Non-contact respiratory health monitoring device

Abstract

The non-contact health monitoring device can perform a beamsteering process that creates a plurality of beamsteered radar data streams from a received radar data stream. The non-contact health monitoring device can determine, for each spatial region radar data stream, a respiratory displacement of the user in relation to time. The non-contact health monitoring device can analyze, for each spatial region radar data stream, the respiratory displacement of the user in relation to time. The non-contact health monitoring device can output a screening result based on analyzing the respiratory displacement of the user.

Description

[0001] Cross-referencing of related applications

[0002] This application is related to PCT application US2019 / 031,290, filed May 8, 2019, entitled "Sleep Tracking and Vital Sign Monitoring Using Low Power RadioWaves". This application is also related to non-provisional application No. 16 / 990,705, filed August 11, 2020, entitled "Contactless Sleep Detection and Disturbance Attribution" (Agent's File No.: 090421-1183285). This application is also related to application filed August 11, 2020, entitled "Contactless Sleep Detection and Disturbance Attribution for Multiple Users". This application is also related to non-provisional application No. 16 / 990,714 entitled "Users" (Agent's file number: 090421-1183289). This application is also related to non-provisional application No. 16 / 990,720 entitled "Contactless Cough Detection and Attribution" (Agent's file number: 090421-1183290), filed on August 11, 2020. This application is also related to non-provisional application No. 16 / 990,726 entitled "Precision Sleep Tracking Using a Contactless Sleep Tracking Device" (Agent's file number: 090421-1190042), filed on August 11, 2020. This application is also related to application No. 16 / 990,726 entitled "Initializing Sleep Tracking" (Agent's file number: 090421-1190042), filed on August 11, 2020. This relates to non-provisional application No. 16 / 990,746 (Agent's Case No.: 090421-1198051) concerning "on a Contactless Health Tracking Device". For all purposes, the full disclosure of these applications is incorporated herein by reference. Background Technology

[0003] While people sleep, they may experience health problems they are unaware of, especially if they tend to sleep alone. While a person may experience symptoms such as frequent awakenings, fatigue, excessive sleepiness, or poor motor performance, they may not recognize these symptoms or be able to attribute them to a medical problem present in their sleep. For example, abnormal breathing that may be associated with sleep apnea can cause a person to experience some or all of these obvious symptoms. Such individuals may benefit from monitoring their health attributes, such as breathing, during sleep. Summary of the Invention

[0004] Various embodiments related to non-contact health monitoring devices are described. In some embodiments, a non-contact health monitoring device is described. The device may include a housing. The device may include a radar sensor, which may include multiple antennas. The radar sensor may be housed within the housing. The device may include a processing system, which includes one or more processors, communicating with the radar sensor, and is housed within the housing. The processing system may be configured to receive radar data streams from the radar sensor for each of the multiple antennas, thereby receiving multiple radar data streams. The processing system may be configured to perform a beam steering process that creates multiple target space region radar data streams from the received multiple radar data streams. The processing system may be configured to determine a user's respiratory displacement for each of the multiple target space region radar data streams. The processing system may be configured to analyze the user's respiratory displacement for each of the multiple target space region radar data streams. The processing system may be configured to output screening results based on time-correlated analysis of the user's respiratory displacement for each of the multiple target space region radar data streams.

[0005] Embodiments of such devices may include one or more of the following features: screening results may indicate that a user may be experiencing anomalous breathing. Beamguidance processes may include applying Weighted Delay and Summation (WDAS) beamguidance to create multiple target space region radar data streams. A processing system configured to determine the user's respiratory displacement for each of the multiple target space region radar data streams may include the processing system being configured to determine phase values ​​for each of the multiple target space region radar data streams. A processing system configured to analyze the user's respiratory displacement for each of the multiple target space region radar data streams may include the processing system being configured to apply a pre-trained machine learning model. The pre-trained machine learning model may be a neural network. Each of the multiple target space regions may correspond to a different part of the user's body. The multiple target space regions may include at least five target space regions. The processing system may also be configured to use a state machine to determine, based on the multiple radar data streams, that the user is asleep and not moving. The device may also include a wireless network interface housed in a housing and communicating with the processing system. The device may also include a touchscreen display housed in a housing and communicating with the processing system. The device may also include a microphone housed in the housing and communicating with the processing system. The device may also include a speaker housed in the housing and communicating with the processing system. The processing system may also be configured to receive verbal commands regarding the analysis of a user's respiratory displacement. The processing system may also be configured to transmit the verbal request to a cloud-based server system via a wireless network interface. The processing system may also be configured to receive commands via the wireless network interface in response to transmitting the verbal request to the cloud-based server system. The processing system may be further configured to output filtering results via a touchscreen display, at least in part, based on the received commands.

[0006] In some embodiments, a method for performing non-contact respiratory health monitoring is described. The method may include receiving radar data streams for each of a plurality of antennas in a radar subsystem, thereby receiving a plurality of radar data streams. The method may include performing a beamguiding process by one or more processors, which may create a plurality of target space region radar data streams from the received plurality of radar data streams. The method may include determining a user's respiratory displacement for each of the plurality of target space region radar data streams by one or more processors. The method may include analyzing a user's respiratory displacement for each of the plurality of target space region radar data streams by one or more processors. The method may include outputting screening results by one or more processors analyzing a user's respiratory displacement in a time-related manner based on each of the plurality of target space region radar data streams.

[0007] Implementations of this method may include one or more of the following features: the screening results may indicate that the user may be experiencing paradoxical breathing. The beamguided process may include applying weighted delay-sum (WDAS) beamguided guidance to create multiple target space region radar data streams. Determining the user's respiratory displacement for each target space region radar data stream may include determining phase values ​​for each target space region radar data stream. Analyzing the user's respiratory displacement for each target space region radar data stream may include applying a pre-trained machine learning model. The pre-trained machine learning model may be a neural network. Each target space region may correspond to a different part of the user's body. The multiple target space regions may include at least three target space regions. The method may also include using a state machine based on the multiple radar data streams to determine that the user is asleep and not moving.

[0008] In some embodiments, a non-contact health monitoring device is provided. The device may include a housing. The device may include a radar sensor comprising multiple antennas, wherein the radar sensor is housed within the housing. The device may include a processing system comprising one or more processors communicating with the radar sensor, the processing system also housed within the housing. The processing system may be configured to receive radar data streams from the radar sensor for each of the multiple antennas, thereby receiving multiple radar data streams. The processing system may be configured to perform a beamguided process that creates multiple target spatial region radar data streams from the received multiple radar data streams. The processing system may be configured to determine a user's respiratory displacement for each of the multiple target spatial region radar data streams. The processing system may be configured to analyze the user's respiratory displacement for each of the multiple target spatial region radar data streams. The processing system may be configured to output screening results based on time-correlated analysis of the user's respiratory displacement for each of the multiple target spatial region radar data streams. A first target spatial region radar stream in the target spatial region radar stream may correspond to the user's chest region. The second target space region radar stream in the target space region radar stream can correspond to the user's abdominal region. This analysis may include detecting a phase difference between respiratory displacement in the abdominal region and respiratory displacement in the chest region, sufficient to indicate anomalous breathing conditions. The output filtering results may include information representing the detected anomalous breathing conditions. Attached Figure Description

[0009] A further understanding of the nature and advantages of the various embodiments can be achieved by referring to the following accompanying drawings. In the drawings, similar parts or features may have the same reference numerals. Furthermore, various parts of the same type can be distinguished by adding a dash after the reference numerals and a second label to distinguish similar parts. If only the first reference numeral is used in the specification, the description applies to any similar parts having the same first reference numeral, regardless of the second reference numerals.

[0010] Figure 1 A diagram defining paradoxical breathing is shown.

[0011] Figure 2 An example of monitoring a user's breathing during sleep is shown.

[0012] Figure 3 A block diagram of an embodiment of a non-contact health monitoring system is shown.

[0013] Figure 4A An example of a non-contact health monitoring system is shown.

[0014] Figure 4BAn embodiment of frequency-modulated continuous wave radar radio waves output by a radar subsystem is shown.

[0015] Figure 5A An embodiment of a phase capture module that can be used as part of a non-contact health monitoring system is shown.

[0016] Figure 5B An embodiment of a non-contact health monitoring system for monitoring multiple spatial areas is shown.

[0017] Figure 6 An embodiment of a beamguide module for a non-contact sleep tracking device is illustrated, which targets the direction of sleep tracking.

[0018] Figure 7 The illustration shows an embodiment of an antenna layout for a radar subsystem that can be used in conjunction with a beamguide module of a contactless sleep tracking device.

[0019] Figure 8 Another embodiment of a beamguide module for targeting the direction of performing sleep tracking is illustrated.

[0020] Figure 9 An embodiment of a health tracking system is shown, which uses radar to detect human interactions and utilizes integrated beamguide to perform health monitoring.

[0021] Figure 10 An exploded view of an embodiment of a contactless health tracking device is shown.

[0022] Figure 11 An embodiment of a state system for determining when a person is asleep is shown.

[0023] Figure 12 An embodiment of a method for performing non-contact monitoring of a user's breathing is shown. Detailed Implementation

[0024] In paradoxical breathing, a person's chest moves out of sync with the diaphragm (or more generally, the abdomen). Figure 1 Figure 100 illustrates the definition of paradoxical breathing. Normal breathing example 110 shows a person's chest 101 and abdomen 102 tending to move approximately synchronously with each other, as shown by an expanding chest / abdomen 130 and a contracting chest / abdomen 131. That is, when a person breathes, the chest 101 and abdomen 102 tend to move outward and inward approximately simultaneously. In example 111, the chest 101 and abdomen 102 move inward and outward synchronously, producing a constructed additive movement waveform when the individual waveforms representing the movement of the person's chest and abdomen are summed.

[0025] Paradoxical breathing example 120 involves the chest 101 moving outward (expanding) when the abdomen 102 moves inward, and the chest 101 moving inward when the abdomen 102 moves outward, as shown in chest expansion / abdominal contraction state 140 and chest contraction / abdominal expansion state 141. This condition results in a reduction in the amount of air entering and exiting the lungs, thereby reducing the amount of oxygen present in the lungs and consequently lowering the oxygenation level in the patient's blood. Symptoms that a person may experience from paradoxical breathing may include: frequent awakenings, fatigue, excessive sleepiness (when awake), and / or poor motor performance. In addition, paradoxical breathing may be a symptom of larger health problems, such as sleep apnea. In paradoxical breathing, when the waveforms indicating the movement of the chest 101 and abdomen 102, as shown in example 121, are added together, there is a completely or significantly disruptive disturbance. Additionally or alternatively, the embodiments described in detail herein can help identify respiratory effort and patency of the airways, thus allowing for the estimation of the presence and differences between central and neurological (e.g., brain, brainstem, spine), physiological (e.g., genetic / chemostatic, heart failure, altitude, anesthetics, etc.), endogenous (e.g., neurological, muscular diseases) and exogenous (e.g., obstructive or restrictive lung diseases) problems.

[0026] The embodiments described in detail herein focus on non-contact health monitoring devices that can determine that paradoxical breathing or certain other respiratory conditions have occurred and can provide an indication of this condition to the user, possibly along with a recommendation to seek professional medical advice. This non-contact monitoring can occur while the user is sleeping. The non-contact health monitoring device can be installed at the bedside and, once set up, can determine when the user is asleep and can monitor the user for paradoxical breathing and / or other respiratory conditions that can be detected based on movement of the user's chest and / or abdomen. The non-contact health monitoring device uses radar to monitor various parts of the user's chest and / or abdomen. (For simplicity, the remainder of this document refers only to the user's chest; however, it should be understood that the term chest can be interpreted to include at least a portion of the user's abdomen.) Movement of different parts of the user's chest can be used to determine whether paradoxical breathing has occurred. Other respiratory conditions can be monitored using systems and methods similar to those described in detail herein. For example, similar systems and methods can be used to identify shallow breathing and irregular breathing.

[0027] Figure 2An embodiment 200 for monitoring a user's breathing while sleeping is illustrated using the embodiments described in detail herein. In embodiment 200, a user 201 sleeps on a bed 202. A non-contact health monitoring device 210 (“device 210”) is located beside the bed, for example, on a bedside table. Radar can be used to detect movement of the user 201 attributed to breathing. Sheets, blankets, clothing, and other inanimate objects may be penetrated by radio waves emitted by the non-contact health monitoring device 210, which may be millimeter waves. Therefore, the user 201's breathing can be monitored even though blankets, sheets, and clothing are in a direct path between the non-contact health monitoring device 210 and the user 201's chest.

[0028] Device 210 can monitor spatial regions 220. Each spatial region 220 may correspond to a different area of ​​the chest of user 201. Spatial regions 220 may partially overlap. In other embodiments, spatial regions 220 are spaced further apart using beamguide to increase, decrease, or eliminate overlap. In some embodiments, two adjacent spatial regions overlap by less than 50%. The number of spatial regions 220 may vary depending on the embodiment. In some embodiments, as few as two spatial regions may exist. In some embodiments, three, four, five, or more spatial regions 220 may exist. In other embodiments, six spatial regions may exist, as shown in embodiment 200.

