PPG optical sleep and respiration monitoring method and system based on gravity sensing intelligent ring

Abstract

The invention provides a PPG optical sleep and respiration monitoring method and system based on a gravity sensing intelligent ring, and the method comprises the steps: synchronously collecting a PPG signal and a three-dimensional gravity acceleration signal of a hand through the intelligent ring, carrying out the adaptive filtering of the PPG signal, extracting a time-domain pulse wave and a frequency-domain heart rate variability feature, and carrying out the detection of the PPG signal. Body movement intensity and body position change frequency are calculated according to the gravitational acceleration signals, and body movement feature vectors are generated. And then fusing the pulse wave characteristics, the heart rate variability characteristics and the body movement characteristics, constructing a multi-dimensional physiological parameter matrix, analyzing and dividing sleep stages based on spatial-temporal correlation of the matrix, and detecting abnormal respiratory rhythm. And finally, dynamically adjusting a judgment threshold value of the abnormal respiratory rhythm, and outputting a sleep quality evaluation result and a respiratory event alarm. According to the invention, the accuracy and real-time performance of sleep monitoring can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent health monitoring. More specifically, the present invention relates to a PPG optical monitoring sleep and breathing method and system based on a gravity sensing intelligent ring. Background Art

[0002] In the field of modern health monitoring, with the continuous improvement of people's attention to their own health, especially the emphasis on sleep quality and respiratory health, traditional sleep monitoring methods mainly rely on polysomnography (PSG). This device needs to be used in hospitals or professional institutions, and multiple electrodes and sensors are connected to the patient to record physiological signals such as electroencephalogram, electrocardiogram, and electromyogram. Although this method can provide comprehensive sleep analysis, due to the complex device, inconvenient wearing, and the need for professional operation, its application in the home environment is limited. In recent years, with the development of wearable device technology, sleep monitoring devices based on photoplethysmography (PPG) and acceleration sensors have gradually emerged. The PPG sensor reflects heart rate and blood flow by detecting blood volume changes, while the acceleration sensor is used to monitor body movement. However, most of these devices in the prior art can only provide simple sleep duration or heart rate monitoring, have limited ability to accurately divide sleep stages and detect abnormal respiratory rhythms, and lack adaptive adjustment to environmental factors, resulting in insufficient accuracy and reliability of monitoring results.

[0003] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: The existing wearable sleep monitoring devices have deficiencies in signal acquisition accuracy, feature extraction depth, and the ability to analyze complex sleep states, and cannot meet the user's needs for accurate sleep monitoring and respiratory health warning. Summary of the Invention

[0004] The present invention provides a PPG optical monitoring sleep and breathing method and system based on a gravity sensing intelligent ring.

[0005] In the first aspect of the present invention, a PPG optical monitoring sleep and breathing method based on a gravity sensing intelligent ring is provided, including:

[0006] S1. Synchronously collect the user's hand PPG signal and three-dimensional gravity acceleration signal through the intelligent ring;

[0007] S2. Perform adaptive filtering on the PPG signal, and extract time-domain pulse waves and frequency-domain heart rate variability features;

[0008] S3. Calculate the body movement intensity and body position change frequency according to the gravity acceleration signal, and generate a body movement feature vector;

[0009] S4. Integrate the pulse wave features, heart rate variability features, and body movement features to construct a multi-dimensional physiological parameter matrix;

[0010] S5. Based on the spatio-temporal correlation analysis of the multi-dimensional matrix, divide the sleep stages and detect abnormal respiratory rhythms;

[0011] S6. Dynamically adjust the determination threshold for abnormal respiratory rhythms and output the sleep quality assessment results and respiratory event alarms.

[0012] Further, the step S1 includes:

[0013] S11. Collect the hand blood flow signal I(λ,t) through a multi-wavelength PPG sensor, where λ is the light source wavelength and I(λ,t) is the light intensity signal at wavelength λ;

[0014] S12. Obtain the gravity acceleration components G x (t), G y (t), G z (t) using a three-axis accelerometer, where G x (t), G y (t), G z (t) are the acceleration components in the X, Y, and Z axis directions respectively;

[0015] S13. Use an anti-motion interference algorithm to perform baseline calibration on the original signal to meet the signal-to-noise ratio SNR≥20dB, where SNR is the power ratio of the signal to the noise.

