Human detection and identification based on characteristic signals

Abstract

One or more sensors are configured to detect characteristics of moving objects and living subjects for human identification or verification. One or more processors, such as those in the sensor system or controlling the sensors, may be configured to process signals from the one or more sensors to identify a person. The processing may include estimating characteristics from the signals, such as respiratory rate, respiratory depth, degree of movement, and heart rate. The sensors may be radio frequency non-contact sensors with automatic detection control to change detection control parameters based on the identification of a living organism, thereby avoiding sensor interference.

Description

[0001] Divisional Application Declaration

[0002] This application is a divisional application of the Chinese Patent Application for Invention No. 201680026853.X, filed on November 8, 2017, which entered the Chinese State Phase on November 8, 2017, from the PCT International Application No. PCT / EP2016 / 058789, filed on April 20, 2016, with the title "Detection and Identification of Humans by Signature Signals", and which claims priority to U.S. Provisional Patent Application No. 62 / 306, 1 12, filed March 7, 2016. TECHNICAL FIELD

[0003] The present technology relates to circuits and sensors for detecting features of moving objects and living subjects. More particularly, the present technology relates to such sensors in, for example, health monitors, such as range-gated radio frequency motion sensing, with particular focus on signal features for human identification or verification. BACKGROUND

[0004] Continuous wave (CW) Doppler radar motion sensors emit a continuous wave radio frequency (RF) carrier and mix the transmitted RF with the returned echo to produce a difference frequency equal to the Doppler shift produced by a moving target. These sensors do not have a well-defined range gate (i.e., they can receive signals from both near and far objects, where the received signal is a function of the radar cross section). This can result in false triggers, i.e., motion artifact interference. These sensors can also have undesirably high sensitivity at close range, resulting in false triggers.

[0005] Pulse Doppler motion sensors are described in U.S. Patent No. 4,197,537 to Follen et al. A short pulse is transmitted and its echo is mixed with the transmitted pulse. The pulse width defines a range-gated region. When the transmitted pulse ends, the mixing ends and the target returns, arriving after the end of the transmitted pulse, which is not mixed and thus gated out.

[0006] A differential pulse Doppler motion sensor, disclosed in U.S. Patent No. 5,966,090 to McEwan, "Differential Pulse Radar Motion Sensor", alternately transmits at two different pulse widths. It then subtracts the Doppler response from each width to produce a range-gated "Doppler" sensing region with a relatively constant response versus range.

[0007] Pulse radars such as those described in U.S. Patent No. 5,361,070, "Ultra-Wideband Radar Motion Sensor," to McEwen produce an extremely narrow sensing region related to the width of the transmitted pulse. A dual pulse Doppler radar motion sensor as described in U.S. Patent No. 5,682,164, "Pulse Homodyne Field Disturbance Sensor," to McEwen transmits a first pulse and generates a second pulse mixed with echoes from the first pulse after a delay. Thus, a range-gated sensing band is formed with a defined minimum and maximum range. UWB radar motion sensors have the disadvantage of not being globally RF management acceptable as active emitters. These sensors also have difficulty in sensing objects at intermediate ranges and in some embodiments are susceptible to RF interference.

[0008] A modulated pulse Doppler sensor is described in U.S. Patent No. 6,426,716 to McEwen. A range-gated microwave motion sensor includes adjustable minimum and maximum detection ranges. The device includes an RF oscillator with associated pulse generation and delay elements to produce a transmit and mixer pulse, a single transmit (TX) / receive (RX) antenna or a pair of separate TX and RX antennas, and an RF receiver including a detector / mixer with associated filtering, amplification, and demodulation elements to produce a range-gated Doppler signal from the mixer and echo pulses.

[0009] In U.S. Patent No. 7,952,515, McEwen discloses a special holographic radar. The patent adds range gating to a holographic radar to limit the response to a particular down target zone. McEwen states that a clearer, more cluttered radar hologram of the imaged surface can be obtained, particularly when penetrating materials to image internal image planes or segments. Range gating enables a stacked hologram technique where multiple imaged surfaces can be stacked in the down target direction.

[0010] In U.S. Patent No. 7,994,968, McEwen discloses an RF amplitude sampler for a holographic radar. McEwen describes that the RF amplitude sampler can finely resolve the interference patterns produced by a narrowband holographic pulse radar.

[0011] In U.S. Patent Application Publication No. 2014 / 0024917, McMahon et al. describe a sensor for physiological sensing configurable to generate an oscillating signal for emitting radio frequency pulses for range-gated sensing. The sensor can include a radio frequency transmitter configured to emit pulses and a receiver configured to receive reflected pulses of the emitted radio frequency pulses. The received pulses can be processed to detect physiological features such as motion, sleep, respiration, and / or heartbeat.

[0012] Wearable wristbands such as the Nymi introduce another parameter that can be used to authenticate the system - the ECG shape (morphology) that is specific to each user. (https: / / www.nymi.com / )

[0013] US 20100191076 Al (Aaron Lewicke, Yi Zhang, John D. Hatlestad) describes day / night respiration rate monitoring.

[0014] US 8232866 (William R. McGrath, Ashit Talukder) describes remote long range biometric identification using microwave electrocardiogram signals.

[0015] US 832100 (Seth Snyder, Jasper Speicher) describes a biometric data display system and method.

[0016] WO 2003000015 (Mark D Wiederhold, Rodney P Meyer, Steven A Israel, John MIrvine) describes identification by analysis of physiometric changes.

[0017] US 8762733 (P. Alexander Derchak, Lance Myers) describes systems and methods for identity confirmation using physiological biometrics to determine a physiological fingerprint.

[0018] US 20110040574 (Ho Chung Nicholas Fung, Chu Yong Sang) describes a health monitoring system using biometric identification.

[0019] US 6993378 (Mark D. Wiederhold, Steven A. Israel, Rodney P. Meyer, John M. Irvine) describes identification by analysis of physiometric changes.

[0020] There can be a need to improve sensors for sensing such features used to identify or authenticate and / or their signal processing.

[0021] It is desirable to be able to identify a person for authentication and other purposes.

[0022] The advent of wearable and non-contact physiological and behavioral data capture has led to the need to detect and identify a particular person from their person biological features "fingerprint" both excluding data (artifacts) from another person and ensuring the compliance or use of those sensors (and potentially attachment or incorporation of services or treatments). Such features can be derived from physiological and behavioral signals.

[0023] In addition to detecting that sensor data has been collected from a particular user and their micro- and macro-environment, it is desirable for a system to detect deviations from normal (healthy) signals and to be tolerant of such changes (e.g. deteriorating conditions or improved conditions as a result of a treatment / therapy). SUMMARY

[0024] One aspect of some embodiments of the present technology relates to a sensor for detecting physiological features such as using radio frequency signals.

[0025] Another aspect of some embodiments of the present technology relates to a sensor for authenticating a person from detected signal features such as using radio frequency signals.

[0026] Some versions of the present technology can include a method or system for monitoring physiological parameters of one or more persons, such as identifying a person for health monitoring. The system can include one or more sensors for monitoring physiological parameters of one or more persons. The system can include one or more processors configured to process signals from the one or more sensors to identify a person, the processing including estimating characteristics, including respiratory characteristics, cardiac characteristics, or movement characteristics, such as any one or more of detected respiratory rate, detected respiratory depth, detected degree of movement, and detected heart rate, the estimating involving analyzing any one or more of the detected characteristics.

[0027] In some versions, the one or more sensors include radio frequency non-contact sensors. The detecting can include detecting physiological features during sleep of a person. The processing can include detecting sleep stages. The processing can include detecting deep sleep. The processing can include detecting REM sleep. The detecting can include detecting physiological features during wake hours of a person. The one or more sensors can include any one or more of the sensors described throughout the detailed description. The estimating of the one or more processors can include detecting and analyzing any one or more features of the sensor signals described throughout the detailed description.

[0028] In some versions, the one or more sensors can be configured to minimize interference, such as radio frequency (RF) interference, between at least two of the one or more sensors. The one or more sensors can be configured to minimize interference, such as RF interference, by modifying control parameters of the distance gating pulse time, the transmit power level of a pulse, such as an RF pulse, the detection frequency of a pulse, such as an RF pulse, and / or adjusting the positioning of a controllable antenna. The one or more sensors can communicate via wired or wireless circuitry.

[0029] In some versions, a biometric parameter can be applied to dynamically adjust the performance of the one or more sensors in order to optimize for physiological cognition of independent human origin and to exclude other origins. The system can include a control processor in communication with the one or more sensors, the control processor in communication with the one or more sensors to adjust detection control parameters of the one or more sensors based on the identification of the person or animal by the one or more sensors. The detection control parameters of the one or more sensors can include one or more of distance gating, RF center frequency, and RF power level.

[0030] In some versions, the system can include a control processor in communication with the one or more sensors. The control processor can be configured to initiate another sensor system based on the identification of the person or animal by the one or more sensors. The other sensor system can include a video camera. Optionally, in the system, based on the identification that the person is a previously monitored person, the processor records or can record (e.g., such as record data related to the person's identity in a database) the detected biometric. In some versions of the system, based on the identification that the person is not a previously monitored person, the processor refrains or can be configured to refrain from recording the detected biometric. In some versions, the processor can be configured to initiate biometric detection of a particular person. Optionally, the processor can compare a newly detected biometric to the initiated biometric to identify the person. The evaluation can include a comparison between the newly detected biometric and the initiated biometric.

[0031] Optionally, the evaluation can include classifying characteristics determined from the signal. The characteristics can include one or more of: a spectral peak ratio; a set optimizer flag vector; a peak to valley ratio; a filtered respiration rate; a respiration variation measure; an in-band power of the sensor signal; a distance of the sensor signal; a final respiration rate; a ratio of a maximum amplitude to a minimum amplitude of a respiration cycle; a high-band power of the sensor signal; an average respiration rate; a periodic leg movement activity detection; a roll over detection or roll detection; and a post-processed movement.

[0032] Optionally, the estimation may include classifying the characteristics determined by the signal, wherein these characteristics include one or more of the following: cardiac parameters, skin conductance parameters, exercise intensity parameters, respiratory parameters, blood pressure parameters, cough parameters, snoring parameters, and sleep parameters.

[0033] The estimate may include a comparison of the identified characteristic with historical characteristics. In some versions, the estimate may further include calculating the mean and / or standard deviation over a period of time from the identified characteristic.

[0034] In some versions, one or more processors of the monitoring system estimate received data detected by one or more sensors from another monitoring system. The estimation of the received data may include determining sensing equivalence among multiple health monitoring devices. Each monitoring device may include a classifier and a subject classifier, each configured to estimate characteristics from the signal to identify a person. The subject classifier may be further configured to estimate historical characteristics to identify a person. The health monitoring device or system may further include a camera sensitive to infrared light and infrared emitters. In some versions, the monitoring system, such as one or more processors of the health monitoring device, may be configured to detect events from the signal and associate the event with a specific portion of video obtained using the camera that includes the event. The monitoring device or system may further include a battery and a coil for wirelessly charging the battery.

[0035] In some versions, the one or more processors may be configured to control the one or more sensors to change sensor detection power, sensor frequency, sensor distance gating, or other control parameters used for sensing when biometrics of an indicating animal are detected. The one or more processors may be configured to access setting parameters of any one or more light, sound, and / or environmental devices based on identification of a person associated with the setting parameters. The one or more processors may be configured to trigger settings for any one or more light, sound, and / or environmental devices using setting parameters associated with detected biometrics of the identified person.

[0036] In some versions, the system may include a central controller that communicates with the collection of sensors. The central controller may be configured to provide confirmation of the location of identified persons within a building. In some versions, the one or more sensors may include night lighting. The one or more sensors may include a pass-through outlet. The one or more sensors may include an AC plug and an AC power pass-through outlet for powering the sensor. The one or more sensors may include a USB plug and a USB pass-through adapter for powering the sensor. The one or more sensors may include a network interface for wireless or wired network communication.

[0037] In some versions, the one or more processors can be configured to set the operation of the respiratory therapy device based on the identified person. This setting operation may allow treatment to be performed using the respiratory therapy device. This setting operation may alter or change the treatment provided by the respiratory therapy device.

[0038] Optionally, the one or more processors may be configured to retrain to identify the person when processing the biometrics estimated in the identification via a respiratory therapy device. The one or more processors may be configured to adjust the operation of the one or more sensors when determining the quality of the sensor signals. The one or more processors may be configured to identify the person based on the quality of the detected biometrics, depending on the specific biometrics. The one or more processors may be configured to run a registration process to initialize a baseline of biometrics for person identification. This registration process may include a guided breathing phase or a spontaneous breathing phase. In some versions, the one or more processors are configured to exclude biometrics involved in the identification of the person when radio frequency interference is detected.

[0039] In some versions, one or more processors of the system can be configured to set the operation of the alarm based on the identified person. The one or more processors can be configured to identify the primary user from one or more other users. The system can be configured to track parameters of one or more users over time to establish classification characteristics based on at least one of one or more respiratory, cardiac, or movement characteristics. In some such cases, at least one of the one or more respiratory, cardiac, or movement characteristics includes at least one of the following respiratory parameters: distance, changes between breaths, shape, and inspiratory-to-expiratory ratio.

[0040] Optionally, the one or more processors may be configured to classify the user's identity based on features determined during the classification process. This classification process may include any one or more of the following: neural networks, hidden-layer Markov models, logistic regression processing, linear kernel support vector machines, and radial kernel support vector machines. The classification process may include principal component analysis of features or sets of features prior to classification. The classification process may include real-time and offline features. The classification process may include late integration of multiple classifiers and their outputs, or the outputs of classifiers, to produce output posterior probabilities, such as probabilities regarding the identity of a user or primary user.

[0041] In some cases, the system or method may include multiple sensors configured to detect the same or different persons, and the system may automatically adjust parameters of the multiple sensors, such as sensing control parameters, including at least one of the following: distance, power, frequency, detection direction, and radiometric pattern.

[0042] Other aspects, features, and advantages of this technology will become clear from the following detailed description taken in conjunction with the accompanying drawings, which are part of this disclosure and illustrate the principles of the technology by way of example. Other aspects of this technology will become clear from the appended claims. Attached Figure Description

[0043] Other exemplary embodiments of this technology will now be described with reference to the accompanying drawings, in which:

[0044] Figure 1 This is an illustration of an exemplary detection device applicable to some versions of radio frequency sensors using this technology;

[0045] Figure 2A This is a simplified diagram illustrating the conceptual structure and process flow for estimating sensor signals in some versions applicable to this technology;

[0046] Figure 2B This is a diagram illustrating further processing of sensor signals used to detect exemplary physiological indicators;

[0047] Figure 3 and Figure 4 A simplified diagram shows the processing system and workflow used to calculate human sleep and daytime biometrics, respectively.

[0048] Figure 5 The process used to calculate characteristic properties in order to generate a biometric fingerprint is illustrated;

[0049] Figure 6 and Figure 7 The system processes for registration (training) and subsequent authentication or exclusion are shown.

[0050] Figure 8 The heart rate and respiratory rate signals of subject "User A" during sleep (approximately 8.3 hours) are shown, which were recorded using a non-contact pulsed continuous wave RF sensor at a distance of <1.8 m;

[0051] Figure 9 Show Figure 8 Histograms of heart rate and respiratory rate of subject "User A" during sleep (approximately 8.3 hours), recorded using a non-contact pulsed continuous wave RF sensor at a distance of <1.8 m;

[0052] Figure 10 The study shows heart rate and respiratory rate signals of subject "User B" during sleep (only more than 6 hours), which were recorded using a non-contact pulsed continuous wave RF sensor at a distance of <1.8 m.

[0053] Figure 11 Show Figure 10Histograms of heart rate and respiratory rate of subject "User B" during sleep (approximately 8.3 hours), recorded using a non-contact pulsed continuous wave RF sensor at a distance of <1.8 m;

[0054] Figure 12 The cross-spectral density between HR and BR is shown for a rhythmic respiratory rate of 0.1 Hz. Registration may include rhythmic respiratory periods as well as spontaneous breathing;

[0055] Figure 13 This is a simplified diagram illustrating two sensors monitoring two humans in a faulty configuration. Biometric quality is reduced because these sensors are receiving superimposed respiratory and heart rate signals from each person and are detecting movement in both individuals.

[0056] Figure 14 This is a simplified diagram illustrating two sensors monitoring two humans. The biometric quality is excellent because these sensors are configured to minimize distance and power in order to maximize clear, separate biometric data.

[0057] Figure 15 This is a simplified diagram illustrating two sensors monitoring a human in a faulty configuration. Biometric quality is degraded, so sensor _a "sees" the human _b and detects duplicate biometrics in at least one of these sensors.

[0058] Figure 16 This is a simplified diagram illustrating two sensors in a configuration that monitor a human to minimize distance and power in order to maximize clear, distinct biometrics. Sensor _a enters a power-saving / search mode. Sensor _b detects the biometrics of human _b.

[0059] Figure 17 This is a simplified diagram showing two sensors monitoring a bedroom where an animal (dog) has entered the room and is lying on the bed. Sensor _a enters a power-saving / search mode, while sensor _b monitors non-human heart rate and respiratory rate signals.

[0060] Figure 18 This is a simplified diagram illustrating three sensors monitoring three humans in the presence of another living organism (e.g., a dog).

[0061] Figure 19 The process of an exemplary classification system for identifying users is shown.

[0062] Figure 20 The process of an exemplary classification system for identifying two (or more) users in a sleep (or wake) environment is shown, which compares sensor data to check whether the detected user is the expected user on that side of the bed.

[0063] Figure 21A A system according to the present technology is shown. A patient 1000 wearing a patient interface 3000 receives a supply of pressurized air from an RPT device 4000. The air from the RPT device 4000 is humidified in a humidifier 5000, and the air is directed to the patient 1000 along an air circuit 4170. A bed companion 1100 is also shown.

[0064] Figure 21B An RPT device 4000 with a nasal mask-type patient interface 3000 is shown for use on a patient 1000.

[0065] Figure 21C The RPT device 4000 with a full-face mask-type patient interface 3000 is shown for use on patient 1000.

[0066] Figure 22 A non-invasive patient interface 3000 in the form of a nasal mask is shown.

[0067] Figure 23A An RPT device 4000 according to one form of the present technology is shown.

[0068] Figure 23B A schematic diagram of the pneumatic circuit of one form of the RPT device 4000 according to the present technology is shown. Upstream and downstream directions are indicated.

[0069] Figure 23C A schematic diagram of the electrical components of an RPT device 4000 according to one aspect of the present technology is shown.

[0070] Figure 23D A schematic diagram is shown of an algorithm 4300 implemented in an RPT device 4000 according to one aspect of the present technology. Figure 23D In the image, the solid arrow indicates the actual flow of information, such as via electrical signals.

[0071] Figure 23E This is to illustrate one aspect of the technology. Figure 23D The flowchart of the method 4500 executed by the treatment engine module 4320.

[0072] Figure 24 The humidifier 5000 is shown. Detailed Implementation

[0073] 1. Background

[0074] Looking ahead to the future of healthcare and human health, three interconnected categories are evident: (i) the trend of collecting and monitoring data from our bodies; (ii) health budgets under pressure; and (iii) medical automation.

[0075] As people become more health-educated (early pioneers in the 'self-quantification' movement), they also live longer, requiring more physicians, and there is some evidence that the 'power curve' is shifting towards the consumer. For products to succeed in this space, they must attach themselves to people's lives and become indispensable. Prevention is cheaper than cure, and the health system must shift towards outcome-based measurements—closer to the ancient Chinese system, where doctors were paid for keeping people healthy, not for being paid when they were sick (i.e., the opposite of the typical Western model of paying for medicine).

[0076] Consumer ambiguity and the availability of medical devices and services have led to the automation of medication and the emergence of expert systems. These changes facilitate the ability to monitor chronic diseases at home, as payers (e.g., insurance companies) can seek to bridge funding gaps and proactively manage hospitalizations in a planned manner.

[0077] 2. User Identity

[0078] Therefore, there is a need for systems that can identify specific patterns of users (i.e., identify users, referred to here as general biometric “fingerprints”) to prevent imitation and also detect abnormal changes in the biometric and behavioral data of the identified users.

[0079] It should be noted that, unlike real human fingerprints or possible iris scans, the “biological fingerprint” referred to in this work must be a vague (less precise) estimate of the user’s identity and include both biometric and behavioral aspects. These physiological and behavioral parameters are expected to gradually develop for the user over time (e.g., as they become more ill, become healthier, and undergo other life changes).

[0080] The system outlined herein is designed to detect specific patterns in a user. Examples of uses include verifying a user, potentially confirming compliance with prescription treatments and / or authorizing payments or other motivations to improve or manage a user's health and fitness levels, or promoting treatment compliance (e.g., reducing insurance premiums or eligibility for payments or reducing burden). In some cases, detection of a specific user can be used to ensure that the current user being tested for biometric health parameters is the same as the previous user whose health parameters were previously tested (e.g., between different periods of sensor use). In this case, analysis of health parameters from historical tests across multiple periods can confirm a single user without mixing detected health parameters from different users. Therefore, sensor records of detected parameters can be based on biometric confirmation / verification of previously identified users.

