How to assess the severity of obstructive sleep apnea and / or its effects on daytime sleepiness

Abstract

The invention relates to a method for determining obstructive sleep apnea and / or a severity scale of its impact, comprising the steps of: - defining a severity scale of obstructive sleep apnea and / or its impact, - providing two EEG measurement signals from an electroencephalogram at electroencephalogram points of the international 10-20 EEG system, - dividing the EEG measurement signals into frequency bands, - determining at least one inter-frequency modulation index using data from at least two different frequency bands, - determining the severity scale of obstructive sleep apnea and / or its impact using the modulation index between at least one frequency. The invention further relates to a device for carrying out this method.

Description

[Technical field]

[0001] The present invention relates to a method for determining severity criteria for obstructive sleep apnea and / or effects such as daytime sleepiness. [Background technology]

[0002] Sleep apnea is a nocturnal respiratory disorder in which prolonged or partial cessation of breathing occurs repeatedly during sleep. By definition, a prolonged or partial cessation of breathing is a cessation or partial cessation of breathing that lasts longer than 10 seconds. Short cessations of breathing of less than 10 seconds are generally not considered to be a health hazard. However, if longer or partial cessations of breathing occur more than five times per hour during sleep, serious health consequences may be considered.

[0003] Obstructive sleep apnea (OSA) is common in adults. The disease may result in neuronal damage or OSA may itself be a neurological disorder. Stimulation of the hypoglossal nerve is an effective treatment for many OSA patients, supporting the argument that OSA is a neuronal / neuromuscular disorder. Thus, evidence of a dissociation between movement-associated cortical neuronal populations and peripheral upper cervical motor units has been identified in OSA. Furthermore, there is evidence from functional MRT studies for a dissociation between sensorimotor and other cortical regions in OSA.

[0004] Sleepiness, a commonly reported effect of OSA, has a significant association with functional connectivity within the sensorimotor network. Over 55 years, functional neuroimaging studies in adults have demonstrated hypoperfusion in sensorimotor regions in patients with severe OSA. Furthermore, neuroimaging studies of interhemispheric interactions have identified a strong correlation between sleepiness and activity in the bidirectional precentral gyrus, a key part of the sensorimotor network, following sleep deprivation.

[0005] Sleep apnea, or the severity of sleep apnea, is diagnosed by cardiopulmonary polysomnography, usually in a sleep laboratory. Severity is currently divided into three severity groups:

[0006] The measuring sensors are attached to the patient suffering from obstructive sleep apnea for one night or two consecutive nights. During a period of approximately 8 hours, 18 different physiological signals are recorded: electroencephalogram (EEG, 4 signals), electrooculogram (EOG, 2 signals), electromyogram (EMG, 3 signals) at the chin and both lower legs, electrocardiogram (EKG, 1 signal), measurement of pulse rate and oxygen saturation in the blood (pulse oximetry, 2 signals), respiratory flow measurement through the nose and mouth (2 signals), respiratory effort at the abdomen and chest (2 signals), a snoring microphone at the neck (1 signal) and body position (1 signal).

[0007] As part of the cardiopulmonary polysomnography (KRPSG), the severity of OSA is calculated using the RDI (Respiratory Disturbance Index), i.e. the number of apneas, hypopneas, and so-called RERAs (Respiratory Effort Related Arousals) per hour of sleep. Individuals with an RDI greater than 15 per hour of sleep have clinically severe OSA and should seek immediate treatment according to current guidelines. Individuals with an RDI between 5 and 15 per hour have mild OSA and do not necessarily require immediate treatment, and individuals with an RDI less than 5 per hour are healthy (not suffering from OSA).

[0008] The Epworth Sleepiness Scale (EES) is used to subjectively assess daytime sleepiness in patients with OSA (Johns 1991). It is an eight-question questionnaire that provides a subjective (self-perceived) assessment of daytime sleepiness tendencies in monotonous everyday situations (no tendency = 0 points; maximum tendency = 3 points; i.e. cumulative score can vary from 0 to 24 points). A total EES score of >10 is highly likely to indicate significant, clinically relevant daytime sleepiness.

[0009] This measurement is then evaluated automatically or manually using algorithms built into commercially available KRPSG software systems that use 8 of the 18 KRPSG signals mentioned above to calculate the RDI. To ensure quality, it is essential that the visual evaluation is performed by experts (caregivers, medical technical assistants, physicians) with different levels of knowledge in each case.

[0010] To calculate the EES, a subjective assessment of daytime sleepiness using the ESS questionnaire is required.

[0011] This method is relatively complicated. Summary of the Invention [Problem to be solved by the invention]

[0012] The object of the present invention is to provide a preferably automated method for recording and in particular evaluating measurement data, which allows easy generation of reports based on data relating to measurements of the severity of OSA and the associated subjectively assessed degree of daytime sleepiness in OSA patients, in particular allowing corresponding reports to be easily generated without the involvement of an expert. [Means for solving the problem]

[0013] This object is solved by a method for determining a degree criterion of obstructive sleep apnea and / or its influence by evaluation of measurement data according to claim 1, in which a measured degree of obstructive sleep apnea and / or its influence is first determined and then two EEG measurement signals in an electroencephalogram of the 10-20 international EEG system are provided. The EEG measurement signals are divided into frequency bands, in particular filtered accordingly. At least one inter-frequency modulation index is determined from the data in at least two different frequency bands. The degree criterion of obstructive sleep apnea and / or its influence is then determined using the at least one inter-frequency modulation index.

[0014] In this method, only two EEG signals are provided and evaluated. The claimed method uses obstructive sleep apnea and its severity criteria and generates suitable data for correlation, thereby allowing a report to be generated regarding obstructive sleep apnea and its severity. The provision and evaluation of the data can be fully automated.

