Foldable arc-shaped panel type sleep respiration monitoring device and method

Abstract

The invention belongs to the technical field of sleep health monitoring, non-contact vital sign perception and intelligent hardware, and discloses a foldable arc panel type sleep respiration monitoring device and method, which comprises a millimeter wave radar module, a programmable coding metasurface panel, a controller of the programmable coding metasurface panel, an exhaled plume sensing array, a processor, a memory and a control interface / bus, the bed head can be quickly fixed, unfolded for monitoring and folded for storage through the base assembly, the telescopic sliding rail assembly, the multi-degree-of-freedom rotating joint assembly and the inward folding arc-shaped panel assembly. The processor controls the metasurface to switch according to a codebook and to synchronously form a coding-observation sequence with a radar acquisition window, selects a coding subset to lock monitoring based on quality indexes such as a respiratory spectrum peak signal-to-noise ratio and spectrum peak stability, and triggers relocking scanning when the quality is reduced; when the electronic reconstruction still cannot restore mass, the slide rail and the joint are driven to perform a small-step automatic fine tuning search under mass gating conditions, and a user is allowed to manually perform coarse tuning.

Description

Technical Field

[0001] This invention belongs to the fields of sleep health monitoring, non-contact vital sign sensing and smart hardware technology, and particularly relates to a foldable arc-shaped panel-type sleep breathing monitoring device and method. Background Technology

[0002] Breathing during sleep is significantly affected by factors such as changes in sleeping position, obstruction by pillows and blankets, and changes in head and neck angle and distance. In existing non-contact monitoring solutions, the echo quality of single-view millimeter-wave radar is prone to fluctuation when the user is lying on their side, turning over, or deviating from the angle. This leads to a decrease in the effective monitoring duty cycle, unstable respiratory spectrum peaks, false alarms or missed alarms for abnormal events, and poor consistency of monitoring results across different sleeping positions.

[0003] In addition, many non-contact solutions rely on a single modal signal to infer respiratory status, lacking an independent chain of physical evidence to support the "existence / strength of exhaled airflow," which can easily lead to false alarms or missed alarms in complex sleeping positions and occluded environments.

[0004] Meanwhile, some solutions rely on fixed installations (such as ceiling or wall mounting) to obtain a relatively stable viewpoint, but such fixed deployments limit the monitoring location and make it difficult to meet the continuous monitoring needs of travel accommodation (hotels, business trips); if portable devices are used, without structural self-adaptation and stabilization strategies, the error of turning over or placing the device will still introduce large fluctuations.

[0005] Therefore, there is an urgent need for a sleep breathing monitoring device and method that does not require wearing, does not require frequent manual alignment by the user, can maintain a stable and effective signal under changing sleeping positions and occlusion environments, has the ability to fuse multiple pieces of evidence, and can be folded, stored, and deployed in a portable manner.

[0006] Based on the above analysis, the problems and shortcomings of the existing technology are as follows:

[0007] (1) In the existing non-contact monitoring scheme, the echo quality of single-view millimeter-wave radar is prone to fluctuation when the side-lying is blocked, the body turns over or the angle deviates, resulting in a decrease in effective monitoring duty cycle, unstable respiratory spectrum peaks, false alarms or missed alarms of abnormal events, and poor consistency of monitoring results in different sleeping positions.

[0008] (2) Many non-contact solutions rely on a single modal signal to infer the respiratory state, lacking an independent physical evidence chain to support the "existence / strength of exhaled airflow", which is prone to false alarms or missed alarms in complex sleeping positions and occlusion environments.

[0009] (3) Some solutions rely on fixed installation (e.g., ceiling or wall) to obtain a relatively stable viewpoint. However, such fixed deployments limit the monitoring location and make it difficult to meet the continuous monitoring needs of out-of-town accommodation (hotels, business trips). If portable devices are used, without structural self-adaptation and stabilization strategies, the error of turning over or placing the device will still introduce large fluctuations. Summary of the Invention

[0010] To address the problems existing in the prior art, the present invention provides a foldable arc-shaped panel-type sleep breathing monitoring device and method.

[0011] This invention is implemented as follows: a foldable arc-shaped panel-type sleep apnea monitoring device includes:

[0012] A) A millimeter-wave radar module, used to collect echo signals from the human chest and neck region during sleep monitoring. The millimeter-wave radar module operates in the frequency band of 24GHz to 81GHz.

[0013] B) A programmable coded metasurface panel, consisting of an array of multiple controllable units, each controllable unit having at least two discrete electromagnetic response states;

[0014] C) A metasurface controller, connected to the programmable coded metasurface panel, for switching the discrete electromagnetic response state of the controllable unit between at least K coded states according to a preset codebook, where K≥16;

[0015] D) Exhaled plume flow sensor array, comprising at least 4 plume flow sensor nodes, wherein the plume flow sensor nodes include a carbon dioxide sensor and / or a humidity sensor and / or a temperature sensor.

[0016] E) A processor, a memory, and a control interface or bus, wherein the memory stores program instructions executable by the processor;

[0017] F) A portable, retractable, foldable mechanism for mounting and positioning the millimeter-wave radar module, the programmable coded metasurface panel, and the exhaled plume sensor array near the bedside, the portable, retractable, foldable mechanism comprising:

[0018] F1) Base assembly, configured to be fixed to the headboard, bed frame or bedside table by clamping and / or magnetic structures;

[0019] F2) A retractable slide rail assembly, connected to the base assembly, configured to adjust the monitoring height and having a locking mechanism;

[0020] F3) A multi-degree-of-freedom rotary joint assembly, connected to the telescopic slide rail assembly, provides at least pitch and yaw rotational degrees of freedom;

[0021] F4) An inwardly foldable arc-shaped panel assembly connected to the multi-degree-of-freedom rotary joint assembly, the arc-shaped panel assembly including a middle section and at least one wing section, the wing section being connected to the middle section via a folding hinge and being foldable inward to form a stowed state;

[0022] The millimeter-wave radar module, the programmable coded metasurface panel, and the exhaled plume flow sensor array are disposed on the middle section and / or wing section of the arc-shaped panel assembly.