[0081] This technology relates to methods (such as using software algorithms) and systems for reading data from one or more sensors and services, processing the data, and adjusting system parameters based on newly collected data and previously collected datasets. Some or all of these sensors aggregate physiological data readings from the system's end user.

[0082] One example of this software algorithm and system may include a smartphone with embedded sensors such as an accelerometer, compass, Global Positioning System (GPS) (positioning / location awareness), and a heart rate monitoring device for step-by-step counting. Other exemplary systems may include a smartwatch containing sensors such as an accelerometer (for measuring movement), a heart rate monitor (e.g., using optical or other methods), a galvanic skin response (GSR) measurement device, a blood pressure monitor (e.g., derived from photoelectric pulse wave signals), and / or a respiration sensor (e.g., based on a wireless RF biomotion sensor discussed above or herein).

[0083] Additional data can be collected from the user via wireless lines through the 'Lab-on-a-Chip' bonding module, or directly from the drug dispenser via identification chips such as RFID (Radio Frequency ID), or from measuring devices such as spirometers (maximum flow meters).

[0084] 3. Detection of physiological and behavioral information

[0085] In one embodiment, the device may be placed beside the bed (or inside, under, or on the bed, or worn by the user) to collect physiological parameters of the patient while in bed, such as heart rate, respiratory rate, respiratory depth (e.g., shallow / short or deep), degree of movement, and other relevant parameters.

[0086] This system can operate continuously without user input. For example, the device can have non-triggered operation, allowing the sensors to sense (e.g., 24 / 7 / 12) whether a user is nearby (e.g., at all times). In this regard, devices such as distance-gated RF sensors can continuously monitor a space (e.g., a bedroom) to distinguish between the primary user and one or more other users. The system can track user parameters that change over time to establish characteristic classifications based on movement patterns and breathing (e.g., distance, changes between breaths, shape, inhalation and exhalation ratios). Therefore, the system can be configured to establish classification characteristics.

[0087] The system can lock onto the primary (dominant) or initial user over time and establish differentiating characteristics. Additionally, the system can track user habits, such as which side of the bed the user sleeps on and the time the user goes to bed each day of the week. The system can use classifiers, such as neural networks (e.g., hidden-layer Markov models), or, for a simpler implementation, logistic regression, to determine categorical characteristics. Offline and real-time subject categorical characteristics allow the system to determine the probability of detecting which person, the user or another user, such as the user's partner. Offline processing also allows the system to be retrained based on real-time parameters that can be computed / determined during the sensing / monitoring period. In this regard, real-time classification can occur when a user is sensed within the sensor's sensing range. Offline processing (such as using offline parameters that can be computed / determined after the sensing / monitoring period) can occur when the user is no longer within the sensor's range using previously sensed data. Offline characteristics (or multiple offline characteristics) or offline parameters are the product of offline processing. For example, instead of being calculated and / or classified based on the most recent parameters (“real-time”), the offline processing step allows for post-hoc analysis over the entire monitoring / sensing period (e.g., one night) or multiple periods (e.g., multiple nights). This processing can occur, for example, after the bed partner wakes up and leaves the bedroom / sensor area. This allows for a comprehensive review of all data collected during the night / sleep period. Real-time characteristics or parameters can be calculated and / or applied for classification using recent data, such as during a specific sensing / monitoring period.

[0088] Other techniques, such as linear kernel or radial kernel support vector machines (SVMs), can be implemented for classification. The computation of these features can be optimized by “bleaching” the feature set using principal component analysis (PCA) before classification in which many very similar features are used (i.e., reducing any redundant data, such as dimensional data, before further processing). Multiple classifiers can be used, which use the “late integration” of the outputs to form the output posterior probabilities.

[0089] When applied to multiple sensors, sensor parameters such as distance, power, frequency, detection orientation, and / or emission patterns can be automatically adjusted to support areas where many people may be present, such as patients in a hospital ward, by detecting the same or different people by multiple sensors (e.g., placed on either side of a bed) and sharing data over a network. In this regard, emission patterns relate to the detection pattern or detection area (i.e., the three-dimensional (3D) sensing space—including any rear flaps that may be present behind the sensor). Detection patterns can relate to distance and orientation. Sensors can be adjusted to regulate their distance (e.g., using near-range and far-range gating), their emission power level (which also affects distance, as even within far-range gating, the SNR (signal-to-noise ratio) may be lower, making effective further detection closer to the previous point). Power can also relate to energy conservation through low-power / battery use when there are no people in the sensing environment. Radio frequency dielectric resonant oscillators (DROs) can consume significant power if used with sensors. The frequency can be changed for coexistence (e.g., to avoid interference) or regulatory requirements of different countries. Automated detection orientation allows for improper device setup (e.g., if the first user did not optimally specify it, the system can automatically adjust to a suitable setting without prompting the user to move it). For example, the sensor may have a configurable antenna capable of dynamically changing its frequency and radiation characteristics in a controlled and reversible manner.

[0090] Therefore, the system can be configured to detect and verify users by detecting parameters such as movement, breathing, and / or heart rate using non-contact sensors such as RF.

[0091] For example, such as Figure 1 As shown, some embodiments of this technology can implement a sensing or detection device 100 suitable for detecting physiological characteristics of a user or patient in the vicinity of the sensing or detection device. The sensor can be a standalone sensor or can be coupled to other devices, such as a respiratory therapy device, to provide an automated therapeutic response based on analysis of physiological characteristics detected by the sensor of that device. For example, a respiratory therapy device with a controller and a flow generator can be configured to have this sensor and can be configured to adjust the pressure therapy generated at a patient interface (e.g., a mask) in response to physiological characteristics detected by the sensor. Exemplary respiratory therapy devices are described in International Patent Application No. PCT / US2015 / 043204, filed July 31, 2015, the entire disclosure of which is incorporated herein by reference.

[0092] A typical sensor in this device may employ a transmitter to emit radio frequency waves, such as those used for distance gating sensing. Optionally, a receiver included in a combined device with a transmitter may be configured to receive and process reflected versions of these waves. Signal processing, such as using a processor that activates the sensor with the device, may be employed to deduce physiological characteristics based on the received reflected signals.

[0093] For example, such as Figure 2A As shown, the transmitter emits a radio frequency (RF) signal toward the subject, such as a human. Typically, the RF signal originates from a local oscillator (LO). The reflected signal is then received, amplified, and mixed with a portion of the initial signal; the output of this mixer can then be filtered. The resulting signal may contain information about the person's movement, breathing, and heart activity, and is referred to as the raw motion sensor signal.

[0094] Figure 2B This is a simplified diagram illustrating some potential processing of raw sensor signals to produce indications of physiological characteristics. The raw signal will typically contain components reflecting a combination of body movement, respiration, and cardiac activity. Body movement can be identified, for example, using zero-crossing or energy envelope detection algorithms (or more complex algorithms), which can be used to form "start of movement" or "stop of movement" indicators. For example, such movement detection algorithms can be implemented according to the method disclosed in U.S. Patent Application Publication No. 2009 / 0203972, the entire disclosure of which is incorporated herein by reference. Respiratory activity is typically in the range of 0.1 to 0.8 Hz and can be derived by filtering the initial signal using a bandpass filter having a passband in this region. Cardiac activity is reflected as a higher frequency signal and can be approximated by filtering with a bandpass filter having a passband in the range of 1 to 10 Hz.

[0095] This breathing and movement sensor can be a distance-gated RF motion detector. The sensor can be configured to receive a DC power supply input and provide four analog motion channel outputs, which have synchronized and orthogonal components of the breathing and movement signals of a person within the detection range. In the case of a pulsed RF motion sensor, distance gating can help limit motion detection to only preferred areas or ranges. Therefore, detection using the sensor can be within a defined distance of the distance sensor.

[0096] As another example, the type of sensor used can be radio frequency (RF) based, such as the ResMed SleepMinder family of non-contact sensors, which uses at least some of the technologies described in the McEwan and McMahon patent documents mentioned above. Measurement and data processing techniques described in International Publications WO2010 / 036700, WO2010 / 091168, WO2008 / 057883, WO2007 / 143535, and WO2015 / 006364, which are incorporated herein by reference, can also be used. Alternatively, alternative technologies can be used, utilizing accelerometers, piezoelectric or UWB (ultra-wideband radio frequency) pads, passive infrared, or other optical devices such as cameras with skin color and motion detection. For example, the device may implement motion detection algorithms or sensor technologies based on any of the methods and sensors disclosed in any of the following: U.S. Patent Application Publication No. 2009 / 0203972, International Patent Application No. PCT / US14 / 045814; U.S. Provisional Patent Application No. 62 / 149,839, filed April 20, 2015, and U.S. Provisional Patent Application No. 62 / 207,670, filed together with it on the same date, the entire disclosure of which is incorporated herein by reference. Furthermore, any radio frequency sensor described in U.S. Provisional Patent Application No. 62 / 205,129, filed August 14, 2015, may be implemented in any version of this technology, the entire disclosure of which is incorporated herein by reference.

[0097] Microphones can also be used to monitor and classify sound patterns consistent with breathing rate, chronic cough, or snoring, and to separate these sounds from background noise such as fans, road noise, and similar noises. This is known as “nighttime sleep” monitoring, although it can also be performed, for example, during daytime naps; it is targeted at the user in a sleep environment. Additionally, breathing rate, heart rate, and other such physiological signals can also be detected using alternating sensors, such as those based on non-contact or contact-related technologies.

[0098] Once a user is out of bed, their physiological parameters can optionally be monitored via sensors worn on the body. This could be, for example, a clip-on device, a stick-on-patch device (leather or lab-on-a-chip), or a wristwatch-style device; the collected sensor parameters include movement and walking (via accelerometer), position, conductance skin response (GSR), heart rate and respiratory rate (via optical, electrical, or mobile devices), and some or all of other relevant parameters. This is known as “daytime / wake-up monitoring,” and it is targeted at users who are awake and active. Such devices can be clip-on earrings, watch or wristband style elements, such as Fitbit, Basis, Apple, Withings, or other products.

[0099] Biometric data from one or more sensors can also be fused with video / camera image data to further increase confidence in "activity" (i.e., live human) detection or to identify specific users using techniques such as facial recognition, skin color detection, and / or micro-outburst detection. This fusion of data adds another factor to the analysis. For example, skin color detection can be used as a first step in facial recognition, gesture recognition, and other applications, and is robust to scaling, rotation, and occlusion.

[0100] In some embodiments, respiratory rate during the day and / or night can be captured by wearable respiratory sensors, such as those clipped to a waistband or bra, a chest strap (e.g., a spirit.io device), an accelerometer in a nasal cannula, or by extracting waveforms from PPG (optical plethysmography) signals. This data may also be used for biometric identification as described herein.

[0101] It should be noted that pressure and other data from the implantable device (i.e., implanted within the person being monitored) can be combined with values ​​collected by the system. Additionally, baseline data collected outside the bedroom can be correlated. Therefore, in the biometric verification / identification of a person / user, data from one or more sensors can be estimated, such as by establishing a baseline of detected biometric data attributable to a specific user and / or by comparing newly detected data with previously collected data to confirm that the newly detected data is attributable to that specific user. Thus, the system can compare current parameters with historical parameters.

[0102] This system can aggregate and / or combine data from contact sensor counts (e.g., counts from body-worn sensors) and non-contact sensors (e.g., Doppler, RF sensors), as well as video / camera image data from a video detection system. Therefore, the system may be able to collect data 24 / 7 or a portion thereof to suit a user's lifestyle.

[0103] The algorithms and methods described herein can be implemented on one or more processors accessing database storage devices, such as computing devices (e.g., PCs, servers, cloud services, smart device applications, or variants). The algorithm may be based on a classification system analyzing one or more data sources and may utilize historical data stored in the database. Rule thresholds, templates, and / or storage models may vary based on adaptive probability weights, user-specific, and group-based demographic data.

[0104] 3.1 Sensor Feedback

[0105] This system can provide feedback loops to users or third parties (e.g., via PCs or smart devices such as smartphones or tablets), and optionally to monitoring centers for review, validation, and optional intervention. The system can also provide feedback on the therapeutic effects of treatment devices, such as continuous positive airway pressure (CPAP), adaptive servo ventilation (ASV), and / or biphasic positive airway pressure (BIP) machines used to assist users with various respiratory problems. Such devices are described in more detail herein.

[0106] In some embodiments, the sensor may be integrated with or incorporated into respiratory therapy devices such as CPAP devices, flow generators, or similar devices (e.g., respiratory pressure therapy devices (RPTs)), or configured to communicate with them. For example, when detecting the biometrics of a person undergoing treatment, the system may cross-check that the intended user of the system is receiving the correct treatment. Thus, the system may optionally be able to reconfigure a PAP device to the intended machine parameters (pressure or flow scheme, etc.) in, for example, a CPAP, ASV, or recognition-based dual-camera setup. If the intended person cannot be identified (e.g., a previously detected and identified user), the system may also flag an alarm to the monitoring center and / or the user. Examples of this therapy device have been described in more detail previously and in later sections of this specification.

[0107] Similarly, when anticipated biomarkers are detected, the system can utilize the subject's respiratory rate and heart rate to better suit the intended user for treatment. For example, upon detecting an elevated heart rate, the system can input a specific respiratory curve into the device providing respiratory assistance to reduce the heart rate to the desired range.

[0108] In some embodiments, when receiving heart rate feedback for CPAP, biphasic, ASV, or other treatments, increases in heart rate can be tracked, increases in irregular heart rate changes (which may indicate arrhythmia) can be tracked, and / or long-term trends in heart rate dynamics and respiratory rate dynamics can be tracked, both with and without treatment.

[0109] In this way, feedback can be provided to the monitoring server.

[0110] In an occupational health environment, this biometric data can reduce and / or prevent the abuse of monitoring systems. For example, miners or truck drivers may undergo screening, diagnosis, and monitoring for sleep-disordered breathing and have sleep quality analysis devices installed in their living quarters or beds (e.g., in the cab / berth of a vehicle). Some types of sensors may be placed on the bed rather than on the side, above, or below it. By ensuring that specific subjects / users are actually being monitored using the biometric identification, monitoring information can be sent to a central system such as a scheduling system (i.e., ensuring all users sleep in their assigned beds) and, for example, implemented to ensure that the intended / identified person gets sufficient sleep. In this respect, the system can detect who is sleeping in the bed and track their sleep quality, referencing a fatigue management system. In one instance, a user may have sleep apnea but is not complying with their treatment (e.g., not wearing a CPAP mask). Thus, the system may issue an alert that the user is at increased risk (especially if the user requires treatment for occupational health and safety reasons). In some embodiments, the system may require permission from the system user or operator before tracking and / or identifying the user.

[0111] Typically, in some versions, once sensed, biometric data is identified (e.g., by categorized “fingerprints”) as belonging to a specific person, and this data can be stored along with the person’s identity, such as in a database.

[0112] During the daytime, physiological sensors such as the Plessey EPIC capacitive sensor, radar, or wearable devices can be used to monitor professional drivers and / or operators of heavy machinery or other personnel in safety-critical applications (e.g., air traffic controllers).

[0113] An exemplary set of “awake” and “sleep” data is described below. The signal collection method from “awake” can be applied to “sleep” and vice versa.

[0114] 4. Collection of "Awake" Time Data

[0115] Here are some examples of how the system can utilize information:

[0116] 4.1 Heart rate

[0117] a. Heart rate (HR) data can be collected from the user on a continuous or semi-continuous basis. This can be via a chest strap (e.g., a sports monitoring strap such as those provided by Polar) or a wristwatch that performs heart rate monitoring (e.g., a basic watch, a non-motor ECG, a plethysmogram, a projected pacemaker, or a similar graph). Ideally, a low-user-impact device should be used so that it is suitable for daily monitoring over long time periods (e.g., days, months, or years).

[0118] b. Analyze HR data to generate an estimate of heart rate variability (HRV).

[0119] c. Compare the recorded HRV parameters (such as short-term and long-term heart rate fluctuations) with historical and demographic (expected) parameters from healthy and chronically ill subjects, and provide the characteristics to the classifier. Additionally, a smoother version of the low-throughput HRV based on filtered signals can be utilized; this is primarily used to analyze longer-term fluctuations and compare with the non-trend version.

[0120] 4.2 Skin conductance response

[0121] a. The skin conductance response (GSR, also known as the skin conductance response) can be recorded by a wearable device (e.g., a basic watch or other commercially available GSR measurement device).

[0122] b. Use GSR signals as an alternative measurement of sympathetic “emergency” activity.

[0123] c. The extracted GSR and HRV signals are combined to generate an estimate of the ratio between sympathetic and parasympathetic activity. The system uses this balance of sympathetic and parasympathetic activity as an estimate of the normal state of disease progression (e.g., increased sympathetic activity and decreased parasympathetic activity).

[0124] 4.3 Exercise Intensity

[0125] a. Changes in exercise intensity and duration are captured from 'daytime' sensors, recorded in a database, and specific trends and changes are analyzed.

[0126] b. A superposition model of circadian rhythms / sleep can show different patterns of activity and rest periods throughout the day (e.g., office workers after getting out of bed, then commuting, then in the morning, coffee time, meetings, lunch walks, followed by intermittent afternoon naps [the strongest sleepiness usually occurs between 1 pm and 3 pm], increased activity during breaks, increased activity during the time of returning home, decreased activity in the evening during reading / watching TV, and a further significant decrease during sleep time [the strongest sleepiness usually occurs between 2 am and 4 am]).

[0127] c. Energy expenditure can be estimated by combining heart rate and GSR data with a step counter. Alternatively, a standalone step counter (pedometer) can be used as a substitute for exercise intensity (e.g., using a Nike Fuel band, FitBit, JawboneUp, or similar wearable device).

[0128] 4.4 Respiratory parameters

[0129] a. Breathing rate, depth, or activity (described in the "Sleep" section below)

[0130] 4.5 Blood Pressure Parameters

[0131] a. Derived from wearable optical volumetric sensors

[0132] 5. Collection of "Sleep" Time Data

[0133] Here are some examples of how the system utilizes 'sleep monitoring' information:

[0134] 5.1 Respiratory rate, depth, and activity (movement) level

[0135] An algorithm is implemented to detect a user's respiratory rate and dynamic patterns. The algorithm adaptively tracks a person's baseline respiratory rate, movement characteristics, and respiratory waveform shape over days, weeks, months, and / or years to establish their respiratory dynamic profile.

[0136] The algorithm module establishes an adaptive baseline for the user, taking into account respiratory rate parameters such as median, mean, interquartile range, skewness, kurtosis, minimum and maximum respiratory rates over a time period (e.g., 24 hours), primarily (but not excluding) targeting the time the person is asleep. Additionally, it tracks the shape of the inspiratory / expiratory waveform and short-, medium-, and long-term respiratory fluctuations. The reasonableness of the user's baseline can also affect these readings.

[0137] There is some overlap with activity detection performed in “awake” monitoring because the algorithmic processing steps and associated digital signal processing groups are used to determine physiologically repetitive and variable movements, including those caused by chest movement due to breathing, rocking detection and elimination, rolling in bed, and coarse and fine movements caused by multiple actions (including abrasions) (e.g., due to physical stimulation or discomfort).

[0138] 5.2 heart rate

[0139] Heart rate can also be estimated in a 'nighttime' bedroom environment by contact sensors (e.g., wearable devices such as basic watches using optical methods and ECG electrodes and devices) or non-contact sensors such as Sleep Master, using techniques such as microwave-based time-frequency transformation (by projecting heart rate signals—mechanical movements of the heart detected non-invasively from the skin surface).

[0140] 5.3 Coughing and snoring

[0141] Using digital sampling of audio signals recorded via a microphone, the algorithm can detect characteristic patterns of snoring, nasal congestion, coughing, or difficulty breathing. This is implemented using digital filter banks, frequency decomposition, and spectral analysis or morphological processing to search for 'clustering' noise (i.e., cough signals). These events can optionally be cross-correlated with movement and breathing patterns.

[0142] 6. Signal Processing

[0143] For one exemplary implementation of in-bed monitoring, the present invention analyzes dual-channel (in-phase I and quadrature Q) signals recorded by a radio frequency RADAR digitized using an ADC module suitable for digitization. These RF signals can be continuous waves or pulses (e.g., applied to ResMed Sleep Master 5.8 GHz and 10.525 GHz sensors, devices utilizing the FMCW method, or other devices). In some cases, these sensors are those described in U.S. Patent Application Publication No. 2014 / 0024917, the entire contents of which are incorporated herein by reference. These signals are sent to a filter bank, thereby including a series of digital filters with bandpass filtering suitable for detecting and removing low-frequency oscillation information. Phase information in the dual channels is compared to generate a clockwise / counterclockwise pattern. Hysteresis and glitches are applied to suppress signal ghosting, and the resulting signal indicates the direction of the movement source relative to a reference sensor frame. Peak / valley detection and signal tracking are additionally used to assist this processing. Thus, the system can determine whether the movement is moving toward or away from the sensor, and whether the direction has changed.

[0144] The spectral content of the signal can be calculated using Fast Fourier Transform and peak discovery (frequency domain) or via time-frequency processing (such as using discretized continuous microwave transform, appropriate basis selection, and peak lookup). Residual low-frequency components can also be processed to determine trends over longer time scales.