[0016] Further, the specific steps of the adaptive filtering in the step S2 include:

[0017] S21. Construct a band-pass filter bank with a passband range set to 0.5Hz - 5Hz to suppress high-frequency motion noise;

[0018] S22. Calculate the time-domain features of the pulse wave, including the peak interval ΔT p , the rising slope K r , and the trough area A t , where ΔT p is the time difference between adjacent pulse wave peaks, K r is the slope of the rising edge of the waveform, and A t is the integral area of the trough region;

[0019] S23. Extract the frequency-domain heart rate variability features as shown in the following formula:

[0020]

[0021] where HRV is the heart rate variability feature, P HR (f) is the heart rate power spectral density, and f is the frequency variable.

[0022] Furthermore, step S3 includes:

[0023] S31. Calculate the body movement intensity of the moving variance, where E m is the body movement energy intensity;

[0024] S32. Detect the body position flipping event. When the angle change Δθ > 90° and the duration Δt < 1 s, it is marked as flipping, where Δθ is the angle change of the gravity vector and Δt is the duration of the flipping event;

[0025] S33. Generate the body movement feature vector F m = [E m , N f , T s , where N f is the number of body position flips, and T s is the proportion of the resting period.

[0026] Furthermore, the fusion processing of step S4 includes:

[0027] S41. Normalize the pulse wave time domain features, HRV, and F m to eliminate the dimension difference;

[0028] S42. Construct the fusion matrix M = [ΔT p , K r , A t , HRV, E m , N f , T s , where M is the multi-dimensional physiological parameter matrix;

[0029] S43. Calculate the correlation coefficient matrix C ij of each dimension of the matrix, and screen the strongly correlated feature groups with |C ij | > 0.7, where C ij is the correlation coefficient between the i-th dimension and the j-th dimension feature.

[0030] Furthermore, step S5 includes:

[0031] S51. Divide the sleep stages according to the strongly correlated feature groups, including the wakefulness stage, light sleep stage, deep sleep stage, and REM stage;

[0032] S52. Detect abnormal respiratory rhythm. When the standard deviation of the respiratory interval σ b > σ th and the sudden drop in HRV ΔHRV > 30%, it is determined as abnormal respiratory rhythm, where σ b is the standard deviation of the respiratory interval, σ th is the dynamic threshold, and ΔHRV is the decline amplitude of heart rate variability;

[0033] S53. Update the decision threshold σ using a sliding window mechanism th = μσ hist +(1 - μ)σ curr , where μ = 0.6 is the forgetting factor, σ hist is the historical threshold mean, and σ curr is the calculated value of the current window.

[0034] Furthermore, the step S6 includes:

[0035] S61. Adjust the breathing threshold in real time according to the environmental temperature and humidity. The correction formula is

[0036] σ th ′ = σ th ×[1 + α(T - T0)]

[0037] where T is the environmental temperature, T0 is the reference temperature, and α is the temperature compensation coefficient; α

[0038] S62. Output the sleep quality score S q = ω1D s + ω2N a + ω3R e , where D s is the deep sleep duration, N a is the number of awakenings, R e is the number of abnormal breathing times, and ω1, ω2, ω3 are weight coefficients;

[0039] S63. When S q < S min for three consecutive cycles, trigger the device vibration alarm, where S min is the lowest score threshold.

[0040] Furthermore, the calculation method of the frequency domain heart rate variability in the step S23 is as follows:

[0041] Perform Hilbert transform on the PPG signal, extract the envelope signal e(t), and calculate its power spectral density:

[0042] P HR (f) = |FFT[e(t)]| 2 / T w

[0043] where T w is the analysis window duration, FFT is the fast Fourier transform, and e(t) is the envelope signal.

[0044] Furthermore, the calculation formula of the correlation coefficient matrix in the step S43 is:

[0045]

[0046] where x ik is the i-th dimensional eigenvalue of the k-th sample, μ i is the i-th dimensional feature mean, σ i is the standard deviation, and n is the total number of samples.

[0047] In a second aspect of the present invention, there is provided a PPG optical monitoring sleep and respiratory system based on a gravity sensing smart ring, including:

[0048] An acquisition module for synchronously acquiring the user's hand PPG signal and three-dimensional gravity acceleration signal through the smart ring;

[0049] A feature extraction module for adaptively filtering the PPG signal and extracting time-domain pulse waves and frequency-domain heart rate variability features;

[0050] A vector generation module for calculating body movement intensity and body position change frequency according to the gravity acceleration signal and generating a body movement feature vector;

[0051] A matrix construction module for fusing pulse wave features, heart rate variability features and body movement features to construct a multi-dimensional physiological parameter matrix;

[0052] A partitioning module for partitioning sleep stages and detecting abnormal respiratory rhythms based on spatio-temporal correlation analysis of the multi-dimensional matrix;

[0053] A dynamic adjustment module for dynamically adjusting the determination threshold of abnormal respiratory rhythms and outputting sleep quality assessment results and respiratory event alarms.

