WO2008132736A2

WO2008132736A2 - Method and device for characterizing sleep

Info

Publication number: WO2008132736A2
Application number: PCT/IL2008/000562
Authority: WO
Inventors: Armanda Baharav; Zvi Shinar; Shulamit Eyal
Original assignee: Hypnocore Ltd.
Priority date: 2007-05-01
Filing date: 2008-04-28
Publication date: 2008-11-06
Also published as: WO2008132736A3

Abstract

Method and device for determining an incident of a sleep related respiratory event, from ECG data and oximetry data are disclosed. The method comprises determining occurrence of an attenuation in ECG derived respiratory signal below a first threshold, determining occurrence of a desaturation in the oximetry data below a second threshold, determining occurrence of an attenuation in ECG derived respiratory signal below a third threshold with a corresponding desaturation in the oximetry data below a forth threshold, determining occurrence of an attenuation in ECG derived respiratory signal below a fifth threshold with a corresponding detection of arousal; and determining the incident of a sleep related respiratory event if any of the occurrences exists.

Description

METHOD AND DEVICE FOR CHARACTERIZING SLEEP

FIELD OF THE INVENTION

[0001] The present invention relates to sleep disorders. More specifically the present invention relates to a method and a device for characterizing sleep.

BACKGROUND OF THE INVENTION [0002] The growing interest in sleep and its disorders, including their influence on health, well-being and public safety (e.g. involvement of fatigued drivers in traffic accidents) have caused a continuously increasing need to perform sleep investigations for both research and clinical purposes. Substantial research has been undertaken directed toward understanding the nature of sleep and of sleep disorders. These researches yielded considerable information concerning human patterns of sleep and wakefulness, and of physiological activities occurring during human sleep. In addition, substantial information has been obtained concerning various sleep disorders.

[0003] It is common to define, for a normal healthy individual, a succession of three states of being, known as Wakefulness, Rapid-Eye-Movement (REM) sleep and Non-REM (NREM) sleep. NREM sleep is subdivided into four stages, which are enumerated from Stagel to Stage 4, in accordance with an increasing threshold to the arousing influence of external stimuli. These stages are also known as the depth of sleep.

[0004] Several sleep disorders and symptoms are associated with increased upper airway resistance, for example, snoring and Obstructive Sleep Apnea (OSA). The prevalence of OSA is about 4% of adult population, and it increases to up to 50% of male adults above the age of 65.

[0005] Insomnia as a primary disorder has a prevalence of about 10% of the general population. According to the "2005 sleep in America poll" by the National Sleep Foundation, 54% of adult population reported that they experienced at least one symptom of insomnia at least several nights a week.

[0006] Determination of body position during sleep may also assist in diagnosing sleep disorders originating from frequent body position changes during sleep. Many sleep disorders, in particular snoring, sudden infant death syndrome and OSA, are position- dependent. Knowing the body position during sleep is important for study, diagnosis and treatment strategy of such sleep disorders.

[0007] To date, sleep stages are monitored and examined clinically with a polysomnograph (PSG)₅ which provides data regarding the electrical activity of brain, muscles and eye movement during sleep. The PSG data are analyzed in accordance with a gold standard procedure attributed to Rechtschaffen and Kales (R&K) [Rechtschaffen A., Kales A., eds., "A manual of standardized terminology, techniques and scoring system for sleep staging in human subjects", Washington DC: US Government Printing Office, NIH Publication 204, 1968]. The R&K criteria are primarily based on the analysis of three collected bio-signals: (i) electroencephalogram (EEG), (ii) electrooculogram (EOG), and (iii) electromyogram (EMG). The standard procedure is as follows: EEG signals are derived primarily from the cortex of the brain. At the same time an EMG signal which monitors muscle activity, generally from some of the muscles of the head (i.e. submental) is measured, together with left eye and right eye EOG (signals produced by eyeball movements relative to the skull). These EEG, EMG and EOG signals are conventionally recorded on a multi-channel physiological recorder (digital polysomnograph).

[0008] Sleep related respiratory events are commonly detected based on the following signals: nasal pressure or oronasal flow, respiratory inductive plethysmography (RIP), respiratory effort (1-2 leads) or inductive belts, oxygen saturation, electrocardiogram (ECG), body position and a microphone for snoring. Apnea or hypopnea events require a reduction of at least 50% in the respiration signal, or a lesser reduction associated with either cortical arousal or a >3% reduction in the oxygen saturation signal [American academy of sleep medicine task force. "Sleep related breathing disorders in adults: recommendations for syndrome definition and measurement techniques in clinical research." Sleep 1999, 22: 667-689]. The detection of respiratory events is considered in relation to sleep / wake stages and arousals obtained from the analysis of the three signals mentioned in the previous section.