[0145] Sleep-disordered breathing patterns (including cyclic or periodic patterns, such as Cheyne-Stokes respiration) can provide aspects of user identification in addition to triggering symptom exacerbations. It should be noted that if such an SDB episode is detected, systems that recognize subsequent treatments such as CPAP can be automatically retrained to recognize altered biometrics (i.e., identification before and after CPAP intervention). Therefore, if treatment affects respiratory / cardiac / mobility activities, the system can recalibrate / retrain its recognition process.

[0146] Cardiac patterns such as atrial fibrillation and flutter (including paroxysmal), ventricular tachycardia, and other patterns can also provide identification. In such cases, if a user is seeking treatment for an arrhythmia, inputting treatment-related information into the system can reduce identification failures.

[0147] 7. Forming "fingerprints"

[0148] A key aspect of handling multiple possible input signals is a robust estimate of the quality of each signal. Just as the goal is to estimate a user's biometric "fingerprint" and track and capture the evolution of this biometric marker from health to illness, it is recognized that signals can be of poor quality, interrupted, or manipulated in undesirable ways (i.e., tampered with). Poor signal quality is not necessarily caused by interference between sensors. For example, a wearable heart monitor may misalign with the body, resulting in unusable or misleading signals. Correlation and comparison with data from other simultaneously worn moving sensors can help distinguish intermittent poor signals due to body movement (e.g., potential expected signal interruptions due to movement of other sensors) from longer periods of questionable signal quality caused by poor sensor positioning.

[0149] Users may decide to attempt to deceive the counter by placing the step counter on some other artificial or natural source of movement instead of the user, thus giving artificially high readings. Statistical and / or other parameters of the temporal change in detected movement can be used to distinguish the user's predictive biometric "fingerprint" from invalid or tampered values.

[0150] The system used to manage the pattern recognition process can accept multiple input parameters and ensures that a quality estimator has been executed to discard or label any poor quality data. It is recognized that input data can be dimensionally redundant, and procedures such as randomized PCA (principal component analysis) can be applied to reduce any such redundancy before further processing.

[0151] Figure 3Describe a system for calculating human biometrics (or multiple biometrics) in a sleep environment. For analog sensors, data is digitized via an ADC; for digital sensors, the signal is used directly. Preprocessing is performed, such as filtering for noise sources using bandpass filtering (e.g., primarily 50 / 60 Hz). Heart rate, respiratory rate, and movement / activity are extracted using time-frequency methods (e.g., by using short-time Fourier analysis or microwave analysis). The presence of movement and respiratory signals (physiological signals) in the sleep environment is then detected by field sensors such as a distance-gated motion sensor module (or other types of sensors, such as piezoelectric-based pad sensors or wrist-worn sensors). Sleep stages are then performed on the decomposed motion signal containing the physiological signals. Biometric characteristics are extracted, and a “fingerprint” / biometric is estimated and classified as human or non-human, regardless of whether the human was previously known to the system. Behavioral data may include long-term trends in sleep duration (e.g., typical long-term trends in weekday / weekend variations), bedtime (weekday / weekend), sleep efficiency, number of awakenings, sleep onset time, REM sleep percentage, and deep sleep percentage, among others—where these data are influenced by user behavior, such as voluntarily limiting sleep due to social activities on weekends.

[0152] Figure 4 This paper describes a system for calculating human biometrics from physiological parameters and optionally by combining behavioral characteristics (e.g., daytime activity patterns, location, peak or variability of heart / respiratory rate at specific times of the day). Daytime signals may contain both physiological signals and behavioral data. It can be seen that sleep time and daytime detection can be considered in combination or separately, depending on the available sensors and intended use.

[0153] Figure 8 The heart rate (top) and respiratory rate (bottom) of subject "User A" during sleep (approximately 8.3 hours) are shown, recorded using a non-contact pulsed continuous wave RF sensor at a distance of <1.8 m. Figure 9 The histogram represents the heart rate and respiratory rate of "User A". Figure 10 The study showed that different subjects, "User B," had signals of heart rate and respiratory rate during sleep (only exceeding 6 hours), and... Figure 11 The middle section shows a histogram of the relevant heart rate and respiratory rate. These signals can be estimated using the processing techniques described in this paper.

[0154] For example, Figure 5 The diagram illustrates the processing used to compute characteristic features in order to generate a biometric fingerprint. User identification can be performed by differentiating biometric parameters input into a user classifier (see, for example...). Figure 5This allows for the calculation and combination of characteristics from cardiac and respiratory signals. Optionally, if a good quality HR (heart rate) is unavailable / not detected, the system can regress to BR (respiratory rate) under consideration for some or all of the treatment periods; thus, the system can rely on different biomarkers to identify individuals based on the quality of the detected biomarkers. Figure 5 Any features identified (e.g., two or more) shown or described below are estimated as part of a biometric fingerprint.

[0155] Respiratory signal related parameters

[0156] - Variation in respiratory rate throughout the day and / or night (this variability is a characteristic of the user).

[0157] -This can be measured in breaths or longer time scales—for example, 30, 60, 90 seconds or longer.

[0158] - Stability over time (related to this variability)

[0159] -Standard deviation of respiratory rate

[0160] - Breathing depth (shallow, deep, etc.) and the relative amplitude of adjacent breaths

[0161] -Mean or average respiratory rate

[0162] - The truncated mean excludes outliers (e.g., at 10%).

[0163] -Awake or asleep (e.g., the user's sleep state at the time of detection)

[0164] - Spikes in respiratory rate (sudden acceleration or deceleration) observed during the quiescent period and REM sleep.

[0165] -Median (50th percentile)

[0166] - Interquartile range (25th - 75th percentile)

[0167] -5th-95th percentile

[0168] -10th to 90th percentile

[0169] - Histogram shape

[0170] -skewness

[0171] - Peak state

[0172] - Peak frequency over time

[0173] - The ratio of the second harmonic to the third harmonic at the peak frequency

[0174] - Percentage of valid data (valid physiologically reliable data)

[0175] - Autocorrelation of individual signals

[0176] - Characteristic patterns in the spectrum

[0177] -Awake or asleep

[0178] -Relative percentage of REM and deep sleep

[0179] Heart / Heart Signals

[0180] - Heart rate variability (between heartbeats (e.g., derived from a projected cardiac chart) and over longer periods within a defined movement window (e.g., 30 seconds, 60 seconds, 90 seconds)).

[0181] - Variability over time (variability between heartbeats / respiration)

[0182] -average value

[0183] -Cutoff mean (10%)

[0184] -Standard deviation

[0185] -Median (50th percentile)

[0186] - Interquartile range (25th - 75th percentile)

[0187] -5th-95th percentile

[0188] -10th to 90th percentile

[0189] - Histogram shape

[0190] -skewness

[0191] - Peak state

[0192] -Stability over time

[0193] - Peak frequency over time

[0194] - The ratio of the second harmonic to the third harmonic at the peak frequency

[0195] - Percentage of valid data (valid physiologically reliable data)

[0196] -Awake or asleep

[0197] - Autocorrelation of individual signals

[0198] - Characteristic patterns in the spectrum

[0199] Cardiopulmonary signals

[0200] -Amplitude squared cross-spectral density (in moving window)

[0201] - Cross-coherence

[0202] - Respiratory sinus arrhythmia peak

[0203] - The LF / HF ratio, which indicates the parasympathetic / sympathetic balance in the autonomic nervous system.

[0204] - Cross-correlation and cross-coherence (or cross-spectral density) of estimated cardiac and respiratory signals.

[0205] - Characteristic movement patterns over a longer time scale, i.e., statistical behavior observed in the signal.

[0206] - Movement patterns during the detection and comparison of these cardiac and respiratory signals (e.g., during sleep, some users may have restful sleep and some may have less restful sleep).

[0207] For example, a reference distribution of movement and / or other characteristics can be compared with the calculated distribution and can be used as a comparison for testing variability-limiting histograms, such as nonparametric Kolmogorov-Smirnov goodness of fit. Parameter extraction can be achieved using time-frequency analysis techniques (such as STFT or microwave).

[0208] While no single parameter allows for differentiation of a user (e.g., average heart rate or average respiratory rate at rest or during specific sleep stages such as deep sleep), more advanced systems can combine multiple features and therefore favor early integration, sending feature sets to the classifier. If training data (labeled data) is available, a supervised classification system can be employed, providing the system with a large training dataset to generate model parameters. Feedback from the first use of data (daytime or nighttime signals) can update the user-specific classifier so that the biometric "fingerprint" increases with user-specific accuracy. This can be achieved via a registration step, where sample acquisition templates are stored in a database. Subsequently, a matching step is performed to verify identity. When such detailed training data is lacking, semi-supervised or unsupervised learning of feature hierarchies (e.g., deep learning) methods using techniques such as sparse coding (e.g., LCC local coordinate encoding, extracted from the image processing domain) are employed.

[0209] A decision-level neural network is used after the template matching step so that the decision can be verified.

[0210] A user's physiological fingerprint can be referenced to check if the "correct" (i.e., expected) user is detected. This reduces / eliminates ambiguity that can occur when a non-contact sensor, placed, for example, on a bedside table, monitors user A. User A may get out of bed (e.g., to the bathroom or to work, etc.), and then a different user B (e.g., a bed partner, baby, or pet) may move into the sensor's range. A basic sensor data processor detects user B's parameters and then confounds this data with user A. A more advanced processor will detect user B's "fingerprint" / pattern and distinguish these from A. For static sensors with a known set of users, the system can access supervised training data as part of the processing's learning. For unknown users (i.e., unknown to the system), semi-supervised (i.e., with knowledge of user A but not of user B, etc.) pattern recognition can be performed, or an unsupervised system can be used if no distinguishing markers are available to the system. Semi-supervised or unsupervised systems may require data from multiple days / nights to generate a "fingerprint" of the user's physiological and behavioral parameters.

[0211] During system registration using non-contact or minimal-contact sensors in the bedroom, subjects may be asked to perform specific breathing exercises in front of the sensors, such as guided deep breathing signals. The system can register users when guided by specific deep and shallow breathing patterns with defined inspiratory / expiratory phases. The level of coherence with the detected heart rate signal and other HR characteristics can be used as a baseline input. More complex registrations will require recording spontaneous breathing over longer periods (e.g., overnight signals) to train the system and / or monitor daytime sensor parameters.

[0212] exist Figure 6 and Figure 7 The document provides the algorithmic processing steps for the registration process and the subsequent identity verification. Figure 6 In the initial setup, the system is configured for registration. This can be automated or manual; for example, to register sleep periods, it can be manually initiated, but the system can then automatically scan data from nighttime periods to identify baselines and calculate characteristics to generate a user's biometric signature ("fingerprint") (e.g., model weights for a classifier). This is then stored in a database for subsequent biometric queries and comparisons. Daytime registration aggregates data from one or more monitoring devices and may also process user behavioral data to generate the signature. Behavioral data can be assigned lower weights than physiological data because aspects of these behaviors may make it easier for malicious users to simulate respiratory and cardiac signals.

[0213] exist Figure 7The initial steps involve detecting valid signals (e.g., "presence" detected in an RF sensor field, exceeding sensor baseline activity on an accelerometer, etc.) and detecting a human (indicating detected respiratory and cardiac signals), calculating characteristics, and a machine learning algorithm outputting a biometric estimate. The system can then access a database to examine appropriate sleep and / or wakefulness models (based on the detected signals). Based on the probability of detecting a specific user, the system accesses an Identity and Access Management (IAM) system to examine potential candidates. Statistical comparisons are performed, and a decision is made to validate the human biometrics as valid (and allow or examine the relevant level of authorization—or its separate database) or exclude the biometrics (unvalidated user).

[0214] Figure 8 and Figure 10 Sample images of heart rate and respiratory rate of user A and user B (as detected by RF sensors) are shown. Figure 9 and Figure 11 The relevant histograms of these signals are presented.

[0215] Figure 12 The cross-spectral density between segments of heart rate and simultaneous respiratory rate is shown for subjects breathing rhythmically at 0.1 Hz (6 deep breaths per second). Peaks in the cross spectrum at 0.1 Hz (such methods can be used to characterize relationships between systems) show the combined effect on heart rate and respiratory rate.

[0216] The detection and exclusion of data from animals sleeping in bed, such as dogs and cats, can be performed using this system. Detection and exclusion of the BR and HR components of a second person in bed are also possible, for example, for cases involving longer-range sensors detecting portions of the second person's nighttime recordings, or for monitoring the entire night after the first person has left.

[0217] More typically, when employers or insurers provide users with the motivation to use specific monitoring technologies (e.g., to indicate activity / exercise over a period of time), the estimated biometric parameters can be used to check whether the user is actually authorized to use the technology or not another party (unintentionally or intentionally imitating).

[0218] The relationship between specific sleep stages and respiratory rate and / or heart rate parameters can provide additional understanding of a user's specific biometrics, as respiratory rate and heart rate statistics are not specifically under the subject's automatic control during sleep. For example, the system can be configured to detect whether a user is in REM (dreaming) or deep sleep. For example, the system can implement methods for determining sleep states or stages, such as those disclosed in International Patent Application No. PCT / US2013 / 060652, filed September 19, 2013, and International Patent Application No. PCT / US2014 / 045814, filed July 8, 2014, the entire disclosure of which is incorporated herein by reference. Preferably, deep sleep can be selected so that the effects of daytime stress factors are least apparent. REM sleep can show greater respiratory changes in some cases, but it can also be more significantly affected by SDB episodes, PTSD (post-traumatic stress disorder), or other problems. Light sleep (stage 2) or stage 1 light sleep parameters can also be analyzed.

[0219] Strictly speaking, if a person is asleep, they cannot consciously regulate their breathing in a specific pattern, thus reducing the possibility of signal tampering. Therefore, separate wakefulness and sleep models of a user can be created. In practice, separate models of deep sleep, REM sleep, or other sleep stages can be created. Therefore, in some cases, fingerprints of any or every one of these stages can be used during the authentication process. These sleep stage-specific fingerprints can then be estimated during identification using suitable features determined in that specific sleep stage, so that the identification of the person can be performed in conjunction with that specific sleep stage.

[0220] In this regard, the system can provide analysis of the data under analysis, as well as trends over multiple nights.

[0221] Additionally, the breathing patterns detected by the PAP device or RPT can form part of the user's "fingerprint." Patient interfaces or patient circuits such as catheters and / or masks, or other such devices such as those described in more detail herein, may include additional sensing such as an accelerometer, or the ability to detect movement of the catheter (e.g., CPAP tubing), thereby adding movement-characteristic information to the extracted breathing parameters. In this way, the PAP device can examine the user's biometric "fingerprint" for compliance purposes.

[0222] It should be noted that a simple increase in respiratory rate can be associated with non-chronic conditions such as the common cold.

[0223] A holistic, multi-parameter analysis of an individual's breathing allows the system to perceive its environment and correlate relevant data with the correct user. For example, an insurance company could offer discounts to users based on whether they meet certain health improvement goals. These goals could include reducing average heart rate, decreasing respiratory rate, or increasing exercise intensity.

[0224] Changes in skin conductance response and / or skin temperature recorded by simple motion recorder sensors or wearable sensors can be used to augment this biometric classification. For specific heart rate and GSR changes, these can be cross-referenced with exercise intensity to indicate whether a stress event or activity mediates a particular change. For each individual, changes in heart rate, respiratory rate, and GSR in response to exercise or a stress event can be exhibited in a specific order. Therefore, a user's inherent stress level and ability to cope with stress can be used as biometric markers when combined with other available parameters.

[0225] Longer-term patterns of exercise intensity and steps taken can be correlated with specific user behaviors.

[0226] This system can be configured to detect the anti-theft function of monitoring devices (e.g., devices with RF sensors), i.e., detecting the biometrics of unauthorized or unknown users. If location data is available from the attached device, such as if location data is included as part of a "fingerprint," this determination can be made with reference to location data (e.g., GPS coordinates).

[0227] This system can be used in hospital settings to automatically separate reads from different patients as they move to different monitoring beds within the hospital or are discharged and the beds are reused. The system can automatically associate hospital / clinic data collected from the monitoring device with a second device given to the patient to take home for long-term monitoring.

[0228] It should be noted that this is quite resistant to replay attacks, so an attacker / nuisance could attempt to simulate or replay RADAR signals, for example, under the control of a second RF sensor. The first RF sensor can be configured to detect emissions and interference from the second RF sensor and flag an alarm signal to the system. In practice, the system can provide continuous verification because physiological signals can provide fresh biometrics every few seconds.

[0229] The system can feed control signals back to RF sensors by recognizing the biometrics of the first and second persons in the bed, so as to adjust the distance and / or power level of the sensors to detect the desired person (e.g., the first person).

[0230] Therefore, in some cases, the system can aggregate multiple information sources and identify users by intelligently processing this data and / or by providing specific information to the user and device, such as identification by readings during night / sleep. Other versions of this system may target only certain parameters or may not adapt to user-specific parameters.

[0231] The potential benefits of some versions of this technology may include:

[0232] 1. This system can detect physiological patterns and identify humans based on cardiac and / or respiratory signals.

[0233] 2. These signals can be captured via pulsed radio frequency (RF) sensors or sensor arrays.

[0234] 3. The system can identify a user’s biometric parameters based on heart rate and / or respiratory rate and / or amplitude recorded by non-contact sensors, thus confirming that the recorded signals come from a given user.

[0235] 4. The system can update biometrics during sleep and wakefulness phases.

[0236] 5. Reduce / remove conscious variability in signals applied during sleep (i.e., reduce the risk of misclassification or user "faking" breathing patterns that can also affect heart rate) (e.g., as noted in OH&S usage).

[0237] 6. The system can preferably update biometrics during deep (optimal) or REM (suboptimal) sleep.

[0238] 7. The system can communicate with the server to examine existing identity templates / models and determine identities.

[0239] 8. The system can register users during either the waking or sleeping phase.

[0240] 9. The system can register users when using specific deep and shallow breathing patterns and depth guidance with defined inhalation / exhalation phases.

[0241] 10. Systems using pulsed radio frequency (RF) sensors can detect and notify the introduction of unauthorized second RF sensors to mitigate “replay” attacks against the system.

[0242] 11. The system can detect a person's health status and receive input regarding treatment / therapy, allowing it to be retrained to a new health state. This includes a key for the flow generator and intelligent adaptation of the adaptive servo ventilator (ASV) to the heart rate. The system ensures that appropriate personnel use treatment devices previously designed for them, and can be reconfigured to settings more suitable for new / unknown users.

[0243] 12. The system can update biometric data from wearable sensors, such as databases or libraries, during the day and at night, including respiratory rate, heart rate, movement patterns, skin conductance, and blood pressure (which is also an accelerator for daytime respiratory rate and uses “activity” data from other sensors such as video).

[0244] 13. The system can detect when a second biometric is detected for some or all of the recorded time periods, so that the data of a second or third person is not focused on the first person and is erroneously processed by the system.

[0245] 14. This system can detect sleeping or awake animals such as dogs or cats within the sensor area.

[0246] 15. This system can send control signals to sensors / processors to adjust sensor behavior (power / distance gating / directivity) by detecting two people (two biometric features) on the bed to better detect the first person (further details and examples of adjusting sensors based on biometric features and selecting the lowest sensor power for a given situation).

[0247] 16. The system can be cloud-based, enabling it to track a person's biometrics across multiple fixed or mobile sensors in different locations, so that the sensors do not need to move with the person. For example, the system can track a person from a first sensor in a first room to a second sensor in a second room, and automatically verify the physiological data from the first and second sensors.

[0248] 8.0 Integration of Multi-Sensor and Biometric Feedback

[0249] In cases where multiple sensors work together, such as communicating on wired or wireless links (simultaneously or as part of a paired process) in a system with a control processor that has remote positioning or co-positioning with the sensors, characteristic biometric parameters can be used to dynamically adjust the performance of one or more sensors to optimize physiological identification from independent human sources and exclude other sources.

[0250] In the case of two or more adjacent sensors exchanging information to minimize RF interference or other types of interference, the biometric identity and / or related biometric qualities detected by each sensor can be used as a basis for adjusting the control parameters of one or more sensors. For example, adjustable parameters may include the distance gating pulse timing and transmit power level of these sensors (within permissible regulatory limits) or the RF detection frequency (e.g., center frequency). Such changes are most readily implemented in digital sensor designs, although control signals can be provided to analog sensor devices configured to allow this provision.

[0251] The sensor can be configured to use the lowest possible amount of power to achieve good signal quality for biometrics. This is advantageous for environmental reasons and can also increase the sensor's ability to be battery-powered. Additionally, it minimizes the possibility of interference from other sensors or devices.

