JP2020036949A - Automated control for detection of flow limitation - Google Patents

Abstract

To provide improved methods and apparatus for detection, diagnosis and/or treatment of respiratory conditions such as the conditions related to obstructive sleep apnea hypopnea syndrome (OSAHS) or obstructive sleep apnea (OSA).SOLUTION: A respiratory flow limitation detection device, which can include an airway pressure treatment generator, determines a flow limitation measure based on one or more shape indices for detecting partial obstruction and a measure of a patient's ventilation or respiratory duty cycle. The shape indices is based on function(s) that ascertain the likelihood of the presence of M-shaped breathing patterns and/or chair-shaped breathing patterns. The measure of ventilation may be based on analysis of current and prior tidal volumes to detect a less than normal patient ventilation. The duty cycle measure may be a ratio of current and prior measures of inspiratory time to respiratory cycle time to detect an increase in the patient's inspiratory cycle time relative to the respiratory cycle time.SELECTED DRAWING: Figure 5

Description

[Cross-reference of related applications]
This application is based on Australian Provisional Patent Application No. 2007902561, filed May 11, 2007, and U.S. Provisional Patent Application No. 60/200, filed August 17, 2007, the disclosures of which are incorporated herein by reference. Claim the benefit of the filing date of 965,172.

[Field of the Invention]
The proposed technique is a method and apparatus for detecting, diagnosing, and / or treating a respiratory condition, such as sleep apnea hypopnea syndrome (OSAHS) or a condition associated with obstructive sleep apnea (OSA). About.

Sullivan in US Patent No. 5,199,424 issued April 6, 1993.
The application of continuous positive airway pressure (CPAP) has been used as a means to treat the occurrence of obstructive sleep apnea, as described by Lynch et al. The patient is connected to a source of positive pressure air via a nasal mask or nasal prongs. The air source breathed by the patient is slightly above atmospheric pressure. It has been found that the application of continuous positive airway pressure results in what can be described as an "air splint" that supports and stabilizes the upper airway, thus eliminating the occurrence of upper airway obstruction. Application of continuous positive airway pressure is effective in relieving both snoring and obstructive sleep apnea, and is often effective in treating central and mixed apnea.

U.S. Patent No. 5,549, issued to Gruenke on August 27, 1996.
No. 106 discloses a device aimed at relieving a patient of respiration to treat mixed and obstructive sleep apnea. The device increases the intranasal air pressure delivered to the patient's airways just prior to inspiration, and then decreases the pressure to ease expiratory effort.

In U.S. Pat. No. 5,245,995, Sullivan describes how snoring and abnormal breathing patterns can be measured by inspiratory and expiratory pressure measurements during sleep, resulting in pre-occlusion seizures or early signs of other forms of respiratory disease. We are examining whether it can be detected. In particular, patterns of respiratory parameters are monitored, and CPAP pressure is detected upon detection of a pattern that is predefined to result in increased airway pressure, ideally to eliminate the occurrence of obstructive strokes and other forms of respiratory disease. Can be raised.

Berthon in US Patent No. 5,704,345, issued January 6, 1998.
As described by Jones, various techniques for sensing and detecting abnormal breathing patterns indicative of obstructive breathing are known. Berthon-Jones is apnea, snoring,
And a method based on detecting events such as respiratory flow flattening is described. The treatment pressure may be adjusted automatically in response to the detected condition.

As described by Wickham in International Patent Application PCT / AU01 / 01948 (Publication No. WO02 / 18002), flow flattening decisions may be further based on various weighting factors. Weighting factors are applied to areas of airflow to increase sensitivity to various types of respiratory obstruction.

Other methods of detecting occlusion have also been used. For example, US Pat. No. 5,490,
In U.S. Pat. No. 5,502 and 5,803,066, Rapport attempts to minimize the flow of air from the flow generator while attempting to ensure that no flow restriction appears in the patient's airway. A method and apparatus for optimizing positive pressure, which is controlled in a controlled manner, is disclosed.
The controlled positive pressure into the patient's airway is adjusted by detecting flow restriction from the shape of the inspiratory flow waveform. The pressure setting is raised, lowered, or maintained, depending on whether a flow restriction has been detected and the action taken earlier by the system.

In U.S. Patent No. 5,645,053, Remmers describes a system for automatically and continuously adjusting the level of nasal pressure to an optimal value during an OSA (obstructive sleep apnea) procedure. Parameters related to the shape of the time profile of the intake flow, such as the degree of roundness and flatness of the intake profile, are determined. OSA treatment is then performed by automatically re-evaluating the applied pressure and continuously searching for the minimum pressure required to properly inflate the patient's pharyngeal airway.

Despite the availability of such devices to treat OSA, certain sleep obstruction events have not yet been treated by the use of certain devices. Therefore, new methods of automatic detection and treatment of occlusion events are desired.

According to aspects of the present technology, devices and methods are provided with improved automatic detection and / or treatment of sleep disordered breathing or flow restriction.

According to another aspect of the present technology, an improved detection and / or treatment of a flow waveform representing a partial occlusion is provided.

According to yet another aspect of the present technology, there is provided an apparatus and method for a more rapid response to an indication of partial occlusion or flow restriction.

According to yet another aspect of the present technology, to more accurately detect an obstruction event, a flow restriction or partial obstruction, such as a flow restriction waveform or its shape index, may be measured by airway measurements or detected secondary conditions. An apparatus and method for optimizing an indicator are provided.

Aspects of the present technology include determining a respiratory flow index, determining a shape index representing a pattern of flow restriction from the respiratory flow index, and determining a ventilation index or a breath load cycle index from the respiratory flow index. Pertaining to a method for detecting a flow restriction, which may include the step of: deriving a flow restriction index as a function of the determined shape index and one or both of the determined ventilation index and the duty cycle index.

In certain embodiments, the shape index may be an index of flattening, an index of “M” shaping, an index such as chair shaping and / or roundness, or any other index that may represent a partial occlusion. . The ventilation index may be a tidal volume index, such as the ratio of the current tidal volume to the previous tidal volume. The breath duty cycle index may be a ratio, such as the ratio of the current breath inspiration time to the breath cycle time, and the ratio of the prior average breath inspiration time to the breath cycle time.

These methods may be performed by a flow restriction detector and / or by a flow restriction pressure therapy device. For example, these methods may include increasing the pressure as a condition that the shape index represents the presence of an M-shaped breath in the respiratory airflow, and that the ventilation index decrease sufficiently to represent less than normal ventilation. May be implemented to adjust the treatment pressure of the pressure treatment device. In some cases, the treatment pressure value may be such that the shape index is M in the respiratory airflow.
It may be adjusted or increased as a condition of representing the presence of shape breath and of increasing the duty cycle index.

In one embodiment of the present technology, an apparatus for detecting a flow restriction includes a patient interface communicating a flow of respiratory gas, and a flow sensor coupled to the patient interface for generating a flow signal representative of the flow of respiratory gas through the patient interface. And a controller coupled to the flow sensor and processing the flow signal, the controller for deriving a flow restriction index based on the shape index and one or both of the ventilation index and the breath load cycle index. , Configured to control the method of detection. The apparatus includes a controller and a controller configured to calculate the pressure demand as a function of the flow restriction indicator and to configure the flow generator according to the pressure demand, such as the regulation described herein. It may further include a flow generator coupled to the patient interface.

In one embodiment of the present technology, a system for flow restriction detection includes an interface means for communicating a flow of breathing gas, and a flow signal coupled to the interface means and representing a flow of breathing gas through the interface means. The flow measurement means may include a processing means coupled to the flow measurement means for processing the flow signal, the processing means comprising a flow restriction index based on the shape index and either the ventilation index or the breath load cycle index. , To process the method of detection. The system is coupled to the processing means and the interface means, and controls the flow of the controlled breathing gas through the interface means, such that the processing means may be configured to calculate the pressure demand as a function of the flow restriction indicator. It may further comprise flow means for generating and setting the flow generator according to the pressure demand.

In another embodiment, the technology may be an information-bearing medium having processor-readable information such that a device for detecting respiratory flow restriction can be controlled. The processor-readable information determines a measure of respiratory flow, determines a shape index representing a pattern of flow restriction from the measure of respiratory flow, determines a ventilatory measure and / or a breath duty cycle measure from the measure of respiratory flow; A control instruction may be included that derives a flow restriction index as a function of the determined shape index and one or both of the determined ventilation index and breath duty cycle index. The processor readable information may further include instructions for calculating the pressure demand as a function of the flow restriction indicator and adjusting the pressure as described herein.

Another aspect of the present technology relates to generating pressure or flow control information or signals that are proportionally derived as a function of a measure of flow restriction. For example, a controller or processor
It may be configured to control or determine a pressure setting or a flow rate setting for the respiratory treatment device. Based on the respiratory flow index, an occlusion index or shape index, representing the degree of occlusion or flow restriction, may be determined or calculated by the device. The device may also, in some cases, determine a ventilation index from the respiratory flow index that indicates the degree of change in ventilation. Therapeutic pressure setting or flow rate setting is a proportional function of one or both of (1) the degree of occlusion or flow restriction, and (2) the degree of change in ventilation.
May be derived by the device.

  Further embodiments of the present technology are apparent from the following disclosure.

The present technology is described by way of example, and not by way of limitation, in the figures of the accompanying drawings where like reference numerals refer to like elements.

It is a figure which shows the airflow of the patient who breathes normally first, the airflow of the patient who has airway constriction first, and the patient who is finally awake. FIG. 9 shows a flattening index for a patient having an obstruction indicated by an “M” shaped breathing pattern. FIG. 9 illustrates a flattening index with or without multi-breath averaging for a patient having an obstruction indicated by an “M” shaped breathing pattern that leads to a breathing pattern indicating awakening from sleep for the patient. FIG. 3 illustrates an example of components of a device for detection and / or treatment of flow restriction or partial obstruction. FIG. 3 illustrates the appropriate steps of the detection or treatment device in detecting a flow restriction or partial obstruction. FIG. 8 is a diagram illustrating an example of a fuzzy membership function of a variable based on a single breath flattening shape index. It is a figure showing the example of the fuzzy membership function of the variable based on m shape breath shape index. FIG. 4 is a diagram illustrating an example of a fuzzy membership function of a variable based on a permeability index. FIG. 5 is a diagram illustrating an example of a fuzzy membership function of a variable based on a breath load cycle ratio. FIG. 4 is a diagram illustrating an example of an output membership function of flow restriction measurement. FIG. 7 is a diagram illustrating an example of a defuzzification operation in determining a flow restriction or a partial occlusion index using the centroid method. FIG. 7 illustrates an example of a function that modifies the ability to detect a flow restriction or partial occlusion index based on an index of ventilation. FIG. 3 is a graphical comparison of a calculated flow restriction index FFL of the technique described herein and a conventional flattening index determined from the same flow signal. FIG. 7 shows a flow signal derived from nasal flow and the output index of a continuous RERA detector quantifying the likelihood of respiratory arousal. FIG. 7 is a diagram showing the ratio of peak expiratory flow to average expiratory flow, which may be an index of sleep-wake-wake detection. FIG. 3 shows two superimposed flow signals, a filtered flow signal and an unfiltered flow signal recording a sequence of obstructive breaths during low frequency snoring. FIG. 4 illustrates the response of an FIR filter embodiment useful for filtering airflow signals. FIG. 9 is a diagram plotting an example of a function functioning as basic vectors B1 and B2 for detecting M-shaped respiration. Fig. 4 is a graph of a typical M-shaped breath flow signal. FIG. 6 is a graph of a flow signal showing an M-shaped breath enhanced by simple chair shaping. FIG. 4 is a graph of a membership function of a variable based on the ratio of peak inspiratory flow to average inspiratory flow, which is useful for detecting an M-shaped breath having a chair shape. Figure 7 is a graph of a membership function of a variable based on inspiratory peak position that is useful for detecting M-shaped breaths having a chair shape. FIG. 4 shows a graph of a flow signal, a graph of a determined flatness index, and a graph of a determined chair shaping index. FIG. 11 is a graph of a flow signal of a patient experiencing upper airway resistance with increasing respiratory effort ending during sleep awakening at t = 420, and a graph of a corresponding single breath flattening index. 25 is a graph of an example of a flow restriction index, a ventilation index, and a duty cycle index based on the flow signal of FIG. 24. 25 is a graph of a continuous obstruction index or a continuous flattening index based on the sequence of breaths shown in the flow signal of FIG. 24. FIG. 7 shows a function of a weight reduction coefficient that can be used for correcting a flow restriction index. FIG. 4 is a diagram showing a graph of a respiratory flow signal, a graph of an instant snoring signal, and a graph of three snoring indices. It is a figure showing the classification which handles snoring based on a plurality of snoring indices. FIG. 6 is a graph of an M-shaped breath marked with accurate and incorrect trim results. FIG. FIG. 5 includes a graph of flow signal including M-shaped breath versus time and a graph of inspiratory time determined by two methods. FIG. 6 shows a graph of a function useful for reducing the weight of the flow restriction index depending on the leak condition. FIG. 8 is a diagram of a further function that helps to reduce the weight of a flow restriction indicator or a partial occlusion indicator (eg, FFL) based on a leak condition. 5 is a graph of a function suitable for reducing the weight of a flow restriction or partial occlusion indicator depending on the level of mask pressure. 4 is a graph of a function suitable for reducing the weight of a flow restriction or partial obstruction index by an obstruction index. Fig. 4 is a graph of a function suitable for reducing the weight of the snoring index as a function of treatment pressure. 5 is a graph of a function suitable for reducing the weight of a snoring index based on a ventilation index representing breath. FIG. 4 is a histogram of normalized expiratory peak positions for various groups of patients. Figure 7 is a graph of a function of a weight reduction factor to be applied to a flow restriction or partial occlusion index based on a normalized expiratory peak position value.

