Sleep apnea type distinguishing method based on millimeter wave radar

Abstract

The invention discloses a sleep apnea type distinguishing method and device based on a millimeter wave radar, and a computer readable storage medium. The method comprises the following steps: acquiring a millimeter wave radar echo signal of a monitored object in a monitoring area; extracting body movement signal characteristics, thoracic cavity reflection signal energy characteristics and respiratory signal amplitude characteristics from the echo signals; on the basis of the extracted body movement signal features, the chest reflection signal energy features and the respiratory signal amplitude features, constructing a feature sequence with a time sequence; and inputting the feature sequence into a pre-trained classification model based on a self-attention mechanism to perform dynamic correlation analysis, and determining the sleep apnea type of the monitored object. The method has the advantages of non-contact type, accurate distinguishing of sleep apnea types and the like.

Description

Technical Field

[0001] The present invention relates to the field of millimeter wave radar technology, and in particular to a method, device, and computer-readable storage medium for distinguishing sleep apnea types based on millimeter wave radar. Background Art

[0002] Sleep Apnea Syndrome (SAS) is a disease characterized by repeated nocturnal breathing pauses, which can be divided into obstructive Sleep Apnea (OSA), central Sleep Apnea (CSA) and mixed Sleep Apnea (MSA).

[0003] Currently, the main technical solutions for diagnosing sleep apnea syndrome include contact polysomnography (PSG) and non-contact monitoring technologies. PSG is the "gold standard" for clinical diagnosis. It uses multiple sensors worn on the body to record physiological signals such as EEG, ECG, respiratory airflow, and blood oxygen saturation to assess the condition. As an alternative, non-contact technology based on millimeter-wave radar has also emerged. This technology transmits electromagnetic waves to the human body and analyzes the echo signals to remotely monitor micro-movements caused by breathing and heartbeat.

[0004] However, the above-mentioned existing technical solutions all have their own defects and shortcomings. For polysomnography (PSG), its equipment is usually very complex and expensive, and requires the user to attach multiple electrode patches or wear sensors on the body. This is not only cumbersome to operate, but also seriously interferes with the user's normal sleep, affecting the naturalness and authenticity of the monitoring data. As for existing millimeter-wave radar technology, its functions are relatively limited. Most solutions can only identify apnea events, that is, determine whether there is "presence" or "absence" of apnea, but cannot effectively distinguish between different types of apnea such as obstructive sleep apnea (OSA), central sleep apnea (CSA), and mixed sleep apnea (MSA). Some technologies can only identify a single type. Because different types of apnea have different causes and treatments, this lack of classification ability limits its clinical guidance value.

[0005] Therefore, the industry urgently needs a new type of sleep apnea monitoring technology. This technology not only has the advantages of non-contact and non-disturbance of millimeter-wave radar, which is suitable for long-term, comfortable home environment monitoring, but also can break through the limitations of existing radar technology and accurately distinguish different types of sleep apnea, thereby providing users with more targeted health warnings and medical references. Summary of the Invention

[0006] The embodiments of the present application provide a method for distinguishing sleep apnea types based on millimeter-wave radar, aiming to accurately distinguish the types of apnea events during sleep in a non-contact manner.

[0007] To achieve the above objectives, the present invention provides a method for distinguishing sleep apnea types based on millimeter-wave radar, comprising: Acquire millimeter-wave radar echo signals of monitored objects within the monitoring area; The following physiological features are extracted from the echo signal: Body motion signal characteristics that characterize changes in the overall trunk motion state of the monitored subject, chest cavity reflection signal energy characteristics that characterize changes in chest cavity motion energy of the monitored subject, and respiratory signal amplitude characteristics that characterize changes in respiratory airflow of the monitored subject; Constructing a time-series feature sequence based on the extracted body motion signal features, the chest cavity reflection signal energy features, and the respiratory signal amplitude features; The feature sequence is input into a pre-trained classification model based on a self-attention mechanism for dynamic correlation analysis to determine the sleep apnea type of the monitored subject.

[0008] In one embodiment, before extracting three types of physiological features from the echo signal, the method further includes: Adopting adaptive wavelet to perform denoising on the echo signal; Perform fast Fourier transform on the denoised echo signal to obtain the range spectrum; Processing the range spectrum using constant false alarm rate detection to locate a target range unit corresponding to the chest area of ​​the monitored object; The phase information of the target distance unit is unwrapped and differentially processed to extract the micro-motion signal containing the physiological characteristics.

[0009] In one embodiment, extracting body motion signal features that characterize changes in the overall torso motion state of the monitored subject includes: Based on the frequency domain representation of the echo signal, the proportion of energy of the frequency band to which the body motion signal belongs to the total signal energy and the frequency range of the frequency band to which the body motion signal belongs are calculated, and the energy proportion and frequency range of the body motion signal are used together as the body motion signal features.

[0010] In one embodiment, extracting energy features of chest cavity reflection signals representing energy changes of chest cavity motion of a monitored subject includes: The mean, standard deviation and kurtosis of the energy value sequence corresponding to the chest area of ​​the monitored object in the echo signal are calculated within a preset time window and are collectively used as the energy features of the chest reflection signal.

[0011] In one embodiment, extracting a respiratory signal amplitude feature representing changes in respiratory airflow of a monitored subject includes: performing bandpass filtering on the echo signal to obtain a respiratory signal waveform; The maximum amplitude, minimum amplitude and number of peaks of the respiratory signal waveform within a preset time window are extracted and used as the respiratory signal amplitude feature.

[0012] In one embodiment, constructing a time-sequential feature sequence includes: splicing the body motion signal feature, the chest cavity reflection signal energy feature, and the respiratory signal amplitude feature into a multidimensional feature vector; Performing windowing processing on the multidimensional feature vectors arranged in chronological order using a preset sliding window to form an input sequence; Position encoding is performed on the input sequence to represent temporal information in the feature sequence.