[0023] The processor is configured to execute:

[0024] (1) Control the metasurface controller to switch to the i-th encoding state according to the codebook, and after confirming the completion of the encoding state switch, trigger the millimeter-wave radar module to collect echo signals in the acquisition window corresponding to the encoding state;

[0025] (2) Calculate respiratory micro-motion quality index for the echo signal corresponding to at least some of the encoded states, wherein the respiratory micro-motion quality index includes at least the respiratory peak signal-to-noise ratio index and the peak stability index;

[0026] (3) Select a coded subset as a locked state based on the respiratory micro-movement quality index, perform continuous monitoring in the locked state, and use a threshold hysteresis strategy to trigger a relock scan to update the coded subset when the locked quality decreases.

[0027] (4) Extract breathing effort related parameters from the echo signal and exhaled airflow related parameters from the exhaled feather sensor array, and fuse the breathing effort related parameters and the exhaled airflow related parameters to generate a sleep breathing abnormality event indication signal and / or risk warning signal for output.

[0028] (5) When the overall quality index is lower than the first threshold T_low and the duration is not less than t_low, and the overall quality index still does not meet the holding condition after the relock scan, the retractable slide rail assembly and / or multi-degree-of-freedom rotary joint assembly are driven to perform small-step automatic fine-tuning search under quality gating conditions so that the overall quality index is improved and enters the locked state; and the user is allowed to manually coarsely adjust the retractable slide rail assembly and / or multi-degree-of-freedom rotary joint assembly.

[0029] Furthermore, the confirmation of the completion of the encoding state switch includes the metasurface controller outputting a ready indication signal, and the processor triggering millimeter-wave radar acquisition after a delay of Δt after receiving the ready indication signal, where Δt is 0.5ms to 20ms;

[0030] The threshold hysteresis strategy includes a first threshold T_low and a second threshold T_high, where T_high > T_low; when the overall quality index is lower than T_low for a duration not less than t_low, a re-lock scan is triggered; when the overall quality index is higher than T_high for a duration not less than t_high, the lock is maintained.

[0031] t_low is 0.5s to 10s, and t_high is 1s to 30s.

[0032] Furthermore, the codebook is a low-correlation codebook, where the absolute value of the normalized cross-correlation coefficient of the configuration vectors of any two encoding states is no greater than 0.3;

[0033] The respiratory spectrum peak signal-to-noise ratio index is calculated based on the ratio of the main peak amplitude to the noise floor within the 0.08Hz to 0.6Hz respiratory frequency band;

[0034] The peak stability index is calculated based on the variance of the respiratory peak frequency and / or amplitude within N consecutive time windows, where N≥3 and the time window length is 5s~30s;

[0035] Before calculating the respiratory micro-motion quality index, the processor performs distance gating on the echo signal. The gating center distance is determined by the human body principal reflection distance obtained from the initial scan, and the gating width is ±0.1m to ±0.6m.

[0036] Furthermore, when the encoded subset contains multiple encoded states, the processor switches between the multiple encoded states in a polling manner according to the duty cycle to obtain multipath diversity observations.

[0037] Furthermore, the re-locking scan is a two-level scan, with the first-level scan determining the candidate code set and the second-level scan determining the code subset from the candidate code set; the size of the candidate code set is 4 to 32.

[0038] The spacing between adjacent nodes of the exhaled feather flow sensor array is 5mm to 30mm, and the distance from the array plane to the mouth and nose area is 50mm to 400mm.

[0039] The processor performs environmental baseline adaptive correction and condensation interference suppression on the signals from the exhaled plume sensor array;

[0040] The fused sleep breathing abnormality event indication signal includes at least a pause event indication and / or a hypopnea event indication and / or a breathing irregularity event indication;

[0041] The retractable slide rail assembly is a two-stage or three-stage sleeve telescopic structure or a screw / rack drive structure, and is equipped with anti-fall-back locking and / or travel limit.

[0042] Furthermore, the multi-degree-of-freedom rotary joint assembly is provided with a limiting structure and / or a force-limiting clutch structure;

[0043] The arc-shaped panel assembly includes two wing segments, left and right, which are connected to the middle segment via folding hinges and can be folded inward to cover the middle segment.

[0044] Another object of the present invention is to provide a foldable arc-shaped panel-type sleep apnea monitoring method comprising:

[0045] S1) Switch the programmable coded metasurface to the i-th coded state according to the codebook, and trigger the millimeter-wave radar to collect echoes in the corresponding acquisition window after confirming the completion of the switch;

[0046] S2) Calculate the respiratory micromotion quality index and select a coding subset for locking;

[0047] S3) A threshold hysteresis strategy is used to trigger a relock scan to update the encoded subset when the locking quality deteriorates;

[0048] S4) Extract breathing effort-related parameters and exhaled airflow-related parameters and fuse them to output abnormal event indication signals and / or risk warning signals;

[0049] S5) When the comprehensive quality index is lower than T_low and remains not less than t_low, and the holding condition is not met after relocking, the telescopic slide rail assembly and / or multi-degree-of-freedom rotary joint assembly are driven to perform small-step automatic fine-tuning search and enter the lock under quality gating conditions.