Patients with OSA often have apnea or hypopnea during sleep that is terminated only by waking the patient. These frequent events cause sleep fragmentation and stimulation of the sympathetic nervous system. This can have serious consequences for the patient, such as daytime sleepiness (which may be accompanied by a car accident), poor mental activity, memory problems, depression, and high blood pressure. OS
Patients with A may snore loudly and, at the same time, interfere with their partner's sleep. The best form of treatment for patients with OSA is continuous positive airway pressure (CPAP) applied by a blower (compressor) through a connecting hose and mask.
It is. Positive pressure prevents narrowing of the patient's airway during inspiration, thus preventing frequent apnea or hypopnea and their sequelae.

Typically, patients undergo two sleep surveys that are monitored using a number of sensors whose records are retrieved, known as polysomnography. The first study (without treatment) confirms the diagnosis of OSA and the second study is used to titrate the patient to the correct treatment pressure.
Pressure requirements change over night due to changes in position, posture and sleep state. Doctors suggest pressures that may cover any chance (titration pressure). An alternative to this is an automatic machine that adapts the pressure to the needs of the patient, a therapy known as automatic positive airway pressure or APAP. An advantage of APAP therapy is that it is compatible with pressure requirements that may change over various time scales. For example,
Pressure requirements change during the night with posture and sleep status, possibly during the weekend with alcohol consumption, change over months due to the beneficial effects of the therapy itself, weight loss or May change over the years due to increase. In addition, APAP machines typically increase pressure as needed, so that patients can sleep at comfortable pressures well below any therapeutic maximum. Finally, as APAP machines operate at potentially lower pressures, the effects of mask leakage tend to be somewhat improved.

A known algorithm used to automatically set patient pressure on APAP machines is called ResMed AutoSet. Overall, the AutoSet device and its algorithms are excellent for treating OSA patients. The ResMed AutoSet algorithm responds to three situations: flow limitation, snoring (audible noise), and apnea. Automatic pressure setting is also described in US Pat. No. 5,704,345, the contents of which are expressly incorporated herein by cross-reference. Flow restriction is a hydrodynamic property of a so-called "extruded tube" that carries a fluid flow. The pharynx of a patient with OSA is an example of an extruded tube (even though it is a muscle extruded tube rather than a simple passive tube being investigated on a benchtop). In essence, flow restriction is due to the flow decreasing downstream pressure (ie, increasing the flow driven differential pressure) (assuming the upstream pressure is kept constant) in the constricted tube carrying the flow. It is in a state where it is not increased any more. In a patient with a stenotic upper airway, this condition is the same as a situation in which the patient no longer has adequate ventilation and increased respiratory effort no longer increases inspiratory flow. FIG. 1 shows that such a patient first breathes normally at 102, then an airway constriction occurs at 104, and despite the greater negative excursion of the canteen pressure (an indicator of respiratory effort), it is imperative. An example is shown in which the tidal volume or flow rate cannot be increased. Eventually, the patient opens the airway,
Awaken at 106 to take several large breaths to restore gas homeostasis.

The ResMed AutoSet algorithm monitors patient flow and increases pressure when detecting flow restriction or snoring. Apneas are rarely encountered because they are usually preceded by a period of flow restriction (sometimes referred to as partial obstruction) or snoring. As a backup, the pressure is also increased when apnea is detected. If flow turbulence is not measured, the pressure can be reduced slowly and, if possible, an equilibrium pressure is achieved that allows the patient to sleep without waking. Since the AutoSet algorithm responds proportionally, a metric is used for each condition to which the algorithm responds. The metrics are used: a flattening index for flow restriction, a calibrated RMS index of the sound averaged during snoring inspiration, and the length of the detected apnea.

The flattening index is a dimensionless feature (eg, a real number) calculated using the patient's inspiratory waveform. The flattening index attempts to measure in principle how flat the top of the waveform is. A feature of the flow restriction is that the downstream pressure is sufficiently low to keep the tube constricted, while the flow velocity is maintained at a substantially constant value, independent of changes in drive pressure. In patients with flow-limited breathing, this is the same as the inspiratory waveform including the flat top (ie, constant inspiratory flow rate).

For example, depending on the selected scale of the index, a normal breath may produce a flattening index of about 0.2, while a significantly flattened waveform may have a flattening index of about 0.1 or less. May generate a chemical index. In the case of APAP therapy, limits are typically set (eg, 0.19 on some machines) and pressure is increased proportionally to the extent that the flattening index is below the threshold. To reduce the effects of noise and increase specificity, a typical pressure setting algorithm may use a moving average over five breath points. A possible disadvantage of the five breath average is that it slows down the detection of flow restriction, as flattening requires five abnormal breaths to be averaged before falling to its final low point. It is. The three heuristic weightings are to the extent that greater reductions in flattening are needed as leakage increases, as treatment pressure increases, and as evidence of valve-like breathing increases. Flattening may be applied to a threshold that causes an increase in pressure. These rules of thumb help prevent potential pressure spikes despite reduced information (flow signal).

The flattening index is a good indicator of flow restriction, but is designed to detect certain situations. However, it has been found that some specific implementations do not cover some rare situations. The fields in which such observations were made are listed below.

A 1.5 breath moving average slows the detection of flow restriction. This is shown in FIG. In FIG. 2, the upper trace shows a plot of a conventional 5 breath moving average flattening index. The lower trace shows an indicator of respiratory flow. The patient begins to occlude little by little and the flattened trace drops stepwise at 202 due to a five breath average. Towards the right side of the graph, the occlusion becomes more severe and gradually begins to become more "M" in shape. As shown, the flattening index eventually begins to reverse direction and increases, rather than decreases, as the occlusion worsens.
2. Since the various intake shapes may average out to give a completely new shape,
The five breath moving average can be significant. This is shown in FIG. FIG. 3 shows a sequence of so-called M-shaped obstructed breaths at 302, ending with awakening at 304, followed by some reasonably normal recovery breath at 306.
As shown in the graph, both single breath flattening and conventional flattening are high at the end of the M breath series (the former reaches the maximum more quickly), and after awakening, conventional flattening actually 0
. It drops to less than one, not because the breath is flattened, but rather because M breaths are averaged together with normal breaths, producing a pseudo-flat shape.
3. The flattening index is not designed to detect M breath. In fact, the flattening index is M
High when shape breathing occurs. This is shown in FIGS.
4. The flattening index can cause an increase in pressure, independent of the current ventilation or sleep state of the patient-user of the device.
5. Heuristics applied to reduce the weight of flattening can cause undertreatment in certain patients.
6. The flattening index is subject to normal random fluctuations that have significance to the sensitivity and specificity of any algorithm that uses the flattening index to detect flow restriction.

Although M-shaped breaths may be rare, it is still desirable to develop additional methods and equipment to detect flow restrictions and / or improve existing methods and equipment.

Referring to FIG. 4, the present technology relates to a pressure delivery and / or flow restriction detection device that may include a flow generator, such as a servo-controlled blower 402. The device typically provides a patient interface, such as a mask 406, and a flow of air or breathing gas to the patient, and / or
Alternatively, it further includes an air delivery conduit 408 that carries the patient. Blower 402 may be connected to air delivery conduit 408 and mask 406. Exhaust gas can be vented through an exhaust pipe 413. In some cases, a flow sensor 404f and / or a pressure sensor 404p may be utilized. For example, mask flow is measured using a pneumotachograph and a similar device such as a differential pressure transducer or a device utilizing a bundle of tubes or ducts to derive a flow signal F (t). Sometimes. Similarly, mask pressure may be measured at a pressure hole using a pressure transducer to derive a pressure signal P _mask (t). The pressure sensor 404f and the flow sensor 404p are only symbolically represented in FIG. 4 because it is understood that other structures and other equipment may be implemented to measure flow and pressure. Because it is done. The flow signal F (t) and the pressure signal _Pmask (t) are _combined with one or more analog-to-digital (A / D) converters / samplers (shown) Z)
May be sent to the controller or the microprocessor 415 via

Alternatively, the flow signal f (t) and / or the pressure signal P _mask (t) may be provided with the current (
By monitoring I) and the speed (S) and / or speed (w) of the motor, it may be estimated or calculated in relation to the blower motor. In some cases, the blower motor speed may be kept nearly constant and the pressure change in the mask controls the opening of a servo valve that may variably bypass / vent or deliver airflow to the mask. May be implemented. Further, other types of patient interfaces may be used in place of the mask, and flow sensors and / or pressure sensors may measure flow and pressure at locations selective with respect to the patient interface.

The controller or processor 415 is configured and adapted to implement the methods or algorithms described in more detail herein, and includes an integrated chip, memory, and / or other control instructions, data, or information. Storage medium. For example, instructions programmed using the control method may be encoded on an integrated chip in the memory of the device, or such instructions may be loaded as software or firmware using a suitable medium. Sometimes.

With such a controller, the device can be used to set the speed of the blower, or
Regulating a pressure delivery formula that is used to operate ventilation with a release valve and that may be based on a sensing method related to the pressure delivery formula as shown in the examples detailed herein. By doing so, it can be used for many different pressure treatments. Alternatively, pressure therapy is performed on the device without including a pressure therapy component, such that the device may be used to diagnose, detect, and / or quantify the presence of a flow restriction. Sometimes.

The methods or algorithms proposed herein for detecting and / or treating a breathing pattern representing a partial occlusion can be implemented using a controller as a fuzzy logic control system. In general, fuzzy logic is a technique that translates ideas about system behavior into computer code that is based on real numbers for control system parameters. However, the problem at hand is basically one of statistical pattern recognition, which may be implemented using other techniques, such as the alternatives discussed in more detail herein. There is. Thus, although the following techniques have been proposed in terms of fuzzy logic control systems, it will be appreciated that there are other techniques to introduce into the control system based on the desired control inputs and typical output functions.

To this end, embodiments of the control system of the present technology can be implemented according to the following general approach using a partial occlusion or flow restriction indicator, such as a fuzzy flow restriction (FFL) indicator. Using such an index is an approach that may improve some of the observations described above with respect to using a conventional flattening index as a system control variable. The flow restriction indicator FFL uses multiple feature pattern recognition to improve the sensitivity and specificity of the algorithm's response to flow restriction (or more precisely, partial upper airway obstruction). In embodiments, a general approach using the flow restriction indicator FFL may include some or all of the following steps.

1. Monitor the patient's respiratory flow.
2. Extract inspiratory and expiratory waveforms from the flow signal.
3. Features are calculated from each component of the waveform or, where appropriate, from the composite waveform. Multiple features may be considered to constitute a pattern.
4. Compute fuzzy input variables from raw variables. This step constitutes the mapping of the input variables to the new space. Data mining of the appropriate population of patients allows for a good estimation of the parameter space.
5. Combine the derived fuzzy variables using fuzzy logic to produce a fuzzy output. This step can be formulated using matrices that allow an intuitive understanding by the layman.
6. De-fuzzy the fuzzy output to generate a crisp value (real number) to be used by the pressure control algorithm.
7. Combine fuzzy outputs from multiple sources.
8. Adjust treatment pressure based on FFL results.

