of therapy based on a predefined sleep period. In this scenario, stimulation would start automatically when the system detects that the patient has entered a recumbent position, streamliningthe entire process and further reducingthe need for patient interaction with the device during normal use. This feature enhances patient comfort by preventing unnecessary stimulation when the patient is temporarily upright and ensuring therapy resumes effectively when appropriate.

[0284] Figure 27 illustrates a flowchart detailingthe algorithm forthe Roll Stimulation Pause feature. The process begins with the definition of each recumbent position, which could be supine, left, right, or prone 2705 and a duration of Roll Pause T 2710. The system then checks the patient’s current position 2715 and sets that position as the patient’scurrent position 2720. The algorithm then monitors the patient's position 2730 while therapy is enabled 2715. If the algorithm detects a change in the recumbent position, it initiates a temporary cessation of stimulation 2735, pausingfor a predetermined Roll Pause Time (Pause T) 2740. Once this pause time elapses 2745, the stimulation automatically resumes 2725. During this paused period, the algorithm may also trigger the collection of baseline respiratory data to establish new signal interpretation bounds in the patient's updated position.

[0285] The supine, right, left, and prone position can be detected by accelerometers, gyroscopes, inertial measurement units, and pressure sensors. Accelerometers and gyroscopes can be built into the cuff electrode, or on a detachable device forthe patient to wear that allows communication with the controller.

[0286] Figure 28 illustrates a flowchart detailingthe Baseline Signal Collection algorithm. In some examples, this may be used to support the treatment of OSA and / or CSA by establishing baseline respiratory parameters when stimulation is temporarily paused. The algorithm operates by collecting data on respiratory metrics — e.g., inspiration time, expiration time, and / or full tidal volume — during periods without stimulation. This can be used to generate a clear baseline for subsequent therapy adjustments.

[0287] The Baseline Signal Collection algorithm begins by accessing the Respiration Statistics Database 2805 and defining the parameters for valid inspiration time, expiration time, and full tidal volume 2810. The number of respiratory cycles 2815 to be used for averaging is set, and the system then monitors and filters the respiratory signal 2820. As cycles are detected, data are added to the database 2825 and indexed with a counter variable i 2830.

[0288] A decision step 2835 checks whetherthe number of cycles has exceeded N. If not, one respiratory cycle is detected 2840, the counter i is incremented 2845, and the loop continues. When the number of cycles i exceeds N, the system calculates the average tidal volume across N cycles 2850 and adds the average tidal volume to the database 2855.

[0289] Priorto initiating or resumingstimulation, the system checks at step 2860 whether stimulation is on. If stimulation is active, the algorithm terminates; if not, baselinedata collection continues. This process can be used to establish valid bounds for interpreting respiratory signals. This assists the system in distinguishing between normal breathing patterns and abnormal ones associated with OSA and CSA (e.g., to facilitate application of stimulation to restore proper respiratory function).

[0290] The respiratory signal used in the algorithm 2800 can be derived from various sources, including thoracic impedance (TTI) sensors or accelerometers, depending on the system configuration. In some cases, both sources may be used concurrently or in combination, providing increased fidelity to the patient’s breathing patterns.

[0291] By establishing a baseline of respiratory data, the process shown in Figure 28 can be used to assist in improving accuracy and / or effectiveness of OSA and / or CSA treatments. This may be used to enhance patient outcomes (e.g., by more accurately determiningwhen to apply stimulation to the patient) while also maintaining the intrinsic drive to breathe.

[0292] Figure 29 presents a detailed flowchart of the Inspiration-Predictive Stimulation (IPS) algorithm, designed to initiate stimulation just before a predicted breath. The process starts by setting parameters for the minimum and maximum acceptable durations of both the inspiration and expiration phases 2905. These limits ensure that abnormal breathing patterns, which may fall outside these thresholds, are not used in the calculations.