[0054] According to the above embodiments of the present invention, it has at least the following beneficial effects: The present invention synchronously acquires PPG signals and three-dimensional gravity acceleration signals through a smart ring, and combines adaptive filtering technology to extract pulse waves and time-domain and frequency-domain features, which can effectively suppress motion interference, improve signal quality, and ensure the accuracy and reliability of monitoring data. At the same time, based on spatio-temporal correlation analysis of the multi-dimensional physiological parameter matrix, sleep stages can be accurately divided, including the wakefulness period, light sleep period, deep sleep period and REM period, and abnormal respiratory rhythms can be detected in real time, providing comprehensive sleep health monitoring for users. In addition, the design of dynamically adjusting the determination threshold of abnormal respiratory rhythms and correcting the respiratory threshold according to environmental temperature and humidity further improves the adaptability and flexibility of the monitoring system, enabling it to maintain stable monitoring performance under different environmental conditions.

[0055] The present invention also constructs a multi-dimensional physiological parameter matrix by fusing pulse wave features, heart rate variability features, and body movement features, and screens strongly correlated feature groups for sleep stage classification and respiratory event detection, which can effectively eliminate the dimension difference and improve the efficiency and accuracy of feature fusion. At the same time, based on the sleep quality scoring and respiratory abnormality warning mechanism, real-time health feedback and warning can be provided for users to help them timely discover potential sleep problems and thus better manage their own health. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] By referring to the following detailed description with reference to the accompanying drawings, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown by way of example and not by way of limitation, wherein:

[0057] Figure 1 is a schematic flow chart of a PPG optical monitoring sleep and breathing method based on a gravity sensing smart ring provided by an embodiment of the present invention;

[0058] Figure 2 is a schematic structural diagram of a PPG optical monitoring sleep and respiratory system based on a gravity sensing smart ring provided by an embodiment of the present invention;

[0059] Figure 3 schematically shows a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] The principles and spirit of the present invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and then implement the present invention, and do not limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to convey the scope of the present invention fully to those skilled in the art.

[0061] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: completely hardware, completely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0062] It should be noted that any element number in the drawings is for illustration and not limitation, and any naming is only for distinction and does not have any limiting meaning.

[0063] The following refers to Figure 1 , Figure 1Schematic diagram of the PPG optical monitoring sleep and respiration method based on a gravity-sensing smart ring provided by an embodiment of the present invention. As Figure 1 shown, a PPG optical monitoring sleep and respiration method 100 based on a gravity-sensing smart ring includes:

[0064] S1. Synchronously collect the user's hand PPG signal and three-dimensional gravity acceleration signal through the smart ring;

[0065] S2. Perform adaptive filtering on the PPG signal to extract time-domain pulse waves and frequency-domain heart rate variability features;

[0066] S3. Calculate the body movement intensity and body position change frequency according to the gravity acceleration signal, and generate a body movement feature vector;

[0067] S4. Integrate the pulse wave features, heart rate variability features, and body movement features to construct a multi-dimensional physiological parameter matrix;

[0068] S5. Based on the spatio-temporal correlation analysis of the multi-dimensional matrix, divide the sleep stages and detect abnormal respiratory rhythms;

[0069] S6. Dynamically adjust the determination threshold for abnormal respiratory rhythms, and output the sleep quality assessment result and respiratory event alarm.

[0070] It should be noted that the present invention proposes a PPG optical monitoring sleep and respiration method based on a gravity-sensing smart ring. PPG (photoplethysmogram) is an optical technology that reflects cardiovascular activities by detecting blood volume changes, while gravity sensing measures the movement and body position changes of the hand through a three-axis accelerometer. The core of this method lies in synchronously collecting PPG signals and three-dimensional gravity acceleration signals to achieve comprehensive monitoring of sleep and respiration states. Pulse wave features and heart rate variability features are extracted through adaptive filtering, and a multi-dimensional physiological parameter matrix is constructed in combination with the body movement feature vector, and then the sleep stages are divided and abnormal respiratory rhythms are detected. This method can not only provide accurate sleep monitoring, but also dynamically adjust the determination threshold for respiratory abnormalities to adapt to different users' physiological states and environmental changes.