In accordance with this method, a system embodiment maps widely varying input parameters to variables that can be easily interpreted using fuzzy logic principles. For example, a shape index, such as a flattening index, may be mapped to a fuzzy variable, such as high flattening, normal flattening, low flattening, and the like. Each one of these variables depends on the value of the flattening index as determined by the mathematical function associated with the variable and, based on input data associated with the flattening decision, such as respiratory airflow, Take a value between 1 and 1. Similarly, other fuzzy variables are based on appropriate mathematical functions and input data associated with the determined ventilation,
And it is possible to map a ventilation index such as low ventilation. Yet another fuzzy variables also low _T i _{-on-T tot,} ordinary _T i _{-on-T tot,} or respiratory duty cycle indicators, such as _high-T i _{-on-T tot} (e.g., respiratory cycle (The ratio of the time of inspiration (T _i ) to the total time of (T _tot )). These variables may then be further mapped to fuzzy outputs that may be used to generate control variables. The fuzzy output associated with the flow restriction indicator is, for example,
It may be mild, moderate, and severe, and may be based on other fuzzy variables, such as the embodiments described above. For example, a fuzzy logic controller may enforce the conditions represented by the following statement embodiments:
If (a) (low planarization and low ventilation), the flow restriction is severe.
(B) If (high planarization and high T _i -on-T _tot ), the flow restriction is mild.

Zero reverse fuzzified output that can be used as a flow restriction indicator (no flow restriction)
And 1 (severe). With such a system, the input space can be classified in a non-linear sense using easily interpretable rules without any loss of real precision.

In fact, using the accuracy of such a flow restriction index, which represents the degree of blockage in a continuous range from 0 to 1, the index is a response pressure that is suitable for handling the conditions detected using this index. Or it can be implemented proportionally to derive flow control settings.
For example, without having to use a threshold that would have been compared to such an indicator to determine whether a pressure adjustment should be made, the indicator can be set in a respiratory therapy device or in such a device. May be performed more directly in determining treatment settings for For example, the occlusion indicator may serve as a proportional function (eg, indicator × adjustment amount) that may be applied directly to the pressure or flow rate adjustment. The resulting adjustment may then be applied to the treatment settings of the respiratory treatment device. As described with respect to the techniques described in more detail herein, the "FFL" indicator may be implemented to serve as such a proportional treatment setting function.

As further shown in FIG. 5, in such a system, the flow restriction comprises: (a) determining a shape index representing a pattern of flow restriction from the respiratory flow index shown in step 502; (B) determining a ventilation index or breath duty cycle index from the respiratory flow index shown in step 504; (c) a determined shape index shown in step 506; and a determined ventilation index. Or deriving a flow restriction indicator as a function of one or both of the determined respiratory cycle indicators. Using such a method, sophisticated adjustments to the treatment pressure may be made. For example, increases to treat flow restriction include (a) that the shape index indicates the presence of an M-shaped or flattened breath in the respiratory airflow, and (b) that the ventilation index meets normal ventilation. If both are determined to decrease sufficiently to indicate that no pressure change may be based on the condition that a pressure change may occur. Further increases to treat flow restriction include both (a) the shape index representing the presence of M-shaped breath in the respiratory airflow, and (b) an increase in duty cycle index has been detected. If is determined, it may be based on other conditions, such as a pressure change may occur. With such conditions, an unnecessary therapeutic response would appear when the shape index, ventilation index, or respiratory duty cycle index used alone did not accurately predict the level of flow restriction. Given without therapeutic change.

Thus, in one embodiment, the system is characterized by the use of multiple (eg, four) input features to derive the flow restriction indicator FFL. The input features are (1
) Single breath flattening (SBF) shape index, (2) M shape index (MS), (3) air permeability (V
R) and / or (4) the ratio of inspiratory time to total breath time (T _i -on
-T _tot ) may be included.

An embodiment of the shape index implemented as a single breath flattening index SBF is preferably 5
US Pat. No. 5,704,345, except that it is performed without using a respiratory moving average
May be calculated in a manner similar to the approach disclosed in US Pat. Details of performing the single breath flattening shape index determination are disclosed in Section A herein.

An exemplary implementation of determining the shape index of an M-shape is described in Section B herein. The illustrated calculation of the M-shape feature varies from zero (ie, the intake flow waveform is not M-shaped) to 1 (ie, the intake flow waveform is deterministically M-shaped). I can do it. The M shape index may in some cases be augmented by another shape index, such as a simple “chair shape” index. An exemplary determination of a chair shape index (eg, fuzzy “chairness”) is also detailed in Section B.

An embodiment of determining a suitable ventilation index may be based on a tidal volume calculation. For example, the ventilation index may be a ratio that includes the current tidal volume and the previous tidal volume such that the indicator may represent a change in ventilation or tidal volume. In one suitable embodiment, the indicator may be a permeability VR that can be calculated as follows.

1. The absolute value of the respiratory flow is obtained, and the absolute value is filtered using a simple low-pass filter (time constant = 3 minutes).
2. Divide the filter output by 2 to create a "3-minute vent."
3. Calculate the tidal volume (basically the average of inspiratory volume and expiratory volume).
4. Divide the tidal volume by the length of the breath, i.e. yield the average breath flow rate, then divide this value by 3 minutes ventilation to get an indication of how much this breath matches the previous ventilation. Cause.

Normally, VR vibrates in a narrow range of about 1. In the case of severe upper airway obstruction, the VR may decrease to a value much less than one, and in the case of complete obstruction, it may decrease to zero. Breathing (wake-up breathing) has a VR much greater than 1 (eg, 1.5 to 2). Further details of the calculation of the specific permeability VR are contained in section G herein.

As described above, the system may optionally utilize a respiratory duty cycle indicator during the derivation of the flow restriction indicator FFL. Since the indicator may be based on the ratio of the current duty cycle to the previous one, the indicator may represent a change in the respiratory duty cycle. Exemplary details of calculating the respiratory duty cycle ratio TTR are further described in Section G herein. For example, a suitable breath duty cycle index is T _i -o
It may be based on the calculation of the _{nT tot} variable. The selected variable has a (non-plausible) limit of zero (no inspiration) and 1 (no inspiration). Since the base T _i -on-T _tot can vary from patient to patient, the ratio of T _i -on-T _{tot to} the immediately preceding mean, ie, the ratio of T _i -on-T _tot herein. Is calculated. T _i average value of the immediately preceding _{-on-T tot,} each _T i _{-on-T tot} calculated
It can be calculated by putting the values into a simple low pass filter. T _i -o
When n-T _tot increases compared to the previous history, TTR exceeds one. T _i -on-
T _tot has a value around 0.4, TTR is usually 1, and we will “compensate”, ie, increase our tidal volume, even though the patient has an obstructive upper airway. Is greater than one.

As described above, these input features are mapped to several variables by mathematical functions. Thus, the shape index SBF is, for example, six fuzzy variables, ie, B_L
OW_FLATTENING, EXTRA_LOW_FLATTENING, VERY_
LOW_FLATTENING, LOW_FLATTENING, NORMAL_FLA
It may be mapped to TTENING and HIGH_FLATTENING. These fuzzy memberships are illustrated by sample functions, graphed in FIG. 6, and presented in left-to-right order with each variable listed above. As an example of applying the membership function, if the shape index SBF is calculated and results in a value of 0.05, the variable B_LO when applied to each of the membership functions
W_FLATTENING results in a value of 1.0 and a value of zero for all other fuzzy variables in the graph of FIG. Other suitable functions and variables may be selected as needed.

Similarly, an example of membership functions and variables based on the calculation of the shape index of the M shape is shown in FIG. This graph shows LOW_M-SHAPE, NORMAL_M-SHA
PE, HIGH_M-SHAPE, VERY_HIGH_M-SHAPE, and B_
The appropriate function for HIGH_M-SHAPE is shown. Examples of membership functions and variables based on the calculation of the permeability index are shown in FIG. This graph is VERY
_LOW_VENTILATION, LOW_VENTILATION, NORMAL_
VENTILATION, HIGH_VENTILATION, and VERY_HI
The appropriate function for GH_VENTILATION is shown. Finally, an example of membership functions and variables based on the calculation of the breath duty cycle rate is shown in FIG. This _{graph, T i -on-T tot _LOW} , T i -on-T tot _NORMAL, T i
_{-On-T tot _HIGH, T i} -on-T tot _VERY_HIGH, and, T
shows a suitable function for _{i -on-T} tot _EXTRA_HIGH.

Based on some or all of these mapped variables, the system may then apply or evaluate rules based on the combination of variables during the derivation of the flow restriction index FFL. For example, fuzzy rules are shown in the following table based on pre-specified fuzzy variables. These matrices further express in natural language ideas about what can constitute a partial occlusion.

Table A
Flattening index and ventilation index

Table A provides a summary of a set of fuzzy rules that can be applied using the fuzzy variables associated with the single breath flattening shape index and the permeability index. For example, the "Ultra High" column for the fuzzy variable for the flattened shape index and the "
The rules in bold ("light to medium", that is, light to medium) in the "low" line express the following fuzzy rules.

  If (very high flatness and low airflow), the FFL is "mild to moderate".

This and other rules shown in the table can be evaluated, and "AND" is a fuzzy AND. The common output result for each output response (eg, "mild to moderate") is then fuzzy ORed together. For example, two rules having a "medium" output (e.g., a measure of moderate flow restriction) are fuzzy-or'd to produce a final "medium" output. This is equivalent to the following equation.

Medium = (VERY_HIGH_FLATTENING AND VERY_LOW_
VENTILATION) OR (EXTRA_HIGH_FLATTENING A
ND LOW_VENTILATION)

Based on the application of these rules during the derivation of the flow restriction index, there are at least three things: 1) the FFL index increases as the flattening severity increases (eg, as the shape index decreases). 2) If flatness appears and ventilation is reduced, FFL
The indicator is even more severe, and 3) if the ventilation is high (eg, the appearance of a recuperated (large) breath, etc.), the response to the flattened shape index is adjusted during rule enforcement. It is obvious.

Flow restriction may not be accompanied (at least initially) by a reduction in tidal volume (eg, permeability). In these cases, the patient may have increased the duty cycle,
Maintain your tidal volume by extending inspiration time as a percentage of total inspiration-expiration time. This tendency can be measured or determined in this embodiment using a duty cycle indicator (eg, TTR) that becomes greater than one with increasing inspiratory time.

The system converts the flattened shape index information to the breath load cycle using the rules expressed by Table B below in a manner similar to the manner in which the flattened shape index information is combined with a ventilation index (eg, VR). Further coupled with an indicator (eg, TTR).

Table B
Flattening and breath cycle index

Similar to Table A, Table B represents fuzzy rules of flow obstruction based on the index of the breathing duty cycle and the flattening shape index and the corresponding natural language description. For example, the bold (“medium to heavy”, ie, medium to heavy) rule in the “extremely high” column and the “ultrahigh” row is the following fuzzy rule.

_{(EXTRA_HIGH_FLATTENING AND T i -on-T} tot _V
ERY_HIGH), the FFL is moderate to severe.

This rule states that if the inspiratory breathing pattern waveform tends to be severely flat and if inspiration is extended moderately by consideration of the duty cycle index, then the flow restriction index FFL is moderate to severe. Express matters. As described with respect to Table A, the results of the rules for the common output functions of this table can be combined using a fuzzy OR operation.

In a manner similar to the manner in which the flow restriction index is derived using the flattened shape index information and one or both of the ventilation index and the breath load cycle index, the flow restriction may be M-shaped and / or enhanced M-shaped. It may also be derived using both the inhalation detection index and one or both of the ventilation index or the breath duty cycle index. This derivation aspect is shown in Tables C and D.

Table C
M-shaping or enhanced M-shaping and ventilation index

Table C expresses the same rules as the chart above, except that it pertains to fuzzy variables based on the M-shaped shape index and / or chair shape index. As is evident from the nature of the choice of output function in the table, one rule is to prevent a response to the detected M-shaped breathing pattern if the patient has normal or higher average ventilation. It is intended to be. Thus, flow restriction indicators may occur during certain types of awake and REM sleep, but the system does not respond to "behavioral" M-shaped breathing patterns that do not represent the current flow restriction Is derived as follows. For example, the rule represented by the entry in the top row of the last column of Table C indicates that when a good ventilation index (eg, “Ultra High”) is present, the output function is associated with a bad “Ultra High” M shape index. Specifies that the index is "zero".