[0029] Beamguided targeting can be used to individually target each space region 220 (e.g., space regions 220-1, 220-2, 220-3, 220-4, 220-5, and 220-6). Beamguided can be performed using a device 210 with multiple antennas, through processing performed by device 210. Thus, a single radar chirp of radio waves can be transmitted, and the reflected waves can be received using multiple antennas of device 210. The data obtained from the received reflected radio waves can have a beamguided process applied to effectively target a specific space region 220. This same data can be applied multiple times to the beamguided process (e.g., using different weights), allowing the same dataset to be used to target each space region 220. For example, various weights can be applied to create a beamform 215 to target space region 220-6. A separate set of weights can be applied to the same data to perform beamguided targeting of space region 220-5. Because multiple beamguided processes are used to process the same dataset based on reflected radio waves, movement occurring in one space region 220 can be monitored simultaneously and independently with movement in other space regions 220.

[0030] Figure 3A block diagram of an embodiment of a contactless health monitoring system 300 (“System 300”) is shown. System 300 may include: a contactless health monitoring device 301 (“Device 301”); a network 360; and a cloud-based server system 370. Device 301 may represent… Figure 2 An embodiment of device 210. Device 301 may include: a processing system 310; a data storage unit 318; a radar subsystem 320; an environmental sensor suite 330; a display 340; a wireless network interface 350; and a speaker 355. Typically, device 301 may include a housing that houses all the components of device 301. Figure 9 and Figure 10 Further details regarding this possible housing are provided.

[0031] Processing system 310 may include one or more processors configured to perform various functions, such as those of radar processing module 312, sleep state detection engine 314, and respiration monitoring engine 316. Processing system 310 may include one or more dedicated or general-purpose processors. Such dedicated processors may include processors specifically designed to perform the functions detailed herein. Such dedicated processors may be ASICs or FPGAs, which are general-purpose components physically and electrically configured to perform the functions detailed herein. Such general-purpose processors may execute dedicated software stored using one or more non-transitory processor-readable media, such as random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD).

[0032] The radar subsystem 320 (also known as a radar sensor) can be a single integrated circuit (IC) that transmits, receives, and outputs data indicating the received reflected waveform. The output of the radar subsystem 320 can be analyzed using the radar processing module 312 of the processing system 310. (About...) Figure 4A Further details are provided regarding radar subsystem 320 and radar processing module 312.

[0033] Device 301 may include one or more environmental sensors, such as all, one, or some combination of environmental sensors provided as part of environmental sensor kit 330. Environmental sensor kit 330 may include: a light sensor 332; a microphone 334; a temperature sensor 336; and a passive infrared (PIR) sensor 338. In some embodiments, several or multiple instances of all these sensors may be present. For example, in some embodiments, multiple microphones may be present. Light sensor 332 may be used to measure the amount of ambient light present in the general environment of device 301. Microphone 334 may be used to measure the level of ambient noise present in the general environment of device 301. Temperature sensor 336 may be used to measure the ambient temperature of the general environment of device 301. PIR sensor 338 may be used to detect moving living organisms (e.g., people, pets) within the general environment of device 301. Other types of environmental sensors are possible. For example, a camera and / or humidity sensor may be incorporated as part of environmental sensor kit 330. As another example, an active infrared sensor may be included. In some embodiments, some data, such as humidity data, may be obtained from a nearby weather station that has data available via the Internet. In some embodiments, active acoustic sensing methods may be implemented, including but not limited to sonar and ultrasound, and including a single or array of sound sources and / or receivers. Such arrangements may be used as one or more auxiliary sensing modes in conjunction with other sensors and methods described herein.

[0034] In some embodiments, one, some, or all of the sensors in the environmental sensor suite 330 may be external to the device 301. For example, one or more remote environmental sensors may communicate with the device 301 directly (e.g., via a direct wireless communication method, via a low-power mesh network) or indirectly (e.g., via one or more other devices, via a low-power mesh network, via a network access point, via a remote server).

[0035] Device 301 may include various interfaces. Display 340 may allow processing system 330 to present information for viewing by one or more users. Wireless network interface 350 may allow communication using a wireless local area network (WLAN) such as WiFi-based networks. Speaker 355 may allow output of sound such as synthesized speech. For example, a response to a spoken command received via microphone 334 may be output via speaker 355 and / or display 340. Spoken commands may be analyzed locally by device 301 or may be transmitted via wireless network interface 350 to a cloud-based server system 370 for analysis. Responses based on the analysis of spoken commands may be sent back to device 301 via wireless network interface 350 for output via speaker 355 and / or display 340. Additionally or alternatively, speaker 355 and microphone 334 may be configured together for active acoustic sensing, including ultrasonic acoustic sensing. Additionally or alternatively, other forms of wireless communication are also possible, such as using low-power wireless mesh network radios and protocols (e.g., Thread) to communicate with various smart home devices. In some embodiments, a wired network interface, such as an Ethernet connection, can be used to communicate with the network. Furthermore, the evolution of wireless communication towards fifth-generation (5G) and sixth-generation (6G) standards and technologies offers greater throughput and lower latency, enhancing mobile broadband services. 5G and 6G technologies also provide new services for vehicle-to-everything (V2X), fixed wireless broadband, and the Internet of Things (IoT) through control and data channels. Such standards and technologies can be used for communication via device 301.

[0036] Low-power wireless mesh network radios and protocols can be used to communicate with power-constrained devices. Power-constrained devices can be dedicated battery-powered devices. Such devices may rely entirely on one or more batteries for power, thus keeping the amount of power used for communication low, reducing the frequency of needing to replace one or more batteries. In some embodiments, power-constrained devices may have the ability to perform communication over relatively high-power networks (e.g., WiFi) and low-power mesh networks. Power-constrained devices may not frequently use relatively high-power networks to conserve power. Examples of such power-constrained devices include environmental sensors (e.g., temperature sensors, carbon monoxide sensors, smoke sensors, motion sensors, presence detectors) and other forms of remote sensors.

[0037] It is worth noting that some embodiments of device 301 do not have any camera or video camera. By not including a vehicle-mounted camera, nearby users can be assured of their privacy. For example, device 301 can typically be installed in a user's bedroom. For a variety of reasons, users do not want to be in such a private space or have a camera pointed at them while they are sleeping. In other embodiments, device 301 may have a camera, but the camera lens may be obstructed by a mechanical lens shutter. To use the camera, the user may need to physically open the shutter to allow the camera to see the environment around device 301. When the shutter is closed, user privacy can be ensured relative to the camera.

[0038] Wireless network interface 350 allows wireless communication with network 360. Network 360 may include one or more public and / or private networks. Network 360 may include a private local wired or wireless network, such as a home wireless LAN. Network 360 may also include a public network, such as the Internet. Network 360 may allow device 301 to communicate with a remotely located cloud-based server system 370.

[0039] The cloud-based server system 370 can provide various services to the device 301. Regarding respiratory data, the cloud-based server system 370 may include processing and storage services for respiratory-related data. Although Figure 3 The embodiments involve processing system 310 performing sleep state detection and monitoring user breathing; however, in other embodiments, such functions may be performed by cloud-based server system 370. Furthermore, in addition to using data storage 318 to store breathing data, or as an alternative, breathing-related data may be stored by cloud-based server system 370, for example, mapped to a public user account linked to device 301. If multiple users are monitored, breathing data may be stored and mapped to a master user account or the account of the corresponding user.

[0040] The cloud-based server system 370 can additionally or alternatively provide other cloud-based services. For example, device 301 can be additionally used as a home assistant device. The home assistant device can respond to voice queries from a user. In response to detecting a spoken voice trigger phrase, device 301 can record audio. The audio stream can be transmitted to the cloud-based server system 370 for analysis. The cloud-based server system 370 can perform a speech recognition process, using a natural language processing engine to understand queries from the user, and provide responses output by device 301 as synthesized speech, output to be displayed on display 340, and / or commands to be executed by device 301 (e.g., increasing the volume of device 301) or sent to another smart home device. Furthermore, queries or commands can be submitted to the cloud-based server system 370 via display 340, which may be a touchscreen. For example, device 301 can be used to control various smart home devices or home automation devices. Such commands can be sent directly by device 301 to the device to be controlled or can be sent via the cloud-based server system 370.

[0041] Based on the data output by the radar processing module 312, the sleep state detection engine 314 can be used to determine whether the user is likely asleep or awake. The sleep state detection engine 314 can proceed via a state machine, such as regarding... Figure 11 Alternatively, the state identified by using this state machine can be used to determine whether a user is awake or asleep. For example, if it is determined that the user is in bed and has been still for at least a period of time, the user can be identified as asleep. The output of the sleep state detection engine 314 can be used by the breathing monitoring engine 316 to determine when the user's breathing should be analyzed and recorded and / or when the breathing monitoring engine 314 should be activated or deactivated. For example, the user's breathing can only be analyzed when the user is asleep and at rest (except for movement due to vital signs).

[0042] Data storage 318, which may include one or more non-transitory processor-readable media, may store data obtained from the respiratory monitoring engine 316. The obtained data may include indications of the type, time, duration, and / or intensity of abnormal respiratory conditions (e.g., paradoxical breathing) experienced by the user. Other respiratory-related data, such as the user's respiratory rate (e.g., breaths per minute), may be tracked and stored. Changes in respiratory rate may be monitored over nighttime and longer time periods, such as days, weeks, months, and years. In some embodiments, additionally or alternatively, a cloud-based server system 370 may be used to store such data.

[0043] The respiratory data output engine 319 can compile or summarize data received from the machine learning model 540 and / or stored by the data storage 318, and can output such a summary of data using the display 340 and / or speaker 355. For example, a summary of the user's breathing, including any observed medical problems, can be output automatically in the morning or in response to a user request for a summary from the previous night. The summary may include a visualization of the user's breathing rate and / or an indication of any identified medical conditions (and, possibly, a definition of the identified medical conditions). The summary may also include trends identified over a longer period, such as whether the medical condition is improving or worsening. Some or all of the summary can be output via speaker 355 using synthesized speech. The summary can also be stored locally using the data storage 318 or remotely using a cloud-based server system 370.

[0044] Figure 4A An embodiment of a beamguided non-contact health monitoring system 400 (“System 400A”) is illustrated. System 400A allows for non-contact health assessments of a user. Non-contact health assessments may include status determination, such as whether the user is in bed, stationary in bed, or moving in bed. The user’s status can be used to determine whether health data about the user should be recorded and / or whether other systems should be activated. For example, System 500B may be activated only when the user is determined to be in bed and stationary. When the user is determined to be in bed and stationary, small movements of the user can be monitored. These small movements of the user’s body may correspond to vital signs, such as breathing and heart rate. This vital sign data can be recorded, stored, and analyzed.

[0045] System 400A may include: radar subsystem 405 (which may represent an embodiment of radar subsystem 320); radar processing module 410 (which may represent an embodiment of radar processing module 312); and beam guidance module 430. Beam guidance performed by beam guidance module 430 can focus radar reflections from areas where the user may be present and ignore or at least reduce the use of radar reflections from objects that cause interference (such as nearby walls or other stationary objects).

[0046] Radar subsystem 405 may include an RF transmitter 406, an RF receiver 407, and radar processing circuitry 408. The RF transmitter 406 may transmit radio waves, for example, in the form of a continuous wave (CW) radar. The RF transmitter 406 may use a frequency modulated continuous wave (FMCW) radar. The FMCW radar may operate in burst mode, such as a reference... Figure 4BIn detail, a conversion process is performed to create virtual continuous samples. Radar subsystem 405 can operate in burst mode or continuous sparse sampling mode. In burst mode, RF transmitter 406 can output one frame or multiple chirp strings with relatively short intervals between chirps. Each frame may be followed by a relatively long period of time until the next frame. In continuous sparse sampling mode, no frames or chirp strings are output; instead, chirps are output periodically. The intervals between chirps in continuous sparse sampling mode can be longer in duration than the intervals between intra-frame chirps in burst mode. In some embodiments, radar subsystem 405 can operate in burst mode, but the raw chirp waterfall data from each burst can be combined (e.g., averaged) to create simulated continuously sparse sampled chirp waterfall data. In some embodiments, raw waterfall data collected in burst mode may be more suitable for gesture detection, while raw waterfall data collected in continuous sparse sampling mode may be more suitable for sleep tracking, vital sign detection, and general health monitoring. Gesture detection can be performed by other hardware or software components that use the output of radar subsystem 405 (not shown).

[0047] The frequency of the transmitted radio waves may scan repeatedly from low to high frequencies (or vice versa). The power level used for transmission can be very low, giving the radar subsystem 405 an effective range of a few meters or less. The RF receiver 407 may include one or more antennas and may receive radio waves reflected from nearby objects by the radio waves emitted by the RF transmitter 406. The reflected radio waves can be interpreted by the radar processing circuitry 408. The radar processing circuitry 408 may output raw waveform data, which may also be referred to as raw chirp waterfall data, for analysis by a separate processing entity. The radar subsystem 405 may be implemented as a single integrated circuit (IC), or the radar processing circuitry 408 may be a separate component from the RF transmitter 406 and the RF receiver 407. In some embodiments, the radar subsystem 405 is integrated as part of the device 301 such that the RF transmitter 406 and the RF receiver 407 are pointed in the same direction as the display 340. In other embodiments, an external device including the radar subsystem 405 may be connected to the device 301 via wired or wireless communication. For example, the radar subsystem 405 may be an add-on device to a home assistance system.