[0071] Specifically, the acquisition of PPG signals is accomplished by multi-wavelength sensors on the smart ring. These sensors can detect changes in light intensity at different wavelengths, thereby reflecting the blood flow in the hand. For example, the selection of the light source wavelength can be adjusted according to different monitoring targets, usually in the range of 600 - 1000 nanometers, to ensure sensitivity to oxyhemoglobin in the blood. The gravitational acceleration signal is obtained through a triaxial accelerometer, which measures the acceleration components in the X, Y, and Z directions respectively. In the signal processing stage, the passband range of the adaptive filter bank is set to 0.5Hz - 5Hz. This range can effectively suppress high-frequency motion noise while retaining the physiological characteristics of the pulse wave. The time-domain characteristics of the pulse wave include peak interval, rising slope, and trough area, and these parameters can reflect the contraction and relaxation of the heart. The heart rate variability characteristics are calculated by analyzing the heart rate power spectral density, which can reflect the activity state of the autonomic nervous system. The body movement feature vector is generated by calculating the moving variance of the body movement intensity and the body position flipping event, and these features can reflect the body activity during sleep.

[0072] Preferably, to further improve the monitoring accuracy, a more detailed analysis of the time-domain characteristics of the pulse wave can be performed. For example, high-precision timestamps can be used for calculating the peak interval to ensure that the measurement accuracy of the time difference between adjacent pulse wave peaks reaches the millisecond level. The calculation of the rising slope can be achieved by fitting the linear part of the rising edge of the pulse wave, so as to more accurately reflect the intensity of heart contraction. In the extraction of body movement features, the detection of body position flipping events can be realized by setting an angle change threshold (such as greater than 90°) and a duration threshold (such as less than 1 second), which can effectively distinguish normal body position adjustments from involuntary movements during sleep.

[0073] Furthermore, for the extraction of heart rate variability characteristics, the Hilbert transform can be used to extract the envelope signal of the PPG signal, and then calculate its power spectral density. This method can better reflect the dynamic changes of heart rate and provide a more reliable basis for sleep stage classification and detection of abnormal respiratory rhythms.

[0074] In some embodiments, step S1 includes:

[0075] S11. Collect the hand blood flow signal I(λ,t) through a multi-wavelength PPG sensor, where λ is the light source wavelength and I(λ,t) is the light intensity signal at wavelength λ;

[0076] S12. Obtain the gravitational acceleration components G x (t), G y (t), G z (t), where G x (t), G y (t), G z(t) are the acceleration components in the X, Y, and Z axis directions respectively;

[0077] S13. Use an anti-motion interference algorithm to perform baseline calibration on the original signal, ensuring that the signal-to-noise ratio SNR ≥ 20 dB, where SNR is the power ratio of the signal to the noise.

[0078] It should be noted that the signal acquisition steps mentioned in the present invention include collecting the hand blood flow signal through a multi-wavelength PPG sensor and obtaining the gravitational acceleration components through a triaxial accelerometer. The PPG sensor is a sensor based on the optical principle, which reflects the blood flow in the blood vessels by emitting light of a specific wavelength and detecting the change in the intensity of the reflected light. The multi-wavelength PPG sensor can simultaneously collect the light intensity signals at multiple wavelengths, thus more comprehensively reflecting the optical characteristics of the blood. The triaxial accelerometer is a sensor that can measure the acceleration of an object in three-dimensional space. By detecting the acceleration components in the X, Y, and Z directions, it can accurately reflect the motion state and body position change of the hand. In addition, in order to improve the signal quality, an anti-motion interference algorithm is used to perform baseline calibration on the original signal to ensure that the signal-to-noise ratio reaches more than 20 dB, thereby providing a high-quality data basis for subsequent feature extraction and analysis.

[0079] Specifically, the signal collected by the multi-wavelength PPG sensor can be expressed as the relationship between the light intensity signal and the light source wavelength. For example, red light (wavelength of about 660 nm) and infrared light (wavelength of about 940 nm) can be selected as the light sources because these two wavelengths of light have different absorption characteristics for oxyhemoglobin and deoxyhemoglobin in the blood, thus being able to reflect the oxygenation state of the blood. The gravitational acceleration components collected by the triaxial accelerometer respectively correspond to the X, Y, and Z directions, and these directions can be defined according to the wearing position and direction of the smart ring. For example, the X axis can be defined as the direction along the finger pointing to the wrist, the Y axis is perpendicular to the finger plane, and the Z axis is consistent with the finger pointing direction. In the signal processing stage, the role of the anti-motion interference algorithm is to remove the noise interference caused by hand movement, thereby improving the purity of the signal. The signal-to-noise ratio (SNR) is the ratio of the signal power to the noise power. Through algorithm optimization, it is ensured that the collected signal has a high signal-to-noise ratio, thus guaranteeing the accuracy of subsequent analysis.