[0009] Furthermore, some apnea screening tests are partially accepted, using only one or two channels recordings, typically pulse oximetry and/or respiratory movements recording using oronasal pressure, oronasal thermistor, abdomen or thorax piezoelectric or inductive sensors. These screening tests have many limitations, primarily they suffer from low sensitivity, and hence are recommended only as confirmation for suspected OSA, and are not accepted for ruling out OSA in a symptomatic patient [Netzer N et al, "Overnight pulse oximetry for sleep disordered breathing in adults." Chest 2001, 120: 625-633].

[0010] An example of another data-set for characterizing sleep parameters is described in US patent application 2006/0235315 (Akselrod; Solange et al., incorporated herein by reference)

[0011] In their patent application Akselrod; Solange et al. perform analysis of acquired data for characterizing sleep, sleep stages, body position and sleep disorders by data derived from signals of electrical activity recorded from a chest of a sleeping subject, such as ECG signals. Note however that the analysis does not include classification of the respiratory events as of central or obstructive origin.

[0012] It is desired to provide a novel method, device, and system that combine measuring several sleep parameters (e.g. the electrical activity of the heart (e.g. ECG), and pulse oximetry (SpO2 and pulse wave) signals, with or without one or more EMG (electromyogram) signals) and enables accurate characterization of sleep disorders.

SUMMARY OF THE INVENTION

[0013] There is thus provided, according to some embodiments of the present invention,a method for determining an incident of a sleep related respiratory event, from ECG data and oximetry data, the method comprising: [0014] determining occurrence of an attenuation in ECG derived respiratory signal below a first threshold,

[0015] determining occurrence of a desaturation in the oximetiy data below a second threshold,

[0016] determining occurrence of an attenuation in ECG derived respiratory signal below a third threshold with a corresponding desaturation in the oximetry data below a forth threshold,

[0017] determining occurrence of an attenuation in ECG derived respiratory signal below a fifth threshold with a corresponding detection of arousal; and

[0018] determining the incident of a sleep related respiratory event if any of the occurrences exists. [0019] Furthermore, in accordance with some embodiments of the present invention, the method further comprises using EMG data to classify the sleep related respiratory event as a central sleep related respiratory event or obstructive sleep related respiratory event.

[0020] Furthermore, in accordance with some embodiments of the present invention, the method further comprises determining sleep stages using the ECG data and pulse transit time derived from the ECG data and a pleth data.

[0021] Furthermore, in accordance with some embodiments of the present invention, the method further comprises determining sleep stages using EMG data, ECG data and pulse transit time derived from the ECG data and a pleth data. [0022] Furthermore, in accordance with some embodiments of the present invention, the method further comprises receiving the ECG data and the oximetry data from a remote user over the Internet.

[0023] Furthermore, in accordance with some embodiments of the present invention, the method further comprising providing information on the sleep related respiratory event over the Internet to the user.

[0024] Furthermore, in accordance with some embodiments of the present invention, there is provided a device for determining an incident of a sleep related respiratory event, comprising:

[0025] an ECG sensor, [0026] a oxygen saturation sensor,

[0027] a processor provided with an algorithm for determining an incident of a sleep related respiratory event, from ECG data and oximetry data, the algorithm comprising:

[0028] determining occurrence of an attenuation in ECG derived respiratory signal below a first threshold, [0029] determining occurrence of a desaturation in the oximetry data below a second threshold,

[0030] determining occurrence of an attenuation in ECG derived respiratory signal below a third threshold with a corresponding desaturation in the oximetry data below a forth threshold, [0031] determining occurrence of an attenuation in ECG derived respiratory signal below a fifth threshold with a corresponding detection of arousal; and [0032] determining the incident of a sleep related respiratory event if any of the occurrences exists.

[0033] Furthermore, in accordance with some embodiments of the present invention, the device further comprises an EMG sensor. [0034] Furthermore, in accordance with some embodiments of the present invention, the algorithm of the processor further comprises using EMG data to classify the sleep related respiratory event as a central sleep related respiratory event or obstructive sleep related respiratory event.

[0035] Furthermore, in accordance with some embodiments of the present invention, the algorithm of the processor further comprises determining sleep stages using EMG data, ECG data and pulse transit time derived from the ECG data and a pleth data.

[0036] Furthermore, in accordance with some embodiments of the present invention, the algorithm of the processor further comprises determining sleep stages using the ECG data and pulse transit time derived from the ECG data and a pleth data. [0037] Furthermore, in accordance with some embodiments of the present invention, the device is further provided with a communication link wherein the algorithm of the processor further comprises receiving the ECG data and the oximetry data from a remote user over the Internet using the communication link.