[0293] A key parameter, the pre-inspiration stimulation time 2910, is defined in milliseconds. This parameter sets the timingfor when stimulation will begin relative to the start of the next predicted breath. The firmware then monitors and filters the respiratory signal 2915, identifies the start of inspiration by detecting minima 2920, and the start of expiration by detecting maxima 2925. These extrema help the system accurately track the phases of each respiratory cycle.

[0294] The firmware maintains a rolling cycle buffer 2930 that stores data from recent respiration cycles. A decision step 2935 determines whether the detected inspiration and expiration periods fall within valid limits. Data that do not meet the criteria are excluded 2940, while valid data are included 2945. Using this historical buffer, the system predicts thetiming of the next inspiration based on N cycles 2950. For each new breath, stimulation is initiated just before the predicted inspiration time at step 2955, using the predefined preinspiration stimulation time to ensure precise delivery.

[0295] The algorithm assumes that the respiratory signal being processed is valid and consistent with the established parameters. A final decision step 2960 determines whether therapy has ended. Possible modifications to the algorithm include replacing the local minimum and maximum detection method with baseline zero-crossings, which could simplify the calculation process. Additionally, the respiratory signal could be acquired using either thoracic impedance (TTI) sensors, accelerometers, or a combination of both, providing flexibility in how respiratory data is collected and interpreted.

[0296] Figure 30 presents a flowchart of the Respiration-Locked Stimulation (RLS) algorithm, which is designed to synchronize the stimulation cycle with the patient’s respiratory rate. This approach ensures that stimulation consistently occurs at a specific phase within the respiratory cycle. The algorithm is modeled after a traditional Phase- Locked Loop (PLL), where a desired phase shift between the stimulation and respiratory cycles is established, and the stimulation cycle is continuously modulated to maintain this shift.

[0297] The algorithm begins by setting the stimulation rate range 3005, measured in cycles per minute (CPM), typically between 12 and 20 CPM. Parameters forthe minimum and maximum valid durations of inspiration and expiration are defined 3010, and a stimulation phase delay 3015 is set in degrees, specifying the timing of stimulation relative to the respiratory cycle.

[0298] The firmware then monitors and filters the respiratory signal 3020, detecting the start of inspiration at minima 3025 and the start of expiration at maxima 3030. A cycle buffer 3035 is updated, and a decision step 3040 determines whether the periods are valid. Data that do not meet the defined limits are excluded 3045. Using this buffer, the system calculates the initial stimulation rate 3050, based on the average respiratory rate of the recent cycles.

[0299] A stimulation delay is then computed 3055 using the phase delay parameter and the average breath rate, determining the timing of stimulation relative to expiration. Stimulation is applied after this delay 3060, and the cycle repeats. The algorithm continuously monitors the phase difference between the respiratory signal and the stimulation 3065. If the phase error exceeds a defined threshold 3070, the stimulation rate is adjusted and the delay is recalculated 3075 to maintain synchronization.

[0300] Key assumptions include the validity of the respiratory signal being processed. Potential modifications include replacingthe local minimum / maximum detection method with baseline zero-crossings, or applying a fixed time delay rather than recalculating it from the phase delay. The respiratory signal may be acquired from thoracic impedance (TTI) sensors, accelerometers, or both, providing flexibility in how respiration data is collected.

[0301] This flowchart illustrates the continuous monitoring and adaptive modulation of the stimulation cycle to ensure it remains synchronized with the patient’s breathing pattern, thereby optimizing the effectiveness of stimulation therapy.

[0302] Figure 31 illustrates the flowchart of the Adaptive Stimulation Amplitude algorithm, designed to adjust the stimulation amplitude dynamically based on the patient’s respiratory signal, with the aim of enhancing respiratory effort when necessary. The algorithm responds to variations in tidal volume, adaptingthe amplitude of stimulation to align with the patient’s current breathing demands.