[0015] One measure for the severity of obstructive sleep apnea is, for example, the Disordered Breathing Index. One effect of obstructive sleep apnea is daytime sleepiness. The degree of daytime sleepiness is indicated, for example, using the Epworth Sleepiness Scale (ESS).

[0016] The 10-20 International EEG System is an internationally accepted method of describing the positioning of scalp electrodes as part of an EEG measurement, and in particular as part of a polysomnographic sleep study.

[0017] It is particularly advantageous if the EEG measurement signals are recorded at home during sleep, but it is also possible to record the data in a sleep laboratory or at a doctor's office.

[0018] The method is particularly easy to implement if, according to a preferred embodiment, the criteria for the severity of obstructive sleep apnea and its effects, in particular the breathing disorder index and / or daytime sleepiness, are determined on the basis of only two EEG measurement signals.

[0019] It has proven to be particularly advantageous if the electroencephalography points at which EEG measurement signals are recorded are points C3 and C4, as the measurement data obtained at these points can be used to determine with relatively high accuracy a breathing disorder index or a measure of daytime sleepiness.

[0020] In order to eliminate measurement disturbances at one measurement point, it is advantageous for the measurement signal to be split into separate frequency bands for each measurement point.

[0021] The present invention is based on the following observations.

[0022] Inter-frequency coupling (CFC) is a fundamental feature of brain oscillatory activity and is strongly correlated with brain function. CFC comprises different patterns: phase synchronization, amplitude simultaneous modulation, and phase-amplitude coupling (PAC). In this invention, PAC was analyzed because it represents neuronal coding and information transmission within local micro- and macro-scale neuronal ensembles in the brain. Low-frequency oscillatory activity represents the coordination of information flow between brain regions by modulating the excitability of local brain ensembles. The phase in this low-frequency oscillatory activity affects both the rhythm of high-frequency activity and the firing rate of individual neurons. Therefore, phase-amplitude inter-frequency coupling (PACFC) seems to promote effective interactions between similar phase-preferential neurons and synchronization of high-frequency bands during slower specific rhythm phases.

[0023] Understanding the neurophysiological connectivity patterns and oscillatory dynamics in sensorimotor cortical regions may help to further clarify the central nervous system (CNS) pathophysiological processes in OSA. Due to the fact that OSA is clinically associated with cognitive impairment, alertness impairment, and vigilance impairment during wakefulness (especially excessive daytime sleepiness), a better understanding and modeling of such connectivity may provide a basis for predicting related clinical phenomena.

[0024] The present invention characterizes the functional network connectivity of sensorimotor regions in treatment-naive OSA patients. For this purpose, it was first tested whether sleep stage-specific PACFC modulation of theta-gamma differed between patients with and without severe OSA. This evaluation was based on the results of previous studies that showed a significant role of theta oscillations during different sleep stages. Furthermore, it was analyzed whether these modulations were specific to certain sleep stages. To evaluate whether possible functional dissociation in sensorimotor regions was frequency-specific, PACFC modulation of delta-alpha was compared between patients with and without severe OSA as a control experiment. Finally, it was analyzed whether such dissociation correlated with clinical parameters, such as patient-reported sleepiness outcome (Epworth Sleepiness Scale; EES).

[0025] Previous studies have used inter-frequency coupling for sleep stage classification, especially in healthy adults with OSA. However, these studies did not evaluate the significance of this classification in terms of clinical applicability to OSA patients. In the present invention, frequency-specific CFC-based modulation index is used not only for sleep stage classification, but also to further demonstrate that the same modulation index can predict clinical scores such as ESS. Thus, these results hopefully help resolve either frequency-specific, sleep stage-specific, or global functional separation in sensorimotor regions in OSA patients. Furthermore, this can provide an objective, neurophysiological surrogate marker to quantify patient-reported subjective sleepiness, as well as the severity of respiratory disease in OSA patients.

[0026] In a preferred embodiment of the method, at least one inter-frequency modulation index is determined by phase-amplitude inter-frequency coupling. For this purpose, the measurement signal is preferably divided into at least two of the following frequency bands: a low frequency band of 0.1-1 Hz, a delta band (1-3 Hz), a theta band (4-7 Hz), an alpha band (8-13 Hz), a beta band (14-30 Hz) and a gamma band (31-100 Hz).

[0027] In detail, the two measurement signals are divided into two frequency bands, namely the alpha band (8-13 Hz) and the delta band (1-3 Hz), where the amplitude envelope is preferably determined from the alpha band and the phase of the measurement signals is determined from the delta band by phase / amplitude cross-frequency coupling.

[0028] Additionally or alternatively, the measurement signal is divided into two frequency bands, namely theta band (4-7 Hz) and gamma band (31-100 Hz), where the amplitude envelope is preferably determined from the gamma band and the phase of the measurement signal is determined from the theta band by phase / amplitude frequency coupling.

[0029] The EEG measurement signals required for the method are preferably stored in a database or memory module, so that they are available regardless of the time and place of recording the EEG measurement signals. This makes it possible to create a correlation database, for example, in which criteria of the severity of obstructive sleep apnea and / or its effects are correlated to inter-frequency modulation indices.

[0030] In a preferred embodiment, a correlation database is provided with correlation data between at least one inter-frequency modulation index and a severity criterion for obstructive sleep apnea and / or its effects. The severity criterion for obstructive sleep apnea and / or its effects is then determined from the at least one inter-frequency modulation index using data from the correlation database.

[0031] Providing a correlation database allows a particularly easy and accurate assignment of inter-frequency modulation indices to obstructive sleep apnoea and / or its effects.

[0032] Preferably, the severity criteria of obstructive sleep apnea and / or its effects, in particular the breathing disorder index and / or daytime sleepiness, are determined by a support vector machine based on at least one inter-frequency modulation index.