In addition, most M-shaped breaths exhibit moderate to severe obstructiveness and show reduced ventilation. Thus, the other rules in Table C allow detection of flow restriction to system pressure changes to address situations such as this detected M-shaped breath and low ventilation indicators such as reduced ventilation.

As described with respect to Table A, the results of the rules for the common output functions of this table can be combined using a fuzzy OR operation.

  Table D presents the rules for the breath duty cycle index.

Table D
M shaping or enhanced M shaping and duty cycle index

Table D represents the specific fuzzy rules and corresponding natural language descriptions for the breath duty cycle index and the flow occlusion or flow restriction index derived from the M-shaping and / or augmented M-shaping decisions. In this embodiment, the matrix (rather than permeability) an M shape information T _i
“Coupled” with the −on-T _tot ratio. The matrix may be considered to be related to the presence of M-breaths, ie, the presence of M-shaped patterns, with a tendency for the breath duty cycle to increase. This combination identifies a flow restriction during breathing when the patient's ventilation is approximately "normal" and the patient is "compensating" by taking longer inspirations.

As described with reference to Table A, the results of the rules for the common output functions of this table can be combined using a fuzzy OR operation.

The results of the rules applied based on Tables A, B, C and D may then be applied to the output function. FIG. 10 shows a suitable fuzzy output membership function that may be applied with this result. Output membership functions include negative, zero, mild, mild to moderate, moderate, moderate to severe, and severe.

As an example, consider Table D. When a table in individual fuzzy rules are calculated _{(e.g., (EXTRA_HIGH_MSHAPE AND T i -on-} T tot _VERY_
If HIGH), the FFL is moderate to severe) and the outputs are collected and fuzzy ORed together. For example, there are three medium to severe outputs in Table D that need to be fuzzy ORed. Once this has been done for all outputs, it is fed into the inverse fuzzification function of FIG. 10 and the centroid method may be used to provide the crisp output as in FIG. This results in a real number between 0 and 1.25. If individual outputs from all tables are available, the highest output may be used for the value of FFL.

In determining a single "crisp" indicator in the defuzzification step just mentioned, a centroid or other such defuzzification operation may be performed. In the centroid method, the crisp value of the output variable is determined by finding the value associated with the centroid of the value of the output function. An embodiment of such a method is shown in FIG. FIG. 11 shows mild =
10 illustrates the application of the centroid method calculated for fuzzy inputs when 0.1, mild to medium = 0.5, medium = 0.8. The calculation result is 0.61064. Those skilled in the art will understand other ways to calculate the crisp index in light of the present description.

In one embodiment of the system, the flow restriction indicator may be derived to avoid treating other potential detection conditions. For example, awake breaths can be flat or M-shaped in shape, thus representing a flow restriction based on a decision using the shape index, but since breaths actually extend inspiratory time, It may not represent the actual flow restriction in the patient. Thus, a flow restriction index may be derived to not treat such breaths as restricted flow. One way of performing such a derivation is to modify the fuzzy output of the fuzzy rules from Tables A, B, C and D to adjust the combination using the breath duty cycle index TTR with the function shown in FIG. It is to be. In essence, the fuzzy outputs resulting from rules based on a combination of duty cycle indicators TTR (eg, the rules in Tables B and D) are multiplied by the output of the function of FIG. 12 before defuzzification. Thus, if the permeability is less than or equal to about 1, the result of each particular rule cannot be changed. With an increase in the permeability index VR, which indicates an increased likelihood of awakening, the output of the rules in Tables B and D is:
In accordance with the function example of FIG. 12 and by a multiplication operation, the weight is gradually reduced until the air permeability index VR approaches 1.5. At that point, the result of the multiplication operation causes the output of the affected rule to be zero.

The above building blocks (for example, the shape index, fuzzy membership functions and variables,
Flow restriction index F in the flow restriction detection system on the basis of fuzzy rules).
The calculations of the FL embodiment can be further summarized using the following exemplary steps.

1. Breath may be framed from the patient flow signal in a typical manner to identify and extract data representing inspiratory and expiratory waveforms from the flow signal.
2. The inspiratory waveform may optionally be trimmed using a particular M-shape,
Any forward pose may be removed. Such a method is described in Section F herein.
3. The M shape determination and the fuzzy chairness determination may be made based on the typical calculations shown in Section B.
4. The augmented M-shape may, in some cases, be a shape index of the M-shape and a chair shape index (eg,
Fuzzy chairness).
5. The filtered single breath flattening (SBF) shape index may be determined based on the typical calculations shown in Section A.
6. Permeability (VR) and T _i -on-T _tot ratio (TTR) may be determined based on the typical calculations shown in Section G.
7. The SBF, VR, TTR and augmented M-shape may be fuzzified into variables like the typical membership function described above.
8. The fuzzy rules then follow Tables A, B, C and / or D as described above.
May be applied according to the matrix
9. The fuzzy rules from Tables B and D that use the breath duty cycle index TTR may be modified in some cases by the "restricted awake breath" function described above with respect to FIG.
10. Each fuzzy rule matrix is collected and de-fuzzified, as described separately above.
11. This results in a fuzzy output that is all fuzzy-ORed to produce the flow restriction index FFL as described below.
FFL = fuzzy OR (SBF-VR, SBF-TTR, M-shape-VR, M
-Shape-TTR)
Where SBF-VR is the inverse fuzzification result based on the rules in Table A,
SBF-TTR is the result of reverse fuzzification based on the rules in Table B,
M-shape-VR is an inverse fuzzification result based on the rules in Table C,
M-shape-TTR is the result of reverse fuzzification based on the rules in Table D.
12. Next, a fuzzy persistent flattening index (FPF) may be calculated and possibly fuzzy ORed with the flow restriction index FFL according to the following equation: Index FP
A typical calculation of F is described in Section C herein. The adjustment of the FFL index can be performed by the following equation.
FFL = Fuzzy OR (FFL, FPF)
13. The FFL may then utilize, for example, the calculations described in Section D herein to determine the degree to which breath framing (eg, detecting inspiratory or expiratory waveforms) may be incorrect. Will be modified. The resulting incorrect breath framing factor may be multiplied by FFL using the following equation:
13. FFL = FFL × bad breath framing coefficient This final FFL value may in some cases be used in a ring buffer of a length, such as 3, for example, and the final FFL value used by the pressure setting algorithm will be May be based on a moving average of the FFL values of A suitable expression for this operation may be:

FIG. 13 shows a conventional flattening index and a flow restriction index F derived by the steps of the above summary.
4 provides a graph comparison diagram with FL. The index and index are based on the same flow signal shown in the lower trace. The flow signal in the lower trace includes a series of obstructed breaths ending in arousal followed by some fairly normal resting breath, and more obstructions. Breath is initially "flat" and gradually becomes M-shaped. In the upper trace, the conventional flattening index falls first, then rises, and finally falls sharply during a recovery breath. The middle trace flow restriction index FFL rises steadily due to both breath flattening and M-shaped, and then falls to zero when breathing in the flow signal becomes more normal.

The flow restriction indicator may be implemented under the control of a pressure therapy device. In one such embodiment, the indicator may be implemented using one or more of the following:

1. The flow rate is measured with a flow generator (FG).
2. Measure pressure with FG.
3. Calculate the pressure drop between the FG and the mask using the flow rate of step 1.
4. The pressure at the mask is calculated as the pressure obtained by subtracting the pressure drop calculated in step 3 from the pressure at the FG.
5. The pressure at the mask is used to calculate the flow through the vent in the mask (sometimes called intentional leak).
6. Subtract the vent flow from the FG measured flow to produce the sum of the patient flow (respiratory flow) and the unintentional (mask or mouth) leak.
7. Filter the signal from step 6 to extract the DC (unintentional leakage) component.
8. The DC component calculated in the previous step is subtracted from the flow calculated in step 6 to produce patient flow (respiratory flow).
9. For example, the method presented in Section H herein lightly filters the patient flow to remove unwanted high frequencies.
10. Frame the breath using patient flow in the usual way.
11. For example, any of the forward poses may be clipped from the front of the inspiration by the method set forth in Section F herein.
12. Once the complete breath (inspiration + expiration) is framed, calculate the following features.
a. The length of any apnea preceding the breath.
b. Inhalation snoring index (eg, the average of the snoring signal during the current inspiration being analyzed).
c. A filtered single breath flattening index (SBF) value for the currently analyzed breath (eg, according to the method set forth in Section A herein).
d. For example, the value of the M-shape index for the current inspiration and the value of the fuzzy chairness for the current inspiration, according to the method set forth in Section B herein. The M-shape values and the fuzzy chairness values are fuzzy-or'd to produce enhanced M-shape features.
e. Permeability (VR) values and T _i -on-T _tot ratios (TTR) for breaths currently being analyzed, according to the methods described in Section G herein.
f. FFL as described in detail above using SBF, VR, TTR and enhancement M
Is calculated.
g. Calculate the valve-like leak rate using the currently analyzed exhalation.
h. Measure unintentional leaks at the end of the exhalation currently being analyzed.
i. Check the current setting (ie, EPAP) for the mask pressure.
j. The exhaled breath currently being analyzed is used to calculate a normalized expiratory peak position NEPL, for example, by the method described in Section K herein.
k. Check the value of the last peak disturbance.
13. The weight reduction coefficient is calculated as in the following pseudo code. Section J describes examples of individual weight reduction functions that may be used.
a. deweight = 1.0
b. deweight * = FFL_function_of_Leak (leak
)
c. valve-like leak * = valve_like_leak_fu
nction_of_leak (leak)
d. dweight * = FFL_function_of_valve_like
_Leak (valve-like leak)
e. deweight * = FFL_function_of_pressure
(EPAP)
f. deweight * = FFL_function_of_NEPL (NEP
L)
g. deweight * = FFL_function_of_Jaming (
jamming)
14． Current threshold value required for pressure increase (current_crit_FFL)
Is then calculated using the following equation:
Current_crit_FFL = 1.0−deweight * (1.
0-crit_FFL)
15. The standard value of unmodified crit_FFL is 0.05. Therefore, if deweight = 1.0 after all weight reductions have been applied, current_cri
t_FFL = 0.05. Alternatively, if deweight = 0.0, cu
rent_crit_FFL = 1.0.
16. The value of the pressure rise associated with the FFL (FFL “module”) can now be calculated and “predetermined”.
a. dp = 1.0 * (FFL-current_crit_FFL)
b. if (dp> 0.0) then
｛
max_dp = (EPAP-range-max-EPAP)
if (dp> max_dp) then dp = max_dp
if (dp> 0.0) then FFL_prescriptio
n + = dp
｝
else
FFL_prescription = decay (FFL_prescr
option, Ti + Te, 20)
The last step simply decays the pressure exponentially over the time of breath, preferably with a time constant of about 20 minutes.
17． This time, the pressure rise due to snoring is calculated, may be "predetermined", and may utilize a separate weight reduction function as shown in Section J herein.
a. Current_crit_snore = Snore_function_o
f_Pressure (EPAP)
b. Insp_snore * = Snore_function_of_VR (VR
)
c. if (Insp_snore> Current_crit_snore)
then
｛
dp = 1.5 * (Insp_snore−Current_c
rit_snore)
max_dp1 = ttot * 0.2
if (dp> max_dp1) then dp = max_dp
1
max_dp2 = (ETAP_range_max−ETAP)
if (dp> max_dp2) then dp = max_dp
2
if (dp> 0.0) then-snore_prescript
ion + = dp
｝
else snow_prescription = decay (sno
re_prescription, Ti + Te, 20)
18. This time, the pressure rise due to apnea may be calculated.
if ((apnoea_airway_closed && (apnoea_d
uration> 10)) then
｛
Head_room = (Max_Pressure_Apnoea−E
PAP)
if (head-room> 0.0) then
｛
new_epap = Max_Pressure_Apneaea-h
ead_room * exp (-ExpRiseTimeApnoea * apno
ea_duration)
if (new_eap> EPAP_range_max) the
n
new_epap = EPAP_range_max
apnea_prescription + = (new-eapp-E
PAP)
｝
｝
else appnoea_prescription = decay (ap
noea_prescription, Ti + Te, 20)
19. New EPAP settings may be calculated thereafter.
a. EPAP = EPAP_range_min + apnea_prescr
option + snore_prescription + FFL_prescr
option
b. New EPAP settings are subsequently sometimes be achieved by increasing the FG pressure so that the mask pressure approaches the new EPAP value at the maximum slew rate per 1.0cmH ₂ O. Similarly, the treatment pressure is preferably increased only while the patient is inhaling.