[0050] Another object of the present invention is to provide a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the foldable arc-shaped panel sleep breathing monitoring method.

[0051] Another object of the present invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the foldable arc-shaped panel-type sleep breathing monitoring method.

[0052] Another objective of the present invention is to provide an information data processing terminal for implementing the foldable arc-shaped panel-type sleep breathing monitoring device.

[0053] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0054] By using a programmable coded metasurface to achieve electronic reconstruction of observation conditions, virtual multi-view and multi-path diversity observations can be obtained to improve echo quality and stability under changing sleeping positions and occlusion environments.

[0055] Exhaled airflow parameters are provided by an exhaled feather sensor array as an independent chain of evidence, and are fused with breathing effort parameters extracted by millimeter-wave radar to generate abnormal event indications and risk warnings, thereby improving the reliability of the judgment.

[0056] The portable, foldable curved panel structure (headboard fixed base, retractable slide rail, multi-degree-of-freedom rotating joint and inward foldable curved panel) enables rapid deployment and storage. When electronic reconstruction is insufficient to restore quality, the slide rail / joint is driven to perform small-step automatic fine-tuning under quality gating conditions to search for the best monitoring point, thereby improving data consistency and effective monitoring duty cycle.

[0057] By using a programmable coded metasurface to reconstruct electronic observation conditions and achieve multipath diversity, the echo quality and stability can be improved under conditions of changing sleeping posture and occlusion.

[0058] By using a quality-indicator-driven lock / relock closed loop and threshold hysteresis strategy, we can improve the effective monitoring of duty cycle and avoid frequent switching caused by short-term overturning.

[0059] By providing an independent chain of evidence for airflow through an exhaled plume sensor array and fusing it with radar effort information, the reliability of anomaly event detection is improved.

[0060] Its portable, foldable curved panel structure enables rapid deployment and storage, making it suitable for home and travel accommodation scenarios.

[0061] When electronic reconstruction is insufficient to restore observation quality, the slide rail / joint is driven to perform small-step automatic fine-tuning under quality gating conditions. This can reduce the precipitous quality drop caused by flipping / placement errors and improve the consistency of monitoring in different attitude segments. It also supports manual coarse adjustment by the user, taking into account portability, quietness and safety.

[0062] (1) The expected benefits and commercial value of the technical solution of this invention after transformation are as follows:

[0063] This invention enables home-based sleep apnea monitoring in a non-contact manner, reducing discomfort and reliance on consumables associated with adhesive / tube-based data collection and improving user compliance. Simultaneously, the foldable curved panel and bedside mounting base facilitate rapid deployment and storage, reducing packaging, transportation, and on-site installation costs. It is suitable for applications in homes, primary healthcare, and sleep management services, thus demonstrating promising prospects for productization and large-scale application.

[0064] (2) The technical solution of this invention fills a technical gap in the industry both domestically and internationally:

[0065] For the multimodal integrated home monitoring architecture of "foldable curved panel + programmable coded metasurface electronic reconfiguration + exhaled plume sensor array (optional BCG)", existing technologies mostly focus on single radar monitoring or single airflow / body motion monitoring, and lack a systematic solution that can achieve continuous locking and relocking in dynamic scenarios such as side lying, turning over, and occlusion. This invention provides a device structure and method for the above scenarios, which is expected to fill the gap in portable deployment and stable monitoring in this field.

[0066] (3) The technical solution of the present invention solves a technical problem that people have long wanted to solve but have never been able to solve successfully:

[0067] Home-based non-contact monitoring is prone to echo degradation, data incompleteness, and event misjudgment under conditions such as changes in sleeping posture, obstruction by blankets, and deviation in installation angle, making it difficult to continuously acquire usable respiratory information over a long period. This invention achieves dynamic compensation for observation conditions through codebook scanning and quality index evaluation, combined with lock / relock closed-loop control and (optional) silent gating with small-step fine-tuning; and reduces false alarms and missed alarms through "breathing effort-airflow parameter" fusion discrimination, thereby improving the continuous effective monitoring duration and the reliability of abnormal event discrimination.

[0068] (4) The technical solution of the present invention overcomes technical bias:

[0069] Addressing the preconceived notions in existing technologies that "non-contact monitoring is difficult to obtain stable and usable respiratory parameters in scenarios such as side-lying, turning over, and occlusion" and "it is difficult to distinguish between effort and airflow abnormalities by relying solely on a single sensor source," this invention maintains echo quality through encoded metasurface electronic reconstruction and quality-gated closed-loop control, and introduces an exhaled plume sensor array as independent evidence for fusion discrimination. This enables stable abnormal indications and risk warnings to be obtained even in the aforementioned complex scenarios, thereby overcoming related technical biases. Attached Figure Description

[0070] Figure 1 This is a structural block diagram of the foldable arc-shaped panel-type sleep breathing monitoring device provided in an embodiment of the present invention.

[0071] Figure 2 This is a schematic diagram of the timing of coded metasurface codebook switching and radar synchronous acquisition provided in an embodiment of the present invention.

[0072] Figure 3 This is a flowchart of the quality index calculation and locking / relocking closed loop provided in the embodiments of the present invention.

[0073] Figure 4 This is a schematic diagram of the exhaled feather flow sensor array layout provided in an embodiment of the present invention.

[0074] Figure 5This is a schematic diagram of the process for generating abnormal event indications and risk warnings by fusing effort parameters and airflow parameters according to an embodiment of the present invention.

[0075] Figure 6 This is a schematic diagram of the structure of the portable foldable curved panel provided in an embodiment of the present invention.

[0076] Figure 7 This is a flowchart of the foldable arc-shaped panel-type sleep breathing monitoring method provided in an embodiment of the present invention.