[0048] For radar subsystem 405, if FMCW is used, an unambiguous FMCW range can be defined. Within this range, the distance to an object can be accurately determined. However, outside this range, detected objects may be incorrectly interpreted as being closer than objects within the unambiguous range. This incorrect interpretation may be due to the frequency of the mixed signal and the sampling rate of the ADC used by the radar subsystem to convert the received analog signal into a digital signal. If the frequency of the mixed signal is higher than the Nyquist sampling rate of the ADC, the digital data representing the reflected radar signal output by the ADC may be incorrectly represented (e.g., as a lower frequency representing a closer object).

[0049] When using device 301 to monitor sleep patterns and vital statistics, the user can be indicated to be the person closest to device 301. Another person or animal may be present in the bed. The unambiguous FMCW range may need to be defined far enough, such as two meters, that two people (or approximately the width of the bed) fall within the unambiguous FMCW range of radar subsystem 405. Two meters may be an ideal distance, as it is approximately the width of a commercial king-size bed (e.g., a king-size bed).

[0050] Radar subsystem 405 may include multiple antennas to receive reflected radar radio waves. In some embodiments, three antennas may be present. These antennas may be arranged in an "L" pattern, such that two antennas are horizontally orthogonal and two antennas are vertically orthogonal, wherein one antenna is used for both horizontal and vertical arrangement, as per [reference to...]. Figure 7 Detailed description. By analyzing the phase difference in the reflected radio waves, weighting can be applied to vertically and / or horizontally target the received radar beam. In other embodiments, the antenna can be aligned with different patterns.

[0051] Horizontal beam guidance can be performed for several reasons. First, horizontal aiming can be performed to compensate for the transmitting radar pointing towards objects that cause interference. For example, if a user's headboard is against a wall, the headboard and / or wall can occupy a significant portion of the radar subsystem 405's field of view. Radar reflections from the headboard and / or wall may not be used to determine data about the user; therefore, it may be beneficial to de-emphasize reflections from the wall and / or headboard and emphasize reflections obtained from away from the wall and / or headboard. Thus, by weighting the received radar signal, the receiving beam can be made horizontally away from the wall and headboard. Second, horizontal beam guidance can be performed to target multiple spatial areas, such as... Figure 2 The spatial region 220 allows for the analysis of different parts of the user's chest.

[0052] In system 400A, beam guidance module 430 is used to process the raw radar waterfall data received from radar subsystem 405 before processing is performed by radar processing module 410. Therefore, beam guidance module 430 can serve as a preprocessing module prior to analysis by radar processing module 410 and can be used to highlight areas where one or more users are expected to be present. Beam guidance module 430 can be implemented using hardware, software, or firmware; therefore, beam guidance module 430 can be implemented using one or more processors identical to those used in radar processing module 410.

[0053] The beam guidance module 430 may include a channel weighting engine 431 and a beam guidance system 432. The channel weighting engine 431 can be used to perform a training process to determine a series of weights to be applied to the received radar signals from each antenna before mixing them together. The channel weighting engine 431 can perform the training process when the monitoring area is determined to be empty. During this time, the strength of signals received from large static objects (e.g., walls, headboards) can be analyzed, and weights can be set to make the beam horizontal (and possibly vertical) away from such objects. Therefore, by guiding the received radar beam through the channel weighting engine 431, the amount of reflection in a static environment can be minimized for a specific distance range from the device (e.g., up to one meter). This training can also be performed when a user is present. That is, the received beam of the radar subsystem 405 can be guided to where movement is detected, or specifically, to where the user's vital signs are detected.

[0054] The beam guidance system 432 can use the weights determined by the channel weighting engine 431 to individually weight the received reflected radar signals from each antenna. The received signals from each antenna can be weighted and then combined for processing by the radar processing module 410. Figure 6-8 Further details are provided regarding various embodiments of how the beamguide module 430 is implemented. The beamguide module 430 can be used in conjunction with any other embodiments detailed herein.

[0055] Raw waveform data can be transferred from beamguide module 430 to radar processing module 410. Radar processing module 410 may include one or more processors. One or more dedicated or general-purpose processors may be used to perform the functions of radar processing module 410. Dedicated processors may include processors specifically designed to perform the functions detailed herein. Such dedicated processors may be ASICs or FPGAs, which are general-purpose components physically and electrically configured to perform the functions detailed herein. General-purpose processors may execute dedicated software stored using one or more non-transitory processor-readable media, such as random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD). Radar processing module 410 may include: a moving filter 411; a frequency emphasizer 412; a range-vitals transformation engine 413; a range gating filter 414; a spectrum summation engine 415; and a neural network 416. Each component of radar processing module 410 may be implemented using software or firmware or as dedicated hardware.

[0056] The raw chirped waterfall or waveform data output by radar subsystem 405 and processed by beamguide module 430 can be received by radar processing module 410 and processed first using motion filter 411. In some embodiments, it is important that motion filter 411 is the initial component for performing filtering. That is, the processing performed by radar processing module 410 is non-interchangeable in some embodiments. Typically, vital sign determination and sleep monitoring can occur when the monitored user is sleeping or attempting to sleep in bed. In such an environment, there may be very little movement. Such movement may be attributed to the user moving in bed (e.g., turning over while trying to fall asleep or asleep) and the user's vital signs, including movement caused by breathing and movement caused by the monitored user's heartbeat. In such an environment, most of the radio waves emitted from RF transmitter 406 may be reflected by static objects near the monitored user, such as mattresses, spring mattresses, bed frames, walls, furniture, bedding, etc. Therefore, most of the raw waveform data received from radar subsystem 405 may be unrelated to user movement and the user's vital signs.

[0057] Motion filter 411 may include a waveform buffer that buffers “chirps” or segments of received raw waveform data. For example, sampling may occur at a rate of 10 Hz. In other embodiments, sampling may be slower or faster. In some embodiments, motion filter 411 may buffer twenty seconds of received raw waveform chirps. In other embodiments, buffered raw waveform data for shorter or longer durations is buffered. Filtering may be performed on the buffered raw waveform data to remove raw waveform data indicating stationary objects. That is, for a moving object, such as the chest of a monitored user, the user’s heart rate and respiratory rate will affect the distance and velocity measurements performed by radar subsystem 405 and output to motion filter 411. This movement of the user will cause “jitter” in the raw waveform data received during the buffering period. More specifically, jitter refers to the phase shift caused by radio waves reflected from a moving object. Instead of using reflected FMCW radio waves to determine the speed of a moving object, the phase shift caused by movement in the reflected radio waves can be used to measure vital statistics, including heart rate and respiratory rate, as detailed herein.

[0058] For stationary objects, such as furniture, zero phase shift (i.e., no jitter) will appear in the raw waveform data during the buffer period. The motion filter 411 can subtract this raw waveform data corresponding to the stationary object, so that the raw waveform data indicating motion is passed to the frequency accentuator 412 for further analysis. For the remaining processing of the radar processing module 410, the raw waveform data corresponding to the stationary object can be discarded or otherwise ignored.

[0059] In some embodiments, an infinite impulse response (IIR) filter is incorporated as part of the motion filter 411. Specifically, a single-pole IIR filter can be implemented to filter out raw waveform data that does not indicate motion. Thus, the single-pole IIR filter can be implemented as a high-pass, low-impedance filter that prevents raw waveform data indicating motion below a certain frequency from being transmitted to the frequency accentuator 412. The cutoff frequency can be set based on known limitations of human vital signs. For example, the expected respiratory rate is between 10 and 60 breaths per minute. The filter can exclude motion data indicating frequencies below 10 breaths per minute. In some embodiments, a bandpass filter can be implemented to exclude raw waveform data indicating high-frequency motion that is impossible for human vital signs. For example, for a person at rest or near rest, a heart rate higher than the respiratory rate is unlikely to exceed 150 beats per minute. Raw waveform data indicating higher frequencies can be filtered out by the bandpass filter.

[0060] In some embodiments, the frequency of the raw waveform data passed to the frequency accentuator 412 by the moving filter 411 can be further fine-tuned. For example, during the initial configuration phase, the user can provide information about the monitored user (e.g., himself, his child), such as age data. Table 1 indicates typical respiratory rates for different ages. Similar data may exist for heart rate. The filter can be configured to exclude data from the expected respiratory rate, heart rate range, or both.

[0061] Birth - 6 weeks 30-60 6 months 25-40 3 years old 20-30 6 years old 18-25 10 years old 17-23 adult 12-18 65-80 years old 12-28 >80 years old 10-30

[0062] Table 1

[0063] The vital signs of the monitored user being measured are periodic pulse events: the user's heart rate may vary over time, but the heart can be expected to beat periodically. This beating is not a sine function, but can be understood as a pulse event, more akin to a square wave with a relatively low duty cycle, causing movement of the user's body. Similarly, the user's respiratory rate may vary over time, but breathing is a periodic function performed by the user's body, similar to a sine function, except that exhalation is typically longer than inhalation. Furthermore, a specific window of waveform data is analyzed at any given time. Because of the specific time window in which the waveform data is analyzed, even a perfect sine wave within that window will result in spectral leakage in the frequency domain. The frequency components resulting from this spectral leakage should be de-emphasized.

[0064] Frequency emphasis unit 412 can work in conjunction with distance-vitality transformation engine 413 to determine one (e.g., respiration) or two (e.g., respiration plus heart rate) frequency components of the raw waveform data. Frequency emphasis unit 412 can use frequency windowing, such as a 4DHamming window (other forms of windowing are possible, such as Hann windows), to emphasize important frequency components of the raw waveform data and de-emphasize or remove waveform data attributable to spectral leakage outside the defined frequency window. This frequency windowing may reduce the amplitude of the raw waveform data, which could be due to processing artifacts. Using frequency windowing can help reduce the impact of data-related processing artifacts while preserving data that can be individually determined for heart rate and respiratory rate.

[0065] For fixed bedside FMCW radar-based monitoring devices, they can be located within 1 to 2 meters of one or more users being monitored to detect respiration and heart rate (e.g., using...). Figure 4B The radar emitted by the radar emphasizes that a 2D Hamming window of 10 to 60 bpm (0.16 Hz to 1 Hz) respiratory rate and 30 to 150 bpm (0.5 to 2.5 Hz) heart rate provides a sufficiently good signal for reliable measurement without prior knowledge of the patient's age or medical history.

[0066] Since heart rate and respiratory rate are periodic pulse events, they can be represented by different fundamental frequencies in the frequency domain, but each may have many harmonic components at higher frequencies. One of the main purposes of the frequency emphasis unit 412 is likely to prevent frequency fluctuations of the harmonics of the monitored user's respiratory rate from affecting the frequency measurement of the monitored user's heart rate (or vice versa). Although the frequency emphasis unit 412 can use a 2D Hamming window, it should be understood that other window functions or isolation functions can be used to help isolate the frequency fluctuations of the monitored user's respiratory rate from the frequency fluctuations of the monitored user's heart rate.

[0067] The range-vitality transformation engine 413 analyzes the received motion-filtered waveform data to identify and quantify the magnitude of movement at a specific frequency. More specifically, the range-vitality transformation engine 413 analyzes phase jitter over time to detect relatively small movements caused by user vital signs (e.g., respiratory rate and heart rate) with relatively low frequencies. The analysis by the range-vitality transformation engine 413 can assume that the frequency components of the motion waveform data are sinusoidal. Furthermore, the transformation used by the range-vitality transformation engine 413 can also identify the frequency at which the distance is observed. The frequency, amplitude, and distance can be determined, at least in part, because the radar subsystem 405 uses an FMCW radar system.

[0068] Before applying the transformation by the distance-vitality transformation engine 413, the distance-vitality transformation engine 413 can perform a zero-padding process to add multiple zeros to the raw waveform data of the moving filter. By performing the zero-padding process, the resolution in the frequency domain can be effectively improved, allowing for more accurate low-rate measurements (e.g., low heart rate, low respiratory rate). For example, zero-padding can help numerically improve the resolution to detect a difference of half a breath per minute compared to the resolution of one breath per minute without zero-padding. In some embodiments, three to four times the number of zeros can be added compared to the buffered sample size of the raw waveform data. For example, if 20 seconds of buffered raw waveform data is being analyzed, 60 to 80 seconds of zero-padding can be added to the sample. Specifically, a range of three to four times the zero-padding of the sample has been found to significantly improve resolution without making the transformation process overly complex (and therefore processor-intensive).

[0069] To determine the amount of zero-padding to be performed, equations 1-3 can be used. In equation 1, RPM_resolution is ideally less than 1.

[0070]

[0071] n_FFT_slow_time_min=(nearest_power_of_2)(60*chirp_rate) Equation 2

[0072] In some embodiments, a chirp rate of 30 Hz can be used. Such a frequency provides sufficient margin relative to the Nyquist limit of the upper limits of respiratory and heart rates. Therefore, n_FFT_slow_time_min could be 2048. Given a 20-second window to estimate respiratory statistics, Equation 3 yields a value of 600.

[0073] n_chirps_for_respiration = 20 * chirp_rate = 600 Equation 3

[0074] This value of 600 is less than the required vitals-FFT size, causing the distance-vitality transformation engine 413 to perform 3x to 4x zero-padding. The balance between how much zero-padding to perform can be based on the associated increase in frequency resolution and the computational cost required to perform the FFT. A 3x to 4x zero-padding has been found to provide sufficient resolution for heart rate and respiratory rate while reducing the amount of computation required.