[0303] The process starts by settingtidalvolume parameters for each recumbent position (supine, right, left, prone) 3105. These values establish the baseline for determining when and how to adjust stimulation amplitude. In addition, parameters for the minimum and maximum phasic amplitude 3110 are defined, setting the lower and upper bounds for amplitude duringstimulation cycles.

[0304] Each stimulation cycle begins atthe minimum phasic amplitude 3115, providing a baseline level of stimulation. The system then monitors tidal volume for each cycle 3120. A decision step determines whether the tidal volume is low 3125. If low, the phasic amplitude is increased 3130, and stimulation is delivered at the adjusted amplitude3135. Another decision step determines whether the tidal volume is high 3140. If high, the phasic amplitude is decreased 3145, after which stimulation continues.

[0305] The algorithm assumes that the maximum tidal volume for each stable position has already been determined from previous baseline signal collection, ensuring that the system has a reference point for appropriate amplitude adjustments.

[0306] Possible modifications to the algorithm include acquiring the respiratory signal using thoracic impedance (TTI) sensors, accelerometers, or both, offering flexibility depending on the system configuration.

[0307] Similarly, Figure 32 illustrates an embodiment of an algorithm 3200 that adapts stimulation duration, which modulates the duration of each stimulation pulse in response to changes in respiratory patterns associated with sleep disordered breathing, including both OSA and CSA. The process begins by setting tidal volume parameters for each recumbent position 3205, followed by defining minimum and maximum phasic duration limits 3210. The algorithm initializes with the phasic duration set to the maximum 3215, and stimulation begins 3220.

[0308] The system then monitors tidal volume for each cycle 3225. A decision step checks whether tidal volume is low 3230. If so, the phasic duration is increased 3235, and stimulation is delivered with the adjusted duration 3240. Another decision step checks whethertidal volume is high 3245. If high, the phasic duration is decreased 3250, after which stimulation continues.

[0309] This algorithm works in a similar manner to the “amplitude adjust” algorithm of Figure 31 , monitoring the same respiratory signal but focusing on adjusting the length of the stimulation pulses. When the patient exhibits weaker or shallower breathing, the algorithm increases the stimulation pulse duration to provide longer support. As the patient’s tidal volume increases and respiratory effort becomes more robust, the duration is reduced, ensuringthe patient is not overstimulated. This continuous adjustment maintains an optimal balance between the timing of stimulation and the patient’s respiratory needs.

[0310] Together, these two algorithms — adaptive in both amplitude (Figure 31) and duration (Figure 32) — create a flexible, patient-responsive stimulation system for treatingboth OSA and CSA with a single lead. These algorithms ensure that both the intensity and duration of stimulation are fine-tuned to the patient’s respiratory cycle in real-time, providing a highly personalized therapy to improve respiratory function. As agnostic algorithms, they can be applied regardless of whether the disordered breathing is obstructive or central in nature, and they work in harmony to increase therapeutic effect while decreasing discomfort and / or overstimulation for the patient.

[0311] Figure 33 (Intermittent Stimulation) discloses a type-specific algorithm, more relevant to certain OSA or CSA presentations, that reduces stimulation frequency by delivering stimulation to every second or third respiratory cycle. The process begins by definingthe duty cycle 3305, where “N” denotes the number of cycles between successive stimulations. The counter3310 is initialized, and respiratory cycles are monitored 3315. With each monitored cycle, the counter is incremented 3320, and a decision step 3325 checks whether the counter has reached N.

[0312] If the count equals N, stimulation is delivered 3330, the counter is reset 3335, and the process continues. If therapy remains active 3340, the algorithm loops back to monitor further respiratory cycles.

[0313] For instance, a 1 :2 duty cycle applies stimulation to every second cycle, while a 1 :3 duty cycle applies stimulation to every third cycle. This method is designed to conserve energy while ensuring sufficient respiratory support through the refractory effect of nerve stimulation. The assumption is that this refractory effect will maintain effective respiration between stimulations, thereby decreasing energy usage without compromising therapeutic efficacy.

