Sleep stimulation method, apparatus, device, medium, and program product

Abstract

This application provides a sleep stimulation method, apparatus, device, medium, and program product. The sleep stimulation method includes: acquiring a target EEG signal of a subject to be stimulated; acquiring a first candidate slow-wave event from the target EEG signal; acquiring target feature information related to the first candidate slow-wave event; acquiring a stimulation score based on the target feature information, the stimulation score being used to assess the appropriateness of applying sleep stimulation to the first candidate slow-wave event; predicting the target time when the target EEG signal reaches the preset target phase based on the phase information of the first candidate slow-wave event and the preset target phase when the stimulation score meets preset stimulation conditions; determining the stimulation time at which to apply sleep stimulation to the subject based on the target time; and applying sleep stimulation to the subject at the stimulation time. This allows sleep stimulation to be applied near or directly at the preset target phase, thereby improving the user's sleep quality.

Description

Technical Field

[0001] This application relates to the field of sleep assistance, and more specifically, to a sleep stimulation method, device, equipment, medium, and program product. Background Technology

[0002] In the field of sleep aids, especially during the deep sleep stage of non-rapid eye movement (NREM) sleep, external stimuli correlated with the slow-wave phase of the electroencephalogram (EEG) can enhance slow-wave activity and improve sleep quality to some extent. Common forms of stimulation include auditory stimulation, vibration stimulation, and bone conduction stimulation. Current technologies typically involve first detecting slow-wave events based on EEG signals, then predicting the target time of the target phase, and finally applying sleep stimulation at the corresponding target time.

[0003] Although some slow-wave events meet the basic detection threshold, their waveform stability is poor, resulting in excessive prediction error of the target time of the target phase. This leads to the application of sleep stimulation at the wrong phase, which not only fails to improve the user's sleep quality but may even wake the user from sleep. Summary of the Invention

[0004] The purpose of this application is to provide a sleep stimulation method, apparatus, device, medium, and program product to solve the problem that in related technologies, the application of sleep stimulation at the wrong phase due to the lack of screening of slow-wave events may wake the user from sleep.

[0005] In a first aspect, embodiments of this application provide a sleep stimulation method, including: Acquire the target EEG signal of the subject to be stimulated; Obtain a first candidate slow-wave event from the target EEG signal; Obtain target feature information related to the first candidate slow-wave event; the target feature information includes at least one of the following: phase correlation feature information, consistency feature information, physiological stability feature information, historical stimulus effect feature information, and target EEG signal segments obtained from the target EEG signal based on a time window related to the first candidate slow-wave event; A stimulation score is obtained based on the target feature information, and the stimulation score is used to assess the suitability of implementing sleep stimulation on the first candidate slow wave event. When the stimulation score meets the preset stimulation conditions, the target time when the target EEG signal reaches the preset target phase is predicted based on the phase information of the first candidate slow wave event and the preset target phase. The timing for applying sleep stimulation to the subject to be stimulated is determined based on the target time. At the stimulation time, a sleep stimulus is applied to the subject to be stimulated.

[0006] In the above implementation, a stimulation score is obtained before the target EEG signal reaches the target phase at the target time, i.e., the suitability of applying sleep stimulation to the first candidate slow-wave event is determined. Only when the stimulation score meets the preset stimulation conditions is the target EEG signal's target phase at the predicted time, and the stimulation time is determined based on the target time, so that sleep stimulation can be applied to the subject at the stimulation time. In this way, the first candidate slow-wave events in the target EEG signal can be effectively screened based on the stimulation score, and the suitability of applying sleep stimulation to the first candidate slow-wave events can be determined. This is to improve the user's sleep quality and avoid blindly determining the target time when the first candidate slow-wave events are not suitable for sleep stimulation, thus reducing energy consumption.

[0007] Secondly, this application provides a sleep stimulation device, comprising: The first acquisition module is configured to acquire the target EEG signal of the subject to be stimulated. The second acquisition module is configured to acquire a first candidate slow-wave event from the target EEG signal; The third acquisition module is configured to acquire target feature information related to the first candidate slow-wave event; the target feature information includes at least one of the following: phase correlation feature information, consistency feature information, physiological stability feature information, historical stimulus effect feature information, and target EEG signal segments acquired from the target EEG signal based on a time window related to the first candidate slow-wave event; The fourth acquisition module is configured to acquire a stimulus score based on the target feature information; the stimulus score is used to assess the suitability of implementing sleep stimulation on the first candidate slow-wave event. The prediction module is configured to, when the stimulus score meets the preset stimulus conditions, predict the target time when the target EEG signal reaches the preset target phase based on the phase information of the first candidate slow wave event and the preset target phase. The determination module is configured to determine the stimulation time for applying sleep stimulation to the subject to be stimulated based on the target time; A stimulation module is configured to apply sleep stimulation to the object to be stimulated at the stimulation time. In a third aspect, this application provides an electronic device including a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the sleep stimulation method of the first aspect described above.

[0008] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the sleep stimulation method of the first aspect described above.

[0009] Fifthly, this application provides a computer program product comprising a computer program that, when executed by a processor, implements the sleep stimulation method of the first aspect described above. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A schematic flowchart of a sleep stimulation method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a sleep stimulation device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0012] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0013] Example 1: To address the problem in related technologies where excessively large prediction errors in the target time of the target phase lead to the application of sleep stimulation at the wrong phase, potentially waking the user from sleep, this application provides a sleep stimulation method. See also... Figure 1 As shown, Figure 1 This is a flowchart illustrating the sleep stimulation method provided in the embodiments of this application, including: Step S101: Obtain the target EEG signal of the subject to be stimulated.

[0014] In this embodiment, multiple candidate EEG signals can be acquired, and any one of the candidate EEG signals can be selected as the target EEG signal. The multiple candidate EEG signals are obtained by detecting the subject in a sleep state through different EEG channels at the same time period.

[0015] The alternative EEG signals can be scalp EEG signals, ear EEG signals, or a combination of both. Scalp EEG signals refer to EEG signals collected by electrodes placed on the head, forehead, near the mastoid process, or other scalp locations. Ear EEG signals refer to EEG signals collected by electrodes placed in the ear canal, concha, cymba concha, behind the ear, or around the ear. Step S102: Obtain the first candidate slow wave event from the target EEG signal.

[0016] In this embodiment, a candidate slow-wave event refers to an event identified from an electroencephalogram (EEG) signal that may correspond to a slow-wave activity. The first candidate slow-wave event is the candidate slow-wave event identified from the target EEG signal that may correspond to a slow-wave activity.

[0017] In this embodiment, the EEG signal can be filtered according to a preset frequency band, and then peak-valley detection can be performed on the filtered EEG signal to obtain candidate slow-wave events. Correspondingly, a candidate slow-wave event is a waveform segment from one peak value to the next adjacent valley value. Alternatively, slow-wave segments can be identified based on zero-crossing positions; correspondingly, a candidate slow-wave event is a waveform segment from one negative zero-crossing to the next positive zero-crossing. Candidate slow-wave events can also be determined based on a combination of peak-valley amplitude and period length; correspondingly, a candidate slow-wave event is a waveform segment from one valley value to the next adjacent peak value. Alternatively, the EEG signal can be input into a preset slow-wave event recognition model to obtain candidate slow-wave events, or a preset reference slow-wave waveform can be used to perform a matching operation on the EEG signal to obtain candidate slow-wave events.

