Electrical stimulation device and method for controlling electrical stimulation device

Abstract

The invention provides an electrical stimulation device and a control method of the electrical stimulation device. The electrical stimulation device comprises an electrode which is in contact with a target object in a use state and is used for performing electrical stimulation on a target tissue of the target object; the sensor is used for collecting breathing data of the target object, and the breathing data comprises at least one of motion data, expiration temperature data, thoracic pressure data, electrical impedance data, blood oxygen data and posture data; the controller is electrically connected with the sensor and the electrode and is used for determining the predicted sleep state of the target object based on the breathing data collected by the sensor; determining an output parameter of a stimulation signal based on the predicted sleep state; and based on the output parameter of the stimulation signal, controlling the electrode to output the stimulation signal so as to stimulate the target tissue. The sleep quality of the target object wearing the electrical stimulation device is improved, and the stimulation effect of the stimulation signal is further improved.

Description

Technical Field

[0001] This disclosure relates to the field of neurostimulation technology, and in particular to an electrical stimulation device and a method for controlling the electrical stimulation device. Background Technology

[0002] Users with sleep apnea may experience apnea and hypoventilation during sleep due to the physiological relaxation of the muscles in the pharynx and tongue, causing the tongue to obstruct the posterior pharyngeal passage. Therefore, neuro- or muscular stimulation is necessary for patients with sleep apnea to enhance and maintain the patency of the upper airway, thereby reducing or eliminating airway obstruction caused by the posterior displacement of the tongue.

[0003] In related technologies, patients turn on an electrical stimulation device before falling asleep. Once turned on, the device directly outputs stimulation signals in a specific mode until the patient manually turns it off after falling asleep. However, the stimulation signals output by the device can affect the patient's sleep state. The fixed stimulation mode is difficult to automatically adapt to changes in the patient's sleep time and lacks uniformity among different patients, thus seriously affecting the user's sleep quality. Summary of the Invention

[0004] In view of this, the present disclosure provides an electrical stimulation device and a control method for the electrical stimulation device, which can effectively improve the output time of the electrical stimulation signal and enhance the user experience.

[0005] According to a first aspect of the present disclosure, an electrical stimulation device is provided, the electrical stimulation device comprising: An electrode that, in use, comes into contact with a target object and is used to electrically stimulate target tissues in the target object's body. A sensor for collecting respiratory data of the target object, the respiratory data including at least one of the following: motion data, expiratory temperature data, intrathoracic pressure data, electrical impedance data, blood oxygen data, and posture data; A controller, electrically connected to the sensor and the electrode, is used for: Based on the respiratory data collected by the sensor, the predicted sleep state of the target object is determined; Based on the predicted sleep state, the output parameters of the stimulation signal are determined; Based on the output parameters of the stimulation signal, the electrode is controlled to output a stimulation signal to stimulate the target tissue.

[0006] In some embodiments, determining the predicted sleep state of the target object based on the respiratory data collected by the sensor includes: Based on the respiratory data, the predicted respiratory state of the target object in the next respiratory cycle is determined; Based on the posture data and electrical impedance data in the respiratory data, the sleep position signal of the target object is determined; Based on the motion data and intrathoracic pressure data in the respiratory data, the sleep movement signals of the target subject are determined; The predicted sleep state of the target object is determined based on the blood oxygen data in the respiratory data, the predicted respiratory state, sleep posture signals, and sleep movement signals.

[0007] In some embodiments, the respiratory data, used to determine the predicted respiratory state of the target object in the next respiratory cycle, includes: Based on the respiratory data, determine the current respiratory state of the target object; Based on the target object's current respiratory state, determine the predicted respiratory state of the target object in the next respiratory cycle.

[0008] In some embodiments, determining the current respiratory state of the target object based on the respiratory data includes: In response to the determination that there is an interference signal in the respiratory signal of the target object, a sampling window is determined according to a preset respiratory frequency range and preset respiratory motion characteristics, wherein the respiratory signal is determined based on the motion data in the respiratory data; Based on the preset respiratory motion characteristics, feature extraction is performed on the respiratory signal within the sampling window, and feature weights are determined. The simulated respiratory signal is determined based on the reference respiratory rate and reference respiratory phase; The reference respiratory rate and reference respiratory phase are adjusted according to the feature weights to minimize the residual between the respiratory signal and the simulated respiratory signal, and to determine the current respiratory rate and current respiratory phase in the current respiratory state.

[0009] In some embodiments, determining the predicted respiratory state of the target object in the next respiratory cycle based on the target object's current respiratory state includes: Based on a pre-built respiratory state space model; Feature vectors are obtained by extracting features from the current breathing state. Based on the state vector corresponding to the current breathing state and the feature vector, the predicted breathing state of the target object in the next breathing cycle is predicted through the state space model.

[0010] In some embodiments, determining the output parameters of the stimulation signal based on the predicted sleep state includes: In response to the predicted sleep state indicating that the target object has begun to enter sleep, the output parameters of the stimulation signal are determined based on the sleep data of the target object within a historical time window. In response to the predicted sleep state indicating that the target object is awake, the output of the stimulation signal is stopped.

[0011] In some embodiments, the controller is further configured to: Based on the difference between the actual sleep state and the predicted sleep state of the target object, a correction function for the sleep breathing model is determined. The sleep breathing model is modified based on the aforementioned correction function.

[0012] In some embodiments, the controller is further configured to: The respiratory state space model is updated based on the difference between the actual respiratory state and the predicted respiratory state of the target object in the next respiratory cycle. Based on the predicted respiratory state, the acquisition parameters of the sensor are adjusted, wherein the acquisition parameters of the sensor include at least one of the following: acquisition frequency, resolution, and power.

[0013] According to a second aspect of the present disclosure, a method for controlling an electrical stimulation device is provided, the method comprising: Respiratory data of a target object is acquired based on sensors, wherein the respiratory data includes at least one of the following: motion data, intrathoracic pressure data, electrical impedance data, blood oxygen data, and posture data; Based on the respiratory data, the controller determines the predicted sleep state of the target object; Based on the predicted sleep state, the output parameters of the stimulation signal are determined by the controller; Based on the output parameters of the stimulation signal, the controller controls the electrodes to output the stimulation signal.