[0013] In one embodiment, the pre-trained classification model based on the self-attention mechanism is a Transformer network model; Analyzing the feature sequence based on the Transformer network model includes: In the multi-head self-attention layer, attention weights are calculated based on the feature sequence to capture the dynamic correlation between features in the sequence; In the feedforward network layer, a nonlinear transformation is performed on the output of the multi-head self-attention layer; In the classification layer, the Softmax function is used to normalize the results processed by the feedforward network layer to output probability values ​​corresponding to different sleep apnea types.

[0014] In one embodiment, after determining the sleep apnea type of the monitored subject, the method further includes: When it is determined that the duration of the apnea of ​​the monitored object exceeds a preset time threshold, a preset alarm event is executed.

[0015] To achieve the above-mentioned objectives, an embodiment of the present application further proposes a millimeter-wave radar-based sleep apnea type differentiation device, comprising a memory, a processor, and a millimeter-wave radar-based sleep apnea type differentiation program stored in the memory and executable on the processor. When the processor executes the millimeter-wave radar-based sleep apnea type differentiation program, the millimeter-wave radar-based sleep apnea type differentiation method as described in any one of the above items is implemented.

[0016] To achieve the above-mentioned objectives, an embodiment of the present application further proposes a computer-readable storage medium, on which a millimeter-wave radar-based sleep apnea type differentiation program is stored. When the millimeter-wave radar-based sleep apnea type differentiation program is executed by a processor, the millimeter-wave radar-based sleep apnea type differentiation method as described in any of the above items is implemented.

[0017] Based on the above embodiments, the method for distinguishing sleep apnea types based on millimeter-wave radar proposed in this application has the following beneficial effects: 1. Achieved high-precision differentiation of sleep apnea types, greatly improving diagnostic value: The technical solution of this application constructs a feature sequence by simultaneously extracting three types of complementary information: body motion signal features, chest reflection signal energy features, and respiratory signal amplitude features. This feature sequence is then analyzed using a classification model based on a self-attention mechanism. This allows for accurate differentiation of three types of sleep apnea: obstructive sleep apnea (OSA), central sleep apnea (CSA), and mixed sleep apnea (MSA).

[0018] 2. Achieves completely non-contact, comfortable monitoring, especially suitable for long-term home monitoring: This application's technical solution uses millimeter-wave radar to acquire signals. The entire process eliminates the need for the monitored subject to wear or attach any sensors or electrode patches. This fundamentally addresses the pain points of traditional polysomnography (PSG) equipment, which is complex, disrupts sleep, and leads to poor user compliance. This non-invasive, non-disruptive monitoring method is particularly suitable for children who are sensitive to touch or the elderly who require long-term, regular monitoring.

[0019] 3. The monitoring system is robust and the data foundation is reliable: The technical solution of this application includes a complete and rigorous signal preprocessing process before extracting physiological features. Adaptive wavelet denoising filters out noise, and fast Fourier transform combined with constant false alarm rate detection automatically and accurately locates the chest position of the monitored subject. Phase unwrapping and differential analysis are then used to extract micro-motion signals with a high signal-to-noise ratio. This series of processes ensures that the data source for subsequent feature extraction and model analysis is pure and reliable, thereby improving the stability of the entire system and its adaptability to environmental changes.

[0020] 4. Intelligent alarm function and high practicality: The alarm function of this application's technical solution is not simply triggered, but rather incorporates a determination of the duration of apnea. Only when the duration of apnea exceeds a clinically significant preset threshold will the system trigger a preset alarm event. This design effectively filters out brief, harmless respiratory pauses, avoiding unnecessary false alarms that harass users or guardians, making the warning function more accurate, reliable, and user-friendly. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0022] Figure 1 This is a module structure diagram of an embodiment of a device for distinguishing sleep apnea types based on millimeter-wave radar according to the present invention; Figure 2 FIG2 is a flow chart of an embodiment of a method for distinguishing sleep apnea types based on millimeter-wave radar according to the present invention.

[0023] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0024] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0025] To better understand the above technical solutions, exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. Instead, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0026] It should be noted that in the claims, any reference signs placed between brackets shall not be construed as limiting the claims. The presence of "comprising" in the text does not exclude the presence of components or steps not listed in the claims. The quantifier "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The present invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim that lists several means, several of these means may be embodied by the same hardware item. The use of "first", "second", and "third" etc. does not indicate any order and these words may be interpreted as names.

[0027] like Figure 1 As shown, Figure 1 It is a structural diagram of a server 1 (also called a sleep apnea type differentiation device based on millimeter-wave radar) in the hardware operating environment involved in an embodiment of the present invention.

[0028] The server of the embodiment of the present invention is a device with display function such as "Internet of Things devices", smart air conditioners, smart lights, smart power supplies with networking functions, AR / VR devices with networking functions, smart speakers, self-driving cars, PCs, smart phones, tablet computers, e-book readers, portable computers, etc.

[0029] like Figure 1 As shown, the server 1 includes: a memory 11 , a processor 12 and a network interface 13 .

[0030] The memory 11 includes at least one type of readable storage medium, including flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the server 1, such as the hard disk of the server 1. In other embodiments, the memory 11 may also be an external storage device of the server 1, such as a plug-in hard disk equipped on the server 1, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc.

[0031] Furthermore, the memory 11 may include both an internal storage unit of the server 1 and an external storage device. The memory 11 can be used not only to store application software installed on the server 1 and various data, such as the code of the millimeter-wave radar-based sleep apnea type classification program 10, but also to temporarily store data that has been output or is about to be output.

[0032] In some embodiments, the processor 12 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip, configured to execute program codes or process data stored in the memory 11 , such as executing the millimeter-wave radar-based sleep apnea type differentiation program 10 .

[0033] The network interface 13 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface), and is generally used to establish a communication connection between the server 1 and other electronic devices.

