Systems and methods for screening, diagnosis, detection, monitoring and / or treatment
By using photoplethysmography sensors and machine learning algorithms in a home sleep apnea testing device, the problem of the inability to distinguish between central and obstructive breathing events in existing technologies has been solved, enabling accurate diagnosis and monitoring of sleep-disordered breathing.
Patent Information
- Application Number
- CN202380032164.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-02-03
- Filing Date
- 2023-02-01
- Publication Date
- 2024-11-26
AI Technical Summary
Existing home sleep apnea testing devices (such as PAT HSAT) cannot effectively distinguish between central and obstructive respiratory events and lack monitoring channels for airflow and breathing effort, resulting in insufficient diagnostic accuracy.
By acquiring data using a photoplethysmography (PPG) sensor at the finger, and combining it with machine learning algorithms and classifiers, respiratory effort-related information is extracted to distinguish between central and obstructive respiratory events.
It enables accurate detection and differentiation of central sleep apnea in the home environment, improves the sensitivity and specificity of diagnosis, and can effectively guide further in-laboratory PSG validation.
Smart Images

Figure CN119031880A_ABST
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application claims the benefit of U.S. Provisional Application No. 63 / 306,413, filed on February 3, 2022, the entire disclosure of which is hereby incorporated by reference. Background Art Technical Field
[0003] The technology relates to one or more of screening, detection, diagnosis, monitoring, treatment, prevention and improvement of breathing-related disorders. The technology also relates to medical devices or equipment and their uses. In some embodiments, the technology relates to screening, detection, diagnosis, monitoring, treatment, prevention and / or improvement of sleep breathing disorders, such as central or obstructive apnea.
[0004] 2.2 Description of Related Technology
[0005] Sleep apnea (SA) is a common sleep disorder affecting millions of people worldwide[1],[2]. Apnea events are defined as cessation of airflow during sleep lasting 10 seconds or longer, while hypopnea events are characterized by a reduction in airflow rather than a complete cessation[3],[4]. The severity of a patient's SA can be expressed by their apnea-hypopnea index (AHI), which is simply the number of apnea and hypopnea events per hour of sleep[5],[6],[7].
[0006] SA can be obstructive SA (OSA) or central SA (CSA) in nature. OSA is the more common form of SA, and some of its symptoms and consequences are fatigue, daytime sleepiness, arrhythmias, and systemic hypertension [3]. During OSA, respiratory effort persists, but the upper airway is mechanically obstructed, resulting in airflow interruption. On the other hand, CSA is characterized by the absence of neural drive for breathing during sleep [8], [9]. Both types of respiratory events can coexist in one patient. Respiratory events may also be of mixed origin, characterized by a lack of respiratory drive followed by obstructive breathing. Typically, CSA is the primary diagnosis when at least 50% of respiratory events are scored as of central origin. CSA can be characterized by the cAHI (central AHI), which is calculated as the number of central respiratory events per hour of sleep.
[0007] Although the symptoms of CSA often resemble those of OSA, the choice of therapy depends on the type of SA. Treatments for OSA include lifestyle measures (e.g., weight loss), mandibular advancement devices, surgery, and most commonly, continuous positive airway pressure (CPAP)
[10] ,
[11] . Comparatively, there are a variety of therapies for CSA, some of which differ significantly from those used for OSA. These approaches include pharmacological interventions, oxygen, nocturnal mechanical ventilation (via nasal mask or tracheostomy and endotracheal intubation), and diaphragmatic pacing [3],
[13] . Therefore, it is important for clinicians to distinguish between CSA and OSA in patients with SA in order to make the best treatment decisions.
[0008] The gold standard for diagnosing SA is polysomnography (PSG)
[14] ,
[15] . PSG utilizes multiple sensory channels, including electroencephalogram, electrooculogram, electromyogram, electrocardiogram, and pulse oximetry, as well as airflow and respiratory effort to assess the underlying causes of sleep disturbances
[16] . However, due to the inconvenience of performing in-laboratory PSG, and in part due to the global COVID pandemic, home sleep apnea testing (HSAT) based on peripheral arterial tonometry (PAT) has rapidly gained popularity and currently comprises the most widely deployed category of HSAT
[17] . PAT HSAT allows for minimally invasive multi-night testing and can be provided in a completely disposable form
[18] ,
[19] . An example of such a system is called NightOwl PAT HSAT, which is described by Massie et al.
[20] . It consists of a fingertip-sized probe that works with cloud-based analysis software. ( Figure 1 ). The analysis determines respiratory events from a peripheral arterial tone signal determined by a photoplethysmographic signal generated by a sensor.
[0009] Compared to PSG, PAT HSAT has a major disadvantage, namely, a reduced channel design (i.e., fewer sensors and signals). PAT-based HSAT derives most of its modalities from finger photoplethysmography (PPG), from which it derives blood oxygen saturation (SpO2), pulse rate (PR), and PAT. Essentially, this means that PAT HSAT does not have access to information sources such as airflow reduction and cortical arousals, and must instead rely on alternative information sources to infer the presence of respiratory events
[18] .
[0010] Another challenge for PAT HSAT is the distinction between types of respiratory events. Both types of events are characterized by a decrease in airflow. However, central respiratory events are accompanied by a cessation of respiratory effort, as traditionally inferred from the abdominal and thoracic respiratory effort bands of PSG (which monitor respiration-related fluctuations around the abdomen and chest) [5]. Figure 2 A comparison of respiratory effort during central and obstructive respiratory events is shown.
[0011] PAT HSAT devices that only use a finger PPG probe for respiratory event detection do not have airflow and respiratory effort channels (signals) that can be used for respiratory event type labeling.
[0012] It is desirable to have an automated method to detect CSA in patients with SA using only PPG data acquired at the finger by PAT HSAT. A validated PPG-based CSA detection method can be used to flag CSA in patients, which enables referral to an in-laboratory PSG for further verification.
[0013] I. Other Work
[0014] Recently, Pillar et al. published a study using WatchPAT TM In a validation study of a novel CSA detection method for PAT HSAT
[21] , additional sensors were placed on the chest. In addition to the finger PPG, the system included an optional snoring and body position (SBP) sensor located below the sternal notch. The SBP sensor was used to derive the upper chest wall respiratory motion that is attenuated during central respiratory events. According to Pillar et al.
[21] , the systolic upstroke of the PPG exhibits greater variability during obstructive respiratory events compared to central respiratory events. The method uses upstroke variability along with respiratory effort data obtained by the SBP sensor to identify central events.
[0015] Lazazzera et al. also published a method using PPG and SpO2 signals to detect respiratory events and classify them into apneas and hypopneas and their origin (central or obstructive)
[22] . Their method used PPG features, SpO2 features, pulse rate time domain features, and pulse rate frequency domain features to classify respiratory events. However, they did not report any patient performance analysis or the accuracy of cAHI estimation.
[0016] 2.2.1 Screening / diagnosis / surveillance systems
[0017] Screening and diagnosis generally describe the identification of a disorder from its signs and symptoms. Screening typically gives a true / false result, such as indicating whether a patient's disorder is severe enough to warrant further investigation, while diagnosis can yield clinically actionable information. Screening and diagnosis tend to be one-time processes, while monitoring the progression of a disorder can continue indefinitely. Some screening / diagnostic systems are only for screening / diagnosis, while some can also be used for monitoring.
[0018] As previously mentioned, polysomnography (PSG) is a conventional system for diagnosing / monitoring cardiopulmonary disorders, and a clinical expert is usually involved to apply the system. PSG usually involves placing 15 to 20 contact sensors on a person to record various biosignals, such as electroencephalogram (EEG), electrocardiogram (ECG), electrooculogram (EOG), electromyogram (EMG), etc. PSG for sleep apnea involves observing the patient in a clinic for two nights, one night for pure diagnosis, and the second night for titration of treatment parameters by a clinician. A clinical expert may be able to fully diagnose or monitor a patient based on visually observed PSG signals. However, there are situations where a clinical expert may not be available or may not be able to afford a clinical expert. Therefore, PSG is both expensive and inconvenient. In particular, it is not suitable for home diagnosis / monitoring.
[0019] Another convenient home screening / diagnosis / monitoring system includes a nasal cannula, a pressure sensor, a processing device, and a recording device. The nasal cannula is a device including two hollow open protrusions that are configured to be slightly inserted into the patient's nostrils non-invasively so as to interfere with the patient's breathing as little as possible. The hollow protrusion is in fluid communication with a pressure transducer via a Y-shaped tube. The pressure transducer provides a data signal representing the pressure at the entrance of the patient's nostrils (nasal pressure). It has been shown that the nasal pressure signal is a satisfactory substitute for the nasal flow signal generated by a flow transducer in series with a sealed nasal mask because the nasal pressure signal is comparable in shape to the nasal flow signal. The processing device can be configured to analyze the nasal pressure signal from the pressure transducer in real time or near real time to detect and classify SDB events, thereby monitoring the patient's condition. Screening or diagnosis may require similar analysis, but not necessarily in real time or near real time. Therefore, the recording device is configured to record the nasal pressure signal from the pressure transducer for later offline or "batch" analysis by the processing device for screening / diagnosis purposes.
[0020] 2.2.2 Therapy
[0021] A range of therapies have been used to treat or improve respiratory conditions. In addition, such therapies can be used by otherwise healthy individuals to prevent the onset of respiratory disorders.
[0022] Nasal continuous positive airway pressure (CPAP) therapy has been used to treat obstructive sleep apnea (OSA). The mechanism of action is that the continuous positive airway pressure acts as a pneumatic splint and can prevent upper airway occlusion by pushing the soft palate and tongue forward and away from the posterior oropharyngeal wall.
[0023] High flow therapy (HFT) does not rely on positive airway pressure within a sealed patient interface, but rather relies on delivering air at a therapeutic flow rate to near the entrance of the patient's airway via an unsealed patient interface that may be significantly open to the atmosphere. High flow therapy has been used to treat OSA, CSR, and COPD. Air delivered to the airway at a high flow rate, relative to typical respiratory flow, flushes out the patient's anatomical dead space and reduces the amount of re-breathed CO2, thereby improving the efficiency of gas exchange. High flow therapy can be used in conjunction with respiratory pressure therapy.
[0024] Non-invasive ventilation (NIV) therapy provides ventilation support to patients through the upper airway to help patients breathe fully and / or maintain adequate oxygen levels in the body by doing some or all of the work of breathing. NIV is provided via a non-invasive patient interface. NIV has been used to treat CSR, OHS, COPD, NMD, and chest wall disorders.
[0025] 2.2.3 Diagnosis and treatment system
[0026] These therapies can be provided by a therapeutic system or device.Such systems and devices can also be used to diagnose a condition without treating the condition.
[0027] The treatment system may include a respiratory therapy device (RT device) such as a respiratory pressure therapy device (RPT device), or a respiratory flow therapy device such as a high flow therapy device (HFT device), an air circuit, a humidifier, and a patient interface.
[0028] 2.2.3.1 Patient interface
[0029] The patient interface can be used to couple the respiratory device to its user, for example by providing a flow of breathable gas. The flow of breathable gas can be provided to the user's nose and / or mouth via a mask, to the user's mouth via a tube, or to the user's trachea via a tracheostomy tube. Depending on the therapy to be applied, the patient interface can, for example, form a seal with the patient's facial area to facilitate delivery of gas at a pressure sufficiently different from the ambient pressure (e.g., a positive pressure of about 10 cmH2O) to achieve therapy. For other forms of therapy, such as delivery of oxygen, the patient interface may not include a seal sufficient to facilitate delivery of an air supply to the airway at a positive pressure of about 10 cmH2O.
[0030] 2.2.3.2 Respiratory therapy (RT) devices
[0031] Air pressure generators are known in a range of applications, such as industrial-scale ventilation systems. However, air pressure generators for medical applications have specific requirements that are not met by more general purpose air pressure generators, such as reliability, size, and weight requirements for medical devices. Furthermore, even devices designed for medical use may have disadvantages, including one or more of comfort, noise, ease of use, efficacy, size, weight, manufacturability, cost, and reliability.
[0032] One known RPT device for treating sleep-disordered breathing is the S9 Sleep Therapy System manufactured by ResMed, which demonstrates CPAP therapy. Another example of an RPT device is a ventilator. Ventilators such as the ResMed Stellar ventilator for adults and pediatrics TM The company has developed a series of invasive and non-invasive non-dependent ventilation therapies for a range of patients, for the treatment of a variety of conditions such as, but not limited to, NMD, OHS and COPD.
[0033] An RPT device typically includes a pressure generator, such as a motor-driven blower or a compressed gas reservoir, and is configured to supply pressurized air to the patient's airway. The outlet of the RPT device is connected via an air circuit to a patient interface such as those described above.
[0034] There remains a need for a system that can not only monitor the health of users, but also help those users understand how to improve their own quality of life, such as by understanding their symptoms and triggers, and how they respond to medications.
[0035] In addition, it is desirable that the system be able to access and process as much data as possible related to the user's condition. In this regard, it is also desirable that the system be able to monitor the patient as much as possible during the day and night, not just when the patient is in bed. Summary of the invention
[0036] The present technology is intended to provide a medical device for screening, monitoring, diagnosing, ameliorating, treating or preventing respiratory disorders with improved one or more of comfort, cost, efficacy, ease of use and manufacturability.
[0037] The technology may include a system, device, and / or method that detects CSA in SA patients, such as using only PPG data acquired at a finger by a PPG sensor such as a PAT HSAT. For example, the technology may be implemented to discriminate / distinguish / differentiate CSA from OSA events. In this regard, the technology may detect such events by discrimination. That is, such discrimination may distinguish one type of event from another type of event having similar characteristics (e.g., different types of apnea events) in order to distinguish between them.
[0038] In some embodiments, the method of the system / device may involve extracting and analyzing respiratory effort related information from finger PPG data, such as by deriving a signal that resembles PSG-based respiratory effort as closely as possible. The extracted respiratory signal may then be used to extract features for use in a machine process or machine learning process or classifier, such as an ensemble of tree classifiers, designed / trained to distinguish between two types of respiratory events that may have similar features. This distinction may be part of a device for a screening, monitoring, and / or therapy system.
[0039] Some embodiments of the present technology may include a processor-readable medium having processor-executable instructions stored thereon, which, when executed by one or more processors, cause the one or more processors to distinguish sleep disordered breathing events from signals generated by a photoplethysmography (PPG) sensor. The processor-executable instructions may include instructions to receive data representing a signal from a PPG sensor. The processor-executable instructions may include instructions to derive a PPG-based breathing signal from the data representing the signal from the PPG sensor. The processor-executable instructions may include instructions to calculate first-level features from the PPG-based breathing signal. The processor-executable instructions may include instructions to calculate second-level features from the first-level features. The processor-executable instructions may include instructions to evaluate the first-level features and the second-level features, such as in a classifier. The processor-executable instructions may include instructions to generate an output indication identifying a type of sleep disordered breathing based on the evaluation.
[0040] In some embodiments, the output indication identifying the type of sleep disordered breathing may characterize multiple events of a sleep session. The processor executable instructions may also include instructions for deriving a peripheral arterial tension measurement (PAT) signal from data representing a signal from a PPG sensor. The processor executable instructions may also include instructions for detecting one or more respiratory events based on the PAT signal. The output indication identifying the type of sleep disordered breathing may characterize one or more respiratory events. The PPG sensor may be a finger sensor. The output indication may identify a central sleep apnea type. Central sleep apnea types may include mixed central and obstructive sleep apnea types. The output indication may identify an obstructive sleep apnea type.
[0041] In some embodiments, the instructions for calculating the first-level features may include instructions for calculating values from multiple windows of the PPG-based respiratory signal. For each of the multiple windows, the calculated values may include any one or more of the range, interquartile range, variance, and multiple local maxima. The multiple windows may include a window of a first uniform size moving along the PPG-based respiratory signal. The multiple windows may also include a window of a second uniform size moving along the PPG-based respiratory signal, the second uniform size being shorter than the first uniform size. The instructions for calculating the second-level features from the first-level features may include instructions for selecting a subset of values for each of the first-level features, wherein the subset of values may be associated with a subset of windows corresponding to the occurrence of detected respiratory events in the multiple windows. The instructions for calculating the second-level features from the first-level features may include instructions for calculating one or more values of the second-level features from the selected subset of values. The one or more calculated values of the second-level features may include any one or more of the following: a minimum value, an average of the lowest three values, and one or more percentile values.
[0042] In some embodiments, the instructions for deriving a PPG-based respiration signal from data representing a signal from a PPG sensor may include instructions for filtering a baseline modulation of the data representing the signal from the PPG sensor to generate a filtered signal. The instructions for deriving a PPG-based respiration signal from data representing a signal from a PPG sensor may include instructions for generating an envelope signal from peaks of the filtered signal. The instructions for deriving a PPG-based respiration signal from data representing a signal from a PPG sensor may also include instructions for normalizing the envelope signal by subtracting a moving average window of the envelope signal and dividing the envelope signal by a moving interquartile range window of the envelope signal. In some embodiments, the evaluation may be performed in a classifier, or the evaluation may be performed in a classifier that may include an ensemble of tree classifiers.
[0043] In some embodiments, the processor-executable instructions may further include instructions to detect a vasoconstriction event from data representing a signal from a PPG sensor. The output indication may be based on the detected vasoconstriction event. The instructions to detect a vasoconstriction event may include instructions to filter the PPG signal data representing the signal from the PPG sensor to produce a filtered PPG signal. The instructions to detect a vasoconstriction event may include instructions to process the filtered PPG signal to detect a trough attributable to vasoconstriction. The processor-executable instructions may further include instructions to omit one or more values of one or both of the first-level features and the second-level features associated with the detected vasoconstriction event from evaluation (e.g., of a classifier).
[0044] In some embodiments, the processor executable instructions may also include instructions for detecting an arrhythmic event from data representing a signal from a PPG sensor. The output indication may be based on the detected arrhythmic event. The instructions for detecting an arrhythmic event may include instructions for detecting one or more irregular peak-peak intervals in the PPG signal associated with an arrhythmia. The processor executable instructions may also include instructions for omitting data corresponding to at least one of the one or more respiratory events from an evaluation (e.g., of a classifier), the at least one respiratory event may be associated with the detected one or more irregular peak-peak intervals. The instructions for detecting one or more irregular peak-peak intervals may include instructions for calculating the absolute value of the difference between two adjacent peak-peak intervals. The instructions for detecting one or more irregular peak-peak intervals may include instructions for calculating the average of the absolute values of the differences between adjacent peak-peak intervals of the PPG signal. The instructions for detecting one or more irregular peak-peak intervals may include instructions for comparing the absolute value and the average value. The processor executable instructions may also include instructions for generating a signal for controlling the operation of a respiratory therapy device based on the output indication. The control operation may include controlling the pressure or flow therapy of a blower of the respiratory therapy device. The instructions to receive data representing a signal from a PPG sensor may include instructions for receiving the data at a server.
[0045] Some implementations of the present technology may include a server accessing a processor-readable medium as described herein, wherein the server may be configured to receive a request to download processor-executable instructions of the processor-readable medium to a processing device over a network.
[0046] Some embodiments of the present technology may include a processing device that may include one or more processors and a processor-readable medium as described herein, or the processing device may be configured to access processor-executable instructions using a server as described herein. The processing device may be a respiratory therapy device. The processing device may be configured to generate a pressure therapy or a flow therapy.