[0075] The distance-vitality transformation engine 413 can perform a series of Fourier transforms (FTs) to determine the frequency components of the received raw waveform data output by the frequency accentuator 412. Specifically, the distance-vitality transformation engine 413 can perform a series of fast Fourier transforms (FFTs) to determine the specific frequencies of the waveform data and the amplitude of the waveform data at those frequencies.

[0076] Waveform data acquired over a time period can be represented using multiple dimensions. The first dimension (e.g., along the y-axis) can be associated with multiple samples of waveform data from a specific chirp, and the second dimension (e.g., along the x-axis) is associated with a specific sample index of waveform data collected across multiple chirs. The third dimension of the data (e.g., along the z-axis) represents the intensity of the waveform data.

[0077] Multiple FFTs can be performed based on the first and second dimensions of the waveform data. An FFT can be performed along each of the first and second dimensions: an FFT can be performed for each chirp, and an FFT can be performed for each specific sample index on multiple chirs occurring during that time period. An FFT performed on waveform data for a specific reflection chirp can indicate one or more frequencies, which, in FMCW radar, indicate the distance to the object reflecting the emitted radio waves. An FFT performed for specific sample indices across multiple chirs can measure the frequency of phase jitter across multiple chirs. Therefore, an FFT in the first dimension can provide the distance to the presence of vital statistics, and an FFT in the second dimension can provide the frequency of vital statistics. The output of an FFT performed in both dimensions indicates: 1) the frequency of vital statistics; 2) the range of the measured vital statistics; and 3) the magnitude of the measured frequencies. In addition to the values ​​due to the presence of vital statistics in the data, there may be noise, for example, filtered using a spectral summation engine 415. This noise may be partly due to the fact that heart rate and respiration are not perfect sine waves.

[0078] To clarify, the transformation performed by the distance-vitality transformation engine 413 differs from the range-Doppler transformation. Instead of analyzing changes in velocity (as in the range-Doppler transformation), it performs analysis on the periodic changes in phase shift over time as part of the distance-vitality transformation. The distance-vitality transformation is adjusted by tracking phase changes (called phase jitter) to identify small movements (e.g., respiration, heart rate) occurring over relatively long time periods. As previously mentioned, zero-filling is performed to allow sufficient resolution to accurately determine heart rate and respiratory rate.

[0079] Distance gating filter 414 is used to monitor a defined range of interest and exclude waveform data due to movement outside the defined range of interest. For the arrangement detailed herein, the defined range of interest can be 0 to 1 meter. In some embodiments, the defined range of interest may vary or may be set by the user (e.g., through training or a setup process) or by a service provider. In some embodiments, the goal of this arrangement may be to monitor the person closest to the device (and exclude or isolate data from any other person further away, such as someone sleeping next to the person being monitored). In other embodiments, if both individuals will be monitored, the data can be separated, as per [the relevant documentation / reference]. Figure 12Detailed description. Therefore, the range-signal transformation engine 413 and the range gating filter 414 are used to separate, exclude, or remove motion data attributed to objects outside the defined range of interest, and to sum the energy of the motion data attributed to objects within the defined range of interest. The output of the range gating filter 414 may include data having a defined range within the allowable range of the range gating filter 414. This data may also have a frequency dimension and an amplitude dimension. Therefore, the data may have three dimensions.

[0080] The spectrum summing engine 415 can receive the output from the distance gating filter 414. The spectrum summing engine 415 can be used to transfer the energy of the harmonic frequencies of the measured heart rate and respiratory rate and add the harmonic frequency energy to the fundamental frequency energy. This function can be called the Harmonic Sum Spectrum (HSS). Heart rate and respiratory rate are not sinusoidal; therefore, in the frequency domain, harmonics will appear at frequencies higher than the fundamental frequencies of the user's respiratory rate and heart rate. One of the main purposes of the spectrum summing engine 415 is to prevent the harmonics of the monitored user's respiratory rate from affecting the frequency measurement of the monitored user's heart rate (or vice versa). The HSS can be performed in second order by adding the original spectrum to a downsampled instance of the spectrum (multiplied by a factor of two). This process can also be applied to higher-order harmonics, such that their respective spectra are added to the spectrum at the fundamental frequency.

[0081] At this stage, for a person lying still in bed (excluding movement due to breathing and heart rate), two main frequency peaks are expected in the frequency data. However, if the monitored user is moving, such as turning over in bed, the energy will be significantly distributed across the entire spectrum (a broader distribution). Such large body movement may appear as numerous small peaks in the frequency data. If the bed is empty and not occupied, there may be no or almost no frequency components above the background noise, as the motion filter 411 has previously filtered the raw waveform data corresponding to a static object. The distribution and amplitude of frequency peaks in the spectrum can be used to determine whether the user is likely awake or asleep.

[0082] The spectrum summation engine 415 can output feature vectors indicating heart rate (e.g., heartbeats per minute) and respiratory rate (e.g., breaths per minute). The feature vectors can indicate both frequency and amplitude. A neural network 416 can be used to determine whether the heart rate and / or respiratory rate indicated in the output of the feature vectors from the spectrum summation engine 415 should be considered valid. Therefore, the heart rate and respiratory rate output by the spectrum summation engine 415 can be stored, presented to the user, and / or the output of the neural network 416 can be considered valid. The neural network 416 can be trained (e.g., using supervised learning performed using a set of training data) to output one of three states by performing spectrum analysis, such as those shown in Table 2. Vital statistics can be considered valid when it is determined that the user is present and the detected movement is due to the user's vital signs.

[0083] Each state in Table 2 is associated with a different distribution of spectral energy and spectral sparsity. Spectral energy refers to the total energy of the entire spectrum detected due to movement within the monitored area. Spectral sparsity indicates whether movement tends to be distributed across a wide frequency range or concentrated at a few specific frequencies. For example, spectral sparsity is high if energy peaks occur at a few frequencies, such as when a user's vital signs are detected (but not other movement). However, spectral sparsity is low if peaks (exceeding a threshold) or some other form of determination based on a threshold criterion at least in part based on amplitude occur at many frequencies.

[0084] As an example, movement caused by vital signs such as a heartbeat can indicate significant movement (e.g., high spectral energy) at a specific frequency (e.g., high spectral sparsity); movement caused by a user moving a limb may also indicate significant movement (high spectral energy), but may have low spectral sparsity. A neural network can be trained to distinguish each state based on the spectral energy distribution output by the spectral summation engine 415. Therefore, two features can be provided to the neural network 416, representing a first value of spectral energy and a second value representing spectral sparsity.

[0085] The output of the spectrum summing engine 415 can be characterized as an eigenvector with a first dimension of frequency and a second dimension of amplitude. A first value representing the spectral energy can be calculated by determining the maximum amplitude present in the eigenvector output by the spectrum summing engine 415. This maximum amplitude value can be normalized to a value between 0 and 1. A second value representing the spectral sparsity can be calculated by subtracting the median amplitude of the eigenvector from the maximum amplitude. Again, here, the calculated sparsity can be normalized to a value between 0 and 1.

[0086] Table 2 illustrates how spectral energy and spectral sparsity features are used as features of a trained neural network to perform classification generalizations of the state of the monitored area.

[0087] User presence and movement based solely on vital signs high high User presence and movement (limb and torso movement) high Low No users exist. Low Low

[0088] Table 2

[0089] The state of the monitoring area classified by the neural network 416 can be used to determine the sleep state of the monitored user, or more generally, to determine whether the user is moving or still in bed. The state of the monitoring area determined by the classification performed by the neural network 416 can be further used to determine whether the vital statistics output by the spectrum summation engine 415 should be trusted or ignored. For accurate vital statistics determination, heart rate and respiratory rate can be identified as potentially accurate when the neural network 416 determines that the user is present and stationary (i.e., without large body movements; however, movement occurs due to breathing and / or heartbeat). In some embodiments, the vital statistics output by the spectrum summation engine 415 can be stored locally (e.g., to mitigate privacy concerns); in other embodiments, the vital statistics output can be transmitted to a cloud-based server system 370 for remote storage (alternative to or supplementing this local data storage).

[0090] The neural network 416 can initially be trained using a large amount of training data consisting of amplitude and frequency feature vectors, which has been appropriately labeled as classifications mapping spectral energy and spectral sparsity to corresponding real states of the monitored region. Alternatively, the neural network 416 can initially be trained using a large amount of training data consisting of amplitude and frequency feature vectors, which has been appropriately classified as mapping pairs of spectral energy and spectral sparsity, each pair being appropriately labeled as a corresponding real state of the monitored region. The neural network can be a time-independent, fully connected neural network. In some embodiments, machine learning arrangements, classifiers, or artificial intelligence forms other than neural networks can be used.

[0091] In other embodiments, instead of using spectral energy and spectral sparsity values ​​as features for the neural network, a neural network, possibly with additional front-end convolutional layers, can be trained to directly use the output of the range gating filter 414. Instead, one embodiment of the convolutional network can analyze the frequency and amplitude data of the range gating filter 414 output to perform classification of the user's state. The convolutional neural network can be trained to utilize offline training based on a set of spectral measurements mapped to the true state of the monitored area before the end user uses the system 400A.

[0092] Sleep states determined by neural network 416 can be stored along with time data in data storage 318. Vital statistics output by spectrum summation engine 415 can be stored in a vital statistics database when neural network 416 indicates the presence and stillness of the monitored user. Other vital statistics may be discarded or flagged to indicate they are unlikely to be accurate. Data stored in data storage 318 and the vital statistics database can be stored locally at device 301. In some embodiments, storage occurs only at device 301. Such an implementation can help alleviate concerns about the transmission and remote storage of health-related data. In some embodiments, the monitored user can choose to have sleep data and vital statistics transmitted, stored, and analyzed externally, such as by a cloud-based server system 370 via a network interface (e.g., wireless network interface 350). Storage by cloud-based server system 370 can have significant benefits, such as allowing users to remotely access such data, enabling access to healthcare providers, or participating in research. Users can retain the ability to delete or otherwise remove data from cloud-based server system 370 at any time.

[0093] In some embodiments, the radar processing module 410 may be located entirely or partially remotely from device 301. While the radar subsystem 405 may need to be local to the monitored user, processing by the radar processing module 410 can be moved to a cloud-based server system 370. In other embodiments, a smart home device that performs local communication with device 301 (e.g., via LAN or WLAN) may perform some or all of the processing by the radar processing module 410. In some embodiments, local communication protocols, such as those involving mesh networking, may be used to transmit raw waveform data to the local device that will perform the processing. Such communication protocols may include Wi-Fi, Bluetooth, Thread, or the IEEE 802.11 and 802.15.4 family of communication protocols. Similar to processing, the storage of sleep data and vital statistics may occur at the cloud-based server system 370 or at another smart home device in the home where device 301 is located. In other embodiments, the radar processing module 410 may be integrated with the radar subsystem 405 as a single component or component system.

[0094] The sleep data and vital statistics stored in the sleep data storage device 318 can be used to provide users with short-term and long-term trends related to their sleep patterns, vital statistics, or both, through the sleep data compilation engine 319. For example, each morning, charts, statistics, and trends can be determined by the sleep data compilation engine 319 based on the data stored in the sleep data storage device 318 and output for display via a display 340. Charts from the previous night's sleep data, as well as one or more charts indicating respiratory rate and heart rate during the previous night, can be presented. The sleep data compilation engine 319 can output similar charts, trends, and statistics for significantly longer time periods, such as weeks, months, years, and even years. Other uses of the sleep data and vital statistics are possible. For example, if certain triggers regarding heart rate, respiratory rate, and / or sleep patterns are triggered, medical professionals can be notified. Additionally or alternatively, notifications can be output to users indicating that the collected data may relate to or indicate healthy individuals. In some cases, specific sleep problems, such as sleep apnea, may be identified. Sleep data can be output via speaker 355 using synthesized speech (e.g., in response to the user waking up, in response to a dictated user command, or in response to input from the user via a touchscreen such as display 340). Such sleep data can also be represented on display 340 in graphical and / or textual form.

[0095] System 400A may additionally include a beam guidance module 430. Beam guidance module 430 may include a channel weighting engine 431, which may be implemented using software, firmware, and / or hardware, similar to components of radar processing module 410. Beam guidance module 430 is shown separately from radar processing module 410 because it processes data received from radar subsystem 405 to emphasize data received from a particular direction and de-emphasize data received from other directions. Beam guidance module 430 may be implemented using the same hardware as radar processing module 410. For example, beam guidance module 430 may be a software process that modifies radar data received from radar subsystem 405 before applying motion filter 411. Device 301 may be a surface-top device designed to be placed in a specific location, connected to a continuous power source (e.g., a household power outlet), and interacted with via voice and / or touchscreen. Therefore, radar subsystem 405 may remain pointed towards a portion of its surroundings for a considerable period of time (e.g., hours, days, weeks, months). Generally, the beam guidance module 430 can be used to map the environment (e.g., a room) where the device 301 is located and guide the sensing direction of the radar subsystem 405 to the area within the field of view of the radar subsystem 405 where the user is most likely to be present.

[0096] Targeting an area within the field of view of radar subsystem 405 can help reduce the number of missed and false alarms caused by the movement of objects other than the user. Furthermore, targeting can help compensate for the angle and position of device 301 relative to where the user is sleeping. (For example, device 301 may be located on a bedside table at a different height than the user's bed. Additionally or alternatively, radar subsystem 405 device 301 may not be directly pointed at the user's sleeping position on the bed.)