[0080] Preferably, during signal acquisition, the light source selection and signal processing algorithm of the multi-wavelength PPG sensor can be further optimized. For example, in addition to red light and infrared light, green light (with a wavelength of approximately 530 nm) can be added as a light source to improve the detection sensitivity of microvascular blood flow. In addition, the anti-motion interference algorithm can be implemented through adaptive filtering techniques, such as using a Kalman filter to dynamically adjust the filtering parameters according to the characteristics of the real-time signal, thereby more effectively removing motion noise. In the signal processing of the triaxial accelerometer, the stationary state and the motion state can be distinguished by setting an acceleration threshold. For example, when the acceleration value exceeds a certain threshold (such as 0.1 g), it is determined as the motion state, and the corresponding signal calibration mechanism is triggered. These optimization measures can further improve the accuracy and reliability of signal acquisition, providing higher-quality data support for subsequent sleep and respiration monitoring.

[0081] In some embodiments, the specific steps of the adaptive filtering in step S2 include:

[0082] S21. Construct a band-pass filter bank with a passband range set to 0.5 Hz - 5 Hz to suppress high-frequency motion noise;

[0083] S22. Calculate the time-domain characteristics of the pulse wave, including the peak interval ΔT p , the rising slope K r , and the valley area A t , where ΔT p is the time difference between adjacent pulse wave peaks, K r is the slope of the rising edge of the waveform, and A t is the integral area of the valley region;

[0084] S23. Extract the frequency-domain heart rate variability characteristics as shown in the following formula:

[0085]

[0086] where HRV is the heart rate variability characteristic, P HR (f) is the heart rate power spectral density, and f is the frequency variable.

[0087] It should be noted that the adaptive filtering step in the present invention is a key link in realizing PPG signal processing. Its purpose is to extract effective information that can reflect the physiological state by constructing a band-pass filter bank and calculating the time-domain and frequency-domain characteristics of the pulse wave. The role of the band-pass filter bank is to remove high-frequency motion noise and low-frequency interference and retain the physiological characteristics in the signal. The time-domain characteristics of the pulse wave include the peak interval, rising slope, and valley area, which can reflect the contraction and relaxation states of the heart, while the frequency-domain heart rate variability characteristics evaluate the activity of the autonomic nervous system by analyzing the heart rate power spectral density. These steps together provide a high-quality signal basis for subsequent sleep and respiration monitoring.

[0088] Specifically, the passband range of the band-pass filter bank is set from 0.5 Hz to 5 Hz. This range can effectively suppress high-frequency motion noise while retaining the physiological characteristics of the pulse wave. For example, the lower limit of 0.5 Hz can remove low-frequency interference related to breathing, while the upper limit of 5 Hz can filter out high-frequency noise caused by rapid hand movement. In the extraction of the time-domain characteristics of the pulse wave, the peak interval refers to the time difference between adjacent pulse wave peaks, which reflects the stability of the heart rate; the rising slope refers to the slope of the rising edge of the pulse wave, which can reflect the strength of cardiac contraction; and the trough area refers to the integral area of the pulse wave trough region, which reflects the blood flow during diastole. The frequency-domain heart rate variability characteristics are calculated by analyzing the heart rate power spectral density, and the core is to evaluate the regulatory ability of the autonomic nervous system through the power ratio in a specific frequency band. For example, the power ratio in the frequency band from 0.04 Hz to 0.4 Hz can reflect the changes in heart rate variability, thus providing an important basis for sleep stage classification and detection of abnormal respiratory rhythms.

[0089] Preferably, in order to further improve the effect of adaptive filtering, the filter design and feature extraction method can be optimized. For example, an adaptive filter such as a Kalman filter or a wavelet transform filter can be used to dynamically adjust the filter parameters according to the real-time characteristics of the signal, thereby more effectively removing motion noise. In the extraction of the time-domain characteristics of the pulse wave, the detection of the peak amplitude can be increased to reflect the intensity changes of each heartbeat. For the calculation of the frequency-domain heart rate variability characteristics, in addition to the power ratio in the frequency band from 0.04 Hz to 0.4 Hz, the ratio of high-frequency (HF) and low-frequency (LF) power (LF / HF) can also be introduced to more comprehensively evaluate the balance state of the autonomic nervous system.