[0047] Some embodiments of the present technology may include a method for a server to access a processor-readable medium as described herein, which may include receiving a request at the server to download processor-executable instructions of the processor-readable medium to an electronic processing device over a network. The method may include sending the processor-executable instructions to the electronic processing device in response to the request.
[0048] Some embodiments of the present technology may include a method for one or more processors to distinguish sleep disordered breathing events from signals generated by a photoplethysmography (PPG) sensor. The method may include accessing any processor-readable medium described herein using one or more processors. The method may include executing processor-executable instructions of the processor-readable medium in one or more processors.
[0049] Some embodiments of the present technology may include a method in which one or more processors distinguish sleep disordered breathing events from a signal generated by a photoplethysmography (PPG) sensor. The method may include receiving data representing a signal from a PPG sensor. The method may include deriving a PPG-based breathing signal from the data representing a signal from the PPG sensor. The method may include calculating a first-level feature from the PPG-based breathing signal. The method may include calculating a second-level feature from the first-level feature. The method may include evaluating the first-level feature and the second-level feature, such as in a classifier. The method may include generating an output indication identifying a type of sleep disordered breathing based on the evaluation, such as by a classifier.
[0050] In some embodiments, the output indication identifying the type of sleep disordered breathing characterizes multiple events of a sleep session. The method may also include deriving a peripheral arterial tonometry (PAT) signal from data representing a signal from a PPG sensor. The method may also include detecting one or more respiratory events based on the PAT signal. The output indication identifying the type of sleep disordered breathing characterizes one or more respiratory events. The PPG sensor may be a finger sensor. The output indication may identify a central sleep apnea type. Central sleep apnea types may include mixed central and obstructive sleep apnea types. The output indication may identify an obstructive sleep apnea type.
[0051] Calculating the first level features may include calculating values from a plurality of windows of the PPG-based respiratory signal. For each of the plurality of windows, the calculated values may include any one or more of a range, an interquartile range, a variance, and a plurality of local maxima. The plurality of windows may include a window of a first uniform size that moves along the PPG-based respiratory signal. The plurality of windows may also include a window of a second uniform size that moves along the PPG-based respiratory signal, the second uniform size being shorter than the first uniform size. Calculating the second level features from the first level features may include selecting a subset of values for each of the first level features, wherein the subset of values may be associated with a subset of windows in the plurality of windows that correspond to occurrences of detected respiratory events. Calculating the second level features from the first level features may include calculating one or more values of the second level features from the selected subset of values. The one or more calculated values of the second level features may include any one or more of the following: a minimum value, an average of the lowest three values, and one or more percentile values. Deriving the PPG-based respiratory signal from the data representing the signal from the PPG sensor may include filtering a baseline modulation of the data representing the signal from the PPG sensor to generate a filtered signal. Deriving the PPG-based respiratory signal from the data representing the signal from the PPG sensor may include generating an envelope signal from a peak of the filtered signal. Optionally, deriving the PPG-based respiration signal from the data representing the signal from the PPG sensor may further include normalizing the envelope signal by subtracting a moving average window of the envelope signal and dividing the envelope signal by a moving interquartile range window of the envelope signal. The evaluation may be performed in the classifier, or the evaluation may be performed in a classifier that may include an ensemble of tree classifiers.
[0052] In some embodiments, the method may further include detecting a vasoconstriction event from the data representing the signal from the PPG sensor. The output indication may be based on the detected vasoconstriction event. Detecting the vasoconstriction event may include filtering the PPG signal data representing the signal from the PPG sensor to produce a filtered PPG signal. The method may include processing the filtered PPG signal to detect troughs attributable to vasoconstriction. The method may include omitting one or more values of one or both of the first level features and the second level features associated with the detected vasoconstriction event from an evaluation (e.g., of a classifier). The method may include detecting an arrhythmia event from the data representing the signal from the PPG sensor. The output indication may be based on the detected arrhythmia event. Optionally, detecting the arrhythmia event may include detecting one or more irregular peak-peak intervals in the PPG signal associated with an arrhythmia. Optionally, the method may include omitting data corresponding to at least one of the one or more respiratory events from an evaluation (e.g., of a classifier), the at least one respiratory event may be associated with the detected one or more irregular peak-peak intervals. Optionally, detecting the one or more irregular peak-peak intervals may include calculating the absolute value of the difference between two adjacent peak-peak intervals. Detecting one or more irregular peak-to-peak intervals may include calculating an average of absolute values of differences between adjacent peak-to-peak intervals of the PPG signal. Detecting one or more irregular peak-to-peak intervals may include comparing the absolute value and the average value.
[0053] In some embodiments, the method may further include generating a signal for controlling the operation of the respiratory therapy device based on the output indication. Controlling the operation may include controlling the pressure or flow therapy of a blower of the respiratory therapy device. Optionally, receiving data representing the signal from the PPG sensor may include receiving the data at a server.
[0054] Some embodiments of the present technology may include a processor-readable medium having processor-executable instructions stored thereon, which, when executed by one or more processors, cause the one or more processors to detect a sleep disordered breathing event from a signal generated by a photoplethysmography (PPG) sensor. The processor-executable instructions may include instructions to receive data representing a signal from a PPG sensor. The processor-executable instructions may include instructions to detect a vasoconstriction event and / or an arrhythmia event from the data representing the signal from the PPG sensor. The processor-executable instructions may include instructions to evaluate the data representing the signal from the PPG sensor to generate an output indication identifying a sleep disordered breathing event based on the evaluation and the detected vasoconstriction event and / or the detected arrhythmia event.
[0055] In some embodiments, the processor-readable medium may include instructions to detect a vasoconstriction event, wherein the instructions to detect a vasoconstriction event may include instructions to detect a signal sag attributable to vasoconstriction. Optionally, the instructions to detect a signal sag may include instructions to filter PPG signal data representing a signal from a PPG sensor to produce a filtered PPG signal. The instructions to detect a signal sag may include instructions to process the filtered PPG signal to detect a trough attributable to vasoconstriction. The processor-readable medium may include instructions to evaluate data representing a signal from a PPG sensor, the instructions to generate an output indication based on the detected vasoconstriction event. To generate an output indication based on the detected vasoconstriction event, the instructions to evaluate data representing a signal from a PPG sensor may include instructions to ignore one or more values of one or more features calculated from the data associated with the detected vasoconstriction event.
[0056] In some embodiments, the processor-readable medium may include instructions for detecting an arrhythmia event, wherein the instructions for detecting an arrhythmia event may include instructions for detecting an irregular zone attributable to an arrhythmia. Optionally, the instructions for detecting an irregular zone attributable to an arrhythmia may include instructions for detecting one or more irregular peak-peak intervals in a PPG signal associated with an arrhythmia. The instructions for detecting one or more irregular peak-peak intervals may include instructions for calculating an absolute value of a difference between two adjacent peak-peak intervals. The instructions for detecting one or more irregular peak-peak intervals may include instructions for calculating an average of the absolute values of the differences between adjacent peak-peak intervals of the PPG signal. The instructions for detecting one or more irregular peak-peak intervals may include instructions for comparing the absolute value and the average. The processor-readable medium may include instructions for evaluating data representing a signal from a PPG sensor, the instructions generating an output indication based on the detected arrhythmia event. To generate an output indication based on the detected arrhythmia event, the instructions for evaluating data representing a signal from a PPG sensor may include instructions for ignoring at least one respiratory event detected from the data representing a signal from a PPG sensor that may be associated with the detected irregular zone. The processor executable instructions may also include instructions for generating a signal for controlling the operation of the respiratory therapy device based on the output indication. The controlling operation may include controlling the pressure or flow therapy of a blower of the respiratory therapy device. The instructions for receiving data representing the signal from the PPG sensor may include instructions for receiving the data at the server.
[0057] Some implementations of the present technology may include a server accessing any processor-readable medium as described herein, where the server may be configured to receive a request to download processor-executable instructions of the processor-readable medium to a processing device over a network.
[0058] Some embodiments of the present technology may include a processing device, which may include: one or more processors; and a processor-readable medium as described herein, or the processing device may be configured to access processor-executable instructions using a server as described herein. The processing device may be a respiratory therapy device. The processing device may be configured to generate a pressure therapy or a flow therapy.
[0059] Some embodiments of the present technology may include a method for a server to access any processor-readable medium described herein. The server method may include receiving a request at the server to download processor-executable instructions of the processor-readable medium to an electronic processing device over a network. The server method may include sending processor-executable instructions to the electronic processing device in response to the request.
[0060] Some embodiments of the present technology may include a method in which one or more processors detect a sleep disordered breathing event from a signal generated by a photoplethysmography (PPG) sensor. The method may include accessing any processor-readable medium described herein using one or more processors. The method may include executing processor-executable instructions of the processor-readable medium in one or more processors.
[0061] Some embodiments of the present technology may include a method of detecting a sleep disordered breathing event from a signal generated by a photoplethysmography (PPG) sensor. The method may include receiving data representing a signal from the PPG sensor. The method may include detecting a vasoconstriction event and / or an arrhythmia event from the data representing the signal from the PPG sensor. The method may include evaluating the data representing the signal from the PPG sensor to generate an output indication identifying a sleep disordered breathing event based on the evaluation and the detected vasoconstriction event and / or the detected arrhythmia event.
[0062] In some embodiments, the method may include detecting a vasoconstriction event, wherein detecting a vasoconstriction event may include detecting a signal dip attributable to vasoconstriction. Detecting signal dips may include filtering PPG signal data representing a signal from a PPG sensor to produce a filtered PPG signal. Detecting signal dips may include processing the filtered PPG signal to detect a trough attributable to vasoconstriction. The method may include evaluating data representing the signal from the PPG sensor to generate an output indication based on the detected vasoconstriction event. To generate an output indication based on the detected vasoconstriction event, evaluating data representing the signal from the PPG sensor may include ignoring one or more values of one or more features calculated from the data that may be associated with the detected vasoconstriction event.
[0063] In some embodiments, the method may include detecting an arrhythmic event, wherein detecting an arrhythmic event may include detecting an irregular zone attributable to an arrhythmia. Optionally, detecting an irregular zone attributable to an arrhythmia may include detecting one or more irregular peak-peak interval zones in a PPG signal associated with an arrhythmia. Detecting one or more irregular peak-peak interval zones may include calculating an absolute value of a difference between two adjacent peak-peak intervals. Detecting one or more irregular peak-peak interval zones may include calculating an average of the absolute values of the differences between adjacent peak-peak intervals of the PPG signal. Detecting one or more irregular peak-peak interval zones may include comparing the absolute value and the average value. The method may include evaluating data representing a signal from a PPG sensor to generate an output indication based on the detected arrhythmic event. To generate an output indication based on the detected arrhythmic event, evaluating the data representing the signal from the PPG sensor may include ignoring at least one respiratory event detected from the data representing the signal from the PPG sensor that may be associated with the detected irregular zone. The method may also include generating a signal for controlling the operation of a respiratory therapy device based on the output indication. The controlling operation may include controlling a pressure or flow therapy of a blower of the respiratory therapy device. Receiving data representing a signal from a PPG sensor may include receiving the data at a server.
[0064] Other features of the present technology will become apparent upon consideration of the information contained in the following detailed description, abstract, drawings, and claims. BRIEF DESCRIPTION OF THE DRAWINGS
[0065] The present technology is illustrated by way of example and not limitation in the figures of the accompanying drawings in which like reference numerals refer to similar elements, including:
[0066] Figure 1 A peripheral arterial tonometry (PAT) device 102 is shown that may be implemented for SA detection as described herein, according to examples of the present technology.
[0067] Figure 2 A graph comprising airflow and effort signals determined from a PSG (polysomnography) showing by comparison a typical central event and a typical obstructive respiratory event. Both events are accompanied by cessation of airflow, but a central event is characterized by a lack of respiratory effort, which is reflected in the flat region of the respiratory effort band signal.
[0068] Figure 3 is a graph of a segment of a PPG signal and its envelope showing the modulation of respiratory effort.
[0069] Figure 4 Included are two graphs showing the respiratory effort signal extracted from the finger PPG and the respiratory effort channel of the PSG by comparison.
[0070] Figure 5 An event marking procedure is shown.
[0071] Figure 6 Included are two graphs showing a PPG signal (top graph) and a respiratory effort signal derived from the PPG signal (bottom graph).
[0072] Figure 7 Included are three graphs showing a PPG signal (top graph), a respiratory effort signal derived from the PPG signal (middle graph), and a respiratory effort signal from PSG (bottom graph).
[0073] Figure 8 is a graph of sensitivity and specificity data for an example embodiment of the present technology.
[0074] Fig. 9 Included are scatter plots of central sleep apnea predicted by example embodiments of the present technology and expert labeling of such events by two different expert groups.
[0075] Fig.10 Negative predictive value (NPV) and positive predictive value (PPV) and their sum are shown for various central apnea-hypopnea index (cAHI) cut-off thresholds for screening patients.
[0076] Fig.11 is a table of demographic and clinical characteristics of a dataset involved in training example embodiments of the present technology.
[0077] Fig.12 is a table of performance endpoints for example implementations of the present technology.
[0078] Fig.13 is a confusion matrix illustrating the classification performance of an example of the present technique.
[0079] Fig.14A are example methods by which one or more processors in a system such as the present technology may detect or distinguish different types of sleep disordered breathing events.
[0080] Fig. 14B
[0066] Another example method by which one or more processors in a system such as the present technology may distinguish between different types of sleep disordered breathing events.
[0081] Fig. 14C
[0013] An example method in which one or more processors in a system such as the present technology detects (such as by distinguishing between different types) sleep disordered breathing events based on detection of arrhythmic events and / or detection of vasoconstrictive events.
[0082] Fig.15AA system in accordance with the present technology is shown. A patient 1000 wearing a patient interface 3000 receives a supply of pressurized air from an RPT device 4000. Air from the RPT device 4000 is humidified in a humidifier 5000 and passed along an air circuit 4170 to the patient 1000. A bed partner 1100 is also shown.
[0083] Fig. 15B An RPT device 4000 is shown being used by a patient 1000 wearing a nasal mask 3000 .
[0084] Fig. 15C An RPT device 4000 is shown in use by a patient 1000 wearing a full face mask 3000 .
[0085] Fig.16 A non-invasive patient interface 3000 is shown in the form of a nasal mask.
[0086] Fig.17A A RPT device 4000 according to one form of the present technology is shown.
[0087] Fig. 17B A schematic diagram of a pneumatic circuit for an RPT device 4000 according to one form of the present technology is shown. The upstream and downstream directions are indicated.
[0088] Fig. 17C A schematic diagram of the electrical components of an RPT device 4000 in accordance with one aspect of the present technique is shown.
[0089] Fig.17D 4300 is a schematic diagram of an algorithm 4300 implemented in an RPT device 4000 according to an aspect of the present technique. Fig.17D , solid arrows represent actual information flow, for example via electronic signals.
[0090] Fig.17E is shown according to one form of the present technology by Fig.17D Flowchart of method 4500 performed by therapy engine module 4320.
[0091] Fig.18 A humidifier 5000 is shown. DETAILED DESCRIPTION
[0092] Before further describing the present technology in detail, it should be understood that the present technology is not limited to the specific examples described herein, which may vary. It should also be understood that the terms used in the present invention are only for describing the specific examples discussed herein and are not intended to be limiting.
[0093] The following description is provided about each example that can share one or more common characteristics and / or features. It should be understood that one or more features of any one example can be combined with one or more features of another example or other examples. In addition, in any one of the examples, any single feature or combination of features can constitute another example.
[0094] The present technology relates to processes for detecting or distinguishing central and obstructive respiratory events from data representing signals from a finger photoplethysmographic sensor. The methods described can be incorporated into systems, such as for minimally invasive central sleep apnea screening, and provide improved devices, such as for detecting, screening, monitoring, diagnosing or treating such sleep-related breathing problems.
[0095] The PAT sensing device 102 utilizes a fingertip mounted PPG probe
[20] . PPG uses optical technology to detect changes in blood volume in the microvascular bed of tissue
[23] ,
[24] . PPG measurements are used to derive changes in arterial oxygen saturation (SpO2), pulse rate (PR), and peripheral arterial tone, which are then used to detect respiratory events
[20] . Peripheral arterial tone (PAT) refers to the tension of the smooth muscle tissue of the peripheral arteries. When the muscle tension of the peripheral arteries increases, the diameter of the arteries decreases, resulting in reduced perfusion, thereby reducing the pulsatile blood volume in the peripheral tissues. The latter is picked up as a decrease in the PPG signal swing between systole and diastole. The PAT signal can be derived from the PPG signal from the PPG sensor, such as by the method described in PCT / EP2021 / 067532, the disclosure of which is incorporated herein by reference. The PPG derived signal is called the PAT signal and can be derived by such a trend of decreased pulsatile blood volume. The PAT sensing device analyzes the concurrence of a decrease in SpO2, a surge in pulse rate, and an increase in peripheral arterial tone. As airflow decreases, oxygen supply to the lungs decreases from baseline, resulting in a significant drop in SpO2. Near the end of a respiratory event, sympathetic nervous system activity surges to awaken the patient, resulting in restoration of ventilation, a surge in pulse rate, and the release of norepinephrine into the bloodstream. Norepinephrine then binds to α-adrenaline receptors that innervate arterial smooth muscle tissue in the fingers, causing a sudden vasoconstriction, or synchronous increase in PAT picked up by the PPG sensor
[20] .
[0096] Such a PPG sensor may be implemented for detection of respiratory events, such as by detecting signal features indicative of apnea and / or hypopnea events within a peripheral arterial tonometry (PAT) signal that may be derived from the sensor's PPG signal. As discussed in greater detail herein, the PPG signal may also be processed to extract respiratory effort related information from which signal processing, computational, and / or classification techniques may be applied to detect or distinguish between central and obstructive types of respiratory events.
[0097] Thus, as described in more detail in the context of the examples herein, any one or more of the following techniques may be applied to signals from a PPG sensing device. Fig.14A As shown, some embodiments may include a method 13000 of one or more processors distinguishing sleep disordered breathing events from a signal generated by a photoplethysmography (PPG) sensor. The method may include, at process 13010, receiving data representing a signal from a PPG sensor. The method may include, at process 13040, deriving a PPG-based breathing signal (e.g., a breathing effort signal) from the data representing the signal from the PPG sensor. The method may include, at process 13050, calculating first-level features from the PPG-based breathing signal. The method may include, at process 13060, calculating second-level features from the first-level features. The method may include, at process 13070, evaluating the first-level features and / or the second-level features, such as in a classifier. The method may include, at process 13080, generating an output indication identifying a type of sleep disordered breathing based on an evaluation, such as by a classifier. This type identification may be based on respiratory events or based on sleep sessions. For example, the type indication may be for central or obstructive apnea events, or for a certain interval, such as a session (e.g., such as a night of sleep or several nights of sleep wearing a screening device), which may be characterized as a session affected by central sleep apnea. At 13090, the method may also include generating an output using or based on the generated indicator and / or evaluation, such as by comparing the indicator to a threshold value. Such output may include electronic communication or transmission of the indicator and / or control instructions for operating the therapy device, such as settings for the device.