[0034] The network may be the Internet, a cloud network, a wireless fidelity (Wi-Fi) network, a personal area network (PAN), a local area network (LAN), and / or a metropolitan area network (MAN). Various devices in the network environment may be configured to connect to the communication network according to various wired and wireless communication protocols. Examples of such wired and wireless communication protocols may include, but are not limited to, at least one of the following: Transmission Control Protocol and Internet Protocol (TCP / IP), User Datagram Protocol (UDP), Hypertext Transfer Protocol (HTTP), File Transfer Protocol (FTP), ZigBee, EDGE, IEEE 802.11, Light Fidelity (Li-Fi), 802.16, IEEE 802.11s, IEEE 802.11g, multi-hop communication, wireless access point (AP), device-to-device communication, cellular communication protocol, and / or Bluetooth communication protocol, or a combination thereof.

[0035] Optionally, the server may further include a user interface, which may include a display and an input unit such as a keyboard. Optional user interfaces may also include standard wired and wireless interfaces. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display, which may also be referred to as a display screen or display unit, is used to display information processed by server 1 and to display a visual user interface.

[0036] Figure 1 Only the server 1 having components 11-13 and the sleep apnea type differentiation program 10 based on millimeter wave radar is shown. It can be understood by those skilled in the art that Figure 1 The structure shown does not constitute a limitation on the server 1 , and the server 1 may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0037] In this embodiment, the processor 12 may be configured to call the millimeter-wave radar-based sleep apnea type differentiation program stored in the memory 11 and perform the following operations: Acquire millimeter-wave radar echo signals of monitored objects within the monitoring area; The following physiological features are extracted from the echo signal: Body motion signal characteristics that characterize changes in the overall trunk motion state of the monitored subject; The energy characteristics of the chest cavity reflection signal representing the energy changes of the chest cavity motion of the monitored subject, and the amplitude characteristics of the respiratory signal representing the changes of the respiratory airflow of the monitored subject; Constructing a time-series feature sequence based on the extracted body motion signal features, the chest cavity reflection signal energy features, and the respiratory signal amplitude features; The feature sequence is input into a pre-trained classification model based on a self-attention mechanism for dynamic correlation analysis to determine the sleep apnea type of the monitored subject.

[0038] Based on the hardware architecture of the millimeter-wave radar-based sleep apnea classification device described above, an embodiment of the present invention's millimeter-wave radar-based sleep apnea classification method is proposed. This millimeter-wave radar-based sleep apnea classification method aims to accurately distinguish apnea events during sleep using a non-contact method.

[0039] Reference Figure 2 , Figure 2 This is an embodiment of a method for distinguishing sleep apnea types based on millimeter wave radar according to the present invention. The method for distinguishing sleep apnea types based on millimeter wave radar includes the following steps: S10: Acquire the millimeter-wave radar echo signal of the monitored object in the monitoring area.

[0040] Specifically, a frequency-modulated continuous wave (FMCW) millimeter-wave radar can be deployed in the sleeping environment, for example, fixed to the wall directly above the headboard, 1.5-1.6 meters above the ground, with the radar aimed at the chest area of ​​the monitored subject at an angle of approximately 45°. The radar continuously samples within a set range at a preset sampling interval, obtaining raw echo signals that reflect the physiological activity of the monitored subject. Before formal monitoring begins, the system can also first determine whether the monitoring bed is occupied. For example, by clustering the radar point cloud and performing tracking algorithms, it can confirm that a target has entered the set monitoring bed position range before initiating subsequent signal processing and analysis.

[0041] S20. Extract the following physiological features from the echo signal: a body motion signal feature representing changes in the overall trunk motion state of the monitored subject, a chest cavity reflection signal energy feature representing changes in chest cavity motion energy of the monitored subject, and a respiratory signal amplitude feature representing changes in respiratory airflow of the monitored subject.

[0042] Regarding the three types of physiological characteristics in step S20, the respiratory signal amplitude characteristic is the basis for determining whether an apnea event has occurred. Regardless of the type of apnea, the common characteristic is the cessation of respiratory airflow. Therefore, by analyzing the amplitude changes of the respiratory signal (for example, whether the amplitude approaches zero), the presence or absence of an apnea event can be determined.

[0043] Secondly, body motion signal characteristics are key to distinguishing central sleep apnea (CSA) from obstructive sleep apnea (OSA). Central sleep apnea is caused by the failure of the central nervous system to issue respiratory commands, resulting in complete inactivity of respiratory muscles (such as the diaphragm and intercostal muscles), manifesting as a stillness of the entire chest, abdomen, and even trunk. In obstructive sleep apnea, central commands are normal, and the patient continues to strive for breathing. However, the upper airway is blocked, resulting in sustained or even more intense chest and abdominal movements. By analyzing the overall trunk movement, it is possible to effectively distinguish the "completely still" central sleep apnea from the "still struggling" obstructive sleep apnea.

[0044] Furthermore, the energy characteristics of the chest reflection signal provide strong evidence and refinement of obstructive sleep apnea (OSA). During obstructive sleep apnea, due to airway obstruction, the subject unconsciously increases their respiratory effort in an attempt to overcome the obstruction. This manifests as dramatic and regular fluctuations in the energy of the chest movement (i.e., the radar echo signal energy). Quantifying these energy variations (for example, analyzing their amplitude or statistical distribution) allows for more reliable identification of obstructive sleep apnea.

[0045] In other words, by combining the above three types of physiological characteristics, a complete logical chain can be formed: first, use "respiratory amplitude" to determine whether there is a pause, then use "body movement characteristics" to distinguish whether there is respiratory effort (distinguishing between CSA and OSA), and then use "thoracic energy" to confirm the pattern of respiratory effort, thereby achieving a comprehensive and accurate distinction between the three types.