[0038] Furthermore, in accordance with some embodiments of the present invention, the algorithm of the processor further comprises providing information on the sleep related respiratory event over the Internet to the user using the communication link.

[0039] Furthermore, in accordance with some embodiments of the present invention, the device further comprises an amplifier for amplifying signals received from the sensors.

[0040] Furthermore, in accordance with some embodiments of the present invention, said oxygen saturation sensor is selected from a group of sensors consisting of a pulse waveform sensor and SpO2 sensor.

BRIEF DESCRIPTION OF THE DRAWINGS [0041] In order to better understand the present invention, and appreciate its practical applications, the following Figures are provided and referenced hereafter. It should be noted that the Figures are given as examples only and in no way limit the scope of the invention. Like components are denoted by like reference numerals.

[0042] Fig. 1 illustrates a block-diagram of the information processing path in a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0043] Fig. 2 illustrates a block-diagram of the information processing path in an internet web-based device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0044] Fig. 3 illustrates a block-diagram of the main steps preformed by a processor in a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0045] Fig. 4 is a flow chart of preliminary processing of the data by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention. [0046] Fig. 5 is an illustration of RR interval and pulse transient time (PTT) for understanding the definition of the terms used in the text.

[0047] Fig. 6 is a flow chart of oximetry data validity-check by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention. [0048] Fig. 7 is an illustration of an example of pleth and ECG graphs for performing a validity check by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0049] Fig. 8 is a flowchart describing the detection of arousal and awakening events by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0050] Fig. 9 is an illustration of an example of supporting indication of an arousal event based on a decrease in a pulse wave amplitude (PWA) parameter by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention. [0051] Fig 10 is an illustration of an example of supporting indication of an arousal event based on an increase in submental EMG amplitude by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0052] Fig 11 is a flowchart describing the first part of detection of respiratory and de- saturation events by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0053] Fig. 12 is a flowchart describing the second part of detection of respiratory and desaturation events that are described in Fig. 11.

[0054] Fig. 13a is an illustration of an example of detection of obstructive respiratory events from EMG amplitude by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0055] Fig. 13b is an illustration of an example of detection of central respiratory events from EMG amplitude by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0056] Fig. 14 is a block-diagram describing the evaluation of sleep stages by a device for the characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

[0057] Fig. 15 is a flow chart describing the incorporation of the information based on respiratory and de-saturation events, sleep stages and arousal events by a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention.

DESCRIPTION OF EMBODIMENTS

[0058] The present invention relates to a method, device, and system for characterizing sleep, and more particularly, to a method, device, and system for an efficient determination of wake and sleep stages, as well as sleep related respiratory events (also known as apnea and hypopnea events or breathing disorder) and their severity, using data derived from signals of electrical activity of the heart, such as electrocardiogram (ECG), and photoplethysmography (PPG) based signals, such as pulse oximeter and an option to include data derived from one or more electrical activity of muscles, i.e electromyogram (EMG).

[0059] An embodiment of the present invention discloses a method, device, and system that combine acquisition of the electrical activity of the heart (e.g. ECG), and pulse oximetry (SpO2 and pulse wave) signals, with or without one or more EMG (electromyogram) signals and enables the accurate characterization of sleep, detection of sleep stages, sleep related respiratory events, awakenings and arousals, and body position.

[0060] By analyzing the electrical activity of the heart the following information can be derived during sleep: respiratory data series, awakenings and arousals, autonomic function, body position and with some limitations - sleep stages.

[0061] By analyzing the oxygen saturation and pulse wave the following can be derived, during sleep: sensor reliability, oxygen saturation levels, desaturation events, and arousals from the pulse waveform envelope. [0062] By integrating data recorded in parallel from the electrical activity of the heart with data from the oxygen saturation and pulse wave of a sleeping person the following additional information or improvements can be derived: better estimation of the oxygen sensor reliability, pulse-transient-time (PTT) for estimating sympathetic nervous system (SNS) activity, better characterization of sleep stages, better characterization of arousals and awakenings - hence having better estimate of insomnia, better characterization of respiratory events and estimating the severity of Obstructive Sleep Apnea Syndrome (OSAS) by cross referencing respiratory data series with arousals and desaturations.

[0063] Adding EMG information recorded from any respiratory muscle to the above data can further improve the characterization of respiratory events by distinguishing between central and obstructive origin of each respiratory event. Adding EMG information recorded from any muscle that lowers its tonus during REM stage to the ECG, SpO2 and pleth data can further improve the characterization of sleep stages and arousals.