[0098] As another example, Fig. 14BAs shown, some embodiments may include one or more processors for detecting sleep disordered breathing events from a signal generated by a photoplethysmography (PPG) sensor, such as by differentiation, method 14000. The method may include, at process 14010, receiving data representing a signal from a PPG sensor. The method may include, at process 14020, deriving a peripheral arterial tonometry (PAT) signal from the data representing the signal from the PPG sensor. The method may include, at process 14030, detecting one or more respiratory events based on the PAT signal. The method may include, at process 14040, deriving a PPG-based respiratory signal from the data representing the signal from the PPG sensor. The method may include, at process 14050, calculating first-level features from the PPG-based respiratory signal. The method may include, at process 14060, calculating second-level features from the first-level features. The method may include, at process 14070, evaluating the first-level features and / or the second-level features, such as in a classifier. The method may include, at process 14080, generating an output indication identifying a sleep disordered breathing type for each of the one or more detected respiratory events based on an evaluation, such as by a classifier. At 14090, the method may also include generating an output using or based on the generated indicator and / or evaluation, such as by comparing the indicator to a threshold value. Such an output may include electronic communication or transmission of the indicator and / or control instructions for operating the therapy device, such as settings of the device.
[0099] As another example, Fig. 14C As shown, the processing of the present technology may include a method 15000 for detecting sleep disordered breathing events (e.g., one or more) from a signal generated by a photoplethysmography (PPG) sensor. The method, at 15010, receives data representing a signal from a PPG sensor. The method may include, at 15025, detecting a vasoconstriction event and / or an arrhythmia event from the data representing the signal from the PPG sensor. The method may include, at 15070, evaluating the data representing the signal from the PPG sensor to generate an output indication of identifying a sleep disordered breathing event based on the detected vasoconstriction event based on the evaluation and the detected vasoconstriction event and / or the detected arrhythmia event. At 15090, the method may also include generating an output using or based on the generated indicator and / or the evaluation, such as by comparing the indicator to a threshold value. Such an output may include an electronic communication or transmission of an indicator and / or a control instruction for operating a therapy device, such as a setting of the device.
[0100] Although such processes are shown in a sequential order, such order is not strictly required, such as where some processes can be performed in parallel or in a different order, or where certain aspects are performed without other aspects.
[0101] Such techniques / processes may be performed by one or more processors or other processing devices 104 or computers, such as a cloud-based processing system (e.g., one or more servers) or other processing devices that communicate with or receive (directly or indirectly) signals or data from the PAT sensing device 102, so as to function as the system 100. Such processing devices 104 or devices may include, for example, a controller, a central controller, or other processor described herein, such as a processor of an RT or RPT, or a remote processing device or server in communication with such a controller. Thus, the PPG-based determination using the sensing device may be used as a basis for controlling or setting the operation of the RT device (locally or remotely), such as making therapy changes (e.g., control of pressure or flow therapy), as discussed in greater detail herein. Thus, the techniques of one or more methods described herein may be implemented as a computer program or processor control instructions stored in a non-transitory computer-readable storage medium, such as a memory accessible by a processing device described herein. In addition, as described herein, the characteristics of the signal may be evaluated in a machine algorithm or process, which may be a machine learning process of one or more processors, such as a classifier. However, such processor-based evaluation may also or alternatively be implemented as a rule-based logic decision algorithm or other computer evaluation involving a logic reasoning engine configured to evaluate the sensor data or features calculated therefrom using appropriate thresholds. Such thresholds may be determined through empirical analysis or study.
[0102] Details of example aspects of such methods may be considered further herein.
[0103] i. Extracting breathing signal from raw PPG
[0104] As mentioned earlier, the PPG signal measured at the fingertips contains respiratory information. This information is present because changes in chest pressure throughout the respiratory cycle affect blood flow to the extremities of the body
[20] . Therefore, the PPG signal amplitude oscillates synchronously with the respiratory cycle. This amplitude modulation can be isolated to retain the signal representing respiratory effort. Figure 3 The PPG signal and the respiratory modulations present in the signal are shown. The present technique can utilize the presence or absence of these respiratory modulations in the finger PPG to infer the type of respiratory event.
[0105] To this end, respiratory modulation can be extracted from the raw PPG signal by obtaining or generating an envelope of the signal. For example, the raw PPG signal can be filtered at a suitable frequency (e.g., 0.15 Hz), such as by a high-pass filter, to remove slowly varying baseline modulation. The respiratory rate at rest is typically 0.2 Hz or higher (normal respiratory rate for adults ranges from 0.2-0.33 Hz), and therefore, respiratory modulation is substantially unaffected by such low-end filtering. Subsequently, the peak envelope of the filtered PPG signal can be calculated, such as using a peak detector or other envelope detection method. The envelope signal can optionally be normalized by subtracting a moving average window of the envelope (e.g., a 6 second average) and dividing it by a moving window interquartile range window (e.g., a 30 second moving window interquartile range window). The initial moving average window (e.g., a 6 second window) can help remove signal baseline fluctuations while retaining respiratory modulation. The latter moving window (e.g., a 30 second window) can help normalize any steady-state changes in the signal amplitude. The resulting signal, which may be referred to as a PPG-derived respiratory effort (PPGDR) signal, may then be further analyzed or processed, such as for classification by a classifier, such as by determining or calculating a classification feature from the PPGDR signal. Figure 4 A comparison of a segment of PPGDR with the corresponding respiratory effort segment from the chest effort band of the PSG is shown.
[0106] ii. Feature calculation
[0107] Therefore, after the PPGDR signal is extracted, features can be extracted from the PPGDR signal, for example, by using one or more moving windows or sliding window operations. Such a "window" in signal processing generally involves a sample subset of a time-based signal, and different windows may include different subsets of samples taken from a time-based signal. For example, two window sizes with shorter and longer durations (e.g., 11 seconds and 21 seconds) can be implemented. The window can be shifted stepwise (such as a step of 1 second) on or along the PPGDR signal. Then, the first-level feature values can be calculated for the signal samples of each window so as to attribute them to a one-second interval at the center of each window. This results in a set of first-level feature values for each first-level feature derived from a shorter window and a set of first-level feature values for each first-level feature of a longer window.
[0108] The first level features calculated for each window may include, for example, any one or more of the range, interquartile range, variance, and number of PPGDR local maxima (peaks). Optionally, the range of the signal may also be calculated for other window sizes such as 5 seconds and 7 seconds.
[0109] For local maximum related features, local maximum detection may be performed on the PPGDR signal (the minimum peak-to-peak distance may be set, for example, to about 250 milliseconds), where the peak-to-peak interval represents the duration of the respiratory cycle. The absence of a local maximum indicates a lack of respiratory effort.
[0110] Then the second level features can also be calculated, such as by deriving their values from the first level features, or selecting a subset of these values from the first level features. For example, the second level features can be determined so that they are related to a specific respiratory event detected in the signal from the PAT sensing device. In this regard, the feature design can take into account the variability of the duration and position of the flat zone in the PPGDR signal during the respiratory event. Therefore, for each respiratory event detected by evaluating the PAT sensing device signal (such as by determining the detection method of the flat zone in the PAT signal), a moving window feature value or a first level feature value can be selected, for which the center second overlaps the respiratory event period detected in time. That is, the second level features can be determined from the first level feature values derived from the window overlapping with the respiratory event. Calculate / determine the second level features from these selected first level feature values. Such second level features can include, for example, the minimum value of the selected first level feature value, the average value of the lowest three values of the selected first level feature value, and any one or more of one or more percentiles of the selected first level feature value such as the 10th and / or 25th percentile. This results in the second level feature value of each first level feature.
[0111] For local maximum related features, the aggregation operation includes calculating the minimum value and the number and proportion of windows with 0, 1 and 2 peaks. The resulting aggregated feature set for each respiratory event can be used to support the subsequent steps described in this article.
[0112] iii. Signal Rejection
[0113] In embodiments such as using only the PPG signal to extract the PPGDR signal, the quality of the type determination (e.g., classification) depends largely on the underlying quality of the PPG data. Strong vasoconstriction reflected by a sudden drop in the amplitude of the PPG pulse often contains low signal-to-noise ratio episodes and cannot be adequately offset by normalization operations. This significantly reduces the PPGDR quality. Therefore, the system can be configured to identify vasoconstriction to improve the quality of sleep disordered breathing event detection and / or type differentiation, so that sleep disordered breathing event detection and / or type differentiation is based on vasoconstriction event detection. For example, vasoconstriction events can be detected from the signal from the sensor by detecting signal dips in the PPG and / or PAT signal, such as by detecting / identifying troughs (periods or regions affected by vasoconstriction) that are significantly prominent in a filtered form of the PPG signal. Such filtering of the PPG signal can be performed to remove baseline offsets and perfusion offsets that may otherwise affect trough detection. Figure 6 An example of such a vasoconstriction affected area is shown. Alternatively, such a period or area may be detected by detecting a signal drop period in a PAT signal derived from a PPG signal.
[0114] Other methods for detecting such signal dips may also, or alternatively, be implemented for such vasoconstriction event detection. For example, a PAT or PPG signal may be processed to detect a "V" shape, which may be characterized by detecting decreasing and increasing slopes in the signal or its baseline having a central minimum. Such characterization may be implemented, for example, with curve fitting or best fit techniques, such as with least squares and / or linear regression methods and / or correlations for finding a curve in the signal that correlates to a predetermined curve function of an expected shape (e.g., a "V" shape). Other methods may include filtering, such as detecting peak-to-trough trends or amplitude trends, which may indicate a sudden signal drop. In some embodiments, the signal may be processed to track peaks and / or troughs in the signal envelope for detecting such sudden dips.
[0115] Such periods of vasoconstriction-affected time periods detected in the PAT signal or PPG signal, such as when identified by a trough detection method, can be ignored by the system. Therefore, the corresponding periods of PPGDR extracted from such vasoconstriction-affected regions can be omitted from the above-mentioned sleep disordered breathing detection or differentiation process (e.g., classification). For example, if a respiratory event includes a vasoconstriction-affected region, features (or their values) derived from that region portion of the PPGDR signal can be omitted to compile an aggregated feature set for respiratory event classification.
[0116] iv. Event Rejection
[0117] Another problem with the quality of the PPGDR signal occurs when motion or other artifacts affect the quality of the PPG or when severe arrhythmias are present. Therefore, the system can be configured to identify arrhythmias to improve the quality of sleep disordered breathing event detection and / or type differentiation, so that sleep disordered breathing event detection and / or type differentiation is based on arrhythmic event detection. For example, arrhythmic events can be detected from the signal from the sensor by detecting irregular periods in the PPG and / or PAT signals that indicate signal fluctuations of arrhythmias. For example, when the peak-to-peak intervals of the PPG are irregular due to artifacts or severe arrhythmias, the respiratory modulation present in the PPG will be severely affected by such irregularities, and the PPGDR extracted from these areas will show false oscillations that are unrelated to the true respiratory modulation.
[0118] To ensure that PPGDR can be used for respiratory event type prediction, irregular PPG peak-to-peak interval regions can be detected, and any respiratory event with more than a certain percentage (e.g., 20%) of irregular PPG peak-to-peak intervals relative to the total number of PPG peak-to-peak intervals of respiratory events can be rejected.
[0119] In some embodiments, the system can identify irregular peak-to-peak intervals when the absolute value of the difference between two adjacent peak-to-peak intervals exceeds a certain percentage (e.g., 20%) of the average of those values for other intervals of the PPG signal. Thus, in an example, if more than 20% of the respiratory events have PPG peak-to-peak intervals that are irregular, then respiratory events can be rejected. Figure 7 An example of a segment of such a rejected PPG signal and the corresponding PPGDR compared to the respiratory effort channel of the PSG is shown. The proportion of rejected respiratory events per recording can be understood as the Event Rejection Ratio (ERP).
[0120] Other methods for detecting such irregular periods of depression of the PPG signal may also, or alternatively, be implemented for such arrhythmia event detection. For example, the PPG signal may be processed to detect other characteristic points of cardiac information in the PPG signal, such as determining such intervals on a point-to-point basis, rather than relying on the peak value (i.e., peak-to-peak) of the signal for interval detection. For example, such characteristic points may be troughs, such that trough-to-trough intervals may be determined and applied to the irregular interval detection method described previously. Another example characteristic point may be a dicrotic notch, such that notch-to-notch intervals may be implemented in the method. In some versions, detection of such intervals may be improved by preprocessing the PPG signal for detection. For example, the PPG signal may be filtered, such as by bandpass filtering, and characteristic point-based intervals may be detected from the preprocessed signal or the filtered signal.
[0121] v. Machine breathing event type recognition
[0122] The aggregate feature set can then be applied to a machine algorithm as described above, such as a machine learning algorithm or a trained classifier. Examples of this embodiment may include a tree classifier set, a neural network, a support vector machine, or other machine learning algorithms / models. The output of the classifier may then include a type identification that distinguishes central respiratory events and obstructive respiratory events, such as by generating a suitable representative label that designates a respiratory event as central or obstructive. Such event labels may be applied by the system, such as for screening, monitoring, diagnosis, and / or treatment. For example, the system may estimate or predict a central apnea hypopnea index (cAHI) by counting the number of central events predicted by a computer model / classifier and dividing the count by an estimated value of total sleep time (TST), which may also be determined by a PAT sensing device. In some versions, if only non-rejected respiratory events are used for prediction, the estimated or predicted cAHI may be inferred by dividing it by 1 minus ERP to take into account the total number of respiratory events (rejected and non-rejected) of the patient. Thus, in the case of a screening application, the estimated or predicted cAHI can then be compared to a cutoff threshold to provide an indication of whether the user should receive further care (e.g., monitoring, further screening, or therapy) when the cAHI indicates that the user is experiencing significant central sleep apnea that may require additional intervention.
[0123] In the case of training a computer model or classifier using patient data, expert analysis event type labels (central vs. obstructive) can be used to mark respiratory events that can be inferred by detecting flat areas in the patient's PAT sensing device signal. Such respiratory events identified by expert review or identified with a high degree of certainty can be used as labeled events (ground truth events) for training any suitable classifier.
[0124] vi. Statistical performance analysis
[0125] A. Dataset-Example Validation
[0126] Various aspects of the above techniques were implemented, evaluated and validated in the study. Respiratory information in the PPG data embedded in the dataset was extracted and used to train a set of tree classifiers that predict the centrality or obstructiveness of each respiratory event. Classifier performance was evaluated using leave-one-patient cross-validation, where expert analysis of the PSG was used as the ground truth. A second independent analysis of the PSG was also evaluated against the ground truth to allow benchmarking of the PPG-based approach.
[0127] The dataset used in this study included 266 patients with suspected SA who were prospectively recruited at four different centers, three in the United States (all within the United Health Services Group, Miami, FL) and one in Belgium (Ziekenhuis Oost Limburg, ZOL, Genk). The U.S. arm of the study was approved by the Aspire IRB, part of the WIRB-Copernicus Group. The Belgian arm of the study was approved by the ZOL Ethics Committee.
[0128] All patients were scheduled for an overnight in-laboratory PSG. The CSA prevalence of the patient cohort was not stratified to maintain the normal cohort prevalence of CSA. Due to the expected target population of the cohort matching research device, this resulted in the most representative prevalence-dependent performance endpoints, such as negative predictive value (NPV) and positive predictive value (PPV). Qualified laboratory technicians at each participating research center were responsible for setting up the equipment and capturing PSG data. During the setup of PSG, the PAT HSAT was attached to the middle finger of the hand where the PSG pulse oximeter was applied.
[0129] Patient refusal
[0130] Because of the small proportion of retained events used for cAHI estimation, predictions made for patients with a large number of rejected events would be unreliable. Therefore, patients with ERPs above 0.3 were rejected and removed from further analysis.
[0131] All PSG data were scored in duplicate by two independent centers blinded to each other. The first scoring was performed by a team of sleep technicians at the center where the patient was admitted (local analysis). Another independent scoring was performed by scorers from Cerebra Medical (CM, Canada). These studies were first analyzed using their Michele Sleep Scoring System (MSSS) and subsequently supplemented by expert technicians using complete manual scoring. Malhotra et al.
[25] confirmed in a multicenter trial that the MSSS supplemented with manual edits by expert scorers was more robust than the results of a single scorer. Because of this conclusion, CM's analysis was used as an expert benchmark (expert analysis) to compare with local analysis and PPG-based methods. All PSG data were scored according to the latest AASM scoring rules using the recommended hypopnea score 1A (3%) rule
[26] .
[0132] In the model training of the study, each respiratory event inferred by PAT HSAT is marked using expert analysis event type labels (central vs. obstructive). Due to the fact that for model training, we only want to retain the respiratory events identified with a relatively high level of certainty, so the respiratory events identified by expert analysis but not identified by local analysis are removed. The respiratory events that experts and local analysis agree on for their existence (but not necessarily type) are used to mark PAT HSAT events (basic fact events). PAT HSAT events that do not overlap with any basic fact events or within 10 seconds after any basic fact events are removed from the training data. This removal of expert analysis or PAT HSAT events is performed only for training sets rather than model validation. Each expert analysis annotation is used for PAT HSAT event marking at most once.
[0133] Figure 5 The marking procedure is shown by means of an example. If a PAT HSAT event has a unique overlapping expert analysis annotation, the PAT HSAT event is marked according to the corresponding expert analysis event type. If there is more than one annotation event overlapping with the PAT HSAT event, if any overlapping expert analysis annotation is a central event, the PAT event is marked as central, otherwise the respiratory event is marked as obstructive. Since the PAT HSAT respiratory event position may be slightly offset relative to the expert analysis annotation, if no other annotation overlaps with the PAT HSAT event, any previous expert analysis annotation up to 10 seconds is matched with the PATH SAT event. Each expert analysis annotation is used for PAT HSAT event marking at most once. In a total of 44,420 PAT HSAT events, 26,208 have corresponding expert analysis annotations and are marked according to the above procedure. In the 26,208 marked PAT HSAT events, 2,881 are central (or mixed), and 23,327 are obstructive.
[0134] Since airflow channels are necessary to mark the accurate start and end of respiratory events, PAT HSAT has imperfect boundaries for respiratory events. Therefore, it is not easy to infer via PAT HSAT whether an episode without respiratory effort is followed by an episode with respiratory effort, corresponding to a central event that may cover a mixed event. For this reason, the expert analysis labels are binarized into centrality and obstructiveness, and mixed events are considered centrality. Therefore, the cAHI of the reported implementation method is an estimate of the sum of central and mixed respiratory events per hour of sleep.
[0135] vii.cAHI prediction and assessment
[0136] Different from training, for evaluation, all 44,420 PAT HSAT events were retained, regardless of whether there was a corresponding PSG annotation. In order to maximize the utilization of the strongly unbalanced data set caused by the low prevalence of CSA, training and evaluation were performed via patient leave-one-out cross validation. For each patient, a tree classifier ensemble was trained on the labeled data of the remaining patients (ensemble method: bagging, false positive cost: 2.5, number of trees: 30, maximum number of splits: 150). The false positive cost parameter of the classifier is used to adjust the predicted event sensitivity and specificity, in this case to prevent over-detection of CSA. Using this classifier, the non-rejected PAT HSAT events of the considered patient are predicted. The predicted cAHI is calculated as the number of central events predicted by the model divided by the PAT HSAT total sleep time (TST) estimate. Since only non-rejected events are used for prediction, the predicted cAHI is inferred by dividing it by 1 minus ERP to take into account the total number of respiratory events (rejected and non-rejected) of the patient.
[0137] The performance of the method was evaluated for cAHI cutoffs of 5, 10 and 15. The evaluation was performed on a patient basis by labeling the patient as central if the cAHI was above a given cutoff.