[0077] Figure 8 This is a comparison chart of the signal-to-noise ratio of respiratory signals under different lateral lying angles according to an embodiment of the present invention.

[0078] Figure 9 This is a comparison chart of the respiratory event detection performance of embodiments of the present invention.

[0079] Figure 10 This is a schematic diagram of signal quality recovery during the automatic fine-tuning process according to an embodiment of the present invention.

[0080] In the diagram: 1. Millimeter-wave radar module; 2. Programmable coded metasurface panel; 3. Metasurface controller; 4. Exhaled plume flow sensor array; 5. Processor; 6. Memory; 7. Control interface / bus; 8. Mattress BCG sensor module (optional); 9. Base assembly; 10. Clamping structure; 11. Magnetic base (optional); 12. Telescopic slide rail assembly; 13. Locking mechanism; 14. Multi-degree-of-freedom rotary joint assembly; 15. Arc-shaped panel middle section; 16. Arc-shaped panel wing section; 17. Folding hinge; 18. Anti-condensation assembly; 19. Battery; 20. Cable routing channel. Detailed Implementation

[0081] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0082] Example 1: Device Hardware Structure and System Connection

[0083] like Figure 1 and Figure 6 As shown, the present invention provides a foldable arc-shaped panel-type sleep breathing monitoring device, the hardware system of which mainly includes a sensing module, a control module and a mechanical structure module.

[0084] 1. Sensing Module:

[0085] Millimeter-wave radar module 1: The operating frequency band is 24GHz to 81GHz. It is used to transmit electromagnetic waves to the chest and neck area of ​​the human body and receive their echoes in order to detect micro-movements caused by breathing.

[0086] Programmable coded metasurface panel 2: Composed of an array of M×N controllable units, each unit can switch between at least two discrete electromagnetic response states (such as "0" and "1") under the control of metasurface controller 3. By configuring different coding sequences, the pointing, shape, and focusing characteristics of the radar beam can be dynamically changed.

[0087] Exhaled plume sensor array 4: consists of at least 4 sensor nodes arranged at a spacing of 5mm to 30mm. Each node integrates one or more of carbon dioxide (CO2), humidity and temperature sensors to capture the physicochemical characteristics of exhaled airflow from the mouth and nose at close range.

[0088] (Optional) Mattress BCG Sensing Module 8: Based on piezoelectric film or distributed pressure sensors, it is laid under the mattress to acquire body movement (Ballistocardiogram, BCG) signals caused by heartbeat and breathing, as an auxiliary and backup signal source.

[0089] 2. Control Module

[0090] Processor 5 and memory 6: Processor 5 is the core computing and control unit of the device, responsible for running all algorithm processes of the present invention; memory 6 is used to store preset low correlation codebooks (the absolute value of the normalized cross-correlation coefficient of any two codewords is ≤0.3), algorithm parameters (such as thresholds T_low and T_high), control programs and historical data.

[0091] Metasurface controller 3: Receives the encoded index command issued by processor 5 and drives the unit array of metasurface panel 2 to switch to the corresponding state.

[0092] Control Interface / Bus 7: Provides power and data communication channels for each module.

[0093] 3. Portable foldable mechanical structure module:

[0094] This module is key to enabling the device to be portable, rapidly deployed, and adaptively adjusted. Figure 6 As shown, it includes:

[0095] Base assembly 9: Securely attaches the device to the headboard, bed frame or bedside table via clamping structure 10 (such as spring clips and / or optional magnetic base 11).

[0096] The retractable slide rail assembly 12 and locking mechanism 13 connect the base and the upper structure, allowing for vertical height adjustment. After adjustment, the locking mechanism 13 secures the assembly, ensuring no slippage during monitoring. Its stroke design ensures that the exhaled plume array 4 can be adjusted to an optimal range of 50mm to 400mm from the user's mouth and nose.

[0097] Multi-DOF rotary joint assembly 14: Provides at least two rotational degrees of freedom, pitch and yaw, allowing the upper panel to be aligned with the user's chest and neck area. The joint has built-in limit and force-limiting clutches to prevent hard collisions with people or furniture.

[0098] The inwardly foldable arc-shaped panel assembly consists of a central arc-shaped panel 15 and two left and right arc-shaped panel wings 16 connected by folding hinges 17. The wings 16 can fold inward to cover the central section 15, achieving compact storage. The millimeter-wave radar module 1, the programmable coded metasurface panel 2, and the exhaled plume flow sensor array 4 are all integrated on this arc-shaped panel. The lower edge of the panel is also equipped with an anti-condensation component 18, such as a hydrophobic coating or a micro heating film, to reduce the interference of condensation on airflow sensing.

[0099] Each module is connected by cables within the wiring channel 20 and is powered by the built-in battery 19 or an external power source.

[0100] Example 2: Initial Deployment, Self-calibration, and Encoding Status Scan Locking Process

[0101] Upon first use, the user should clamp the device to the headboard using the base 9, manually unfold the curved panel, and roughly orient it. After starting the device, the system will execute the following automatic initialization procedure:

[0102] Step S21: Initial Spatial Pose Optimization. Processor 5 controls the metasurface to perform a rapid full codebook scan (e.g., traversing 32 states) and simultaneously triggers radar acquisition. By analyzing the echoes, the initial distance R0 of the human body's primary reflector is determined (e.g., 1.5 meters). Simultaneously, the processor evaluates the current airflow signal strength. If the calculation finds the array is too far from the user, the slide rail 12 is automatically driven forward for fine adjustments (e.g., 5mm at a time) under quality gating until the combined radar signal strength and airflow signal strength reach their optimal values, completing the deployment self-calibration.