[0018] Step S103: Obtain target feature information related to the first candidate slow wave event; the target feature information related to the first candidate slow wave event includes at least one of the following: phase correlation feature information, consistency feature information, physiological stability feature information, historical stimulus effect feature information, and target EEG signal segments obtained from the target EEG signal based on the time window related to the first candidate slow wave event. Phase-related feature information may include at least one of the following: phase prediction error estimate, phase position confidence interval width, phase distribution entropy corresponding to the stimulus time, and waveform feature information of at least one candidate slow-wave sleep cycle in the target EEG signal segment.

[0019] The phase prediction error estimate can be determined based on the deviation between the actual phase of the historical EEG signal at the historical stimulus time and the preset target phase. The historical stimulus time is determined based on the historical target time, which is the target time for predicting that the historical EEG signal will reach the preset target phase. The phase prediction error can be used to characterize the estimation of the prediction error for the phase corresponding to the stimulus time, and to characterize the degree of uncertainty in the prediction of the phase corresponding to the stimulus time. The smaller the phase prediction error estimate, the smaller the expected deviation between the predicted phase and the actual phase, and the more accurate and reliable the phase prediction; the larger the phase prediction error estimate, the higher the uncertainty of the phase prediction.

[0020] The phase confidence interval width can be used to characterize the phase range covered by the preset confidence interval corresponding to the phase at the stimulus time. Correspondingly, the phase confidence interval width can be used to characterize the uncertainty of the phase prediction corresponding to the stimulus time. The smaller the phase confidence interval width, the more concentrated the phase prediction results, the narrower the confidence interval, the more certain the phase prediction, and the higher the reliability. Conversely, the larger the phase confidence interval width, the more dispersed the phase prediction results, the wider the confidence interval, and the higher the uncertainty of the phase prediction.

[0021] The phase distribution entropy corresponding to the stimulus moment is calculated based on the probability distribution of the phases corresponding to historical stimulus moments. The phase distribution entropy corresponding to the stimulus moment can be used to characterize the uncertainty of the phase prediction at that moment. A smaller entropy value indicates a more concentrated phase probability distribution, resulting in a more certain and reliable phase prediction; a larger entropy value indicates a more dispersed phase probability distribution, resulting in higher uncertainty in the phase prediction.

[0022] The consistency feature information includes multiple candidate EEG signals of the subject to be stimulated. These multiple candidate EEG signals are obtained by different EEG channels detecting the subject to be stimulated at the same time. The target EEG signal is any one of the multiple candidate EEG signals.

[0023] The historical stimulus effect feature information includes the preset historical stimulus effect and the frequency domain feature information obtained after performing Fourier transform on the target EEG signal segment. The historical stimulus effect includes the historical response information of the subject to be stimulated within a preset time period after the sleep stimulus was applied to the subject before the current time. The historical response information is used to characterize the magnitude of improvement or decline in sleep quality.

[0024] Historical stimulus effects may include, but are not limited to: changes in slow wave amplitude after several recent stimuli; changes in slow wave frequency power after several recent stimuli; phase hit performance after several recent stimuli; arousal status after several recent stimuli; and body movement after several recent stimuli.

[0025] Phase hit performance characterizes the match between the actual output time of the stimulus and the predetermined target phase after the stimulus is applied to the subject for sleep. It can be characterized by at least one of the following: the deviation between the actual phase at which the sleep stimulus is applied and the predetermined target phase; whether the phase at which the sleep stimulus is applied falls within the predetermined phase tolerance range; the average phase deviation of several recent stimuli; and the phase hit rate of several recent stimuli.

[0026] Physiological information includes physiological signals of the subject to be stimulated within a time window associated with the first candidate slow-wave event, including at least one of head vibration frequency, respiratory rate, and heart rate.

[0027] Step S104: Obtain a stimulus score based on the target feature information. The stimulus score is used to assess the suitability of implementing sleep stimulation on the first candidate slow-wave event. In one optional implementation of this application, the target feature information includes phase-related feature information, which includes waveform feature information of at least two candidate slow-wave sleep cycles in the target EEG signal segment. Obtaining a stimulus score based on the target feature information includes: obtaining a first similarity between at least two candidate slow-wave sleep cycles based on their waveform feature information; and obtaining a stimulus score based on the first similarity. Thus, by obtaining the first similarity between two candidate slow-wave sleep cycles, the waveform stability of the target EEG signal segment can be assessed. This allows for prediction when the target EEG signal reaches a preset target phase at a target time, provided the target EEG signal segment waveform is stable, thereby reducing prediction deviation at the target time and improving the overall accuracy of phase prediction.

[0028] In this embodiment, waveform similarity can be obtained between waveform feature information of at least two candidate slow-wave sleep cycles, and this waveform similarity is the first similarity. Alternatively, the cycle length of each candidate slow-wave sleep cycle can be obtained from the waveform feature information of each candidate slow-wave sleep cycle; the first similarity between the at least two candidate slow-wave sleep cycles can be obtained based on the difference in cycle length between the at least two candidate slow-wave sleep cycles; the first similarity is inversely proportional to the difference in cycle length. Alternatively, the interval duration between adjacent peaks and valleys in each candidate slow-wave sleep cycle can be obtained from the waveform feature information of each candidate slow-wave sleep cycle; the first similarity between the at least two candidate slow-wave sleep cycles can be obtained based on the difference in interval duration between the at least two candidate slow-wave sleep cycles; the first similarity is inversely proportional to the difference in interval duration. In this way, the first similarity can be obtained.

[0029] In the embodiments of this application, the first similarity can be directly used as the stimulus score.

[0030] In one optional implementation of this application, when the target feature information includes phase-related feature information, and the phase-related feature information includes waveform feature information of at least one candidate slow-wave sleep cycle in the target EEG signal segment, obtaining a stimulus score based on the target feature information includes: obtaining a second similarity between each candidate slow-wave sleep cycle and the reference slow-wave waveform based on the waveform feature information of each candidate slow-wave sleep cycle and the waveform feature information of a preset reference slow-wave waveform; and obtaining a stimulus score based on the second similarity. Thus, by obtaining the second similarity between the candidate slow-wave sleep cycle and the reference slow-wave waveform, a similarity assessment between the target EEG signal segment and the reference slow-wave waveform can be achieved. This allows for prediction when the target EEG signal reaches the target phase at a target time, where the target EEG signal segment and the reference slow-wave waveform are highly matched, thereby reducing the prediction deviation of the target time and improving the overall accuracy of phase prediction.