[0138] i. General Concepts
[0139] Statistical analysis was performed using MATLAB (Version 2020a, MathWorks, USA). Patient demographic information was obtained during the clinical study and reported in the results. For cAHI prediction, Pearson correlation with the ground truth was calculated. For each cAHI cutoff, cohort sensitivity, specificity, NPV and PPV, accuracy, and Cohen's Kappa values were calculated, and a receiver operating characteristic (ROC) curve was generated. The ROC curve was generated by obtaining the classifier probability score for each respiratory event and changing the probability cutoff threshold for binary classification. For all sensitivities, specificities, PPVs, and NPVs, accuracy and Cohen's Kappa endpoint parameters and 95% confidence intervals were calculated. Construct a classifier confusion matrix. Confusion matrices were constructed for four cAHI categories: cAHI <5, 5≤cAHI<10, 10≤cAHI<15, and cAHI≥15. A significance level of α (p value) of 0.05 was determined.
[0140] Where possible, the same analysis was performed to compare local and expert analyses.
[0141] ii. Endpoint Analysis
[0142] Since there are no standardized performance targets established for cAHI prediction, the performance of local analysis relative to expert analysis was used as a performance benchmark. In agreement with Pillar et al.
[21] , cAHI cutoffs of 10 and 15 were used for performance analysis. AHI cutoffs of 5 and 30 are also standard in the literature, so performance analysis for a cAHI cutoff of 5 is also reported.
[0143] As there were only 6 patients with cAHI ≥ 30, no analysis was performed for this cutoff value because the performance measure is difficult to generalize due to the small sample size.
[0144] Endpoints used for performance evaluation were sensitivity, specificity, NPV, PPV, accuracy, and Cohen's Kappa of the PPG-based method at the cAHI cutoffs examined.
[0145] II. Results
[0146] A. Patient refuses
[0147] After removing patients with an ERP above the 0.3 cutoff from further analysis, 245 of the original 266 patients remained. All statistical analyses and evaluations were performed on the final group of patients accepted.
[0148] B. Demographic Information
[0149] As listed in Table I, patients were predominantly male (60%), middle-aged (mean 53.9 years, STD 13.7), and overweight (mean BMI 29.9 kg / m2 and STD 5.9). The mean AHI was 31.1 (STD 24.8).
[0150] Twenty-four patients had no clinical sleep apnea (AHI < 5), 56 patients had mild sleep apnea (5 < AHI < 15), 64 patients had moderate sleep apnea (15 < AHI < 30), and 101 patients had severe sleep apnea (AHI > 30).
[0151] 213 patients had a cAHI<5, 11 patients had 5≤cAHI<10, 10 patients had 10≤cAHI<15, 5 patients had 15≤cAHI<30, and 6 patients had a cAHI≥30.
[0152] C. Queue Performance
[0153] Table II shows the sensitivity, specificity, NPV, PPV, accuracy, and Cohen's Kappa of the PPG-based approach for three cAHI cutoffs relative to the expert analysis. The same is shown for the local analysis relative to the expert analysis. For a cAHI cutoff of 10, the approach achieved 81% sensitivity, 99% specificity, 90% PPV, 98% NPV, 97.6% accuracy, and a Cohen's Kappa of 0.84.
[0154] Figure 8 The ROC curves for each cAHI cutoff are shown. The maximum AUC of our method was found to be 0.98 at a cAHI cutoff of 15.
[0155] Fig. 9 Scatter plots of cAHI predicted by the PPG-based method and local analysis cAHI relative to expert analysis cAHI are shown. Pearson correlation was 0.81 for PPG-based prediction (p-value < 0.001) and 0.72 for local analysis (p-value < 0.001).
[0156] Table III shows the confusion matrix for PPG-based predictions for the four cAHI intervals and the same analysis for the local analysis.
[0157] The 4-level accuracy of cAHI predictions derived from this matrix was 89.4% for PPG-based predictions and 89.0% for local analysis. Cohen's Kappa was 0.573 for PPG-based predictions and 0.544 for local analysis.
[0158] III. Discussion
[0159] A. General Concepts
[0160] The results show that the example method described above allows accurate cAHI prediction based on PPG data obtained with a finger probe, and does so without the need for any additional sensor modality, such as the chest probe described by Pillar et al.
[21] .
[0161] B. PAT HSAT performance
[0162] The performance of this embodiment was found to be comparable to local analysis of simultaneously acquired PSGs.
[0163] Another comparison could be made between the automated method and the performance of the PAT HSAT cAHI assay described by Pillar et al.
[21] (note that, as mentioned previously, this method combines PPG and chest accelerometery modalities). The study population described by Pillar et al.
[21] was preselected for CSA. Therefore, the PPV and NPV endpoints could not be compared because of the large difference in prevalence between their study and ours.
[0164] The method described by Pillar et al.
[21] achieved a sensitivity of 71.4% and a specificity of 98.6% for a cAHI cutoff of 10. It achieved a sensitivity of 66.7% and a specificity of 100% for a cAHI cutoff of 15. In comparison, the performance endpoints of the automated methods listed in Table II show excellent performance for the sum of sensitivity and specificity for both cAHI cutoffs.
[0165] C. Selection of cAHI cutoff for CSA screening; clinical implications and trade-offs
[0166] The choice of the most desirable cAHI cutoff for CSA screening should consider the trade-off between NPV and PPV. Since patients flagged by automated methods can be referred for a full in-laboratory PSG, having a high PPV is important to avoid unnecessary follow-up examinations, whereas a high NPV is required to avoid false-negative tests.
[0167] Fig.10 PPV, NPV and their sum for varying cAHI cutoffs are shown. For cAHI cutoffs between 2.5 and 15, NPV remains relatively stable at about 95%, reaching a maximum of 99.1% at a cAHI cutoff of 13.5. PPV increases from 57.8% at a cAHI cutoff of 2.5 to 94.7% at a cAHI cutoff of 9.5. At a cutoff of 10.5, it begins to decrease approximately monotonically. The cAHI cutoff corresponding to the highest NPV and PPV sum of 19.3 is at a cAHI cutoff of 10.5.
[0168] When choosing a cAHI, in addition to maximizing NPV and PPV, another consideration is choosing the desired sensitivity to detect a plausible CSA advantage. Major CSA occurs when 50% or more of events are central in nature. Therefore, if we wish to detect patients with major CSA with an AHI as low as 5, we would need to set the cAHI cutoff to 2.5. If we wish to detect only patients with major CSA with an AHI of 15 or greater, we would need to set the cAHI cutoff to 7.5. An optimal cAHI cutoff would be around 10 to flag major CSA in patients with an AHI of 20 or greater. Given the above, Fig.10The figure in reveals the clinician's flexibility in choosing a cAHI cutoff based on this desired minimum AHI that can flag major CSA. Within the range of cAHI cutoffs between 2.5 and 10, lower cAHI cutoffs were generally associated with lower PPVs and therefore higher rates of unnecessary follow-up examinations.
[0169] C. Conclusion
[0170] In summary, this study validated the disclosed exemplary aspects of the method for detecting CSA in SA patients using only PPG data acquired at the finger. The performance endpoints of the method at the examined central apnea-hypopnea index (cAHI) cutoff levels of 5 and 10 were higher than or equal to the PSG benchmark, and were not significantly lower than the PSG benchmark at the cAHI cutoff level of 15. The method achieved 81% sensitivity, 99% specificity, 90% positive predictive value, and 98% negative predictive value at a cAHI cutoff of 10 events per hour of sleep. Therefore, the PPG-based CSA detection method can, for example, be used to mark the patient's CSA, which can enable referral to an in-laboratory PSG for further confirmation. The method can accurately mark the significant presence of CSA, which in turn can help make more optimized therapy decisions.
[0171] 5.1 Optional Example Processing System
[0172] As previously described, in one form, the present technology may include an apparatus or device for treating and / or monitoring respiratory disorders, which may participate in and / or be configured for any of the aforementioned processes. The apparatus or device may be an RT or RPT device 4000 for supplying a pressurized air flow to a patient 1000 via an air circuit 4170 leading to a patient interface 3000. The air flow may be pressure controlled (for respiratory pressure therapy) or flow controlled (for flow therapy such as high flow therapy HFT). Therefore, the RPT device may also be configured to be used as a flow therapy device. In the following description, reference may be made to Figure 7-10 Consider RT or RPT devices.
[0173] 5.2 Patient Interface
[0174] A non-invasive patient interface 3000 according to one aspect of the present technology includes the following functional aspects: a seal-forming structure 3100, an inflatable chamber 3200, a positioning and stabilizing structure 3300, a vent 3400, a connection port 3600 for connecting to an air circuit 4170, and a forehead support 3700. In some forms, the functional aspects may be provided by one or more physical components. In some forms, one physical component may provide one or more functional aspects. In use, the seal-forming structure 3100 is arranged to surround the entrance of the patient's airway to facilitate the supply of pressurized air to the airway.
[0175] 5.3RPT device
[0176] The RPT device 4000 according to one aspect of the present technology includes mechanical and pneumatic components 4100, electrical components 4200, and is programmed to execute one or more algorithms 4300. The RPT device 4000 can have an outer housing 4010 formed in two parts: an upper portion 4012 and a lower portion 4014. In one form, the outer housing 4010 can include one or more panels 4015. The RPT device 4000 can include a chassis 4016 that supports one or more internal components of the RPT device 4000. The RPT device 4000 can include a handle 4018.
[0177] The pneumatic path of the RPT device 4000 may include one or more air path items, such as an inlet air filter 4112, an inlet muffler 4122, a pressure generator 4140 capable of supplying pressurized air (e.g., a blower 4142), an outlet muffler 4124, and one or more transducers 4270, such as a pressure sensor 4272 and a flow sensor 4274.
[0178] One or more of the air path items may be located within a removable unitary structure, which will be referred to as a pneumatic block 4020. The pneumatic block 4020 may be located within the outer housing 4010. In one form, the pneumatic block 4020 is supported by or formed as part of the chassis 4016.
[0179] The RPT device 4000 may have a power source 4210, one or more input devices 4220, a central controller 4230, a therapy device controller 4240, a pressure generator 4140, one or more protection circuits 4250, a memory 4260, a transducer 4270, a data communication interface 4280, and one or more output devices 4290. The electrical components 4200 may be mounted on a single printed circuit board assembly (PCBA) 4202. In an alternative form, the RPT device 4000 may include more than one PCBA 4202.
[0180] 5.3.1 Mechanical and pneumatic components of RPT device
[0181] The RPT device 4000 may include one or more of the following components in an integral unit. In the alternative, one or more of the following components may be located as respective separate units.
[0182] 5.3.1.1 Air filter
[0183] An RPT device 4000 according to one form of the present technology may include an air filter 4110, or multiple air filters 4110.
[0184] In one form, the air inlet filter 4112 is located at the beginning of the pneumatic path upstream of the pressure generator 4140 .
[0185] In one form, an air outlet filter 4114 (eg, an antimicrobial filter) is located between the outlet of the pneumatic block 4020 and the patient interface 3000 .
[0186] 5.3.1.2 Silencer
[0187] An RPT device 4000 according to one form of the present technology may include a silencer 4120 , or a plurality of silencers 4120 .
[0188] In one form of the present technology, an inlet muffler 4122 is located in the pneumatic path upstream of the pressure generator 4140 .
[0189] In one form of the present technology, an outlet muffler 4124 is located in the pneumatic path between the pressure generator 4140 and the patient interface 3000 .
[0190] 5.3.1.3 Pressure generator
[0191] In one form of the present technology, the pressure generator 4140 for supplying pressurized air is a controllable blower 4142. For example, the blower 4142 may include a brushless DC motor 4144 having one or more impellers housed in a volute. The pressure generator 4140 may be capable of generating an air supply or air flow, for example, at about 120 liters per minute, at a positive pressure ranging from about 4 cmH2O to about 20 cmH2O, or in other forms up to about 30 cmH2O.
[0192] The pressure generator 4140 is under the control of the therapy device controller 4240.
[0193] In other forms, the pressure generator 4140 may be a piston driven pump, a pressure regulator connected to a high pressure source (e.g., a compressed air reservoir), or a bellows.
[0194] 5.3.1.4 Transducer
[0195] The transducer may be internal to the RPT device, or external to the RPT device, such as a finger PPG sensor or PAT sensing device as previously described. An external transducer may be located, for example, on or form part of an air circuit (e.g., a patient interface). An external transducer may be in the form of a non-contact sensor such as a Doppler radar motion sensor that sends or transmits data to the RPT device.
[0196] In one form of the present technology, one or more transducers 4270 are located upstream and / or downstream of the pressure generator 4140. The one or more transducers 4270 are constructed and arranged to generate data representative of a corresponding characteristic of the air flow, such as flow, pressure, or temperature at that point in the pneumatic path.
[0197] In one form of the present technology, one or more transducers 4270 are located proximate to the patient interface 3000 .
[0198] In one form, the signal from transducer 4270 may be filtered, such as by low pass filtering, high pass filtering, or band pass filtering.
[0199] 5.3.1.5 Anti-overflow valve
[0200] In one form of the present technology, an anti-spillback valve 4160 is located between the humidifier 5000 and the pneumatic block 4020. The anti-spillback valve is constructed and arranged to reduce the risk of water flowing upstream from the humidifier 5000, for example to the motor 4144.
[0201] 5.3.1.6 Air circuit
[0202] The air circuit 4170 according to one aspect of the present technology is a conduit or tube that, when in use, is constructed and arranged to allow air flow to travel between two components, such as the pneumatic block 4020 and the patient interface 3000 .
[0203] 5.3.1.7 Oxygen delivery
[0204] In one form of the present technology, supplemental oxygen 4180 is delivered to one or more points in the pneumatic pathway, such as upstream of the pneumatic block 4020 , to the air circuit 4170 , and / or to the patient interface 3000 .
[0205] 5.3.2 RPT device electrical components
[0206] 5.3.2.1 Power supply
[0207] In one form of the present technology, the power source 4210 is internal to the outer housing 4010 of the RPT device 4000. In another form of the present technology, the power source 4210 is external to the outer housing 4010 of the RPT device 4000.
[0208] In one form of the present technology, the power supply 4210 provides power only to the RPT device 4000. In another form of the present technology, the power supply 4210 provides power to both the RPT device 4000 and the humidifier 5000.
[0209] 5.3.2.2 Input device
[0210] In one form of the present technology, the RPT device 4000 includes one or more input devices 4220 in the form of buttons, switches, or dials to allow a person to interact with the device. The buttons, switches, or dials may be physical devices or software devices accessible via a touch screen. The buttons, switches, or dials may be physically connected to the external housing 4010 in one form, or may be in another form in wireless communication with a receiver electrically connected to a central controller 4230.
[0211] In one form, input device 4220 may be constructed and arranged to allow a person to select values and / or menu options.
[0212] 5.3.2.3 Central Controller
[0213] In one form of the present technology, the central controller 4230 is a processor suitable for controlling the RPT device 4000, such as an x86 INTEL processor.
[0214] A central controller 4230 adapted to control a RPT device 4000 according to another form of the present technology comprises a processor based on an ARM Cortex-M processor from ARM Holdings. For example, an STM32 series microcontroller from STMICROELECTRONICS may be used.
[0215] Another central controller 4230 suitable for controlling the RPT device 4000 according to another alternative form of the present technology comprises a member selected from the ARM9 based 32-bit RISC CPU family. For example, the STR9 series of microcontrollers from ST MICROELECTRONICS may be used.
[0216] In some alternative forms of the present technology, a 16-bit RISC CPU may be used as the central controller 4230 of the RPT device 4000. For example, a processor from the MSP430 series of microcontrollers manufactured by TEXAS INSTRUMENTS may be used.
[0217] In another form of the present technology, the central controller 4230 is a dedicated electronic circuit. In another form, the central controller 4230 is an application specific integrated circuit (ASIC). In another form, the central controller 4230 includes discrete electronic components.
[0218] The central controller 4230 is configured to receive input signals from one or more transducers 4270 , one or more input devices 4220 , and the humidifier 5000 .
[0219] The central controller 4230 is configured to provide output signals to one or more of the output device 4290 , the therapy device controller 4240 , the data communication interface 4280 , and the humidifier 5000 .
[0220] In some forms of the present technology, the central controller 4230 is configured to implement one or more methods described herein, such as one or more algorithms 4300 represented as computer programs stored in a non-transitory computer-readable storage medium (such as memory 4260 or other memory described herein). In some forms of the present technology, the central controller 4230 may be integrated with the RPT device 4000, as previously discussed. However, in some forms of the present technology, some of the methods may be performed by a remotely located device or server (such as the server previously mentioned). For example, the remotely located device or server may determine control settings for delivery to a ventilator or other RT device, such as by detecting breathing-related events and distinguishing them by type by analyzing stored data such as from any sensor described herein.
[0221] Although the central controller 4230 may include a single controller that interacts with various sensors 4270, data communication interfaces 4280, memory 4260, and other devices, the functionality of the controller 4230 may be distributed among more than one controller. Thus, the term "central" as used herein is not meant to limit the architecture to a single controller or processor that controls other devices. For example, an alternative architecture may include a distributed controller architecture involving more than one controller or processor, which may optionally communicate electronically (wired or wirelessly) directly or indirectly with the previously described finger sensors or a server that communicates with the finger sensors, such as for implementing any of the methods described herein. This may include, for example, a separate local (i.e., within the RPT device 4000) or remotely located controller that executes some of the algorithms 4300, or even more than one local or remote memory that stores some of the algorithms. Additionally, when represented as a computer program, the algorithm may include high-level human-readable code (e.g., C++, Visual Basic, other object-oriented languages, etc.) or low-level / machine-level instructions (Assembler, Verilog, etc.). Depending on the functionality of the algorithm, such code or instructions may be burned into a controller (e.g., an ASIC or DSP), or may be a run-time executable program ported to a DSP or general-purpose processor which is then specifically programmed to perform the tasks required by the algorithm.
[0222] 5.3.2.4 Clock
[0223] The RPT device 4000 may include a clock 4232 connected to a central controller 4230 .
[0224] 5.3.2.5 Therapy device controller
[0225] In one form of the present technology, the therapy device controller 4240 is a therapy control module 4330 which forms part of an algorithm 4300 executed by the central controller 4230 .
[0226] In one form of the present technology, the therapy device controller 4240 is a dedicated motor control integrated circuit. For example, in one form, the MC33035 brushless DC motor controller manufactured by ONSEMI is used.
[0227] 5.3.2.6 Protection circuit
[0228] An RPT device 4000 according to the present technology may include one or more protection circuits 4250 .
[0229] One form of protection circuit 4250 according to the present technology is an electrical protection circuit.
[0230] One form of protection circuit 4250 according to the present technology is a temperature or pressure safety circuit.
[0231] 5.3.2.7 Memory
[0232] According to one form of the present technology, the RPT device 4000 includes a memory 4260, such as a non-volatile memory. In some forms, the memory 4260 may include static RAM powered by a battery. In some forms, the memory 4260 may include volatile RAM.
[0233] Memory 4260 may be located on PCBA 4202. Memory 4260 may be in the form of EEPROM or NAND flash memory.
[0234] Additionally or alternatively, the RPT device 4000 includes memory 4260 in a removable form, such as a memory card manufactured in accordance with the Secure Digital (SD) standard.
[0235] In one form of the present technology, memory 4260 (such as one of the memories previously described) acts as a non-transitory computer-readable storage medium having stored thereon computer program instructions representing one or more methods described herein, such as one or more algorithms 4300.