[0252] 8.1 Biometric Control Using Multi-Sensor Arrays

[0253] The sensors can be configured to operate close to each other while maintaining minimal RF interference. In this regard, two sensors can be placed on opposite sides of the bed and programmed to monitor an individual sleeping in a normal setup. For example, the biometrics of a particular user can be detected by each sensor, and each sensor can be initialized with this user data so that the sensor includes a baseline of data attributable to that user. However, during further operation, as... Figure 13 As shown, the quality of the sensors can be degraded because they can receive overlapping signals from each person. For example, sensor _a, programmed to detect signals from human _a, can receive overlapping respiratory, heart rate, and / or movement signals from human _b, thus reducing the signal received for human _a. Similarly, sensor _b, programmed to detect signals from human _b, can receive overlapping respiratory, heart rate, and / or movement signals from human _a, thus reducing the signal received for human _b.

[0254] To avoid this degradation, these sensors can be adjusted to minimize interference. For example, such as Figure 14 As shown, two sensors placed on opposite sides of the bed and programmed to monitor a single individual sleeping in a normal setting can be programmed to minimize interference by adjusting the distance and power of each sensor. In this regard, sensors _a and _b, respectively programmed to receive biometrics from human _a and human _b, can be configured to minimize distance and power to maximize clear, individual biometrics. For example, the sensors can reduce their detection range when detecting accidents or using biometrics not specifically identified by a given user's "fingerprint". Figure 14 In the example, when sensor_b detects biometrics that it considers associated with its initial user and those that it does not, sensor_b can decrease (e.g., incrementally) its detection distance (e.g., via power or distance gating) until it detects only biometrics that it considers associated with its initial user. Sensor_a can be programmed similarly. Therefore, the quality of biometrics received by sensors_a and_b can be excellent because these sensors detect biometric signals from only a single human.

[0255] In some embodiments, two sensors are placed on opposite sides of the bed and programmed to monitor the sleep of a single individual; these sensors may monitor only a single individual. For example, as... Figure 15 As shown, a sensor_b programmed to detect signals from a human_b can receive respiratory, heart rate, and / or movement signals from the human_b as required. That is, for example, the sensor can identify its initial user by the biometric signals it detects. However, a sensor_a programmed to detect signals from a human_a (currently nonexistent) can receive respiratory, heart rate, and / or movement signals from the human_b. That is, for example, the sensor can recognize that the detected biometric signals do not originate from its initial user. Therefore, biometric quality degrades when the biometric signals detected by sensor_a are repeated by sensor_b or when there is additional interference.

[0256] To avoid duplicate biometrics or other interference, sensors programmed to receive an individual's biometrics can enter a power-saving mode and / or a search mode when detecting a unique, unidentified biometric feature, where the sensor reduces or terminates, for example, sensing or sensing distance. Therefore, as... Figure 16 As shown, one sensor can go into sleep mode when the activity sensor can operate with minimal interference. In this regard, sensor_b can be programmed to detect breathing, heart rate, and / or movement signals from human_b, and sensor_a can be programmed to detect signals from human_a (currently not present). Sensor_a can avoid repeating biometrics detected by sensor_b and enters a power-saving / search mode when it is determined that human_a is not present. The power-saving and / or search mode can limit the sensor's distance and power while the sensor waits for an individual to return. This prevents the reception of duplicate biometrics.

[0257] In some embodiments, more than one pair of sensors may be used. For example, such as Figure 18 As shown, three sensors can be used in close proximity. In this regard, the first sensor "α" (sensor_a) can be located on the bedside table on the first side of the bed and is directed to the first human (human_a), who can be the initial user of sensor_a. Human_b, sharing the bed with human_a, can be monitored by the second sensor "β" (sensor_b), which is located on the bedside table on the second opposite side of the bed and is directed to human_b, who can be the initial user of sensor_b. In one instance, human_b can be a large subject with a body mass index (BMI) greater than human_a or human_c. Additionally, human_c can sleep in a bed located on the side of the bed with a thin partition away from human_b and human_a. Human_c can be the initial user of sensor_c and can be monitored by sensor_c.

[0258] The system can be configured to operate sensor_a at an odd frequency, sensor_b at an even frequency, and sensor_c at an odd frequency. This minimizes RF interference between the sensor groups.

[0259] Biometric signals can then be detected by each sensor device. Based on the detected biometric signals, humans a, b, and c can be identified based on "fingerprints." In this regard, sensor a can detect a segment of biometric signal that is identified as human b (on the other side of the bed) when human a is not detected (e.g., human a has left the room). In other words, human b is identified as human whose biometrics have been detected by both sensors a and b. Therefore, the system control processor can send or generate control signals to sensor a to adjust the distance gating to shorten the distance and continue monitoring the detected biometrics. Additionally, the system control processor can optionally reduce the power level of sensor a.

[0260] When sensor_a detects human_a, the sensor can be adjusted or separately activated to enter normal power mode. When the configuration is implemented such that sensor_a correctly detects the biometrics of human_a (when human_a is actually in the room) and does not detect the biometrics of human_b, the control parameters of the detection configuration can be stored as optimal settings.

[0261] Additionally, when an individual or animal intended for monitoring enters the sensor's range, the sensor can be programmed to further control changes. For example, such as... Figure 18 As shown, a dog "dog_a" can enter the bed by climbing onto a human to reach the center of the bed. These sensors can detect non-human biometrics, and in such detections, these sensors can adjust their distance gating and / or power parameters (e.g., sensor_a and sensor_b) to minimize the biometrics of dog_a, thereby supporting the preferred detection of human_a and human_b using sensors_a and sensor_b respectively.

[0262] In this way, the behavior of the sensor can be adjusted once or multiple times during the monitoring period, such as by a sensor control processor or other system control processor, including any detection method described herein.

[0263] In cases where optimizing sensor parameters does not improve biometric quality, one or more users may be prompted to adjust the orientation of one or more sensors.

[0264] In examples of multi-sensor setups, such as Figure 18The examples shown (including sensors_a,_b, and_c) may optionally include processors programmed for location checking (e.g., GPS or other input data) and / or setting detection frequencies to minimize interference. In this regard, the processor may determine and utilize RF detection frequencies suitable for the positioning sensor. For example, if the sensor determines its location is in the United States, it may access data (e.g., tables regarding frequencies and geographic locations) and set the target detection center frequency and FCC spectrum mask, along with other settings from tables relevant to the United States. Similar parameters are available in parts of the European Union according to appropriate ETSI (European Telecommunications Standards Institute) spectrum masks, and center frequencies are permitted in those regions (or indeed in other countries based on local regulations). Thus, the center frequency of sensor_a may be automatically tuned to 10,527,125 kHz, with a power level of 43.8%, and the distance gating (time of flight) may be adjusted to provide detection exit at 92.5 cm. Sensor_b has an auto-adjustable center frequency of 10,525,000 kHz and a power level of 93.2%, with a distance gating adjusted to provide a stop at 165.2 cm. Sensor_c has an auto-adjustable center frequency of 10,522,375 kHz and a power level of 80.0%, with a distance gating adjusted to provide a stop at 150.0 cm.

[0265] In one instance, human a exhibited a median heart rate of 51.4 bpm and an interquartile range of 6.3 bpm during deep sleep, and a median heart rate of 55.7 bpm and an interquartile range of 9.1 bpm during REM sleep. The biometrics of human a are best described using the median and the shape of the resulting histogram, as well as the moderate to high coherence between respiratory and cardiac signals. The respiratory rate of human a was 13.1 br / min (± / 3.2 br / min) during deep sleep and 13.9 br / min (± / 1.1 br / min) during REM sleep.

[0266] Human _b exhibited a mean heart rate of 77.5 bpm throughout the night, with significant accelerations of nearly 30 bpm, and increased interbeat variability due to underlying intermittent arrhythmias. The mean respiratory rate of human _b, 18.2 br / min, increased to over 20 br / min during periods of significant heart rate acceleration. The br / rate variation during REM sleep was 4.5 br / min. Respiratory cessation (apnea and hypoventricular activity) was detected in the respiratory recordings. A characteristic bradycardia / tachycardia sequence in the heart rate was associated with these episodes of respiratory cessation. HR / BR coherence was generally low due to significant variability in both HR and BR. Skewness and kurtosis plots indicated an uneven distribution.

[0267] The human sleep pattern shows an average heart rate of 59.8 bpm, which is particularly stable during deep sleep. During REM sleep, variability increases significantly to oscillations between 2 and 23 bpm. The average respiratory rate of the human sleep pattern is 17 br / min, with a minimum of 15.2 br / min.

[0268] When dog_a enters the space between human_a and human_b, sensor_b can adapt to human_a being slightly closer to the previously described position. For example, when detecting unidentified / initialized biometrics, the processor controlling the sensor can control the sensor's parameters, such as its detection power, frequency, or other control parameters used for sensing. For example, the center frequency can be adjusted to 10,525,000 kHz, its power level to 84.9%, and distance gating is adjusted to provide an exit at 145.6 cm. This will help distinguish between dogs and humans by limiting the parameters of sensor_a, so that only human_a is detected (e.g., by adjusting distance, power level, or detection direction), and dog_a is excluded (i.e., not detected); similarly, the parameters of sensor_b are automatically adjusted to detect only human_b and exclude dog_a. The system can also tag detected unknown biometrics, and additionally, respiratory rate, respiratory depth, and heart rate parameters are consistent with the animal. While the average dog takes 24 breaths per minute at rest, this high respiratory rate is characteristic of chronic diseases such as congestive heart failure or COPD (chronic obstructive pulmonary disease) in humans, but is accompanied by shallow breathing (relative to the population average or baseline) and varying heart rate variability. When dogs or cats have undergone strenuous exercise or been exposed to high temperatures, a faster respiratory rate and open breathing (shortness of breath) are expected over a period of time. The typical respiratory rate range for dogs is approximately 20–30 br / min, and heart rate is typically 60–100 bpm for large breeds, 100–140 bpm for small breeds, and approximately 220 bpm in puppies. The typical respiratory rate range for cats is 20–40 br / min, with heart rates between 140 and 220 bpm. Total lung capacity tends to increase in mammals with increasing body mass; cats or dogs are, on average, lighter than adult humans, and there are different patterns in infants and children. Therefore, the sensor control parameters can also be adjusted when detecting one or more such biomarkers, which are recognized by the processor to indicate non-human animals.

[0269] Similarly, as a further example, if human _a leaves the bed, sensor _a can enter an energy-saving / search mode, where the center frequency is 10,527,125 kHz and the power level is 0 to 100%, and the distance strobe (time of flight) is adjusted to provide an exit at a distance of 50-300 cm. Given the existing knowledge of the presence of sensor _b, sensor _a's search mode could limit the power percentage and exit distance to minimize potential interference pulses entering sensor _b.

[0270] In some embodiments, controllable antennas (e.g., using classical phased array antennas or digital beamforming) can be used to adjust the coverage of these sensors and minimize interference. For example, the sensing “cones” of sensors_a and_b will be adjusted to completely exclude any biometric evidence from dog_a.

[0271] Optionally, in some cases, sensor-based biometric identification can trigger other systems, such as when the sensor does not identify the initial user. For example, as... Figure 17 As shown, the system's processor can trigger or send an alarm (e.g., to the smart devices of human_a and human_b) to alert to unexpected biometrics (e.g., dog biometrics). As another example, the processor can activate a network camera or trigger an audible alarm (e.g., via a speaker) upon this detection.

[0272] By detecting specific biometrics, customized configurations of light and / or sound and / or environmental parameters can be set (i.e., controlled by a controller / processor performing the identification). As an example, consider a system where two partners share a bedroom. When the first partner is detected, the system turns on the smart TV to a sports channel (or via an online service such as Netflix or a similar stream) and turns off the master bedroom lights. After a certain period, if the user is detected as asleep, soft jazz music is played via the Hi-Fi system, and the TV is turned off. Once the user is in deep sleep, the Hi-Fi volume is gradually reduced and then turned off. If a second partner is detected in the bedroom, the Hi-Fi / radio is configured to play popular music / the latest top 20 tracks from a stored music collection or popular music via a streaming service (e.g., Spotify, Deezer, Apple); additionally, the room lights dim, and the bedside lamps are configured with orange / red hues and low white / blue content. The TV is configured to display a sunset video sequence (with low backlight intensity) or simulate a fireplace. Once the second partner is asleep, the music volume gradually decreases / the TV dims until the user is in deep sleep, at which point both are turned off. When two partners are indoors simultaneously, motion TV / video streams are directed to the first partner's wireless headphones, and pop music is directed to the second partner's wireless headphones (or music pillow). When available, a TV polarizer is configured to direct motion content to the first user and fireplace video to the second user. Thus, biometric detection of one or more partners allows for a customized bedroom (or practically living room) experience. Similarly, for the wake-up experience, an alarm can be customized based on the first and second partners. For example, the first partner might want to wake up later than the second. In this case, a low-amplitude alarm with focused light can be used by the second partner to avoid waking the first partner. Such preferences can be programmed into the system and are based on identification access performed by the system's processor.

[0273] For example, the system can also alert the user if another animal and / or human (unidentified) enters the sensor's sensing area. For example, such as... Figure 17As shown, when a dog (or actually another animal or human) enters the sensor-sensing space, the system triggers an alarm. Optionally, images or videos of the event can be triggered and stored and / or transmitted to authorized personnel for viewing. In this regard, a triggered camera (e.g., an attached network camera, low-light camera, infrared camera with an integrated illuminator, thermal camera, or other type) or webcam can send video of unidentified individuals to authorized personnel. To understand this, an alarm can also be triggered if unknown human biometrics such as a thief / intruder are detected within the area of ​​one or more sensors. When video and / or audio sensors are accessible or under the control of the system, video and / or audio can be recorded or transmitted in real time for viewing by a monitoring center before contact with police. This biometric detection can also be used to reduce false activation of intruder alarm systems when implemented in conjunction with other detection devices such as PIR (passive infrared), other microwave intruder sensors, glass break detectors (vibration or acoustic), magnetic contact switches, vibration sensors, triggered video motion, or initiated pressure (such as pads).

[0274] As can be seen, such biometric detection implemented in apartment buildings or office buildings helps firefighters check the human (biometric) counts provided by the fire alarm system (and optionally associate with access card systems in appropriate locations). These RF sensors can continue to function in smoke or high-heat conditions compared to standard video content (and RF does not have the same privacy issues as fully video processing, such as its potential use in bathrooms / bedrooms). If user biometrics are registered with a system such as a central controller or computer that receives identifications from a collection of sensors, this fire alarm computer can output a list of names and locations within the building / structure. Therefore, the central computer can maintain data identifying the location of each sensor and, in an emergency, combine this list with identifications of people within the detection range of the sensors in the sensor collection.

[0275] This RF sensor can be powered in many ways. For example, the RF sensor can be placed in a pass-through AC outlet (i.e., plugged into a power outlet / AC socket, and a very thin backpack AC outlet is provided; a USB outlet can also be provided on the outlet panel to provide a second use for the device). Night lighting features can also be provided on the device, for example, placed in a hallway, and activated by detecting movement using the RF sensor. The RF sensor will sense within the outlet / AC outlet area (e.g., covering part or all of a room) and transmit data via Wi-Fi or other wireless communication devices or via power lines (e.g., via powerline adapters or similar standards). A different example is the provision of a pass-through USB adapter or cable. The RF sensor and communication (e.g., via Wi-Fi or on its USB link) are integrated into the adapter or cable and draw power from the 5V provided in the USB manual. This allows the RF sensor to be placed anywhere a USB cable is provided, such as plugged into a wall-mounted phone / tablet charger, plugged into a laptop, etc. It can be seen that by plugging this device into a laptop or tablet, the biometrics of the user in front of or near the device can be checked. Either an AC outlet device or a USB online device provides a simple (plug-in only) capability for monitoring biometrics in a space with extremely low cost.

[0276] 9.0 Other methods for identification / differentiation

[0277] As discussed earlier, sensors can employ identification methods to distinguish life forms in the vicinity of the sensor. In other examples below, multiple life forms can be distinguished to enable the identification of a specific user.

[0278] 9.1 Introduction

[0279] A large proportion of adults sleep with partners, who may have potentially different sleep patterns. A partner may go to bed before the primary (or dominant) user of the sleep sensor, or may remain in bed after the primary user has left. To avoid mistaking the partner for the primary user, the system can employ methods to differentiate them. In some cases, this may allow for separate assessment of the primary user's sleep (i.e., separate from the partner's sensor data).

[0280] 9.2 Methods

[0281] 9.2.1 Experimental Design

[0282] For the initial study, five healthy participants were recruited. Each participant slept with a partner. A 10.5 GHz non-contact sensor was placed beside the participant's (or user's) bed, on their bedside unit, facing their chest. See also Figure 1Non-contact sensors were set up to record data 24 hours a day for the duration of the study. The primary user (the user below) also provided daily details about when they went to bed, when their partner went to bed, when they woke up in the morning, and when their partner woke up.

[0283] Sensor data was segmented into 24-hour records, with start and end times equal to the initial time the device was turned on. For each record, a series of annotations were manually created (based on daily bedtime and bedtime entries provided by the user) to identify (a) user data, (b) partner data, (c) absent, and (d) 'insufficient information'.

[0284] 9.2.2 Sensor Signals and Preprocessing

[0285] An exemplary sensor system outputs two similar voltage signals, representing the sensor's I and Q signals. The sensor operates by generating two shortwave radio frequency energies of 10.5 GHz. The first pulse serves as the primary transmit pulse, and the second pulse is a mixer pulse. The first pulse is reflected off a nearby object to form an echo pulse received back by the sensor. The distance the echo pulse travels introduces a time delay in the arrival of the echo signal back to the sensor. This time delay causes a phase shift between the echo signal and a reference mixer pulse. By adding ('mixing') the echo pulse to the mixer pulse within the receiver, a signal proportional to any phase shift of the echo pulse is generated. Moving targets (such as human breathing) generate variable phase shifts, which can be detected by electronic devices in the circuit (de Chazal, P., Fox, N., O'Hare, E. et al. Sleep / wake measurement using a non-contact biomotion sensor, Journal of Sleep Research, 2011, 20: 356–366).

[0286] Filter the sensor data before analysis (filter low-throughput and anti-aliasing, and in parallel, filter high-throughput and anti-aliasing for motion analysis). Analyze the data within a time period (e.g., a 30-second time interval or other suitable time interval).

[0287] An initial presence / absence detection is performed, followed by a sleep / wake cycle. Motion and activity levels can be quantified for each time period. The respiratory rate is calculated in 1 Hz. A complete sleep stage analysis (identifying light, deep, and REM sleep) is also possible.

[0288] 9.2.3 Feature Extraction and Research

[0289] Then, features are extracted (e.g., 25 features (Table 1)) (e.g., every 30-second interval of each recording). Therefore, the processor can be configured to calculate or determine any one or more features from the following table in the identification process. For initial studies, one-way ANOVA can be used, for example, to compare derived features used by users and partners over known time periods to assess whether each individual feature significantly distinguishes users from partners. Each feature can also be compared to the mean of the complete recordings and the mean of historical data from a specific subject. This will significantly increase the number of available features.

[0290] Table 1 (Exemplary Feature Names and Descriptions)

[0291]

[0292]

[0293] When implementing a classifier system to identify / recognize specific individuals, characteristic values ​​for users and partners can be determined or calculated over a time period (e.g., multiple nights, one evening, one or more time periods). Any one or more of these calculated characteristics can be used individually or by combining multiple characteristics to provide a statistically significant distinction between users and partners. When evaluated individually using one-way ANOVA, the characteristics in Table 1 were found to significantly distinguish user data from partner data, where p-values ​​(where p < 0.05 is considered statistically significant) were less than or equal to 0.05. This indicates that these characteristics are suitable for inclusion in a classifier model to distinguish user data from partner data.

[0294] 9.2.4 Exemplary Classifier Training

[0295] This classifier can be employed using data from sensor records. For example, for each record, all partner data can be used as part of the training set, and an equal number of user timelines can be randomly selected to complete the training set.

[0296] Feature selection paths (such as sequential forward feature selection) can be used within cross-validation (e.g., tenfold cross-validation) to select a subset of these features that can be combined using classification methods such as logistic regression to best identify the presence of a user or only their partner. The selected features can then be used as a classifier for implementation in a processing device to identify people.

[0297] 9.2.5 Classifier Performance Evaluation

[0298] The trained classifier can be tested using data recorded from sensors. The test set can include all partner data and all user data.

[0299] 9.3.0. Other Exemplary Classifiers

[0300] 9.3.1 Separate classifiers for moving and non-moving cells

[0301] Many potential characteristics provide meaningless (i.e., non-a number / NaN or zero) outputs during data segments involving movement or during non-movement (respiration) data segments. For example, during movement segments, setting the optimizer and respiratory rate characteristics may result in NaN. Similarly, during clear breathing segments, the movement characteristic is equal to zero when no movement is detected. Therefore, in some versions, detection characteristics (such as those from Table 1) can be grouped into discrete groups, such as groups with characteristics that are meaningful during movement and groups with characteristics that are meaningful during clear breathing and when no movement is detected. These groups have some overlapping characteristics that are meaningful in both scenarios. A classifier model can then be applied to each characteristic group, and a movement marker set by analyzing the sensor signal can be used as a trigger to select the breathing or movement classifier model to use at each time period (see [link to relevant documentation]). Figure 19 Therefore, based on the movement marker, different sets of characteristics can be considered in the classification of a particular time period as being related to a particular user (e.g., user or partner).