[0033] In summary, phase-amplitude inter-frequency coupling indicates the modulation of high frequency power in the C3 / C4 EEG signal by low frequency phase in the same (C3 / C4) EEG signal. The modulation index (MI) of this inter-frequency coupling (CFC) is then used to identify the phase-amplitude assignment between the phase-modulation frequency band (e.g. delta) and the amplitude-correction frequency band (e.g. gamma). For the measurement of the inter-frequency coupling measurement, only the signals recorded by the brain are used, i.e. only the signals from the C3 and C4 electrodes. After estimating the MI, a support vector machine learning algorithm is used to predict the RDI as well as the ESS (Epworth Sleepiness Scale).

[0034] In a preferred embodiment of the described method, the measurement signal is recorded during sleep, which is divided into sleep stages, whereby at least one inter-frequency modulation index is determined depending on the sleep stage, so that the allocation of the inter-frequency modulation index and the degree criterion of obstructive sleep apnea and / or its influence can be indicated depending on the sleep stage. This allows a very accurate report of the severity of sleep apnea and daytime sleepiness to be made, since it is confirmed that the individual inter-frequency modulation indexes are different for the individual sleep stages depending on the severity of sleep apnea and daytime sleepiness.

[0035] It is particularly advantageous if all method steps are performed automatically.The automated classification of sleep stages is advantageously performed when measurement signals are recorded during sleep.

[0036] Another object of the invention is a device for carrying out the method for determining obstructive sleep apnea and / or severity criteria of its effects, and in particular for implementing the described method, said device comprising a headgear with two sensors for recording EEG signals, in particular at points C3 and C4.

[0037] Since only two measurement points are required to carry out the method according to the invention, it is sufficient for the headgear to comprise only two sensors, in particular for determining the EEG signal at points C3 and C4.

[0038] A data memory and a data processing module may also be provided. The data memory is provided for storing the EEG signals and making them available for further evaluation. The data processing module is provided for determining an inter-frequency modulation index from the data according to the invention. The data processing module is further designed to determine a degree criterion of obstructive sleep apnea and / or its effects from the inter-frequency modulation index.

[0039] In a preferred embodiment, the headgear is a beanie, a cap which may also have a chin piece, or a headband.

[0040] The present invention has the advantage that RDI and ESS are determined in patients with OSA using only two EEG electrodes (instead of the standard eight of the eighteen electrodes mentioned above), making the method less complex and less expensive than known methods.

[0041] Due to the small amount of different measurement signals, the corresponding device for carrying out the method can be designed in a way that disturbs the test subject less.This is particularly advantageous for measurements during sleep, because the measurement device for recording the measurement signals does not disturb sleep or disturbs sleep much less compared to known devices.Furthermore, both RDI and ESS are determined completely automatically by the present invention, so that an objective method for determining / predicting daytime sleepiness is available.

[0042] Preferred embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. [Brief description of the drawings]

[0043] [Figure 1] FIG. 1 is a schematic diagram of the data acquisition and data analysis process. [Diagram 2] FIG. 13 shows MI differences in theta-gamma CFC. [Diagram 3] FIG. 1 shows MI differences in delta-alpha CFC. [Figure 4] FIG. 1 illustrates SVM classification of different sleep stages. [Diagram 5] FIG. 13 shows SVM prediction of RDI and ESS. [Figure 6] FIG. 1 shows correlation of clinical parameters with CFC measurements. [Figure 7] FIG. 1 shows the posterior distribution of the analyzed groups. [Figure 8] FIG. 1 shows a measuring device for recording EEG signals at C3 and C4. [Figure 9] FIG. 2 shows a first alternative embodiment of a measuring device for recording EEG signals of C3 and C4. [Figure 10] FIG. 1 shows a second alternative embodiment of a measuring device for recording EEG signals of C3 and C4. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0044] A total of 170 patients were included in the study carried out using the method according to the invention. These patients were divided into a main dataset and a validation dataset. The main dataset included 86 participants, 22 females, age group 27-84 years, 44 subjects with moderate or severe OSA. The validation dataset included 84 participants, 28 females, age group 35-75 years, 42 subjects with moderate or severe OSA. The data used for the analysis were retrospectively evaluated. Consecutive datasets of patients after applying the inclusion and exclusion criteria mentioned below were included for the analysis. Therefore, a retrospective, non-randomized, case-control study design was used. All patients first visited the clinic of the Sleep Medicine Center of the University Hospital, complaining of complaints about snoring and / or daytime sleepiness. All patients were first diagnosed with OSA based on the PSG recordings used in the study. Therefore, all patients were treatment naive and had not previously used positive airway pressure therapy with mandibular progression devices or upper airway surgery or therapy.

[0045] Participant inclusion and exclusion criteria were based on conditions that influenced the development and severity of confirmed OSA and / or EEG recordings. Data from adult patients (age >18 years) who presented with complaints of snoring and / or daytime sleepiness in the clinic, had not been previously treated for sleep-related respiratory disorders, and had not undergone nocturnal polysomnography in our sleep medicine center were included. Participants with neurodegenerative (e.g. Parkinson's disease) or neuroinflammatory (e.g. multiple sclerosis) diseases, history of stroke, heart failure-stage 3 or 4 according to the New York Heart Association (NYHA), chronic obstructive pulmonary disease (COPD), any psychiatric illness were excluded from the study. Additionally, subjects who regularly used sedatives, benzodiazepines, serotonin reuptake inhibitors, or other psychotropic medications, subjects with any type of malignant disease, subjects who had undergone anatomical cranial or cervical radiation therapy, subjects who had undergone surgery on intracranial structures, or subjects who had undergone surgery for the treatment of sleep-related respiratory disorders (snoring or OSA) were further excluded from the study.Patient-reported excessive daytime sleepiness (EDS) outcomes were recorded using the Epworth Sleepiness Scale (ESS).