[0236] 5.3.2.8 Transducer
[0237] The transducer may be internal to the device 4000, or external to the RPT device 4000. The external transducer may be located, for example, on the air delivery circuit 4170 (e.g., at the patient interface 3000) or form part of the air delivery circuit. The external transducer may be in the form of a non-contact sensor such as a Doppler radar motion sensor that sends or transmits data to the RPT device 4000.
[0238] 5.3.2.8.1 Flow
[0239] The flow transducer 4274 according to the present technology can be based on a differential pressure transducer, such as the SDP600 series differential pressure transducer from SENSIRION. The differential pressure transducer is in fluid communication with the pneumatic circuit, one of each pressure transducer being connected to a corresponding first point and a second point in the flow restriction element.
[0240] In one example, the central controller 4230 receives a signal from the flow transducer 4274 representing the total flow rate Qt.
[0241] 5.3.2.8.2 Pressure
[0242] A pressure transducer 4272 in accordance with the present technology is positioned in fluid communication with the pneumatic path. An example of a suitable pressure transducer 4272 is a sensor from the HONEYWELL ASDX series. An alternative suitable pressure transducer is a sensor from the NPA series from GENERAL ELECTRIC.
[0243] In use, the central controller 4230 receives a signal from the pressure transducer 4272. In one form, the signal from the pressure transducer 4272 is filtered before being received by the central controller 4230.
[0244] 5.3.2.8.3 Motor speed
[0245] In one form of the present technology, a motor speed transducer 4276 is used to determine the rotational speed of the motor 4144 and / or blower 4142. A motor speed signal from the motor speed transducer 4276 may be provided to the therapy device controller 4240. The motor speed transducer 4276 may be, for example, a speed sensor such as a Hall effect sensor.
[0246] 5.3.2.9 Data Communication System
[0247] In one form of the present technology, a data communication interface 4280 is provided and connected to the central controller 4230. The data communication interface 4280 may be connected to a remote external communication network 4282 and / or a local external communication network 4284. The remote external communication network 4282 may be connected to a remote external device 4286. The local external communication network 4284 may be connected to a local external device 4288.
[0248] In one form, the data communications interface 4280 is part of the central controller 4230. In another form, the data communications interface 4280 is separate from the central controller 4230 and may include an integrated circuit or processor.
[0249] In one form, remote external communication network 4282 is the Internet. Data communication interface 4280 may connect to the Internet using wired communication (eg, via Ethernet or fiber optic) or wireless protocols (eg, CDMA, GSM, LTE).
[0250] In one form, the local external communications network 4284 utilizes one or more communications standards, such as Bluetooth or consumer infrared protocols, and may optionally communicate with any of the sensors described herein.
[0251] In one form, the remote external device 4286 is one or more computers, such as a cluster of networked computers and / or a server as described herein. In one form, the remote external device 4286 can be a virtual computer rather than a physical computer. In either case, such a remote external device 4286 can be accessed by a properly authorized person such as a clinician.
[0252] The local external device 4288 may be a personal computer, a mobile phone, a tablet, or a remote control device.
[0253] 5.3.2.10 Output devices including optional display and alarm
[0254] Output devices 4290 according to the present technology may take the form of one or more of visual, audio, and tactile units.The visual display may be a liquid crystal display (LCD) or a light emitting diode (LED) display.
[0255] 5.3.2.10.1 Display Driver
[0256] The display driver 4292 receives characters, symbols, or images to be displayed on the display 4294 as input, and converts them into commands that cause the display 4294 to display the characters, symbols, or images.
[0257] 5.3.2.10.2 Display
[0258] The display 4294 is configured to visually display characters, symbols, or images in response to commands received from the display driver 4292. For example, the display 4294 may be an eight-segment display, in which case the display driver 4292 converts each character or symbol (such as the digit "0") into eight logic signals that indicate whether the eight corresponding segments are to be activated to display the particular character or symbol.
[0259] 5.3.3RPT device algorithm
[0260] 5.3.3.1 Preprocessing module
[0261] A pre-processing module 4310 in accordance with the present technology receives as input raw data from a transducer 4270 (e.g., a flow sensor 4274 or a pressure sensor 4272) and performs one or more processing steps to calculate one or more output values to be used as input to another module (e.g., a therapy engine module 4320).
[0262] In one form of the present technology, the output values include interface or mask pressure Pm, respiratory flow Qr and leak flow Ql.
[0263] In various forms of the present technology, the pre-processing module 4310 includes one or more of the following algorithms: pressure compensation 4312, ventilation flow estimation 4314, leakage flow estimation 4316, respiratory flow estimation 4317, ventilation volume determination 4311, target ventilation volume determination 4313, respiratory rate estimation 4318 and backup rate determination 4319.
[0264] 5.3.3.1.1 Pressure compensation
[0265] In one form of the present technology, the pressure compensation algorithm 4312 receives as input a signal indicative of the pressure in the pneumatic path near the outlet of the pneumatic block 4020. The pressure compensation algorithm 4312 estimates the pressure drop in the air circuit 4170 and provides as output an estimated pressure Pm in the patient interface 3000.
[0266] 5.3.3.1.2 Ventilation Flow Estimation
[0267] In one form of the present technology, the ventilation flow estimation algorithm 4314 receives as input an estimated pressure Pm in the patient interface 3000 and estimates a ventilation flow Qv of air from the vent 3400 in the patient interface 3000 .
[0268] 5.3.3.1.3 Leakage flow estimation
[0269] In one form of the present technology, the leak flow estimation algorithm 4316 receives the total flow Qt and the ventilation flow Qv as inputs and estimates the leak flow Ql. In one form, the leak flow estimation algorithm 4316 estimates the leak flow Ql by averaging the difference between the total flow and the ventilation flow Qv over a period long enough to include several respiratory cycles (e.g., about 10 seconds).
[0270] In one form, the leak flow estimation algorithm 4316 receives as input the total flow Qt, ventilation flow Qv, and estimated pressure Pm in the patient interface 3000, and estimates the leak flow Ql by calculating the leak conductance and determining the leak flow Ql as a function of the leak conductance and pressure Pm. The leak conductance can be calculated as the quotient of the low-pass filtered non-ventilation flow equal to the difference between the total flow Qt and the ventilation flow Qv and the low-pass filtered pressure square root Pm, where the low-pass filter time constant has a value long enough to include several respiratory cycles (e.g., about 10 seconds). The leak flow Ql can be estimated as the product of the leak conductance and the pressure function Pm.
[0271] 5.3.3.1.4 Respiratory flow estimation
[0272] In one form of the present technology, a respiratory flow estimation algorithm 4317 receives as input a total flow Qt, a ventilation flow Qv, and a leak flow Ql, and estimates a respiratory flow Qr of air to the patient by subtracting the ventilation flow Qv and the leak flow Ql from the total flow Qt.
[0273] In other forms of the present technology, the respiratory flow estimation algorithm 4317 provides a value that serves as a surrogate for the respiratory flow Qr. Possible surrogates for respiratory flow include:
[0274] - Patient chest breathing exercise 1000
[0275] - Current consumed by the pressure generator 4140
[0276] -Motor speed of pressure generator 4140
[0277] - Patient transthoracic impedance 1000
[0278] The respiratory flow surrogate value may be provided by a transducer 4270 in the RPT device 4000 (eg, a motor speed sensor 4276) or a sensor external to the RPT device 4000 (such as a respiratory motion sensor or a transthoracic impedance sensor).
[0279] 5.3.3.1.5 Determination of ventilation volume
[0280] In one form of the present technology, a ventilation determination algorithm 4311 receives an input of respiratory flow Qr and determines a measurement value Vent indicative of current patient ventilation.
[0281] In some embodiments, the ventilation determination algorithm 4311 determines a measure of ventilation, Vent, which is an estimate of actual patient ventilation.
[0282] In one such embodiment, the measure of ventilation Vent is half the absolute value of the respiratory flow Qr, optionally filtered by a low pass filter such as a second order Bessel low pass filter with a corner frequency of 0.11 Hz.
[0283] In one such embodiment, the measurement of ventilation Vent is an estimate of total alveolar ventilation (i.e., non-anatomical dead space ventilation). This requires an estimate of anatomical dead space. The patient's height (or arm span in the case of severe skeletal deformities) can be used as a good predictor of anatomical dead space. Total alveolar ventilation is then equal to the measurement of actual patient ventilation (e.g., as determined above) minus the product of the estimated anatomical dead space and the estimated spontaneous respiratory rate Rs.
[0284] In other embodiments, the ventilation determination algorithm 4311 determines a measure of ventilation Vent that is approximately proportional to the actual patient ventilation. One such embodiment estimates the peak respiratory flow Qpeak during the inspiratory portion of the cycle. If the flow waveform shape does not vary much (here, the shapes of two breaths are considered similar when the flow waveforms of the breaths normalized in time and amplitude are similar), then this procedure and many other procedures involving sampling the respiratory flow Qr produce a measure that is approximately proportional to ventilation. Some simple examples include the median of positive respiratory flow, the median of the absolute values of respiratory flow, and the standard deviation of flow. Any linear combination of any order statistics of the absolute values of respiratory flow using positive coefficients, and even some any linear combination of any order statistics of the absolute values of respiratory flow using both positive and negative coefficients, is approximately proportional to ventilation. Another example is the average of the respiratory flow in the middle K proportions (by time) of the inspiratory portion, where 0<K<1. If the flow waveform shape is constant, there are an arbitrarily large number of measurements that are exactly proportional to ventilation.
[0285] In other forms, the ventilation determination algorithm 4311 determines a measurement of ventilation Vent that is not based on respiratory flow Qr, but rather is a surrogate for current patient ventilation, such as oxygen saturation (SaO2) or partial pressure of carbon dioxide (PCO2) obtained from a suitable sensor attached to the patient 1000.
[0286] 5.3.3.1.6 Determination of target ventilation
[0287] In one form of the present technology, the central controller 4230 takes as input a current ventilation measurement Vent and executes one or more target ventilation determination algorithms 4313 for determining a target value Vtgt for the ventilation measurement.
[0288] In some forms of the present technology, there is no target ventilation determination algorithm 4313 and the target ventilation Vtgt is predetermined, for example by hard coding during configuration of the RPT device 4000 or by manual input via the input device 4220 .
[0289] In other forms of the present technology, such as adaptive servo-ventilation (ASV) therapy (described below), the target ventilation determination algorithm 4313 calculates the target ventilation Vtgt based on a value Vtyp indicative of the typical recent ventilation of the patient 1000.
[0290] In some forms of adaptive servo-ventilation therapy, the target ventilation Vtgt is calculated as a high proportion of the typical recent ventilation Vtyp, but less than the typical recent ventilation. Such forms of the high proportion may be in the range of (80%, 100%) or (85%, 95%) or (87%, 92%).
[0291] In other forms of adaptive servo-ventilation therapy, the target ventilation Vtgt is calculated to be slightly greater than an integer multiple of the typical recent ventilation Vtyp.
[0292] The typical recent ventilation Vtyp is a value at which the distribution of the measured value Vent of the current ventilation at multiple moments on some predetermined time scale tends to be concentrated, that is, a measurement of the central tendency of the measured value of the current ventilation in the recent history. In one implementation of the target ventilation determination algorithm 4313, the recent history is on the order of a few minutes, but in any case should be longer than the time scale of the Cheyne-Stokes rise and fall cycle. The target ventilation determination algorithm 4313 can use any of a variety of known measurements of central tendency to determine the typical recent ventilation Vtyp based on the measured value Vent of the current ventilation. One such measurement is the output of a low-pass filter on the measured value Vent of the current ventilation, where the time constant is equal to 100 seconds.
[0293] 5.3.3.1.7 Respiration rate estimation
[0294] In one form of the present technology, a respiratory rate estimation algorithm 4318 receives as input a respiratory flow Qr of a patient 1000 and generates an estimate of the patient's spontaneous respiratory rate Rs.
[0295] The respiratory rate estimation algorithm 4318 may estimate the spontaneous respiratory rate Rs during periods when the patient 1000 is breathing spontaneously, i.e., when the RPT device 4000 is not delivering "backup breaths" (described below). In some forms of the present technology, the respiratory rate estimation algorithm 4318 estimates the respiratory rate during periods when the servo-assist (defined as pressure support minus minimum pressure support) is low (less than 4 cmH2O in one embodiment) because such periods are more likely to reflect spontaneous breathing efforts.
[0296] In some forms of the present technology, the respiratory rate estimation algorithm 4318 estimates the respiratory rate during periods of sleep breathing, because the respiratory rate during these periods may be significantly different from the respiratory rate during wakefulness. Anxiety generally leads to higher respiratory rates than during sleep. When patients focus on their own breathing process, their respiratory rate is generally lower than the respiratory rate during normal wakefulness or sleep. Techniques such as those described in patent application No. PCT / AU2010 / 000894 published as WO 2011 / 006199 (the entire disclosure of which is hereby incorporated by reference herein) can be used to identify periods of wakefulness breathing from the respiratory flow rate Qr.
[0297] In some forms of the present technology, the respiratory rate estimation algorithm 4318 estimates the spontaneous respiratory rate Rs as the inverse of one of a variety of well-known statistical measures of central tendency of the respiratory duration Ttot during the period of interest. In such measures, it is desirable to reject outliers, or at least be robust to outliers. One such measure, the trimmed mean, in which lower and higher K proportions of the sorted respiratory durations are discarded and the mean is calculated based on the remaining respiratory durations, is robust to outliers. For example, when K is 0.25, this is equivalent to discarding the upper and lower quartiles of the respiratory duration Ttot. The median is another robust measure of central tendency, although this occasionally gives unsatisfactory results when the distribution is strongly bimodal. A simple mean can also be used as a measure of central tendency, although it is sensitive to outliers. An initial interval filtering stage can be employed, in which consecutive time intervals corresponding to unreliable respiratory rates (e.g., greater than 45 breaths / minute or less than 6 breaths / minute) are excluded from the mean calculation as outliers. Other filtering mechanisms that can be used alone or in combination with interval filtering are to exclude any breaths that are not part of a sequence of N consecutive spontaneous breaths, where N is some small integer (e.g., 3), and to exclude early and late breaths of a sequence of consecutive spontaneous breaths, such as excluding the first and last breaths of a sequence of four breaths. The rationale for the latter mechanism is that the first and last breaths of a sequence of spontaneous breaths (in particular, and early and late breaths in general) may be atypical; for example, the first spontaneous breath may occur due to arousal, and the last spontaneous breath may be longer due to reduced respiratory drive, which results in a backup breath that ends the sequence of spontaneous breaths.
[0298] In some forms of the present technology, the respiratory rate estimation algorithm 4318 makes an initial estimate of the spontaneous respiratory rate Rs using an initial estimation period to enable subsequent processing in the therapy engine module 4320 to begin, and then continuously updates the estimate of the spontaneous respiratory rate Rs using estimation periods longer than the initial estimation period to improve statistical robustness. For example, the initial estimation period may be 20 minutes of suitable spontaneous breathing, but the estimation period may then be gradually increased up to some maximum duration, such as 8 hours. Rather than using a rolling window of that duration for the estimate, a low pass filter of the respiratory duration may be used, with the response time gradually getting longer (more precisely, the angular frequency gradually decreasing) as the session progresses.
[0299] In some forms, a suitably processed short-term (e.g., 10 minute) measure of central tendency (such as a trimmed mean) may be input to a suitable low pass filter to give an estimate of Rs that varies over a time scale of hours or longer. This has the advantage that potentially large amounts of respiration duration data do not need to be stored and processed, as might occur if a trimmed mean needed to be calculated from a moving window of respiration duration data spanning hours or days.
[0300] In some forms of the present technology, breathing rate measured over a short period of time, particularly within a breath, may also be used in place of breathing duration in the above-mentioned measure of central tendency, giving generally similar but not identical results.
[0301] 5.3.3.1.8 Determination of backup rate
[0302] In one form of the present technology, a backup rate determination algorithm 4319 receives as input a spontaneous breathing rate estimate Rs provided by a breathing rate estimation algorithm 4318, and returns a "backup rate" Rb. The backup rate Rb is the rate at which the RPT device 4000 will deliver backup breaths (i.e., continue to provide ventilatory support) to the patient 1000 in the absence of significant spontaneous breathing effort.
[0303] In one form of the pre-processing module 4310, the backup rate determination algorithm 4319 is not present, and instead the backup rate Rb is provided to the RPT device 4000 manually, for example via the input device 4220, or is hard-coded when the RPT device 4000 is configured.
[0304] In a form called adaptive backup rate, the backup rate determination algorithm 4319 determines the backup rate Rb as a function of the spontaneous breathing rate Rs. In one embodiment, the function determines the backup rate Rb as the spontaneous breathing rate Rs minus a constant such as 2 breaths per minute. In another embodiment, the function determines the backup rate Rb as the spontaneous breathing rate Rs multiplied by a constant slightly less than 1.
[0305] In a form called variable backup rate, the backup rate determination algorithm 4319 determines the backup rate Rb as a function of time. The backup rate Rb is initialized to a value called the autonomous backup rate (SBR), which is a certain fraction of the final target backup rate called the continuous timed backup rate (STBR). The fraction can be two-thirds, or three-quarters, or other positive values less than 1. SBR is the reciprocal of the timeout period and the backup breath when the most recent inspiration is a spontaneous (i.e., unobstructed triggered) breath. STBR can be predetermined (e.g., by manual input or hard coding as described above) or set to a certain typical breathing rate, such as 15bpm. Since the previous spontaneous breath, over time, the backup rate Rb increases from SBR to STBR. The increase can be according to a predetermined curve, such as a series of steps, or a continuous linear curve. The curve is selected so that the backup rate Rb reaches STBR after a predetermined interval. The interval can be measured in units of time, such as 30 seconds, or measured relative to the patient's breathing, such as 5 breaths.
[0306] In some forms of variable backup rate, the predetermined interval at which the backup rate Rb increases from SBR to STBR may be a function of the adequacy of the current ventilation. In one embodiment suitable for servo-ventilation, where there is a target value Vtgt for a measure of ventilation, the backup rate approaches STBR more quickly to the extent that the current measure of ventilation Vent is less than the target ventilation Vtgt.
[0307] In one form of a variable backup rate called an adaptive variable backup rate, a backup rate determination algorithm 4319 determines the backup rate Rb as a function of the currently estimated spontaneous breathing rate Rs provided by the respiratory rate estimation algorithm 4318, and a function of time. As determined by the variable backup rate, the adaptive variable backup rate determines that the backup rate Rb is increased from SBR to STBR within a predetermined interval, and the predetermined interval can be a function of the current ventilation adequacy. STBR can be initialized to a standard respiratory rate, such as 15bpm. Once a reliable estimate of the spontaneous breathing rate Rs can be obtained from the respiratory rate estimation algorithm 4318, the STBR can be set to the currently estimated spontaneous breathing rate Rs multiplied by a certain constant. SBR can be set to a certain fraction of STBR, as in the variable backup rate. In one form, for example, a fraction of two-thirds can be set to a lower value, such as 0.55, during the initial estimation period of the spontaneous breathing rate Rs to accommodate the occasional long breathing duration of patients with relatively low respiratory rates (such as 12 breaths per minute).
[0308] In some forms, the constant that is multiplied by the current estimated spontaneous breathing rate Rs to obtain the STBR may be slightly higher than 1, such as 1.1, to provide more aggressive ventilation during apnea, which may be desirable in short-term apnea. The constant may be slightly lower than 1, such as 0.8, particularly if difficulty resynchronizing with the patient as the patient struggles to return is a problem in a particular patient. A lower backup rate makes resynchronization easier by prolonging the expiratory pause during which resynchronization typically occurs.