In yet another embodiment of the present technology, the flow restriction indicator FFL may be implemented, as the case may be, as part of a respiratory effort related arousal (RERA) detector. AASM in 1999
The Task Force defined RERA as follows:
"A series of breaths characterized by increasing respiratory effort leading to arousal from sleep but not meeting the criteria for apnea or hypopnea. These events require that both of the following criteria be met:
1. A pattern of progressively more negative esophageal pressure, ending with a sudden pressure change to a smaller negative level and arousal.
2. The event lasts more than 10 seconds. "

In 2000, it was held at New York University School of Pharmacy and was published in Sleep, vol. 23, no.
6, p / 763-771, "Non-Invasive Detective."
on of Respiratory Effort-Related Arousal
s (RERAs) by a Nasal Cannula / Pressure Tran
sducer System "is a nasal cannula / pressure transducer system with an R
Proven and reliable for the detection of ERA.

By utilizing the techniques described herein, the RERA detector may be based on the actual flow signal obtained from the flow generator. For example, a flow restriction indicator according to any of the methods described above may be determined based on the flow signal. The arousal index may therefore be derived as a further function of the flow restriction index function and the ventilation increase index.

Thus, in one embodiment, the RERA detector may be based on the following method.
There is a previous flow restriction (eg, FFL is greater than 0 (FFL>0));
If the ventilation step change follows (eg, breath), RERA is detected.

Preferably, the indicator is implemented as a continuous variable so that a threshold adjustment based on experimental data can be made. This may be an alternative to a Boolean threshold for each input parameter that causes information loss). The following algorithm can be used.

1. Keep track of the three most recent FFL values in the rolling buffer.
2. Track the three most recent ventilation rates (VR is the ratio of average tidal volume to current 3 minute ventilation).
3. Sum the last three FFL values and limit the result to the range [0.0: 1.0].
4.2 pieces of computing the latest VR difference _(i.e., VR _{n -VR n-1} and _VR n -VR
_n-2 ).
5. Obtain the maximum value of step 4 and at the same time limit this maximum value to a value greater than or equal to zero.
6. Multiply the result of step 3 by the result of step 5 to produce the RERA result.
7. The result of step 6 may be normalized by taking the square root of this result.
8. If the result of step 7 is above the threshold, a threshold may be set such that RERA is recorded.
9. There are breath (s) with an RERA score below the threshold between breaths recorded as RERA.
10. In the extreme case of UARS, the buffer size is chosen as 3 since there are only three detectable breaths between awakenings.

As shown in FIG. 14, the lower signal represents a continuous RERA detector index according to the method. The lower trace shows the detection of respiratory effort-related arousals based on the awakeness of the upper trace representing the flow signal.

The RERA detector may be further implemented as part of the respiratory distress index. The RERA detector may be further used in the diagnostic mode to calculate the RDI (Respiratory Disturbance Index) according to the following equation:
RDI = RERAs + apnea + hypopnea per hour

Alternatively, RERAs may be reported by the treatment device as an indication of the effectiveness of the treatment. Finally, the RERA index described herein may be used as an input to a treatment algorithm. For example, the RERA index may be determined over some time frame, such as per hour, and the result may be "outside" to set the threshold (or gain) of the flow restriction index to increase pressure. Loop controller ". An example of an outer loop controller is assigned to ResMed, Inc. and is described in WO 2005/051470 (PCT), the disclosure of which is incorporated herein by reference.
/ AU2004 / 001652).

The ratio of peak expiratory flow to average expiratory flow may be used as another type of index for sleep-wake-wake detection. The ratio of peak expiratory flow to average expiratory flow is: sleep-wake /
It may be a part of awakening detection. FIG. 15 includes a histogram comparing the wake-up standard to the two sleep data sets. Awakening tends to "flatten", i.e., the peaks are closer to the mean in value. By comparing a threshold, such as 0.5, with the ratio, it may be considered that RERA is detected.

Section A-Single Breath Flattening One embodiment of the single breath flattening index may be calculated by the following method as follows.

1. Frame your breath.
2. Extract the inspired part of the breath.
3. Cut out any backward or forward poses as needed.
4. Interpolate the resulting inspiration on a standard grid of N points (typically N = 65).
5. Divide the value of point y by a factor such that the breath area is normalized to 1 by the unity base length.
6. Compute the rms value of the difference of the point's middle half from one.

The flow signal used to calculate flattening may be referred to (among others) as "patient flow". This flow signal may be the result of filtering the raw flow signal with a filter such as a "10 Hz" filter, but the filter type may vary. The filter preferably reduces unwanted signal content such as that caused by turbulence and snoring. Filters may slightly attenuate cardiac content. Filters are a trade-off between accurate detection of zero crossings on the one hand and rejection of unwanted signals on the other hand. For example, if the filter is made more aggressive to reject all cardiac oscillations, breath detection may be incorrect. However, upon completion of breath detection, further filtering may be performed as needed.

In some cases, the filtered single breath flattening may be determined as follows.

1. The raw stream is filtered in a conventional manner to provide a patient stream.
2. Patient flow is conventionally used to frame the breath.
3. Patient flow is continuously provided to an FIR filter that further attenuates unwanted signal components.
4. Since FIR filters have a constant phase delay with respect to frequency, it is possible to choose the points needed for the flattening calculation simply by considering the filter delay.
5. The flattening is calculated in the usual way.

FIG. 16 shows two superimposed flow signals recording a sequence of obstructive breaths where the normal flattening calculation (in this case) may have been adversely affected by low frequency snoring.
One signal is filtered (and delay compensated) such that snoring is clearly not noticeable. The filtered signal gives a reasonably low value of the flattening calculated based on the signal breath compared to a calculation based on another signal where snoring is clearly noticed.

In the illustrated signal of FIG. 16, the last breath was calculated as 0.21 (based on the unfiltered signal) and 0.13 (based on the filtered signal) in the conventional signal. It has a flattening value. The latter value indicates the required pressure increase, while the former does not. As an alternative to such filtering, conventional five breath point averaging may also achieve this filtered result at the expense of significant response delay.

A typical (and not computationally intensive) filter used is a boxcar FIR filter. The length 12 boxcar filter has the response shown in FIG. 17 at a sampling frequency of 50 Hz. The filter's impulse response is insignificant if we are interested in the middle-half signal of each inspiration in the first place.

Section B-M Shape Index and Augmented M Shape A suitable method of determining an occlusion index, such as the M shape index, may be achieved using the following algorithm. This index detects the presence of an "M" shaped breath pattern. Such an index may be generally considered to be an indicator of a "u" shaped breath pattern. The occlusion indicator may be augmented or modified by additional occlusion indicators. For example, as described in more detail herein, the occlusive index is a first occlusive index, such as an "M" shape index, and a second occlusive index, such as an "h" or chair shape index. May be derived as a function of the index.

For example, in some embodiments, each inspiration may be interpolated on a grid of N points, such as N = 65. In the present embodiment, two basic functions are calculated as follows.
t = i / (N-1) where i varies from 0 to N-1.
B1 = sin (πt)
B2 = sin (3πt)
These primitives may then be stored for use with all subsequent calculations of the M-shaping index.

Each inspiration is then extracted and interpolated on a grid of N points. The two coefficients are then calculated as follows:
F1 = sum (B1 · fs)
F2 = sum (B2 · fs)
Where fs represents the interpolated intake point and • represents the dot product operator.

The final shape values are obtained by normalization as follows:

This shape factor is then constrained to vary between zero (pure sine wave) and 1 (super-M shaping). FIG. 18 plots the appropriate functions that serve as the base vectors B1 and B2.

A typical M-shaped breath flow signal is plotted in the graph of FIG. Based on the above method, the calculation of the plotted breath is as follows.
F1 = 4.6082
F2 = 2.6538
Shape index = 0.50
A typical non-flow-limiting breath may have an M-shape index of only about 0.2.

As described above, M-shaped breaths may be enhanced by simple chair shaping. This is illustrated in FIG. 20, which plots a flow signal with a flow restriction that produces an enhanced shape. The flow-limited intake shown in FIG. 20 (from t = 0 to approximately t = 2 seconds) has a typical “chair” shape. Such inspiration is characterized by two features: 1) a high ratio of peak inspiratory flow to average inspiratory flow, and 2) 0 (chair with left back as shown in FIG. 20) or 1 (illustrated). The right side may be characterized by a normalized peak position near one of the back chairs. In one embodiment, these indices may be calculated as follows.
Normalized intake peak position (Norm PeakLoc):

Where:
t ₀ is the time of the start of intake,
t _end is the _end of inspiration,
_tpeak is the time of the peak inspiratory flow rate.
Ratio of peak intake flow to average intake flow (RPMIF):
RPMIF _{= Q peak} / _Q '
Where:
Q _peak is the maximum flow rate during inspiration,
Q 'is the average flow rate during inspiration.

During sleep, normal inspiration has a normalized peak position of approximately 0.5 and a peak to average inspiratory flow ratio of 1.35. These fuzzified features are used to measure “chairness” based on the graphs of FIGS. 21 and 22. The result of the above equation is applied to the mathematical function of each graph.

Once these fuzzy variables have been calculated, the final fuzzy chairness index is calculated as fuzzy, and the results are as follows:
Fuzzy chairness = Fuzzy logical AND (Fuzzy peak vs. average, Fuzzy peak position)

Based on this determination, the intake of FIG. 20 has a fuzzy chairness of 0.82. The scaling of the fuzzy chairness feature is such that the fuzzy chairness feature is directly fuzzy-orable with the M-shaped feature due to the combined shape index. A fuzzy chairness higher than approximately 0.3 means a flow restriction. Fuzzy chairness, in Japanese,
It is described as "If the ratio of inspiratory peak to inspiratory average is high and the peak inspiratory flow position is near the beginning or end of inspiratory, the inspiratory is chair shaped." FIG.
A typical flow sequence is shown where the intake shape is initially flat such that the F-index is less than approximately 0.2. However, the shape changes the chair shape to the extent that the SBF index (SB flattening) is higher than approximately 0.2, but the fuzzy chairness or chair shape index is higher than approximately 0.3.

Section C-Continuous Flattening The embodiment of the flow restriction indicator FFL outlined above is designed to respond to flow restriction with both high sensitivity and high specificity in a timely manner. In some cases, arousal cannot be prevented due to an increase in upper airway resistance. Consider the trace shown in FIG. 24 of a patient experiencing upper airway resistance, where the increased respiratory effort ends at waking from sleep at t = 420 (upper panel breath). The lower panel of FIG. 24 shows the single breath flattening index.

FIG. 25 shows why the FFL flow restriction index did not result in a pressure increase in this series. Patients maintain their tidal volume (eg, about 1 VR) without extending their inspiratory time as a function of total breath time (eg, about 1 TTR). Thus, despite the single breath flattening index achieving a low value, the FFL flow restriction index does not reach the level needed to properly increase the pressure.

To address this situation, the system may introduce another occlusion indicator in some cases. For example, a first occlusion index may be derived by filtering a second occlusion index, such as a flattening index. This can give the device certain historical values related to past occlusions. In some cases, the historical value is determined by normal breathing (such as when the historical or filtered occlusion index is an indicator of a continuous or persistent past occlusion).
If (for example, non-blocking) is realized, it may be reset.

For example, "fuzzy persistent flattening" may be implemented. As the name implies, this fuzzy persistent flattening index essentially responds to a consistently low value of the flattening index. The indicator is further implemented to respond slowly to the FFL flow limit indicator without interfering with the flow limit indicator FFL and without overtreating the patient. Therefore, the system preferably filters the single breath flattening index in the following manner.

A simple first order autoregressive digital filter is
_{Y n = y n-1 +} G (x n -y n-1)
Where G is the gain of the filter. A time constant twice as fast as the length of five breaths may be defined. So, for example, if initially your breath is considered to be 4 seconds long,
Time constant = τ = 5 × 4/2 = 10
It is. To take into account that breaths may not be 4 seconds in length, the breath detection algorithm determines the current respiration rate (RR, per minute) that can be determined as the average of the last 5 breaths detected. (Respiration rate). This means
τ = 5 × 60 / 2RR
It can be implemented as an adaptive time constant given as Therefore, RR is
If there are 15 breaths per minute that could be a typical RR, the time constant is calculated to be 10 as before.