[0014] In some embodiments, determining the predicted sleep state of the target subject based on the respiratory data includes: Based on the respiratory data, the predicted respiratory state of the target object in the next respiratory cycle is determined; Based on the posture data and electrical impedance data in the respiratory data, the sleep position signal of the target object is determined; Based on the motion data and intrathoracic pressure data in the respiratory data, the sleep movement signals of the target subject are determined; The predicted sleep state of the target object is determined based on the blood oxygen data in the respiratory data, the predicted respiratory state, sleep posture signals, and sleep movement signals.

[0015] Determining the predicted respiratory state of the target object in the next respiratory cycle based on the respiratory data includes: Based on the respiratory data, determine the current respiratory state of the target object; Based on the target object's current respiratory state, determine the predicted respiratory state of the target object in the next respiratory cycle.

[0016] In some embodiments, determining the current respiratory state of the target object based on the respiratory data includes: In response to the determination that there is an interference signal in the respiratory signal of the target object, a sampling window is determined according to a preset respiratory frequency range and preset respiratory motion characteristics, wherein the respiratory signal is determined based on the motion data in the respiratory data; Based on the preset respiratory motion characteristics, feature extraction is performed on the respiratory signal within the sampling window, and feature weights are determined. The simulated respiratory signal is determined based on the reference respiratory rate and reference respiratory phase; The reference respiratory rate and reference respiratory phase are adjusted according to the feature weights to minimize the residual between the respiratory signal and the simulated respiratory signal, and to determine the current respiratory rate and current respiratory phase in the current respiratory state.

[0017] In some embodiments, determining the predicted respiratory state of the target object in the next respiratory cycle based on the target object's current respiratory state includes: Based on a pre-built respiratory state space model; Feature vectors are obtained by extracting features from the current breathing state. Based on the state vector corresponding to the current breathing state and the feature vector, the predicted breathing state of the target object in the next breathing cycle is predicted through the state space model.

[0018] In some embodiments, determining the output parameters of the stimulation signal based on the predicted sleep state includes: In response to the predicted sleep state indicating that the target object has begun to enter sleep, the output parameters of the stimulation signal are determined based on the sleep data of the target object within a historical time window. In response to the predicted sleep state indicating that the target object is awake, the output of the stimulation signal is stopped.

[0019] In some embodiments, the method further includes: Based on the difference between the real-time respiratory state and the predicted respiratory state of the target object in the next respiratory cycle, a correction function for the respiratory state space model is determined. The respiratory state space model is modified based on the aforementioned correction function.

[0020] In some embodiments, the controller is further configured to: The respiratory state space model is updated based on the difference between the actual respiratory state and the predicted respiratory state of the target object in the next respiratory cycle. Based on the predicted respiratory state, the acquisition parameters of the sensor are adjusted, wherein the acquisition parameters of the sensor include at least one of the following: acquisition frequency, resolution, and power.

[0021] The technical solutions provided in this disclosure may have the following beneficial effects: By collecting respiratory data from the target subject using sensors, the controller determines the subject's predicted sleep state based on this data. The controller then determines the output parameters of the stimulation signal based on the predicted sleep state to control the electrodes' output of the stimulation signal. This allows the electrical stimulation device to adaptively adjust the stimulation signal according to the target subject's sleep state, avoiding interference during the sleep-onset phase, improving sleep quality when wearing the device, and ultimately enhancing the stimulation effect.

[0022] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0024] Figure 1 This disclosure is a structural diagram of an electrical stimulation device according to an exemplary embodiment; Figure 2 This disclosure is a flowchart illustrating a method for determining a predicted sleep state according to an exemplary embodiment; Figure 3 This disclosure is a flowchart illustrating a method for determining respiratory rate and phase according to an exemplary embodiment; Figure 4 This disclosure illustrates a flowchart for determining a predicted respiratory state according to an exemplary embodiment; Figure 5 This disclosure is a flowchart illustrating a method for determining an electrical stimulation signal according to an exemplary embodiment; Figure 6 This disclosure is a flowchart illustrating a method for determining an electrical stimulation signal according to an exemplary embodiment. Detailed Implementation

[0025] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0026] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0027] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0028] For users with sleep-disordered breathing, an electrical stimulation device can be used during sleep to electrically stimulate the muscles of the tongue or the hypoglossal nerve, causing the target tissues to contract and maintaining an open upper airway. In related technologies, users typically wear and turn on the electrical stimulation device when preparing to fall asleep, allowing it to output stimulation signals. However, users generally do not immediately enter a sleep state after preparing to fall asleep. If stimulation signals are output during the precise sleep-falling stage, the signals may cause a slight fever or even discomfort in the stimulated area, affecting the user's normal sleep-falling process and increasing the time required to fall asleep. Furthermore, if the user wakes up or gets out of bed after falling asleep, the stimulation signals will also prevent the user from falling back asleep, leading to a rejection of the stimulation therapy and reducing sleep quality.

[0029] Based on this, this disclosure provides an electrical stimulation device, which can be an implantable device, specifically, such as a tongue muscle stimulator, which provides electrical stimulation to the user's tongue muscles by inserting the stimulator into the user's mouth. Alternatively, it can be an implantable device, specifically, such as a hypoglossal nerve stimulator, which provides electrical stimulation to the user's hypoglossal nerve after implantation. The electrical stimulation device can adjust the output stimulation signal to the user's tongue muscles or hypoglossal nerve according to the user's sleep state, so as to maintain the patency of the airway while ensuring the user's sleep quality. The electrical stimulation device includes: An electrode that, in use, contacts a target object and is used to electrically stimulate the target tissue of the target object. A sensor for collecting respiratory data of the target object, the respiratory data including at least one of the following: motion data, expiratory temperature data, intrathoracic pressure data, electrical impedance data, blood oxygen data, and posture data; A controller, electrically connected to the sensor and the electrode, is used for: Based on the respiratory data collected by the sensor, the predicted sleep state of the target object is determined; Based on the predicted sleep state, the output parameters of the stimulation signal are determined; Based on the output parameters of the stimulation signal, the electrode is controlled to output a stimulation signal to stimulate the target tissue.