[0031] Similarly, in this embodiment, for the waveform feature information of each candidate slow-wave sleep cycle, the waveform similarity between the waveform feature information of the candidate slow-wave sleep cycle and the reference slow-wave waveform can be obtained. This waveform similarity is then considered the second similarity. Alternatively, the cycle length of the candidate slow-wave sleep cycle can be obtained from its waveform feature information; a second similarity can be obtained based on the second cycle length difference between the cycle length of the candidate slow-wave sleep cycle and the cycle length of the reference slow-wave waveform; the second similarity is inversely proportional to the second cycle length difference. Alternatively, the interval duration between adjacent peaks and valleys in the candidate slow-wave sleep cycle can be obtained from its waveform feature information; a second similarity can be obtained based on the second interval duration difference between the interval duration between adjacent peaks and valleys in the candidate slow-wave sleep cycle and the interval duration between adjacent peaks and valleys in the reference slow-wave waveform; the second similarity is inversely proportional to the second interval duration difference. This allows for the acquisition of the second similarity.

[0032] Similarly, in the embodiments of this application, the second similarity can also be directly used as the stimulus score.

[0033] The reference slow wave waveform can be user-defined or an average waveform template obtained based on historical slow wave statistics.

[0034] In one optional implementation of this application, when the target feature information includes consistency feature information, and the consistency feature information includes multiple candidate EEG signals of the subject to be stimulated, a third similarity can be obtained among the multiple candidate EEG signals; a stimulation score is then obtained based on the third similarity. Since the multiple candidate EEG signals are obtained by detecting the subject in a sleep state using different EEG channels at the same time period, obtaining the third similarity among the multiple candidate EEG signals verifies the validity of the target EEG signal, thereby ensuring the accuracy of subsequent phase prediction.

[0035] In detail, phase consistency among multiple candidate EEG signals can be obtained. Phase consistency characterizes the degree of phase alignment of multiple candidate EEG signals in the time domain. The degree of phase alignment is positively correlated with the temporal overlap of peak and trough positions in the waveform features of multiple candidate EEG signals; among them, the third similarity is phase consistency. Since the higher the temporal overlap of peak and trough positions of multiple candidate EEG signals, the more similar these candidate EEG signals are in phase rhythm, indirectly reflecting the stronger the confidence of the currently captured EEG signal. Therefore, phase consistency among multiple candidate EEG signals can be used to verify whether the target EEG signal is reliable and trustworthy.

[0036] Alternatively, the similarity of waveform features among multiple candidate EEG signals can be obtained; this similarity is termed the third similarity. Since the more similar the waveform features of multiple candidate EEG signals, the more convergent their phase rhythms are, indirectly reflecting a higher confidence level in the currently captured EEG signal. Therefore, the similarity of waveform features among multiple candidate EEG signals can be used to verify the reliability and credibility of the target EEG signal.

[0037] Alternatively, multiple second candidate slow-wave events can be obtained from each candidate EEG signal, along with the timestamp of each second candidate slow-wave event in its respective candidate EEG signal. Based on the timestamp of each second candidate slow-wave event in its respective candidate EEG signal, time alignment matching is performed on all second candidate slow-wave events across different candidate EEG signals to determine at least one target candidate slow-wave event set with time alignment matching from all candidate EEG signals. Time alignment matching is performed on all second candidate slow-wave events in the target candidate slow-wave event set. A third similarity is obtained between multiple candidate EEG signals based on the number of second candidate slow-wave events in the target candidate slow-wave event set. The third similarity is positively correlated with a target ratio, which represents the ratio between the number of all second candidate slow-wave events in the target candidate slow-wave event set and the number of all candidate EEG signals.

[0038] The more second-candidate slow-wave events in the target candidate slow-wave event set, the more similar these candidate EEG signals are in phase rhythm, indirectly reflecting a stronger confidence level in the currently captured EEG signal. Therefore, a third similarity can be obtained between multiple candidate EEG signals based on the number of second-candidate slow-wave events in the target candidate slow-wave event set, and further, the reliability and credibility of the target EEG signal can be verified based on the third similarity.

[0039] In this embodiment, if multiple candidate EEG signals are identical in phase rhythm, the total number of second candidate slow-wave events in any target candidate slow-wave event set should be equal to the total number of candidate EEG signals. However, if any candidate EEG signal is subject to external interference, its phase rhythm may deviate from that of other candidate EEG signals, resulting in the total number of second candidate slow-wave events in the target candidate slow-wave event set being less than the total number of candidate EEG signals. Therefore, a third similarity can be obtained between multiple candidate EEG signals based on the number of second candidate slow-wave events in the target candidate slow-wave event set.

[0040] In another optional implementation of this application, when the target feature information includes historical stimulus effect feature information and target EEG signal fragments, the target EEG signal fragments and historical stimulus effect feature information can be input into a preset stimulus effectiveness probability prediction model to obtain the stimulus effectiveness probability. The stimulus effectiveness probability is used to characterize the probability that the sleep quality of the subject will improve after applying sleep stimulation. Finally, a stimulus score is obtained based on the stimulus effectiveness probability. In this way, the probability that the sleep quality of the subject will improve after applying sleep stimulation can be estimated, that is, it can be determined whether applying sleep stimulation to the subject at the target time can achieve the expected effect.

[0041] In another optional implementation of this application, where the target feature information includes physiological stability feature information, and the physiological feature information includes the physiological signals of the subject to be stimulated within a time window related to the first candidate slow-wave event, the arousal risk probability of the subject to be stimulated can be obtained based on the physiological signals. The arousal risk probability is used to assess the probability that the subject to be stimulated will be aroused after sleep stimulation is applied. Finally, a stimulation score is obtained based on the arousal risk probability. The arousal risk probability is positively correlated with at least one of head vibration frequency, respiratory rate, and heart rate. In this way, the probability that the subject to be stimulated will be aroused after sleep stimulation is applied at the stimulation time can be predicted, so as to subsequently determine whether to apply sleep stimulation to the subject at the stimulation time based on the arousal risk probability.

[0042] In another optional implementation of this application embodiment, when the target feature information includes phase-related feature information and the phase-related feature information includes a phase prediction error estimate, the stimulus score can be obtained based on the phase prediction error estimate; wherein the stimulus score is inversely proportional to the phase prediction error estimate.

[0043] In another optional implementation of this application embodiment, when the target feature information includes phase-related feature information and the phase-related feature information includes the phase position confidence interval width, the stimulus score can be obtained based on the phase position confidence interval width; wherein, the stimulus score is inversely proportional to the phase position confidence interval width.

[0044] In another optional implementation of this application embodiment, when the target feature information includes phase-related feature information, and the phase-related feature information includes the phase distribution entropy corresponding to the stimulus time, the stimulus score can be obtained based on the phase distribution entropy corresponding to the stimulus time; wherein, the stimulus score is inversely proportional to the phase distribution entropy corresponding to the stimulus time.

[0045] In the embodiments of this application, the various factors in the target feature information, namely phase correlation feature information, consistency feature information, physiological stability feature information, historical stimulus effect feature information, and target EEG signal segments obtained from the target EEG signal based on the time window related to the first candidate slow wave event, can be used individually or in combination for obtaining the stimulus score.

[0046] For example, in the embodiments of this application, one of the first similarity, second similarity, third similarity, and stimulus effectiveness probability can be directly used as the stimulus score. Alternatively, the stimulus score can be obtained by weighting the first similarity, second similarity, third similarity, stimulus effectiveness probability, phase distribution entropy corresponding to the stimulus time, phase position confidence interval width, phase prediction error estimate, and arousal risk probability according to preset weights.