[0064] As was previously clarified, integration the data series allows interpretation mat cannot be made from either of the data series alone. Thus the limitations of other solutions that are based on the acquisition and analysis of only one of the above mentioned data series are overcome. An embodiment of the present invention provides information on sleep quality and architecture, thus allows for improved diagnosis of sleep related breathing disorder in comparison with screening devices based on pulse oximetry. It also provides much more specific details compared with ECG based sleep analysis devices. [0065] Reference is now made to define and clarify the terms used in the text that follows:

[0066] The terms "oximetry pulse waveform" "pulse wave" and "pleth" are used interchangeably herein and refer to the pulse waveform data series obtained by the photo- plethysomograph (PPG) sensor. [0067] The terms "sleep related respiratory events", "respiratory events" and "breathing disorder" are used interchangeably herein and refer to apnea and hypopnea events which can be of central, obstructive or mixed origin.

[0068] The terms "arousals" and "arousals and awakenings" events are used interchangeably herein. The main difference between awakenings and arousals is at the scale at which these non-sleep periods affect the ECG data.

[0069] Reference is now made to sleep characterization interpretations derived from the integration of various data series in accordance to an embodiment of the present invention:

[0070] The accuracy and reliability of the recordings of the oxygen detectors (pleth and SpO2 data) are better verified by utilizing ECG data.

[0071] The different stages of sleep are better identified by integrating in the analysis both ECG data and PTT data.

[0072] Respiratory events (apnea and hypopnea events) are better identified by integrating in the ECG analysis information obtained from the SpO2 data. [0073] Further improvement in the characterization of respiratory events can be obtained by classifying the sleep related respiratory event as a central sleep related respiratory event or obstructive sleep related respiratory event by adding EMG information from respiratory muscle to the above information.

[0074] Better identification of arousal and awakening states are obtained by integrating in ECG data analysis information obtained by analyzing additional time series recordings such as pleth and/or EMG (EMG recordings of muscle that lower their tonus during REM stage).

[0075] Improve in the characterization of each of the above mentioned sleep parameters improves in the characterization of the other sleep parameters (illustrated in Figure 15 that follows). For example: a correct determination of the different sleep and wake stages enables the discarding of false respiratory events determinations obtained when the examined person was awake.

[0076] Reference is now made to the Figures of an embodiment of the present invention:

[0077] Fig. 1 illustrates a block-diagram of the information processing path in a device for characterizing the physiological states of a sleeping person in accordance with an embodiment of the present invention. The text that follows suggests the technical requirements for constructing a device for the implementation of the information processing path, illustrated in the Figure. An acquisition device (10) is composed of several sensors and amplifier (20). The different sensors record simultaneously the electrical activity of .the heart (e.g. ECG) (12), pulse waveform (pleth) (14), oxygen saturation level in arterial blood (SpO2) (16), and, optionally, record one or more muscle tone (EMG signal) (18). The sensor for recording the activity of the heart is composed of at least 2 electrodes and one reference electrode. The acquiring of the pleth and SpO2 is usually done by the same detector, i.e. a photo-plethysomograph (PPG) operating with red and infrared light capable of recording percent of oxygenated blood (SpO2) and pulse waveform (pleth). The sensor for recording a single EMG signal is composed of two electrodes capable of recording electrical activity of a muscle. The amplifier (20) is capable of amplifying, digitizing and storing the data received from the acquisition unit (10). Preferred digitization of the electrical activity of the heart is a sample rate of 300Hz and quantization of 0.5μV (relative to input). Preferred digitization of the PPG includes averaging for 1 sec for the blood saturation, and 100Hz for pulse waveform. Preferred digitization of the EMG is a sample rate of 300Hz. The processor (22) is capable of applying the various algorithms, as shown in Fig. 3, on the digitized data and to produce a report that summarizes sleep analysis to be sent to an output device (24).