[0030] In some embodiments, the structural diagram of the electrical stimulation device may be as follows: Figure 1 As shown in the diagram, the controller is electrically connected to the electrodes and sensors. The sensors collect respiratory data of the target subject during sleep and transmit the data to the controller. Based on the respiratory data collected by the sensors, the controller determines the predicted sleep state of the target subject and then determines the output parameters of the stimulation signal used to stimulate the tongue muscles or hypoglossal nerve of the target subject. Finally, the control electrodes output stimulation signals according to the output parameters to stimulate the tongue muscles or hypoglossal nerve of the target subject, causing the tongue muscles to contract, maintaining the airway patency, and preventing respiratory obstruction.

[0031] The target tissue can be the target object's tongue muscles, such as the geniohyoid, geniohyoid, and palatohyoid muscles, or the target object's hypoglossal nerve. When the electrical stimulation device is an implantable device, the electrodes are attached to the target object's tongue muscles so that the stimulation signal output by the device can stimulate the target object's tongue muscles. When the electrical stimulation device is an implantable device, the electrodes are implanted under the target object's tongue so that the device can stimulate the target object's hypoglossal nerve. Sensors can include motion sensors, temperature sensors, pressure sensors, impedance sensors, blood oxygen sensors, and posture sensors. Motion sensors collect motion data of the target object during sleep, such as turning over or kicking. The temperature sensor can be placed in the target object's nasal cavity or nostrils to collect the temperature of the exhaled air. The pressure sensor can be placed outside the target object's chest cavity or implanted in the target object's chest to collect chest pressure data to determine the target object's lung pressure, thereby determining the target object's respiratory amplitude and rhythm. The impedance sensor can be installed on the chest and abdomen of the target object to determine the target object's respiratory rhythm by collecting impedance changes in the chest and abdomen. The blood oxygen sensor is used to collect the blood oxygen concentration in the target object. If the blood oxygen saturation in the target object is lower than a threshold, it indicates that the target object's breathing is obstructed, and external oxygen cannot enter the lungs in time to supply the target object's body consumption. The posture sensor can detect continuous and significant changes in the target object's body, such as changes in sleep posture, such as changing from supine to lateral. It should be noted that the respiratory data includes at least the entire time series of the target object within the current respiratory cycle, or respiratory data from multiple respiratory cycles within a historical time period, such as respiratory data from the past ten minutes. The respiratory cycle can be determined based on the respiratory phase, from the inspiratory phase of 0-180° to the expiratory phase of 180-360°, which can be considered as a complete respiratory cycle.

[0032] After the sensor collects the respiratory data of the target object, the respiratory data can be transmitted to the controller via a data transmission channel. Upon receiving the respiratory signal, the controller can predict the target object's respiratory state in the next respiratory cycle and estimate the target object's predicted sleep state based on the predicted respiratory state. Then, using a sleep breathing model, combined with the predicted respiratory state and the target object's physiological changes during sleep, the predicted sleep state of the target object is determined. Based on the predicted sleep state, the timing, duration, and intensity of the output stimulus signal are determined. Finally, the control electrode outputs a stimulus signal according to the output parameters to stimulate the muscles of the target object's tongue to contract, thereby keeping the target object's airway open.

[0033] In some embodiments, the process of determining the predicted sleep state of the target subject based on respiratory data is described in [reference needed]. Figure 2 The flowchart shown includes: S201, Based on the respiratory data, determine the predicted respiratory state of the target object in the next respiratory cycle.

[0034] After the sensor collects the respiratory data of the target object, the respiratory data can be transmitted to the controller via a data transmission channel. Upon receiving the respiratory signal, the controller can predict the target object's respiratory state in the next respiratory cycle based on a respiratory state-space model. The predicted respiratory state can include the target object's motion state, respiratory rate, phase, amplitude, blood oxygen content, and the frequency and phase of respiratory attenuation. Based on the target object's predicted respiratory state, information such as the risk of airway obstruction, the duration of inspiratory and expiratory breathing in the next respiratory cycle, and changes in blood oxygen saturation can be determined.

[0035] In some embodiments, the current respiratory state of the target object can be determined first based on the respiratory data; and then the predicted respiratory state of the target object in the next respiratory cycle can be determined based on the current respiratory state of the target object.

[0036] First, a respiratory state vector for the target object within the current respiratory cycle can be constructed based on respiratory data collected by sensors. Then, this respiratory state vector is input into a preset respiratory state space model, and the respiratory state is predicted based on this model to obtain the predicted respiratory state for the next respiratory cycle. The respiratory state vector includes parameters such as motion state, respiratory rate, phase, amplitude, blood oxygen content, respiratory attenuation frequency, and respiratory attenuation phase. The state transition matrix in the respiratory state space model can be determined based on a sequence of respiratory state data over a historical time period to estimate the optimal matrix. For example, it can be determined using maximum likelihood estimation, expectation-maximization algorithms, or least squares methods.

[0037] In some embodiments, determining the respiratory status of a target object during the current respiratory cycle can first involve determining multiple respiratory indicators of the target object, which may specifically include: Based on the intrathoracic pressure data, electrical impedance data, and expiratory temperature data in the respiratory data, determine the target object's current respiratory rate, respiratory amplitude, and respiratory phase; based on the intrathoracic pressure data, electrical impedance data, blood oxygen data, and expiratory temperature data in the respiratory data, determine the target object's respiratory attenuation frequency and respiratory attenuation phase; based on the blood oxygen data in the respiratory data, determine the target object's current blood oxygen content; based on the motion data and posture data in the respiratory data, determine the target object's motion state.

[0038] After acquiring the respiratory data collected by the sensors, the respiratory data can be preprocessed. For example, the respiratory data can be sampled synchronously to ensure that the timestamps of all sensor data are aligned (using hardware synchronization or software interpolation). The respiratory data can also be processed by normalization, and different sensor signals can be normalized to eliminate the influence of dimensions.