[0046] In some embodiments, in order to obtain pure physiological information that can be used for analysis from the original radar echo full of noise and environmental clutter, a systematic preprocessing process needs to be performed on the echo signal before the three physiological features are formally extracted. This process aims to convert the original, complex echo signal into a time series signal that can accurately reflect the micro-movement of the monitored subject's chest cavity. Specifically, the preprocessing process includes the following steps: S51. De-noise the echo signal using an adaptive wavelet filter. Specifically, the system uses an adaptive wavelet filter to de-noise the echo signal. Raw radar signals are inevitably contaminated by circuit noise and environmental electromagnetic interference. Adaptive wavelet de-noising is a highly effective processing method. By selecting an appropriate wavelet basis (e.g., Daubechies 6, or DB6) and decomposing the signal into different scales, this method dynamically calculates the noise threshold for each layer of wavelet coefficients based on the signal's statistical characteristics. This effectively suppresses noise while maximally preserving the non-stationary signal details generated by human physiological activity.

[0047] For example, when using adaptive wavelet denoising, the noise standard deviation estimate is calculated The calculation formula is: ; Among them, D1 is the first layer wavelet detail coefficient, 0.6745 is the empirical correction coefficient of Gaussian distribution, and the adaptive soft threshold T calculation formula is: ; Where N is the signal length, and the threshold T of each layer can be dynamically adjusted through the noise standard deviation calculated above, thereby effectively shrinking the wavelet coefficients, retaining the effective signal and suppressing the noise.

[0048] S52. Perform Fast Fourier Transform (FFT) processing on the denoised echo signal to obtain a range spectrum. Specifically, after the frequency modulated continuous wave signal emitted by the radar is reflected by the monitored object, its round-trip delay is proportional to the distance of the monitored object. By performing FFT processing on the echo signal (for example, 128-point FFT), the time domain signal can be transformed into the range domain to form a "range-amplitude" spectrum. ,in is the range frequency, and r is the range unit. On this range spectrum, different peaks correspond to reflecting objects at different distances within the radar detection range.

[0049] S53. Use constant false alarm rate (CFAR) detection to process the distance spectrum to locate the target distance unit corresponding to the chest area of ​​the monitored object. Specifically, after obtaining the distance spectrum, the system needs to automatically and accurately find which peak comes from the chest of the monitored object. CFAR is an adaptive detection algorithm that traverses the entire distance spectrum through a sliding window. For the "unit to be detected" in the center of the window, the algorithm calculates the average noise level of the "reference units" on both sides of it, and sets a dynamic detection threshold based on the noise level. When the signal strength of the unit to be detected is significantly higher than this dynamic threshold, the system determines that there is a target here. In this way, the distance unit (bin) where the chest of the monitored object is located can be robustly locked without being interfered with by other stationary objects in the room (such as furniture).

[0050] For example, when using a CFAR detector, the protection cell length L=16 and the reference cell length M=32 (16 cells on the left and 16 cells on the right).

[0051] S54. The phase information of the target distance unit is unwrapped and differentially processed to extract the micro-motion signal containing the physiological characteristics. Specifically, the phase of the radar echo signal is extremely sensitive to the tiny displacement of the target. The millimeter-level or even submillimeter-level movement of the chest due to breathing and heartbeat will cause regular changes in the phase. However, the original phase value is wrapped in the range of -π to +π, and there is a "2π ambiguity" phenomenon. Therefore, it is necessary to first restore it to a continuously changing real phase through a phase unwrapping algorithm. Then, by calculating the difference between the continuous phases of adjacent radar frames, the static phase offset can be eliminated, and finally a time series signal that only reflects the tiny displacement of the chest between two sampling intervals is obtained. This signal is the "micro-motion signal", which carries all the physiological information required for subsequent analysis.

[0052] For example, the formula for calculating the phase difference between two adjacent frames is: ; in, is a continuous phase.

[0053] It can be understood that through a series of continuous signal processing steps such as adaptive wavelet denoising, fast Fourier transform, constant false alarm rate detection, and phase unwrapping and differentiation, it is possible to automatically and robustly locate the chest position of the monitored target from the original radar echo full of noise and clutter, and extract micro-motion signals with a high signal-to-noise ratio that can accurately reflect subtle physiological activities such as breathing and heartbeat, providing a high-quality and reliable data foundation for subsequent high-precision physiological feature extraction and sleep apnea type differentiation.

[0054] In some embodiments, extracting body motion signal features that characterize changes in the overall trunk motion state of the monitored object includes: based on the frequency domain representation of the echo signal, calculating the proportion of the energy of the frequency band to which the body motion signal belongs to the total signal energy and the frequency range of the frequency band to which the body motion signal belongs, and taking the energy proportion and frequency range of the body motion signal together as the body motion signal features.

[0055] Specifically, large-scale body movements (such as rolling over or moving limbs) and steady breathing exhibit significantly different characteristics in the frequency domain of radar echo signals. Therefore, by transforming the signals into the frequency domain for analysis, these two types of movements can be effectively distinguished.

[0056] Specifically, the system first performs time-frequency analysis on the preprocessed micro-motion signal. For example, the short-time Fourier transform (STFT) technique can be used. A window function (such as a Hanning window) of a specific length (e.g., 256 points) is slid along the time axis to segment the micro-motion signal and perform a Fourier transform. This generates a time-frequency spectrum that simultaneously displays the temporal variations of the signal's frequency components. This time-frequency spectrum is the so-called "frequency domain representation of the echo signal." Like a map, it clearly depicts how energy is distributed across different frequencies at each point in time.

[0057] Once the time-spectrogram is obtained, the proportion of the energy in the frequency band to which the body motion signal belongs to the total signal energy can be calculated, and its frequency range can be determined. First, based on prior knowledge or experimental data, the frequency ranges corresponding to different physiological activities must be defined. For example, the frequency of steady breathing typically ranges from 0.1Hz to 0.8Hz, while the Doppler shift generated by large body movements falls into a relatively higher frequency band, such as 0.8Hz to 5Hz. Then, within each time window, by integrating or summing the time-spectrogram data, the total signal energy and the signal energy falling within the predefined "body motion signal frequency band" (e.g., 0.8Hz-5Hz) can be calculated. Dividing the latter by the former yields the energy proportion of the body motion signal. Furthermore, by searching for the lowest and highest frequencies with non-zero energy within the body motion signal frequency band, the actual frequency range of the body motion signal within that time window can be determined.