[0103] Step S22: Encoding status scanning and synchronous acquisition. For example... Figure 2 As shown in the timing diagram, the processor 5 sends the encoding index command to the metasurface controller 3 in sequence according to the preset low correlation codebook; after the metasurface controller 3 completes the corresponding encoding state switch, it returns a switch completion confirmation to the processor 5; after confirmation, the processor 5 triggers the millimeter-wave radar module 1 to perform data acquisition in the acquisition window W_k corresponding to the i-th encoding state after a delay Δt, and obtains the echo data D_k, thereby constructing a sequence of one-to-one correspondence between "encoding state - observation data".

[0104] Step S23: Calculation and initial locking of respiratory micro-motion quality index. Processor 5 processes the echo data D_k obtained in step S22 for each coded state. First, a distance gate with a width of ±(0.1m~0.6m) is applied with the initial distance R0 as the center to filter out irrelevant environmental clutter. Then, time-frequency analysis (such as short-time Fourier transform) is performed on the gated signal, and the main respiratory spectrum peak is extracted within each time window (such as 30 seconds) in the respiratory frequency band of 0.08Hz~0.6Hz.

[0105] Breathing spectrum peak signal-to-noise ratio (SNR): Calculated as the ratio (dB) of the amplitude of the main peak to the average amplitude of the noise floor within the frequency band.

[0106] Peak stability index: Calculate the variance of the main respiratory peak frequency value within N consecutive time windows (N≥3).

[0107] Processor 5 combines the above two metrics to score each coding state. One or more states with the highest scores are selected to form an initial locked subset. If the locked subset contains multiple states, the system will perform polling observations according to a set duty cycle to achieve multipath diversity and enhance robustness to local occlusion.

[0108] Example 3: Dynamic anti-interference monitoring based on threshold hysteresis and two-stage relocking

[0109] During continuous monitoring, the system dynamically maintains signal quality through closed-loop control.

[0110] Step S31: Threshold Hysteresis Monitoring. Processor 5 calculates the comprehensive quality index Q (weighted by SNR and stability index) in real time under the current locked state. The system presets two thresholds: a relock trigger threshold T_low (e.g., 0.4) and a high quality maintenance threshold T_high (e.g., 0.7), and T_high > T_low. Simultaneously, time conditions are set: duration t_low (e.g., 5 seconds) and t_high (e.g., 10 seconds).

[0111] The comprehensive quality index Q can be calculated as follows: Q = w1·Norm(SNR) + w2·Norm(1 / (Var_f+ε)); where Var_f is the variance of the peak respiratory frequency over N consecutive time windows, ε is a constant to prevent division by zero, w1+w2=1, and Norm(·) is a monotonically normalized function that maps the index to the interval [0,1].

[0112] Step S32: Quality Degradation and Relock Trigger. When a user rolls over, causing occlusion and the overall quality index Q to remain below T_low until it reaches t_low, the system determines that the current lock state has failed and automatically triggers the relock scan process.

[0113] Step S33: Two-stage relocking scan. To balance speed and accuracy, a two-stage relocking mechanism is used:

[0114] Level 1 Scan (Rapid Coarse Screening): Within 1-2 seconds, quickly traverse all available coded states (e.g., 32 states), collecting only short-term data (e.g., 1 second), and quickly calculate the preliminary quality score for each state. Filter out a set of candidate states (ideally 4-16 states) whose scores exceed a certain threshold (e.g., Q>0.5).

[0115] Secondary scanning (refined evaluation): For each state in the candidate set, continuous observations are performed for a longer period (e.g., 3-5 seconds) to calculate a more stable and reliable average quality index. Finally, the state with the highest index (or the top few states) is selected as the new locked subset.

[0116] Step S34: Lock Recovery. The system switches to the new locked subset for monitoring, and the overall quality index Q rises accordingly. When Q remains above T_high and reaches t_high, the system considers itself to have entered a stable, high-quality monitoring phase and maintains the current lock. This closed-loop mechanism ensures that when the user routinely turns over (within 30-60 degrees), the system can restore stable monitoring within 10-20 seconds through pure electronic reconstruction, such as... Figure 8 The experimental data show that monitoring interruptions were effectively avoided.

[0117] Example 4: Robust processing and feature extraction of exhaled plume signals

[0118] To reliably extract respiratory airflow evidence from complex environments, the data from the exhaled plume sensor array 4 were specifically processed.

[0119] Step S41: Environmental baseline adaptive correction. Since indoor temperature, humidity, and CO2 background values ​​change slowly, processor 5 maintains a dynamic background baseline (such as the moving average over the past 10 minutes) for each sensor node. The real-time signal value is subtracted from its corresponding instantaneous baseline value to obtain the net change signal purely caused by respiratory events.

[0120] Step S42: Condensation Interference Detection and Suppression. In cold environments, warm, humid exhaled air may condense on the sensor surface. Processor 5 has dedicated judgment logic: if the reading of a humidity sensor node jumps to a saturation value (e.g., 99%RH) and remains there within a very short time (<0.1 seconds), accompanied by a slight positive temperature change, it is determined to be a condensation event.

[0121] Suppression strategy: Within a time window (e.g., 5 seconds) after a condensation event occurs, significantly reduce or remove the fusion weight of the node's data. At the same time, rely on other unaffected nodes in the array or use spatial interpolation algorithms to maintain the continuity of the airflow signal, thereby effectively avoiding false "airflow disappearance" misjudgments caused by condensation.

[0122] Step S43: Exhaled airflow feature parameter extraction. From the calibrated and purified array signal, the following features are extracted as exhaled airflow related parameters:

[0123] CO2 fluctuation range: the difference between the peak and trough of CO2 concentration during each respiratory cycle.