[0039] During different stages of respiration, intrathoracic pressure, intrathoracic impedance, and expiratory temperature all change with the respiratory process. For example, intrathoracic pressure increases during inspiration and returns to normal during expiration; intrathoracic impedance increases during inspiration (air entering reduces conductivity) and decreases during expiration; and expiration temperature is significantly higher than inspiration (exhaled air ≈ body temperature, inhaled air ≈ room temperature). Therefore, based on intrathoracic pressure, impedance, and expiratory temperature data from respiratory data, the current respiratory rate, respiratory amplitude, and respiratory phase of a target subject can be determined. For example, based on impedance data during respiration, spectral analysis can be used to extract characteristics of the impedance data to determine the respiratory rate. Alternatively, peak testing can be used to analyze the minimum values ​​in pressure data and / or expiratory temperature, and the respiratory cycle can be determined based on the interval between adjacent minimum values, thus determining the respiratory rate. Furthermore, multi-sensor cross-validation can be used to compare the frequencies calculated from the three signals, identify outliers, and thus determine the respiratory rate.

[0040] It should be noted that during sleep, changes in body position may weaken the respiratory signals captured by the motion sensor, and environmental noise may also significantly interfere with the respiratory signals. Figure 2 The diagram shows that the respiratory signal includes significant noise interference, causing the respiratory cycle boundaries to become blurred. This further leads to lower accuracy in estimating respiratory rate and phase. If the estimation accuracy of respiratory rate and phase is low, the accuracy of determining electrical stimulation parameters based on the current respiratory state will also be low, further resulting in lower accuracy in stimulating the hypoglossal nerve or tongue muscles of the target subject, making it impossible to accurately stimulate these organs.

[0041] Therefore, in some embodiments, feature extraction can be performed on respiratory signals carrying interference signals to improve the accuracy of determining respiratory rate and respiratory phase. For details, please refer to [link to relevant documentation]. Figure 3 The flowchart shown includes: S301, in response to determining that there is an interference signal in the respiratory signal of the target object, a sampling window is determined according to a preset respiratory frequency range and preset respiratory motion characteristics.

[0042] S302, extract features from the respiratory signal within the sampling window based on the preset respiratory motion features, and determine the feature weights.

[0043] S303 determines the simulated respiratory signal based on the reference respiratory rate and reference respiratory phase.

[0044] S304, the reference respiratory frequency and reference respiratory phase are adjusted according to the feature weights to minimize the residual between the respiratory signal and the simulated respiratory signal, and the current respiratory frequency and current respiratory phase in the current respiratory state are determined.

[0045] The preset respiratory frequency range can be determined based on the target object's historical respiratory data, and the preset respiratory motion characteristics can be determined by feature extraction from the target object's historical respiratory data. The respiratory signal is determined based on motion data within the respiratory data; for example, it can be determined based on the target object's chest displacement data. Based on the respiratory characteristics in the respiratory signal, it can be determined whether interference signals exist in the respiratory signal. For example, if data points in the respiratory signal exceed a preset amplitude range, interference signals can be identified. Alternatively, if the rate of change (slope) between adjacent sampling points in the respiratory signal exceeds a preset threshold, interference signals can be identified.

[0046] If interference signals are detected in the respiratory signal of the target object, the length of the sampling window can be determined based on a preset respiratory frequency range and preset respiratory motion characteristics to ensure that the sampling window contains a specific number of respiratory cycles. For example, if the preset respiratory frequency range is 3-6 breaths per minute, the sampling window length can be determined to be 100 seconds to ensure that the sampling window contains at least 5 respiratory cycles. Alternatively, a sampling window containing smooth, periodic, and sinusoidal motion can be selected from the respiratory signal based on preset respiratory motion characteristics. By selecting a sampling window, computational resources can be avoided in invalid frequency ranges, interference signals in the sampling window can be initially reduced, and subsequent computational efficiency can be improved.

[0047] After determining the sampling window, features can be extracted from the respiratory signal within the sampling window based on preset respiratory motion characteristics, and feature weights can be determined. Specifically, a filter can be designed based on the preset respiratory motion characteristics. The filter is used to perform convolution calculations on the respiratory signal within the sampling window. The convolution calculation enhances signal components (i.e., respiratory signals) similar to the preset respiratory motion characteristics in the sampling window, while weakening dissimilar components (noise), thereby improving the signal-to-noise ratio of the respiratory signal within the sampling window. After performing convolution calculations on the respiratory signal within the sampling window, a weighted feature vector of the respiratory signal within the sampling window can be obtained. The feature weights represent the weights of different respiratory frequencies.

[0048] The reference breathing frequency can be determined based on the preset breathing frequency range, for example, selecting a breathing frequency within the preset breathing frequency range. The reference breathing frequency can also be determined based on the feature weights, for example, determining the breathing frequency corresponding to the maximum weight value within the feature weights as the reference breathing frequency. After determining the breathing frequency, a corresponding reference breathing phase can be selected based on the breathing frequency. The reference breathing phase can also be determined based on the preset respiratory motion characteristics. After determining the reference breathing frequency and reference breathing phase, a breathing model can be constructed based on the reference breathing frequency and reference breathing phase, and a simulated breathing signal can be obtained based on the breathing model. The breathing model can be a sine wave function; substituting the reference breathing frequency and reference breathing phase into the sine wave function yields the simulated breathing signal.

[0049] The reference respiratory rate and reference respiratory phase are adjusted according to the feature weights to minimize the residual between the respiratory signal and the simulated respiratory signal, and to determine the current respiratory rate and current respiratory phase in the current respiratory state.

[0050] Then, the residual between the respiratory signal and the simulated respiratory signal is calculated, and a cost function is constructed based on the feature weights and the residual between the respiratory signal and the simulated respiratory signal. By minimizing the cost function, the values ​​of the reference respiratory frequency and the reference respiratory phase are iteratively adjusted until the cost function converges. The reference respiratory frequency at this point is determined as the current respiratory frequency in the current respiratory state, and the reference respiratory phase at this point is determined as the current respiratory phase in the current respiratory state. In some embodiments, the residual between the respiratory signal and the simulated respiratory signal can be minimized using algorithms such as gradient descent, Gauss-Newton, or Levenberg-Marquardt to obtain the current respiratory frequency and the current respiratory phase.