[0097] When, for example, it is determined that no user is present based on the low spectral energy and low spectral density in Table 2, the optimal beam guidance process can be performed by the channel weighting engine 431 and the beam guidance system 432. Even without users present, analysis can be performed to determine which directional alignment of the radar subsystem 405 provides minimal clutter.

[0098] Figure 4B An embodiment of a chirp timing diagram 400B for frequency-modulated continuous wave (FMCW) radar radio waves output by a radar subsystem is illustrated. Chirp timing diagram 400B is not drawn to scale. Radar subsystem 405 can typically output radar in the mode of chirp timing diagram 400B. Chirp 450 represents a continuous pulse of radio waves whose frequency scans from low frequency to high frequency. In other embodiments, a single chirp can continuously scan from high frequency to low frequency, from low frequency to high frequency, and then back to low frequency, or from high frequency to low frequency and then back to high frequency. In some embodiments, the low frequency is 58 GHz and the high frequency is 63.5 GHz. (For such frequencies, the radio waves may be referred to as millimeter waves.) In some embodiments, the frequency is between 57 and 64 GHz. The low and high frequencies can vary depending on the embodiment. For example, the low and high frequencies can be between 45 GHz and 80 GHz. Frequency selection can be at least partially chosen to comply with government regulations. In some embodiments, each chirp comprises a linear scan from low frequency to high frequency (or vice versa). In other embodiments, an exponential or some other pattern may be used to scan frequencies from low to high or from high to low.

[0099] Chirp 450, which can represent all chirps in chirp timing diagram 400B, can have a chirp duration 452 of 128 microseconds. In other embodiments, the chirp duration 452 can be longer or shorter, for example, between 30 μs and 600 μs. In some embodiments, a period of time may elapse before a subsequent chirp is emitted. The pause between chirps 456 can be 205.33 microseconds. In other embodiments, the pause between chirps 456 can be longer or shorter, for example, between 10 μs and 1 ms. In the illustrated embodiment, the chirp period 454, including chirp 450 and the pause between chirps 456, can be 333.33 μs or 500 μs. This duration varies depending on the selected chirp duration 452 and the pause between chirps 456.

[0100] The output, consisting of multiple chirps separated by pauses between chirps, can be referred to as a burst 458 or frame 458. Frame 458 may include twenty chirps. In other embodiments, the number of chirps in frame 458 may be more or less, for example, between 1 and 100. The number of chirps present in frame 458 can be determined based on the average power to be radiated. The FCC or other regulatory agencies may set a maximum amount of power that is allowed to be radiated into the environment on average. For example, there may be a duty cycle requirement that limits the duty cycle of any 33-millisecond time period to less than 10%. In a particular example with 20 chirps per frame, each chirp may have a duration of 128 μs, and the duration of each frame is 33.33 ms. The corresponding duty cycle is (20 frames) * (.128 ms) / (33.33 ms), which is approximately 7.8%. By limiting the number of chirps within frame 458 before the pauses between frames, the total amount of power output can be limited. In some embodiments, the peak EIRP (Effective Isotropic Radiated Power) may be 13 dBm (20 mW) or less, for example, 12.86 dBm (19.05 mW). In other embodiments, the peak EIRP is 15 dBm or less, and the duty cycle is 15% or less. In some embodiments, the peak EIRP is 20 dBm or less. That is, at any given time, the average power radiated by the radar subsystem may never exceed these values. Furthermore, the total radiated power over a time period may be limited. In some embodiments, a duty cycle may not be required.

[0101] Frames can be transmitted at a frequency of 30 Hz (33.33 ms), as shown in time period 460. In other embodiments, the frequency can be higher or lower. The frame frequency may depend on the number of chirps within a frame and the duration of the inter-frame pause 462. For example, the frequency may be between 1 Hz and 50 Hz. In some embodiments, chirps can be transmitted continuously, causing the radar subsystem to output a continuous stream of chirps interspersed with inter-chirp pauses. Due to the transmission of chirps and the processing of received chirp reflections, a trade-off can be made to save the average power consumed by the device. Inter-frame pause 462 represents a period of time during which no chirp is output. In some embodiments, inter-frame pause 462 is significantly longer than the duration of frame 458. For example, the duration of frame 458 may be 6.66 ms (the chirp period 454 is 333.33 μs and there are 20 chirs per frame). If there is a 33.33 ms interval between frames, then inter-frame pause 462 may be 26.66 ms. In other embodiments, the duration of the inter-frame pause 462 can be longer or shorter, for example, between 15 ms and 40 ms.

[0102] exist Figure 4BIn the illustrated embodiment, a single frame 458 and the beginning of subsequent frames are shown. It should be understood that each subsequent frame can be structurally similar to frame 458. Furthermore, the transmission mode of the radar subsystem can be fixed. That is, chirps can be transmitted according to the chirp timing diagram 400B regardless of the presence of a user, the time of day, or other factors. Therefore, in some embodiments, the radar subsystem always operates in a single transmission mode, regardless of environmental conditions or the activity being monitored. A continuous string of frames similar to frame 458 can be transmitted when device 301 is powered on.

[0103] Figure 5A An embodiment of a phase acquisition module 500A that can be used as part of a non-contact health monitoring system is shown. The phase acquisition module 500A can be used as part of a non-contact health monitoring system 500B. The phase acquisition module 500A can represent a modified embodiment of the radar processing module 410. The phase acquisition module 500A may include a moving filter 411, a frequency emphasis unit 412, and a phase analysis engine 502. Each of these components can be implemented using a software process executed by a general-purpose or dedicated processor (or multiple processors). Dedicated hardware can be implemented to perform the functions of the phase acquisition module 501. The moving filter 411 and the frequency emphasis unit 412 can be implemented as described above. Figure 4A It functions as detailed in the radar processing module 410. For the phase acquisition module 500A, the frequency emphasis unit 412 can be configured to emphasize the frequency components attributable to respiration (and, possibly, deemphasize the frequencies related to heart rate).

[0104] Phase analysis engine 502 is typically used to provide an instantaneous indication of the phase present in the waveform data output by the beam guidance module for a specific reflective radar chirp. For the waveform data received by phase analysis engine 502, a Fast Fourier Transform (FFT) can be performed to determine the phase value (reference absolute phase value) and amplitude of each frequency component of the reflected chirp. For each reflected chirp in the waveform data, phase analysis engine 502 outputs data indicating the frequency, the range at which the frequency was detected, the phase of the frequency component, and the amplitude of the frequency component.

[0105] Radar subsystem 405 may include an analog-to-digital converter (ADC) to convert received reflected waveform data into digital waveform data. (FMCW may also include mixing performed prior to the ADC conversion, in which the received reflected waveform is mixed with the waveform being transmitted.) Since all ADCs have limited resolution, changes in distance to a moving target object (e.g., a user's chest) are sufficient to cause a phase change, potentially requiring a phase difference to be observable in the digital data output by the ADC. Equation 2 illustrates the effect of distance to the moving object on the phase.

[0106]

[0107] In Equation 4, for FMCW, f0 represents the lowest frequency of the chirp; f1 represents the highest frequency of the chirp; c represents the speed of light; and R represents the range (distance) from the radar sensor to the moving object. chirp This indicates the duration (in seconds) of a single emitted chirp. (See also: Regarding...) Figure 4B As described in detail, a relatively small change in R over the frequency range used for chirping (e.g., due to movement of the user's chest during breathing) results in a phase change that is large enough to be detected as part of the digital waveform data in the digital data output by the ADC. Therefore, Equation 4 confirms that the phase can be used to accurately measure chest displacement caused by the user's breathing.

[0108] Figure 5B An embodiment of a non-contact health monitoring system 500B (“System 500B”) for monitoring multiple spatial areas is shown. System 500B can utilize the output of the same radar subsystem 405 as System 400A. Component 516 can correspond to... Figure 3 The respiratory monitoring engine 316. In some embodiments, before activating and executing component 516 using one or more processors, it is determined that the user has been in bed and still for at least a period of time or some other time-based criterion. Component 516 can be executed using one or more general-purpose processors that execute dedicated software. In other embodiments, one or more dedicated processors can be configured to execute the functions of system 500B.

[0109] Each monitored spatial region can correspond to a different target part of the user's chest, for example Figure 2 The spatial region 220 in the system 500B may include: radar subsystem 405; beam guidance module 430; phase acquisition module 501; radar phase processing module 510; feature bundle compiler 520; and machine learning model 540.

[0110] Radar subsystem 405, etc. Figure 4A As shown, it functions. Radar subsystem 405 includes multiple antennas for receiving the chirp of the reflected radar. Individual waveform data is output for each receiving antenna of radar subsystem 405. Figure 4AIn this system 500, a single-beam guidance module 430 is present. By having a single-beam guidance module, a single direction can be targeted. The beam guidance module 430 can determine a general spatial region including the user's chest. Once the general spatial region is determined, the region can be further refined by beam guidance towards multiple spatial regions within the general spatial region. Conversely, the system 500 contains multiple receiver-side digital beam guidance modules 530. Each of these beam guidance modules can function as detailed regarding beam guidance module 430. Each of the beam guidance modules 530 can use different weights to beam guide towards different parts of the user's chest. For example, refer to... Figure 2 Beam guidance module 530-1 can use weights to focus on spatial region 220-1, beam guidance module 530-2 can use different weights to focus on spatial region 220-2, and so on. According to embodiments, there are different numbers of beam guidance modules 530, such as 2, 3, 4, 5 or more target modules.

[0111] Each beamguide module 530 receives the same waveform data from the radar subsystem 405. Therefore, for a specific received reflection chirp, each beamguide module 530 targets a different spatial region. Because each beamguide module 530 operates based on the same waveform data from the radar subsystem 405, the beamguided waveform data output by the beamguide module 520 corresponds to the same point in time. Figure 6 and Figure 8 Various embodiments of how the beam guidance module 530 can be implemented are shown.

[0112] Beamguide module 430 may have been used to determine the approximate orientation of the user's chest. This orientation can serve as a baseline for determining the orientation to various spatial regions corresponding to different parts of the user's chest. After beamguide module 430 determines the orientation that produces strong measurements of the user's vital signs, this orientation can be incremented horizontally by a fixed or variable angle to produce each spatial region to be monitored. Weights can be determined by each beamguide module 530 based on the channel weights determined by channel weighting engine 431 and the desired angle that a particular spatial region should form relative to the center of the user's chest. The larger the number of beamguide modules 530, the smaller the angles between spatial regions may be.

[0113] The waveform data output from each beamguide module 530 has been weighted to target specific spatial regions corresponding to different overlapping or non-overlapping areas of the user's chest. The weighted waveform data output from the beamguide module 530 is then processed using a separate phase acquisition module 501. Each of the phase acquisition modules 501 is as follows: Figure 5AIt functions as detailed in the phase acquisition module 500A. Since each phase acquisition module in phase acquisition module 501 is directed to a different spatial region of the user's chest, the phase, frequency, and amplitude values ​​output by each phase acquisition module 501 correspond to a specific region of the user's chest. For each received reflected chirp, each phase acquisition module 501 outputs an indication of phase, amplitude, and frequency to the corresponding radar phase processing module in radar phase processing module 510.

[0114] Each radar phase processing module 510 may include multiple components. As an example, the components of radar phase processing module 510-1 are indicated as: phase selector 511-1 and phase unwrapper 512-1. Each of the other radar phase processing modules 510 may have similar components, but they are not shown for simplicity.

[0115] Each of the phase selectors 511, such as phase selector 511-1, can be used to select the phase component expected to correspond to the user's breathing in the monitored spatial region. Phase selector 511-1 determines the frequency signal with the maximum amplitude based on spectral data received from phase acquisition module 501-1. Since the user has been determined to be stationary (e.g., by neural network 416), there is no other movement in the environment, and filtering has been performed to emphasize frequencies corresponding to breathing within a specific distance range, it can be expected that the frequency component with the maximum amplitude is attributable to the user's breathing. The phase of this frequency component with the maximum amplitude is determined by phase selector 511-1, and other data can be discarded. Therefore, the output of phase selector 511-1 can be a phase value between -π and π (or 0 and 2π). When processing waveform data corresponding to multiple chirps, it can be expected that the phase value output by phase selector 511-1 will increase or decrease.

[0116] The phase selector of the other radar processing module 510 functions similarly. It can be expected that the phase value output by the phase selector of the other radar processing module 510 corresponds to the phase of a specific spatial region of the monitored user's chest. The greater the movement of the user's chest, the greater the expected change in phase over time.

[0117] Each radar phase processing module in radar phase processing module 510 may include a phase unwrapper, such as phase unwrapper 512-2. Phase entanglement may occur because the user's chest may rise and fall beyond a single wavelength. Phase unwrapper 512 can be used to eliminate ambiguity in phase changes exceeding a single wavelength. Various phase unwrapping algorithms can be used on the instantaneous phase value of each chirp output from phase selector 511-1. For example, phase unwrapper 512 can use Itoh's sample-by-sample phase unwrapping algorithm to obtain phases that can exceed a single wavelength range represented by -π and π (or 0 and 2π). Therefore, the output of phase unwrapper 512-1 is a phase value that can be greater than a single wavelength and can account for movement of the user's chest over a distance greater than a single wavelength. The phase value output by phase unwrapper 512-1 can be fed to feature bundle compiler 520.