[0047] Step S105: If the stimulus score meets the preset stimulus conditions, predict the target time when the target EEG signal reaches the preset target phase based on the phase information of the first candidate slow wave event and the preset target phase. The preset target phase represents the pre-set phase position where sleep stimulation is to be applied. The preset target phase can be the rising edge of a slow wave, near the peak, near the trough, near the zero crossover, or other preset phase regions.

[0048] Step S106: Determine the timing of applying sleep stimulation to the target subject based on the target timing.

[0049] In one optional implementation of this application, the stimulus time may be the same as the target time.

[0050] In another optional implementation of this application, the target time can be compensated based on device processing delay, signal transmission delay, transducer start-up delay, stimulus propagation delay, stimulus duration and / or individualized phase deviation calibration amount to determine the stimulus time.

[0051] Step S107: Apply sleep stimulation to the subject at the stimulation time.

[0052] Sleep stimulation can be one of the following: acoustic stimulation signals, vibration stimulation signals, or bone conduction stimulation signals.

[0053] Acoustic stimulus signals represent stimulus signals that act on the user's auditory system via air conduction, such as pitch, short-tone pulses, broadband noise segments, or combinations thereof. Vibrational stimulus signals represent stimulus signals that act on localized areas of the user's body via mechanical vibration, such as micro-vibrations output through headphones, headbands, eyeglass temples, the occiput, or other contact points. Bone conduction stimulus signals represent stimulus signals that are perceived by the user by coupling mechanical vibrations to the skull or periauricular structures via bone conduction units.

[0054] In this embodiment of the application, applying sleep stimulation to the subject at the stimulation time includes: The system determines whether applying sleep stimulation to the subject at the stimulation time satisfies preset constraints. These constraints include at least one of the following: the stimulation time falls within a preset allowed stimulation period; the interval between the stimulation time and the previous application of sleep stimulation is greater than or equal to a preset minimum stimulation interval; and the total number of stimulations within a unit time period does not reach a stimulation threshold. If applying sleep stimulation to the subject at the stimulation time satisfies the preset constraints, then sleep stimulation is applied. If applying sleep stimulation to the subject at the stimulation time does not satisfy the preset constraints, then sleep stimulation is not applied.

[0055] Because repeatedly applying sleep stimulation to a subject within a short period can easily cause the subject to wake up, it is essential to determine whether applying the sleep stimulation at that time meets preset constraints beforehand. Specifically, this means confirming whether the stimulation time falls within a preset allowed stimulation period, whether the interval between the stimulation time and the previous application of sleep stimulation is greater than or equal to a preset minimum stimulation interval, and whether the total number of stimulations within a given time period reaches a stimulation threshold. Only if these preset constraints are met will the sleep stimulation be applied. This prevents the subject from waking up due to repeated stimulation within a short period.

[0056] The sleep stimulation method provided in this application obtains a stimulation score before the target EEG signal reaches the target time of the preset target phase, i.e., it obtains the suitability of applying sleep stimulation to the first candidate slow-wave event. Only when the stimulation score meets the preset stimulation conditions is the target time of the target EEG signal to reach the preset target phase predicted, and the stimulation time is determined based on the target time, so as to apply sleep stimulation to the subject to be stimulated at the stimulation time. In this way, the first candidate slow-wave events in the target EEG signal can be effectively screened based on the stimulation score, and the suitability of applying sleep stimulation to the first candidate slow-wave events can be determined. This is to improve the user's sleep quality, and at the same time, it can avoid blindly determining the target time when the first candidate slow-wave events are not suitable for applying sleep stimulation, thus reducing energy consumption.

[0057] Example 2: This embodiment is based on Embodiment 1 and provides an illustrative example of this application: During the subject's sleep, physiological signals and alternative EEG signals are collected. The subject can be a human. In other words, physiological and EEG signals can be collected during the user's sleep.

[0058] The alternative EEG signals include, but are not limited to, ear EEG signals, scalp EEG signals, and other EEG signals that reflect slow-wave activity during sleep.

[0059] Physiological signals include, but are not limited to: inertial measurement unit (IMU) signals, head vibration signals, rate of change of posture, body motion indicators, respiratory rate, heart rate, pulse signals, and motion artifact indicators.

[0060] Based on the physiological signals obtained, micro-arousal indicators can be acquired. That is, the probability of a user's arousal risk can be obtained based on physiological signals.

[0061] Microarousal indicators are used to characterize the tendency for short-term awakenings or shallow sleep during sleep. Examples include short-term increases in high-frequency power, increased electromyographic activity, increased body movement, short-term changes in respiration or heart rate, and sudden changes in brain electrical activity. High-frequency power represents brain electrical frequencies higher than the slow-wave band. The process involves: acquiring a first candidate slow-wave event from any candidate EEG signal, i.e., acquiring a first candidate slow-wave event from the target EEG signal. Acquisition methods include, but are not limited to: filtering the target EEG signal according to a predetermined frequency band and then detecting peak-valley structures; identifying slow-wave segments based on zero-crossing locations; jointly determining based on peak-valley amplitude and period length; identifying slow-wave morphology based on a preset reference slow-wave waveform; and performing slow-wave event recognition based on a preset slow-wave event recognition model.

[0062] Candidate slow wave events can be characterized by information such as start point, end point, zero crossover point, peak point, valley point, half-cycle length, full cycle length, event window, or current phase.

[0063] For the first candidate slow-wave event, target feature information for calculating the stimulus score is obtained. Target feature information may include all candidate EEG signals, frequency domain feature information obtained after Fourier transforming the target EEG signal segment, preset historical stimulus effects, estimated phase prediction error, phase position confidence interval width, phase distribution entropy corresponding to the stimulus time, target EEG signal segments obtained from the target EEG signals to which the first candidate slow-wave event belongs based on a time window associated with the first candidate slow-wave event, and waveform feature information of at least one candidate slow-wave sleep cycle within the target EEG signal segment.

[0064] The phase confidence interval width characterizes the phase range covered by the preset confidence interval corresponding to the target time, and is used to characterize the uncertainty of the phase prediction corresponding to the target time. The smaller the phase confidence interval width, the more concentrated the phase prediction results, the narrower the confidence interval, the more certain the phase prediction, and the higher the reliability; the larger the phase confidence interval width, the more dispersed the phase prediction results, the wider the confidence interval, and the higher the uncertainty of the phase prediction.

[0065] The phase distribution entropy at the stimulus moment is a characterization of the entropy value calculated based on the probability distribution of the phase at the target moment. It is used to characterize the uncertainty of the phase prediction at the stimulus moment. The smaller the entropy value, the more concentrated the phase probability distribution, and the more certain and reliable the phase prediction; the larger the entropy value, the more dispersed the phase probability distribution, and the higher the uncertainty of the phase prediction.

[0066] Historical stimulus effects may include, but are not limited to: changes in slow wave amplitude after several recent stimuli; changes in slow wave frequency power after several recent stimuli; phase hit performance after several recent stimuli; arousal status after several recent stimuli; and body movement after several recent stimuli.