[0051] When calculating respiratory amplitude, a weighted average can be calculated based on the peak values ​​of the impedance data and the peak values ​​of the intrathoracic pressure data, and the exhaled volume can be determined based on the integral area of ​​the respiratory temperature data, thus determining the respiratory amplitude. When determining the respiratory phase, a pre-constructed respiratory cycle template can be used. First, the respiratory cycle can be segmented, for example, using the rising edge of the respiratory temperature data (the start of exhalation) as the cycle starting point. Then, key point detection can be performed, such as determining the end of inspiration based on the peak times of the intrathoracic pressure data and / or impedance data, and determining the end of exhalation based on the time when the expiratory temperature data falls back to the baseline, thus determining the respiratory phase.

[0052] Based on the thoracic pressure, electrical impedance, blood oxygen, and expiratory temperature data in the respiratory data, the respiratory attenuation frequency and respiratory attenuation phase of the target subject are determined. The respiratory attenuation frequency refers to the frequency of abnormal breathing patterns caused by repeated upper airway collapse during sleep, including apnea frequency and / or hypoventilation frequency. Apnea frequency refers to the complete cessation of airflow for a period exceeding a threshold, but chest and abdominal respiratory movements still occur (due to obstruction). Apnea frequency can be determined based on the number of apneas per hour of sleep; for example, if the target subject experiences 30 apneas per hour, the apnea frequency is 30 apneas / hour. Specifically, the respiratory data collected by the sensors is characterized by the disappearance of expiratory temperature data (indicating no airflow through the nasal cavity), but thoracic pressure and / or electrical impedance data indicating that the target subject is in a respiratory state; for example, a decrease in the amplitude of thoracic pressure fluctuations (weakened inspiratory effort) / a decrease in the amplitude of electrical impedance changes (insufficient chest expansion). Simultaneously, blood oxygen data indicates a decrease in blood oxygen saturation in the target subject. Hypoventilation frequency refers to a partial reduction in respiratory airflow that persists above a threshold, accompanied by a decrease in blood oxygen saturation or mild arousal. The frequency of hypoventilation can be determined based on the number of times hypoventilation occurs during sleep per hour. For example, if 20 hypoventilations occur per hour, the hypoventilation frequency is 20 times / hour. Specifically, the respiratory data collected by the sensors are characterized by a decrease in the peak expiratory temperature or a shortened plateau phase (weakened airflow), but a decrease in the amplitude of intrathoracic pressure fluctuations (weakened inspiratory effort) / a decrease in the amplitude of impedance changes (insufficient thoracic expansion). At the same time, blood oxygen data indicates a decrease in blood oxygen saturation in the target subject.

[0053] The attenuation phase of respiration refers to a specific stage within a single respiratory cycle where ventilation efficiency decreases due to partial or complete upper airway collapse. This primarily includes inspiratory phase obstruction, where the upper airway collapses during inspiration, restricting airflow but maintaining respiratory effort; and expiratory phase restriction, where airflow is prolonged during expiration due to airway narrowing or dynamic collapse. Detecting the attenuation phase of inspiration requires steps such as cycle segmentation and feature extraction. First, the start of expiration (cycle start) is marked by the rise in expiratory temperature, and the period from the temperature drop to its lowest point until the impedance / pressure reaches its peak is defined as the inspiratory phase. Then, based on the characteristics of intrathoracic pressure data, impedance data, and blood oxygenation data, attenuation data of the inspiratory phase are determined. Specifically, the intrathoracic pressure data shows increased negative pressure fluctuations (sawtooth waveform, reflecting effort to counteract obstruction), the impedance data shows a slower rate of increase (due to restricted airflow and limited chest expansion), and the blood oxygenation data shows a decrease in blood oxygen saturation. The expiratory phase decreases by first determining the timing of the peak in intrathoracic pressure data, the peak in electrical impedance data, or the time when expiratory temperature data returns to baseline. Then, the expiratory phase decreases by determining the timing of the plateau in expiratory temperature data and the failure of intrathoracic pressure to return to baseline during expiration.

[0054] Furthermore, the current blood oxygen content of the target subject can be determined based on the blood oxygen data in the respiratory data. The motion state of the target subject can also be determined based on the motion and posture data in the respiratory data. The motion state characterizes the body movements or postures of the target subject during sleep. The number of turns and the standard deviation of acceleration (reflecting the amplitude of movement) can be determined based on the time-domain features in the motion data; frequency-domain features can be extracted based on Fourier transform to determine the target subject's movements, such as the typical frequency of turning over (0.1–0.3 Hz); and the continuity of movements can be determined based on periodic features. The spatial relationships of the target subject's body movements can also be extracted from the posture data to determine the target subject's sleep posture; for example, the prone position can be identified based on the angle between the midpoint of the shoulder and the key point of the nose (angle < critical value).

[0055] After obtaining the current respiratory state of the target object, the controller can predict the target object's respiratory state in the next respiratory cycle based on the current respiratory state. For the method of determining the predicted respiratory state, please refer to [link to relevant documentation]. Figure 4 The flowchart shown includes: S401, extract features from the current breathing state to obtain a feature vector.

[0056] S402, based on the state vector corresponding to the current breathing state and the feature vector, predict the predicted breathing state of the target object in the next breathing cycle through a pre-constructed breathing state space model.

[0057] The state transition matrix in the respiratory state space model can be determined based on a sequence of respiratory state data over a historical time period. When extracting features from the current respiratory state, a convolutional feature extraction network can be used. Feature extraction is performed on the current breathing state to obtain the feature vector corresponding to the current breathing state. Then, the current breathing state and the feature vector can be input together into a pre-constructed breathing state space model to predict the breathing state of the next breathing cycle. In one embodiment, the calculation method for predicting the breathing state can be found in formula (1):

[0058] (1) in, Indicates a prediction of respiratory status. Indicates the current respiratory status. Represents the state transition matrix. Let G represent the eigenvectors, and G be the weight matrix.

[0059] The weight matrix is ​​used to map the feature vectors to the state space, and its elements reflect the contribution strength of different features to the state in the next cycle. The weight matrix can clearly quantify the influence of features on the state, which may include prior information containing relationships such as respiratory cycle, amplitude, and blood oxygen content. For example, a prolonged respiratory cycle corresponds to a decreased frequency, a decreased respiratory amplitude corresponds to a decreased blood oxygen saturation, a rapid decrease in blood oxygen saturation corresponds to a weakening respiratory rate and an increased respiratory rate, and the current weakening respiratory rate corresponds to a continued weakening in the next respiratory cycle, etc.