[0302] 9.3.2 Developing a classifier using a unique user-based metric

[0303] In some versions, features can also be categorized or grouped as either subject-based or signal quality-based, and can be used to create two classifiers. Signal quality-based features are useful when training a model using real-life data (i.e., annotated data of two people in bed, where sensors capture data for 24 hours a day), where the user and partner are alone for some time periods and on specific sides of the bed (the user is always within a certain distance of the sensor, and the partner is at a different distance). Using this dataset, signal quality-based features, along with subject-based features, can be used to develop a 'set' classifier. Figure 19 Alternatively, by excluding signal quality-based characteristics and including only subject-based characteristics such as respiratory rate and movement, unannotated data from two different subjects can be combined to develop a 'subject classifier' (see [link to relevant documentation]). Figure 19Typical recordings typically do not include significant periods of partner data. In a typical recording, the user begins recording when they are in bed and ready to go to sleep, and ends when they wake up in the morning. When the primary user is not in bed, only partner data will be present; when they are present and the partner is not 'visible' at all, the signal reflects the primary user. Since both user and partner data will likely be captured using the same settings in this scenario, all signal quality-based characteristics will be excluded during the development of this 'subject classifier'. These classification methods may be better suited to different scenarios due to the nature of the training data used; that is, a classifier may be better suited to identifying partner data at the beginning or end of sleep, while a subject classifier may be better suited to identifying subjects taking naps during the day.

[0304] 9.3.3 Use of Historical Data

[0305] A general classification model can be used for the initial usage period, one night, or possibly longer. This allows for the aggregation of subject-specific historical data, which can be used to improve model performance over time. This data can be updated periodically, thus providing a gradually better-fitting and more accurate classifier as the usage period extends. Reference Figure 19 Exemplary methods for this classification model are illustrated. For example, during initial use or a setup period, a setup classification process at 1920 may estimate a set of features from feature extraction unit 1922 for user identification. Setup classification may employ processes for respiratory classification and / or movement classification at 1924 and 1926, respectively. A subject-specific classification process 1928 may also be included. A more specific subject classification process 1928 may also employ processes for respiratory classification and / or movement classification at 1930 and 1932, respectively. Process 1928 may estimate a set of features obtained from feature extraction unit 1922 that differs from the set of features estimated by the setup classification process at 1920; however, such sets of features may overlap.

[0306] The classifier combiner process 1940 can select, at 1940, classifications such as those from individual or combined subject classification and setting classification processes. The output is the probability (percentage) of user presence, thus making a binary decision. If the probability is less than a certain threshold, the output is "partner present," and if the probability is greater than the threshold, the output is "user present." The threshold can be based on variables that change over time, including (but not limited to) sleep duration or daytime. The classifier combiner can function to selectively select the outputs of different classifier processes as they change over time (e.g., the number of periods or aggregated usage data). For example, initially, the classifier combiner can select only the setting classification and subsequently (e.g., over a time period or number of periods) it can instead select only the subject classification. Optionally, the combiner can selectively modify the weights assigned to the two recognition outputs from different classification processes over time. For example, it can assign greater weight to the recognition made by the initial setting classification process and gradually increase the recognition weight of the subject classification process over time. Separate respiratory and movement classification processes exist to account for possible NaN or zero characteristics in one or the other data stream (see section “8.3.1 Separate Classifiers for Movement and Non-movement”); separate cardiac characteristic classifiers may also be included.

[0307] 9.4.0 Alternative Machine Learning Models

[0308] Alternatively, other supervised, semi-supervised, or unsupervised machine learning models can be implemented. For example, anomaly detection, support vector machines, or clustering classification methods can be used in other system implementations.

[0309] 10.0 Other methods for differentiating multiple users / subjects

[0310] Figure 20 This illustrates a process that can be implemented in a system for distinguishing N subjects monitored by N devices, such as in cases where multiple devices communicate with each other and / or where a centralized processing device (e.g., a server) processes data from sensors. Figure 20 The image shows only two sensor devices, 100-A and 100-B.

[0311] For example, each sensor device can access or process data received from other nearby sensor devices and apply a phased classification process. For example, Figure 20Sensor devices 1 and 2 can each initially assess whether another device has detected or is detecting a subject at 2010a, 2010b, such as by comparing sensor data from the two sensors or characteristics extracted from the sensor data. The "characteristic extraction" item refers to how data from device 2 will be processed on both device 1 (in the dashed area designated 100a) and device 2 (in the dashed area designated 100b). Similarly, data from device 1 will be processed on both devices. This allows for comparison between the data collected by the two devices and determines whether they are monitoring the same person or two individuals. If they are monitoring the same person, data with excellent signal quality can be used, and classification will be used to determine which is the user. If the devices are monitoring two individuals, then this figure ensures that the primary user is identified, and data analysis continues.

[0312] If this is not the case, the analysis on the device in the detection mode can proceed normally in 2012a and 2012b. That is, each device can then continue to independently monitor each different user by collecting / estimating the sensed data of each user (e.g., respiratory characteristics, cardiac characteristics, sleep characteristics, etc.).

[0313] However, if several devices detect the same pattern (i.e., the same person is detected by two sensors), parallel classifiers (e.g., two or more) can be processed on each device. Each device can then implement a classifier to confirm / identify whether the detected subject is a previously monitored user, such as by implementing a comparison with... Figure 19 The aforementioned process. For example, sensor data may be applied to set up classification processes 2020a, 2020b and / or subject-specific classification processes 2028a, 2028b designed to identify subjects (such as using historical data). Classification combiners 2040a, 2040b may then determine whether the data pertains to a previously monitored subject. If the subject is a known user, sensor monitoring then continues at 2012a, 2012b to continue monitoring the identified user, such as in a cumulative manner relative to previous periods. If the subject is not a known user, the sensor data may be discarded, or the sensors may continue monitoring the new user at 2012a, 2012b in a non-cumulative manner relative to any sensor / monitoring data from existing periods attributed to another user.

[0314] The potential advantage of this approach lies in its applicability to multiple individuals (e.g., two people using two sensors in a room) and its tailoring to specific individuals (e.g., two people) in a manner that does not require biometric analysis for a larger group. In this respect, the distinction can be a bounded problem associated with a limited group (e.g., family / bedmates or hospital patients). Furthermore, it allows the outputs of multiple sensors to share communication to achieve system-level separation of sleep / breathing / movement / heart rate, even when both sensors detect a single bed user when their detection ranges overlap. Therefore, it can significantly improve the robustness of deployed systems across a wide range of bed shapes and sizes, bedside tables, people, and other contexts.

[0315] What will be understood is that even Figure 20 The diagram illustrates two parallel or simultaneous sensor processes that can monitor more than two people by repeating the process block with other devices and sharing the data.

[0316] In this regard, various techniques can be implemented to share sensor data by sensor devices and to share sensor data between these sensor devices. For example, wireless communication (e.g., Wi-Fi Direct (Self-Organizing)) can be optionally implemented between sensors to enable sensor-to-sensor (point-to-point) communication. Other forms of communication (e.g., wired or other wireless communication) can also be implemented.

[0317] Additional optional features in 11.0

[0318] (i) In some cases, the sensor device can be integrated into a wirelessly rechargeable product (e.g., using a Qi wireless charger). Therefore, this rechargeable sensor product may have a battery (e.g., a lithium battery) and a coil to allow charging. The sensor product can then be charged during the day when placed near a wireless charger. After charging, it can then be inserted into a retaining structure for guiding the sensor product for use during monitoring (i.e., near a user at night). For example, the sensor may mate with a wall mount, stand, or bed seat, where the rechargeable sensor product is received. The wireless charger can then be used to charge other devices (e.g., smartphones) at night.

[0319] (ii) In some cases, the sensor devices may each include a microphone and / or an integrated camera sensitive to infrared light (e.g., an unfiltered IR or additionally optimized camera device for infrared detection). An infrared emitter (e.g., one or more IR LEDs with a diffuser / lens) may also be included. The sensor devices can then record user movement in a dark bedroom via their non-contact sensors and camera (recorded in camera memory and / or transmitted wirelessly). When events (e.g., SDB attacks, specific movement types such as PLM / RLS, abnormal movement, respiratory rate, as detected via non-contact or minimal contact sensors) or other events such as snoring, wheezing, coughing (e.g., detected via a non-contact microphone / converter) are detected in real time or at the end of the night, the processor can associate / correlate the detected events with food segments acquired during a normal time frame detected by the IR camera. The associated / correlated segments (and optionally periodic time frames) can then be evaluated together (e.g., for display on a display device such as a mobile phone or computer) to give the user an easy way to index the video to view the major events of the night. Thus, this allows the user to, for example, view SDB events in the video. For example, they can see themselves stop breathing and then resume breathing. Event-indexed videos can be viewed on smart devices (such as sensor devices with integrated monitors / displays or smartphones, tablets, or computers that communicate with sensor devices) and can, for example, allow for extremely quick viewing of nighttime sleep, for example, via events.

[0320] (iii) This camera-implemented sensor can also be implemented for remote viewing, such as in real-time or near real-time. For example, the sensor can be used to allow remote visual monitoring of the sleep and other states of patients or other users (e.g., the monitored subject waking up, being in deep sleep, etc.). Thus, the sensor can be used as an infant monitor / remote health professional monitor, such as in a clinic or hospital setting.

[0321] 12.0 Treatment

[0322] A range of treatments can be used to treat or improve respiratory symptoms, such as some versions of this technology, including treatment devices. Furthermore, these treatments can be used by other healthy individuals to prevent the development of respiratory disorders.

[0323] Continuous positive airway pressure (CPAP) therapy has been used to treat obstructive sleep apnea (OSA). The mechanism of action is that continuous positive airway pressure acts as an air splint and can prevent upper airway obstruction by pushing the soft palate and tongue forward and away from the posterior oropharyngeal wall.

[0324] Non-invasive ventilation (NIV) therapy provides ventilatory support to patients through the upper airway to help them breathe fully and / or maintain adequate oxygen levels in the body by performing some or all of the work of breathing. NIV support is delivered via a non-invasive patient interface. NIV has been used to treat CSR, OHS, COPD, NMD, and chest wall disorders.

[0325] 12.1 Diagnostic and Treatment Systems

[0326] In some versions of this technology, treatment can be provided through a therapeutic system or device that works in conjunction with the previously mentioned identification device / system. Such systems and devices can also be used to diagnose conditions without treating them.

[0327] The treatment system may include a respiratory pressure therapy device (RPT device), an air circuit, a humidifier, a patient interface, and data management.

[0328] A patient interface can be used to connect a breathing device to its user, for example, by providing a breathable airflow. The breathable airflow can be provided to the patient's nose and / or mouth via a mask, to the mouth via a tube, or to the user's trachea via a tracheostomy tube. Depending on the treatment to be applied, the patient interface can form a seal with, for example, the patient's facial area, thereby facilitating the delivery of gas at a pressure sufficiently different from ambient pressure (e.g., a positive pressure of about 10 cmH2O) to achieve the treatment. For other forms of treatment, such as oxygen or high-flow-rate air delivery, the patient interface may not include a seal sufficient to deliver air at a positive pressure of about 10 cmH2O to the airway.

[0329] 12.2 Respiratory Pressure Therapy (RPT) Device

[0330] Air pressure generators are known in a range of applications, such as industrial-scale ventilation systems. However, medical air pressure generators have specific requirements that are not met by more general air pressure generators, such as the reliability, size, and weight requirements of medical devices. Furthermore, even devices designed for medical use may have disadvantages related to one or more of the following: comfort, noise, ease of use, efficiency, size, weight, manufacturability, cost, and reliability.

[0331] One known RPT device for treating sleep-disordered breathing is the S9 Sleep Therapy System, manufactured by ResMed, which has been shown to be effective with CPAP therapy. Another example of an RPT device is a CPAP machine, such as the ResMed Stellar. TM The series of adult and pediatric ventilation machines can provide a range of patients with invasive and non-invasive, non-dependent ventilation therapy for the treatment of a variety of conditions, such as, but not limited to, NMD, OHS, and COPD.

[0332] RPT devices typically include a pressure generator, such as an electric blower or a compressed gas reservoir, and are 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. 12.1 Optional Exemplary Treatment Devices

[0333] As previously mentioned, in one form, this technology may include a device or apparatus for treating and / or monitoring respiratory symptoms. This device or apparatus may be an RPT device 4000 for supplying pressurized air to a patient 1000 via an air circuit 4170, thereby forming a patient interface 3000. In the following description, the RPT device may be referenced to Figure 21- Figure 24 Let's consider it.

[0334] 12.2 Patient Interface

[0335] 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 inflation chamber 3200, a positioning and stabilizing structure 3300, an air exchange port 3400, a connection port 3600 for connection 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, a single physical component may provide one or more functional aspects. In use, the seal-forming structure 3100 is configured to surround an inlet to the patient's airway to facilitate the supply of pressurized air to the airway.

[0336] 12.3 RPT device

[0337] According to one aspect of the present technology, an RPT device 4000 includes mechanical and pneumatic components 4100, electrical components 4200, and is programmed to execute one or more algorithms 4300. The RPT device 4000 may have an outer housing 4010, which is formed in two parts: an upper portion 4012 and a lower portion 4014. In one form, the outer housing 4010 may include one or more panels 4015. The RPT device 4000 includes a chassis 4016 that supports one or more internal components of the RPT device 4000. The RPT device 4000 may include a handle 4018.

[0338] The pneumatic path of the RPT device 4000 may include one or more air path objects, such as an inlet air filter 4112, an inlet silencer 4122, a pressure generator 4140 (e.g., a blower 4142) capable of supplying pressurized air, an outlet silencer 4124, and one or more converters 4270, such as a pressure sensor 4272 and a flow rate sensor 4274.

[0339] One or more air path components may be positioned within a removable, separate structure, referred to as pneumatic block 4020. Pneumatic block 4020 may be positioned within an outer housing 4010. In one embodiment, pneumatic block 4020 is supported by, or forms part of, a chassis 4016.

[0340] The RPT device 4000 may include a power supply 4210, one or more input devices 4220, a central controller 4230, a treatment device controller 4240, a pressure generator 4140, one or more protection circuits 4250, a memory 4260, a converter 4270, a data communication interface 4280, and one or more output devices 4290. 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.

[0341] 12.3.1 Mechanical and pneumatic components of the RPT unit

[0342] The RPT device 4000 may include one or more of the following components in an integral unit. In an alternative form, one or more of the following components may be positioned as separate units.

[0343] 12.3.1.1 Air Filter

[0344] One form of RPT device 4000 according to the present technology may include one or more air filters 4110.

[0345] In one configuration, the inlet air filter 4112 is positioned at the beginning of the pneumatic path upstream of the pressure generator 4140.

[0346] In one configuration, an outlet air filter 4114, such as an antibacterial filter, is positioned between the outlet of the pneumatic block 4020 and the patient interface 3000.

[0347] 12.3.1.2 Muffler

[0348] One form of RPT device 4000 according to the present technology may include one or more mufflers 4120.

[0349] In one embodiment of this technology, the inlet silencer 4122 is positioned in the pneumatic path upstream of the pressure generator 4140.

[0350] In one embodiment of this technology, the outlet silencer 4124 is positioned in the pneumatic path between the pressure generator 4140 and the patient interface 3000.

[0351] 12.3.1.3 Pressure Generator

[0352] In one form of this 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 enclosed in a volute. The pressure generator 4140 may be able to generate an air supply or airflow, for example, at a rate up to about 120 liters per minute, and at a positive pressure in the range of about 4 cmH2O to about 20 cmH2O, or in other forms up to about 30 cmH2O.

[0353] The pressure generator 4140 is controlled by the treatment device controller 4240.

[0354] In other forms, the pressure generator 4140 may be a piston-driven pump, a pressure regulator (e.g., a compressed air reservoir) connected to a high-pressure source, or a bellows.

[0355] 12.3.1.4 Converter

[0356] The transducer can be located inside or outside the RPT device. An external transducer can be located, for example, on or part of an air circuit such as a patient interface. An external transducer can be in the form of a non-contact sensor, such as a Doppler radar motion sensor that transmits or transfers data to the RPT device.

[0357] In one form of this technology, one or more converters 4270 are located upstream and / or downstream of pressure generator 4140. The one or more converters 4270 may be constructed and configured to generate relevant characteristics, such as velocity, pressure, or temperature, representing the airflow at that point in the pneumatic path.

[0358] In one form of this technology, one or more converters 4270 are positioned adjacent to the patient interface 3000.

[0359] In one embodiment, the signal from converter 4270 may be filtered, for example, by low-pass filtering, high-pass filtering, or band-pass filtering.

[0360] 12.3.1.5 Anti-overflow valve

[0361] In one embodiment of this technology, an anti-backflow valve 4160 is positioned between the humidifier 5000 and the pneumatic block 4020. The anti-backflow valve is constructed and configured to reduce the risk of water flowing upstream from the humidifier 5000 to, for example, the motor 4144.

[0362] 12.3.1.6 Air Circuit

[0363] According to one aspect of the technology, the air circuit 4170 is a conduit or tube that is constructed and configured in use to allow airflow between two components, such as the pneumatic block 4020 and the patient interface 3000.

[0364] 12.3.1.7 Oxygen Delivery

[0365] In one form of this technology, supplemental oxygen 4180 is delivered to one or more points in a pneumatic path, such as upstream of pneumatic block 4020, and then to air circuit 4170 and / or patient interface 3000.

[0366] 12.3.2 Electrical Components of RPT Unit

[0367] 12.3.2.1 Power Supply

[0368] In one embodiment of this technology, the power supply 4210 may be located inside the outer housing 4010 of the RPT device 4000. In another embodiment of this technology, the power supply 4210 may be located outside the outer housing 4010 of the RPT device 4000.

[0369] In one embodiment of this technology, power supply 4210 supplies power only to RPT device 4000. In another embodiment of the invention, power supply 4210 supplies power to both RPT device 4000 and humidifier 5000.

[0370] 12.3.2.2 Input Device

[0371] In one form of this technology, the RPT device 4000 includes one or more input devices 4220 in the form of buttons, switches, or dials to allow personnel to interact with the device. The buttons, switches, or dials can be physical devices or software devices accessed via a touchscreen. In one form, the buttons, switches, or dials can be physically connected to an external housing 4010, or in another form, the buttons, switches, or dials can communicate wirelessly with a receiver electrically connected to a central controller 4230.

[0372] In one form, the input device 4220 may be constructed and configured to allow a person to select values ​​and / or menu options.

[0373] 12.3.2.3 Central Controller

[0374] In one form of this technology, the central controller 4230 is a processor such as an x86 Intel processor suitable for controlling the RPT device 4000.

[0375] According to another form of the present technology, the central controller 4230 suitable for controlling the RPT device 4000 includes a processor based on an ARM Cortex-M processor from ARM Holdings. For example, an STM32 series microcontroller from STMicroelectronics can be used.

[0376] According to another alternative form of the present technology, the central controller 4230 suitable for controlling the RPT device 4000 includes a component selected from a family of ARM9-based 32-bit RISC CPUs. For example, a microcontroller from STMicroelectronics' STR9 series can be used.

[0377] In some alternative forms of this technology, a 16-bit RISC CPU can be used as the central controller 4230 of the RPT device 4000. For example, a processor from the MSP430 family of microcontrollers manufactured by Texas Instruments can be used.

[0378] In another form of this 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 yet another form, the central controller 4230 includes discrete electronic components.

[0379] The central controller 4230 can be configured to receive input signals from one or more converters 4270, one or more input devices 4220, and humidifier 5000.

[0380] The central controller 4230 is configured to provide output signals to one or more of the output device 4290, the treatment device controller 4240, the data communication interface 4280, and the humidifier 5000.

[0381] In some forms of this 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, which are stored in a non-transitory computer-readable storage medium such as memory 4260. In some forms of this technology, as previously discussed, the central controller 4230 may be integrated with the RPT device 4000. However, in some forms of this technology, some methods may be performed by a remote positioning device. For example, a remote positioning device may determine the control settings of a ventilator or detect respiratory-related events by analyzing stored data from sensors such as those described herein.

[0382] While the central controller 4230 may include a single controller interacting with various sensors (e.g., converter 4270), a data communication interface 4280, a memory 4260, and other means, the functionality of controlling 4230 may be distributed across more than one controller. Therefore, the term "central" as used herein is not intended to limit the structure to a single controller or processor controlling other means. For example, alternative structures may include a distributed controller structure involving more than one controller or processor. This could include, for example, a separate local (i.e., within the RPT device 4000) or remotely located controller executing some algorithms 4300, or even more than one local or remote memory storing some algorithms. Additionally, algorithms, when represented as computer programs, may include high-level human-readable code (e.g., C++, Visual Basic, other object-oriented languages, etc.) or low / machine-level instructions (Assembler, Verilog, etc.). Depending on the function of the algorithm, this code or instructions may be burned into a controller, such as an ASIC or DSP, or executable to be loaded onto a DSP or general-purpose processor, which is then specifically programmed to perform the tasks required by the algorithm.

[0383] 12.3.2.4 Clock

[0384] RPT device 4000 may include clock 4232, which is connected to central controller 4230.