[0124] Humidity rise slope: The average rate of change of the rising segment of the humidity curve at the start of exhalation.

[0125] Array spatial gradient consistency: Analyze the spatial distribution pattern of signal changes at each node at the same moment. The gradient direction should roughly point to the airflow source (mouth and nose). This indicator is used to verify the authenticity of the signal source.

[0126] Example 5: Dual-modal information fusion for identifying sleep apnea events

[0127] The core advantage of this invention lies in its integration of two independent types of physical evidence: "breathing effort" (radar) and "exhaled airflow" (feather array). Processor 5 executes the following fusion discrimination logic:

[0128] Step S51: Time Alignment and Parameter Extraction. The respiratory effort-related parameters (including respiratory cycle, chest and abdominal movement amplitude, and respiratory rhythm disorder index) extracted in real time from the radar echo are precisely synchronized with the exhaled airflow-related parameters extracted from the plume array.

[0129] Step S52: Rule-based fusion judgment. Processor 5 performs real-time analysis based on a preset expert rule base, as shown in the following example:

[0130] Rule A (Detection of Apnea): If the radar signal shows that the periodic chest and abdominal movements lasting more than 10 seconds have significantly weakened or disappeared (abnormal effort), and at the same time the plume array signal shows that the periodic fluctuations of CO2 and humidity have also completely disappeared (disappearance of airflow), then an "apnea event" is determined to have occurred, and a high-confidence alarm is output.

[0131] Rule B (Identifying Hypoventilation): If the radar signal shows that the respiratory effort amplitude has decreased by more than 50% of the baseline value but has not completely disappeared, and at the same time the plume array signal shows that the CO2 fluctuation amplitude of the exhaled airflow has decreased by more than 30% synchronously, and the duration exceeds 10 seconds, then it is determined as a "hypoventilation event".

[0132] Rule C (Interference Elimination): If the radar signal is temporarily disrupted or disappears due to the user turning over significantly, but the plume array continues to detect regular and normal-amplitude airflow fluctuations, the system will mark this period as "motion interference," without triggering a breathing abnormality alarm, and will only record it.

[0133] Fusion discrimination utilizes the breathing effort characteristics of millimeter-wave radar and the airflow characteristics of plume arrays for cross-verification, which can reduce false alarms and false negatives caused by interference such as turning over and obstruction (see [link]). Figure 9 (Example comparison).

[0134] Example 6: Cascaded Adaptive: Search and Recovery from Electronic Reconfiguration to Mechanical Fine-tuning

[0135] When the user's body position changes drastically (e.g., lying on their side at an angle greater than 60 degrees or facing away from the device), exceeding the capability of metasurface electronic beam reconstruction, the system activates a second-layer protection mechanism.

[0136] Step S61: Electronic Reconstruction Failure Determination. When the overall quality index Q is consistently lower than T_low due to extreme body position, and after the relocking scan described in Example 3, the quality scores of all coding states still cannot restore Q to above T_high (i.e., Q_max < T_high), the system determines that the pure electronic reconstruction has failed.

[0137] The quality gating conditions may include one or more of the following: the safe distance in front of the device is greater than the preset threshold d_safe; the user is in a relatively static state or not in a state of significant turning over; the drive noise and drive speed meet the nighttime quiet limit; the power / power supply meets the drive conditions; the joints and slide rails have not triggered the travel limit or overload protection.

[0138] Step S62: Mechanical automatic fine-tuning search under quality gating. The processor 5) starts the preset search mode under safe conditions (such as detecting no obstacles ahead). It controls the multi-degree-of-freedom rotary joint 14) and the telescopic slide rail 12) to perform small-step search in a limited "search space" (such as yaw ±30°, pitch ±15°, depth ±5cm) according to the sequence of "angle first, then depth".

[0139] Step S63: Search-Evaluate-Lock Loop. Each time the machine completes a small adjustment (e.g., a 5° deflection), the system pauses, and the metasurface and radar perform a rapid codebook scan and quality assessment. For example... Figure 10 The timeline shown demonstrates a clear step-like increase in the overall quality index Q. Each "step" represents a successful pose adjustment.

[0140] Step S64: Restore Locking. When the evaluation quality Q in a certain pose reaches and stabilizes under the locking condition, the processor 5) stops the mechanical search and locks the slide rail and joint; then the pose is maintained and monitoring continues using the locking / relocking strategy described in Example 3 to improve monitoring continuity.

[0141] Example 7: Experimental Verification: Quantitative Comparison of Key Performance

[0142] To objectively demonstrate the technical effectiveness of the present invention, a series of comparative experiments were conducted.

[0143] Experiment 7.1: Verification of resistance to lateral obstruction. The respiratory spectrum peak SNR was compared between the fixed-beam radar and the device of this invention at different angles from supine to lateral. The results show that the SNR of the device of this invention decreases more slowly as the lateral angle increases, and is generally higher than that of the control group (specific values ​​depend on prototype parameters, placement distance, and environment).

[0144] Experiment 7.2: Event Detection Validation. Using PSG-annotated data as a reference, the detection metrics of radar-only, airflow-only, and fusion algorithms were compared. Example results show that the fusion algorithm outperforms the single-mode scheme overall in terms of sensitivity, specificity, and F1 score (see [link to experiment 7.2]). Figure 9 Example).

[0145] Experiment 7.3: Demonstration of Automatic Mechanical Fine-Tuning Recovery. Simulating a scenario where extreme body positions cause electronic reconstruction to fail, the system triggers a small mechanical step search and recovers to the monitored pose that meets the lock-and-hold conditions within tens of seconds (see [link to experiment 7.3]). Figure 10 Example).