[0060] After obtaining the predicted respiratory state of the target object in the next respiratory cycle, the controller can further determine the physiological changes of the target object during sleep based on the respiratory data, such as sleep posture changes and physiological indicator changes.

[0061] S202, determine the sleep position signal of the target object based on the posture data and electrical impedance data in the respiratory data.

[0062] The sleep position signal data of the target object refers to the signal of changes in body posture (such as supine, lateral, prone, etc.) during sleep. When determining the sleep position signal of the target object, the changes in the target object's body values ​​along the X, Y, and Z axes and the electrical impedance data can be extracted based on the posture data detected by the accelerometer. These changes are then combined with preset judgment conditions for position changes to determine the target object's sleep position signal. For example, the target object's sleep posture can be determined based on the static gravity component in the posture data, where supine corresponds to Z-axis ≈ 1g, X / Y ≈ 0g; left lateral position corresponds to X-axis ≈ -1g, Y / Z ≈ 0g; and prone corresponds to Z-axis ≈ -1g. Alternatively, the target object's position change can be determined through dynamic roll detection, such as the peak value of the gyroscope's Y-axis angular velocity (typically a rolling angular velocity > 50° / s). A decrease in abdominal impedance in the electrical impedance data (due to abdominal pressure and improved electrode contact) can characterize a change in the target object from supine to lateral position.

[0063] S203, determine the sleep movement signal of the target object based on the motion data and thoracic pressure data in the respiratory data.

[0064] The sleep motion signals refer to the body movement signals caused by limb movements, turning over, or brief micro-awakenings that occur during sleep. The timing of the start and end of movement can be determined based on motion data. Combining this with intrathoracic pressure data can further determine whether the target's respiratory status has changed. For example, if the pressure value reaches a peak after the motion data indicates the start of movement, it indicates that the target's limbs are moving. Conversely, if the intrathoracic pressure data changes but the motion data indicates no limb movement, it can be determined that the target may have simply coughed or the motion sensor may have shifted.

[0065] S204, based on the blood oxygen data in the respiratory data, the predicted respiratory state, the sleep position signal, and the sleep movement signal, determine the predicted sleep state of the target object.

[0066] When determining the predicted sleep state of a target subject based on blood oxygen data, the predicted respiratory state, sleep position signals, and sleep movement signals, a sleep breathing model can be used. This sleep breathing model can be a hierarchical hybrid model. First, feature extraction is performed on the blood oxygen data, predicted respiratory state, sleep position signals, and sleep movement signals. Specifically, features such as frequency, amplitude, and inspiratory slope can be extracted from the respiratory signals; features such as supine / lateral lying classification and turning frequency can be extracted from the sleep position signals; duration features can be extracted from the sleep movement signals; and features such as mean, descent slope, and minimum value of blood oxygen saturation can be extracted from the blood oxygen data. Then, based on the feature information of the predicted respiratory state, a hidden Markov model is used to calculate the probability of being in different sleep stages. These sleep stages can include NREM (Non-Rapid Eye Movement Sleep), REM (Rapid Eye Movement Sleep), and wakefulness. Weight parameters are generated based on the features of the sleep position signals and sleep movement signals to correct the probabilities of different sleep stages; and the sleep breathing stages are validated using features extracted from the blood oxygen data. In one embodiment, the predicted sleep state of the target object can be determined by referring to the method shown in formula (2), as follows: (2) in, Indicates a prediction of sleep state. This represents a sleep breathing model. Indicates a prediction of respiratory status. Indicates sleep body movement signals, and This indicates the blood oxygen content.

[0067] After obtaining the predicted sleep state of the target subject, the output parameters of the stimulus signal can be determined based on the predicted sleep state. These output parameters include: whether a stimulus signal is output, the timing of the stimulus signal output, the duration of the stimulus signal output, and the intensity of the stimulus signal.

[0068] In some embodiments, determining the output parameters of the stimulation signal based on the predicted sleep state includes: in response to the predicted sleep state indicating that the target object has begun to fall asleep, determining the output parameters of the stimulation signal based on the sleep data of the target object within a historical time window; and in response to the predicted sleep state indicating that the target object is awake, stopping the output of the stimulation signal.

[0069] The target object begins to enter sleep when its predicted sleep state changes from wakefulness to NREM or REM. Therefore, the output parameters of the stimulus signal can be determined based on the target object's predicted sleep state within the historical event window. For example, when the target object begins to enter sleep, the output parameters of the stimulus signal can be determined based on the target object's sleep data over the past ten minutes. Specifically, the output parameters can be determined based on the respiratory status data within the sleep data. In one embodiment, the next respiratory cycle can be determined first, as detailed in formula (3): (3) in, This is a respiratory state cycle, through the state... Each dimension The weighted average calculation uses the weight of the k-th dimension. This indicates that the next respiratory cycle can be obtained.

[0070] In one embodiment, the stimulus output timing can be determined based on a preset time bias and the predicted start time of respiration in the respiratory state. It should be noted that the stimulus signal can be emitted before the start of the next respiratory cycle. The preset time bias can be determined based on nerve conduction delay and muscle mechanical response delay, typically ranging from 200ms to 500ms, and can be calibrated according to different target subjects. Based on the respiratory state cycle determined in the predicted respiratory state, the start time of the next respiratory cycle can be determined. Then, before the start time of the next respiratory cycle, the stimulus output timing is determined based on the preset time bias.

[0071] The dynamic time bandwidth refers to the time domain window corresponding to the frequency range of effective physiological changes in the respiratory signal (e.g., 0.1-2 Hz corresponds to 0.5-10 seconds), which can be determined by the power spectral density (PSD) of the respiratory signal. The predicted duration of the respiratory state is the next respiratory state cycle. After determining the next respiratory state cycle, the stimulation duration can be determined according to the method shown in formula (4), as follows: (4) Where h is the stimulation duration, For dynamic time bandwidth.