[0314] Figure 34 (Adaptive Intermittent Stimulation) expands upon the Intermittent Stimulation approach with an algorithm that incorporates a real-time feedback mechanism to adjust stimulation frequency based on monitored respiratory parameters, such as tidal volume, heart rate, and heart rate variability. The process begins by setting tidal volume parameters for each recumbent position (supine, right, left, prone) 3405, after which respiratory cycles are monitored 3410.

[0315] Initially, stimulation is applied during every respiratory cycle 3415 to provide immediate therapeutic support for OSA and CSA. The system then monitors tidal volume for each cycle 3420 and checks whether stable breathing is achieved 3425. If respiration remains stable, stimulation is adjusted to occur every other cycle 3430. If tidal volume and other monitored parameters remain stable and within predefined limits, stimulation may be further reduced (e.g., every third cycle). In contrast, if tidal volume or other monitored metrics decrease below a threshold, stimulation reverts to every cycle 3415 to maintain adequate respiratory function.

[0316] The algorithm is designed to continuously monitor respiratory and cardiovascular parameters and adapt stimulation frequency to optimize energy consumption while maintaining respiratory support. It assumes thatthe maximum tidal volume for each stable recumbent position has been previously determined via baseline signal collection.

[0317] Both the Intermittent Stimulation algorithm (Figure 33) and the Adaptive Intermittent Stimulation algorithm (Figure 34) aim to optimize energy efficiency by selectively stimulating respiratory cycles in patients with OSA and CSA. The Intermittent algorithm provides a static reduction in stimulation frequency, while the Adaptive algorithm dynamically adjusts stimulation frequency in response to real-time respiratory and cardiovascular conditions. These methods reduce current consumption in medical devices, such as implantable stimulators, while ensuring effective respiratory therapy.

[0318] Figure 35 illustrates a flowchart describing an agnostic algorithm 3500 that monitors and evaluates the quality of a patient’s respiration by determining a Respiratory Quality Index. The algorithm tracks various metrics related to breathing, including the duration of inspiration and expiration, tidal volume, apnea / hypopnea (AH) events, heart rate, and heart rate variability, both prior to and during stimulation.

[0319] The algorithm begins by accessing respiration data 3505 and defining valid inspiration / expiration times and tidal volume ranges 3510. A respiration statistics database is established and buffers are maintained 3515. The respiratory signal is monitored andfiltered 3520, and data from each respiratory cycle is added to the database 3525. The data is periodically averaged 3530 and the database is updated per cycle 3535.

[0320] A decision step checks whether tidal volume falls below a defined threshold 3540. If so, the system records recovery time 3545, classifies AH events 3550, calculates summary statistics 3555, and counts and classifies AH events by duration 3560. These AH events are categorized as apneas or hypopneas, depending on the rate of tidal volume decline and recovery characteristics. Classification may also be enhanced by comparing thoracic impedance (TTI) signals with accelerometer signals.

[0321] The summary data compiled includes averages, minimum and maximum values, and standard deviations for inspiration time, expiration time, tidal volume, heart rate, heart rate variability, and the number of AH events of various durations. This summary is recalculated after a fixed accumulation period or upon detection of a stable position change.

[0322] The algorithm assumes that the end of an AH event can be reliably detected using respiratory and / or cardiovascular signals, and that apneas can be differentiated from hypopneas using these signals. Possible modifications include using TTI or accelerometer data, or defaulting to TTI while activating accelerometer data when tidalvolume drops below a criticalthreshold.

[0323] This algorithm, detailed in Figure 35, provides a comprehensive method for accumulating and analyzing respiratory and cardiovascular metrics to assess respiratory quality and detect critical events such as apneas and hypopneas in both OSA and CSA patients.