[0385] 12.3.2.5 Treatment device controller

[0386] In one form of this technology, the treatment device controller 4240 is a treatment control module 4330, which forms part of an algorithm 4300 executed by the central controller 4230.

[0387] In one embodiment of this technology, the treatment device controller 4240 is a dedicated motor control integrated circuit. For example, in one embodiment, the MC33035 brushless DC motor controller manufactured by ON Semiconductor is used.

[0388] 12.3.2.6 Protection Circuit

[0389] The RPT device 4000 according to the present technology may include one or more protection circuits 4250.

[0390] According to this technology, one form of protection circuit 4250 is an electrical protection circuit.

[0391] According to this technology, one form of protection circuit 4250 is a temperature or pressure safety circuit.

[0392] 12.3.2.7 Memory

[0393] According to one embodiment of the present technology, the RPT device 4000 includes a memory 4260, such as non-volatile memory. In some embodiments, the memory 4260 may include battery-powered static RAM. In some embodiments, the memory 4260 may include volatile RAM.

[0394] The memory 4260 may be located on PCBA 4202. The memory 4260 may be in the form of EEPROM or NAND flash memory.

[0395] Alternatively or alternatively, the RPT device 4000 includes a removable memory 4260, such as a memory card manufactured according to the Secure Digital (SD) standard.

[0396] In one form of this technology, memory 4260 is used as a non-transitory computer-readable storage medium storing computer program instructions representing one or more methods described herein, such as one or more algorithms 4300.

[0397] 12.3.2.8 Converter

[0398] The converter can be located inside the device 4000 or outside the RPT device 4000. An external converter can be located, for example, on or part of the air delivery circuit 4170 (e.g., at the patient interface 3000). The external converter can be in the form of a non-contact sensor, such as a Doppler radar motion sensor that transmits or transfers data to the RPT device 4000.

[0399] 12.3.2.8.1 Flow rate

[0400] The flow rate converter 4274 according to this technology can be based on a differential pressure converter, such as the SDP600 series differential pressure converter from SENSIRION. The differential pressure converter is in fluid communication with a pneumatic circuit, wherein one of each pressure converter is connected to a corresponding first and second point in the flow-limiting element.

[0401] In one example, the signal representing the total flow rate Qt from the flow converter 4274 is received by the central controller 4230.

[0402] 12.3.2.8.2 Pressure

[0403] The pressure transducer 4272 according to this technology is positioned in fluid communication with a pneumatic path. A suitable example of the pressure transducer 4272 is a sensor from the HONEYWELL ASDX series. An alternative suitable pressure transducer is a sensor from the GENERALELECTRIC NPA series.

[0404] In use, the signal from the pressure converter 4272 is received by the central controller 4230. In one embodiment, the signal from the pressure converter 4272 is filtered before being received by the central controller 4230.

[0405] 12.3.2.8.3 Motor Speed ​​Converter

[0406] In one embodiment of this technology, a motor speed converter 4276 is used to determine the rotational speed of motor 4144 and / or blower 4142. The motor speed signal from the motor speed converter 4276 can be provided to the treatment device controller 4240. The motor speed converter 4276 can be, for example, a speed sensor, such as a Hall effect sensor.

[0407] 12.3.2.9 Data Communication System

[0408] In one embodiment of this technology, a data communication interface 4280 is provided, which is connected to a central controller 4230. The data communication interface 4280 can be connected to a remote external communication network 4282 and / or a local external communication network 4284. The remote external communication network 4282 can be connected to a remote external device 4286. The local external communication network 4284 can be connected to a local external device 4288.

[0409] In one embodiment, the data communication interface 4280 is part of the central controller 4230. In another embodiment, the data communication interface 4280 is separate from the central controller 4230 and may include an integrated circuit or a processor.

[0410] In one embodiment, the remote external communication network 4282 is the Internet. The data communication interface 4280 can connect to the Internet using wired communication (e.g., via Ethernet or fiber optic) or wireless protocols (e.g., CDMA, GSM, LTE).

[0411] In one form, the local external communication network 4284 utilizes one or more communication standards, such as Bluetooth or consumer infrared protocols.

[0412] In one form, the remote external device 4286 may be one or more computers, such as a cluster of networked computers. In another form, the remote external device 4286 may be a virtual computer rather than a physical computer. In either case, this remote external device 4286 may be accessed by appropriately authorized personnel (e.g., a clinician).

[0413] The local external device 4288 can be a personal computer, mobile phone, tablet, or remote control device.

[0414] In one form, the interface can communicate with sensors such as any sensor described herein (including, for example, an RF motion sensor).

[0415] 12.3.2.10 includes optional display and alarm output devices.

[0416] The output device 4290 according to this technology can take the form of one or more visual, auditory, and tactile units. The visual display can be a liquid crystal display (LCD) or a light-emitting diode (LED) display.

[0417] 12.3.2.10.1 Display Driver

[0418] The display driver 4292 receives characters, symbols, or images as input that are intended to be displayed on the display 4294, and converts them into commands that cause the display 4294 to display those characters, symbols, or images.

[0419] 12.3.2.10.2 Monitor

[0420] Display 4294 is configured to visually display characters, symbols, or images in response to commands received from display driver 4292. For example, display 4294 may be an eight-segment display, in which case display driver 4292 converts each character or symbol (such as the number "0") into eight logic signals that indicate whether the eight corresponding segments will be activated to display a specific character or symbol.

[0421] 12.3.3 RPT Device Algorithm

[0422] 12.3.3.1 Preprocessing Module

[0423] According to the present technology, the preprocessing module 4310 receives raw data as input from the converter 4270 (e.g., flow sensor 4274 or pressure sensor 4272) and directs one or more processing steps to calculate one or more output values, which will be used as input to another module such as the treatment engine module 4320.

[0424] In one form of this technology, the output values ​​include the interface or mask pressure Pm, the breathing flow rate Qr, and the leakage flow rate Ql.

[0425] In various forms of this technology, the preprocessing module 4310 includes one or more of the following algorithms: pressure compensation 4312, exhaust flow rate estimation 4314, leakage flow rate estimation 4316, breathing flow rate estimation 4317, ventilation volume determination 4311, target ventilation volume determination 4313, breathing rate estimation 4318, and standby frequency determination 4319.

[0426] 12.3.3.1.1 Pressure Compensation

[0427] In one form of this technology, pressure compensation algorithm 4312 receives as input a signal indicating the pressure in the pneumatic path adjacent to the outlet of pneumatic block 4020. Pressure compensation algorithm 4312 estimates the pressure drop in air circuit 4170 and provides an estimated pressure Pm as output in patient interface 3000.

[0428] 12.3.3.1.2 Exhaust port velocity estimation

[0429] In one form of this technology, the exhaust port velocity estimation algorithm 4314 receives the estimated pressure Pm as input in the patient interface 3000 and estimates the exhaust port air velocity Qv from the exhaust port 3400 in the patient interface 3000.

[0430] 12.3.3.1.3 Leakage velocity estimation

[0431] In one form of this technology, the leakage velocity estimation algorithm 4316 receives a total flow velocity Qt and an exhaust flow velocity Qv as inputs, and estimates the leakage velocity Ql. In another form, the leakage velocity estimation algorithm 4316 estimates the leakage velocity Ql by calculating the average of the difference between the total flow velocity and the exhaust flow velocity Qv over a period long enough to include multiple breathing cycles (e.g., about 10 seconds).

[0432] In one form, the leakage flow rate estimation algorithm 4316 receives the total flow rate Qt, the exhaust flow rate Qv, and the estimated pressure Pm as inputs to the patient interface 3000, and estimates the leakage flow rate Ql by calculating the leakage conductance and determining that the leakage flow rate Ql is a function of the leakage conductance and the pressure Pm. The leakage conductance can be calculated as the quotient of the low-pass filtered non-exhaust flow rate of the difference between the total flow rate Qt and the exhaust flow rate Qv, and the low-pass filtered square root of the pressure Pm, wherein the low-pass filtered time constant has a value long enough to include multiple respiratory cycles (e.g., approximately 10 seconds). The leakage flow rate Ql can be estimated as the product of the leakage conductance and the pressure function Pm.

[0433] 12.3.3.1.4 Respiratory Flow Estimation

[0434] In one form of this technology, the breathing flow estimation algorithm 4317 receives the total flow rate Qt, the exhaust port flow rate Qv, and the leakage flow rate Ql as inputs, and estimates the breathing air flow rate Qr by subtracting the exhaust port flow rate Qv and the leakage flow rate Ql from the total flow rate Qt.

[0435] In other forms of this technology, the respiratory flow estimation algorithm 4317 provides a value that serves as a proxy for the respiratory flow Qr. Possible proxies for respiratory flow include:

[0436] -Patient 1000's chest breathing movements

[0437] - Current drawn from pressure generator 4140

[0438] - Motor speed of pressure generator 4140

[0439] - Transthoracic resistivity of patient 1000

[0440] The respiratory flow proxy value can be provided by a converter 4270 (such as a motor speed sensor 4276) in the RPT device 4000 or by a sensor outside the RPT device 4000 (such as a respiratory motion sensor or a transthoracic resistivity sensor).

[0441] 12.3.3.1.5 Determining the ventilation rate

[0442] In one form of this technology, the ventilation volume determination algorithm 4311 receives the respiratory flow rate Qr as input and determines the current patient ventilation volume measurement value Vent.

[0443] In some implementations, the ventilation volume determination algorithm 4311 determines the ventilation volume measurement value Vent as an estimate of the actual patient's ventilation.

[0444] In one such implementation, the ventilation volume measurement Vent is half the absolute value of the respiratory flow Qr, which is optionally filtered by a low-band filter such as a second-order Bessel low-band filter at a 0.11 Hz angular frequency.

[0445] In one such implementation, the ventilation volume measurement Vent is an estimate of total alveolar ventilation (i.e., non-anatomical dead space ventilation). This requires estimating the anatomical dead space. Patient 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 actual patient ventilation volume measurement, for example as determined above, minus the product of the estimated anatomical dead space and the estimated spontaneous respiratory rate Rs.

[0446] In other embodiments, the ventilation volume determination algorithm 4311 determines a ventilation volume measurement Vent that is proportional to the actual patient ventilation broadly. One such embodiment estimates the peak respiratory flow Qpeak during the inspiratory portion of the cycle. This and many other procedures that involve sampling the respiratory flow Qr produce measurements that are proportional to the ventilation broadly, as long as the flow velocity waveform shape does not change too much (here, the shapes of two breaths obtained when the flow velocity waveforms of the breaths normalized by time and amplitude are similar). Some simple examples include the median positive respiratory flow, the median of the absolute value of the respiratory flow, and the flow standard deviation. Any linear combination of the absolute values of the respiratory flow using positive coefficients and even some absolute values using both positive and negative coefficients in any order statistic of the blood is approximately proportional to the ventilation. Another example is the average value of the respiratory flow in the middle K proportion (multiplied by time) of the respiratory portion, where 0 < K < 1. This is any large number of measurements that are precisely proportional to the ventilation when the flow waveform shape is constant.

[0447] In other forms, the ventilation volume determination algorithm 4311 determines a ventilation volume measurement Vent that is not based on the respiratory flow Qr, but is a proxy for the current patient ventilation, such as the partial pressure of oxygen saturation (SaO2) or carbon dioxide (PCO2), which is obtained from a suitable sensor attached to the patient 1000.

[0448] 12.3.3.1.6 Target ventilation volume determination

[0449] In one form of the present technology, the central controller 4230 obtains the current ventilation volume measurement Vent as an input and executes one or more target ventilation volume determination algorithms 4313 for determining the target value Vtgt of the ventilation volume measurement.

[0450] In some forms of the present technology, there is no target ventilation volume determination algorithm 4313, and the target ventilation Vtgt is pre-determined, for example, by hard-coding during the configuration of the RPT device 4000 or by manual input via the input device 4220.

[0451] In other forms of the present technology, such as in adaptive servo ventilation (ASV) therapy (described below), the target ventilation volume determination algorithm 4313 calculates the target ventilation Vtgt from a value Vtyp indicating the typical recent ventilation of the patient 1000.

[0452] In some forms of the adaptive servo ventilation therapy, the target ventilation Vtgt is calculated as a high proportion of the typical recent ventilation Vtyp but less than that recent ventilation. In such forms, the high proportion can be within the range (80%, 100%), or (85%, 95%), or (87%, 92%).

[0453] In other forms of adaptive servo ventilation therapy, the target ventilation Vtgt is calculated to be slightly greater than one times the typical latest ventilation Vtyp.

[0454] Typical recent ventilation (Vtyp) is the value around which the distribution of the current ventilation measurement (Vent) tends to cluster over multiple moments on a predetermined time scale; that is, the central tendency measurement of the current ventilation measurement within its recent history. In one implementation of the target ventilation determination algorithm 4313, the recent history is on the order of minutes, but in any case should be longer than the time scale of the Cheyne-Stokes breathing increase and decrease cycles. The target ventilation determination algorithm 4313 can determine the typical recent ventilation (Vtyp) from the current ventilation measurement (Vent) using a well-known type of central tendency measurement. One such measurement is the output of a low-band filter to the current ventilation measurement (Vent), where the time constant is equal to one hundred seconds.

[0455] 12.3.3.1.7 Respiratory Rate Estimation

[0456] In one form of this technology, the respiratory rate estimation algorithm 4318 receives the respiratory flow Qr of patient 1000 as input and generates an estimate of the patient's spontaneous respiratory rate Rs.

[0457] The respiratory rate estimation algorithm 4318 estimates the spontaneous respiratory rate Rs during a period when the patient is spontaneously breathing (R000), i.e., when the RPT device 4000 is not delivering "backup breathing" (described below). In some forms of this technology, the respiratory rate estimation algorithm 4318 estimates the respiratory rate during a period of less than 4 cmH2O in one embodiment when servo-assisted breathing (defined as pressure support minus minimum pressure support) is low, because such a period is more likely to reflect spontaneous breathing activity.

[0458] In some forms of this technology, respiratory rate estimation algorithm 4318 estimates the respiratory rate during sleep breathing periods, since the respiratory rate during these periods can be substantially different from the respiratory rate during wakefulness. Anxiety often results in a respiratory rate higher than the general respiratory rate during sleep. When a patient focuses on their breathing process, their respiratory rate is typically lower than those during normal wakefulness or sleep. Techniques such as those described in patent application number PCT / AU2010 / 000894, published as WO 2011 / 006199 (which is incorporated herein by reference), can be used to identify periods of wakefulness breathing by respiratory flow Qr.

[0459] In some forms of this technique, the respiratory rate estimation algorithm 4318 estimates the spontaneous respiratory rate Rs as one of many well-known statistical measures of the central tendency of the respiratory duration Ttot during the time period of interest. In such measures, it is desirable to exclude outliers or at least make them outlier robust. One such measure, the truncated mean, is outlier robust, where the lower K proportion and upper K proportion of the sorted respiratory durations are discarded and the mean is calculated over the remaining respiratory durations. For example, this reaches the upper and lower quartiles of the respiratory duration Ttot when K is 0.25. The median is another robust measure of central tendency, although this can occasionally produce unsatisfactory results when the distribution is strongly bimodal. The 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, where consecutive time intervals corresponding to illogical respiratory rates (e.g., greater than 45 breaths / min or less than 6 breaths / min) are excluded as outliers from the mean calculation. Other filtering mechanisms, used alone or in combination with interval filtering, exclude any breaths that are not part of a series of N consecutive spontaneous breaths, where N is some small integer (e.g., 3), and exclude the early and late breaths of a series of consecutive spontaneous breaths, such as excluding the first and last breaths of a series of four breaths. The basic principle of the latter mechanism is that the specific first and last breaths of a series of spontaneous breaths, as well as the typical early and late breaths, can be atypical; for example, the first spontaneous breath may occur due to awakening, and the last spontaneous breath may be longer due to a gradual decrease in respiratory drive, resulting in a spare breath at the end of the spontaneous breath series.

[0460] In some forms of this technology, the respiratory rate estimation algorithm 4318 is capable of making an initial estimate of the spontaneous respiratory rate Rs using an initial estimation time period to enable subsequent processing to begin in the treatment engine module 4320, and then continuously updating the estimated value of the spontaneous respiratory rate Rs using an estimation time period longer than the initial estimation time period. For example, the initial estimation time period could be 20 minutes of suitable spontaneous breathing, but the estimation time period could then be gradually increased to a certain maximum duration, such as 8 hours. Instead of a rolling window for this duration used for this estimation, a low-band filter of this respiratory duration can be used, which has a gradually longer reaction time (more precisely, a gradually decreasing corner frequency) as the period continues.

[0461] In some forms, short-term (e.g., 10-minute) measurements of central tendency suitable for processing, such as truncated means, can be inputs suitable for low-band filters to give an estimate Rs for varying time scales of hours or longer. The advantage of this is that potentially large amounts of respiratory duration data do not need to be stored and processed, which may occur when truncated means need to be calculated over moving windows of respiratory duration data spanning hours or days.

[0462] In some forms of this technique, the respiratory rate calculated over a short period of time, specifically within a single breath, can be used instead of the respiratory duration measured by the central tendency described above, thus giving generally similar but not identical results.

[0463] 12.3.3.1.8 Determination of Reserve Frequency

[0464] In one form of this technology, the standby frequency determination algorithm 4319 receives as input the spontaneous respiratory rate estimate Rs provided by the respiratory rate estimation algorithm 4318 and returns a "standby frequency" Rb. The standby frequency Rb is the frequency at which the RPT device 4000 will deliver standby breaths, i.e., continue to provide ventilatory support to the patient 1000 in the absence of significant spontaneous respiratory effort.

[0465] In one form of the preprocessing module 4310, there is no backup frequency determination algorithm 4319, and the backup frequency Rb is instead provided to the RPT device 4000 manually, for example, via the input device 4220, or hard-coded when the RPT device 4000 is configured.

[0466] In one form, called adaptive reserve frequency, the reserve frequency determination algorithm 4319 determines the reserve frequency Rb as a function of the spontaneous breathing rate Rs. In one implementation, this function determines the reserve frequency Rb as the spontaneous breathing rate Rs minus a constant such as 2 breaths / minute. In another implementation, the function determines the reserve frequency Rb as the spontaneous breathing rate Rs multiplied by a constant slightly less than one.

[0467] In one form, called a variable standby frequency, the standby frequency determination algorithm 4319 determines the standby frequency Rb as a function of time. The standby frequency Rb is initialized to a value called the spontaneous standby frequency (SBR), which is a fraction of the final target standby frequency, called the sustained timed standby frequency (STBR). This fraction can be two-thirds, three-quarters, or other positive values ​​less than one. When most of the minimum inspiration is spontaneous (i.e., patient-triggered) breathing, the SBR is the timeout period divided by the standby frequency. The STBR can be predetermined (e.g., by manual input or hard-coding as described above) or set to some typical respiratory rate such as 15 bpm. Over time after a previous spontaneous breath, the standby frequency Rb increases from the SBR to the STBR. This increase can be based on a predetermined curve (such as a series of steps) or a continuous linear curve. The curve is chosen so that the standby frequency Rb reaches the STBR after a predetermined interval. This interval can be measured in units of time, such as 30 seconds, or relative to patient breathing measurements, such as 5 breaths.

[0468] In some forms of variable backup frequencies, the predetermined interval at which the backup frequency Rb increases from the SBR to the STBR can be a function of the adequacy of the current ventilation. In one implementation, suitable for servo ventilation, where there is a target value Vtgt for the ventilation volume measurement, the backup frequency causes the STBR to approach more quickly to the extent that the current expiratory measurement Vent is less than the target ventilation Vtgt.

[0469] In one form of variable backup frequency, called adaptive variable backup frequency, backup frequency determination algorithm 4319 determines the backup frequency Rb as a function of the currently estimated spontaneous respiratory rate Rs provided by spontaneous breathing estimation algorithm 4318 and the time factor. Similar to variable backup frequency determination, adaptive variable backup frequency determination increases the backup frequency Rb from SBR to STBR at predetermined intervals, which may be a function of the current adequacy of ventilation. STBR can be initialized to a standard respiratory rate, such as 15 bpm. Once a reliable estimate of the spontaneous respiratory rate Rs is available from respiratory rate estimation algorithm 4318, STBR can be set to the currently estimated spontaneous respiratory rate Rs multiplied by some constant. SBR can be set to some fraction of STBR, as in variable backup frequency. In one form, this fraction (e.g., two-thirds) can be set to a lower value, such as 0.55, during the estimated time period of spontaneous respiratory rate Rs to accommodate occasional long respiratory durations in patients with relatively long respiratory rates, such as 12 breaths / minute.

[0470] In some forms, the constant multiplied to obtain the currently estimated spontaneous respiratory rate Rs from the STBR can be slightly higher than 1, such as 1.1, to provide more vigorous ventilation during respiratory arrest, which may be desirable in short respiratory arrests. This constant can be slightly lower than 1, such as 0.8, particularly when resynchronization difficulties become a problem in certain patients as they struggle to recover. A lower reserve frequency, achieved through a longer expiratory interval, makes resynchronization easier, and resynchronization often occurs during this period.