[0058] For example, the energy ratio of the body motion signal is extracted , frequency range The calculation formula is: ; ; in They represent the starting frequency to be calculated and the starting frequency and cutoff frequency of the body motion signal to be calculated respectively; Represents the frequency domain amplitude of the respiratory signal; are the maximum and minimum frequency values ​​in the body motion signal.

[0059] It can be understood that through time-frequency analysis methods such as short-time Fourier transform, the signal can be expanded in the time and frequency dimensions, and the frequency bands belonging to large-amplitude limb or trunk movements can be accurately divided. Therefore, by calculating the energy proportion and frequency range of this specific frequency band, it is possible to highly sensitively and quantifiably capture whether the monitored object has overall movements such as turning over and limb twitching that are different from respiratory micro-movements.

[0060] In addition, the body motion signal features can also be extracted by distinguishing the high-frequency / high-energy part from the low-frequency / low-energy part in the echo signal.

[0061] In some embodiments, extracting the energy characteristics of the chest reflection signal that characterizes the energy changes of the chest movement of the monitored object includes: calculating the mean, standard deviation and kurtosis of the energy value sequence corresponding to the chest area of ​​the monitored object in the echo signal within a preset time window, and collectively serving as the energy characteristics of the chest reflection signal.

[0062] Specifically, the system first needs to obtain a sequence that represents the energy variation of chest motion over time. This sequence is derived from the "target distance unit" located during early signal processing. By calculating the echo signal strength or power at each sampling moment within this unit, a sequence of energy values ​​within a preset time window (for example, 60 seconds) is generated. This energy value sequence intuitively records the intensity of chest motion.

[0063] After obtaining the energy value sequence, in order to describe it from multiple dimensions, the system calculates three key statistics of the sequence: mean, standard deviation, and kurtosis. The "mean" of energy reflects the average energy level of chest movement within a time window. The "standard deviation" of energy quantifies the degree of dispersion of the energy sequence. A higher standard deviation means that the energy fluctuations of chest movement are very violent, which is highly correlated with the struggling state during obstructive apnea. The "kurtosis" of energy describes the peak shape of the energy value distribution. A higher kurtosis value usually means that there are some short but extremely high-intensity spikes in the energy sequence, which corresponds to the sudden and explosive breathing efforts of the monitored subject. These three statistics together constitute a multi-dimensional "fingerprint" of the energy pattern of chest movement.

[0064] For example, the mean value of the energy is extracted from the chest cavity reflection signal energy feature. , standard deviation , kurtosis , their respective calculation formulas are as follows: ; ; ; Where N is the number of sampling points in the window, Represents the energy of the sampling point of the respiratory signal time domain waveform.

[0065] It can be understood that by calculating the mean, standard deviation and kurtosis of the energy series, the energy change characteristics of chest movement can be comprehensively quantified from three dimensions: average level, fluctuation amplitude and distribution form. Therefore, it can effectively capture and distinguish normal stable breathing from the intense and irregular breathing efforts in obstructive sleep apnea, providing rich and highly discriminative judgment information for the accurate identification of obstructive sleep apnea events.

[0066] In addition, the energy characteristics of the chest cavity reflection signal can also be extracted by calculating and analyzing the energy time series changes of the signal segment representing the chest cavity movement.

[0067] In some embodiments, extracting a respiratory signal amplitude feature that characterizes changes in the respiratory airflow of the monitored subject includes: performing bandpass filtering on the echo signal to obtain a respiratory signal waveform; and extracting the maximum amplitude, minimum amplitude, and number of peaks of the respiratory signal waveform within a preset time window, collectively as the respiratory signal amplitude feature.

[0068] Specifically, the system first cleanly separates the respiratory signal waveform from the echo signal that contains multiple information such as heartbeat, body vibration, and environmental noise. This can be achieved through a frequency-selective bandpass filter. Considering that the respiratory frequency of normal adults at rest is mainly concentrated in the extremely low frequency band of 0.1Hz to 0.8Hz, the system can apply a digital bandpass filter with the same passband range (0.1-0.8Hz) to filter the micro-motion signal obtained from the previous processing. In this way, the higher-frequency heartbeat signal and body vibration signal, as well as the lower-frequency signal drift, will be effectively filtered out, resulting in a relatively smooth, periodic "respiratory signal waveform" that can clearly reflect each inhalation and exhalation process.

[0069] After obtaining a pure respiratory signal waveform, the system can quantify the waveform morphology within a preset time window (for example, 30 seconds) to extract the respiratory signal amplitude characteristics. This includes extracting the maximum and minimum values ​​of the waveform amplitude within the time window. These two values ​​intuitively reflect the depth of breathing. When breathing stops, these two values ​​will be very close to zero. At the same time, the system will also use the peak detection algorithm to count the number of peaks (or troughs) that appear in the respiratory waveform within the time window. This "peak number" directly corresponds to the number of breaths in the time period and is the basis for calculating the respiratory frequency. When respiratory arrest occurs, the number of peaks in the pause period will be zero.

[0070] It can be understood that by bandpass filtering the radar echo signal in a specific frequency band (such as 0.1-0.8Hz), a pure respiratory signal waveform that can directly reflect the respiratory rhythm and depth can be effectively separated from the mixed micro-motion signal. Therefore, by directly measuring the maximum / minimum amplitude and number of peaks of this waveform, the presence, speed, and strength of breathing can be intuitively and accurately quantified, providing the most basic and direct basis for judging the occurrence of all types of apnea events, and is a necessary prerequisite for subsequent type differentiation.

[0071] In addition, the respiratory signal amplitude feature can also be extracted by filtering to obtain a pure respiratory waveform and measuring its morphology.