[0309] 5.3.3.2 Therapy Engine Module
[0310] In one form of the present technology, a therapy engine module 4320 receives as input one or more of a pressure Pm in the patient interface 3000, a respiratory flow Qr of air to the patient, and an estimate of the spontaneous breathing rate Rs, and provides as output one or more therapy parameters. In various forms, the therapy engine module 4320 includes one or more of the following algorithms: phase determination 4321, waveform determination 4322, inspiratory flow limitation determination 4324, apnea / hypopnea determination 4325, snoring detection 4326, airway patency determination 4327, and therapy parameter determination 4329, such as therapy parameter determination including central vs. obstructive type determination as previously described.
[0311] 5.3.3.2.1 Phase determination
[0312] In one form of the present technology, the phase determination algorithm 4321 receives as input a signal indicative of respiratory flow Qr and provides as output the phase Φ of the current respiratory cycle of the patient 1000 .
[0313] In some forms, referred to as discrete phase determination, the phase output Φ is a discrete variable. One embodiment of discrete phase determination provides a two-value phase output Φ having an inhalation value or an exhalation value, such as values represented as 0 turns and 0.5 turns when the start of spontaneous inhalation and exhalation are detected, respectively. The "triggered" and "cycled" RPT device 4000 effectively performs discrete phase determination because the trigger point and the cycle point are the moments when the phase changes from exhalation to inhalation and from inhalation to exhalation, respectively. In one embodiment of two-value phase determination, the phase output Φ is determined to have a discrete value of 0 (thereby "triggering" the RPT device 4000) when the respiratory flow Qr has a value that exceeds a positive threshold, and is determined to have a discrete value of 0.5 turns (thereby "cycling" the RPT device 4000) when the respiratory flow Qr has a value that is more negative than a negative threshold.
[0314] Another embodiment of discrete phase determination provides a three-value phase output Φ having one of values of inhale, intermediate inspiratory pause, and exhale.
[0315] In other forms, referred to as continuous phase determination, the phase output Φ is a continuous value, such as varying from 0 to 1 turn, or 0 to 2 radians. The RPT device 4000 performing continuous phase determination can trigger and loop when the continuous phase reaches 0 and 0.5 turns, respectively. In one embodiment of continuous phase determination, a fuzzy logic analysis of the respiratory flow Qr is used to determine the continuous values of the phase Φ. The continuous values of the phase determined in this embodiment are generally referred to as "fuzzy phase". In one embodiment of the fuzzy phase determination algorithm 4321, the following rules are applied to the respiratory flow Qr:
[0316] 1. If the respiratory flow is zero and increases rapidly, the phase is 0 turns.
[0317] 2. If the respiratory flow is large and stable, the phase is 0.25 turns.
[0318] 3. If the respiratory flow is zero and decreases rapidly, the phase is 0.5 turns.
[0319] 4. If the respiratory flow is large and stable, the phase is 0.75 turns.
[0320] 5. If the respiratory flow is zero and stable, and the 5-second low-pass filtered absolute value of the respiratory flow is large, the phase is 0.9 turns.
[0321] 6. If the respiratory flow is positive and the phase is exhalation, the phase is 0 turns.
[0322] 7. If the respiratory flow is negative and the phase is inspiration, the phase is 0.5 turns.
[0323] 8. If the absolute value of the 5-second low-pass filter of the respiratory flow is large, the phase increases at a steady rate equal to the patient's respiratory rate, and the time constant of the low-pass filter is 20 seconds.
[0324] The output of each rule can be represented as a vector whose phase is the result of the rule and whose amplitude is the fuzziness of the rule being true. The fuzziness of respiratory flow being "large", "stable", etc. is determined using appropriate membership functions. The results of the rules are represented as vectors and then combined by some function such as taking the centroid. In this combination, the rules can be weighted equally, or differently.
[0325] In another embodiment of continuous phase determination, the inhalation time Ti and the exhalation time Te are first estimated from the respiratory flow Qr. The phase Φ is then determined as half the proportion of the inhalation time Ti that has passed since the previous trigger moment, or 0.5 turns plus half the proportion of the exhalation time Te that has passed since the previous cycle moment (whichever is closer).
[0326] In some forms of the present technology adapted for pressure support ventilation therapy (described below), the phase determination algorithm 4321 is configured to trigger even when the respiratory flow Qr is not significant, such as during apnea. As a result, the RPT device 4000 delivers "backup breaths" in the absence of spontaneous respiratory effort from the patient 1000. For such a form, referred to as spontaneous / timed (S / T) mode, the phase determination algorithm 4321 can utilize the backup rate Rb provided by the backup rate determination algorithm 4319.
[0327] The phase determination algorithm 4321 using "fuzzy phase" can implement the S / T mode using a backup rate Rb by including a "momentum" rule in the fuzzy phase rule. The effect of the momentum rule is that if there is no feature of the respiratory flow Qr, then the continuous phase is transferred forward from exhalation to inspiration at the backup rate Rb, otherwise the respiratory flow will transfer the continuous phase forward through other rules. In one embodiment, the further the measured value of ventilation Vent (described below) is below the target value of ventilation Vtgt (also described below), the higher the momentum rule is weighted in the combination. However, due to the rapid increase in pressure support in response to mild to moderate hypoventilation (relative to the target ventilation), the ventilation may be very close to the target ventilation. It is desirable that when the ventilation is close to the target, the momentum rule is given a lower weight to allow the patient to breathe at a rate significantly lower than the breathing rate at other times (when the patient is not in central apnea) without being unnecessarily pushed by the ventilator to breathe at a higher rate. However, when the momentum rule is given a lower weight when ventilation is above a value below but close to the target ventilation, adequate ventilation can be easily achieved at a relatively high pressure support at a rate much lower than the backup rate. It is desirable to deliver backup breaths at a higher rate because this will enable the target ventilation to be delivered at a lower pressure support. This is desirable for a number of reasons, a key one being to reduce mask leak.
[0328] In summary, in the fuzzy phase determination algorithm 4321 implementing the S / T mode, there is a dilemma when selecting the weight for the momentum rule including the backup rate Rb: if it is too high, the patient may feel "pushed" by the backup rate. If it is too low, the pressure support may be too large. Therefore, it is desirable to provide a method for implementing the S / T mode that does not rely on the above-mentioned momentum rule.
[0329] The phase determination algorithm 4321 (discrete or continuous, without momentum rule) can use the backup rate Rb to implement the S / T mode in a manner called timed backup. Timed backup can be implemented as follows: the phase determination algorithm 4321 attempts to detect the start of inhalation due to spontaneous breathing effort, for example by monitoring the respiratory flow Qr as described above. If the start of inhalation due to spontaneous breathing effort is not detected within a time period after the last triggering moment of duration equal to the inverse of the backup rate Rb (an interval called the backup timing threshold), the phase determination algorithm 4321 sets the phase output Φ to the value of inhalation (thereby triggering the RPT device 4000). Once the RPT device 4000 is triggered and backup breaths begin to be delivered, the phase determination algorithm 4321 attempts to detect the start of spontaneous exhalation, for example by monitoring the respiratory flow Qr, and at the start of spontaneous exhalation, the phase output Φ is set to the value of exhalation (thereby cycling the RPT device 4000).
[0330] If the backup rate Rb increases over time from SBR to STBR, as in the variable backup rate system described above, the backup timing threshold starts out longer and then gradually gets shorter. That is, the RPT device 4000 starts out less alert and gradually becomes more alert with a lack of spontaneous breathing effort as more backup breaths are delivered. Such an RPT device 4000 is less likely to make the patient feel "pushed" if the patient prefers to breathe at a lower rate than standard, while still delivering backup breaths when needed.
[0331] If the STBR in a variable backup rate system is adapted to the patient's estimated spontaneous breathing rate, Rs, as in the adaptive variable backup rate system described above, then the backup breaths will be delivered at a rate adapted to the patient's own recent spontaneous breathing efforts.
[0332] 5.3.3.2.2 Waveform determination
[0333] In one form of the present technology, the therapy control module 4330 controls the pressure generator 4140 to provide a therapy pressure Pt that varies as a function of the phase Φ of the patient's respiratory cycle according to a waveform template Π(Φ).
[0334] In one form of the present technology, the waveform determination algorithm 4322 provides a waveform template Π(Φ), where values in the range of [0, 1] on the phase value Φ domain provided by the phase determination algorithm 4321 will be used by the therapy parameter determination algorithm 4329.
[0335] In one form suitable for discrete or continuous valued phases, the waveform template Π(Φ) is a square wave template having a value of 1 for phase values up to and including 0.5 turns, and having a value of 0 for phase values above 0.5 turns. In one form suitable for continuous valued phases, the waveform template Π(Φ) includes two smooth curved portions, i.e., a smooth curved portion (e.g., raised cosine) rising from 0 to 1 for phase values up to 0.5 turns, and a smooth curved portion (e.g., exponential) decaying from 1 to 0 for phase values above 0.5 turns. An example of such a "smooth and comfortable" waveform template is a "shark fin" waveform template, in which the rise is a raised cosine and the smooth decay is a pseudo-exponential (so that the limit of Π is exactly zero as Φ approaches one turn).
[0336] In some forms of the present technology, the waveform determination algorithm 4322 selects a waveform template Π(Φ) from a library of waveform templates based on the settings of the RPT device 4000. Each waveform template Π(Φ) in the library may be provided as a lookup table of values Π relative to phase values Φ. In other forms, the waveform determination algorithm 4322 calculates the waveform template Π(Φ) "on the fly" using a predetermined functional form that may be parameterized by one or more parameters (e.g., a time constant of a portion that curves in an exponential manner). The parameters of the functional form may be predetermined or dependent on the current state of the patient 1000.
[0337] In some forms of the present technology adapted for discrete two-valued phases of inhalation (Φ=0 revolutions) or exhalation (Φ=0.5 revolutions), the waveform determination algorithm 4322 computes the waveform template Π "on the fly" as a function of the discrete phase Φ and the time t measured from the most recent triggering moment. In one such form, the waveform determination algorithm 4322 computes the waveform template Π(Φ, t) in both portions (inhalation and exhalation) as follows:
[0338]
[0339] where Π i (t) and Π e (t) is the inspiratory and expiratory portion of the waveform template Π(Φ, t), and Ti is the inhalation time. In one such form, the inspiratory portion of the waveform template Π i (t) is a smooth rise from 0 to 1 parameterized by the rise time, and the exhalation part of the waveform template Π e (t) is a smooth decline from 1 to 0 parameterized by the decline time.
[0340] 5.3.3.2.3 Determination of inspiratory flow limitation
[0341] In one form of the present technology, the processor executes one or more algorithms 4324 for detecting inspiratory flow limitation (partial obstruction).
[0342] In one form, algorithm 4324 receives as input a respiratory flow signal Qr and provides as output a measure of the degree to which the inspiratory portion of a breath exhibits inspiratory flow limitation.
[0343] In one form of the present technology, the inspiratory portion of each breath is identified based on the phase Φ estimated at each moment. For example, the inspiratory portion of the breath is the value of the respiratory flow with a phase Φ less than or equal to 0.5. A plurality of evenly spaced points (e.g., 65) representing time points are interpolated by an interpolator along the inspiratory flow-time curve for each breath. The curve described by the points is then scaled by a scalar to have a uniform length (duration / period) and a uniform area to eliminate the effects of changing respiratory rate and depth. The scaled breath is then compared in a comparator with a pre-stored template representing a normal unobstructed breath. At any time during inspiration from the template, breaths that deviate from a specified threshold (usually 1 scaling unit) are rejected, such as those breaths determined by the test element to be caused by coughing, sighing, swallowing, and hiccups. For non-rejected data, the moving average of the first such scaling point of the previous few inspiratory events is calculated by the central controller 4230. For the second such point, this is repeated on the same inspiratory event, and so on. Thus, for example, 65 scaled data points are generated by the central controller 4230 and represent a moving average of the previous few inspiratory events (e.g., 3 events). The moving average of the continuously updated values of (e.g., 65) points is hereinafter referred to as "scaled flow", denoted as Qs(t). Alternatively, a single inspiratory event may be utilized instead of a moving average.
[0344] From the scaled flow, two shape factors relevant to determining partial blockage can be calculated.
[0345] Shape Factor 1 is the ratio of the average of the middle (e.g., 32) scaled flow points to the average of the total (e.g., 65) scaled flow points. When this ratio exceeds 1, the breath will be considered normal. When the ratio is 1 or less, the breath will be considered obstructed. A ratio of about 1.17 is considered the threshold between partially obstructed and unobstructed breaths and is equivalent to a degree of obstruction that allows adequate oxygenation to be maintained in a typical user.
[0346] Shape Factor 2 is calculated as the RMS deviation from unit scaled flow at the intermediate (e.g., 32) points. An RMS deviation of about 0.2 units is considered normal. A zero RMS deviation is considered a completely flow-limited breath. The closer the RMS deviation is to zero, the more flow-limited the breath will be considered.
[0347] Shape factors 1 and 2 may be used as alternatives, or in combination. In other forms of the present technology, the number of sampling points, breaths, and intermediate points may be different from those described above. Furthermore, the thresholds may be different from those described.
[0348] 5.3.3.2.4 Determination of apnea and hypopnea
[0349] In one form of the present technology, the central controller 4230 executes one or more algorithms 4325 for detecting apneas and / or hypopneas.
[0350] In one form, one or more apnea / hypopnea detection algorithms 4325 receive as input the respiratory flow Qr and provide as output a flag indicating that an apnea or hypopnea has been detected.
[0351] In one form, when the function of respiratory flow Qr falls below a flow threshold for a predetermined period of time, apnea is considered to have been detected. The function may determine a peak flow, a relatively short-term average flow, or an intermediate flow between a relatively short-term average flow and a peak flow, such as an RMS flow. The flow threshold may be a relatively long-term measurement of flow.
[0352] In one form, when the function of respiratory flow Qr falls below a second flow threshold within a predetermined time period, hypopnea is considered to have been detected. The function may determine a peak flow, a relatively short-term average flow, or an intermediate flow between a relatively short-term average flow and a peak flow, such as an RMS flow. The second flow threshold may be a relatively long-term measurement of flow. The second flow threshold is greater than a flow threshold for detecting apnea.
[0353] In one form, such respiratory events may be characterized as central or obstructive based at least in part on the aforementioned finger sensor PPG-based type detection.
[0354] 5.3.3.2.5 Detection of snoring
[0355] In one form of the present technology, the central controller 4230 executes one or more snoring detection algorithms 4326 for detecting snoring.
[0356] In one form, the snoring detection algorithm 4326 receives as input the respiratory flow signal Qr and provides as output a measure of the degree to which snoring is present.
[0357] The snore detection algorithm 4326 may include a step of determining the strength of the flow signal in the range of 30-300 Hz. In addition, the snore detection algorithm 4326 may also include a step of filtering the respiratory flow signal Qr to reduce background noise (e.g., the sound of airflow from the blower 4142 in the system).
[0358] 5.3.3.2.6 Determination of airway patency
[0359] In one form of the present technology, the central controller 4230 executes one or more algorithms 4327 for determining airway patency.
[0360] In one form, the airway patency algorithm 4327 receives the respiratory flow signal Qr as input and determines the power of the signal in a frequency range of about 0.75 Hz to about 3 Hz. The presence of a peak in this frequency range is considered to indicate an open airway. The absence of a peak is considered to indicate a closed airway.
[0361] In one form, the frequency range within which the peak is sought is the frequency of a small forced oscillation in the treatment pressure Pt. In one embodiment, the frequency of the forced oscillation is 2 Hz with an amplitude of about 1 cmH2O.
[0362] In one form, the airway patency algorithm 4327 receives as input the respiratory flow signal Qr and determines whether a cardiogenic signal is present. The absence of a cardiogenic signal is considered to indicate a closed airway.
[0363] 5.3.3.2.7 Determination of therapy parameters
[0364] In some forms of the present technology, the central controller 4230 executes one or more therapy parameter determination algorithms 4329 for determining one or more therapy parameters using values returned by one or more other algorithms in the therapy engine module 4320 .
[0365] In one form of the present technology, the therapy parameter is the instantaneous therapy pressure Pt. In one embodiment of this form, the therapy parameter determination algorithm 4329 determines the therapy pressure Pt using the following equation:
[0366] Pt=AΠ(Φ)+P0 (1)
[0367] in:
[0368] -A is the amplitude,
[0369] -Φ is the current value of the phase;
[0370] -Π(Φ) is the value of the waveform template at the current value of the phase (in the range of 0 to 1), and
[0371] -P0 is the base pressure.
[0372] If the waveform determination algorithm 4322 provides the waveform template Π(Φ) as a lookup table of values indexed by phase Φ, the therapy parameter determination algorithm 4329 applies equation (1) by locating the nearest lookup table entry to the current value Φ of the phase returned by the phase determination algorithm 4321, or by interpolating between two entries that span the current value Φ of the phase.
[0373] The values of amplitude A and base pressure P0 may be set by the therapy parameter determination algorithm 4329 in the following manner according to the selected pressure therapy mode.
[0374] 5.3.3.3 Therapy Control Module
[0375] The therapy control module 4330 according to one aspect of the present technology receives as input therapy parameters from a therapy parameter determination algorithm 4329 of the therapy engine module 4320 and controls the pressure generator 4140 to deliver air flow according to the therapy parameters.
[0376] In one form of the present technology, the therapy parameter is a therapy pressure Pt, and the therapy control module 4330 controls the pressure generator 4140 to deliver an air flow having a mask pressure Pm at the patient interface 3000 equal to the therapy pressure Pt.
[0377] 5.3.3.4 Detection of fault conditions
[0378] In one form of the present technology, the processor executes one or more methods for detecting a fault condition 4340. The fault condition detected by the one or more methods may include at least one of the following:
[0379] -Power failure (no power or insufficient power)
[0380] - Transducer fault detection
[0381] - Unable to detect the presence of parts
[0382] -Operating parameters outside recommended ranges (e.g., pressure, flow, temperature, PaO2)
[0383] -The test alarm failed to generate a detectable alarm signal.
[0384] Upon detecting a fault condition, the corresponding algorithm signals the presence of a fault by one or more of the following:
[0385] - Activate audible, visual and / or dynamic (e.g., vibration) alarms
[0386] - Send messages to external devices
[0387] - Event Log
[0388] 5.4 Humidifier
[0389] In one form of the present technology, a humidifier 5000 (e.g., Fig.10 ) to change the absolute humidity of the air or gas for delivery to the patient relative to the ambient air. Typically, the humidifier 5000 is used to increase the absolute humidity of the air flow and increase the temperature of the air flow (relative to the ambient air) before delivery to the patient's airway.
[0390] 5.5 Glossary
[0391] For purposes of the present invention, in some forms of the present technology, one or more of the following definitions may apply. In other forms of the present technology, alternative definitions may apply.
[0392] 5.5.1 General Concepts
[0393] Air: In some forms of the present technology, air may be taken to mean atmospheric air, and in other forms of the present technology, air may be taken to mean some other combination of breathable gases, such as atmospheric air enriched with oxygen.
[0394] Respiratory Pressure Therapy (RPT): The delivery of a supply of air to the airways at a therapeutic pressure that is usually positive relative to atmosphere.
[0395] Continuous Positive Airway Pressure (RPT) Therapy: A respiratory pressure therapy in which the therapy pressure is approximately constant throughout the patient's breathing cycle. In some forms, the pressure at the entrance to the airway is slightly higher during exhalation and slightly lower during inhalation. In some forms, this pressure will vary between different breathing cycles of the patient, for example, increasing in response to detecting an indication of partial upper airway obstruction, and decreasing in the absence of an indication of partial upper airway obstruction.