[0046] Full-night polysomnography (PSG) recordings from 170 fully examined subjects who met the study enrollment criteria were used for analysis. All participants underwent overnight polysomnography (PSG) to record electroencephalograms, electrooculograms, submandibular and pretibial electromyograms, and electrocardiograms. Polysomnography (PSG) was recorded to determine the type and severity of sleep-related respiratory disorders according to current AASM (American Academy Sleep Medicine, Inc.) criteria. Nasal airflow was visualized by measuring impulse pressure using a nasal sensor that determines the pressure oscillations of respiratory airflow. Thoracic and abdominal excursions, oxyhemoglobin saturation (pulse oximeter), and body position were recorded simultaneously. Snoring was recorded using a microphone placed in front of the larynx. PSG recordings were performed in all patients using a commercially available PSG measurement system. All EEG recordings were made from C3 and C4 electrodes at a sampling rate of 200 Hz. Patients were split into a main and a validation dataset. The main dataset included 86 participants, 22 females, in the age group 27-84 years, 44 subjects with moderate or severe OSA. The validation dataset included 84 participants, 28 females, in the age group 35-75 years, 42 subjects with moderate or severe OSA. Recording parameters (EEG bandpass filter (0.05-200) Hz, sampling rate and EEG channels C2-M1 (left mastoid) and C4-M2 (right mastoid)) were identical in both datasets. All polysomnography (PSG) recordings were performed with standardized settings for each patient between 10 pm and 6 am.

[0047] Sleep stages were visually (manually) assessed according to the American Academy of Sleep Medicine guidelines. Sleep-related respiratory events were visually (manually) assessed according to the updated American Academy of Sleep Medicine guidelines. Apnea was detected when the maximum signal deviation fell 90% or more from baseline prior to an event lasting 10 seconds or more. Similarly, hypopnea was detected when the maximum signal deviation fell 30% or more from baseline prior to an event lasting 10 seconds or more, associated with either a 3% or greater arterial oxygen desaturation or cortical arousal. Further classification into obstructive, central, or mixed respiratory apnea events was based on simultaneous assessment of nasal airflow, or chest and abdominal deflections.

[0048] To pre-process the data, the raw EEG data were low-pass filtered (4th order Butterworth filter; limit frequency: 100 Hz) to avoid aliasing, and then wide-pass filtered at 0.5 Hz. To remove artifacts, the data underwent independent component analysis (FastICA) to remove artifacts related to muscles, eye blinks, eye movements, and line noise. The pre-processed data were then split into six different frequency bands for both electrodes: very low frequency (VLF, 0.1–1 Hz), delta (1–3 Hz), theta (4–7 Hz), alpha (8–13 Hz), beta (14–30 Hz), and gamma (31–100 Hz). A schematic of the data recording and analysis process is shown in Figure 1.

[0049] Phase-amplitude inter-frequency coupling (PACFC) describes the modulation of high frequency power by low frequency phase. This CFC modulation index (MI) is then used to identify the phase-amplitude assignment between the phase-modulation frequency band (e.g. delta) and the amplitude-modulation frequency band (e.g. alpha). To calculate the CFC-MI, the following steps were performed: First, the obtained EEG signal was filtered into two frequency bands, delta and alpha. After filtering, the Hilbert transform was applied to both the filtered time series and the amplitude envelope of the other to obtain the phase in one time series. This combined time series then has information in each phase, from the delta oscillation to the amplitude of the alpha rhythm. The possible phase range from -180° to +180° was then divided into 20 units (N) of 18° each, and the Kullback-Leibler distance (KL) was calculated using the following formula:

[0050]

number

[0051]

number

[0052] Regarding Shannon entropy, the KL distance can be used to judge the deviation between the distribution of data and a uniform distribution (U) as follows:

[0053]

number

[0054] Finally, the CFC-MI was calculated for all units. The distribution of the mean amplitudes is uniform across all units and shows no assignment between phase and amplitude. Therefore, the modulation index (MI) can be calculated as follows:

[0055]

number

[0056] Thus, MI is zero when the mean amplitude is distributed over all phases, and is maximum when Dirac delta is obtained for the distribution of phases. Couplings in the frequency bands theta-gamma and delta-alpha were estimated by correlation between the amplitude of high frequency signals and the phase of low frequency signals. CFC was estimated using a 5 second time frame with 50% overlap.

[0057] To analyze the significance of these CFC modulation indices, an SVM algorithm (support vector machine) was used to classify different sleep stages based on the CFC-MI values ​​from both frequency bands. SVM is an effective tool for nonlinear classification between two data sets, searching for the optimal separation threshold between both data sets by maximizing the margin between the closest points of the classes. Here, a polynomial kernel was used for this prediction with good performance and a grid search (min=1, max=10) was used to find some optimal input parameters and gamma (0.25). This selection was confirmed by 10-fold cross-validation, using 75% of the data for training and 25% of the data for testing. To verify the validity of these CFC modulation indices for clinical applicability, an SVM analysis was further applied to predict the clinical scores (RDI and ESS) used in the diagnostic criteria for patients with OSA. Here, a support vector regressor (SVR) analysis was performed. This is a multiple regression method based on machine learning, which can assign confirmed and trained values ​​and represent the prediction accuracy. To obtain a threshold for prediction accuracy, an approach based on statistical inference obtained from the Bayes' theorem credible interval was deployed. A threshold of 75% could distinguish the posterior distribution from a 95% Bayes' theorem credible interval (representing 95% inclusion of data points). Here, the posterior distribution and credible interval were obtained considering all sleep stages in both groups and all modulation indices from the densest interval of 95% of the distributions (range: 0.32-0.89). Thus, prediction accuracy greater than that obtained after 10-fold cross-validation (75%) was considered a highly significant result.