[0072] S30: constructing a time-series feature sequence based on the extracted body motion signal features, the chest cavity reflection signal energy features, and the respiratory signal amplitude features.

[0073] Specifically, after extracting the three physiological features of body motion, chest energy, and respiratory amplitude, the discrete information extracted at each independent time point needs to be integrated and constructed into a structured data format that can be understood and processed by time series analysis models (such as Transformer). The purpose of this step is to convert instantaneous features into a feature sequence with a time dimension, that is, a feature sequence with time series characteristics.

[0074] In some embodiments, constructing a time-sequential feature sequence includes the following steps: S31. Splice the body motion signal features, the chest cavity reflection signal energy features and the respiratory signal amplitude features into a multi-dimensional feature vector. Specifically, at any sampling moment, the system obtains different feature values ​​belonging to three categories. For example, the body motion feature may be 2-dimensional, the chest cavity energy feature is 3-dimensional, and the respiratory amplitude feature is also 3-dimensional. By arranging and connecting (i.e., splicing) the feature values ​​from these different sources in a preset order, a higher-dimensional, unified feature vector can be formed, such as an 8-dimensional feature vector. This multi-dimensional feature vector comprehensively describes the comprehensive physiological state of the monitored object at the sampling moment.

[0075] S32. The multidimensional feature vectors arranged in chronological order are windowed using a preset sliding window to form an input sequence. Specifically, the feature vector of a single time point is not sufficient for time series analysis, so it is necessary to observe continuous changes over a period of time. The sliding window technology is used to achieve this purpose. The system sets a window of a fixed time length (for example, 60 seconds) and slides this window along the time axis. At each position, the window "frames" all continuous multidimensional feature vectors within the time period and combines them into a two-dimensional matrix, which is an "input sequence." Subsequently, the window slides forward a small step according to a preset step size (for example, 15 seconds) and "frames" the next input sequence again. This sliding method with overlap helps to ensure that the feature information at the window boundary is not missed.

[0076] S33. Position encoding is performed on the input sequence to represent the temporal information in the feature sequence. Specifically, a model like Transformer does not have the ability to perceive the order of elements in the sequence. It needs to be explicitly informed of the position of each feature vector in the time window. Position encoding is used to inject this temporal information. The system generates a unique "position encoding vector" with the same dimension as the feature vector for each position in the sequence (for example, from the 1st time point to the 60th time point). This position encoding vector is usually generated by periodic functions such as sine and cosine, and its value is determined only by its position in the sequence. By adding or concatenating this position encoding vector with the feature vector of the corresponding position, a final feature sequence containing rich physiological information and temporal information can be formed, and then fed into the subsequent classification model.

[0077] For example, a 60-second window is used and the window is slid in time order (step length 15 seconds) to generate the input sequence. The sine-cosine function is used to generate the position code to ensure that the temporal sequence of the data is not lost. The specific coding is implemented as follows: ; ; Where pos is the position index; i represents the dimension of the encoding vector (half the number of features); is the hidden layer dimension of the model.

[0078] It can be understood that through a series of processes such as feature splicing, sliding windowing, and position encoding, instantaneous physiological features of different modalities and dimensions are integrated into sequence data with a unified format that can reflect the time evolution process, and the absolute or relative position information is injected into each time step in the sequence, thereby providing an ideal input data format for the subsequent classification model based on the self-attention mechanism, so that the model can not only analyze the physiological state at each moment, but also understand and learn the dependencies, sequences, and dynamic changes of these states in the time dimension, so as to accurately distinguish different types of sleep apnea.

[0079] S40: Input the feature sequence into a pre-trained classification model based on a self-attention mechanism to perform dynamic correlation analysis to determine the sleep apnea type of the monitored subject.

[0080] In this example, the pre-trained classification model is a Transformer network model. Its core advantage lies in its ability to efficiently process long sequences of data and capture long-range dependencies between elements within the sequence. The model's internal processing is primarily accomplished through its core encoder structure, which includes steps such as multi-head self-attention calculations, nonlinear transformations in a feedforward network, and final classification normalization.

[0081] Specifically, analyzing the feature sequence based on the Transformer network model includes the following steps: S41. In the multi-head self-attention layer, attention weights are calculated based on the feature sequence to capture the dynamic correlation between the features in the sequence.

[0082] Specifically, the feature sequence is first fed into the model's multi-headed self-attention layer, where attention weights are calculated based on the feature sequence to capture the dynamic correlations between features within the sequence. The core of the self-attention mechanism is that for each time point in the sequence (i.e., each feature vector), it calculates the strength of the correlation, or "attention score," between that time point and all other time points. This means the model can autonomously learn which other moments in the sequence to focus on when judging the current state. For example, when judging a mixed apnea event, the model may simultaneously attend to the central pause features at the front of the sequence and the obstructive pause features at the back of the sequence. The multi-headed mechanism allows the model to perform multiple such attention calculations in parallel across different representation subspaces, thereby capturing richer and more diverse feature correlation patterns from different perspectives.

[0083] For example, the self-attention weight calculation formula is: ; In it, query represents the query matrix, key represents the key matrix, T represents the transpose, hidden_dim represents the matrix dimension, and value represents the value matrix.

[0084] S42. In the feedforward network layer, a nonlinear transformation is performed on the output of the multi-head self-attention layer.

[0085] Specifically, the output after processing by the multi-head self-attention layer will be sent to the feedforward network layer for nonlinear transformation. The output of the self-attention layer is a new sequence after weighted fusion of contextual information, in which the vector of each position contains the associated information of the entire sequence. The subsequent feedforward network layer (usually a fully connected network consisting of two linear layers and an activation function such as ReLU) will independently perform a nonlinear mapping on each position vector in the new sequence. The purpose of this step is to increase the nonlinear expression ability of the model so that the model can learn more complex and abstract feature combinations. Usually, a complete Transformer encoder layer is composed of a multi-head self-attention layer and a feedforward network layer (as well as the residual connection and layer normalization between them). In some implementations of the present application, the feature sequence will pass through 4 such encoder layers in sequence for in-depth feature extraction layer by layer.