[0324] Figure 36 illustrates a flowchart of an agnostic “therapy pause” algorithm that provides various options for handling the loss of valid respiratory sensor signals, which may occur due to either a lack of respiration or signal interference. The purpose of this algorithm is to prescribe different system behaviors when respiratory signals are deemed invalid. This algorithm is agnostic in nature and does not need to define the specific sleep disorder the patient is suffering from, whether that sleep disorder be OSA, CSA, or MSA.

[0325] The algorithm begins by setting the Signal Loss Recovery (SLR) mode 3605. Stimulation is applied 3610 while the respiratory signal is monitored 3615. A decision step checks whether the respiratory signal is valid 3620. If valid, the system continues stimulation as normal. If invalid, the algorithm checks the defined SLR mode.

[0326] If SLR mode is set to Continue 3625, the system maintains stimulation using the most recent valid cycle 3630. If SLR mode is set to Pause 3635, stimulation is halted until valid respiratory signals are detected again 3640. If SLR mode is set to Burst 3645, a long stimulation pulse is applied 3650 immediately after signal loss, and stimulation then ceases untilvalid signals are once more obtained.

[0327] These options offer flexibility in responding to signal loss, allowing the system to continue stimulation, pause therapy, or apply a temporary burst of stimulation, depending on the chosen setting. This ensures that the system can adapt to varying signal conditions and patient needs in both OSA and CSA.

[0328] Figure 44 illustrates the operation of an exemplary “phase adjust” algorithm that may be used by the Controller. Although the morphology of the Objective Function 4402, which represents the benefit of the stimulation, is shown in Figure 44, in other instances, such as under real-world or arbitrary conditions, the shape of the curve is not known and may change over time. Therefore, the Controller periodically searches for the optimal operating point, or the best Delay 4404 within a given window. For the example shown in Figure 44, the algorithm selects two delay values, A4406 = -150 milliseconds and B 4408 = +120 milliseconds. A third value is chosen as the midpoint between A and B, which is C 4410 = -15 milliseconds. All three delay values are attempted, and the resulting values of the objective function are recorded as F_A 4412, F_B 4414, and F_C 4416.

[0329] In the example shown in Figure 44, the vertical axis could represent the peak flow rate in liters per minute, although other measurements, such as tidal volume, heart rate, or heart rate variability, could also be chosen as the objective function. At this point, the delay value giving the lowest benefit is dropped, and the process is repeated. The algorithm ends when F A 4412 and F B 4414 are both equal to or greater than F C 4416. By repeating this process, the algorithm effectively searches for a delay value that results in thegreatest value of the function F, determining an optimal delay value to increase airflow to the lungs. Variations of the algorithm may use Golden Section Search, Exhaustive Search, Newton’s Method, or AMEOBA.

[0330] Based on the delay value determined to be optimal for a given patient under specific conditions, stimulation with the single lead may be delivered during different portions of the inspiratory or expiratory period. Furthermore, the determination of the optimal stimulation delay could be done to maximize one or more breath characteristics, including but not limited to tidal volume, airflow, airway patency, heart rate, and heart rate variability.Additional Examples:

[0331] Afirst embodiment includes a method to treat sleep apnea by one or more target nerves in a sleeping patient, the method comprising: monitoring sleep biomarker (s) of the sleeping patient using sensors; storing and processing of information indicative of the sleep biomarker(s); determining a respiratory cycle of the sleeping patient from the biomarker(s); determining the ventilation efficacy based on the biomarkers; determining occurrences of sleep apnea from the biomarker(s); deliver energy, such as electrical energy, to the target nerve, such as the phrenic nerve during the respiratory cycle(s) of the sleeping patient; determining a phase in the respiratory cycle to deliver energy to a target nerve in the sleeping patient based on the stored information of the sleep biomarker(s); adjusting the phase and / or amplitude for the delivery of the energy by analyzing the stored information of the sleep biomarker(s) of the sleeping patient while the energy is being delivered, and establishing and maintainingthe synchrony between the respiratory cycle and the stimulation cycle.