[0072] When determining the intensity of the stimulus signal, the intensity of the stimulus signal can be obtained by fitting the respiratory amplitude and blood oxygen content with a polynomial. For details, please refer to formula (5): (5) Among them, , Here, represents the polynomial coefficients, and n represents the degree of the polynomial. This refers to the range of breathing. This refers to blood oxygen content.

[0073] After obtaining the output parameters of the stimulation signal used to stimulate the tongue muscle tissue and hypoglossal nerve, the controller can control the electrodes to output the stimulation signal according to the output parameters to electrically stimulate the hypoglossal nerve or tongue muscle tissue of the target object, thereby ensuring that the stimulation signal can accurately stimulate the user's tongue muscle tissue or hypoglossal nerve, keep the user's airway open, and improve the user's sleep experience.

[0074] If the predicted sleep state indicates that the target subject is awake, it means that the target subject can control tongue muscle contraction independently, without the need for additional stimulation signals from the electrical stimulation device. Therefore, the controller can control the electrodes to stop outputting stimulation signals to avoid affecting the target subject's sleep onset, shorten the sleep onset time, and improve the sleep quality.

[0075] In some embodiments, the controller is further configured to: determine a correction function for the respiratory state space model based on the difference between the real-time respiratory state of the target object in the next respiratory cycle and the predicted respiratory state; and correct the respiratory state space model based on the correction function.

[0076] In one embodiment, the correction function can be found in formula (6): (6) in, Indicates a prediction of sleep state. This represents the actual predicted sleep state. The correction function is then obtained. Then, the correction function can be superimposed on the sleep breathing model to correct the sleep breathing model.

[0077] Stimulation signals are generated by fusion estimation of sleep breathing states, and the adaptability of the stimulation signals during sleep is optimized by incorporating sleep breathing state feedback. When the target subject experiences events such as awakening or getting up during sleep, the electrical stimulation device adaptively adjusts the stimulation signals according to changes in breathing states, avoiding interference with the target subject's sleep process and improving the adaptability of stimulation for obstructive sleep apnea.

[0078] This disclosure also provides a method for determining an electrical stimulation signal, which can be applied to an electrical stimulation device. For the method steps, please refer to [link to relevant documentation]. Figure 5 This includes: S501, acquire respiratory data of the target object based on sensors, wherein the respiratory data includes at least one of the following: motion data, intrathoracic pressure data, electrical impedance data, blood oxygen data, and posture data.

[0079] S502, Based on the breathing data, the controller determines the predicted sleep state of the target object.

[0080] S503, based on the predicted sleep state, the controller determines the output parameters of the stimulation signal.

[0081] S504, based on the output parameters of the stimulation signal, the controller controls the electrodes to output the stimulation signal.

[0082] The sensor collects respiratory data from the target subject during sleep and transmits the data to the controller. Based on this data, the controller predicts the target subject's sleep state and determines the output parameters for the stimulation signal used to stimulate the neuromuscular system. Finally, the control electrodes output the stimulation signal according to these parameters to stimulate the target subject's tongue muscles or hypoglossal nerve, causing the tongue muscles to contract, maintaining an open airway, and preventing respiratory obstruction.

[0083] In some embodiments, determining the predicted sleep state of the target subject based on the respiratory data includes: Based on the respiratory data, the predicted respiratory state of the target object in the next respiratory cycle is determined; Based on the posture data and electrical impedance data in the respiratory data, the sleep position signal of the target object is determined; Based on the motion data and intrathoracic pressure data in the respiratory data, the sleep movement signals of the target subject are determined; The predicted sleep state of the target object is determined based on the blood oxygen data in the respiratory data, the predicted respiratory state, sleep posture signals, and sleep movement signals.

[0084] Determining the predicted respiratory state of the target object in the next respiratory cycle based on the respiratory data includes: Based on the respiratory data, determine the current respiratory state of the target object; Based on the target object's current respiratory state, determine the predicted respiratory state of the target object in the next respiratory cycle.

[0085] In some embodiments, determining the current respiratory state of the target object based on the respiratory data includes: In response to the determination that there is an interference signal in the respiratory signal of the target object, a sampling window is determined according to a preset respiratory frequency range and preset respiratory motion characteristics, wherein the respiratory signal is determined based on the motion data in the respiratory data; Based on the preset respiratory motion characteristics, feature extraction is performed on the respiratory signal within the sampling window, and feature weights are determined. The simulated respiratory signal is determined based on the reference respiratory rate and reference respiratory phase; The reference respiratory rate and reference respiratory phase are adjusted according to the feature weights to minimize the residual between the respiratory signal and the simulated respiratory signal, and to determine the current respiratory rate and current respiratory phase in the current respiratory state.

[0086] In some embodiments, determining the predicted respiratory state of the target object in the next respiratory cycle based on the target object's current respiratory state includes: Based on a pre-built respiratory state space model; Feature vectors are obtained by extracting features from the current breathing state. Based on the state vector corresponding to the current breathing state and the feature vector, the predicted breathing state of the target object in the next breathing cycle is predicted through the state space model.

[0087] In some embodiments, determining the output parameters of the stimulation signal based on the predicted sleep state includes: In response to the predicted sleep state indicating that the target object has begun to enter sleep, the output parameters of the stimulation signal are determined based on the sleep data of the target object within a historical time window. In response to the predicted sleep state indicating that the target object is awake, the output of the stimulation signal is stopped.

[0088] In some embodiments, the method further includes: Based on the difference between the real-time respiratory state and the predicted respiratory state of the target object in the next respiratory cycle, a correction function for the respiratory state space model is determined. The respiratory state space model is modified based on the aforementioned correction function.

[0089] In some embodiments, the controller is further configured to: The respiratory state space model is updated based on the difference between the actual respiratory state and the predicted respiratory state of the target object in the next respiratory cycle. Based on the predicted respiratory state, the acquisition parameters of the sensor are adjusted, wherein the acquisition parameters of the sensor include at least one of the following: acquisition frequency, resolution, and power.

[0090] In some embodiments, the method for determining the electrical stimulation signal may refer to [reference needed]. Figure 6 The flowchart shown is shown.