[0067] The target feature information may also include the peak and valley amplitude of the target EEG signal, the rising slope of the target EEG signal, the falling slope of the target EEG signal, the half-cycle length of the target EEG signal, the full cycle length of the target EEG signal, and the zero-crossing interval of the target EEG signal. Based on the waveform feature information of candidate slow-wave sleep cycles in the target feature information, the correlation between the waveform feature information of each candidate slow-wave sleep cycle and the waveform feature information of the preset reference slow-wave waveform can be obtained, that is, the second similarity can be obtained.

[0068] Based on the waveform feature information of two candidate slow-wave sleep cycles, the consistency of cycle length, peak-valley interval, and phase evolution of the at least two candidate slow-wave sleep cycle events can be obtained, which can also be used to obtain the first similarity.

[0069] It can obtain the correlation between multiple candidate EEG signals (e.g., scalp EEG signals and ear EEG signals), the phase consistency between multiple candidate EEG signals, and the detection consistency of multiple candidate EEG signals for the same candidate slow wave. In other words, it can obtain the third similarity between multiple candidate EEG signals based on target feature information.

[0070] In the embodiments of this application, the input features of the stimulus effectiveness probability prediction model may include, in addition to historical stimulus effects, target EEG signal segments, frequency domain feature information obtained after Fourier transforming the target EEG signal segments, and multimodal fusion features, etc.

[0071] Multimodal fusion features are formed by combining two or more signals from scalp EEG, ear EEG, head vibration, IMU, body movement, respiration, or heart rate. They can be obtained through feature splicing, correlation calculation, consistency analysis, weighted combination, or model fusion to improve the scoring accuracy of candidate slow wave events.

[0072] Based on the obtained input features, these features can be fed into the stimulus effective probability prediction model to obtain the stimulus effective probability.

[0073] Based on the obtained indicators such as first similarity, second similarity, third similarity, arousal risk probability, effective stimulus probability, phase distribution entropy corresponding to the stimulus time, phase position confidence interval width, and phase prediction error estimate, multiple indicators can be weighted according to preset weights to obtain a stimulus score.

[0074] For example, indicators that promote stimulus gains or improve phase hit rate (such as first similarity, second similarity, third similarity, and stimulus effectiveness probability) can be positively weighted; while indicators that indicate arousal risk, artifact risk, or low reliability (such as arousal risk probability, phase prediction error estimate, phase position confidence interval width, and phase distribution entropy corresponding to the stimulus time) can be negatively weighted.

[0075] In another embodiment, the stimulus score can also be directly output from a preset stimulus scoring model. The stimulus scoring model can be a statistical model, classification model, regression model, tree model, or neural network model. Indicators such as first similarity, second similarity, third similarity, arousal risk probability, stimulus effectiveness probability, estimated phase prediction error, phase confidence interval width, and phase distribution entropy corresponding to the stimulus time can be input into the stimulus scoring model to obtain the stimulus score.

[0076] If the stimulus score is higher than the overall score threshold, applying sleep stimulation to the subject at the stimulation time is permitted. Alternatively, if the first similarity is higher than a preset similarity threshold and the arousal risk probability is lower than a preset risk threshold, applying sleep stimulation to the subject at the stimulation time is permitted. Alternatively, if the stimulus effectiveness probability is greater than a preset effectiveness threshold, applying sleep stimulation to the subject at the stimulation time is permitted.

[0077] Alternatively, we can first determine if the first similarity is greater than or equal to a preset similarity threshold. If the first similarity is lower than the preset similarity threshold, then applying sleep stimulation to the subject at the stimulation time is not allowed. If the first similarity is greater than or equal to the preset similarity threshold, we then determine if the stimulus score is greater than or equal to a comprehensive score threshold. If the stimulus score is greater than or equal to the comprehensive score threshold, then applying sleep stimulation to the subject at the stimulation time is allowed. If the stimulus score is lower than the comprehensive score threshold, then applying sleep stimulation to the subject at the stimulation time is not allowed.

[0078] While allowing the application of sleep stimulation to the subject at the stimulation time, it is also possible to determine whether the application of sleep stimulation to the subject at the stimulation time meets preset constraints. The preset constraints include: stimulation refractory period, minimum stimulation interval, upper limit of stimulation times per unit time, i.e. stimulation threshold, and stimulation budget within a sleep stage.

[0079] The refractory period is a period of time during which stimulation is temporarily suspended after a sleep stimulus has been completed, in order to avoid excessive stimulation or increased interference.

[0080] If applying a sleep stimulus to the target at the stimulus time satisfies the preset constraint conditions, then a sleep stimulus will be applied to the target at the stimulus time.

[0081] If applying sleep stimulation to the subject at the stimulation time does not meet the preset constraints, then sleep stimulation will not be applied to the subject.

[0082] Sleep stimuli include, but are not limited to: auditory stimulation, vibration stimulation, bone conduction stimulation, and other non-invasive sensory stimuli.

[0083] In one embodiment, the acoustic stimulation signal may be a short tone pulse, a noise fragment, or a predetermined tone. Vibrational stimulation may be output from a vibration unit in the wearable device to the ear, head, or other contact area. Bone conduction stimulation signals may be output from a bone conduction transducer to the periauricular region, temporal region, mastoid region, or other locations suitable for bone conduction coupling.

[0084] After applying sleep stimulation to the user, response information within a preset time window can be collected. The response information includes EEG signal response and / or physiological signal response, and the parameters, stimulation threshold, or scoring model of the stimulation score can be updated based on the response information.

[0085] The response information may include at least one of the following: slow wave amplitude variation, slow wave frequency band power variation, probability or amplitude variation of subsequent slow wave events, high frequency power variation, micro-arousal index, body movement variation, and actual stimulus phase error.

[0086] Based on the response information, the parameters, stimulus thresholds, or scoring models of the stimulus scores can be updated as follows: updating the weights in the stimulus scores, updating the stimulus thresholds, adjusting the stimulus refractory period, adjusting the stimulus frequency, updating the individualized phase deviation calibration, and updating the model parameters.

[0087] For example: If the slow wave amplitude increases after stimulation, the weight of the benefit term related to the characteristics of the current event is increased; If the high-frequency power increases, the microarousal index rises, or the body movement increases after stimulation, the risk-related weight should be increased or the stimulation threshold should be increased. If the actual stimulus phase error continues to deviate in the same direction in the near future, update the individualized phase deviation calibration. If the number of stimulations per unit time is too high, resulting in a decrease in returns, then the refractory period should be extended or the stimulation frequency reduced.