[0332] Asecond embodiment includes a method for treating sleep apnea comprising collecting patient's data for sleep analysis; detecting OSA, CSA, or mixed sleep apnea from biomarkers; controlling stimulation of the target nerve with the predetermined stimulation parameters and at the timing that would result in desired ventilation efficacy; monitoring patient's biomarkers for response to stimulation.

[0333] Athird embodiment includes a method, wherein the sensors are one or more of accelerometer, transthoracic impedance, pressure sensor, oxygen sensor, electrocardiogram sensor, and microphone.

[0334] Afourth embodiment includes a method, wherein the biomarkers are breath type, ventilation efficacy, and respiration cycle.

[0335] Afifth embodiment includes a method, wherein the breath type is one of normal breath, obstructive sleep apnea, and central sleep apnea.

[0336] A sixth embodiment includes a method, wherein ventilation efficacy is one or more of Inspiratory duty cycle, tidal volume, peak airflow, minute volume, tissue oxygenation and apnea hypopnea index.

[0337] A seventh embodiment includes a method, wherein the inspiration time, expiration time, tidalvolume, heart rate, heart rate variability, and the number of AH events of various durations are derived based on the sensor signals which include one or more of accelerometer, transthoracic impedance, oxygen sensor, pressure sensor, microphone, and electrocardiogram sensor.

[0338] An eighth embodiment includes a method, wherein energy of is a train of electrical stimuli that includes one or more pulses.

[0339] A ninth embodiment includes a method, wherein the pulses are one of unipolar, bipolar, or tripolartype.

[0340] Atenth embodiment includes a method, wherein the respiratory cycle includes the information relatingto the length of the respiratory cycle and time onset of respiration.

[0341] An eleventh embodiment includes a method, wherein delivery of stimulation can be done using one or more of mode switching, mode of synchrony, or agnostic mode.

[0342] Atwelfth embodiment includes a method, wherein monitoring sleep biomarkers can be done beat-to-bet, in ensemble average, or periodically.

[0343] Athirteenth embodiment includes a method, wherein phase adjustment can be done using linear feedback, non-linear feedback, phase locked loop, and optimizers.

[0344] Afourteenth embodiment includes a method, wherein the optimizers include at least one of binary search, exhaustive search, perturbation search, steepest gradient, and downhill simplex.

[0345] Any of the of the first through fourteenth embodiments may be combined with any one or more of any of the other ones of the first through fourteenth embodiments.

[0346] Afifteenth embodiment includes a system for providing treatment for sleep disordered breathing in a patientthat is sleeping, the system comprising: at least one sensor configured to generate a signal based on detection of at least one physical phenomenon of the patient; at least one implantable stimulation electrode configured to apply a stimulation energy to one or more target nerves of the patient; at least one processor configured to perform operations comprising: the method of any or all of the first through fourteenth embodiments.

[0347] While at least one exemplary embodiment of the present invention(s) is disclosed herein, it should be understood that modifications, substitutions and alternatives may be apparent to one of ordinary skill in the art and can be made without departing from the scope of this disclosure. This disclosure is intended to cover any adaptations or variations of the exemplary embodiment(s). In addition, in this disclosure, the terms “comprise” or "comprising" do not exclude other elements or steps, the terms "a" or "one" do not exclude a plural number, and the term “or” means either or both, unless the disclosure states otherwise. Furthermore, characteristics or steps which have been described may also be used in combination with other characteristics or steps and in any order unless the disclosure or context suggests otherwise. This disclosure hereby incorporates by reference the complete disclosure of any patent or application from which it claims benefit or priority.