[0118] The feature bundle compiler 520 receives unwrapped phase values ​​for each monitored spatial region. The number of monitored spatial regions corresponds to the number of dimensions of the phase data received by the feature bundle compiler 520. For each dimension, the unwrapped phase values ​​are buffered for a period of time to create a signal indicating the phase changes over time (e.g., five seconds, ten seconds, etc.). Thus, in the illustrated example, phase values ​​for four regions are received for the feature bundle compiler 520, but in other embodiments, the feature bundle compiler 520 may receive fewer or more dimensions.

[0119] The feature bundle compiler 520 can determine the relative phase shift between different spatial regions. For each dimension of the received phase data, the feature bundle compiler 520 can perform an FFT. The output of each FFT can be frequency, amplitude, and phase. For each dimension, the feature bundle compiler 520 creates a phasor indicating the amplitude and phase. The phasors created by the feature bundle compiler 520 for each dimension are fed into the machine learning model 540 to serve as features for performing classification by the machine learning model 540.

[0120] Machine learning model 540 can receive phasors for each dimension created by feature bundle compiler 520. Machine learning model 540 can be a neural network, such as a feedforward neural network classifier, which can have three layers. Machine learning model 540 can receive bundles of phasors, where each phasor indicates both amplitude and phase. At a high level, machine learning model 540 can identify cases where the bundle includes phasors where each phasor has a significant amplitude but significantly different phases. For example, two vectors (each corresponding to a different spatial region of the user's chest) can have significant amplitudes and opposite or nearly opposite phases. This situation might indicate anomalous breathing.

[0121] To train the machine learning model 540, a set of phasor bundles labeled with ground truth can be created, wherein the bundles are created for users with known respiratory conditions (such as paradoxical breathing) and people without abnormal respiratory conditions. Based on such training data, neural networks or other forms of machine learning models can be trained to perform classification based on such ground truth labeled training data. In some embodiments, more than one type of machine learning module exists to allow the monitoring of multiple types of respiratory abnormalities.

[0122] The classifications output by the machine learning model 540 can be provided to the respiratory data output engine 319, stored using data storage 318, and / or stored by a cloud-based server system 370. If the machine learning model 540 outputs a classification of abnormal breathing conditions, such data can be stored. When the user is determined to be awake, for example in the morning, information instructing a medical professional to check the user for abnormal breathing conditions can be presented. In some embodiments, the respiratory data output engine 319 indicates the presence of a specific type of abnormal breathing condition, such as paradoxical breathing. In some embodiments, the respiratory data output engine 319 uses synthesized speech to indicate the detected abnormal breathing condition. Other data can also be provided to the user, such as indications of the duration of the detected abnormal breathing condition, the severity of the abnormal breathing condition, the number of days since the abnormal breathing condition was detected, etc.

[0123] Figure 6 An embodiment 600 of a beamguide module 610 for aiming at a direction performing sleep tracking and / or respiratory monitoring is illustrated. The beamguide module 610 may represent an embodiment of beamguide module 430 or beamguide module 530. Typically, the beamguide module 610 can apply weights to each radar data stream received from radar subsystem 405, sum the weighted inputs, and output the combined weighted radar data stream to radar processing module 410. The applied weights can introduce a delay into the input of a particular antenna, which can be achieved by the weights being complex values. By introducing a delay into one or more radar data streams received from the antenna, the antenna reception beam can be effectively guided.

[0124] In embodiment 600, three digital radar data streams 620 (620-1, 620-2, 620-3) are received from radar subsystem 205, each corresponding to a separate antenna. Therefore, in this embodiment, three antennas are part of radar subsystem 205. In other embodiments, radar subsystem 205 may have fewer (e.g., two) or more (e.g., four, five, six, seven, or more) antennas, each with a corresponding raw radar data stream output digitally to beamguide module 610.

[0125] Mixer 630 and combiner 640 can represent beam guidance system 432. Each of the radar data streams 620 can be input into a separate mixer in mixer 630. Mixer 630 can be implemented digitally and therefore can represent a software process. Mixer 630-1 mixes radar data stream 620-1 with weights represented by complex values ​​output by channel weighting engine 431. Mixer 630-2 mixes radar data stream 620-2 with weights output by channel weighting engine 431 (which may be the same as or different from the weights applied at mixer 630-1). Mixer 630-3 mixes radar data stream 620-3 with weights output by channel weighting engine 431 (which may be the same as or different from each weight applied at mixers 630-1 and 630-2).

[0126] The channel weighting engine 431, which can represent a software process, can perform a training process to determine values ​​(e.g., complex values) representing the weights that should be output to each mixer 630. In other embodiments, the channel weighting engine 431 can be executed by separate dedicated hardware or hardware incorporated as part of the radar subsystem 405. The digital signal representing the weights output by the channel weighting engine 431 can effectively impose a larger or smaller delay on each radar data stream 620. The weights applied via the mixer 630 can be normalized to 1. Therefore, the sum of the three weights applied in embodiment 600 can be 1.

[0127] Beamguide system 432 and beamguide module 610 can be used to implement weighted delay and summation (WDAS) beamguide via mixer 630. Equation 5 details how WDAS is implemented:

[0128]

[0129] In equation 5, w i The channel weight is represented, which can be a complex value that incorporates phase delay; x i This indicates incoming digital radar data (e.g., FMCW radar chirp) from radar subsystem 405; a i This represents the complex numerical weights of the different received antenna signals with varying amplitudes responsible for the phase delay. The weighting of the channel weighting engine 431 output can be determined by performing a least-squares optimization process. The least-squares optimization process can be performed according to Equation 6.

[0130] mininiae||y-Xw|| 2 Equation 6

[0131] In Equation 6, y represents the vectorized data generated using the target beam. X represents the radar data stream received from radar subsystem 405; w represents the weights to be learned by channel weighting engine 431. As part of the training process to determine the most effective weights targeting the user, various weights can be tested (e.g., in a single pattern, randomly) to attempt to minimize the output of Equation 6. For example, if enough randomized weights are tested, a minimized output value can be expected within a certain amount of error. By minimizing the output value according to the least-squares optimization process, weights corresponding to the beam direction that most closely targets the user's position within the bed can be obtained. These weights can then be used for future monitoring of the user. Retraining can be performed periodically or occasionally to compensate for changes in the user's movement within the bed and / or changes in the orientation and / or position of the sleep detection device.

[0132] Before use, weights can be determined offline to compensate for the known tilt of the radar subsystem, such as... Figure 9 As shown, and indicated by directions 950 and 952. When the user is present, the optimal direction is determined for the user, for example, by sweeping or randomly selecting weights. When the user is not present, one or more directions to stationary objects that produce significant reflections can be determined, so that one or more of these directions can be avoided when aiming at the user.

[0133] It should be understood that the learning process, apart from the least-squares optimization process, can be performed by the channel-weighted engine 431. For example, in some embodiments, the user can assist the training process by providing input indicating the direction from the contactless sleep tracking device to the user's sleep location. In other embodiments, different forms of automated learning processes can be performed to align the beam with the user.

[0134] The channel weighting engine 431 can be triggered to determine the weights on the system 400A that is starting up or turned on. If the system 400A detects movement, for example, via an onboard accelerometer, the channel weights can be recalculated.

[0135] Combiner 640 can receive the sum of weighted radar data streams 635 (e.g., 635-1, 635-2, and 635-3) output by mixer 630. Combiner 640 can output a single summed output 645 to radar processing module 410. By applying at least one weight different from other weights applied by mixer 630 (which causes a delay), the beam can be effectively guided in directions that may have vertical and / or horizontal components. The processing of radar processing module 410 can be as follows: Figure 4A Perform as described in the detailed description.

[0136] Figure 7An embodiment of a possible antenna layout for radar subsystem 700 is shown. Radar subsystem 700 can represent an embodiment of an integrated circuit used as radar subsystem 405. The entire IC can have dimensions of 6.5 mm (length 705) by 5 mm (width 704). In other embodiments, the length 705 by the width 704 of the entire IC is between 7 mm by 7 mm and 4 mm by 4 mm. The illustrated embodiment of radar subsystem 405 has three receiving antennas and one transmitting antenna, but other embodiments can have more or fewer antennas. Radar subsystem 700 can have receiving antennas 710-1, 710-2, and 710-3 distributed in an "L" pattern. That is, as Figure 7 As shown, antennas 710-1 and 710-2 can be aligned on axis 701, and antennas 710-2 and 710-3 can be aligned on axis 702, which is perpendicular to axis 701. The center of antenna 710-2 can be located 2.5 mm or less from the center of antenna 710-1. The center of antenna 710-2 can be located 2.5 mm or less from the center of antenna 710-3.

[0137] The transmitting antenna 710-4 can be arranged separately from the L-shaped pattern of the receiving antennas 710-1, 710-2, and 710-3. That is, in some embodiments, the center of the transmitting antenna 710-4 is not located on the axis of the antenna 710-3, which is parallel to axis 701. In some embodiments, the transmitting antenna 710-4 is located on axis 703, which is parallel to axis 702, at the center of antenna 710-1.

[0138] Each antenna 710 may be a hollow rectangular dielectric resonant antenna (DRA). Each antenna 710 may have the same set of dimensions. Alternatively, each of the receiving antennas 710-1, 710-2, and 710-3 may have the same dimensions, and the transmitting antenna 710-4 may have different dimensions than the receiving antennas. In some embodiments, the transmitting antenna 710-4 has a wider width, for example, 0.2 mm, than the receiving antennas 710-1, 710-2, and 710-3, but the same length.

[0139] In this arrangement, the phase delay introduced by the weighting applied between the radar data streams of antenna 710-1 and antenna 710-2 may affect the vertical direction of the received beam, while the phase delay introduced by the weighting between the radar data streams of antenna 710-2 and antenna 710-3 may affect the horizontal direction of the received beam (assuming that the radar subsystem integrated circuits exist in approximately the same direction within the non-contact sleep tracking device).

[0140] In some embodiments, separate antennas are used for both transmission and reception. For example, antenna 710-4 may be dedicated to transmission, while antennas 710-1, 710-2, and 710-3 may be dedicated to reception.

[0141] Using a radar subsystem (where all antennas reside on a single, relatively compact integrated circuit chip, as described) has proven to achieve a good balance between cost savings, reasonable capability to perform receiver-side beam steering, and a sufficiently wide antenna pattern in the horizontal plane capable of covering common bed sizes (e.g., queen, king, full, twin). Simultaneously, devices incorporating such a radar subsystem allow for placement close enough to the bed (e.g., within 1 meter) that they can also function as a personal assistant, including alarm clock functionality (which can replace an alarm clock), a home control center, and / or an entertainment touchscreen device.

[0142] While the beam guidance module of embodiment 600 does not take into account the arrangement of antennas 710 relative to each other, embodiment 800 adapts to the topology of the antenna arrangement of radar subsystem 700. In other embodiments, antennas 710 may be arranged in a pattern different from "L".

[0143] Figure 8 An embodiment 800 of a beamguide module for aiming to perform sleep tracking and / or health monitoring is shown. In embodiment 800, the antenna arrangement (i.e., antenna topology) of radar subsystem 405 is considered. By considering the antenna topology, more precise beamguide can be performed, which can result in more accurate tracking of a user sleeping in a user's bed. Antenna 710-1 corresponds to radar data stream 620-1, antenna 710-2 corresponds to radar signal stream 620-2, and antenna 710-3 corresponds to radar information stream 620-3. That is, the phase delay added between the data streams of antennas 710-2 and 710-3 in radar subsystem 405 is used for horizontal beamguide, while the phase delay added between the data streams of antennas 710-2 and 710-1 is used for vertical beamguide. Depending on whether the data stream of antenna 710-2 is used for vertical or horizontal beamguide, a separate digitally implemented mixer can be used to apply different weights.

[0144] As in embodiment 600, and in embodiment 800, individual digital radar data streams 620 are received from each antenna of radar subsystem 405. Mixer 830 and combiner 840 may represent... Figure 4AThe beam guidance system 432. In embodiment 800, the beam guidance module 810 has four mixers 830 (830-1, 830-2, 830-3, and 830-4). Similar to embodiment 600, the values ​​(e.g., complex values) output by the channel weighting engine 431 can be mixed with each radar data stream of radar data stream 620. However, different weights can be mixed with radar data stream 620-2, where two weighted outputs are created for horizontal and vertical beam targeting, respectively. Radar data stream 620-1 can have weights applied via mixer 830-1 and can be combined with radar data stream 620-2 (which has weights applied via mixer 830-2) via combiner 840-1. The weights applied at mixers 830-1 and 830-2 can be summed to a normalized value of 1. Radar data stream 620-3 may have weights applied via mixer 830-4, and may be combined with radar data stream 620-2 having weights applied via mixer 830-3 via combiner 840-2. The weights applied at mixers 830-3 and 830-4 may be added together to a normalized value of 1.

[0145] The channel weighting engine 431 can be implemented similarly to that implemented in embodiment 600. The channel weighting engine 431 can perform a least-squares optimization process or some other optimization process to determine the optimal or near-optimal orientation of the received beam. In embodiment 800, the channel weighting engine 431 can generate four outputs for weighting instead of the three outputs in embodiment 600. Therefore, if the pattern or random set of values ​​for the weighted outputs is used as part of the least-squares optimization process, a larger number of output value sets can be tested in embodiment 800 compared to embodiment 600 to obtain an optimized set of output value sets for setting the weights.