[0146] Example 8: Long-term home application and health trend management

[0147] It can be used continuously in a home environment (e.g., for several weeks). Processor 5 records the respiratory rate, respiratory amplitude index, and abnormal event count each night, and can summarize the data into trend information by time period for users to view or for use by the upper-level health management system.

[0148] Optionally, if the device is equipped with a mattress BCG module 8, breathing-related body motion features can be further extracted to assist in triggering relocking or improving continuity when the echo quality is poor.

[0149] To verify the effectiveness of the device and method described in this invention, software simulation and prototype testing were conducted, and a comparative analysis was performed with traditional single-view millimeter-wave solutions from key performance dimensions.

[0150] Experiment 1: Verification of the effect of programmable coded metasurfaces on resisting sleep posture occlusion

[0151] Experimental Objective: A common challenge in sleep monitoring is that users turning over while lying on their side can obstruct or deviate from the radar's main beam, affecting the chest area. This experiment aims to quantitatively verify the programmable coded metasurface dynamic beam scanning and locking mechanism employed in this invention, demonstrating a fundamental improvement in respiratory signal quality acquisition compared to traditional fixed-beam radar in a simulated side-lying scenario.

[0152] Figure 8As shown, this invention constructs a control test environment. The control group uses a single fixed-beam millimeter-wave radar with its beam direction fixedly pointing towards a preset chest target. The experimental group uses a programmable coded metasurface loaded on the front end of the same radar and runs the codebook scanning and quality-driven locking algorithm described in this invention. By gradually changing the target angle to simulate the process of a human body moving from a supine (0°) to a large lateral position (60°), the signal-to-noise ratio (SNR) of the simulated respiratory signal extracted from the target reflection echo at each angle is recorded.

[0153] II. Experiment 2: Accuracy Comparison of Multimodal (Radar + Airflow) Fusion vs. Single-Modal Event Detection

[0154] Experimental Objective: The essence of sleep apnea is "breathing effort exists, but airflow stops." Detection based solely on chest micro-movements (effort) can easily misinterpret stillness as apnea; detection based solely on long-distance airflow can lead to missed detections due to airflow diffusion or environmental interference. This experiment aims to verify the necessity of the proposed dual-evidence fusion method of "radar effort signal + near-range airflow array signal" and its significant effect on improving key indicators in medical testing.

[0155] like Figure 9 As shown, this invention collects simulated sleep data containing known respiratory events (annotated by a professional polysomnography (PSG) scanner). Then, three algorithms are used to perform event detection analysis on the same set of raw sensor data:

[0156] 1. Algorithm A (Single-mode-radar): Only analyzes the chest and abdominal motion amplitude of millimeter-wave radar.

[0157] 2. Algorithm B (Single-Mode-Airflow): Only analyzes the periodic fluctuations of carbon dioxide concentration in the exhaled plume array.

[0158] 3. Algorithm C (Invention - Fusion): Executes the fusion logic in the claims of this invention. For example, when the radar detects the disappearance or significant weakening of periodic chest and abdominal movements, and the airflow array also detects the disappearance or weakening of the corresponding expiratory airflow, it is determined to be a valid "apnea" event.

[0159] Sensitivity: refers to the ability to correctly identify real patients (without missing diagnoses). Algorithm C (fusion) has a sensitivity of 94.3%, significantly higher than the two single-modal algorithms. This means that for real sleep apnea events, the fusion algorithm has the lowest possibility of missing reports.

[0160] Specificity: refers to the ability to correctly identify healthy individuals (without misdiagnosis). Algorithm C also achieved the highest specificity of 96.5%, indicating that it is least likely to misjudge normal actions such as rolling over or holding one's breath as abnormal events.

[0161] F1 score: This is a single metric that comprehensively balances sensitivity and specificity; a higher score indicates better overall performance. Algorithm C has an F1 score of 0.95, which is more than 10 percentage points higher than the best-performing single-modal algorithm (Algorithm B, 0.85).

[0162] The charts and data convey a clear conclusion: perception of a single physical modality has insurmountable limitations and is susceptible to interference, leading to misjudgments or missed detections. This invention introduces an independent chain of exhaled airflow evidence and performs spatiotemporal correlation and cross-validation with breathing effort signals, achieving a mechanism similar to "double verification." This is not merely a technological advancement, but a revolutionary change in judgment logic, upgrading the assessment of sleep apnea events from "indirect speculation" to "direct confirmation," thereby elevating detection reliability to a level suitable for serious home medical scenarios.

[0163] III. Experiment 3: Demonstration of the effectiveness of the automated mechanical fine-tuning search process

[0164] Experimental Objective: The ability of electronic beam reconfiguration has physical limits (e.g., it cannot penetrate thick obstructions or handle 180° turns). This experiment aims to demonstrate how the second-layer protection mechanism of this invention—quality-driven mechanical automatic fine-tuning—works when electronic adaptive methods fail, and to prove that it can automatically and intelligently recover the system from an "unlocked" state, ensuring the continuity of monitoring.

[0165] like Figure 10 As shown, this embodiment of the invention simulates an extreme change in sleep position: the user suddenly and dramatically turns over from an initial ideal supine position (where the overall quality index Q value is close to 1), not only lying on their side but also moving their body away from the device, causing the radar main beam and all electronically adjustable "virtual perspectives" to be unable to effectively capture respiratory micro-movements (Q value drops sharply to below 0.2). The system initiates an emergency recovery process according to preset logic.

[0166] 1. Quality collapse (t1): The Q value curve shows a precipitous drop, indicating that the current monitoring point has failed.