[0471] 12.3.3.2 Healing Engine Module

[0472] In one form of this technology, the treatment engine module 4320 receives one or more of the following as inputs from the patient interface 3000: pressure Pm, respiratory airflow to the patient Qr, and an estimated spontaneous breathing rate Rs, and provides one or more treatment parameters as outputs. In various forms, the treatment engine module 4320 includes one or more of the following algorithms: stage determination algorithm 4321, waveform determination algorithm 4322, inspiratory flow limit determination algorithm 4324, breathing cessation / insufficiency determination algorithm 4325, snoring detection algorithm 4326, airway occlusion determination algorithm 4327, and treatment parameter determination algorithm 4329.

[0473] 12.3.3.2.1 Stage Determination

[0474] In one form of this technology, the phase determination algorithm 4321 receives a signal indicating respiratory flow Qr as input and provides the phase Φ of the current respiratory cycle of the patient 1000 as output.

[0475] In some forms, this is called discontinuous phase determination, where the phase output Φ is a discontinuous variable. One implementation of discontinuous phase determination provides a dual-value phase output Φ with an inhalation value or an exhalation value (e.g., represented as 0 and 0.5 cycles, respectively) at the start of spontaneous inhalation and exhalation. The RPT device 4000, which effectively “triggers” and “cycles,” performs discrete phase determination because the trigger point and cycle point are the time intervals from exhalation to inhalation and from inhalation to exhalation, respectively. In one implementation of dual-value phase determination, the determined phase output Φ has a discrete value of 0 when the respiratory flow Qr exceeds a positive threshold (thus “triggering” the RPT device 4000), and a discrete value of 0.5 cycles when the respiratory flow Qr has a value more negative than a negative threshold (thus “cycling” the RPT device 4000).

[0476] Another implementation of discrete phase determination provides a three-valued phase output Φ having one of the values ​​of inhalation, intermediate inspiratory pause, and exhalation.

[0477] In other forms, this is called continuous phase determination, where the phase output Φ is a continuous value, such as changing from 0 to 1 revolution or from 0 to 2π radians. The RPT device 4000 performing continuous phase determination can trigger and cycle when the continuous phase reaches 0 and 0.5 revolutions, respectively. In one implementation of continuous phase determination, the continuous value of phase Φ is determined using fuzzy logic analysis of the respiratory flow rate Qr. The continuous value of the phase determined in this implementation is typically referred to as the "fuzzy phase." In one implementation of the fuzzy phase determination algorithm 4321, the following rules are applied to the respiratory flow rate Qr:

[0478] 1. If the respiratory flow is zero and increases rapidly, then this phase is 0 cycles.

[0479] 2. If the respiratory flow is a large positive value and is stable, then this phase is 0.25 cycles.

[0480] 3. If the respiratory flow is zero and decreases rapidly, then this phase is 0.5 cycles.

[0481] 4. If the respiratory flow is a large negative value and is stable, then the phase is 0.75 cycles.

[0482] 5. If the respiratory flow is zero and stable and the absolute value of the 5-second low-pass filter for the respiratory flow is large, then this phase is 0.9 cycles.

[0483] 6. If the respiratory flow is positive and the phase is exhalation, then the phase is 0 cycles.

[0484] 7. If the respiratory flow is negative and the phase is inhalation, then the phase is 0.5 cycles.

[0485] 8. If the absolute value of the 5-second low-pass filter for respiratory flow is large, then this phase gradually increases at a steady rate equal to the patient's respiratory rate, with a low-pass filter of 20 seconds.

[0486] The output of each rule can be represented as a vector, where the phase of the vector is the result of the rule and the magnitude of the vector is the degree of fuzziness of the rule as true. The degree of fuzziness for respiratory flow such as "large" and "stable" is determined using a suitable membership function. The results of the rules are then combined using functions such as graph centers, represented as vectors. In this combination, the rules can be weighted equally or differently.

[0487] In another implementation of the continuous phase determination, the inhalation time Ti and exhalation time Te are first estimated by the respiratory flow rate Qr. The phase Φ is then determined as either the general proportion of the inhalation time Ti since the previous trigger time, or 0.5 cycles plus half the proportion of the exhalation time Te since the previous cycle time (whichever is more recent).

[0488] In some forms of this technology, applicable to pressure support ventilation therapy (described below), the phase determination algorithm 4321 is configured to trigger even when the respiratory flow Qr is small, such as during respiratory arrest. Thus, the RPT device 4000 delivers a “backup breath” in the absence of spontaneous respiratory effort from the patient 1000. For this type of form, referred to as spontaneous / timed (S / T) mode, the phase determination algorithm 4321 may utilize a backup frequency Rb provided by the backup frequency determination algorithm 4319.

[0489] The phase determination algorithm 4321 using "fuzzy phases" can implement S / T mode by including a "momentum" rule in the fuzzy phase rules using a spare frequency Rb. The momentum rule serves to advance the continuous phase from exhalation to inhalation at the spare frequency Rb if there is no respiratory flow characteristic Qr that can be further advanced by other rules. In one implementation, the more realistically the ventilation measurement Vent (described below) is lower than the ventilation target value Vtgt (also described below), the higher the weight of the momentum rule in the combination. However, since pressure support increases rapidly in response to mild to moderate pulmonary hypoventilation (relative to target ventilation), ventilation may be quite close to the target. It is desirable to give the momentum rule a low weight when ventilation is close to the target to allow the patient to breathe at a significantly lower rate than at other times (when the patient is not in central respiratory arrest) without the ventilator pushing the breathing at a higher frequency. However, when momentum rules are given low weight, adequate ventilation can be easily achieved at a frequency far lower than the standby frequency under relatively high pressure support when ventilation exceeds a value below but close to the target ventilation. It is desirable for standby breathing to be delivered at a higher frequency because this will allow the target ventilation to be delivered at lower pressure support. This is desirable for many reasons, a key one being to reduce mask leakage.

[0490] In summary, the fuzzy phase determination algorithm 4321 for implementing S / T mode faces a dilemma in selecting the weights of the momentum rule used to incorporate the backup frequency Rb: if the momentum is too high, the patient may feel "pushed forward" by the backup frequency; if the momentum is too low, the support pressure may be excessive. Therefore, it is desirable to provide a method for implementing S / T mode that does not rely on the momentum rule described above.

[0491] The phase determination algorithm 4321 (which can be discrete or continuous in the absence of momentum rules) can be described as a timed standby method using a standby frequency Rb to implement the S / T mode. Timed standby can be implemented as follows: The phase determination algorithm 4321 attempts to detect inhalation initiated due to spontaneous breathing effort, for example, by monitoring the respiratory flow rate Qr as described above. If no inhalation initiated due to spontaneous breathing effort is detected within a time period following a continuous triggering time, which is equal to the reciprocal of the standby frequency Rb (the interval is called the standby timing threshold), the phase determination algorithm 4321 sets the phase output Φ to the inhalation value (thus triggering the RPT device 4000). Once the RPT device 4000 is triggered and standby breathing begins, the phase determination algorithm 4321 attempts to detect the start of spontaneous exhalation, for example, by monitoring the respiratory flow rate Qr, at which point the phase output Φ is set to the exhalation value (thus causing the RPT device 4000 to cycle).

[0492] If the standby frequency Rb increases during the time from SBR to STBR, similar to the variable standby frequency described above, the standby timing threshold begins to lengthen and then gradually shortens. That is, the RPT device 4000 begins with lower alertness and gradually becomes more alert to a lack of spontaneous breathing effort as more standby breaths are delivered. This RPT device 4000 is unlikely to "push forward" the patient when they prefer to breathe at a lower than standard frequency, while still delivering standby breaths when the patient needs them.

[0493] If the STBR in a variable standby frequency system adapts to the patient's estimated spontaneous breathing rate Rs, as in the adaptive variable standby frequency system described above, the standby breath will be delivered at a frequency that adapts to the patient's most recent spontaneous breathing effort.

[0494] 12.3.3.2.2 Waveform Determination

[0495] In one form of this technology, the treatment control module 4330 controls the pressure generator 4140 to provide a treatment pressure Pt that varies according to a waveform template П(Φ) as a function of the phase Φ of the patient's respiratory cycle.

[0496] In one form of this technology, waveform determination algorithm 4322 provides a waveform template П(Φ) with values ​​in the range [0,1], which are provided by stage determination algorithm 4321 in the structural domain of these values ​​Φ for use by treatment parameter determination algorithm 4329.

[0497] In one form, applicable to discrete or continuous value stages, the waveform template П(Φ) is a square wave last-stroke, having a value of 1 for stage values ​​up to and including 0.5 turns and a value of 0 for stage values ​​exceeding 0.5 turns. In another form, applicable to continuous value stages, the waveform template П(Φ) comprises two smooth curve portions: a smooth curve (e.g., raised cosine) rising from 0 to 1 for stage values ​​up to 0.5 turns and a smooth curve (e.g., exponential) decaying from 1 to 0 for stage values ​​exceeding 0.5 turns. An example of a “smooth and comfortable” waveform template is a “shark fin” waveform template, where the rise is raised cosine and the smooth decay is quasi-exponential (so that the limit of П is precisely zero when Φ approaches one turn).

[0498] In some forms of this technology, the waveform determination algorithm 4322 selects a waveform template П(Φ) from a waveform template library according to the RPT device 4000. Each waveform template П(Φ) in the library can be provided as a lookup table of value П for stage value Φ. In other forms, the waveform determination algorithm 4322 calculates the waveform template П(Φ) "on the fly" using a pre-determined functional form, which may be parameterized by one or more parameters (e.g., the time constant of the exponential curvature portion). The parameters of the functional form can be pre-determined or dependent on the current state of the patient 1000.

[0499] In some forms of this technique, applicable to discrete dual-valued phases of inhalation (Φ = 0 cycles) or exhalation (Φ = 0.5 cycles), waveform determination algorithm 4322 "instantly" calculates a waveform template П as a function of the discrete phase Φ and the time t (transition from exhalation to inhalation) measured since the most recent trigger time. In one such form, waveform determination algorithm 4322 calculates the waveform template (Φ, t) in two parts (inhalation and exhalation):

[0500]

[0501] Where П i (t) and П e (t) represents the inspiratory and expiratory portions of the waveform template П(Φ, t), and Ti is the inhalation time. In one such form, the inspiratory portion П of the waveform template... i (t) is a smooth increase from 0 to 1 parameterized by the rise time, and the expiratory portion of the template parameter is П. e (t) is a smooth descent from 1 to 0 parameterized by the descent time.

[0502] 12.3.3.2.3 Determination of Inspiratory Flow Limit

[0503] In one form of this technology, the processor executes one or more algorithms 4324 for detecting inhalation flow limitation (partial obstruction).

[0504] In one form, algorithm 4324 receives the respiratory flow signal Qr as input and provides as output a measure of the degree to which the inspiratory portion of the breath exhibits inspiratory flow restriction.

[0505] In one form of this technique, the inspiratory portion of each breath is identified based on an estimated stage Φ at each moment. For example, the inspiratory portion of a breath is a value of respiratory flow where stage Φ is less than or equal to 0.5. A number of evenly spaced points (e.g., sixty-five) representing time points are interpolated along the inspiratory flow-time curve of each breath using an interpolator. The curve described by these points is then calibrated by a calibrator to have a consistent length (duration / time period) and a consistent area to remove the effects that alter respiratory rate and depth. The calibrated breaths are then compared in a comparator to a pre-scored template representing normal, unobstructed breathing. Breaths from this template that deviate from a specified threshold (typically 1 calibration unit) during inspiration at any time are excluded, such as those caused by coughing, sighing, swallowing, and hiccups, as determined by a test element. For the data that are not excluded, a moving average of the first such calibration point is calculated by a central controller 4230 used for performing several inspiratory events. This is repeated for the same inspiratory event at the second such point, and so on. Therefore, for example, sixty-five calibration data points are generated by the central controller 4230, and these data points represent the moving average of a number of inhalation events (e.g., three events). The moving average of the continuously updated values ​​of these (e.g., sixty-five) points is referred to below as the “calibration flow” and denoted as Qs(t). Alternatively, a single inhalation event may be used instead of the moving average.

[0506] From the calibrated flow, two shape factors related to the determination of partial blockage can be calculated.

[0507] A shape factor of 1 is the ratio of the average of the intermediate (e.g., thirty-two) calibration flow points to the average of the total (e.g., sixty-five) calibration flow points. A ratio greater than 1 indicates normal breathing. A ratio of 1 or less indicates obstructed breathing. A ratio of approximately 1.17 is considered the threshold between partially obstructed and unobstructed breathing, and corresponds to the degree of obstruction that would allow adequate oxygenation to be maintained in a typical user.

[0508] The shape factor 2 is calculated from the RMS deviation per unit calibrated flow, which is obtained for an intermediate (e.g., 32) point. An RMS deviation of approximately 0.2 units is considered normal. A zero RMS deviation is considered to indicate overall flow-limiting breathing. The closer the RMS deviation is to zero, the more flow-limited the breathing will be considered.

[0509] Shape factors 1 and 2 can be used as alternatives or in combination. In other forms of this technique, the number of sampling points, breathing points, and intermediate points may differ from those described above. Additionally, the threshold values ​​may not be those described.

[0510] 12.3.3.2.4 Determination of respiratory arrest and insufficiency

[0511] In one form of this technology, the central controller 4230 executes one or more algorithms 4325 for detecting respiratory arrest and / or insufficient breathing.

[0512] In one form, one or more breathing cessation / insufficiency detection algorithms 4325 receive respiratory flow Qr as input and provide a flag indicating that breathing cessation or insufficiency has been detected as output.

[0513] In one form, respiratory arrest is considered to be detected when a function of respiratory flow Qr decreases below a flow threshold for a predetermined period of time. This function can determine peak flow, relative short-term average flow, or a flow median between relative short-term average and peak flow (e.g., RMS flow). The flow threshold can be a relative long-term flow measurement.

[0514] In one form, insufficiency is considered to be detected when a function of respiratory flow Qr decreases below a second flow threshold for a predetermined period of time. This function can determine the peak flow, the relative short-term average flow, or the median of the relative short-term average and peak flow (e.g., RMS flow). The second flow threshold can be a relative long-term flow measurement. The second flow threshold is greater than the flow threshold used to detect respiratory arrest.

[0515] 12.3.3.2.5 Snoring Detection

[0516] In one form of this technology, the central controller 4230 executes one or more snoring detection algorithms 4326 for detecting snoring.

[0517] In one form, the snoring detection algorithm 4326 receives the respiratory flow signal Qr as input and provides a measure of the degree of snoring presence as output.

[0518] The snoring detection algorithm 4326 may include the step of determining the intensity of the flow rate signal in the range of 30-300 Hz. The snoring detection algorithm 4326 may also include the 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).

[0519] 12.3.3.2.6 Determination of airway patency

[0520] In one form of this technology, the central controller 4230 executes one or more algorithms 4327 for determining airway occupancy.

[0521] In one form, the airway occupancy algorithm 4327 receives the respiratory flow signal Qr as input and determines the signal power in the frequency range of approximately 0.75 Hz to approximately 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.

[0522] In one form, the frequency range of the sought peak value is a small forced oscillation within the treatment pressure Pt. In one embodiment, the forced oscillation has a frequency of 2 Hz and an amplitude of approximately 1 cmH2O.

[0523] In one form, the airway occupancy algorithm 4327 receives the respiratory flow signal Qr as input and determines the presence or absence of a cardiac-generated signal. The absence of a cardiac-generated signal is considered to indicate an airway closure.

[0524] 12.3.3.2.7 Determination of Treatment Parameters

[0525] In some forms of this technology, the central controller 4230 executes one or more treatment parameter determination algorithms 4329 for determining one or more treatment parameters using values ​​returned by one or more other algorithms in the treatment engine module 4320.

[0526] In one form of this technology, the treatment parameter is the instantaneous treatment pressure Pt. In one embodiment of this form, the treatment parameter determination algorithm 4329 uses the following equation to determine the treatment pressure Pt.

[0527] Pt=AΠ(Φ)+P0 (1)

[0528] in:

[0529] -A is the amplitude.

[0530] -Φ is the current stage value;

[0531] -П(Φ) is the waveform template value at the current stage (in the range of 0 to 1), and

[0532] -P0 is the base pressure.

[0533] If waveform determination algorithm 4322 provides a waveform template П(Φ) as a value lookup table indexed by stage Φ, treatment parameter determination algorithm 4329 applies equation (1) by locating the most recent lookup table entry to the current stage value Φ returned by stage determination algorithm 4321 or by inserting between two entries that span the current stage value Φ.

[0534] The values ​​of amplitude A and base pressure P0 can be determined by the treatment parameter algorithm 4329 according to the pressure treatment mode selected in the manner described below.

[0535] 12.3.3.3 Treatment Control Module

[0536] According to one aspect of the present technology, the treatment control module 4330 receives treatment parameters as input from the treatment parameter determination algorithm 4329 of the treatment engine module 4320, and controls the pressure generator 4140 to deliver an airflow according to the treatment parameters.

[0537] In one form of this technology, the treatment parameter is the treatment pressure Pt, and the treatment control module 4330 controls the pressure generator 4140 to deliver an airflow, the airflow having a mask pressure Pm at the patient interface 3000 equal to the treatment pressure Pt.

[0538] 12.3.3.4 Fault Status Detection

[0539] In one form of this technology, the processor executes one or more methods 4340 for detecting a fault state. The fault state detected by the one or more methods may include at least one of the following:

[0540] - Power failure (no power or insufficient power supply)

[0541] - Converter fault detection

[0542] -Detection of faults in the component

[0543] - Operating parameters outside the recommended range (e.g., pressure, flow rate, temperature, PaO2)

[0544] - Test the alarm to generate a fault that can detect alarm signals.

[0545] When detecting a fault condition, the corresponding algorithm uses one or more of the following signals to indicate the presence of a fault:

[0546] -Activate audio, visual, and / or motion (e.g., vibration) alarms.

[0547] - Send messages to external devices

[0548] -Record events

[0549] 12.4 Humidifier

[0550] In one form of this technology, a humidifier 5000 is provided to change the absolute humidity of the air or gas delivered to the patient relative to ambient air (e.g., as shown in the image). Figure 24 (As shown). Typically, the humidifier 5000 is used to increase the absolute humidity of the airflow before it is delivered to the patient's airway and to increase the temperature (relative to ambient air).

[0551] 12.5 Vocabulary List

[0552] For the purposes of this disclosure, one or more of the following definitions may be applied in certain forms of the present technology. Alternative definitions may be applied in other forms of the present technology.

[0553] 12.5.1 General Rules

[0554] Air: In some forms of this technology, air may be considered to mean atmospheric air, and in other forms of this technology, air may be considered to mean some other combination of breathable gases, such as oxygen-rich atmospheric air.

[0555] Respiratory pressure therapy (RPT): This involves delivering an air supply to the airway at a therapeutic pressure, which is typically positive relative to atmospheric pressure.

[0556] Continuous positive airway pressure (CPAP) therapy: This involves treating the patient with a nearly constant respiratory pressure throughout the entire respiratory cycle. In some forms, the pressure at the airway inlet will be slightly higher during expiration and slightly lower during inspiration. In other forms, the pressure will vary between different respiratory cycles, for example, increasing in response to an indication of partial upper airway obstruction and decreasing in the absence of such an indication.

[0557] Patient: A person, whether or not they have a respiratory illness.

[0558] Automated Positive Airway Pressure (APAP) therapy: CPAP therapy in which the treatment pressure is automatically adjustable between a minimum and a maximum, for example, varying with each breath, depending on the presence of an indication of an SBD event.

[0559] 12.5.2 Regarding the respiratory cycle:

[0560] Breathing apnea: According to some definitions, breathing apnea is considered to have occurred when airflow remains below a predetermined threshold for a period of time (e.g., 10 seconds). Breathing apnea is also considered to have occurred when, despite the patient's efforts, some obstruction in the airway prevents airflow. Central breathing apnea is considered to have occurred when breathing apnea is detected due to reduced or absent breathing effort.

[0561] Respiratory rate or respiratory rate (Rs): The patient’s spontaneous respiratory rate, usually measured in breaths per minute.

[0562] Duty cycle: The ratio of inspiratory time (Ti) to total respiratory duration (Ttot).

[0563] Effort (breathing): The work done by a spontaneously breathing person in trying to breathe.

[0564] The expiratory portion of the respiratory cycle: the time period from the start of expiratory flow to the start of inspiratory flow.

[0565] Flow limitation: A state of breathing in which increased effort by the patient does not result in a corresponding increase in flow rate. Flow limitation occurring during the inspiratory portion of the respiratory cycle can be described as inspiratory flow limitation. Flow limitation occurring during the expiratory portion of the respiratory cycle can be described as expiratory flow limitation.

[0566] Hypoventricular hypopnea: a decrease in flow rate, but not a cessation of flow. In one form, hypoventricular hypopnea is considered to have occurred when the flow rate remains below a threshold for an extended period. In one form in adults, any of the following can be considered hypoventricular hypopnea:

[0567] (i) A 30% reduction in the patient's breathing lasts for at least 10 seconds, plus a corresponding 4% reduction in saturation; or

[0568] (ii) The patient’s breathing is reduced (but at least 50%) for at least 10 seconds, accompanied by a decrease in saturation of at least 3% or arousal.