[0086] S43. In the classification layer, a Softmax function is used to normalize the results processed by the feedforward network layer to output probability values ​​corresponding to different sleep apnea types.

[0087] Specifically, after the feature sequence passes through all encoder layers in sequence, its final representation is fed into an independent classification layer. In the classification layer, the Softmax function is used to normalize the results to output probability values ​​corresponding to different types of sleep apnea. The Softmax function can convert a vector of arbitrary values ​​into a probability distribution vector whose elements sum to 1. For example, the original output of the classification layer may be a vector containing four real values, corresponding to the scores of four categories (normal, obstructive, central, and mixed). After processing by the Softmax function, these four scores are converted into four probability values ​​between 0 and 1, and their sum is 1. The category with the highest score is the identified sleep apnea type or non-sleep apnea.

[0088] For example, the Softmax normalization calculation formula is: ; Where z is a vector. 、 Represents the i-th and j-th elements.

[0089] For example, a 60x8 feature sequence is input into a Transformer model with four encoder layers. In the multi-head self-attention sublayer of the first encoder layer, the model calculates a high attention weight between the 10th and 45th time points in the sequence. After deep processing through the four encoder layers, the final sequence representation is fed into the classification layer. The raw score vector output by the classification layer might be [-1.5, 0.5, 3.0, -2.0]. This vector is normalized using the Softmax function and converted into a probability distribution vector, for example: {Normal: 0.01, Obstructive: 0.08, Central: 0.90, Mixed: 0.01}. Based on this probability distribution, the system determines the sleep apnea type within that time window as "central," as its corresponding probability value of 0.90 is the highest.

[0090] It can be understood that due to the adoption of a deep processing architecture based on a multi-head self-attention mechanism and a feedforward network, and in particular because the multi-head self-attention mechanism can learn the dependency between any two positions in a sequence from different subspaces in parallel, thereby dynamically calculating the mutual influence weights of different physiological characteristics in the time dimension, the model can transcend the local dependency limitations of traditional time series models and effectively capture the long-distance, cross-time complex patterns that are crucial for determining the type of sleep apnea, greatly improving the model's ability to understand the dynamic changes in time series and the ultimate classification accuracy.

[0091] It is worth noting that in other embodiments of the present application, classification models such as recurrent neural networks, long short-term memory networks, convolutional neural networks, and support vector machines may also be used.

[0092] In some embodiments, to elevate the method provided herein from a simple classification system to a real-time monitoring tool with practical early warning capabilities, after determining the subject's sleep apnea type, the system further implements an intelligent alarm process. Specifically, if the duration of the subject's apnea exceeds a preset time threshold, a preset alarm event is triggered. This process is designed to provide timely alerts for potentially harmful respiratory events while avoiding unnecessary interference to the user or guardian due to oversensitivity.

[0093] Specifically, the alarm process is not triggered immediately every time apnea is detected, but rather introduces logic to determine the duration of the event. The system maintains a timer in the background. When the classification model first outputs one of the three types of apnea, obstructive, central, or mixed, the timer starts counting. In each subsequent analysis window, if the classification result is still one of the three apnea types, the timer continues to accumulate; conversely, if the classification result changes to a "normal" state, the timer is immediately reset.

[0094] Only when the accumulated duration of the timer exceeds a clinically significant, pre-set time threshold will the system finally execute the preset alarm event. This time threshold is critical because, physiologically, brief (e.g., a few seconds) respiratory pauses can occur during normal sleep and are not necessarily dangerous. By setting a reasonable threshold (for example, according to medical guidelines, respiratory arrest lasting more than 10 seconds is generally defined as an apnea event, which can be set to 20 seconds in a specific implementation of this application), it can ensure that the system only alarms for events that last longer, may cause a drop in blood oxygen, and pose potential health risks. The so-called "execution of a preset alarm event" can include various forms, such as pushing an alarm notification to a bound smartphone app via WiFi or Bluetooth, driving a bedside sound and light alarm to sound or flash, or recording a serious event that requires special attention in the system's monitoring log.

[0095] It can be understood that since the alarm step of the present application does not trigger an alarm immediately every time apnea is detected, but introduces logic to judge the duration of the apnea event and sets a preset time threshold (such as 20 seconds), it can effectively filter out short-term respiratory pauses that are not clinically significant, significantly reduce the false alarm rate of the alarm, avoid unnecessary harassment to users or guardians, and make the alarm function more practical and reliable. An effective reminder is only issued when a longer apnea event that may pose a threat to health occurs.

[0096] Based on the above embodiments, the method for distinguishing sleep apnea types based on millimeter-wave radar proposed in this application has the following beneficial effects: 1. Achieved high-precision differentiation of sleep apnea types, greatly improving diagnostic value: The technical solution of this application constructs a feature sequence by simultaneously extracting three types of complementary information: body motion signal features, chest reflection signal energy features, and respiratory signal amplitude features. This feature sequence is then analyzed using a classification model based on a self-attention mechanism. This allows for accurate differentiation of three types of sleep apnea: obstructive sleep apnea (OSA), central sleep apnea (CSA), and mixed sleep apnea (MSA).

[0097] 2. Achieves completely non-contact, comfortable monitoring, especially suitable for long-term home monitoring: This application's technical solution uses millimeter-wave radar to acquire signals. The entire process eliminates the need for the monitored subject to wear or attach any sensors or electrode patches. This fundamentally addresses the pain points of traditional polysomnography (PSG) equipment, which is complex, disrupts sleep, and leads to poor user compliance. This non-invasive, non-disruptive monitoring method is particularly suitable for children who are sensitive to touch or the elderly who require long-term, regular monitoring.