[0058] Scientific controls were performed to ensure the absence of influence of variables other than the independent variables on the results of the study. It was confirmed whether the estimated PACFC was independent of arousal and intermittent limb movement in these patients. For this purpose, the arousal index and the intermittent limb movement index (PLM) were estimated for each patient, as well as the Pearson correlation between the PACFC and these indices in each sleep stage. Since previous studies have shown a significant correlation between heart rate variability and clinical OSA scores, it was analyzed whether PACFC was influenced by the activity of the autonomic nervous system. For this purpose, the heart rate variability (HRV) was estimated for each patient, respectively, and the Pearson correlation with PACFC was estimated in each sleep stage. HRV was calculated using the standard deviation of the normal-normal interval; the technique will be explained later. Furthermore, the significance of the size of the sample used in the study was determined. For this purpose, a post-hoc Bayes' theorem posterior distribution analysis was estimated for the MI index in the N1 sleep stage between the two groups.

[0059] Of the 86 patients analyzed, 42 patients were diagnosed with a Respiratory Disturbance Index (RDI) below 15 per hour (4 patients with an RDI below 5 per hour, 38 patients with an RDI between 5 and 15 per hour) and 44 patients with clinically severe OSA (30 patients with an RDI between 15 and 30 per hour, 14 patients with an RDI above 30 per hour). These two groups do not differ significantly in age and sex (p>0.05). Details of the demographics, together with the clinical measurements obtained, are shown in Table 1. Statistical analyses performed on the interfrequency coupling (CFC) parameters and their associations with the clinical measurements obtained from these two groups yielded significant results, as explained below.

[0060] Demographic details of all participants included in the study RDI: Respiratory Disorder Index ESS: Epworth Sleepiness Scale

[0061] [Table 1]

[0062] The CFC modulation index (MI) in the theta-gamma frequency band was significantly decreased in all sleep stages in patients with clinically severe, i.e., moderate or severe (RDI >15 per hour) OSA (p<0.001), as shown in Figure 2A. The theta-gamma modulation index was higher during NREM stages N2 and N3 than during N1 and REM sleep stages for both groups. The difference in MI values ​​between both groups was highest during N1, decreased during N2, but increased again during N3 and REM sleep stages.

[0063] A table showing all these values ​​is provided as Supplementary Table 1.

[0064] Figure 2 shows the MI differences in theta-gamma CFC. The rain cloud plot in Figure 2a clearly shows that patients in the RDI > 15 per hour group have significantly lower theta-gamma CFC modulation indices in all sleep stages (NREM and REM) than the RDI < 15 per hour group. Here, the exact same pattern was observed in both the initial and validation patient groups (Figure 2b).

[0065] However, as shown in Figure 3A, CFC-MI in the delta-alpha frequency band was significantly decreased (p<0.001) only during REM and N1, but not in patients with clinically severe OSA in N3 sleep stage compared to patients with mild or no OSA (RDI ≤15 per hour). Furthermore, CFC-MI in NREM sleep stage N2 was higher in patients with severe OSA (i.e., RDI >15 per hour) compared to patients with no OSA (RDI ≤15 per hour) (see Table 1).

[0066] Figure 3 shows the MI difference in delta-alpha CFC. The rain cloud plot in Figure 3a shows that patients in the RDI group of more than 15 per hour have significantly lower delta-alpha CFC modulation index than REM and N1 sleep stages, and have significantly higher MI than patients in the RDI group of less than 15 per hour in N2 sleep stage. The delta-alpha CFC modulation index in NREM-3 (N3) sleep stage is almost identical in both patient groups. It is noteworthy that the exact same pattern is seen in both the initial and validation data sets (Figure 3b).

[0067] SVM analysis (Support Vector Machine) showed significant classification of all four sleep and wake stages using CFC modulation indices away from theta-gamma and delta-alpha frequency bands. Overall classification accuracy was higher than 80% and reached up to 94% for classification of wake stages using theta-gamma CFC-MI, as shown in Figure 4.

[0068] Figure 4 shows the SVM classification of different sleep stages. The bar graphs show the classification accuracy of the support vector machine (SVM) of different sleep stages using theta-gamma and delta-alpha cross-frequency coupling modulation indices (CFC). Groups of 10 bars in each set represent the accuracy obtained for 10-fold cross-validation. The dotted line is shown at 75% accuracy to highlight the significant level of classification obtained for all CFC metrics used in this study. All accuracy values ​​are shown in Supplementary Table 1.

[0069] Moreover, SVM could predict RDI and Epworth Sleepiness Scale (ESS) with a fair accuracy (more than 75%) in different sleep stages using CFC-Mi in both frequency band pairs. Theta-gamma CFC could significantly predict RDI and ESS in NREM sleep stages (N2 and N3). Delta-alpha CFC in REM sleep stages could significantly predict RDI, and delta-alpha CFC in wakefulness stages could significantly predict ESS. All prediction details are shown in Figure 5.

[0070] Figure 5 shows prediction of RDI and ESS. The scatter plot shows the prediction accuracy of the Support Vector Machine (SVM) of the Respiratory Disorder Index (RDI) and the Epworth Sleepiness Scale (ESS) using modulation indices from theta-gamma and delta-alpha interfrequency coupling (CFC). Groups of 10 points in each set represent the accuracy obtained for 10-fold cross-validation. Accuracy above 75% is obtained for significance and is indicated by a dotted line in the plot.

[0071] Of the 84 patients analyzed from this dataset, 42 patients were diagnosed with a Respiratory Disturbance Index (RDI) of 15 or less per hour, and 42 patients were diagnosed with clinically severe OSA (3 patients with an RDI of 15-30 per hour, and 39 patients with an RDI of >30 per hour). Again, there were no significant differences in age and gender in this dataset (p>0.05). Demographic details are shown in Table 1.

[0072] The statistical analysis performed for this dataset yielded very similar results to the first (main) dataset, confirming most of the results. CFC-Mi in the theta-gamma frequency band was also reduced in clinically severe OSA patients (RDI > 15 per hour). This was similar to the main findings, which showed higher modulation even during NREM-N2 and N3 sleep stages (Figure 2B). Similarly, as can be seen in the main dataset (Figure 3B), CFC-MI in the delta-alpha frequency band was also significantly reduced only during REM and N1, but not N2 and N3 sleep stages in patients with clinically severe OSA.