[0091] First, the controller, in step 1, controls multiple sensors to collect respiratory data from the target subject while wearing the electrical stimulation device. Then, in step 2, the respiratory data is analyzed to determine the target subject's sleep position, sleep movement, and blood oxygen saturation. Next, in step 3, a sleep apnea model is invoked to estimate the target subject's predicted sleep state based on their sleep position, sleep movement, and blood oxygen saturation. In step 4, based on the target subject's predicted sleep state, the output parameters of the electrical stimulation device are determined, and the electrodes are controlled to output stimulation signals to stimulate the hypoglossal nerve or tongue muscles of the target subject. In step 5, the sensor acquisition parameters are updated based on the estimated predicted sleep state and the target subject's actual predicted sleep state. Finally, in step 6, the sleep apnea model is updated based on the estimated predicted sleep state and the target subject's actual predicted sleep state to improve the accuracy of the sleep apnea model in predicting the target subject's predicted sleep state.

[0092] For the foregoing method embodiments, in order to simplify the description, they are described as a series of actions. However, those skilled in the art should understand that this disclosure is not limited to the described order of actions, because according to this disclosure, some steps may be performed in other orders or simultaneously.

[0093] Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by this disclosure.

[0094] The foregoing has described specific embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention applied herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not claimed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0096] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

[0097] The above description is merely a preferred embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. An electrical stimulation device, characterized in that, The electrical stimulation device includes: An electrode that, in use, comes into contact with a target object and is used to electrically stimulate target tissues in the target object's body. A sensor for collecting respiratory data of the target object, the respiratory data including at least one of the following: motion data, expiratory temperature data, intrathoracic pressure data, electrical impedance data, blood oxygen data, and posture data; A controller, electrically connected to the sensor and the electrode, is used for: Based on the respiratory data collected by the sensor, the predicted sleep state of the target object is determined; Based on the predicted sleep state, the output parameters of the stimulation signal are determined; Based on the output parameters of the stimulation signal, the electrode is controlled to output a stimulation signal to stimulate the target tissue.

2. The electrical stimulation device according to claim 1, characterized in that, Determining the predicted sleep state of the target subject based on the respiratory data collected by the sensor includes: Based on the respiratory data, the predicted respiratory state of the target object in the next respiratory cycle is determined; Based on the posture data and electrical impedance data in the respiratory data, the sleep position signal of the target object is determined; Based on the motion data and intrathoracic pressure data in the respiratory data, the sleep movement signals of the target subject are determined; Based on the blood oxygen data in the respiratory data, the predicted respiratory state, sleep posture signals, and sleep movement signals, the predicted sleep state of the target object is determined by a sleep breathing model.

3. The electrical stimulation device according to claim 2, characterized in that, Determining the predicted respiratory state of the target object in the next respiratory cycle based on the respiratory data includes: Based on the respiratory data, determine the current respiratory state of the target object; Based on the target object's current respiratory state, determine the predicted respiratory state of the target object in the next respiratory cycle.

4. The electrical stimulation device according to claim 3, characterized in that, Determining the current respiratory state of the target object based on the respiratory data includes: In response to the determination that there is an interference signal in the respiratory signal of the target object, a sampling window is determined according to a preset respiratory frequency range and preset respiratory motion characteristics, wherein the respiratory signal is determined based on the motion data in the respiratory data; Based on the preset respiratory motion characteristics, feature extraction is performed on the respiratory signal within the sampling window, and feature weights are determined. The simulated respiratory signal is determined based on the reference respiratory rate and reference respiratory phase; The reference respiratory rate and reference respiratory phase are adjusted according to the feature weights to minimize the residual between the respiratory signal and the simulated respiratory signal, and to determine the current respiratory rate and current respiratory phase in the current respiratory state.

5. The electrical stimulation device according to claim 3, characterized in that, The step of determining the predicted respiratory state of the target object in the next respiratory cycle based on the target object's current respiratory state includes: Feature vectors are obtained by extracting features from the current breathing state. Based on the state vector corresponding to the current breathing state and the feature vector, the predicted breathing state of the target object in the next breathing cycle is predicted through a pre-constructed breathing state space model.

6. The electrical stimulation device according to claim 1, characterized in that, The step of determining the output parameters of the stimulation signal based on the predicted sleep state includes: In response to the predicted sleep state indicating that the target object has begun to enter sleep, the output parameters of the stimulation signal are determined based on the sleep data of the target object within a historical time window. In response to the predicted sleep state indicating that the target object is awake, the output of the stimulation signal is stopped.

7. The electrical stimulation device according to claim 2, characterized in that, The controller is also used for: Based on the difference between the actual sleep state and the predicted sleep state of the target object, a correction function for the sleep breathing model is determined. The sleep breathing model is modified based on the aforementioned correction function.

8. The electrical stimulation device according to claim 5, characterized in that, The controller is also used for: The respiratory state space model is updated based on the difference between the actual respiratory state and the predicted respiratory state of the target object in the next respiratory cycle. Based on the predicted respiratory state, the acquisition parameters of the sensor are adjusted, wherein the acquisition parameters of the sensor include at least one of the following: acquisition frequency, resolution, and power.

9. A control method for an electrical stimulation device, characterized in that, The method, applied to the electrical stimulation device as described in any one of claims 1 to 8, comprises: Respiratory data of a target object is acquired based on sensors, wherein the respiratory data includes at least one of the following: motion data, intrathoracic pressure data, electrical impedance data, blood oxygen data, and posture data; Based on the respiratory data, the controller determines the predicted sleep state of the target object; Based on the predicted sleep state, the output parameters of the stimulation signal are determined by the controller; Based on the output parameters of the stimulation signal, the controller controls the electrodes to output the stimulation signal.

10. The method according to claim 9, characterized in that, Determining the predicted sleep state of the target subject based on the respiratory data includes: Based on the respiratory data, the predicted respiratory state of the target object in the next respiratory cycle is determined; Based on the posture data and electrical impedance data in the respiratory data, the sleep position signal of the target object is determined; Based on the motion data and intrathoracic pressure data in the respiratory data, the sleep movement signals of the target subject are determined; Based on the blood oxygen data in the respiratory data, the predicted respiratory state, sleep posture signals, and sleep movement signals, the predicted sleep state of the target object is determined by a sleep breathing model.

Images