Claims

1. CLAIMS1 . A system for providing treatment for sleep disordered breathing in a patient that is sleeping by stimulating a phrenic nerve of the patient, the system comprising: at least one processor configured to: monitor, based on data generated by at least one sensor, at least one biomarker of the patient; determine, based on the at least one biomarker, a respiratory cycle of the patient; control at least one stimulation electrode to apply stimulation energy to one or more phrenic nerves of the patient; determine ventilation efficacy based on the at least one biomarker determined from the data from the at least one sensor based on application of the stimulation energy; and based on the ventilation efficacy, adjust a timing within the respiration cycle for when the stimulation energy is to be applied to the one or more phrenic nerves of the patient.

2. A system for providing treatment for sleep disordered breathing in a patient that is sleeping, the system comprising: at least one processor configured to:monitor, based on data generated by at Least one sensor, at least one biomarker of the patient; determine, based on the at least one biomarker, a respiratory cycle of the patient; control application of stimulation energy, by using at least one stimulation electrode, to one or more phrenic nerves of the patient; determine ventilation efficacy of the patient based on signals generated based on the application of the stimulation energy; and based on the ventilation efficacy, adjust a phase of the respiratory cycle at which the stimulation energy is applied.

3. The system of any one of claims 1 or 2, wherein the ventilation efficacy is calculated using one or more of inspiration time, expiration time, tidal volume, heart rate, heart rate variability, and a number of apnea hypopnea (AH) events of various durations.

4. The system of claim 3, wherein the inspiration time, expiration time, tidalvolume, heart rate, heart rate variability, and the number of AH events of various durations are derived based on the sensor signals which include one or more of accelerometer, transthoracic impedance, oxygen sensor, pressure sensor, microphone and electrocardiogram sensor.

5. The system of any one of claims 1 to 4, wherein the stimulation energy is applied by using a train of electrical stimuli that includes one or more pulses.

6. The system of claim 5, wherein the one or more pulses are one or more of unipolar, bipolar, ortripolar type.

7. The system of any one of claims 1 to 6, wherein the at least one processor is further configured to: establish synchrony between the respiratory cycle and the timing or the phase of the stimulation energy relative to the respiratory cycle.

8. The system of claim 7, wherein the at least one processor is further configured to: determine, based on the at least one biomarker, that the synchrony has lapsed; and alter the timing or the phase of the stimulation that is delivered to restore synchrony.

9. The system of claim 7, wherein the at least one processor is further configured to: based on determination of loss of synchrony, adjust delivery of stimulation energy to be increased; and based on determination of reestablishment of synchrony, adjust delivery of simulation energy to be decreased.

10. A system for providing treatment for sleep disordered breathing in a patient that is sleeping, the system comprising: at least one processor configured to: performing a process to determine at least one biomarker of the patient; determine, based on the at least one biomarker, a respiratory cycle of the patient; cause stimulation energy, via at least one stimulation electrode, to be applied to a phrenic nerve; adjust the application of the stimulation energy to be synchronous with the respiratory cycle; determine, based on the at least one biomarker, thatthe synchrony has lapsed; and take an action to reestablish the synchrony, wherein the action includes at least one of increasing the stimulation energy, adjusting timing of the stimulation, or adjusting a phase of the stimulation.11 . The system of claim 10, wherein an amplitude of the stimulation energy is increased without adjustment of frequency of a phase of the stimulation.

12. The system of claim 10, wherein the at least one processor is further configured to:adjust a phase in the respiratory cycle at which the stimulation energy is applied in combination with the increasing an amplitude.

13. The system of any one of claims 10-12, wherein the at least one processor is further configured to: determine ventilation efficacy based on the at least one biomarker obtained from sensor signals generated based on application of the stimulation energy to the at least one phrenic nerve; and based on the ventilation efficacy, adjust a phase in the respiratory cycle at which the stimulation energy is applied.

14. The system of any one of claims 10-12, wherein the at least one processor is further configured to: determine ventilation efficacy based on the at least one biomarker obtained from sensor signals generated based on application of the stimulation energy to the at least one phrenic nerve; and based on the ventilation efficacy, adjust a timing within the respiration phase for when the stimulation energy is applied to the one or more phrenic nerves of the patient.