[0146] Two outputs 845, output 845-1 and output 845-2, can be output to radar processing module 850. For sleep analysis, output 845-1 and output 845-2 can then be processed separately by radar processing module 850. At a high level, processing by radar processing module 850 can be performed on each output 845 until direction is no longer used during processing. In embodiment 800, separate instances of the moving filter, frequency emphasis unit, and range-signal conversion engine can be applied to each of the outputs 845, and the results can then be averaged or summed. More specifically, output 845-1 can be output to moving filter 851-1, followed by frequency emphasis unit 852-1, and then range-signal conversion engine 853-1. Output 845-2 can be output to moving filter 851-2, followed by frequency emphasis unit 852-2, and then range-signal conversion engine 853-2. The moving filter 851, frequency emphasis unit 852, and range-vitality transformation engine 853 can be as follows: Figure 4A The moving filter 411, frequency enhancer 412, and range-to-vitality transformation engine 413 function as detailed. The outputs of range-to-vitality transformation engines 853-1 and 853-2 can be summed or combined using combiner 854. The output of combiner 854 (which may represent the average value of the outputs of range-to-vitality transformation engines 853) can be processed by range gating filter 414 and subsequent components, as described in [the relevant section]. Figure 4A Detailed description.

[0147] In the first set of embodiments, beamguidance in the vertical direction is not performed in order to analyze a user's breathing using Embodiment 800. In such embodiments, beamguidance can be used solely by employing raw waveform data from horizontally distributed antennas. For example, refer to... Figure 7 Only the raw waveform data obtained from antennas 710-2 and 710-3 can be used for horizontal beamguide. Then, the output 845-2 from beamguide module 810 can be input to the corresponding phase acquisition module in phase acquisition module 501. In such an embodiment, components related to vertical beamforming are not implemented or can be used for beam aiming for sleep tracking purposes.

[0148] In the second set of embodiments, to analyze a user's breathing using beamguided operation performed by embodiment 800, both outputs 845-1 and 845-2 of beamguided module 810 can be used. In such an embodiment implemented as part of system 500B, each of these outputs can be processed using a separate phase acquisition module of phase acquisition module 501 and a separate radar phase processing module of radar phase processing module 510, thereby creating a pair of unwrapped phase values. The two phase values ​​of each pair can then be averaged together and passed to feature beam compiler 520. Therefore, to perform both vertical and horizontal beamforming of embodiment 800, it may be necessary to implement twice the number of phase acquisition modules and radar phase processing modules (in addition to performing averaging) to monitor the same number of spatial regions as when only horizontal beamguided operation is performed.

[0149] For sleep monitoring, since the radar processing module 850 and beam steering module 810 can be executed as a software process by a general-purpose processor (or multiple processors), implementing a more complex mix, weighting, and multiple instances of the moving filter 851, frequency emphasis unit 852, and range-vitality transformation engine 853 may require sufficient processing power to be available. Therefore, assuming such processing power is available, it may not be necessary to make hardware changes to the non-contact sleep tracking device to implement embodiment 800 instead of embodiment 600. In some embodiments, embodiment 600 can be implemented, and embodiment 800 (or vice versa) can be implemented if the sleep tracking results are inaccurate. Advantageously, in some embodiments of the non-contact health monitoring device 300 that include a smart home management device (e.g., a Nest Home Hub) in which radar functionality is integrated, the smart home management device is a network-connected combination of a smart speaker, a home assistant, and a touchscreen-based control and / or entertainment hub, which can improve the parameter calculation methods or even the entire radar processing algorithm through on-site software updates launched as needed via the Internet from a central cloud server.

[0150] Although implemented in the digital domain Figure 6 and Figure 8 Regarding the receiver-side beam guidance in embodiments 600 and 800, the functions of beam guidance modules 610 and 810 can be implemented in the analog domain using analog components. If such beam guidance is performed in the analog domain, a conversion to the digital domain can be performed after such analog beam guidance is performed, thereby providing digital data to radar processing modules 410 or 850.

[0151] Figure 9 An embodiment of a contactless sleep tracking device 900 (“Device 900”) is shown. Alternatively, Device 900 may be referred to as a contactless breathing tracking device. Device 900 may have a front surface including a front transparent screen 940, making the display visible. Such a display may be a touchscreen. The area surrounding the front transparent screen 940 may be an optically opaque region, referred to as a bezel 930, through which the radar subsystem 405 has a field of view of the environment in front of Device 900.

[0152] For the following description, the terms vertical and horizontal generally describe directions relative to the bedroom; vertical refers to a direction perpendicular to the floor, while horizontal refers to a direction parallel to the floor. Since it can be... The radar subsystem of the BGT60 radar chip is generally planar and typically mounted parallel to the bezel 930 to achieve spatial compactness of the device as a whole. Since the antenna within the radar chip is located in the plane of the chip, the receiving beam of the radar subsystem 320 can be pointed in a direction 950 generally perpendicular to the bezel 930 without beam aiming. Due to the offset tilt of the bezel 930 away from a purely vertical direction—in some embodiments provided at approximately 25 degrees to facilitate easy user interaction with the touchscreen functionality of the transparent screen 940—direction 950 can be directed upwards from the horizontal direction at a departure angle 951. Assuming the device 900 is typically mounted on a bedside platform (e.g., a bedside table) at approximately the same height as the top of the mattress on which the user will sleep, it may be advantageous for the receiving beam of the radar subsystem 320 to be targeted in a horizontal direction 952 or a direction approximately horizontal (e.g., between -5° and 5° from the horizontal). Therefore, vertical beam aiming can be used to compensate for the departure angle 951 of the portion of the device 900 where the radar subsystem 320 is located.

[0153] Figure 10 An exploded view of an embodiment of device 900 is illustrated. Device 900 may include: a display assembly 1001; a display housing 1002; a main circuit board 1003; a neck assembly 1004; a speaker assembly 1005; a base plate 1006; a mesh network communication interface 1007; a top sub-board 1008; a button assembly 1009; a radar assembly 1010; a microphone assembly 1011; a rocker switch bracket 1012; a rocker switch plate 1013; a rocker switch button 1014; a Wi-Fi assembly 1015; a power board 1016; and a power bracket assembly 1017. Device 900 may represent embodiments of how device 101 can be implemented.

[0154] Display assembly 1001, display housing 1002, neck assembly 1004, and base plate 1006 may collectively form a housing that accommodates all the remaining components of device 900. Display assembly 1001 may include an electronic display, which may be a touchscreen, that presents information to a user. Therefore, display assembly 1001 may include a display screen, which may include a metal plate that can serve as a ground plane for the display. Display assembly 1001 may include a transparent portion away from the metal plate, allowing various sensors to have a field of view in the general direction in which display assembly 1001 faces. Display assembly 1001 may include an outer surface made of glass or transparent plastic that serves as part of the housing of device 900.

[0155] The display housing 1002 can be made of plastic or other rigid or semi-rigid materials that serve as the housing for the display assembly 1001. Various components, such as the main circuit board 1003, the mesh network communication interface 1007, the top daughter board 1008, the button assembly 1009, the radar assembly 1010, and the microphone assembly 1011, can be mounted on the display housing 1002. The following components can be connected to the main circuit board 1003 using flat cable assemblies: the mesh network communication interface 1007, the top daughter board 1008, the radar assembly 1010, and the microphone assembly 1011. The display housing can be attached to the display assembly 1001 using adhesive.

[0156] The mesh network communication interface 1007 may include one or more antennas and can enable communication with mesh networks such as Thread-based mesh networks. The Wi-Fi component 1015 may be located at a distance from the mesh network communication interface 1007 to reduce the possibility of interference. The Wi-Fi component 1015 can enable communication with Wi-Fi-based networks.

[0157] The radar component 1010, which may include radar subsystem 320 or radar subsystem 405, may be positioned such that its RF transmitter and RF receiver are located away from the metal plate of display component 1001 and at a considerable distance from mesh network communication interface 1007 and Wi-Fi component 1015. These three components may be arranged in an approximate triangle to increase the distance between components and reduce interference. For example, in device 900, a distance of at least 74 mm may be maintained between Wi-Fi component 1015 and radar component 1010. A distance of at least 108 mm may be maintained between mesh network communication interface 1007 and radar component 1010. Furthermore, it may be desirable to minimize the impact of vibrations that may be generated by speaker 1018 on radar component 1010 due to the distance between radar component 1010 and speaker 1018 for device 900. For example, a distance of at least 79 mm may be maintained between radar component 1010 and speaker 1018. Additionally, it may be desirable to minimize any possible interference of the microphone to the received radar signal due to the distance between microphone and radar component 1010. The top subboard 1008 may include multiple microphones. For example, at least 12 mm may be maintained between the nearest microphone and the radar assembly 1010 on the top subboard 1008.

[0158] Other components may also be present. A third microphone assembly, microphone assembly 1011, may be present, which may be rear-facing. Microphone assembly 1011 may work in conjunction with the microphone of the top sub-board 1008 to isolate spoken commands from background noise. Power board 1016 may convert power received from AC power to DC to power the components of device 900. Power board 1016 may be mounted within device 900 using power bracket assembly 1017. Rocker switch bracket 1012, rocker switch board 1013, and rocker switch button 1014 may be used together to receive user input, such as up / down input. Such input may be used, for example, to adjust the volume of sound output through speaker 1018. As another user input, button assembly 1009 may include a user-actuable toggle button. Such user input may be used to activate and deactivate all microphones, for example, when the user desires privacy and / or does not want device 900 to respond to voice commands.

[0159] Figure 11 An embodiment of a state machine 1100 for determining when a person is sleeping is illustrated. Based on data output by the radar processing module 312, the sleep state detection engine 314 can use the state machine 1100 to determine whether a person is sleeping. It should be understood that in some embodiments, the sleep state detection engine 314 is integrated into the functionality of the radar processing module 312 and does not exist as a separate module. The state machine 1100 may include five possible sleep states: going to bed state 1101; not in bed state 1102; moving in bed state 1103; not moving in bed state 1105; and getting out of bed state 1104.

[0160] If no movement indication waveform data is observed, this indicates that the user is not in bed. Due to their vital signs, a user in bed can be expected to always make at least some movement. Therefore, if zero movement is observed, the user can be determined to be in state 1101. After determining state 1101, the next possible state to determine is state 1102. In state 1102, the monitored user is getting into bed. Significant user movement can be sensed, for example, according to Table 2. This can indicate that the user is getting into bed and can cause the state to transition from state 1101 to state 1102.

[0161] Starting from state 1102, movement on the bed can continue to be detected, such as due to the user rolling around, positioning, moving pillows, sheets and / or blankets, reading, etc. While such movement continues to be detected, state 1102 can transition to state 1103. Alternatively, if movement is detected, and then zero movement is detected, this can indicate that state 1105 has been entered by the monitored user getting out of bed. If this occurs, state 1102 can transition to state 1105 and then back to state 1101. Typically, state 1104 can be interpreted as the user being asleep and state 1103 can be interpreted as the user being awake. In some embodiments, more than a threshold amount of time (or some other form of determination using a threshold criterion at least partially based on time) is required in state 1104 to classify the user as asleep, and more than a threshold amount of time (or some other form of determination using a threshold criterion at least partially based on time) is required in state 1103 to classify the user as awake. For example, if a user was previously identified as asleep, movement in bed for less than five seconds can be interpreted as movement while the user was still asleep. Therefore, if a user transitions from state 1104 to state 1103, experiences some movement, and then returns to state 1104 within a short period, the user can be identified as having experienced a "sleep awakening," where the user's sleep was disturbed, but the user was not yet awakened. Such sleep awakenings can be tracked along with the period when the user is determined to be fully awake, or data can be maintained separately from the period when the user is determined to be fully awake.

[0162] Starting from state 1103, the monitored user can be identified as getting out of bed in state 1105 and can become immobile in state 1104. "Immobile" in state 1104 means the monitored user is not performing large movements, but continues to perform small movements due to vital signs. In some embodiments, vital signs are considered accurate and / or stored, recorded, or otherwise used to measure the user's vital signs only when the monitored user's state is determined to be state 1104. Data collected during states 1103 and 1104 can be used to determine the monitored user's general sleep patterns (e.g., duration of tossing and turning, quality of sleep, when deep sleep occurs, when REM sleep occurs, etc.). After the user enters state 1104 for a predefined period of time, the user can be assumed to be asleep until the user exits state 1104. When the user initially transitions to state 1104, it may be necessary for the user to remain in state 1104 for a certain amount of time, such as two to five minutes, to be considered asleep. If the user remains in state 1103 for at least a defined period of time, the user can be identified as awake. However, if a user moves from state 1104 to state 1103 in less than the defined time period and returns to state 1104, the user can be identified as simply moving within their sleep and having been asleep continuously.

[0163] Figure 3-11 Systems, devices, and state machines can be used to execute various methods. Figure 12 An embodiment of a method for performing non-contact monitoring of a user's breathing to determine the presence of abnormal breathing conditions is illustrated. Method 1200 can be performed using a non-contact health monitoring device 301. More specifically, it can be used... Figure 5B The system 500B is used to execute method 1200.