[0088] Example 3: Based on the same inventive concept, this application also provides a sleep stimulation device 200 in its embodiments. Please refer to... Figure 2 As shown, Figure 2 It shows the use of Figure 1 The sleep device of the method shown. It should be understood that the specific functions of device 200 can be found in the description above; to avoid repetition, detailed descriptions are appropriately omitted here. Device 200 includes at least one software functional module that can be stored in memory or embedded in the operating system of device 200 in the form of software or firmware. Specifically: See Figure 2 As shown, the device 200 includes: a first acquisition module 201, a second acquisition module 202, a third acquisition module 203, a fourth acquisition module 204, a prediction module 205, a determination module 206, and a stimulation module 207. Wherein: The first acquisition module 201 is configured to acquire the target EEG signal of the subject to be stimulated; The second acquisition module 202 is configured to acquire a first candidate slow-wave event from the target EEG signal; The third acquisition module 203 is configured to acquire target feature information related to the first candidate slow-wave event; the target feature information includes at least one of the following: phase correlation feature information, consistency feature information, physiological stability feature information, historical stimulus effect feature information, and target EEG signal segments acquired from the target EEG signal based on a time window related to the first candidate slow-wave event; The fourth acquisition module 204 is configured to acquire a stimulus score based on target feature information. The stimulus score is used to assess the suitability of implementing sleep stimulation on the first candidate slow-wave event. The prediction module 205 is configured to predict the target time when the target EEG signal reaches the preset target phase based on the phase information of the first candidate slow wave event and the preset target phase, provided that the stimulus score meets the preset stimulus conditions. The determination module 206 is configured to determine the timing at which the sleep stimulus is applied to the target subject based on the target timing; Stimulation module 207 is configured to apply sleep stimulation to the subject at the stimulation time.

[0089] In one feasible embodiment of this application, the target feature information includes phase-related feature information, which includes waveform feature information of at least two candidate slow-wave sleep cycles in the target EEG signal segment. The fourth acquisition module 204 can be specifically configured to: acquire a first similarity between at least two candidate slow-wave sleep cycles based on the waveform feature information of at least two candidate slow-wave sleep cycles; and acquire a stimulus score based on the first similarity.

[0090] In one feasible embodiment of this application, the fourth acquisition module 204 may be specifically configured to: acquire the waveform similarity between waveform feature information of at least two candidate slow-wave sleep cycles; the waveform similarity between waveform feature information of at least two candidate slow-wave sleep cycles is the first similarity.

[0091] In one feasible embodiment of this application, the fourth acquisition module 204 may be specifically configured to: acquire the cycle length of each candidate slow-wave sleep cycle from the waveform feature information of each candidate slow-wave sleep cycle; acquire the first similarity between the at least two candidate slow-wave sleep cycles based on the difference in cycle length between the at least two candidate slow-wave sleep cycles; the first similarity is inversely proportional to the difference in cycle length.

[0092] In one feasible embodiment of this application, the fourth acquisition module 204 may be specifically configured to: acquire the interval duration between adjacent peaks and valleys in each candidate slow-wave sleep cycle from the waveform feature information of each candidate slow-wave sleep cycle; acquire the first similarity between the at least two candidate slow-wave sleep cycles based on the difference in interval duration between the at least two candidate slow-wave sleep cycles; the first similarity is inversely proportional to the difference in interval duration.

[0093] In one feasible embodiment of this application, where the target feature information includes phase-related feature information, and the phase-related feature information includes waveform feature information of at least one candidate slow-wave sleep cycle in the target EEG signal segment, the fourth acquisition module 204 can be specifically configured to: acquire a second similarity between each candidate slow-wave sleep cycle and the reference slow-wave waveform based on the waveform feature information of each candidate slow-wave sleep cycle and the waveform feature information of a preset reference slow-wave waveform; and acquire a stimulus score based on the second similarity.

[0094] In one feasible embodiment of this application, the target feature information includes consistency feature information, which includes multiple candidate EEG signals. The multiple candidate EEG signals are obtained by detecting the object to be stimulated in different EEG channels at the same time period. The target EEG signal is any one of the multiple candidate EEG signals. The fourth acquisition module 204 can be specifically configured to: acquire the third similarity between the multiple candidate EEG signals; and acquire the stimulation score based on the third similarity.

[0095] In one feasible embodiment of this application, the fourth acquisition module 204 can be specifically configured to: acquire the phase consistency among multiple candidate EEG signals, wherein the phase consistency is used to characterize the phase alignment degree of the multiple candidate EEG signals in the time domain, and the phase alignment degree is positively correlated with the temporal overlap of the peak and valley positions in the waveform feature information of the multiple candidate EEG signals; wherein, the third similarity is the phase consistency.

[0096] In one feasible embodiment of this application, the fourth acquisition module 204 can be specifically configured to: acquire the similarity of multiple candidate EEG signals in waveform feature information; the similarity of multiple candidate EEG signals in waveform feature information is a third similarity.

[0097] In one feasible embodiment of this application, the fourth acquisition module 204 may be specifically configured to: acquire multiple second candidate slow-wave events from each candidate EEG signal, and the timestamp of each second candidate slow-wave event in its respective candidate EEG signal; based on the timestamp of each second candidate slow-wave event in its respective candidate EEG signal, perform time alignment matching on all second candidate slow-wave events in different candidate EEG signals, and determine at least one target candidate slow-wave event set with time alignment matching from all candidate EEG signals; perform time alignment matching on all second candidate slow-wave events in the target candidate slow-wave event set; acquire a third similarity between multiple candidate EEG signals according to the number of second candidate slow-wave events in the target candidate slow-wave event set; the third similarity is positively correlated with a target ratio, and the target ratio represents the ratio between the number of all second candidate slow-wave events in the target candidate slow-wave event set and the number of all candidate EEG signals.

[0098] In one feasible embodiment of this application, the target feature information includes historical stimulus effect feature information and target EEG signal segments; the historical stimulus effect feature information includes preset historical stimulus effects and frequency domain feature information obtained after performing Fourier transform on the target EEG signal segments; the historical stimulus effects include the historical response information of the subject to be stimulated within a preset time period after applying sleep stimulation to the subject before the current time; the historical response information is used to characterize the magnitude of improvement or decline in sleep quality; the fourth acquisition module 204 can be specifically configured to: input the target EEG signal segments and historical stimulus effect feature information into a preset stimulus effectiveness probability prediction model to obtain the stimulus effectiveness probability; the stimulus effectiveness probability is used to characterize the probability of improvement in sleep quality of the subject to be stimulated after applying sleep stimulation to the subject; and obtain a stimulus score based on the stimulus effectiveness probability.

[0099] In one feasible embodiment of this application, the target feature information includes physiological stability feature information, which includes physiological signals of the subject to be stimulated within a time window related to the first candidate slow-wave event. The physiological signals include at least one of head vibration frequency, respiratory rate, and heart rate. The fourth acquisition module 204 can be specifically configured to: acquire the arousal risk probability of the subject to be stimulated based on the physiological signals; the arousal risk probability is used to assess the probability that the subject to be stimulated will be aroused after sleep stimulation is applied; the arousal risk probability is positively correlated with at least one of head vibration frequency, respiratory rate, and heart rate. A stimulation score is obtained based on the arousal risk probability.

[0100] In one feasible embodiment of this application, the target feature information includes phase-related feature information, which includes a phase prediction error estimate. The phase prediction error estimate is determined based on the deviation between the actual phase of the historical EEG signal at the historical stimulus time and the preset target phase. The historical stimulus time is determined based on the historical target time, which is the target time for predicting that the historical EEG signal will reach the preset target phase. The fourth acquisition module 204 can be specifically configured to: acquire a stimulus score based on the phase prediction error estimate; wherein the stimulus score is inversely proportional to the phase prediction error estimate.