[0569] The inspiratory portion of the respiratory cycle: The time period from the start of inspiratory flow to the start of expiratory flow is considered the inspiratory portion of the respiratory cycle.

[0570] Airway openness: The degree to which the airway is open, or the extent to which the airway is open. An open airway is an open airway. Airway openness can be quantified, for example, a value of one (1) indicates an open airway, and a value of zero (0) indicates a closed airway.

[0571] Positive end-expiratory pressure (PEEP): The pressure in the lungs above atmospheric pressure at the end of expiration.

[0572] Peak flow (Q peak): The maximum flow rate during the expiratory portion of the respiratory flow waveform.

[0573] Respiratory flow / airflow, patient flow / airflow (Qr): These synonyms can be understood as the RPT device’s estimate of respiratory airflow, as opposed to the “real respiratory flow” or “real respiratory airflow” experienced by the patient, and are usually expressed in liters per minute.

[0574] Tidal volume (Vt): The volume of air inhaled or exhaled during normal breathing without additional effort.

[0575] Inspiratory time (Ti): The duration of the inspiratory portion of the respiratory flow waveform.

[0576] Expiratory time (Te): The duration of the expiratory portion of the respiratory flow waveform.

[0577] (Total) Time or Respiratory Duration (Ttot): The total duration between the start of the inspiratory portion of a respiratory flow waveform and the start of the inspiratory portion of a subsequent respiratory flow waveform.

[0578] Upper airway obstruction (UAO): This includes both partial and complete upper airway obstruction. This can be associated with a state of flow restriction, where the flow rate increases only slightly or even decreases with an increase in the pressure gradient across the upper airway (Starling resistance behavior).

[0579] Ventilation volume (Vent): A measurement of the total amount of gas exchanged by a patient's respiratory system. A measurement of ventilation volume can include one or both of the inspiratory and expiratory flow rates per unit time. When expressed in volumes per minute (V / min), this quantity is often referred to as "minute ventilation volume." Minute ventilation volume is sometimes given only in volume form and is understood as volumes per minute.

[0580] 12.5.3 RPT Device Parameters

[0581] Flow rate: The instantaneous volume (mass) of air delivered per unit time. Although flow rate and ventilation rate have the same dimension of volume or mass per unit time, flow rate is measured over a much shorter time period. Flow rate is nominally positive for the inspiratory portion of a patient's respiratory cycle and therefore negative for the expiratory portion. In some cases, the reference for flow rate will be a quantile reference, i.e., a quantity having only amplitude. In other cases, the reference for flow rate will be a vector reference, i.e., a quantity having both amplitude and direction. Flow rate will be given the sign Q. 'Flow rate' is sometimes shortened to simply 'flow rate'. Total flow rate Qt is the flow rate of air leaving the RPT device. Exhaust flow rate Qv is the flow rate of air leaving the exhaust port to allow flushing of exhaled gas. Leakage flow rate Ql is the flow rate of air accidentally leaking from the patient interface system. Respiratory flow rate Qr is the flow rate of air received into the patient's respiratory system.

[0582] Leakage: The word "leakage" is considered to refer to undesirable airflow. In one instance, a leak could occur due to an incomplete seal between the mask and the patient's face. In another instance, a leak could occur in a bend in the conduit leading to the surrounding environment.

[0583] Pressure: Force per unit area. Pressure can be expressed in units (including cmH2O, gf / cm²). 2 Measured within the range of (and hectopascals). 1 cmH₂O equals 1 g⁻¹ / cm³. 2And it is approximately 0.98 hectopascals. In this specification, unless otherwise stated, pressure is given in cmH2O. Pressure at the patient interface (mask pressure) is given by the symbol Pm, while treatment pressure is given by the symbol Pt, which represents the target value obtained at the current moment through mask pressure Pm.

[0584] 12.5.4 Ventilation Machine Terminology

[0585] Adaptive Servo Ventilator (ASV): A servo ventilator with a variable target ventilation volume instead of a fixed target ventilation volume. The variable target ventilation volume can be determined from some characteristics of the patient, such as the patient's respiratory characteristics.

[0586] Standby rate: The ventilator parameter that determines the rate of breathing (typically measured in breaths per minute) that the ventilator will deliver to the patient if it is not caused by spontaneous breathing effort.

[0587] Cyclic: Termination of the inspiratory phase of a ventilator. When a ventilator is delivering breaths to a patient who is breathing spontaneously, it is considered cyclic to stop delivering breaths at the end of the inspiratory portion of the respiratory cycle.

[0588] Positive expiratory airway pressure (EPAP): Baseline pressure, to which the pressure that varies during breathing is added to produce the desired mask pressure that the ventilator will attempt to achieve at a given time.

[0589] End-expiratory pressure (EEP): The desired mask pressure that the ventilator will attempt to achieve at the end of the expiratory phase of breathing. If the pressure waveform template П(Φ) is zero at end-expiratory, i.e., П(Φ) = 0 when Φ = 1, then EEP equals EPAP.

[0590] IPAP: The maximum expected mask pressure that the ventilator will attempt to achieve during the inspiratory phase of breathing.

[0591] Pressure support: Indicates the pressure increase during inspiratory breathing that exceeds the pressure increase during expiratory breathing, and generally refers to the pressure difference between the maximum pressure during inspiration and the baseline pressure (e.g., PS = IPAP – EPAP). In some cases, pressure support refers to the difference planned for the ventilator, rather than the actual difference it achieves.

[0592] Servo ventilator: A ventilator that measures the patient's ventilation volume, has a target ventilation volume, and adjusts the pressure support level to bring the patient's ventilation volume toward the target ventilation volume. Servo assist: Pressure support minus minimum pressure support.

[0593] Spontaneous / Timed (S / T): A mode of ventilator or other device that attempts to detect spontaneous breathing in a patient. However, if the device fails to detect breathing within a predetermined time period, it will automatically initiate the delivery of breaths.

[0594] Swing difference: an equivalent term for pressure support.

[0595] Triggered: When a ventilator delivers air to a spontaneously breathing patient, it is believed to be triggered by the patient's effort at the beginning of the inspiratory portion of the respiratory cycle.

[0596] Typical recent ventilation: Typical recent ventilation (Vtyp) is a measure of the central tendency of recent measurements of ventilation over a predetermined time scale.

[0597] Ventilator: A mechanical device that provides pressure support to a patient to perform some or all of the breathing work.

[0598] 12.6 Other Notes

[0599] This patent document contains copyrighted material. Because it appears in the patent documents or records of the Patent and Trademark Office, the copyright owner does not object to any person making a copy of this patent document or patent disclosure, but otherwise retains all copyright rights.

[0600] Unless explicitly stated in the context and a numerical range is provided, it should be understood that every intermediate value between the upper and lower limits of the range, up to one-tenth of the lower limit unit, and any other value or intermediate value within the range are broadly included within the scope of this invention. The upper and lower limits of these intermediate ranges may be included independently within the intermediate range and within the scope of this invention, but are subject to any explicitly excluded boundaries within the range. When the range includes one or both of these boundaries, the range excluding one or both of those included boundaries is also included within the scope of this invention.

[0601] Furthermore, in cases where one or more values ​​described in the present invention are implemented as part of the present invention, it should be understood that such values ​​may be approximate unless otherwise stated, and such values ​​may be used to the extent permitted or required by the practical implementation of the technology for any suitable valid number of digits.

[0602] 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 this invention pertains. While any methods and materials similar to or equivalent to those described herein may be used in the practice or testing of the techniques of this invention, a limited number of exemplary methods and materials are described herein.

[0603] When a particular material is deemed preferably used for constructing a component, an obvious alternative material with similar properties is used as its substitute. Furthermore, unless otherwise stated, any and all components described herein are to be understood as being capable of being manufactured and therefore can be manufactured together or separately.

[0604] It must be noted that, unless the context clearly specifies otherwise, the singular forms “a” and “the” as used herein and in the appended claims include their plural equivalents.

[0605] All publications mentioned herein are incorporated by reference to disclose and describe the methods and / or materials that are the subject of those publications. The publications discussed herein provide only disclosures prior to the filing date of this application. None of this document should be construed as an admission by prior invention that the present invention was not entitled to prior to such publications. Furthermore, the publication dates provided may differ from the actual publication dates, and may require independent verification.

[0606] Furthermore, in interpreting this disclosure, all terms should be interpreted in the broadest and most reasonable manner consistent with the context. The terms “comprising” and “including” should be interpreted as meaning that an element, component, or step referenced in a non-exclusive manner may be presented together, used together, or combined with other elements, components, or steps that are not expressly referenced.

[0607] The main headings used in the detailed description are included for the reader's convenience only and should not be used to limit the subject matter of the invention as found throughout the disclosure or claims. These headings should not be used to interpret the scope or limitation of the claims.

[0608] Although the present invention has been described with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the invention. In some cases, proper nouns, terms, and symbols may imply specific details not required for practicing the present invention. For example, although the terms "first" and "second" may be used, they are not intended to indicate any order unless otherwise specified, but rather to distinguish different elements. Furthermore, although process steps in a method may be described or illustrated in a certain order, this order is not necessary. Those skilled in the art will recognize that this order can be modified, and / or aspects of this order can be performed simultaneously or even concurrently.

[0609] Therefore, it should be understood that various modifications can be made to the exemplary embodiments and other arrangements can be designed without departing from the spirit and scope of the present invention.

[0610] It should also be understood that, unless otherwise indicated, any reference in this document to subjects known in the art does not imply that such subjects are generally known to a person skilled in the art to which this technology pertains.

[0611] Part List

[0612]

[0613]

[0614]

Claims

1. A physiological parameter monitoring system adapted to identify a person to monitor physiological parameters of the identified person, the system comprising: one or more sensors to monitor physiological parameters of one or more persons, the physiological parameters including cardiac parameters, and one or more processors configured to process signals from the one or more sensors to identify a person from an estimate of a biometric signature, the processing including an estimate of parameters including the cardiac parameters, wherein the one or more processors are further configured to determine one or more settings for a respiratory therapy device based on the estimate of the biometric signature.

2. The system of claim 1, wherein, the physiological parameters of the one or more persons further including one or both of respiratory parameters and movement parameters.

3. The system of claim 1, wherein, the monitoring includes detecting physiological characteristics of the person during sleep, wherein the processing includes one or more of: (a) detecting a sleep stage of the person; (b) detecting deep sleep of the person; (c) detecting REM sleep of the person.

4. The system of any one of claims 1 to 3, wherein, the monitoring includes detecting physiological characteristics of the person during wakeful hours.

5. The system of any one of claims 1 to 3, wherein, the one or more sensors include any one or more of: a radio frequency non-contact sensor; a biological motion sensor; a sensor of the respiratory therapy device; and a wearable sensor.

6. The system of any one of claims 1 to 3, wherein, the respiratory therapy device includes a means configured to supply pressurized air to an airway of the person, and wherein the means includes an electrically powered blower or a compressed gas reservoir.

7. The system of claim 6, wherein, the respiratory therapy device is a positive airway pressure device.

8. The system of claim 7, wherein, the respiratory therapy device is an adaptive servo-ventilation device.

9. The system of any one of claims 1 to 3, wherein, the one or more processors are configured to estimate the cardiac parameters to adjust respiratory therapy provided by the respiratory therapy device.

10. The system of claim 9, wherein, the one or more processors are configured to adjust the respiratory therapy upon detecting an elevated heart rate.

11. The system of any one of claims 1 to 3, wherein, the one or more processors are configured to retrain to identify the person if a biometric estimated in the identification is treated by the respiratory therapy device.

12. The system of any one of claims 1 to 3, wherein, the one or more processors are configured to track trends in heart rate, heart rate variability, and / or heart rate dynamics and respiratory rate dynamics.

13. The system of any one of claims 1 to 3, wherein, the one or more sensors are configured to collect heart rate data from the person when therapy is provided to the person by the respiratory therapy device and when therapy is not provided to the person by the respiratory therapy device.

14. The system of any one of claims 1 to 3, wherein, the one or more processors further include a control processor in communication with the one or more sensors, the control processor configured to communicate with the one or more sensors to adjust one or more detection control parameters of the one or more sensors based on the identification of the person.

15. The system of any one of claims 1 to 3, wherein, the estimate includes classifying parameters determined from the signals, the determined parameters including a plurality of: a spectral peak ratio; a set optimizer signature vector; a peak to trough ratio; a filtered respiratory rate; a respiratory variability measure; an in-band power of a sensor signal; a range of a sensor signal; a final respiratory rate; a ratio of a maximum amplitude to a minimum amplitude of a respiratory cycle; a high band power of a sensor signal; an average respiratory rate; a periodic leg movement activity detection; a rollover detection; and a post-processing movement.

16. The system of any one of claims 1 to 3, wherein, The estimation includes classifying parameters determined from the signals, the determined parameters including cardiac parameters and one or more of electrodermal response parameters, exercise intensity parameters, respiration parameters, blood pressure parameters, cough parameters, snore parameters, sleep parameters.

17. The system of claim 16, wherein, The estimation includes comparing the determined parameters to historical parameters of the identified person to determine trends.

18. The system of claim 17, wherein, The estimation includes calculating mean and / or standard deviation values for a period of time from the determined parameters.

19. The system of any one of claims 1 to 3, wherein, The one or more processors are further configured to allow or deny treatment operation by the respiratory therapy device based on the estimation of the biometric signature.

20. The system of any one of claims 1 to 3, wherein, The one or more processors are further configured to classify the identity of the user from parameters determined in the classification process.

21. The system of claim 20, wherein, The classification process includes any one or more of neural networks, hidden layer Markov models, logistic regression processes, linear kernel support vector machines, radial kernel support vector machines, and principal component analysis of the parameters prior to classification.

22. The system of any one of claims 1 to 3, wherein, The processed signals from the one or more sensors are used to determine compliance with a treatment plan.

23. A method of one or more processors of a physiological parameter monitoring system adapted to identify a person to monitor physiological parameters of the identified person, the method comprising: receiving one or more monitored physiological parameters of the person from one or more sensors, the monitored physiological parameters including cardiac parameters; processing signals from the one or more sensors to identify the person from an estimation of a biometric signature, the processing including estimating monitored physiological parameters, the monitored physiological parameters including the cardiac parameters; and determining one or more settings for a respiratory therapy device based on the estimation of the biometric signature.

24. The method of claim 23, wherein, The monitored physiological parameters are used to determine compliance with a treatment plan.

25. A non-transitory processor-readable medium having stored thereon processor- executable instructions which, when executed by one or more processors, cause the one or more processors to identify a person to monitor physiological parameters of the person, the processor-executable instructions comprising: instructions to access one or more monitored physiological parameters of the person from one or more sensors, the monitored physiological parameters including cardiac parameters; instructions to process signals from the one or more sensors to identify the person from an estimation of a biometric signature, the processing including estimating monitored physiological parameters, the monitored physiological parameters including the cardiac parameters; and instructions to determine one or more settings for a respiratory therapy device based on the estimation of the biometric signature.

26. The non-transitory processor-readable medium of claim 25, wherein, The monitored physiological parameters are used to determine compliance with a treatment plan.

27. A physiological parameter monitoring system adapted to identify a person to monitor physiological parameters of the identified person, the system comprising: one or more sensors to monitor one or more physiological parameters of the identified person, and including bio-motion sensors and microphones; and one or more processors configured to process signals from the one or more sensors to identify the person from which the one or more physiological parameters are being detected, the processed signals including bio-motion signals from the bio-motion sensor, the processing including estimation of characteristics including one or more breathing characteristics, heart characteristics, or movement characteristics, from which the one or more processors determine that the current user for which the one or more physiological parameters are being detected is the same as a previous user for which the one or more physical parameters were previously detected, wherein the one or more processors are configured to process the bio-motion signals to detect one or more sleep stages, and to associate the detected one or more sleep stages with the person identified from the estimation of characteristics, and wherein the one or more processors are configured to determine one or more settings of a respiratory therapy device based on the identified person.

28. The system of claim 27, wherein, the one or more processors are configured to process signals from the one or more sensors to identify the person from which the one or more physiological parameters are being detected, the processed signals including bio-motion signals from the bio-motion sensor, the processing including estimation of characteristics including one or more breathing characteristics, heart characteristics, or movement characteristics, from which the one or more processors determine that the current user for which the one or more physiological parameters are being detected is the same as a previous user for which the one or more physical parameters were previously detected, wherein the one or more processors are configured to process the bio-motion signals to detect one or more sleep stages, and to associate the detected one or more sleep stages with the person identified from the estimation of characteristics, and wherein the one or more processors are configured to determine one or more settings of a respiratory therapy device based on the identified person.

29. The system of claim 28, wherein, the one or more bio-signatures include sleep stage specific fingerprints.

30. The system of claim 28, wherein, the one or more processors include a user identification processor configured to utilize a bio-signature of the one or more bio-signatures to estimate characteristics determined from a sleep stage of the person.

31. The system of any one of claims 27-30, wherein, the one or more processors are configured to adjust one or more sensing control parameters for the one or more sensors based on the processing including estimation of characteristics when it is detected that the current user is not the previous user.

32. The system of claim 31, wherein, the one or more sensing control parameters include one or more of: distance, power, frequency, detection direction, and radiation pattern.

33. The system of any one of claims 27-30, wherein, the processed signals are used to determine compliance with a therapy plan.

34. A method of one or more processors of a physiological parameter monitoring system adapted to identify a person to monitor physiological parameters of the identified person, the method comprising: receiving one or more monitored physiological parameters of the identified person from one or more sensors, the one or more sensors including a bio-motion sensor and a microphone; processing signals from the one or more sensors to identify the person from which the one or more physiological parameters are being detected, the processed signals including bio-motion signals from the bio-motion sensor, the processing including estimation of characteristics by the one or more processors including one or more breathing characteristics, heart characteristics, or movement characteristics, from which the one or more processors determine that the current user for which the one or more physiological parameters are being detected is the same as a previous user for which the one or more physical parameters were previously detected, wherein the one or more processors process the bio-motion signals to detect one or more sleep stages, and to associate the detected one or more sleep stages with the person identified from the estimation of characteristics, and wherein the one or more processors are configured to determine one or more settings of a respiratory therapy device based on the identified person.

35. A non-transitory processor-readable medium having stored thereon processor-executable instructions that, when executed by one or more processors, cause the one or more processors to identify a person to monitor physiological parameters of one or more persons, the processor-executable instructions comprising: instructions to access one or more monitored physiological parameters of the identified person from one or more sensors, the one or more sensors comprising a bio-motion sensor and a microphone; instructions to process signals from the one or more sensors to identify the person, the processed signals comprising bio-motion signals from the bio-motion sensor, the processing comprising estimation of characteristics by the one or more processors, the characteristics comprising one or more respiration characteristics, cardiac characteristics, or movement characteristics, from which the one or more processors determine that a current user detecting the bio-parameters is the same as a previous user detecting the bio-parameters, wherein the one or more processors process the bio-motion signals to detect one or more sleep stages and associate the detected one or more sleep stages with the identified person from the estimation of characteristics, and wherein the one or more processors are configured to determine one or more settings of a respiratory therapy device based on the identified person.

36. A physiological parameter monitoring system adapted to identify a person to monitor physiological parameters of the identified person, the system comprising: one or more sensors to monitor physiological parameters of one or more persons, one or more video cameras, and one or more processors configured to process signals from the one or more sensors to identify the person from estimation of a biometric signature, wherein the one or more processors are configured to fuse biometric data from the one or more sensors with image data from the one or more video cameras to increase confidence in detection of a live person, or to identify a particular user, and wherein the one or more processors are configured to determine one or more settings of a respiratory therapy device based on the estimation of the biometric signature.

37. A physiological parameter monitoring system, the system comprising: one or more sensors to monitor physiological parameters of one or more persons, the physiological parameters comprising cardiac parameters, and one or more processors configured to process signals from the one or more sensors to estimate a biometric, the processing comprising estimation of parameters including the cardiac parameters, wherein the one or more processors are further configured to adapt a therapy with a respiration rate and a heart rate of the person when an expected biometric is detected, wherein upon detection of an elevated heart rate, a respiration profile is applied to a breathing assist device to reduce the heart rate to within an expected range.

38. A monitoring system, comprising: a device configured to monitor sleep, comprising one or more sensors configured to detect a degree of movement of a user when in bed; wherein the system comprises one or more processors configured to process signals comprising physiological movement of the user's chest due to breathing, and coarse and fine movement detection to detect user physical stimulation or discomfort.

39. A monitoring system comprising: a device configured to monitor sleep comprising one or more sensors configured to detect movement of a user when in bed, wherein the user's breathing rate, depth, and activity are monitored; and one or more processors configured to detect patterns in the user's breathing rate and patterns in the user's dynamics, wherein the system is configured to adaptively track the user's breathing rate baseline, movement features, and respiratory waveform shape over days, weeks, months, and / or years to establish a profile of the user's respiratory dynamics.

40. The system of claim 39, wherein, the user's breathing rate baseline, movement features, and respiratory waveform shape over days, weeks, months, and / or years are used to determine whether the user's condition is improving or worsening.

Images