The proper gain of the filter is simply
G = 1 / τ
Given by

The value to be filtered is the (filtered) single breath flattening index. In order to prevent spurious values from adversely affecting the filtering process, the system may, in some cases, perform a "peak limit" on the input values. Thus, the value may be determined as a peak-limited flattening (HLF) index as follows:
(SBF index> 0.3),
HLF = 0.3
Otherwise, HLF = SBF index.

The output of the filter is a persistent flattening index (PF), which can be initialized to any high value, such as PF = 1, so as not to affect the treatment initially.

Another feature of the filter is that a gain different from the input for values greater than the current value of PF is used for inputs less than the current value of PF, as follows.
If HLF <PF,
PF = PF + G (HLF-PF)
Otherwise,
PF = PF + 3 * G (HLF-PF)
This asymmetry causes the filter to drop slowly towards any persistently low input,
This means resetting relatively quickly. This helps to improve the noise threshold so that there is a response only to low flattening values consistently during time frames shorter than the flow restriction indicator FFL.

Next, the value of PF is prevented from deviating from a reasonable value as described below.
(PF> 0.2) Then
PF = 0.2
Otherwise, PF = PF

Finally, the PF may be mapped to a fuzzy variable, fuzzy persistent flattening (FPF), for example, by using the following equation:
FPF = 0.0 if (PF> 0.19), otherwise (0.19-PF) / (0
. 19-0.05))
FIG. 26 shows the response of the FPF to the same series as the breath shown in FIG.

Section D-Bad Breath Framing In some cases, the breath detection algorithm may frame the breath incorrectly, or a patient's cough or swallow may cause breathing that provides little information about the current state of the airway. Become. Optional heuristics may be used to reduce breath weights that may not be relevant based on inspiration time. Thus, the derived occlusion index may be adjusted if a breath pattern representing an incorrectly framed breath is detected. For example, FIG. 27 shows a function for a weight reduction factor that can be multiplied by a flow restriction indicator (eg, FFL). Unrealistically long (T _i > 2.5 seconds) or short (T _i
The weight of the breath determined to be <0.7 seconds) is gradually reduced.

Section E-Snoring Entropy Traditionally, snoring has been measured using a calibrated inspiratory snoring index, which can be considered an indicator of obstructiveness. FIG. 28 shows a typical sequence of snores captured during a patient's sleep. The upper panel represents respiratory flow rates, the middle panel represents instantaneous snoring (eg, an index of sound power in the frequency range of interest), and the lower panel represents three snoring indices. Calibrated intake snoring index (named "snoring index" in FIG. 28)
Is calculated as follows:
Inspiration is framed using the respiratory flow signal.
The instantaneous snore signal is adjusted to account for background noise due to current conditions (eg, set pressure, turbine speed, etc.).
The snoring index is calculated as the average of the instantaneous snoring signal during each inspiration (ie, approximately the average acoustic power per inspiration).

This technique is reliable when used to treat snoring by increasing the mask pressure in proportion to the measured snoring index. However, this technique requires that calibration constants be measured for each flow generator and mask combination and stored in the device's non-volatile memory. Such calibration is time consuming and therefore costly and can be affected by aging.

Matthew Alder and others are described in PCT Application No. PCT / AU2007.
In / 000002, an alternative, named Delta Snoring in FIG. 28, is calculated as the difference between the average of the instantaneous snoring signal during inspiration and the average of the instantaneous snoring signal during expiration. Teach that. This delta snoring is essentially a self-calibrating indicator that assumes that the background noise source changes little between inspiration and expiration. But,
This assumption is not always true.

In another alternative, an inspiratory snoring entropy method is implemented. In one embodiment, the measure of obstruction is determined by filtering the measure of respiratory flow in the frequency range associated with snoring (eg, 30 to 300 Hz). The magnitude of the power, or energy, of the filtered signal in this frequency range is then timed to evaluate whether the power signal represents some kind of inspiratory shape or is simply random noise. Checked as a function. For example, a Shannon entropy function may be used.
If the function indicates that the energy signal is nothing more than random information or noise, the snoring index may be adjusted or reduced in weight since true snoring may not have occurred.

Such a method uses the information contained in the instantaneous snoring signal during inspiration. Such indicators do not require calibration. For example, the occlusion index may be calculated as follows.
Assemble a vector S of length n containing the instantaneous snoring value for the current intake.
If n <2, or n> any large value, abort and return zero.
Subtract the lower limit of the snoring vector and add 1.
S = S-min (S) +1
Calculate the area A.

If A is less than any very small value, abort and return zero.
Normalize the snoring vector.
S = S / A
Calculate the Shannon entropy (se) of the normalized snore vector.

In some cases, the result is scaled as follows to approximate the inspiratory snoring index.
Inhalation snoring entropy = 10.0 * se

Both delta snore and snore entropy are plotted in the lower panel of FIG. As shown, the “snoring entropy” trace shows a low average snoring value (t =
100) tend to catch sharp snores. A sensible combination of both delta snoring and snoring entropy has been found to classify snoring correctly in almost all cases. The graph of FIG. 29 shows one such arrangement. The classifier, represented by the plotted lines, states that "if delta snoring is positive, it is addressed, and if delta snoring is negative, it is progressively higher with negative delta snoring to be addressed. Requires snoring entropy. "

Section F-Advanced Forward Pause Trim Certain pressure therapy devices include a feature known as "forward pose trim". This function is now framed to detect peaks and remove any subsequent expiratory pauses from the previous expiration by extrapolating backwards to the appropriate zero crossing that would represent the start of inspiration. It was designed to cut out the front of the intake air. This feature may not be useful for M-shaped breaths, as shown in FIG. FIG. 30 shows an incorrect trim (for the second peak) and an accurate trim (for the forward peak).

The method can be modified to set boundaries where inspiratory peaks can be found that can be assisted with M-shaped breaths. Thus, for an M-shaped breath, the start of inspiration is determined by extrapolation from the first peak, rather than the second peak, of the inspired portion of the breath based on the boundaries within the range of inspiration. For example, this boundary may be calculated as follows:

２．以下のとおり吸気中の時点ｔ_ｌｉｍを見つける。

３．［ｔ_０：ｔ_ｌｉｍ］の範囲内で吸気のピークを見つける。
４．現在の前方ポーズトリムアルゴリズムを進める。 1. Calculate the total intake.

2. Find the time t _lim during inspiration as follows.

3. Find the peak of inspiration within the range of [t ₀ : t _lim ].
4. Advance the current forward pose trim algorithm.

The plot of FIG. 31 shows a sequence of flow-limited breaths in a flow signal in which the number of breaths is similar to the breath shown in FIG. The upper panel of FIG. 31 represents the flow rate plotted against time, and the lower panel represents the calculated inspiratory time (T _i ) with forward pose trim applied. Both the results of the current algorithm and the results of the new algorithm ("corrected" _Ti ) are shown. The current algorithm cut in half certain breath, it is seen that results in a T _i of artificially and inaccurate low. The new algorithm implements "forward pose trim" in a consistent manner.

Section G - vent indicators, such as the calculation permeability of the vent indicators and duty cycle indicator (1) air permeability (VR) (VR) is determined as the ratio of the vent about the current breath to a metaphase ventilation immediately before (V ₃₎ Sometimes. In the present embodiment, the medium period may be a ventilation index filtered using a filter having a time constant τ of 3 minutes. However, other time constants may be appropriate. The filter used may be a simple first order autoregressive filter. Because the time constant of the filter is quite long, the filter takes some time to rise from zero. The index thus changes slowly between a reasonable ventilation value and the filter output during the time [t ₀ : 3 × τ]. VR can be calculated as follows.

通気率の吸気成分：

８．呼気が確認されると、以下の計算をする。
呼気体積：Ｖ_ｅ
呼気時間：Ｔ_ｅ
平均呼気流速：

通気率の呼気成分：

９．以下のとおりＶＲを計算する。
ＶＲ＝（ＶＲ_ｉＴ_ｉ＋ＶＲ_ｅＴ_ｅ）／（Ｔ_ｉ＋Ｔ_ｅ） 1. The gain of the filter G = 1 / _{f s} τ
Set to. Where f _s is the sampling frequency and τ is the time constant of the filter in seconds. For example, if the sampling frequency is 50 Hz and the time constant is 3 minutes,
G is
G = 1 / (50 × 180)
Is calculated by
2. Initialize the filter to a reasonable ventilation value, such as 0.2 l / sec.
3. From the total flow, calculate the properly filtered patient respiratory flow (Q _p ), minus any vent flow, with any mask leaks subtracted.
4. Calculate mid-term ventilation.
_{V 3 = V 3 + G (} Q p -V 3)
5. V ₃ = (3τ−t) /3τ×0.2+t/3τ×V ₃ during transition period 0 <t ≦ 3τ
6. Frame your breath in the usual way.
7. When the intake is confirmed, the following calculation is performed.
Intake volume: _{V i}
Inspiration time: _Ti
Average intake flow velocity:

Inhalation component of air permeability:

8. When expiration is confirmed, the following calculation is made.
Expiration volume: V _e
Expiration time: _Te
Average expiratory flow rate:

Expiratory components of air permeability:

9. Calculate VR as follows.
VR = (VR _{i Ti} + VR _{e Te} ) / ( _Ti + _Te )

(2) Calculation of Duty Cycle Index (eg, TTR) The duty cycle index may be introduced to derive an obstructive index as described above. For example, from a measure of respiratory flow, a second ratio of the duration of the inspiratory portion of the respiratory cycle to the duration of the respiratory cycle as a function of the measure of respiratory flow. Similarly, a second such index may be determined that may follow the first index in time. The occlusiveness index may then be derived as a function of the first ratio and the second ratio.

A suitable duty cycle indicator such as the T _i -on-T _tot rate (TTR) is
It may be determined as the ratio of the current _{(breath) T} i _{-on-T tot} values for _{i -on-T tot} values. The metaphase can be determined by filtering using a 5 minute time constant or other suitable time constant. The filter may be a simple first order autoregressive filter. TTR may be calculated as follows.

1. The gain of the filter G = 1 / _{f s} τ
Set to. Where f _s is the sampling frequency and τ is the time constant of the filter in seconds. For example, if the sampling frequency is 0.25 Hz (approximate breath frequency) and the time constant is 5 minutes, G is
G = 1 / ((1/4) × 300)
Is calculated by
2. Initialize the filter to a reasonable value such as 0.4.
3. From the total flow, calculate the properly filtered patient respiratory flow (Q _p ), minus any vent flow, with any mask leaks subtracted.
4. The mid-term T _i -on-T _tot is calculated as follows.
_{T i T tot (5) =} T i T tot (5) + G (T i T tot -T i T tot (5))

Calculate TTR as follows.
TTR = TiTtot / TiTtot ₅

Section H-Weight Reduction Function The functions graphed in FIGS. 32 and 33 can be used to reduce the weight of the impact of a flow restriction (eg, FFL) indicator or a snoring indicator. this function is,
It can be applied to increase the threshold of an indicator (eg, FFL or snoring) that needs to be exceeded to cause a pressure rise. For example, the weight reduction function may be related to valve-like leakage rates. The valve-like leak index can be calculated in a customary manner such that its value varies between 0 and 5. In order to prevent spurious valve-like leaks from interfering with the pressure build-up when no leaks are actually occurring, the system is implemented to reduce the weight of valve-like leak values to the extent that no complete leak exists. Therefore, the algorithm for determining the weight reduction due to valve-like leakage is as follows.

1. Calculate valve-like leakage from the last intake.
2. Determine the value of complete leak (end of exhalation).
3. The valve-like leakage value is multiplied by the output of a complete leakage function, such as the function shown by the graph on the left of FIG. For example, if the complete leak is less than 0.025 (<), the valve-like leak is set to zero. Alternatively, the total leak is greater than 0.05 (
If>), the value of valve-like leakage does not change, and in the interval between 0.025 and 0.05, the value of valve-like leakage decreases linearly.
4. To output the weight reduction factor, the value of valve-like leakage is used with a leakage function such as the function shown in the graph on the right side of FIG. For example, the output is 1 for values less than 4 (<) and zero for values greater than 5 (>). In the interval between 4 and 5, the output decreases linearly from 1 to zero.