[0396] Patient: A person, whether or not they have a respiratory disease.
[0397] Automatic Positive Airway Pressure (APAP) Therapy: CPAP therapy in which the treatment pressure is automatically adjusted between a minimum and a maximum limit, for example, with each breath, depending on whether there is an indication of an SBD event.
[0398] 5.5.2 Respiratory cycle
[0399] Apnea: According to some definitions, apnea is considered to occur when respiratory flow drops below a predetermined threshold for a period of time (e.g., 10 seconds). Obstructive apnea is considered to occur when some obstruction of the airway does not allow air to flow despite the patient's efforts. Central apnea is considered to occur when apnea is detected due to reduced or absent respiratory effort.
[0400] Respiratory rate (Rs): The rate at which a patient breathes spontaneously, usually measured in breaths per minute.
[0401] Duty cycle: ratio of inhalation time Ti to total breathing duration Ttot.
[0402] Effort (breathing): The work done by a spontaneously breathing person trying to breathe.
[0403] Expiratory portion of the respiratory cycle: the period from the start of expiratory flow to the start of inspiratory flow.
[0404] Flow Limitation: A state in a patient's breathing in which an increase in the patient's effort does not result in a corresponding increase in flow. When flow limitation occurs during the inspiratory portion of the breathing cycle, it may be described as inspiratory flow limitation. When flow limitation occurs during the expiratory portion of the breathing cycle, it may be described as expiratory flow limitation.
[0405] Hypopnea: A decrease in flow, but not a cessation of flow. In one form, a hypopnea is considered to occur when flow decreases below a threshold for a sustained period of time. In one form in adults, a hypopnea is any of the following:
[0406] (i) The patient experiences a 30% decrease in respiration for at least 10 seconds, with an associated 4% desaturation; or
[0407] (ii) The patient's breathing decreases (but less than 50%) for at least 10 seconds with associated desaturations or arousals of at least 3%.
[0408] Inspiratory portion of the respiratory cycle: The period from the start of inspiratory flow to the start of expiratory flow is considered the inspiratory portion of the respiratory cycle.
[0409] Patency (airway): The degree to which an airway is open or the degree to which an airway is patent. A patent airway is open. Airway patency may be quantified, for example with a value of (1) being patent and a value of zero (0) being closed.
[0410] Positive End Expiratory Pressure (PEEP): The pressure above atmospheric pressure that exists in the lungs at the end of exhalation.
[0411] Peak flow (Qpeak): The maximum flow rate during the inspiratory portion of the respiratory flow waveform.
[0412] Respiratory flow / airflow, patient flow / airflow (Qr): These synonymous terms can be understood to refer to the RPT device's estimate of respiratory airflow, as opposed to "true respiratory flow" or "true respiratory airflow", which is the actual respiratory flow experienced by the patient, usually expressed in liters / minute.
[0413] Tidal volume (Vt): The volume of air inhaled or exhaled during normal breathing when no extra effort is exerted.
[0414] Inspiratory time (Ti): The duration of the inspiratory portion of the respiratory flow waveform.
[0415] Exhalation time (Te): The duration of the exhalation portion of the respiratory flow waveform.
[0416] (Total) time or breath duration (Ttot): The total duration between the start of the inspiratory portion of one respiratory flow waveform and the start of the inspiratory portion of the next respiratory flow waveform.
[0417] Upper airway obstruction (UAO): includes partial and complete upper airway obstruction. This may be associated with a state of flow limitation, where flow increases only slightly, or may even decrease, as the pressure difference across the upper airway increases (Starling impedance behavior).
[0418] Ventilation (Vent): A measure of the total amount of gas exchanged by a patient's respiratory system. Measurements of ventilation may include one or both of inspiratory flow and expiratory flow (per unit time). When expressed as volume per minute, this quantity is often referred to as "minute ventilation". Minute ventilation is sometimes given simply as volume, understood as volume per minute.
[0419] 5.5.3RPT device parameters
[0420] Flow: The instantaneous volume (or mass) of air delivered per unit time. Although flow and ventilation have the same volume or mass magnitude per unit time, flow is measured over a much shorter period of time. Flow may be nominally positive for the inspiratory portion of a patient's breathing cycle, and therefore negative for the expiratory portion of a patient's breathing cycle. In some cases, references to flow will be references to a scalar, i.e., a quantity having only magnitude. In other cases, references to flow will be references to a vector, i.e., a quantity having both magnitude and direction. Flow will be given with the symbol Q. "Flow" is sometimes simply abbreviated to "flow". The total flow, Qt, is the flow of air leaving the RPT device. The ventilation flow, Qv, is the flow of air leaving the vent to allow flushing of exhaled gases. The leakage flow, Ql, is the flow that is unintentionally leaking from the patient interface system. The respiratory flow, Qr, is the flow of air received into the patient's respiratory system.
[0421] Leakage: The term leakage is to be considered as unintended air flow. In one example, a leak may occur due to an incomplete seal between the mask and the patient's face. In another example, a leak may occur in the swivel elbow leading to the environment.
[0422] Pressure: Force per unit area. Pressure can be measured in a range of units, including cmH2O, gf / cm 2 and hectopascals. 1 cmH2O is equal to 1 g-f / cm 2 , which is approximately 0.98 hPa. In this manual, pressure is given in cmH2O unless otherwise specified. The pressure in the patient interface is given with the symbol Pm, while the treatment pressure is given with the symbol Pt, which represents the target value to be achieved by the mask pressure Pm at the current moment.
[0423] 5.5.4 Ventilator Terminology
[0424] Adaptive servo ventilator (ASV): A servo ventilator with a variable rather than fixed target ventilation. The variable target ventilation may be derived from some characteristics of the patient, such as the patient's breathing characteristics.
[0425] Backup Rate: A parameter of a ventilator that establishes the rate of breaths (usually in breaths per minute) that the ventilator will deliver to the patient if not triggered by spontaneous breathing effort.
[0426] Cycling: The termination of the inspiratory phase of a ventilator. When a ventilator is delivering a breath to a spontaneously breathing patient, the ventilator is considered to cycle to stop delivering a breath at the end of the inspiratory portion of the respiratory cycle.
[0427] Expiratory Positive Airway Pressure (EPAP): The base pressure to which intra-breath varying pressures are added to produce the desired mask pressure that the ventilator will attempt to achieve at a given time.
[0428] End-expiratory pressure (EEP): The desired mask pressure that the ventilator attempts to achieve at the end of the expiratory portion of a breath. If the pressure waveform template Π(Φ) is zero at the end of exhalation, i.e., Π(Φ) = 0 when Φ = 1, then EEP is equal to EPAP.
[0429] IPAP: The desired mask pressure that the ventilator attempts to achieve during the inspiratory portion of a breath.
[0430] Pressure Support: A number that indicates the increase in pressure during ventilator inspiration over the pressure during ventilator expiration, and usually means the difference between the maximum and base pressures during inspiration (e.g., PS = IPAP - EPAP). In some cases, pressure support means the difference that the ventilator aims to achieve, rather than the difference that is actually achieved.
[0431] Servomotor: A ventilator that measures patient ventilation, has a target ventilation, and adjusts the level of pressure support to bring the patient's ventilation to the target ventilation.
[0432] Servo assist: Pressure support minus minimum pressure support.
[0433] Spontaneous / Timed (S / T): A mode of a ventilator or other device that attempts to detect the onset of a breath in a spontaneously breathing patient. However, if the device cannot detect a breath within a predetermined period of time, the device will automatically initiate the delivery of a breath.
[0434] Swing: A term equivalent to pressure support.
[0435] Triggered: When the ventilator delivers a breath of air to a spontaneously breathing patient, it is considered to be triggered by the patient's effort at the beginning of the inspiratory portion of the respiratory cycle.
[0436] Typical recent ventilation: Typical recent ventilation Vtyp is a value around which recent measurements of ventilation within a predetermined time scale tend to cluster, ie, a measurement of the central tendency of the measurements of ventilation in recent history.
[0437] Ventilator: A mechanical device that provides pressure support to a patient to perform some or all of the work of breathing.
[0438] 5.5.5 Respiratory system anatomy
[0439] Diaphragm: A sheet of muscle that extends across the base of the rib cage. The diaphragm separates the chest cavity, which contains the heart, lungs, and ribs, from the abdominal cavity. As the diaphragm contracts, the volume of the chest cavity increases and air is drawn into the lungs.
[0440] Larynx: The larynx or voice box houses the vocal cords and connects the lower part of the pharynx (hypopharynx) to the trachea.
[0441] Lung: The respiratory organ of humans. The conducting zone of the lungs contains the trachea, bronchi, bronchioles, and terminal bronchioles. The respiratory zone contains the respiratory bronchioles, alveolar ducts, and alveoli.
[0442] Nasal Cavity: The nasal cavity (or nasal fossa) is a large air-filled space above and behind the nose in the middle of the face. The nasal cavity is divided into two parts by a vertical wing called the nasal septum. On either side of the nasal cavity are three horizontal branches called the nasal conchae (singular "concha") or turbinates. The front of the nasal cavity is the nose, while the back joins into the nasopharynx via the choanae.
[0443] Pharynx: The portion of the throat located just below (below) the nasal cavity and above the esophagus and larynx. The pharynx is conventionally divided into three sections: the nasopharynx (epipharynx) (nasal portion of the pharynx), the oropharynx (mesopharynx) (oral portion of the pharynx), and the hypopharynx (hypopharynx).
[0444] 5.6 Other notes
[0445] A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all copyright rights whatsoever.
[0446] Unless the context clearly indicates otherwise and where a range of values is provided, it is understood that each intervening value (to one-tenth of the unit of the lower limit) between the upper and lower limits of that range, and any other stated or intervening value in that stated range, is encompassed within the present technology. The upper and lower limits of these intermediate ranges (which may independently be included in the intermediate ranges) are also encompassed within the present technology, subject to any explicitly excluded limits in the stated ranges. Where the stated range includes one or both limits, ranges excluding either or both of those included limits are also encompassed within the present technology.
[0447] Furthermore, where a value or values are described herein as being implemented as part of the present technology, it should be understood that such values may be approximate unless otherwise indicated and that such values may be used with any suitable significant figures to the extent that practical technical implementation may permit or require it.
[0448] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the technology belongs. Although any methods and materials similar or equivalent to those described herein can also be used in the practice or testing of the technology, a limited number of exemplary methods and materials are described herein.
[0449] When a specific material is identified as being used to construct a component, obvious alternative materials having similar properties may be used as substitutes. Furthermore, unless otherwise specified, any and all components described herein should be understood to be capable of being manufactured, and thus may be manufactured together or separately.
[0450] It must be noted that as used herein and in the appended claims, the singular forms "a," "an," and "the" include plural equivalents unless the context clearly dictates otherwise.
[0451] All publications mentioned herein are incorporated by reference to disclose and describe methods and / or materials as the themes of those publications. The publications discussed herein are provided only for their disclosures before the filing date of the present application. This paper should not be construed as admitting that the present technology is not entitled to be earlier than such disclosures due to prior inventions. In addition, the publication date provided may be different from the actual publication date, which may need to be confirmed independently.
[0452] Furthermore, in interpreting the present invention, all terms should be interpreted in the broadest reasonable manner consistent with the context. In particular, the terms "comprises" and "comprising" should be interpreted as referring to elements, components or steps in a non-exclusive manner, indicating that the referenced elements, components or steps may exist, be utilized, or be combined with other elements, components or steps not explicitly referenced.
[0453] The subject headings used in the detailed description are for the convenience of the reader's reference only and should not be used to limit the subject matter found in the entire disclosure or claims. The subject headings should not be used to interpret the scope of the claims or claim limitations.
[0454] Although the technology herein has been described with reference to specific embodiments, it should be understood that these embodiments only illustrate the principles and applications of the present technology. In some cases, terms and symbols may imply specific details that are not required for practicing the present technology. For example, although the terms "first" and "second" may be used, unless otherwise specified, they are not intended to indicate any order, but can be used to distinguish different elements. In addition, although the process steps in the method can be described or illustrated in sequence, such order is not required. Those skilled in the art will recognize that such order can be modified and / or various aspects thereof can be performed simultaneously or even synchronously.
[0455] It will therefore be appreciated that numerous modifications may be made to the illustrative embodiments and that other arrangements may be devised without departing from the spirit and scope of the present technology. For example, while any aspects of the foregoing methods are generally described for implementation using a PPG sensor and a breathing signal derived using such a sensor, other sensory signals or sensors may be implemented. Thus, other sensors, such as any of the sensors described herein for generating such a breathing signal (e.g., a breathing effort signal), such as a flow sensor, a pressure sensor, a non-contact motion sensor (e.g., a radio frequency motion sensor (e.g., radar) or a sonar sensor), a chest strap sensor, etc., may be implemented for generating any of the foregoing indicators and / or may be used with any aspects of the foregoing classification / assessment techniques.
[0456] References
[0457] [1]A.v.Benjafield et al.,“Estimation of the global prevalence andburden of obstructive sleep apnoea:a literature-based analysis,”The LancetRespiratory Medicine,vol.7,no.8,pp.687–698,Aug.2019,doi:10.1016 / S2213-2600(19)30198-5.
[0458] [2]N.M.Punjabi,“The epidemiology of adult obstructive sleep apnea,”Proceedings of the American Thoracic Society,vol.5,no.2.pp.136–143,Feb.2008.doi:10.1513 / pats.200709-155MG.
[0459] [3]F.-C.Yen,K.Behbehani,S.Member,E.A.Lucas,J.R.Burk,and J.R.Axe,“ANonnvasive Technique for Detecting Obstructive and Central Sleep Apnea,”IEEEtransactions on biomedical engineering,vol.44,no.12,pp.1262–1268,1997,doi:10.1109 / 10.649998.
[0460] [4]J.A.Dempsey,S.C.Veasey,B.J.Morgan,and C.P.O’donnell,“Pathophysiology of Sleep Apnea,”Physiological reviews,vol.90,no.1,pp.47–112,2010,doi:10.1152 / physrev.00043.2008.
[0461] [5]S.M.Caples,A.S.Gami,and V.K.Somers,“Obstructive Sleep Apnea,”Annals of internal medicine,vol.142,no.3,pp.187–197,2005,doi:10.7326 / 0003-4819-142-3-200502010-00010.
[0462] [6]R.G.Smallwood,M.v Vitiello,E.C.Giblin,and P.N.Prinz,“Sleep Apnea:Relationship to Age,Sex,and Alzheimer’s Dementia,”Sleep,vol.6,no.1,pp.16–22,1983,doi:10.1093 / sleep / 6.1.16.
[0463] [7]D.A.Pevernagie et al.,“On the rise and fall of the apnea-hypopneaindex:A historical review and critical appraisal,”Journal of Sleep Research,vol.29,no.4.Blackwell Publishing Ltd,Aug.01,2020.doi:10.1111 / jsr.13066.
[0464] [8]D.J.Eckert,A.S.Jordan,P.Merchia,and A.Malhotra,“Central sleepapnea:Pathophysiology and treatment,”Chest,vol.131,no.2,pp.595–607,2007,doi:10.1378 / chest.06.2287.
[0465] [9]S.Javaheri and J.A.Dempsey,“Central sleep apnea,”ComprehensivePhysiology,vol.3,no.1,pp.141–163,2013,doi:10.1002 / cphy.c110057.
[0466]
[10] C.Sullivan,F.Issa,M.Berthon-Jones,and L.Eves,“Reversal ofobstructive sleep apnoea by continuous positive airway pressure appliedthrough the nares.,”Lancet,vol.1,no.8225,pp.862–862,1981,doi:10.1016 / s0140-6736(81)92140-1.
[0467]
[11] W.J.Randerath et al.,“Non-CPAP therapies in obstructive sleepapnoea,”European Respiratory Journal,vol.37,no.5,p.1000,May 2011,doi:10.1183 / 09031936.00099710.
[0468]
[12] L.Rosenthal et al.,“CPAP therapy in patients with mild OSA:implementation and treatment outcome,”Sleep medicine,vol.1,no.3,pp.215–220,2000,doi:10.1016 / s1389-9457(00)00012-5.
[0469]
[13] R.N.Aurora et al.,“The treatment of central sleep apnea syndromesin adults:Practice parameters with an evidence-based literature review andmeta-analyses,”Sleep,vol.35,no.1.pp.17–40,2012.doi:10.5665 / sleep.1580.
[0470]
[14] K.E.Bloch,“Polysomnography:a systematic review,”Technology andHealth Care,vol.5,no.4,pp.285–305,1997,doi:10.3233 / THC-1997-5403.
[0471]
[15] C.A.Kushida et al.,“Practice Parameters for the Indications forPSG-AASM Practice Parameters SLEEP Practice Parameters for the Indicationsfor Polysomnography and Related Procedures:An Update for 2005,”Sleep,vol.28,no.4,pp.499–521,2005,doi:10.1093 / sleep / 28.4.499.
[0472]
[16] J.V.Rundo and R.Downey,“Polysomnography,”in Handbook of ClinicalNeurology,vol.160,Elsevier B.V.,2019,pp.381–392.doi:10.1016 / B978-0-444-64032-1.00025-4.
[0473]
[17] I.M.Rosen et al.,“Clinical use of a home sleep apnea test:AnAmerican academy of sleep medicine position statement,”Journal of ClinicalSleep Medicine,vol.13,no.10,pp.1205–1207,2017,doi:10.5664 / jcsm.6774.
[0474]
[18] American Academy of Sleep Medicine Task Force,“Sleep-relatedbreathing disorders in adults:recommendations for syndrome definition andmeasurement techniques in clinical research.The Report of an American Academyof Sleep Medicine Task Force.,”Sleep,vol.22,no.5,pp.667–689,1999,doi:10.1093 / sleep / 22.5.667.
[0475]
[19] O.C.Ioachimescu et al.,“Performance of peripheral arterialtonometry–based testing for the diagnosis of obstructive sleep apnea in alarge sleep clinic cohort,”Journal of Clinical Sleep Medicine,vol.6,no.10,pp.1663–1674,Oct.2020,doi:10.5664 / jcsm.8620.
[0476]
[20] F.Massie,D.M.de Almeida,P.Dreesen,I.Thijs,J.Vranken,and S.Klerkx,“An evaluation of the Night Owl home sleep apnea testing system,”Journal ofClinical Sleep Medicine,vol.14,no.10,pp.1791–1796,Oct.2018,doi:10.5664 / jcsm.7398.
[0477]
[21] G.Pillar et al.,“Detecting central sleep apnea in adult patientsusing WatchPAT—a multicenter validation study,”Sleep and Breathing,vol.24,no.1,pp.387–398,Mar.2020,doi:10.1007 / s11325-019-01904-5.
[0478]
[22] R.Lazazzera et al.,“Detection and Classification of Sleep Apneaand Hypopnea Using PPG and SpO2 Signals,”IEEE Transactions on BiomedicalEngineering,vol.68,no.5,pp.1496–1506,May 2021,doi:10.1109 / TBME.2020.3028041.
[0479]
[23] J.Allen,“Photoplethysmography and its application in clinicalphysiological measurement,”Physiological Measurement,vol.28,no.3.Mar.01,2007.doi:10.1088 / 0967-3334 / 28 / 3 / R01.