[0101] In one feasible embodiment of this application, the target feature information includes phase-related feature information, which includes the phase position confidence interval width. The phase position confidence interval width characterizes the degree of uncertainty as to whether the target EEG signal is in the preset target phase at the stimulation time. The fourth acquisition module 204 can be specifically configured to: acquire a stimulation score based on the phase position confidence interval width; wherein, the stimulation score is inversely proportional to the phase position confidence interval width.

[0102] In one feasible embodiment of this application, the target feature information includes phase-related feature information, which includes the phase distribution entropy corresponding to the stimulus time. The phase distribution entropy corresponding to the stimulus time is an entropy value calculated based on the probability distribution of the phase corresponding to the historical stimulus time. The historical stimulus time is determined based on the historical target time, which is the target time when the predicted historical EEG signal reaches the preset target phase. The fourth acquisition module 204 can be specifically configured to: acquire a stimulus score based on the phase distribution entropy corresponding to the stimulus time; wherein, the stimulus score is inversely proportional to the phase distribution entropy corresponding to the stimulus time.

[0103] In one feasible embodiment of this application, the stimulation module 207 can be specifically configured to: determine whether applying sleep stimulation to the subject at the stimulation time satisfies preset constraints; the preset constraints include at least one of the following: the stimulation time is within a preset allowed stimulation period, the interval between the stimulation time and the previous time of applying sleep stimulation is greater than or equal to a preset minimum stimulation interval, and the total number of stimulations within a unit time period does not reach a stimulation threshold; if applying sleep stimulation to the subject at the stimulation time satisfies the preset constraints, then applying sleep stimulation to the subject at the stimulation time.

[0104] In one feasible embodiment of this application, the device 200 may further include a response module configured to acquire response information of the object to be stimulated in response to a stimulus; and, if the response information indicates a decrease in the sleep quality of the object to be stimulated, to increase the minimum stimulation interval and / or decrease the stimulation threshold.

[0105] It should be understood that, for the sake of brevity, some of the content described in Embodiment 1 will not be repeated in this embodiment.

[0106] Example 4: Based on the same inventive concept, this embodiment provides an electronic device, see [link to relevant documentation]. Figure 3 As shown, it includes a processor 301 and a memory 302. Wherein: The processor 301 is used to execute one or more programs stored in the memory 302 to implement the sleep method described above.

[0107] It is understandable that processor 301 can be a processor core or processor chip, or other circuitry capable of program configuration and execution. Memory 302 can be RAM (Random Access Memory), ROM (Read-Only Memory), flash memory, etc., but this is not a limitation.

[0108] It's understandable. Figure 3 The structure shown is for illustrative purposes only; the electronic device may also include components that are more advanced than those shown. Figure 3 The more or fewer components shown, or having the same Figure 3 Different configurations are shown. For example, it may also have an internal communication bus for communication between the processor 301 and the memory 302; or it may have an external communication interface, such as a USB (Universal Serial Bus) interface, a CAN (Controller Area Network) bus interface, etc.; or it may have an information display component such as a display screen, but this is not a limitation.

[0109] For example, the electronic device may be headphones, sleep earplugs, smart glasses, headbands, or other sleep wearable devices.

[0110] Based on the same inventive concept, this embodiment also provides a computer-readable storage medium, such as a floppy disk, optical disk, hard disk, flash memory, USB flash drive, SD (Secure Digital Memory Card), MMC (Multimedia Card), etc., in which one or more programs implementing the above steps are stored. These one or more programs can be executed by one or more processors to implement the above sleep stimulation method. Further details will not be elaborated here.

[0111] Based on the same inventive concept, this embodiment also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described sleep stimulation method.

[0112] For example, a computer program product may be an installation package or a program package.

[0113] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0114] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0115] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0116] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0117] In this article, "multiple" refers to two or more.

[0118] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A sleep stimulation method, characterized in that, include: Acquire the target EEG signal of the subject to be stimulated; Obtain a first candidate slow-wave event from the target EEG signal; Obtain target feature information related to the first candidate slow-wave event; the target feature information includes at least one of the following: phase correlation feature information, consistency feature information, physiological stability feature information, historical stimulus effect feature information, and target EEG signal segments obtained from the target EEG signal based on a time window related to the first candidate slow-wave event; A stimulation score is obtained based on the target feature information; the stimulation score is used to assess the appropriateness of implementing sleep stimulation on the first candidate slow-wave event; When the stimulation score meets the preset stimulation conditions, the target time when the target EEG signal reaches the preset target phase is predicted based on the phase information of the first candidate slow wave event and the preset target phase. The timing for applying sleep stimulation to the subject to be stimulated is determined based on the target time. At the stimulation time, a sleep stimulus is applied to the subject to be stimulated.

2. The method according to claim 1, characterized in that, The target feature information includes phase-related feature information, which includes waveform feature information of at least two candidate slow-wave sleep cycles in the target EEG signal segment. Stimulus scores are obtained based on the target feature information, including: Based on the waveform feature information of the at least two candidate slow-wave sleep cycles, a first similarity is obtained between the at least two candidate slow-wave sleep cycles. The stimulus score is obtained based on the first similarity.

3. The method according to claim 2, characterized in that, Based on the waveform feature information of the at least two candidate slow-wave sleep cycles, a first similarity is obtained between the at least two candidate slow-wave sleep cycles, including: The waveform similarity between the waveform feature information of the at least two candidate slow-wave sleep cycles is obtained; the waveform similarity between the waveform feature information of the at least two candidate slow-wave sleep cycles is the first similarity.

4. The method according to claim 2, characterized in that, Based on the waveform feature information of the at least two candidate slow-wave sleep cycles, a first similarity is obtained between the at least two candidate slow-wave sleep cycles, including: The cycle length of each candidate slow-wave sleep cycle is obtained from the waveform feature information of each candidate slow-wave sleep cycle. A first similarity between the at least two candidate slow-wave sleep cycles is obtained based on the cycle length difference between the at least two candidate slow-wave sleep cycles; the first similarity is inversely proportional to the cycle length difference.

5. The method according to claim 2, characterized in that, Based on the waveform feature information of the at least two candidate slow-wave sleep cycles, a first similarity is obtained between the at least two candidate slow-wave sleep cycles, including: From the waveform feature information of each candidate slow-wave sleep cycle, obtain the interval duration between adjacent peaks and valleys in each candidate slow-wave sleep cycle; A first similarity between the at least two candidate slow-wave sleep cycles is obtained based on the difference in the interval duration between the at least two candidate slow-wave sleep cycles; the first similarity is inversely proportional to the difference in the interval duration.

6. The method according to claim 1, characterized in that, The target feature information includes the phase-related feature information, which includes waveform feature information of at least one candidate slow-wave sleep cycle in the target EEG signal segment. A stimulus score is obtained based on the target feature information, including: Based on the waveform feature information of each candidate slow-wave sleep cycle and the waveform feature information of a preset reference slow-wave waveform, a second similarity between each candidate slow-wave sleep cycle and the reference slow-wave waveform is obtained. The stimulus score is obtained based on the second similarity.