[0164] In block 1210, radio waves are transmitted. The transmitted radio waves can be from a continuous wave radar, such as FMCW. FMCW radar can transmit radio waves, as discussed in... Figure 4B As detailed above, radio waves can be transmitted by the RF transmitter 406 of the radar subsystem 405. At block 1220, reflections of the radio waves can be received, for example, by multiple antennas of the RF receiver 440 of the radar subsystem 405. Reflections received at block 1220 can be reflected from moving objects (e.g., a person sleeping in bed, a person moving in bed) and stationary objects.

[0165] In block 1230, it is determined whether the user is eligible for respiratory analysis. This determination in block 1230 can be performed based on machine learning classification to determine whether the user is in bed and immobile (except for vital signs). The machine learning classification can be performed by neural network 416. Therefore, system 400A can be active, as per [the relevant information]. Figure 4AAs described in the detailed description, the user's state is determined according to state machine 1100, and vital signs, such as the user's breathing and heart rate, are monitored if possible. If block 1230 is determined to be positive, method 1200 can proceed to block 1230, and system 500B can be activated. If block 1230 is determined to be negative, method 1200 can return to block 1210 and continue monitoring whether the user is eligible to perform respiratory analysis.

[0166] In block 1235, beamforming is performed on n spatial regions corresponding to the user's chest in response to the received reflected radar chirp. The approximate orientation of the user's chest can be initially determined by a beam guidance module (e.g., beam guidance module 430) by identifying the direction in which the user's vital signs are detected with maximum amplitude, as detailed with respect to Equation 6. Based on the determined direction, n spatial regions can be executed, each involving different digital receiver-side beam guidance. Each target spatial region can be spaced at an angle based on the direction determined to detect the user's vital signs with maximum amplitude. The spatial regions can partially overlap.

[0167] In block 1240, the respiratory displacement of each spatial region within the nth spatial region can be determined. The amount of movement of the user's chest within the spatial region can be determined most accurately using phase. For each spatial region, a phase acquisition module can be executed, whose output is a phase indication of the frequency associated with the user's breathing. This phase value can be determined and modified as described in detail with respect to phase acquisition module 501 and radar phase processing module 510. Therefore, the output of block 1240 can indicate the instantaneous position of the user's chest. This instantaneous position can be represented by the unwound phase value.

[0168] In some embodiments, processing is performed in parallel for each of the n spatial regions. In other embodiments, processing is performed serially. If serially, after block 1240 is performed for the first spatial region, block 1250 determines whether processing blocks 1235 and 1240 are needed for the next spatial region. If so, method 1200 returns to block 1235 to perform receiver-side beamforming (corresponding to the same received reflective radar chirp) corresponding to another monitored spatial region. The number of spatial regions varies depending on the embodiment. Embodiments may include 2, 3, 4, 5, 6, 7, 8, 9, 10 or more spatial regions.

[0169] After parallel or serial analysis of all spatial regions for a given received reflective radar chirp, an n-dimensional feature vector can be created in block 1260. Block 1260 can be executed by the feature bundle compiler 520, as per [reference needed]. Figure 5BAs detailed above, the creation of n-dimensional feature vectors may include buffering the phase values ​​of each spatial region for a period of time. For each buffered set of phase values, a Fourier transform (e.g., FFT) may be performed to determine the frequency amplitude and phase values. The phase and amplitude values ​​may be used as phasors or feature vectors input to block 1270. In block 1260, n feature vectors are created such that each feature vector corresponds to one of the n monitored spatial regions.

[0170] In block 1270, n feature vectors are analyzed, for example, using a trained machine learning model. The trained machine learning model can be a neural network. The n feature vectors are used as features for the machine learning model to perform classification. The machine learning model may have been trained using a bundle of feature vectors labeled with ground truth, corresponding to people with a specific respiratory condition (e.g., paradoxical breathing) and people with healthy breathing. The machine learning model can be trained to identify users with a specific abnormal respiratory condition. In some embodiments, block 1270 can be performed multiple times using different machine learning models to identify different respiratory-related health conditions. The output of block 1270 can be one or more classifications and timestamps of the data from which the classifications are performed.

[0171] In block 1280, the classification and timestamp data obtained from block 1270 can be stored, analyzed, and / or output. In some embodiments, the classification data is stored for a period of time, such as one night (e.g., from when the user goes to sleep until the user wakes up in the morning, possibly during which the user wakes up once or more during the night). The data can be compiled, stored, and / or output. To notify the user that they may have abnormal sleep patterns, the classification of a user having abnormal sleep patterns may need to be output for at least a threshold time period or meet some other time-related criterion. For example, if an anomalous breathing classification is output for only a few seconds during the entire night, the user may not be informed of the condition. However, if the user is classified as having anomalous breathing for more than one minute during the night, the user may receive a notification that they have experienced anomalous breathing.

[0172] A night report can be provided to the user, indicating whether the user experienced any abnormal respiratory conditions during the night. The night report can be combined with a night report indicating vital sign-related data (e.g., related to heart rate and / or respiratory rate). If an abnormal respiratory condition is detected, the duration of the respiratory condition experienced by the user can also be output. A definition of the respiratory condition can also be output. The night report can be presented using a contactless device display, and / or can be output using synthesized speech. In some embodiments, the night report data can be accessed by a user logged into an account stored on a cloud-based server system (e.g., via an app installed on the user's smartphone or tablet).

[0173] Additionally or alternatively, long-term reports may be generated periodically or occasionally for the user (or accessible upon the user's request). Long-term reports may cover time periods longer than a single night, such as multiple nights, a week, two weeks, a month, six months, a year, etc., and may indicate one or more trends associated with abnormal breathing conditions. For example, a long-term report may indicate whether the user is experiencing breathing abnormalities for longer or shorter periods each night over time. Evidence of increasing duration of breathing abnormalities each night may help persuade the user to seek formal diagnosis and treatment planning from a medical professional.

[0174] In some embodiments, the device performing method 1200 may respond to a spoken command. A user might utter a command requesting a nighttime report and / or a long-term report, such as “Hi Device, how was my sleeping last night?” Upon receiving a spoken command, the device may transmit the spoken command via a wireless network interface to a cloud-based server system for interpretation. In response, the cloud-based server system may indicate the data to be output by the device. The device may receive data from the cloud-based server system, or may receive a command from the cloud-based server system indicating locally stored data to be output. A graphical interface and / or synthesized speech may be output in response to a spoken command including nighttime and / or long-term report information.

[0175] Various configurations may appropriately omit, substitute, or add various processes or components. For example, in alternative configurations, methods may be performed in a different order than those described, and / or stages may be added, omitted, and / or combined. Furthermore, features described regarding certain configurations may be combined in various other configurations. Different aspects and elements of configurations may be combined in a similar manner. Moreover, technology is evolving; therefore, many elements are merely examples and do not limit the scope of this disclosure or the claims.

[0176] Specific details are given in the description to provide a thorough understanding of the example configurations, including their implementations. However, the configurations can be implemented without these specific details. For example, well-known circuits, processes, algorithms, structures, and techniques have been shown without unnecessary detail to avoid obscuring the configurations. This description provides only example configurations and does not limit the scope, applicability, or configuration of the claims. Rather, the foregoing description of the configurations will provide those skilled in the art with an enabling description for implementing the described techniques. Various changes can be made to the function and arrangement of the elements without departing from the spirit or scope of this disclosure.

[0177] Furthermore, the configuration can be described as a process depicted as a flowchart or block diagram. Although each operation can be described as a sequential process, many operations can be executed in parallel or concurrently. Furthermore, the order of operations can be rearranged. The process may have additional steps not included in the diagram. Additionally, examples of methods can be implemented using hardware, software, firmware, middleware, microcode, hardware description languages, or any combination thereof. When implemented in software, firmware, middleware, or microcode, the program code or code segments that perform the necessary tasks can be stored in a non-transitory computer-readable medium such as a storage medium. The processor can execute the described tasks.

[0178] Several example configurations have been described, and various modifications, alternative constructions, and equivalents may be used without departing from the spirit of this disclosure. For example, the above elements may be components of a larger system, wherein other rules may be applied prior to or otherwise modify the application of the invention. Furthermore, multiple steps may be taken before, during, or after considering the above elements.

Claims

1. A non-contact health monitoring device, comprising: case; A radar sensor, comprising a plurality of antennas, wherein the radar sensor is housed within the housing; and A processing system, including one or more processors, communicating with the radar sensor, and housed within the housing, wherein the processing system is configured to: Each of the plurality of antennas receives a radar data stream from the radar sensor, thereby receiving a plurality of radar data streams; The beam guidance process is executed, which creates multiple target space region radar data streams from multiple received radar data streams; The user's breathing displacement is determined for each of the multiple target space region radar data streams. The user's breathing displacement is analyzed for each of the multiple target space region radar data streams; and The user's breathing displacement is analyzed in a time-related manner based on the radar data stream for each of the multiple target space regions, and the filtering results are output. Wherein, the first target space region radar data stream in the target space region radar data stream corresponds to the user's chest region; Wherein, the second target space region radar data stream in the target space region radar data stream corresponds to the user's abdominal region; The analysis includes detecting a phase difference between the respiratory displacement in the abdominal region and the respiratory displacement in the chest region, sufficient to indicate a paradoxical breathing condition; and The output of the filtering results includes information indicating the detected abnormal breathing conditions.

2. The non-contact health monitoring device according to claim 1, wherein, The beam guidance process includes applying weighted delay and summation beam guidance to create radar data streams for the multiple target space regions.

3. The non-contact health monitoring device according to claim 1, wherein, The processing system is configured to analyze the user's breathing displacement for each of the plurality of target space region radar data streams, including the processing system being configured to apply a pre-trained machine learning model.

4. The non-contact health monitoring device according to claim 3, wherein, The pre-trained machine learning model is a neural network.

5. The non-contact health monitoring device according to claim 1, wherein, Each of the multiple target space regions corresponds to a different part of the user's body.

6. The non-contact health monitoring device according to claim 5, wherein, The plurality of target spatial regions includes at least five target spatial regions.

7. The non-contact health monitoring device according to claim 1, wherein, The processing system is also configured to use a state machine to determine, based on the plurality of radar data streams, that the user is in a sleep state and is not moving.

8. The non-contact health monitoring device according to claim 1 further includes: A wireless network interface, which is housed in the housing and communicates with the processing system; as well as A touchscreen display, which is housed in the housing and communicates with the processing system.

9. The non-contact health monitoring device according to claim 8, further comprising: A microphone, which is housed in the housing and communicates with the processing system; as well as A speaker, housed in the housing and communicating with the processing system, wherein the processing system is further configured to: Receive a verbal command to analyze the user's respiratory displacement; This enables the verbal request to be transmitted to a cloud-based server system via the wireless network interface; In response to transmitting the dictated request to the cloud-based server system, a command is received via the wireless network interface; and The filtering results are output via the touchscreen display, at least in part, based on the received commands.

10. A non-contact health monitoring device, comprising: case; A radar sensor, comprising a plurality of antennas, wherein the radar sensor is housed within the housing; and A processing system, including one or more processors, communicating with the radar sensor, and housed within the housing, wherein the processing system is configured to: Each of the plurality of antennas receives a radar data stream from the radar sensor, thereby receiving a plurality of radar data streams; The beam guidance process is executed, which creates multiple target space region radar data streams from multiple received radar data streams; The user's breathing displacement is determined for each of the multiple target space region radar data streams. The user's breathing displacement is analyzed for each of the multiple target space region radar data streams; and The user's breathing displacement is analyzed in a time-related manner based on the radar data stream for each of the multiple target space regions, and the filtering results are output.

11. The non-contact health monitoring device according to claim 10, wherein, The screening results indicate that the user is experiencing abnormal breathing.

12. The non-contact health monitoring device according to claim 10, wherein, The beam guidance process includes applying weighted delay and summation beam guidance to create radar data streams for the multiple target space regions.

13. The non-contact health monitoring device according to claim 10, wherein, The processing system is configured to determine the user's respiratory displacement for each of the plurality of target space region radar data streams, including the processing system being configured to: Determine the phase value of each target space region radar data stream in the plurality of target space region radar data streams.

14. The non-contact health monitoring device according to claim 10, wherein, The processing system is configured to analyze the user's breathing displacement for each of the plurality of target space region radar data streams, including the processing system being configured to apply a pre-trained machine learning model.

15. The non-contact health monitoring device according to claim 14, wherein, The pre-trained machine learning model is a neural network.

16. The non-contact health monitoring device according to claim 10, wherein, Each of the multiple target space regions corresponds to a different part of the user's body.

17. The non-contact health monitoring device according to claim 16, wherein, The plurality of target spatial regions includes at least five target spatial regions.

18. The non-contact health monitoring device according to claim 10, wherein, The processing system is also configured to use a state machine to determine, based on the plurality of radar data streams, that the user is in a sleep state and is not moving.

19. The non-contact health monitoring device according to claim 10, further comprising: A wireless network interface, which is housed in the housing and communicates with the processing system; as well as A touchscreen display, which is housed in the housing and communicates with the processing system.

20. The non-contact health monitoring device according to claim 19, further comprising: A microphone, which is housed in the housing and communicates with the processing system; as well as A speaker, housed in the housing and communicating with the processing system, wherein the processing system is further configured to: Receive a verbal command to analyze the user's respiratory displacement; This enables the verbal request to be transmitted to a cloud-based server system via the wireless network interface; In response to transmitting the dictated request to the cloud-based server system, a command is received via the wireless network interface; and The filtering results are output via the touchscreen display, at least in part, based on the received commands.

Images