[0480]
[24] A.A.Alian and K.H.Shelley,“Photoplethysmography,”Best Practiceand Research:Clinical Anaesthesiology,vol.28,no.4.Bailliere Tindall Ltd,pp.395–406,2014.doi:10.1016 / j.bpa.2014.08.006.
[0481]
[25] A.Malhotra et al.,“Performance of an automated polysomnographyscoring system versus computer-assisted manual scoring,”Sleep,vol.36,no.4,pp.573–582,Apr.2013,doi:10.5665 / sleep.2548.
[0482]
[26] R.B.Berry et al.,“Rules for scoring respiratory events in sleep:Update of the 2007AASM manual for the scoring of sleep and associatedevents,”Journal of Clinical Sleep Medicine,vol.8,no.5,pp.597–619,2012,doi:10.5664 / jcsm.2172.
Claims
1. A processor-readable medium having stored thereon processor-executable instructions, which, when executed by one or more processors, cause the one or more processors to distinguish a sleep disordered breathing event from a signal generated by a photoplethysmography (PPG) sensor, the processor-executable instructions comprising: receiving instructions for data representing a signal from the PPG sensor; instructions for deriving a PPG-based breathing signal from the data representing a signal from the PPG sensor; instructions for calculating first level features from the PPG-based breathing signal; instructions for calculating second level features from said first level features; instructions for evaluating said first level characteristic and said second level characteristic; as well as Instructions are generated based on the evaluation to identify an output indication of a type of sleep disordered breathing. 2 . The processor-readable medium of claim 1 , wherein the output identifying the type of sleep disordered breathing indicates a plurality of events that characterize a sleep session.
3. The processor-readable medium according to any one of claims 1 to 2, wherein the processor-executable instructions further comprise: instructions for deriving a peripheral arterial tonometry (PAT) signal from the data representing a signal from the PPG sensor; as well as Instructions for detecting one or more respiratory events based on the PAT signal. 4 . The processor-readable medium of claim 3 , wherein the output indication identifying a type of sleep disordered breathing characterizes the one or more respiratory events.
5. The processor readable medium of any one of claims 1 to 4, wherein the PPG sensor is a finger sensor.
6. The processor-readable medium of any one of claims 1 to 5, wherein the output indication identifies a central sleep apnea type.
7. The processor readable medium of claim 6, wherein the central sleep apnea type comprises mixed central and obstructive sleep apnea types.
8. The processor-readable medium of any one of claims 1 to 5, wherein the output indication identifies a type of obstructive sleep apnea.
9. The processor-readable medium of claim 1, wherein the instructions for calculating the first-level features comprise: Instructions for calculating values from a plurality of windows of the PPG-based respiratory signal.
10. The processor-readable medium of claim 9, wherein for each of the plurality of windows, the calculated value comprises any one or more of a range, an interquartile range, a variance, and a plurality of local maxima.
11. The processor-readable medium of any one of claims 9 to 10, wherein the plurality of windows comprises a first uniformly sized window that moves along the PPG-based respiration signal.
12. The processor-readable medium of claim 11, wherein the plurality of windows further comprises windows of a second uniform size moving along the PPG-based respiration signal, the second uniform size being shorter than the first uniform size.
13. The processor-readable medium of any one of claims 1 to 12, wherein the instruction to calculate the second-level features from the first-level features comprises: instructions for selecting a subset of values for each of the first level features, wherein the subset of values is associated with a subset of windows of the plurality of windows corresponding to detected occurrences of respiratory events; as well as Instructions for computing one or more values of the second level feature from the selected subset of values.
14. The processor-readable medium of claim 13, wherein the one or more calculated values of the second-level feature include any one or more of: a minimum value, an average of the lowest three values, and one or more percentile values.
15. The processor-readable medium of any one of claims 1 to 14, wherein the instructions to derive a PPG-based breathing signal from the data representing a signal from the PPG sensor comprise: instructions for filtering a baseline modulation of the data representing a signal from the PPG sensor to generate a filtered signal, and Instructions for generating an envelope signal from peaks of the filtered signal.
16. The processor-readable medium of claim 15, wherein the instructions for deriving a PPG-based respiration signal from the data representing a signal from the PPG sensor further comprise instructions for normalizing the envelope signal by subtracting a moving average window of the envelope signal and dividing the envelope signal by a moving interquartile range window of the envelope signal.
17. The processor-readable medium of any one of claims 1 to 16, wherein the evaluating is performed in a classifier, or the evaluating is performed in a classifier comprising an ensemble of tree classifiers.
18. The processor-readable medium of any one of claims 1 to 17, wherein the processor-executable instructions further comprise instructions for detecting a vasoconstriction event from the data representing a signal from the PPG sensor, and wherein the output indication is based on the detected vasoconstriction event.
19. The processor-readable medium of claim 18, wherein the instructions for detecting a vasoconstriction event include instructions for: filtering the PPG signal data representing a signal from the PPG sensor to produce a filtered PPG signal; and The filtered PPG signal is processed to detect troughs attributable to vasoconstriction.
20. The processor-readable medium of any one of claims 18 to 19, wherein the processor-executable instructions further comprise instructions to omit from the evaluation one or more values of one or both of the first and second level features associated with a detected vasoconstriction event.
21. The processor-readable medium of any one of claims 1 to 20, wherein the processor-executable instructions further comprise instructions for detecting an arrhythmic event from the data representing a signal from the PPG sensor, and wherein the output indication is based on the detected arrhythmic event.
22. The processor readable medium of claim 21, wherein the instructions for detecting an arrhythmic event include instructions for: One or more irregular peak-to-peak interval regions in the PPG signal associated with arrhythmia are detected.
23. A processor-readable medium according to claim 22, wherein the processor-executable instructions also include instructions for omitting data corresponding to at least one of the one or more respiratory events from the evaluation, wherein the at least one respiratory event is associated with one or more detected irregular peak-to-peak intervals.
24. The processor-readable medium of claim 23, wherein the instructions for detecting the one or more irregular peak-to-peak intervals comprise instructions for: Calculate the absolute value of the difference between two adjacent peak-to-peak intervals; and calculating an average of the absolute values of the differences between adjacent peak-to-peak intervals of the PPG signal; and The absolute value and the average value are compared.
25. The processor-readable medium of any one of claims 1 to 24, wherein the processor-executable instructions further comprise instructions for generating a signal for controlling operation of a respiratory therapy device based on the output indication.
26. The processor readable medium of claim 25, wherein the controlling operation comprises controlling pressure or flow therapy of a blower of the respiratory therapy device.
27. The processor-readable medium of any one of claims 1 to 26, wherein the instructions to receive the data representative of the signal from the PPG sensor comprise instructions for receiving the data at a server.
28. A server accessing a processor-readable medium according to any one of claims 1 to 27, wherein the server is configured to receive a request to download the processor-executable instructions of the processor-readable medium to a processing device via a network.
29. A processing device comprising: one or more processors; and (a) a processor-readable medium according to any one of claims 1 to 27, or (b) wherein the processing device is configured to access the processor-executable instructions using a server according to claim 28.
30. The treatment device of claim 29, wherein the treatment device is a respiratory therapy device.
31. The treatment device of claim 30, wherein the treatment device is configured to generate pressure therapy or flow therapy.
32. A method for a server to access a processor-readable medium according to any one of claims 1 to 27, the method comprising receiving at the server a request to download the processor-executable instructions of the processor-readable medium to an electronic processing device via a network; and sending the processor-executable instructions to the electronic processing device in response to the request.
33. A method of one or more processors distinguishing sleep disordered breathing events from a signal generated by a photoplethysmography (PPG) sensor, comprising: accessing, with the one or more processors, a processor-readable medium according to any one of claims 1 to 27, and The processor-executable instructions of the processor-readable medium are executed in the one or more processors.
34. A method in which one or more processors distinguish sleep disordered breathing events from a signal generated by a photoplethysmography (PPG) sensor, the method comprising: receiving data representing a signal from the PPG sensor; deriving a PPG-based breathing signal from said data representing a signal from said PPG sensor; calculating first-level features from the PPG-based breathing signal; calculating second-level features from the first-level features; evaluating the first level characteristic and the second level characteristic; as well as An output indication is generated based on the evaluation to identify a type of sleep disordered breathing.
35. The method of claim 34, wherein the output identifying the sleep disordered breathing type indicates a plurality of events characterizing a sleep session.
36. The method according to any one of claims 34 to 35, further comprising: deriving a peripheral arterial tonometry (PAT) signal from said data representing a signal from said PPG sensor; as well as One or more respiratory events are detected based on the PAT signal.
37. The method of claim 36, wherein the output indication identifying a type of sleep disordered breathing characterizes the one or more respiratory events.
38. The method of any one of claims 34 to 37, wherein the PPG sensor is a finger sensor.
39. The method of any one of claims 34 to 38, wherein the output indication identifies a central sleep apnea type.
40. The method of claim 39, wherein the central sleep apnea type comprises mixed central and obstructive sleep apnea types.
41. A method according to any one of claims 34 to 40, wherein the output indication identifies a type of obstructive sleep apnea.
42. The method of claim 34, wherein calculating the first level features comprises: Values are calculated from multiple windows of the PPG-based respiratory signal.
43. The method of claim 42, wherein for each of the plurality of windows, the calculated value comprises any one or more of a range, an interquartile range, a variance, and a plurality of local maxima.
44. The method of any one of claims 42 to 43, wherein the plurality of windows comprises a first uniformly sized window that moves along the PPG-based respiration signal.
45. The method of claim 44, wherein the plurality of windows further comprises windows of a second uniform size moving along the PPG-based respiration signal, the second uniform size being shorter than the first uniform size.
46. The method of any one of claims 34 to 45, wherein calculating the second level features from the first level features comprises: selecting a subset of values for each of the first level features, wherein the subset of values is associated with a subset of windows of the plurality of windows corresponding to detected occurrences of respiratory events; as well as One or more values of the second level feature are calculated from the selected subset of values.
47. The method of claim 46, wherein the one or more calculated values of the second level feature include any one or more of: a minimum value, an average of the lowest three values, and one or more percentile values.
48. The method of any one of claims 34 to 47, wherein deriving a PPG-based breathing signal from the data representing a signal from the PPG sensor comprises: filtering a baseline modulation of the data representing a signal from the PPG sensor to generate a filtered signal, and An envelope signal is generated from the peak values of the filtered signal.
49. The method of claim 48, further comprising wherein deriving a PPG-based respiration signal from the data representing a signal from the PPG sensor further comprises normalizing the envelope signal by subtracting a moving average window of the envelope signal and dividing the envelope signal by a moving interquartile range window of the envelope signal.
50. The method according to any one of claims 34 to 49, wherein the evaluation is performed in a classifier, or wherein the evaluation is performed in a classifier comprising an ensemble of tree classifiers.
51. The method of any one of claims 34 to 50, further comprising detecting a vasoconstriction event from the data representing a signal from the PPG sensor, and wherein the output indication is based on the detected vasoconstriction event.
52. The method of claim 51 , wherein detecting a vasoconstrictive event comprises: filtering the PPG signal data representing a signal from the PPG sensor to produce a filtered PPG signal; as well as The filtered PPG signal is processed to detect troughs attributable to vasoconstriction.
53. The method of any one of claims 51 to 52, further comprising omitting from the evaluation one or more values of one or both of the first and second level features associated with detected vasoconstriction events.
54. The method of any one of claims 34 to 53, further comprising detecting an arrhythmic event from the data representing a signal from the PPG sensor, and wherein the output indication is based on the detected arrhythmic event.
55. The method of claim 54, wherein the detecting an arrhythmic event comprises detecting one or more irregular peak-to-peak intervals in the PPG signal associated with an arrhythmia.
56. The method of claim 55, further comprising omitting from the evaluation data corresponding to at least one of the one or more respiratory events, the at least one respiratory event being associated with the detected one or more irregular peak-to-peak intervals.
57. The method of claim 56, wherein said detecting said one or more irregular peak-to-peak intervals comprises: Calculate the absolute value of the difference between two adjacent peak-to-peak intervals; and Calculating an average of the absolute values of the differences between adjacent peak-to-peak intervals of the PPG signal; as well as The absolute value and the average value are compared.
58. The method of any one of claims 34 to 57, further comprising generating a signal for controlling operation of a respiratory therapy device based on the output indication.
59. The method of claim 58, wherein the controlling operation comprises controlling pressure or flow therapy of a blower of the respiratory therapy device.
60. The method of any one of claims 34 to 59, wherein receiving the data representative of the signal from the PPG sensor comprises receiving the data at a server.
61. A processor-readable medium having stored thereon processor-executable instructions, which, when executed by one or more processors, cause the one or more processors to detect a sleep disordered breathing event from a signal generated by a photoplethysmography (PPG) sensor, the processor-executable instructions comprising: receiving instructions for data representing a signal from the PPG sensor; instructions for detecting vasoconstriction events and / or arrhythmia events from said data representing signals from said PPG sensor; as well as Instructions for evaluating the data representing signals from the PPG sensor to generate an output indication identifying a sleep disordered breathing event based on the evaluation and the detected vasoconstriction event and / or the detected arrhythmia event.
62. The processor-readable medium of claim 61, comprising said instructions to detect said vasoconstriction event, wherein said instructions to detect said vasoconstriction event comprise instructions to detect a signal dip attributable to vasoconstriction.
63. The processor readable medium of claim 62, wherein the instructions for detecting a signal dip include instructions for: filtering the PPG signal data representing a signal from the PPG sensor to produce a filtered PPG signal; and The filtered PPG signal is processed to detect troughs attributable to vasoconstriction.
64. A processor readable medium according to any one of claims 61 to 63, comprising instructions for evaluating the data representing signals from the PPG sensor, the instructions generating the output indication based on a detected vasoconstriction event.
65. The processor-readable medium of claim 64, wherein to generate the output indication based on a detected vasoconstriction event, the instructions to evaluate the data representative of a signal from the PPG sensor include instructions to ignore one or more values of one or more features calculated from the data associated with a detected vasoconstriction event.
66. A processor readable medium according to any one of claims 61 to 65, comprising instructions for detecting the arrhythmic event, wherein the instructions for detecting the arrhythmic event include instructions for detecting irregular areas attributable to an arrhythmia.
67. The processor-readable medium of claim 66, wherein the instructions to detect the irregular region attributable to an arrhythmia include instructions to detect one or more irregular peak-to-peak interval regions in the PPG signal associated with an arrhythmia.
68. The processor readable medium of claim 67, wherein the instructions for detecting the one or more irregular peak-to-peak intervals comprise instructions for: Calculate the absolute value of the difference between two adjacent peak-to-peak intervals; and calculating an average of the absolute values of the differences between adjacent peak-to-peak intervals of the PPG signal; and The absolute value and the average value are compared.
69. A processor readable medium according to any one of claims 66 to 68, comprising instructions for evaluating the data representing a signal from the PPG sensor, the instructions generating the output indication based on a detected arrhythmia event.
70. A processor-readable medium according to claim 69, wherein, to generate the output indication based on the detected arrhythmia event, the instructions for evaluating the data representing the signal from the PPG sensor include instructions for ignoring at least one respiratory event associated with the detected irregularity zone detected from the data representing the signal from the PPG sensor.
71. The processor-readable medium of any one of claims 61 to 70, wherein the processor-executable instructions further comprise instructions for generating a signal for controlling operation of a respiratory therapy device based on the output indication.
72. A processor readable medium according to claim 71, wherein the control operation includes controlling pressure or flow therapy of a blower of the respiratory therapy device.
73. A processor-readable medium according to any one of claims 61 to 72, wherein the instructions to receive the data representing the signal from the PPG sensor include instructions for receiving the data at a server.
74. A server accessing a processor-readable medium according to any one of claims 61 to 73, wherein the server is configured to receive a request to download the processor-executable instructions of the processor-readable medium to a processing device over a network.
75. A processing device comprising: one or more processors; And (a) a processor-readable medium according to any one of claims 61 to 73, or (b) wherein the processing device is configured to access the processor-executable instructions using a server according to claim 74.
76. The treatment device of claim 75, wherein the treatment device is a respiratory therapy device.
77. The treatment device of claim 76, wherein the treatment device is configured to generate pressure therapy or flow therapy.
78. A method for a server to access a processor-readable medium according to any one of claims 61 to 73, the method comprising receiving at the server a request to download the processor-executable instructions of the processor-readable medium to an electronic processing device via a network; and sending the processor-executable instructions to the electronic processing device in response to the request.
79. A method in which one or more processors detect a sleep disordered breathing event from a signal generated by a photoplethysmography (PPG) sensor, comprising: accessing, with the one or more processors, a processor-readable medium according to any one of claims 61 to 73, and The processor-executable instructions of the processor-readable medium are executed in the one or more processors.
80. A method of detecting a sleep disordered breathing event from a signal generated by a photoplethysmography (PPG) sensor, the method comprising: receiving data representing a signal from the PPG sensor; detecting vasoconstriction events and / or arrhythmia events from said data representing signals from said PPG sensor; as well as The data representing signals from the PPG sensor are evaluated to generate an output indication identifying a sleep disordered breathing event based on the evaluation and the detected vasoconstriction event and / or the detected arrhythmia event.
81. The method of claim 80, comprising said detecting said vasoconstriction event, wherein detecting said vasoconstriction event comprises detecting a signal dip attributable to vasoconstriction.
82. The method of claim 81 , wherein detecting a signal dip comprises: filtering the PPG signal data representing a signal from the PPG sensor to produce a filtered PPG signal; as well as The filtered PPG signal is processed to detect troughs attributable to vasoconstriction.
83. A method according to any one of claims 80 to 82, comprising said evaluating said data representative of a signal from said PPG sensor to generate said output indication based on a detected vasoconstriction event.
84. The method of claim 83, wherein to generate the output indication based on a detected vasoconstriction event, the evaluating the data representative of a signal from the PPG sensor comprises ignoring one or more values of one or more features calculated from the data associated with a detected vasoconstriction event.
85. The method of any one of claims 80 to 84, comprising said detecting said arrhythmic event, wherein said detecting said arrhythmic event comprises detecting an irregular zone attributable to an arrhythmia.
86. The method of claim 85, wherein said detecting the irregular region attributable to arrhythmia comprises detecting one or more irregular peak-to-peak interval regions in the PPG signal associated with an arrhythmia.
87. The method of claim 86, wherein said detecting said one or more irregular peak-to-peak intervals comprises: Calculate the absolute value of the difference between two adjacent peak-to-peak intervals; and Calculating an average of the absolute values of the differences between adjacent peak-to-peak intervals of the PPG signal; as well as The absolute value and the average value are compared.
88. A method according to any one of claims 85 to 87, comprising evaluating said data representing a signal from said PPG sensor to generate said output indication based on a detected arrhythmic event.
89. A method according to claim 88, wherein, in order to generate the output indication based on the detected arrhythmia event, the evaluating the data representing the signal from the PPG sensor includes ignoring at least one respiratory event associated with the detected irregularity zone detected from the data representing the signal from the PPG sensor.
90. The method of any one of claims 80 to 89, further comprising generating a signal for controlling operation of a respiratory therapy device based on the output indication.
91. The method of claim 90, wherein the control operation comprises controlling pressure or flow therapy of a blower of the respiratory therapy device.
92. The method of any one of claims 80 to 91, wherein said receiving said data representative of a signal from said PPG sensor comprises receiving said data at a server.
Citation Information
Cited By
Obtaining breath-related sounds from audio recordings
CN116471988A
Obtaining breathing-related sounds from audio recordings
CN116471988B