[0078] Fig. 2 illustrates a schematic block-diagram of a web based embodiment of the present invention. Several acquisition devices (10) (of which a single device is illustrated in Figure 1) are shown connected to the internet web. Each of the acquisition devices can transfer its recorded information via the web (26) to a processor (22). The processor (22), as shown in Fig. 1, is capable of applying the various algorithms to the digitized data and to produce a report that summarizes the sleep analysis course. The report can be directed via the internet web to one or more location provided with an output device (24). [0079] Fig. 3 illustrates a block-diagram of the main steps preformed by the processor (22) described in Fig 1 of the device in an embodiment of the present invention. The first step includes the preparation of the data series (100) (shown in Fig. 4), followed by detection of arousal and awakening events (200) (shown in Fig. 8), respiratory and desaturation events (300) (shown in Fig. 11 and 12), and sleep stages (400) (shown in Fig. 14). The entire information sets that were detected are analyzed and incorporated (500) (shown in Fig. 15). In addition, body position (BP) can be evaluated using a the same method as was used by Akselrod et al. (600) [0080] Fig. 4 is a flowchart of the main elements included in the preparation of the data series (100) described in Fig. 3. First, prominent fiducial points are marked on the ECG data (typically location of the R wave) (110) (shown in Fig 5), and the inter-beat interval is calculated to build the R-R interval (RRI) series. Then the validity of the oximetry data i.e. the SpO2 and pleth data series, is checked (120) (shown in Fig. 6). The next step includes the calculation of the pulse transient time (PTT) (130). The PTT is the time difference between the position of the R wave and the following peak in the pleth data series (illustrated in Fig. 5). In case there is an EMG data series it is recommended to remove the ECG artifact that contaminates it (150). Knowing the exact locations (in time) of the R waves simplifies the process of removing the ECG artifact. For example, by disregarding the EMG data that occurred in the exact time (including some margins) of the R wave occurrence. A margin of at least 0.05 sec from either side of the R wave location is recommended.

[0081] Fig. 5 is an illustration of an RR interval (RRI) and pulse transient time (PTT) for the understanding the definitions of the terms used in the text. The figure shows ECG tracing (123) of 4 heart beats (designated 127a, 127b 127c and 128d) and the corresponding pleth tracing (125). For each beat the first upward deflection within the sharp complex in the ECG is denoted as the R wave. The peak of this wave is the R wave location (and labeled as 'R'). The time difference between consecutive R-s is defined as the RR interval. This interval is inversely related to the instantaneous heart rate. PTT is the time elapsed from the peak of the R wave in the ECG to the corresponding peak in the pleth data.

[0082] Fig. 6 is a flowchart describing oximetry data validity check procedure (120) described in Fig. 4. Note the algorithm-function checks the validity of each small segment of the data series. The first step checks the validity of the SpO2 data (121) by looking for invalid values or extreme slopes. For example, SpO2 values (which are measured in percentage) of above 100% or below 50% are disregarded. In addition, SpO2 data with local slope of above 10% per second are disregarded. Failing to pass the SpO2 validity test automatically disqualifies the corresponding pleth segment (122). Then an autocorrelation between pleth segments is done (124). The autocorrelation is done between pleth segments that correspond to consecutive beats as defined based on the R wave position (illustrated in Fig. 7). If the autocorrelation value of the pleth segment is above a certain threshold (126), for example zero, the segment of the oximetry data is considered valid (128). Otherwise the segment (the specific portion of the data) is considered not valid both for SpO2 and pleth. SpO2 or pleth segments that had failed to pass the validity check will be disregarded in any further calculation. An example of the validity check is illustrated in Fig. 7.

[0083] Figure 7 is an example of oximetry validity check. This figure concentrates only on the validity check of the pleth data. The figure shows the pleth and ECG data in the upper and lower panels (129 and 131 respectively) as a function of time. The R wave locations are indicated as circles on the ECG time series (in 131). The arrows connect each R wave peaks and its corresponding pleth peak (in 129). The autocorrelation of the pleth data was done by comparing the pleth data that corresponds with a specific beat (its location was defined based on the ECG) relative to the pleth data of previous beat. If the autocorrelation value is below a certain threshold, for example zero, this pleth segment is disregarded. Having several nearby segments that are considered disregarded results in disregarding the entire section, For example any section of 10 beats which includes at least 4 disregarded segments (each segment at a size of a single beat) of pleth will result in disregarding the entire 10 beats. Such region will be defined as 'bad' region both for SpO2 and pleth data and their data will be disregarded. An example of a region with 'bad' pleth data can be seen in the figure in the region between the time measurements 4186 and 4192 seconds.

[0084] Fig. 8 is a flowchart describing the detection of arousal and awakening events (200, shown in Fig. 3) from data of electrical activity of the heart (e.g. ECG), pulse oximetry and if available, relevant EMG (obtained from muscle that lower its tonus during REM stages). The main difference between awakenings and arousals is at the scale at which these non- sleep periods affect the ECG data. Specifically the awakening periods, which are typically characterized by trace duration of at least 15 seconds, affect the ECG data in the low frequencies region while the arousals periods, which are typically characterized by trace duration of 3-15 seconds, affect the ECG data in the intermediate-high frequencies region.