[0098] 3. The monitoring system is robust and the data foundation is reliable: The technical solution of this application includes a complete and rigorous signal preprocessing process before extracting physiological features. Adaptive wavelet denoising filters out noise, and fast Fourier transform combined with constant false alarm rate detection automatically and accurately locates the chest position of the monitored subject. Phase unwrapping and differential analysis are then used to extract micro-motion signals with a high signal-to-noise ratio. This series of processes ensures that the data source for subsequent feature extraction and model analysis is pure and reliable, thereby improving the stability of the entire system and its adaptability to environmental changes.

[0099] 4. Intelligent alarm function and high practicality: The alarm function of this application's technical solution is not simply triggered, but rather incorporates a determination of the duration of apnea. Only when the duration of apnea exceeds a clinically significant preset threshold will the system trigger a preset alarm event. This design effectively filters out brief, harmless respiratory pauses, avoiding unnecessary false alarms that harass users or guardians, making the warning function more accurate, reliable, and user-friendly.

[0100] In addition, embodiments of the present invention further provide a computer-readable storage medium. The computer-readable storage medium can be any one of, or any combination of, a hard disk, a multimedia card, an SD card, a flash memory card, an SMC, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disk read-only memory (CD-ROM), or a USB memory device. The computer-readable storage medium includes a millimeter-wave radar-based sleep apnea type differentiation program 10. The specific implementation of the computer-readable storage medium of the present invention is substantially the same as the specific implementation of the millimeter-wave radar-based sleep apnea type differentiation method and server 1 described above, and will not be further described here.

[0101] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0102] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0103] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0104] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0105] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0106] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for distinguishing sleep apnea types based on millimeter wave radar, characterized in that: include: Acquire millimeter-wave radar echo signals of monitored objects within the monitoring area; The following physiological features are extracted from the echo signal: Body motion signal characteristics that characterize changes in the overall trunk motion state of the monitored subject; The energy characteristics of the chest cavity reflection signal that characterize the energy changes of the chest cavity movement of the monitored object, and A respiratory signal amplitude characteristic that characterizes changes in respiratory airflow of the monitored subject; Constructing a time-series feature sequence based on the extracted body motion signal features, the chest cavity reflection signal energy features, and the respiratory signal amplitude features; The feature sequence is input into a pre-trained classification model based on a self-attention mechanism for dynamic correlation analysis to determine the sleep apnea type of the monitored subject.

2. The method for distinguishing sleep apnea types based on millimeter wave radar according to claim 1, wherein: Before extracting three types of physiological features from the echo signal, the method further includes: Adopting adaptive wavelet to perform denoising on the echo signal; Perform fast Fourier transform on the denoised echo signal to obtain the range spectrum; Processing the range spectrum using constant false alarm rate detection to locate a target range unit corresponding to the chest area of ​​the monitored object; The phase information of the target distance unit is unwrapped and differentially processed to extract the micro-motion signal containing the physiological characteristics.

3. The method for distinguishing sleep apnea types based on millimeter wave radar according to claim 1, wherein: Extract body motion signal features that represent changes in the overall torso motion state of the monitored subject, including: Based on the frequency domain representation of the echo signal, the proportion of energy of the frequency band to which the body motion signal belongs to the total signal energy and the frequency range of the frequency band to which the body motion signal belongs are calculated, and the energy proportion and frequency range of the body motion signal are used together as the body motion signal features.

4. The method for distinguishing sleep apnea types based on millimeter wave radar according to claim 1, wherein: Extracting the energy features of the chest cavity reflection signal that characterize the energy changes of the chest cavity motion of the monitored object, including: The mean, standard deviation and kurtosis of the energy value sequence corresponding to the chest area of ​​the monitored object in the echo signal are calculated within a preset time window and are collectively used as the energy features of the chest reflection signal.

5. The method for distinguishing sleep apnea types based on millimeter wave radar according to claim 1, wherein: Extract the respiratory signal amplitude features that characterize the changes in the respiratory airflow of the monitored subject, including: performing bandpass filtering on the echo signal to obtain a respiratory signal waveform; The maximum amplitude, minimum amplitude and number of peaks of the respiratory signal waveform within a preset time window are extracted and used as the respiratory signal amplitude feature.

6. The method for distinguishing sleep apnea types based on millimeter-wave radar according to any one of claims 1 to 5, characterized in that: Construct a temporal feature sequence, including: splicing the body motion signal feature, the chest cavity reflection signal energy feature, and the respiratory signal amplitude feature into a multidimensional feature vector; Performing windowing processing on the multidimensional feature vectors arranged in chronological order using a preset sliding window to form an input sequence; Position encoding is performed on the input sequence to represent temporal information in the feature sequence.

7. The method for distinguishing sleep apnea types based on millimeter wave radar according to claim 6, wherein: The pre-trained classification model based on the self-attention mechanism is a Transformer network model; Analyzing the feature sequence based on the Transformer network model includes: In the multi-head self-attention layer, attention weights are calculated based on the feature sequence to capture the dynamic correlation between features in the sequence; In the feedforward network layer, a nonlinear transformation is performed on the output of the multi-head self-attention layer; In the classification layer, the Softmax function is used to normalize the results processed by the feedforward network layer to output probability values ​​corresponding to different sleep apnea types.

8. The method for distinguishing sleep apnea types based on millimeter wave radar according to claim 6, wherein: After determining the sleep apnea type of the monitored subject, the method further includes: When it is determined that the duration of the apnea of ​​the monitored object exceeds a preset time threshold, a preset alarm event is executed.

9. A device for distinguishing sleep apnea types based on millimeter-wave radar, characterized in that: The invention comprises a memory, a processor, and a millimeter-wave radar-based sleep apnea type differentiation program stored in the memory and executable on the processor, wherein when the processor executes the millimeter-wave radar-based sleep apnea type differentiation program, the millimeter-wave radar-based sleep apnea type differentiation method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a millimeter-wave radar-based sleep apnea type differentiation program, which, when executed by a processor, implements the millimeter-wave radar-based sleep apnea type differentiation method according to any one of claims 1 to 8.

Images