15. The system of any one of claims 1 to 1 , wherein the at least one processor is further configured to:classify, based on one or more generated signal(s) from a sensor, a respiratory condition of the patient as OSA or CSA.

16. The system of claim 15, wherein the at least one processor is further configured to: based on the classification of the respiratory condition as CSA, deliver a fixed train of the stimulation energy without regard to synchrony between the respiratory cycle and stimulation energy.

17. The system of claim 15, wherein the at least one processor is further configured to: based on the classification of the respiratory condition as OSA, adjust a timing or phase in the respiratory cycle at which the stimulation energy is applied to thereby improve ventilation efficacy.

18. The system of any one of claims 1 to 15, wherein the at least one processor is further configured to deliver a fixed train of the stimulation energy without regard to synchrony between the respiratory cycle and stimulation energy.

19. A system for providing treatment for sleep disordered breathing in a sleeping patient, the system comprising: at least one processor configured to :monitor, based on data generated by at Least one sensor, at least one biomarker of the patient; classify, based on the at least one biomarker, a respiratory condition of the patient of at least OSA and CSA; control stimulation energy applied to the one or more phrenic nerves; based on the classification of the respiratory condition as CSA, deliver a fixed train of the stimulation energy to the one or more phrenic nerves without regard to synchrony to a respiratory cycle of the patient; based on the classification of the respiratory condition as OSA: control the application of the stimulation energy to establish synchrony between the respiratory cycle and the stimulation energy; and after establishing the synchrony, adjusting a timing of the phase in the respiratory cycle at which the stimulation energy is delivered to the one or more phrenic nerves .

20. The system of claim 19, wherein the at least one processor is further configured to: determine a ventilation efficacy of the patient based on the at least one biomarker from sensor signals obtained during or after application of the simulation energy while synchronous with the respiratory cycle; and based on the ventilation efficacy, adjust a phase in the respiratory cycle at which the stimulation energy is applied.21 . The system of any one of claims 19 and 20, wherein the processor is further configured to: determine, based on the at least one biomarker, that the application of the stimulation energy is no longer in synchrony with the respiratory cycle; and based on determination that the stimulation energy is no longer synchronous with the respiratory cycle, increase an amplitude of the stimulation energy, adjust a timing, or adjust a phase of the stimulation.

22. The system of claim 21 , wherein the amplitude of the stimulation energy is increased without adjustment of the phase in the respiratory cycle at which the stimulation energy is applied.

23. The system of claim 21 , wherein the at least one processor is further configured to: adjust the phase at which the stimulation energy is applied in the respiratory cycle in combination with the increasing the amplitude.

24. The system of claim 1 , wherein adjustment of the timing does not occur every time the ventilation efficacy is determined or wherein the biomarker is monitored other than every breath cycle of the patient.

25. A system for providing treatment for sleep disordered breathing in a sleeping patient, the system comprising: at least one sensor configured to detect one or more patient parameters; at least one stimulation electrode configured to deliver electrical stimulation energy to a phrenic nerve of the patient; and at least one processor configured to: receive signals from the at least one sensor indicative of the one or more patient parameters; classify a patient state using a decision function that defines a boundary between an asleep state and an awake state based on the one or more patient parameters; and control delivery of the electrical stimulation energy to the phrenic nerve based on the classification of the patient state.

26. The system of claim 25, wherein the one or more patient includes pitch, bedtime, day of the week, and sleep position .T1. The system of claim 26, wherein parameter weights are assigned inversely to variability .

28. The system of claim 26, wherein a change in weighting shifts a classifier boundary between asleep and awake states.

29. The system of claim 26, wherein the boundary is further adjustable by a patient sensitivity setting.

30. The system of claim 26, wherein parameters having highervariability are omitted from analysis to reduce computation and conserve power.

Images