[0085] The initial part includes arousal detection based on ECG data (210). According to a preferred embodiment of the present invention, the RRI series is filtered using a low-pass- filter thereby providing a first series of data. Similarly, for the purpose of determining the arousal periods, the RRI series is filtered using a band-pass-filter thereby providing a second series of data. A typical cutoff frequency for the low-pass-filter is about 0.01 Hz, and typical cutoff frequencies of the band-pass-filter are 0.05 Hz for the low limit and about 0.2 Hz for upper band limit. Awakening periods are defined as a plurality of beats each associated with at least one of the first series of data which is below a predetermined threshold. Arousal periods are defined as a plurality of beats each associated with at least one of the second series of data which is below a predetermined threshold.

[0086] The calculation may result in positive detected arousals periods and suspected periods, defined based on the threshold used. Each of these events follows several tests (215) prior to acceptance as valid arousal (230). Typical thresholds for positively identifying the awakening and arousals events are at 0.85 of the averaged value of the first series and the second series of data, respectively.

[0087] Arousal events can also be identified by weaker indication obtained from the ECG, for example using a threshold of 0.9 of the averaged value in either of the data series (first or second filtered data), which are accompanied by a decrease in the pleth's amplitude (240). The local pulse wave amplitude (PWA) is defined as the difference between the local maximum and the local minimum of the pleth data, an example of the time window for calculating the local maximum (or minimum) can be 1.2 times the average RRI of the entire data. A considered decrease in the PWA (see figure 9) can be defined as plurality of beats each associated PWA values which are below a predetermined threshold. Such threshold can be about 0.7 of the averaged value of PWA of the preceding region.

[0088] In case relevant EMG data is available, weak indications of arousal or awakening events from the ECG, for example using a threshold of 0.9 of the averaged value in either of the data series (first or second RRI filtered data), can also be identified as an event if they are accompanied by an increase in the EMG (250). The increase in EMG data can be observed for example using the calculation of the mean rectified of the EMG data (mrEMG). mrEMG is defined as the moving average of the absolute value of the amplitude of the EMG data. A considered increase in the EMG data (see Figure 10) can be defined as increase in the mrEMG above a predetermined first threshold for a short period surrounded by a longer period in which the mrEMG is above a second (lower) threshold. Such thresholds can be 3 times and 1.5 times the average mrEMG values of the preceding period for the first and second threshold respectively. Recommended periods can be 1 and 3 seconds to be used in the first and second threshold respectively.

[0089] Fig. 9 is an example of the pulse wave amplitude (PWA) (245) as a function of time prior and during an arousal event. The PWA series is calculated as the difference between the local maximum and local minimum of the pleth data in which the window for the local calculation is 1.2 times the average RRI of the entire data. The dashed line (246) indicates the beginning of the arousal event. The decrease in the PWA series at the time of the arousal event relative to the time prior to the event can be easily seen.

[0090] Fig. 10 is an example of the submental mean rectified EMG (mxEMG) (255) as a function of time prior and during an arousal event. The mrEMGT is defined as the moving average of the absolute amplitude of the EMG data. The dashed line (256) indicates the beginning of the arousal event. The increase in the mxEMG series at the time of the arousal event relative to the time prior to the event can be easily seen.

[0091] Fig. 11 is a flowchart describing the first part of the detection of respiratory and desaturation events (300, shown in Fig 3). According to the American academy of sleep medicine task force ["Sleep related breathing disorders in adults: recommendations for syndrome definition and measurement techniques in clinical research." Sleep 1999, 22: 667-689] apnea can be classified using a reduction in a respiratory time series with desaturation or arousal.

[0092] The first stage in the evaluation is the detection of desaturation events based on valid SpO2 data (310). A desaturation event is defined as a decrease in SpO2 below a predefined threshold relative to baseline value. An example of such threshold can be a decrease of 3%.

[0093] Following, attenuations in the ECG derived respiratory (EDR) time series are detected (315). EDR can be extracted from the waveform parameters of the ECG according to Moody et al. [Moody G.B., Mark R.G., Zoccola A., and Mantero S. (1985): "Derivation of respiratory signals from multi-lead ECGs", Comp. Cardiol., 12, pp. 113-6]. In order to detect the attenuations in the EDR time series we define an EDR attenuation parameter which is the minus of the ratio between a moving characteristic of the amplitude of. the EDR over a first time window relative to the moving averaged amplitude of the EDR over a second time window. The moving characteristic can be a predefined percentage of EDR amplitude, such as the percentage 85. It is recommended that the first time window will be larger than the second time window, for example the length of 20 averaged breathing periods for the first time window and the length of 1.5 average breathing periods for the second window. The depth of the attenuation is thus defined by this ratio

[0094] The respiratory events are then detected based on several criteria (317) that combine the information from prior calculations. Respiratory events are identified (325) by any of the following: [0095] Deep attenuation as expressed by low EDR attenuation parameter below a predefined first threshold (320), in which the first threshold might be -4.