7. The method according to claim 1, characterized in that, The target feature information includes consistency feature information, which includes multiple candidate EEG signals of the subject to be stimulated. The multiple candidate EEG signals are obtained by detecting the subject to be stimulated by different EEG channels in the same time period. The target EEG signal is any one of the multiple candidate EEG signals. The stimulus score is obtained based on the target feature information, including: Obtain the third similarity among the multiple candidate EEG signals; The stimulus score is obtained based on the third similarity.

8. The method according to claim 7, characterized in that, Obtaining the third similarity among the plurality of candidate EEG signals includes: The phase consistency among the plurality of candidate EEG signals is obtained. The phase consistency is used to characterize the phase alignment degree of the plurality of candidate EEG signals in the time domain. The phase alignment degree is positively correlated with the temporal overlap of the peak and valley positions in the waveform feature information of the plurality of candidate EEG signals. The third similarity is the phase consistency.

9. The method according to claim 7, characterized in that, Obtaining the third similarity among the plurality of candidate EEG signals includes: The similarity of the multiple candidate EEG signals in waveform feature information is obtained; the similarity of the multiple candidate EEG signals in waveform feature information is the third similarity.

10. The method according to claim 7, characterized in that, Obtaining the third similarity among the plurality of candidate EEG signals includes: Multiple second candidate slow-wave events are obtained from each of the candidate EEG signals, and the timestamp of each second candidate slow-wave event in its respective candidate EEG signal is obtained. Based on the timestamp of each second candidate slow wave event in its respective candidate EEG signal, time alignment matching is performed on all second candidate slow wave events in different candidate EEG signals to determine at least one target candidate slow wave event set with time alignment matching from all the candidate EEG signals; time alignment matching is performed on all second candidate slow wave events in the target candidate slow wave event set. A third similarity is obtained among the plurality of candidate EEG signals based on the number of second candidate slow-wave events in the target candidate slow-wave event set; the third similarity is positively correlated with a target ratio, which represents the ratio between the number of all second candidate slow-wave events in the target candidate slow-wave event set and the number of all candidate EEG signals.

11. The method according to claim 1, characterized in that, The target feature information includes the historical stimulus effect feature information and the target EEG signal segment; the historical stimulus effect feature information includes a preset historical stimulus effect and frequency domain feature information obtained after performing a Fourier transform on the target EEG signal segment; the historical stimulus effect includes the historical response information of the subject to be stimulated within a preset time period after applying sleep stimulation to the subject before the current time; the historical response information is used to characterize the magnitude of improvement or decline in sleep quality. Obtaining the stimulus score based on the target feature information includes: The target EEG signal fragment and the historical stimulus effect feature information are input into a preset stimulus effectiveness probability prediction model to obtain the stimulus effectiveness probability; the stimulus effectiveness probability is used to characterize the probability that the sleep quality of the subject will improve after applying sleep stimulation to the subject. The stimulus score is obtained based on the probability of the stimulus being effective.

12. The method according to claim 1, characterized in that, The target feature information includes the physiological stability feature information, and the physiological feature information includes the physiological signals of the subject to be stimulated within the time window related to the first candidate slow wave event, and the physiological signals include at least one of head vibration frequency, respiratory rate, and heart rate. The stimulus score is obtained based on the target feature information, including: The arousal risk probability of the subject to be stimulated is obtained based on the physiological signals; the arousal risk probability is used to assess the probability that the subject to be stimulated will be awakened after sleep stimulation is applied to the subject; the arousal risk probability is positively correlated with at least one of the head vibration frequency, respiratory rate and heart rate. The stimulus score is obtained based on the arousal risk probability.

13. The method according to claim 1, characterized in that, The target feature information includes phase-related feature information, which includes a phase prediction error estimate. The phase prediction error estimate is determined based on the deviation between the actual phase of the historical EEG signal at the historical stimulus time and the preset target phase. The historical stimulus time is determined based on a historical target time, which is the target time for predicting that the historical EEG signal will reach the preset target phase. A stimulus score is obtained based on the target feature information, including: The stimulus score is obtained based on the phase prediction error estimate; wherein the stimulus score is inversely proportional to the phase prediction error estimate.

14. The method according to claim 1, characterized in that, The target feature information includes phase-related feature information, which includes the phase position confidence interval width. The phase position confidence interval width characterizes the degree of uncertainty as to whether the target EEG signal is in the preset target phase at the stimulation time. The stimulus score is obtained based on the target feature information, including: The stimulus score is obtained based on the phase position confidence interval width; wherein the stimulus score is inversely proportional to the phase position confidence interval width.

15. The method according to claim 1, characterized in that, The target feature information includes phase-related feature information, which includes the phase distribution entropy corresponding to the stimulus time. The phase distribution entropy corresponding to the stimulus time is an entropy value calculated based on the probability distribution of the phase corresponding to historical stimulus times. The historical stimulus time is determined based on historical target times, which are the target times for predicting the arrival of historical EEG signals at the preset target phase. Obtaining a stimulus score based on the target feature information includes: The stimulus score is obtained based on the phase distribution entropy corresponding to the stimulus time; wherein the stimulus score is inversely proportional to the phase distribution entropy corresponding to the stimulus time.

16. The method according to any one of claims 1 to 15, characterized in that, Applying sleep stimulation to the subject at the stimulation time includes: Determine whether applying sleep stimulation to the subject at the stimulation time satisfies preset constraints; the preset constraints include at least one of the following: the stimulation time is within a preset allowed stimulation period, the interval between the stimulation time and the previous time of applying sleep stimulation is greater than or equal to a preset minimum stimulation interval, and the total number of stimulations within a unit time period does not reach a stimulation threshold. If applying sleep stimulation to the subject at the stimulation time satisfies the preset constraint condition, then applying sleep stimulation to the subject at the stimulation time.

17. The method according to claim 16, characterized in that, After applying sleep stimulation to the subject at the stimulation time, the method further includes: Obtain the response information of the object to be stimulated in response to the stimulus; If the response information indicates a decrease in the sleep quality of the subject to be stimulated, the minimum stimulation interval is increased, and / or the stimulation threshold is decreased.

18. A sleep stimulation device, characterized in that, include: The first acquisition module is configured to acquire the target EEG signal of the subject to be stimulated. The second acquisition module is configured to acquire a first candidate slow-wave event from the target EEG signal; The third acquisition module is configured to acquire target feature information related to the first candidate slow-wave event; the target feature information includes at least one of the following: phase correlation feature information, consistency feature information, physiological stability feature information, historical stimulus effect feature information, and target EEG signal segments acquired from the target EEG signal based on a time window related to the first candidate slow-wave event; The fourth acquisition module is configured to acquire a stimulus score based on the target feature information; the stimulus score is used to assess the suitability of implementing sleep stimulation on the first candidate slow-wave event. The prediction module is configured to, when the stimulus score meets the preset stimulus conditions, predict the target time when the target EEG signal reaches the preset target phase based on the phase information of the first candidate slow wave event and the preset target phase. The determination module is configured to determine the stimulation time for applying sleep stimulation to the subject to be stimulated based on the target time; The stimulation module is configured to apply sleep stimulation to the subject to be stimulated at the stimulation time.

19. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the sleep stimulation method according to any one of claims 1 to 17.

20. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the sleep stimulation method according to any one of claims 1 to 17.

21. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the sleep stimulation method according to any one of claims 1 to 17.

Images