The weight reduction of the flow restriction index (eg, FFL) based on the leak (eg, L / s)
This can be done using the function shown in the graph of FIG. For leak values less than 0.5 (<), there is no weight reduction and the output is 1. For values of (>) leakage greater than 0.7, there is a complete weight reduction and the output is zero. For leakage values of 0.5 to 0.6, 1
There is a linear decrease in output from 0 to 0.

Weight reduction of the flow restriction index by pressure, such as the level of the mask CPAP in cmH ₂ O, may be achieved using a function such as the embodiment graphed in FIG. In an embodiment, to a pressure level of less than 10 cm H ₂ O, rather than the weight reduction, relative to a pressure higher than 15 cmH ₂ O, there is partial weight reduction, the output is 0.8. For pressures between 10 and 15, there is a linear decrease in power from 1 to 0.8. The output may then be multiplied by a flow restriction indicator.

Finally, the effects of flow restriction or obstruction indicators can decrease in weight with increasing disturbance. Disturbance is an indicator such as a fuzzy range where the current inspiration or expiration lasts very long. For example, interference may be determined by the methods described in US Pat. No. 6,484,719, the disclosure of which is incorporated herein by reference. High disturbances represent a transient change in leakage, for example, when the patient opens his or her mouth, or when the patient moves to bed and changes the mask position in contact with the face. When the level of disturbance is high, the flow estimate is not accurate and the leakage constant may be reduced to help improve the flow estimate. While the reduction is taking place, it is advisable to avoid the pressure build up until the event has subsided. Avoidance of the pressure rise can be realized by the functional embodiment of FIG. When the jamming reaches 0.25, the system can start de-weighting until the indicator is completely de-weighted at a jamming level of 0.5.
The output of the function may be multiplied by a flow restriction indicator to achieve weight reduction.

Similarly, the snoring indicator may be de-weighted based on various conditions of the system. For example, the snoring value may be reduced in weight by a function of the pressure shown in FIG. This weight reduction reduces the sensitivity of the system to snoring indicators for the purpose of creating a pressure rise. Thus, the value of snoring required for increased treatment pressure increases with increasing pressure.

The inspiratory snoring index may be de-weighted by an index of ventilation, such as an index of "breath". Breath often causes noise simply due to the high peak flow achieved. Utilizing the functional embodiment of FIG. 37, ventilation may be used to generate a reduced weight output coefficient depending on whether the ventilation index is considered to be breath. For example, a ventilation index such as a ventilation rate (VR) may be used as an index of breath. A value of (>) VR greater than 1 may be interpreted as indicative of a breath greater than mid-ventilation. 1
. For values of VR less than 2 (<), the system may avoid reducing the weight of any snoring indicator, and for values of VR> 1.5, the system may be affected by snoring. May be completely reduced. For values of VR between 1.2 and 1.5, the output weight reduction factor may decrease linearly from 1 to zero.

Other calculations that modify the effects of flow restriction or snoring based on leak values, pressure, obstruction and / or ventilation indicators may also be used.

SECTION I Normalized Expiratory Peak Position Normalized expiratory peak position (NEPL) can be used to wake, wake, or otherwise unnatural, extraordinary, or awake from sleep or obstructive sleep flow waveforms , A good indicator of the transition to a waveform representing an unrecognizable event. Thus, the device of the present technology may introduce an indicator of arousal based on peak expiratory flow. For example, arousal may be assessed depending on the location of the peak or normalized peak expiratory flow within the expiratory portion of the respiratory cycle. In such embodiments, the time of the expiratory portion may be defined by a range, and the appearance of a peak within the defined range defines an index that represents arousal. This index may represent arousal if the expiratory peak appears during the late time or expiratory portion of the respiratory cycle.

  Such an index may be calculated as follows:

1. Frame your breath.
2. Separate the expiratory portion of each breath.
3. Find out when peak expiratory flow appears.
4. The time from the beginning of expiration to the peak is divided by the total expiratory time, for example, the index is [0
. 0: 1.0].

Typically, the NEPL is in the range of zero to 0.3 while the patient is sleeping. For example, consider the histogram of FIG. 38, which represents a comparison of three data sets corresponding to NEPL.
(1) Essen-OSA patients on treatment. (2) Concord-Patients who are escalating. (3) Awake-Person breathing on the AutoSet Split airway pressure device available from ResMed.

The graph on the right side of FIG. 38 shows that an awake breather breathes much more than 0.5 (>) breaths than a sleeping patient who breathes less times. Thus, the system can be implemented with a weight reduction function for extraordinary, unnatural, or unexpected events related to sleep based on flow restriction indicators. Suitable functions based on the calculated or determined normalized inspiratory peak position values are shown in the graph of FIG. This function increases the intensity of the flow restriction index required to cause an increase in treatment pressure when the breath becomes more unnatural, unusual, or unrecognizable (eg, abnormally). Used to let.

In the above description and accompanying drawings, specific terms and symbols are provided to provide a thorough understanding of the technology. In some cases, terms and symbols may refer to particular details that are not required to implement the technology. Furthermore, although the techniques herein have been described with reference to particular embodiments, it should be understood that these embodiments are merely illustrative of the principles and uses of the techniques. Therefore, it should be understood that numerous changes can be made to the exemplary embodiments and other mechanisms can be devised without departing from the spirit and scope of the present technology.

For example, a desired flow restriction detection and / or therapy system may be a pattern recognition, such as capturing each inspiration and interpolating on a grid of as many points as needed to maintain important frequency information. Could be based on other techniques for For example, each intake may be interpolated on a 65 point grid. Large sets of training data may be constructed using various types of occlusion waveforms that can be recorded based on clinical evaluations. These waveforms may be pre-categorized as desired, such as mild, moderate, severe, or other similar categories. The 65 points of each waveform may then be input to a classifier (such as a neural network, a support vector machine, or the like) for learning, and a suitable algorithm such as an algorithm based on a genetic algorithm , Arbitrary boundaries may be defined in a 65-dimensional space between the various categories of occlusion. Such a system is then capable of classifying inspiratory waveforms that are measured in the patient against the classifier metrics and used in a control system for patient treatment.

  The potential issues to consider with this approach are:

1. "Dimensional Trouble" refers to the exponential growth of hypervolume as a function of dimension. In other words, as the number of dimensions of the input space increases, the number of learning vectors required to "cover" that space increases exponentially and sufficiently. Covering 65 dimensions can present problems with the speed of this technology.
2. Pre-classification of waveforms for occlusion (and for degree) is cumbersome and there is probably high interobserver variability. Without an increase in CPAP pressure to observe a waveform change, such as "round-up", it is difficult to determine if an occlusion actually exists. Reliable indicators of effort, such as esophageal pressure, can help with the detection problem, but the task of classifying such data in making decisions can still be tricky. For example, REM sleep can generate various, unexpected waveforms.
3. The learning phase requires "significant" numerical resources and a large number of processor cycles.
4. The resulting classifier is difficult to interpret for everyone,
If you use a neural network, depending on how many neurons you end up with)
Working in embedded systems can be numerically intensive.
5. In order to test the resulting classifier, it may be necessary to supply the classifier with sleep studies of everyone on earth.

However, one way to reduce the complexity of the system is to limit the information provided to the system to only "interesting" information. The calculation of unique features, such as the flattening index, does exactly this limitation and serves as a form of signal compression.

Claims (89)