[0073] SVM analysis for sleep stage classification using CFC modulation indices in the theta-gamma and delta-alpha frequency bands showed reproducible results using the validation dataset with a classification accuracy of over 80%, as shown in Figure 4.

[0074] Similarly, this validation dataset also allowed us to reproduce the prediction results for clinical parameters (RDI and Epworth Sleepiness Score (ESS)) with an accuracy of more than 75% when using the same CFC modulation index as in the main dataset, as shown in Figure 5.

[0075] However, the correlation between Epworth sleepiness score and PACFC was also not significant in any of the sleep stages (Figure 6).

[0076] No significant correlations (all p>0.05) were found between wakefulness and the PLM index on PACFC in any sleep stage (Figure 6). 2) Furthermore, no significant correlations were found between heart rate variability and PACFC in any sleep stage, indicating no influence of the autonomic renal system on PACFC (Figure 6).

[0077] Figure 6 shows the correlation of clinical parameters with CFC measurements. Figure 6a shows the correlation coefficient between wakefulness and intermittent limb movement index (PLM) versus phase-amplitude frequency coupling (PACFC) for each sleep stage. For both delta-alpha and theta-gamma, PACFC is separated by columns. Figure 6b shows the correlation coefficient between heart rate variability (HRV) and PACFC in different sleep stages separated by columns. Figure 6c shows the correlation coefficient between Epworth Sleepiness Score (ESS) and PACFC in different sleep stages. Correlations of the main and validation groups are presented separately in each row. r values ​​are presented for the correlations. All correlations were not significant (p>0.05).

[0078] Bayes' theorem posterior distribution showed that 95% of the high density intervals (HDIs) were within the effects obtained in the analyzed data (Figure 7), representing a sufficient sample size for the initial results in this study.

[0079] Figure 7 shows the posterior distribution of the analyzed groups. The right plot shows the distribution histogram of effect sizes representing the 95% high density interval (HDI) found in the analyzed data. This represents a sufficient sample size of subjects included based on the initial results (i.e. inter-frequency coupling of theta-gamma phase-amplitude in N1 sleep stage). The left plot shows the probability distribution with the overlaid posterior predictive distribution on the raw data of each data sample.

[0080] A significant decrease in theta-gamma modulation index (MI) was found in central sensorimotor cortical regions in patients with moderate or severe OSA compared to patients with mild OSA or healthy controls. The decrease in MI during sleep was frequency band specific. It included theta-gamma connectivity during all sleep stages, whereas delta-alpha connectivity was only during REM and N1. Thus, a global decrease in modulation (both theta-gamma and delta-alpha) was identified during REM and N1. Moreover, the differences in MI between stages were clear, so that a classification of sleep stages based on MI values ​​was realized in both data sets in both patient groups. Moreover, during N2 and N3, theta-gamma MI reliably predicted RDI and ESS, and during REM, delta-alpha MI reliably predicted RDI.

[0081] These novel results, showing a functional dissociation between theta and gamma cortical sensorimotor activity during all sleep stages in OSA patients, have pathophysiological and clinical implications and require further investigation.

[0082] Theta-gamma PACFC has been associated with motor, sensory, and cognitive processes. In individuals with an RDI of less than 15 per hour, there is very strong coupling between theta and gamma oscillations in both N2 and N3 stages. In moderate and severe OSA, the modulation index drops quite significantly during N3, while in N2 it drops to a much lower range. During the short wake periods between sleep stages, MI increases significantly in patients with moderate / severe (i.e., significant) OSA compared to patients with an RDI of 15 or less per hour. This may reflect the role of physiological (motor, respiratory, cognitive) compensation for cortical arousal and / or intermittent wake periods to promote central sensorimotor transient increases in patients with severe OSA. The enhancement of synaptic nets associated with arousal can provide the basis for increased connectivity, etc.

[0083] The recorded neural activity in the sensorimotor cortex may either be generated primarily in this anatomical region or may be an epiphenomenon of activity from other subcortical / thalamic or neural brainstem master generators that promote neuromodulation in cranial neural pathways, resulting in a reduction in muscle tone in the upper airway associated with upper airway obstruction that occurs during respiratory events in OSA sleep.

[0084] In patients with focal epileptic seizures, the strength of theta-gamma phase-amplitude coupling during sleep was highest during N3 and lowest during REM. In patients with moderate and severe OSA, theta-gamma PACFC was highest during N2 (see Figure 2). Coupling of fast and slow oscillations was significantly reduced during REM compared to N2 and N3 in all OSA patient groups. This reduction was more characteristic in patients with severe OSA. Strong CFC between high-frequency and slow-wave oscillations during slow-wave sleep was found in the hippocampus of anesthetized primates. EEG measures the summed postsynaptic potentials of synchronously active areas in the cortex and hippocampus spread across the brain, skull, and scalp. Although they often coincide, firing of action potentials is not necessarily related to oscillations of postsynaptic potentials. Thus, oscillations of postsynaptic potentials do not always result in firing of postsynaptic action potentials. It is difficult to distinguish cortical from hippocampal output in humans based solely on surface EEG recordings.

[0085] A significant increase in delta-alpha CFC-MI was identified during N2 in patients with severe OSA compared to patients with an RDI of 15 or less per hour. This finding may represent a compensatory increased activity of the sensorimotor cortex during respiratory events in the Rich-N2 stage to exert better motor control of breathing in more severely affected OSA patients (RDI >15 per hour).