[0096] Deep desaturation event below a predefined second threshold (330), in which the second threshold might be 4. [0097] Combination of a moderate attenuation in the EDR series in which the EDR attenuation parameter is below a predefined third threshold with a moderate desaturation below a predefined forth threshold (335), in which the third threshold might be -2 and the forth threshold might be 3.

[0098] Combination of moderate attenuation in the EDR in which the EDR attenuation parameter is below a predefined fifth threshold followed by an arousal event (340). In which the fifth threshold might be -2.

[0099] Fig. 12 is a flowchart describing the second part of the detection of respiratory and desaturation events. Following the detection of respiratory events (show in Fig. 11) there is a further analysis executed for each event (317) that can be preformed only if relevant EMG data exist, i.e EMG data related to respiratory muscle (355). If there is an increase in the EMG data, as characterized for example by the mrEMG (the moving average of the absolute amplitude of the EMG data) above a predefined threshold, such as 3 times the original value, (365) at the same time as the respiratory event, the event is defined to rise from an obstructive source (370, illustrated in detail in Fig. 13a). If on the other hand the EMG data decreases below a predefined threshold, such as one third of the original value (375), the event is defined to arise from a central source (380, illustrated in detail in Fig. 13b). Otherwise, the event is suspected to arise from obstructive source (385). In case EMG data is not available the respiratory events will not be divided into central and obstructive (360). [00100] Figure 13a and 13b are examples of detection of obstructive respiratory events and detection of central respiratory events (designated 370 and 380 in Fig. 12) from the mean rectified EMG (mrEMG) time series (377) as a function of time. mrEMG is defined as the moving average of the absolute amplitude of the EMG signal. The obstructive events are seen as an increase in the mrEMG when compared to a decrease in the mrEMG in central events.

[00101] Fig. 14 is a block-diagram describing the evaluation steps of sleep stages

(400, shown in Fig. 3). The sleep stages that are identified in accordance with an embodiment of the present invention are: wake, REM sleep and Non-REM sleep which is further divided into two 2 stages. Light sleep (LS) stage which combines Non-REM stages 1 and 2, and slow wave sleep (SWS) stage which combines Non-REM stages 3 and 4. The sleep stages classification is based on calculation of several parameters of the ECG₅ RRI, PTT and if available, relevant EMG data. The first step includes the evaluation of waveform, time domain and frequency domain parameters of the ECG and RjRI signals (410). The ECG waveform parameters includes the extraction of left R wave duration (L- RWD), right R wave duration (R-RWD) and R wave amplitude (RWA) for each R wave. The L-RWD is defined as the time duration between the inflection point just prior to the R wave fiducial point and the R wave fiducial point. The R-RWD is defined as the time duration from the R wave fiducial point and the inflection point just following it. The RWA is the amplitude of the R wave with reference to the minimal value out of the local minimal values obtained in either of its sides. The RRI time domain parameter is a nonlinear parameter indicated as BQ, which is the balance between the number of points in the odd and even quartiles in the phase space constructed by two adjacent RRI values (i.e. Poincare plot of RRI). The RRI frequency parameters are obtained by a time-frequency decomposition (e.g. wavelet analysis) that is performed on the RRI series. The output of such analysis include several frequency domain parameters, that reflect the activity of the sympathetic and parasympathetic nervous system, such as the power of the RKI series in different frequency ranges as a function of time. The recommended frequency bands are very low frequency (VLF) at 0.008-0.04Hz, low frequency (LF) at 0.04-0.15Hz, and high frequency (HF) at 0.15-0.5Hz. Then the time-dependent spectral analyses of the PTT signal (420) need to be evaluated, in order to to obtain its low frequency (CLF) parameter as a function of time. The recommended frequency band for this frequency band is the same as for the LF of the RRI i.e. 0.04-0.15Hz. CLF is known to correlate with sympathetic nervous system activity. In case relevant EMG data is available, i.e EMG data related to muscle that lower its tonus during REM stage, (425) the time and frequency domain parameters of the EMG are needed to be evaluated (430). The recommended EMG time domain parameters are mrEMG (defined above), zero crossing frequency (ZC), defined as the number of times the signal crosses 0 level during a predefined period, and turns which is defined as the number of times the signal derivatives crosses zero level or the number of times the signal changes direction during a predefined period. The recommended time period for ZC and turns parameters is 30 seconds. The EMG frequency parameters are the normalized power (nPWR), defined as the mean of the power spectrum of the signal, and PVAR, defined as the variance of the absolute value spectral coefficients of the signal. The frequency band on which the nPWR and PVAR are calculated is recommended to start at f_L=20Hz and ends at a frequency fj_H for which the range [TL₅ fΗ] contain 95% of the total power in the range [T_L5 sample rate/2]. If the EMG data are available all the above parameters are used as input into a Bayesian classifier, otherwise only the ECG, RRI and PTT parameters will be used. The Bayesian classifier (440) uses a priori probabilities of different sleep stages and a database of these parameters that were calculated for known wake / sleep states, to determine current sleep and wake stages for the whole duration of the recording.