A method for detecting partial obstruction, comprising:
Determining an indicator of respiratory flow;
Determining a shape index representing a pattern of partial obstruction from the index of respiratory flow;
Determining a ventilation index from the respiratory flow index;
Deriving a flow restriction index as a function of the determined shape index and the determined ventilation index;
A method comprising:   The method of claim 1, wherein the pattern of partial occlusion is an M-shaped flow detection function. 3. The method of claim 2, wherein the insufflation index is a ratio that includes a current tidal volume and a previous tidal volume. 4. The method of claim 3, further comprising calculating a pressure demand as a function of a flow restriction indicator. Prior pressure requirements are based on conditions that (a) the shape index represents the presence of an M-shaped breath in the respiratory airflow, and (b) that the ventilation index decreases sufficiently to represent less than normal ventilation. 5. The method of claim 4, wherein the method is increased. 6. The method of claim 5, further comprising determining a second shape index representing a chair pattern of the partial occlusion determined from the respiratory flow index, wherein the derived flow restriction index is a further function of the second shape index. The described method. A method for detecting partial obstruction, comprising:
Determining an indicator of respiratory flow;
Determining a shape index representing a pattern of partial obstruction from the index of respiratory flow;
Determining a duty cycle index from the respiratory flow index;
Deriving a flow restriction index as a function of the determined shape index and the determined duty cycle index;
A method comprising:   The method of claim 7, wherein the pattern of partial occlusion is an M-shaped flow detection function. 9. The method of claim 8, wherein the duty cycle indicator is a ratio that includes a ratio of a current breath inspiration time to a cycle time and a ratio of a prior average breath inspiration time to a cycle time. The method of claim 9, further comprising calculating a pressure demand as a function of a flow restriction indicator. 11. The method of claim 10, wherein the prior pressure requirement is increased as a condition that the shape index indicates the presence of M-shaped breath in the respiratory airflow and that the duty cycle index is increased. 12. The method of claim 11, further comprising determining a second shape index representing a chair pattern of partial occlusion determined from the respiratory flow index, wherein the derived flow restriction index is a further function of the second shape index. The described method. An apparatus for detecting partial occlusion,
A patient interface for transmitting the flow of breathing gas;
A flow sensor coupled to the patient interface for generating a flow signal indicative of a flow rate of the respiratory gas through the patient interface;
A controller coupled to the flow sensor for processing the flow signal;
With
The controller is
(A) determining a shape index representing a pattern of partial occlusion from the flow signal;
(B) determining a ventilation index from the flow signal;
(C) deriving a flow restriction index as a function of the determined shape index and the determined ventilation index;
A device configured to control the device.   14. The apparatus of claim 13, wherein the pattern of partial occlusion is an M-shaped flow detection function. 15. The device of claim 14, wherein the ventilation index is a ratio that includes a current tidal volume and a previous tidal volume. A flow generator coupled to the controller and the patient interface;
The controller is further configured to: (a) calculate the pressure demand as a function of the flow restriction indicator; and (b) set the flow generator in response to the pressure demand.
An apparatus according to claim 13. The controller: (a) the shape index represents the presence of an M-shaped breath in the respiratory airflow; and
17. The apparatus of claim 16, wherein (b) increasing the prior pressure requirement as a condition that the ventilation index is reduced sufficiently to represent less than normal ventilation. The controller is further configured to control a determination of a second shape index representing the chair pattern of the partial occlusion determined from the flow signal, wherein the derived flow restriction index is a further function of the second shape index. An apparatus according to claim 13. An apparatus for detecting partial occlusion,
A patient interface for transmitting the flow of breathing gas;
A flow sensor coupled to the patient interface for generating a flow signal indicative of a flow of breathing gas through the patient interface;
A controller coupled to the flow sensor for processing the flow signal;
With
The controller is
Determining a shape index representing a pattern of partial occlusion from the flow signal;
Determining a duty cycle indicator from the flow signal;
Derivation of a flow restriction index as a function of the determined shape index and the determined duty cycle index,
A device configured to control the device.   20. The apparatus of claim 19, wherein the partial occlusion pattern is an M-shaped flow rate detection function. 21. The apparatus of claim 20, wherein the duty cycle indicator is a ratio including a current breath inspiration time to cycle time ratio and a prior average breath inspiration time to cycle time ratio. A flow generator coupled to the controller and the patient interface;
The controller is further configured to: (a) calculate the pressure demand as a function of the flow restriction indicator; and (b) set the flow generator in response to the pressure demand.
The device according to claim 19. The controller: (a) the shape index indicates the presence of an M-shaped breath in the flow signal; and
(B) increasing the prior pressure demand value as a condition for increasing the duty cycle index.
An apparatus according to claim 1. The controller is further configured to control a determination of a second shape index representing the chair pattern of the partial occlusion determined from the flow signal, wherein the derived flow restriction index is a further function of the second shape index. The device according to claim 19. A system for detecting partial occlusion, comprising:
An interface means for transmitting a flow of breathing gas;
Flow measurement means coupled to the interface means for generating a flow signal representative of the flow of breathing gas through the interface means;
Processing means coupled to the flow measurement means for processing the flow signal;
With
Processing means,
(A) determining a shape index representing a pattern of partial occlusion from the flow signal;
(B) determining a ventilation index from the flow signal;
(C) deriving a flow restriction index as a function of the determined shape index and the determined ventilation index;
A system that is configured to handle   26. The system of claim 25, wherein the pattern of partial occlusion is an M-shaped flow detection function. 27. The system of claim 26, wherein the ventilation index is a ratio that includes a current tidal volume and a previous tidal volume. Flow means coupled to the processing means and the interface means for generating a controlled flow of breathing gas through the interface means;
Processing means is further configured to: (a) calculate the pressure demand as a function of the flow restriction indicator; and (b) set the flow generator in response to the pressure demand.
26. The system of claim 25. Processing means: (a) that the shape index represents the presence of an M-shaped breath in the respiratory airflow; and (b)
29.) The system of claim 28, wherein the pre-pressure requirement is increased as a condition that the ventilation index decreases sufficiently to represent less than normal ventilation. The controller is further configured to control a determination of a second shape index representing the chair pattern of the partial occlusion determined from the flow signal, wherein the derived flow restriction index is a further function of the second shape index. 26. The system of claim 25. A system for detecting partial occlusion, comprising:
An interface means for transmitting a flow of breathing gas;
Flow measurement means coupled to the interface means for generating a flow signal representative of a respiratory flow of breathing gas through the interface means;
Processing means coupled to the flow measurement means for processing the flow signal;
With
Processing means,
Determining a shape index representing a pattern of partial occlusion from the flow signal;
Determining a duty cycle indicator from the flow signal;
Derivation of a flow restriction index as a function of the determined shape index and the determined duty cycle index,
A system configured to control the system.   32. The system of claim 31, wherein the pattern of partial occlusion is an M-shaped flow detection function. 33. The system of claim 32, wherein the duty cycle indicator is a ratio including a current breath inspiration time to cycle time ratio and a prior average breath inspiration time to cycle time ratio. Flow means coupled to the processing means and the interface means for generating a controlled flow of breathing gas through the interface means;
Processing means is further configured to: (a) calculate the pressure demand as a function of the flow restriction indicator; and (b) set the flow generator in response to the pressure demand.
The system according to claim 31. Processing means: (a) that the shape index indicates the presence of an M-shaped breath in the flow signal; and (b)
35.) The system of claim 34, further configured to increase the prior pressure demand as a condition of increasing the duty cycle indicator. The processing means is further configured to control a determination of a second shape index representing the chair pattern of the partial occlusion determined from the flow signal, wherein the derived flow restriction index is a further function of the second shape index. 20. The device of claim 19, wherein: An information storage medium having processor readable information for controlling an apparatus for detecting respiratory partial obstruction, wherein the processor readable information comprises:
Determining an indicator of respiratory flow;
Determining a shape index representing a pattern of partial obstruction from the index of respiratory flow;
Determining a ventilation index from the respiratory flow index;
Deriving a flow restriction index as a function of the determined shape index and the determined ventilation index;
An information storage medium comprising:   38. The information storage medium according to claim 37, wherein the partial occlusion pattern is an M-shaped flow detection function. 39. The information storage medium of claim 38, wherein the ventilation index is a ratio that includes a current tidal volume and a previous tidal volume. 38. The information storage medium of claim 37, wherein the processor readable information further comprises calculating a pressure demand as a function of a flow restriction indicator. Prior pressure requirements are based on conditions that (a) the shape index represents the presence of an M-shaped breath in the respiratory airflow, and (b) that the ventilation index decreases sufficiently to represent less than normal ventilation. 41. The information storage medium of claim 40, which is increased. The processor readable information further comprising determining a second shape index representing a chair pattern of partial occlusion determined from the respiratory flow index, wherein the derived flow restriction index is a further function of the second shape index. The information storage medium according to claim 37, wherein: An information storage medium having processor readable information for controlling an apparatus for detecting respiratory partial obstruction, wherein the processor readable information comprises:
Determining an indicator of respiratory flow;
Determining a shape index representing a pattern of partial obstruction from the index of respiratory flow;
Determining a duty cycle index from the respiratory flow index;
Deriving a flow restriction index as a function of the determined shape index and the determined duty cycle index;
An information storage medium comprising:   44. The information storage medium according to claim 43, wherein the pattern of the partial occlusion is an M-shaped flow detection function. 45. The information storage medium of claim 44, wherein the duty cycle index is a ratio including a ratio of a current breath inspiration time to a cycle time and a ratio of a prior average breath inspiration time to a cycle time. 44. The information storage medium of claim 43, wherein the processor readable information further comprises calculating a pressure demand as a function of a flow restriction indicator. 47. The information-bearing medium of claim 46, wherein the prior pressure requirement is increased as a condition that the shape index indicates the presence of an M-shaped breath in the respiratory airflow and that the duty cycle index is increased. The processor readable information further comprising determining a second shape index representing a chair pattern of partial occlusion determined from the respiratory flow index, wherein the derived flow restriction index is a further function of the second shape index. 44. The information storage medium according to claim 43, wherein: A method for determining a treatment setting of a respiratory treatment device,
Determining an indicator of respiratory flow;
Determining a shape index representing the degree of partial obstruction from the respiratory flow index;
Determining a ventilation index indicative of the degree of ventilation change from the respiratory flow index;
Calculating the treatment setting as a proportional function of both the degree of partial occlusion and the degree of change in ventilation;
A method comprising: 50. The method of claim 49, wherein the shape index is determined from a plurality of breaths within a range of one to three breaths.   50. The method of claim 49, wherein the shape index is an M shape breath index. 50. The method of claim 49, wherein the therapy settings are applied to set up an automatic respiratory therapy device. A device for determining settings for respiratory treatment,
A sensor for determining an indicator of respiratory flow;
(A) A shape index indicating the degree of partial obstruction is determined from the index of the respiratory flow, (b) a ventilation index indicating the degree of change in ventilation is determined from the index of the respiratory flow, and (c) the degree of partial obstruction and the ventilation. A processor configured to derive the treatment setting as a proportional function of both the degree of change of
An apparatus comprising: 54. The apparatus of claim 53, wherein the shape index is determined by the processor from a plurality of breaths in a range of one to three breaths.   54. The device of claim 53, wherein the shape index is an M shape index. The apparatus further comprising a flow generator configured to provide a respiratory pressure therapy to the patient interface, the apparatus further comprising:
A processor configured to control treatment to the patient interface during treatment setting;
The device comprises automatic titration respiratory therapy equipment,
54. The device according to claim 53. An apparatus for determining a treatment setting for respiratory treatment,
Means for determining an indicator of respiratory flow;
Means for determining a shape index representing the degree of partial obstruction from the index of respiratory flow;
Means for determining a ventilation index indicating the degree of change in ventilation from the respiratory flow index;
Means for deriving treatment settings as a proportional function of both the degree of partial occlusion and the degree of change in ventilation;
An apparatus comprising: 58. The apparatus of claim 57, wherein the shape index is determined by the processor from a plurality of breaths in a range of one to three breaths. 58. The device of claim 57, further comprising means for generating and providing a respiratory pressure treatment to the patient in response to the derived treatment settings, wherein the device comprises an automatic titration respiratory treatment device. In a device for treating sleep-disordered breathing, a method for determining a change in treatment,
Determining a first indicator of the degree of partial obstruction from the indicator of respiratory flow;
Determining a second measure of respiratory parameters from the measure of respiratory flow;
Adjusting the first index using the second index to adjust the degree of partial obstruction according to the respiratory parameter;
A method comprising: 61. The method of claim 60, further comprising adjusting the adjusted first indicator upon detection of a breath pattern representing an incorrectly framed breath. 61. The method of claim 60, further comprising generating a treatment setting as a proportional function of the adjusted first indicator.   63. The method of claim 62, wherein the first indicator is a shape index. 64. The method of claim 63, wherein the second indicator represents a degree of change in a ratio of a duration of the inspiratory portion of the respiratory cycle to a duration of the respiratory cycle.   65. The method of claim 64, wherein the second indicator is a function of the relative duration of inspiration and expiration. 66. The method of claim 65, further comprising adjusting the adjusted first indicator as a function of peak flow.   67. The method of claim 66, wherein the peak flow is a peak expiratory flow position.   61. The method of claim 60, wherein the first indicator is a shape index determined from a single breath.   69. The method of claim 68, wherein the shape index is a single breath flattening index. 69. The method of claim 68, wherein breath framing is performed prior to filtering to remove frequencies associated with patient snoring. Flattening is determined prior to filtering to remove frequencies associated with patient snoring;
70. The method of claim 68. In a device for treating sleep-disordered breathing, a method for automatically determining a change in treatment pressure to be delivered by the device,
Determining a first indicator of the degree to which patient inspiratory effort is correlated with patient inspiratory flow;
Determining a second indicator of patient ventilation;
Determining a change in treatment pressure to be a first function of the first index and the second index;
A method comprising:   73. The method of claim 72, wherein the first indicator is an index of partial occlusion. 73. The method of claim 72, wherein the second indicator of patient ventilation is a measure of the degree to which patient ventilation is appropriate. 73. The method of claim 72, wherein the first indicator is an indicator of the relative duration of inspiration and expiration. In a device for treating sleep-disordered breathing, a method for automatically determining an index of a partial obstruction of a patient's airway,
Determining an indicator of respiratory flow;
Determining a first partial obstruction index from the respiratory flow index;
Determining a second partial occlusion index from the respiratory flow index;
Determining a third partial occlusion index as a function of the first partial occlusion index and the second partial occlusion index;
A method comprising:   77. The method of claim 76, wherein the first occlusion indicator is an index representing an M-shaped breath pattern.   80. The method of claim 77, wherein the second occlusion indicator is an index representing a chair-shaped breath pattern. In a device for treating sleep-disordered breathing, a method for automatically determining an index of a partial obstruction of a patient's airway,
Determining an indicator of respiratory flow;
Determining a first partial obstruction index from the respiratory flow index;
Filtering the first partial occlusion index to derive a historical indicator of the partial occlusion;
Resetting the filtering process upon detection of a non-obstructed breath pattern from the respiratory flow index;
A method comprising:   80. The method of claim 79, wherein the first partial occlusion indicator is a flattening index. In a device for treating sleep-disordered breathing, a method for automatically determining an index of a partial obstruction of a patient's airway,
Determining an indicator of respiratory flow;
Filtering an indicator of respiratory flow in a frequency range representing snoring;
Analyzing the magnitude of the power in the frequency domain as a function of time to evaluate whether the power signal represents noise;
A method comprising: 82. The method of claim 81, wherein analyzing comprises applying a Shannon entropy function to power magnitude in the frequency domain. In a device for treating sleep-disordered breathing, a method for automatically determining the start of inspiration,
Determining an indicator of respiratory flow;
Detecting a first peak from the respiratory flow;
Determining the start of a respiratory cycle as a function of the first peak;
With
The method wherein the detection of the first peak is limited to avoid detecting a second peak during the respiratory cycle. In a device for treating sleep-disordered breathing, a method for automatically determining an index of a partial obstruction of a patient's airway,
Determining an indicator of respiratory flow;
Determining a first ratio of the duration of the inspiratory portion during the respiratory cycle to the duration of the respiratory cycle as a function of a measure of respiratory flow;
Determining a second ratio of the duration of the inspiratory portion during the respiratory cycle to the duration of the respiratory cycle as a function of respiratory flow;
Deriving an indication of partial occlusion as a function of the first ratio and the second ratio;
A method comprising: In a device for treating sleep-disordered breathing, a method for automatically determining an index of a partial obstruction of a patient's airway,
Determining an indicator of respiratory flow;
Determining an index of partial obstruction from the index of respiratory flow;
Determining, from the respiratory flow, an index of disturbance that is indicative of a transient change in leakage;
Adjusting the indicator of partial occlusion as a function of the indicator of obstruction;
A method comprising:   86. The method of claim 85, wherein the indication of partial occlusion is a function of detecting an M-shaped pattern breath. In a device for treating sleep-disordered breathing, a method for automatically determining an index of arousal,
Determining an indicator of respiratory flow;
Determining a peak indicator of the expiratory portion of the indicator of respiratory flow;
Determining an index of arousal as a function of the peak index;
A method comprising:   90. The method of claim 87, wherein the function of the peak indicator is a function of a position within the expiratory portion. 89. The method of claim 88, wherein the arousal indicator represents arousal if the peak indicator appears in the second half of the expiratory portion of the respiratory flow indicator.

Images