[0167] 2. Electronic relocking attempt (t1-t2): The system first attempts recovery in a purely electronic manner, i.e., triggering a rescan of the metasurface codebook. However, due to the excessively severe positional deviation, none of the encoded states were able to restore the Q value to an acceptable threshold (red dashed line in the figure, T_high=0.7), and the attempt failed.

[0168] 3. Mechanical Fine-tuning Search Initiation (t3): After electronic means fail, the processor, under quality gating conditions, issues a command to drive the retractable slide rail and multi-degree-of-freedom rotary joint to begin coordinated movement. The gray background area in the figure represents the mechanical fine-tuning in progress.

[0169] 4. Step-like recovery (t3-t5): This is the most representative finding of this experiment. The mechanical adjustments are not blind movements, but rather employ a closed-loop search strategy of "small steps-acquisition-evaluation". The graph shows that the Q-value curve exhibits a clear "step-like" upward trend during this stage. Each "step" represents a tiny mechanical pose adjustment followed by a brief data acquisition and calculation of the new Q-value. This indicates that the system is intelligently exploring the surrounding space and perceiving the quality changes brought about by each adjustment in real time.

[0170] 5. Relocking (t6 and beyond): After several steps of exploration, the system successfully finds a new spatial pose (a new combination of height and angle) that allows the Q value to jump and stabilize above 0.75. Subsequently, mechanical movement stops and locks, and the system re-enters a high-quality, stable locking and monitoring state at this new location.

[0171] This timing diagram vividly illustrates the cascaded adaptive concept of "electronic priority, mechanical backup" in this invention. It demonstrates that the device is not a passive sensor, but an active intelligent system with "environmental perception-decision-action" capabilities. When the primary sensing path is blocked, it can autonomously activate a backup plan, re-establishing optimal monitoring conditions through precise mechanical adjustments. This ensures maximum effective monitoring duty cycle during sleep periods lasting several hours, fundamentally solving the industry problem of data interruption caused by user movement in portable devices.

[0172] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0173] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A sleep breathing monitoring device, characterized in that, Comprising: a millimeter wave radar module, a programmable coded metasurface panel, a metasurface controller, a processor, a memory, an exhalation plume sensing array; The processor is configured to perform the following mechanism coordination process: controlling the programmable coded metasurface to switch between multiple coded states, and after confirming the completion of the corresponding coded state switching, controlling the millimeter wave radar module to collect human body echo signals in the collection window corresponding to the coded state; calculating a respiratory micro-motion quality index for the echo signals corresponding to at least part of the coded states, the respiratory micro-motion quality index including a respiratory spectrum peak signal-to-noise ratio index and a spectrum peak stability index; selecting a coded subset as a lock state based on the respiratory micro-motion quality index, and continuously monitoring in the lock state, and performing a re-scan to update the coded subset when the quality index falls below a re-lock trigger condition; extracting respiratory effort-related parameters from the echo signals and exhalation flow-related parameters from the exhalation plume sensing array, and fusing the two types of parameters to generate a sleep respiratory abnormality event indication signal output.

2. The apparatus of claim 1, wherein The re-lock trigger condition includes that the comprehensive quality index is below a first threshold and lasts for a first time length, the lock retention condition includes that the comprehensive quality index is above a second threshold and lasts for a second time length, and the second threshold is greater than the first threshold.

3. The apparatus of claim 1 wherein, The coded states constitute a low-correlation codebook, and the absolute value of the normalized cross-correlation coefficient between the configuration vectors of any two coded states is not greater than 0.3; The respiratory spectrum peak signal-to-noise ratio index is determined based on the ratio of the main peak amplitude to the noise floor amplitude in the respiratory frequency band; The spectrum peak stability index is determined based on the degree of change of the main peak frequency within a continuous time window.

4. The apparatus of claim 1 wherein, The processor performs range gating on the echo signals before calculating the respiratory micro-motion quality index, the center distance of the range gating is determined by the main reflection distance of the human body, and the gating width is ±0.1m to ±0.6m.

5. The apparatus of claim 1 wherein, Further comprising a mounting mechanism, the mounting mechanism including an adjustable height mechanism and an adjustable angle mechanism; The processor is further configured to: When the comprehensive quality index is below the lock retention condition and lasts for a predetermined time, and the retention condition is still not met after the re-lock scan, the adjustable height mechanism and / or the adjustable angle mechanism is driven to perform a small step search under a quality gating condition, so that the comprehensive quality index rises and re-enters the lock state.

6. The apparatus of claim 5 wherein, The small step search is performed within a preset angle range and height range, and a stroke limit and a backfall prevention structure are provided.

7. The apparatus of claim 5 wherein, The mounting mechanism allows the user to perform manual coarse adjustment, and the manual coarse adjustment does not change the control logic of the automatic fine adjustment mechanism.

8. A sleep breathing monitoring apparatus, characterized by Comprising: a foldable arc-shaped panel assembly, the arc-shaped panel assembly including a middle segment and at least one wing segment, the wing segment being connected to the middle segment through a folding hinge and being capable of being folded inward to form a storage state; The millimeter wave radar module, the programmable coded metasurface panel, and the exhalation plume sensing array are arranged on the middle segment and the wing segment, so that the arc-shaped panel forms a multi-angle perception configuration covering the chest and neck region of the human body in the unfolded state.

9. The apparatus of claim 8 wherein, The wing segment includes left and right wing segments, and the left and right wing segments are respectively connected to the middle segment and folded inward to cover the middle segment.

10. The apparatus of claim 8 wherein, The folding hinge is provided with an angle limiting and force limiting structure to limit the folding range and prevent overload damage.

Images