[00102] Fig. 15 is a flowchart describing the incorporation of the information based on respiratory and desaturation events, sleep stages and arousal events (designated as 500 in Fig. 3). In case a desaturation events occurred while the patient was at a wake stage (510) the desaturation event is disregarded (520). Similar, if a respiratory event occurs while the patient was at a wake stage (530) the respiratory event is disregarded (540). In addition, if there are several close arousals with no respiratory event that occurs at the same time (550), the arousal events may be attributed to a single long event (560). Moreover, if there are long close-in-time arousals, with no respiratory event in between (570) the assessment of the section should change to wake (580). Thus, the incorporation of the "sleep-information" in accordance with the flowchart results in accurate observations of respiratory and desaturation events, sleep staging and arousal events (590). [00103] It should be clear that the description of the embodiments and attached

Figures set forth in this specification serves only for a better understanding of the invention, without limiting its scope.

[00104] It should also be clear that a person skilled in the art, after reading the present specification could make adjustments or amendments to the attached Figures and above described embodiments that would still be covered by the present invention.

Claims

1. A method for determining an incident of a sleep related respiratory event, from ECG data and oximetry data, the method comprising: determining occurrence of an attenuation in ECG derived respiratory signal below a first threshold, determining occurrence of a desaturation in the oximetry data below a second threshold, determining occurrence of an attenuation in ECG derived respiratory signal below a third threshold with a corresponding desaturation in the oximetry data below a forth threshold, determining occurrence of an attenuation in ECG derived respiratory signal below a fifth threshold with a corresponding detection of arousal; and determining the incident of a sleep related respiratory event if any of the occurrences exists.

2. The method as claimed in claim 1, further comprising using EMG data to classify the sleep related respiratory event as a central sleep related respiratory event or obstructive sleep related respiratory event.

3. The method as claimed in claim 1, further comprising determining sleep stages using the ECG data and pulse transit time derived from the ECG data and a pleth data.

4. The method as claimed in claim 3, further comprising determining sleep stages using EMG data, ECG data and pulse transit time derived from the ECG data and a pleth data.

5. The method as claimed in claim 1, further comprising receiving the ECG data and the oximetry data from a remote user over the Internet.

6. The method as claimed in claim 5, further comprising providing information on the sleep related respiratory event over the Internet to the user.

7. A device for determining an incident of a sleep related respiratory event, comprising: an ECG sensor, a oxygen saturation sensor, a processor provided with an algorithm for determining an incident of a sleep related respiratory event, from ECG data and oximetry data, the algorithm comprising: determining occurrence of an attenuation in ECG derived respiratory signal below a first threshold, determining occurrence of a desaturation in the oximetry data below a second threshold, determining occurrence of an attenuation in ECG derived respiratory signal below a third threshold with a corresponding desaturation in the oximetry data below a forth threshold, determining occurrence of an attenuation in ECG derived respiratory signal below a fifth threshold with a corresponding detection of arousal; and deterrnining the incident of a sleep related respiratory event if any of the occurrences exists.

8. The device as claimed in claim 7, further comprising an EMG sensor.

9. The device as claimed in claim 8, wherein the algorithm of the processor further comprises using EMG data to classify the sleep related respiratory event as a central sleep related respiratory event or obstructive sleep related respiratory event.

10. The device as claimed in claim 8, wherein the algorithm of the processor further comprises determining sleep stages using EMG data, ECG data and pulse transit time derived from the ECG data and a pleth data.

11. The device as claimed in claim 7, wherein the algorithm of the processor further comprises determining sleep stages using the ECG data and pulse transit time derived from the ECG data and a pleth data.

12. The device as claimed in claim 7, further provided with a communication link wherein the algorithm of the processor further comprises receiving the ECG data and the oximetry data from a remote user over the Internet using the communication link.

13. The device as claimed in claim 12, wherein the algorithm of the processor further comprises providing information on the sleep related respiratory event over the Internet to the user using the communication link.

14. The device as claimed in claim 7, further comprising an amplifier for amplifying signals received from the sensors.

15. The device as claimed in claim 7, wherein said oxygen saturation sensor is selected from a group of sensors consisting of a pulse waveform sensor and SpO2 sensor.