[0086] Delta band oscillations in spikes and local field potentials in the somatosensory whisker barrel cortex in awake mice are phase-locked to respiration. Thus, respiratory activity directly modulates slow (1-4 Hz) rhythmic neural activity in the somatosensory whisker barrel cortex and indirectly modulates gamma band power via a phase-amplitude coupling mechanism in mice. Our findings provide preliminary evidence for the physiological repercussions of delta and gamma band oscillations in respiratory control, especially in humans with OSA, and should be further validated.

[0087] In particular, delta-alpha CFC-MI remains fairly stable during N3, regardless of OSA severity. Thus, delta-alpha coupling may be included in the brain connectivity process, which remains stable during N3, or may be a surrogate marker for known respiratory stability during N3, when apneas and hypopneas occur at low frequency.

[0088] Assuming that both theta-gamma and delta-alpha modulation indices can reliably predict sleep stage classification according to AASM criteria, it can be deduced that there are very different sleep stage-specific PACFC patterns involving the above-mentioned frequency bands. These different patterns are apparently very robust and involve at least two of the above-mentioned oscillatory channels (gamma-theta, delta-alpha) and can include other oscillatory channels. As the majority of the tested dataset belongs to patients with RDIs of more than 15 per hour, these sleep stage-specific coupling patterns are likely to remain very robust regardless of the degree of associated sleep-related dyspnea.

[0089] The reduced global (theta-gamma and delta-alpha) connectivity during REM in central sensorimotor regions, expressed as reduced MI, in OSA patients with an RDI >15 per hour compared to those with an RDI <15 per hour can provide a surrogate marker for reduced central motor power during REM. This reduced motor power likely affects many muscle groups, especially those controlling upper airway patency, since they are strongly correlated with RDI. The difference in MI is particularly clear in the CFC-MI of delta-alpha (Figure 2). Differential modulation of global and local oscillations during REM sleep has been reported. Thus, multi-frequency (global) MI in sensorimotor regions during REM can serve as a surrogate marker of OSA disease severity. In support of this argument, further analysis (Figure 4) showed that the MI of delta-alpha during REM predicted the mean RDI of both data sets tested with very high reliability.

[0090] Theta-gamma MI during N2 and N3, and delta-alpha MI during brief arousals from sleep, prove to be reliable surrogate markers for patient-reported excessive daytime sleepiness (EDS). These results again indicate possible sleep-related oscillations and stage-specific physiological mechanisms that promote individual alertness and vigilance. Subjective ratings of sleepiness were uniquely associated with increased functional connectivity over a broad range within the sensorimotor network.

[0091] The modulation index in phase-amplitude cross-frequency coupling (PACFC) was the main end result and was tested as a predictor for clinical variables in this study. With regard to its significance, spatial working memory performance, maintenance of multi-item working memory, changes in perceptual outcomes, learning, visual attention, and cognition are some of the functional features associated with PACFC modulation. Furthermore, it has been shown to contribute to BOLD connectivity (dependent on blood oxygen levels) and to have connectivity to brain alterations occurring in several neurological disorders, such as epilepsy, Parkinson's disease, Alzheimer's disease, schizophrenia, obsessive-compulsive disorder (COD), and minimal cognitive impairment (MCI). Although there is ample evidence that PACFC is a potentially promising approach for deciphering brain functions and for some pathologies with a reliable physiological mechanism (low-frequency phase reflects local neuronal excitability, while high-frequency power increases reflect either a global increase in the entire population of synaptic activity or selective activation of connected neuronal sub-networks), there are still several unanswered questions about the origin, causality, and mechanism of these oscillations. The choice of modulation index (MI) in this study is based on the fact that MI has proven to be the most robust to confounding effects of modulators among some of the most commonly used phase-amplitude coupling measures, including data length, signal-to-noise ratio, and sampling rate when approaching the Nyquist frequency.

[0092] The above evidence may open new possibilities for pharmacological or transcranial magnetic stimulation (TMS) interventions in the sensorimotor cortex of OSA patients. TMS during sleep was applied to the corticomotor-somatotopic representation of the tongue. The induced twitching temporarily improved airflow without waking OSA patients. However, the effects on other motor areas and the neurocognitive effects of TMS have not been extensively studied in OSA. The finding that the above CFC modulation indexes in both frequency bands could significantly predict RDI and Epworth Sleepiness Score (ESS) in OSA patients should be further verified in larger studies.

[0093] The results that 1) theta-gamma CFC-MI significantly predicted RDI and ESS in NREM (N2, N3), 2) delta-alpha CFC-MI significantly predicted RDI in REM, and 3) delta-alpha CFC-MI significantly predicted ESS during wakefulness suggest that theta-gamma and delta-alpha metrics of CFC-MI may represent completely different processes in human sleep physiology. Delta-alpha coupling appears to be significant for 1) upper airway movement stability and respiratory control during REM sleep, and 2) attention and vigilance processes as represented by ESS during (cortical) wakefulness and brief wakefulness periods between sleep stages. Theta-gamma phase-amplitude coupling appears to be highly significant for 1) upper airway stability and respiratory control during NREM N2 and N3 sleep, and 2) attention and vigilance related processes (represented by ESS) occurring during N2 and N3 sleep. These results imply that the central sensorimotor regions of the cortex may be a substantial hub in the network regulating sleep and / or sleep-related respiratory activity.After examining a large cohort of patients, the modulation index is finally integrated as an additional metric representing both the severity of dyspnea and daytime sleepiness in patients with OSA.

[0094] Replication of the results in the validation dataset further supports the reproducibility and validity of the results and represents their clinical significance as surrogate markers of diagnosis. Furthermore, scientific control results did not clearly demonstrate the influence of arousal, periodic limb movements, and the autonomic nervous system on PACFC measurements.

[0095] It is suggested that the MI of theta-gamma in sensorimotor cortical areas during N2 and N3, and the CFC-MI of delta-alpha in sensorimotor cortical areas during REM, can be used as a metric of dyspnea during human sleep, and therefore as a measure of OSA severity. Therefore, further analysis of the FCKW of theta-gamma during N2 and N3, and the FCKW of delta-alpha during REM should be performed. Calculation of these MIs in different cortical areas can provide additional insights into the development and diagnosis of OSA. Neurophysiological and neuroimaging studies of thalamocortical connectivity based on the present results can further clarify the mechanism of excessive daytime sleepiness. In addition, it would be interesting to evaluate the effects of established evidence-based treatments for OSA, such as positive airway pressure (PAP) treatment, on PACFC.

[0096] Functional dissociation in the central cortical sensorimotor regions between theta and gamma activity was confirmed in all sleep stages of OSA. Furthermore, significant delta-alpha sensorimotor dissociation occurs during REM and N1 stages in OSA. Thus, sensorimotor dissociation is extensive, shows frequency band and sleep stage specific patterns, and provides further evidence for the presence of central sensorimotor dysfunction in OSA patients. Theta-gamma modulation index during N2 and N3 reliably predicts patient-reported sleepiness. Therefore, modulation index can be used as a surrogate diagnostic predictive marker for sleep dyspnea and for patient-reported excessive daytime sleepiness.

[0097] In summary, the described method is suitable for determining the severity of obstructive sleep apnea and associated daytime sleepiness. The method includes the following steps: - recording two body function data, both body function data being electroencephalographic EEG measurement signals at electroencephalography points C3 and C4, - dividing the measurement signal into frequency bands, said division being performed separately for each measurement point, - determining a modulation index between frequencies; - Determining the respiratory disorder index and daytime sleepiness using a support vector machine based on the inter-frequency modulation index.

[0098] The Disordered Breathing Index and daytime sleepiness are measures of the severity of obstructive sleep apnea and associated symptoms.

[0099] Figures 8 to 10 show embodiments of measuring devices for detecting the measurement signals C3 and C4. These measuring devices are not shown but are part of a device for implementing the method described. The measuring device from Figures 8 and 9 is a headgear, in which two sensors 10 for detecting the measurement signals C3 and C4 are integrated. This headgear is designed here as a cap 12 (Figure 8) or a cap 112 with a chin 114 (Figure 9). Alternatively, interconnected bands 212 can also be provided as headgear, which is worn on the head and in which two sensors 10 for detecting the measurement signals C3 and C4 are integrated (see Figure 10).

Claims

1. 1. A method for determining obstructive sleep apnea and / or severity criteria thereof, comprising: a. defining severity criteria for obstructive sleep apnea and / or its effects; b. Providing two EEG measurement signals in an electroencephalogram at electroencephalogram points of the 10-20 international EEG system; c. dividing the EEG measurement signals into frequency bands; d. determining at least one inter-frequency modulation index using data from at least two different frequency bands; e. determining a measure of severity of obstructive sleep apnea and / or its effects by at least one inter-frequency modulation index; A method of judging by:

2. 2. The method of claim 1, wherein the EEG measurement signals are determined during sleep in a laboratory or at home.

3. 3. The method according to claim 1 or 2, characterized in that the criteria for the severity of obstructive sleep apnea and its impact, in particular the respiratory disorder index and / or the daytime sleepiness impact severity criterion, are determined on the basis of only the two EEG measurement signals.

4. 4. The method according to claim 1, wherein the electroencephalography points at which measurement signals are recorded are points C3 and C4.

5. 4. A method according to claim 3, characterized in that the measurement signal is divided into frequency bands for each measurement point.

6. A method according to any one of claims 1 to 5, characterised in that at least one inter-frequency modulation index is determined by phase-amplitude inter-frequency coupling.

7. 7. The method according to claim 1, wherein the measurement signal is divided into at least two of the following frequency bands: a low frequency band from 0.1 to 1 Hz, a delta band (1 to 3 Hz), a theta band (4 to 7 Hz), an alpha band (8 to 13 Hz), a beta band (14 to 30 Hz) and a gamma band (31 to 100 Hz).

8. 8. A method according to claim 6 or 7, characterized in that the measurement signal is divided into two frequency bands, namely the alpha band (8-13 Hz) and the delta band (1-3 Hz), and a phase-amplitude inter-frequency coupling is determined, the amplitude envelope being preferably determined from the alpha band and the phase of the measurement signal being determined from the delta band by the phase-amplitude inter-frequency coupling.

9. 9. The method according to claim 6, characterized in that the measurement signal is divided into two frequency bands, namely the theta band (4-7 Hz) and the gamma band (31-100 Hz), and a phase-amplitude inter-frequency coupling is determined, the amplitude envelope being preferably determined from the gamma band and the phase of the measurement signal being determined from the theta band by phase-amplitude inter-frequency coupling.

10. A method according to any one of claims 1 to 9, characterized in that a correlation database with correlation data is provided between at least one inter-frequency modulation index and a criterion for the severity of obstructive sleep apnea and / or its effects, and that the criterion for the severity of obstructive sleep apnea and / or its effects is determined from the at least one inter-frequency modulation index by means of data from the correlation database.

11. 11. The method according to claim 1, characterized in that the criteria for obstructive sleep apnea and / or the severity of its impact, in particular the respiratory disorder index and / or the severity of daytime sleepiness impact, are determined using a support vector machine based on at least one interfrequency modulation index.

12. 12. The method according to claim 1, wherein the measurement signals are recorded during sleep and the sleep is divided into sleep stages, and wherein a modulation index between at least one frequency is determined in dependence on the sleep stage.

13. The method according to any one of claims 1 to 12, characterized in that all method steps are performed automatically.

14. A device for implementing a method for determining obstructive sleep apnea and / or severity criteria of its effects, in particular a device for implementing a method according to any one of claims 1 to 13, comprising a headgear with only two sensors for determining EEG signals, said sensors preferably being characterized in that the measurement signals are recorded at points C3 and C4.

15. The device of claim 14 , wherein the headgear is a headband, a cap, or a beanie.

Images