[0090] Furthermore, the sliding window technique can be used to monitor the heart rate variability in real time to adapt to the dynamic changes in physiological states during sleep. These optimization measures can further improve the accuracy and reliability of signal processing, providing more accurate physiological characteristics for sleep and respiration monitoring.

[0091] In some embodiments, step S3 includes:

[0092] S31. Calculate the body movement intensity of the moving variance, where E m is the body movement energy intensity;

[0093] S32. Detect the body position flipping event, and mark it as flipping when the angle change Δθ > 90° and the duration Δt < 1 s, where Δθ is the change amount of the gravity vector angle and Δt is the duration of the flipping event;

[0094] S33. Generate the body movement feature vector F m = [E m, N f , T s , where N f is the number of body position flips, and T s is the proportion of the rest period.

[0095] It should be noted that the step of generating the body movement feature vector in the present invention is achieved by analyzing the gravity acceleration signal. Its purpose is to reflect the body activity of the user during sleep by calculating the body movement intensity and the body position change frequency. The body movement intensity is obtained by calculating the moving variance of the acceleration signal, which can reflect the energy magnitude of the body movement; while the detection of the body position flip event is achieved by monitoring the angular change of the gravity vector. The body movement feature vector integrates this information into a feature set, providing an important kinematic basis for subsequent sleep stage classification and abnormal breathing rhythm detection.

[0096] Specifically, the calculation formula for the body movement intensity is the square root of the sum of the squares of the acceleration signals in three directions, that is where a x , a y , a z are the acceleration components in the X, Y, and Z axis directions respectively. The calculation of the moving variance is to perform variance analysis on the body movement intensity within a certain time window, so as to obtain the dynamic change of the body movement energy. The detection of the body position flip event is achieved by monitoring the angular change amount of the gravity vector. When the angular change exceeds 90° and the duration is less than 1 second, it is determined as one body position flip. The body movement feature vector includes parameters such as the number of body position flips and the proportion of the rest period, and these parameters can comprehensively reflect the body activity during sleep. For example, the number of body position flips can reflect the restlessness of sleep, while the proportion of the rest period can reflect the continuity of sleep.

[0097] Preferably, in order to improve the accuracy and reliability of the body movement feature vector, the calculation of the body movement intensity and the detection of the body position flip event can be optimized. For example, when calculating the body movement intensity, a weighted moving average algorithm can be introduced to perform weighted processing on the acceleration signals within different time windows, so as to more accurately reflect the change trend of the body movement energy. For the detection of the body position flip event, in addition to the threshold judgment of the angular change and the duration, the amplitude change of the acceleration signal can also be combined for comprehensive analysis to avoid misjudgment.

[0098] Furthermore, the body movement feature vector can be further expanded. For example, the statistics of the body movement frequency can be increased, that is, the number of body movement events occurring within a unit time, so as to more comprehensively reflect the body activity during sleep. These optimization measures can further improve the quality of the body movement feature vector, providing richer kinematic information for sleep monitoring.

[0099] In some embodiments, the fusion process in step S4 includes:

[0100] S41. Normalize the time-domain features of the pulse wave, HRV, and F m to eliminate the dimensional differences;

[0101] S42. Construct a fusion matrix M = [ΔT p , K r , A t , HRV, E m , N f , T s , where M is a multi-dimensional physiological parameter matrix;

[0102] S43. Calculate the correlation coefficient matrix C ij for each dimension of the matrix, and select the strongly correlated feature groups with |C ij | > 0.7, where C ij is the correlation coefficient between the i-th and j-th dimensional features.

[0103] It should be noted that the fusion process steps mentioned in the present invention integrate the time-domain features of the pulse wave, heart rate variability features, and body movement features to construct a multi-dimensional physiological parameter matrix. The core of this process is to eliminate the dimensional differences between different features and screen out the feature groups that are of great significance for sleep and breathing monitoring through correlation analysis. The construction of the multi-dimensional physiological parameter matrix provides a comprehensive and integrated physiological data basis for subsequent sleep stage division and abnormal respiratory rhythm detection, enabling the monitoring system to more accurately reflect the user's sleep state and breathing condition.

[0104] Specifically, the first step of the fusion process is to normalize the time-domain features of the pulse wave, heart rate variability features, and body movement features. Normalization is the process of converting data with different dimensions to the same scale, for example, scaling all feature values to between 0 and 1, thereby eliminating dimensional differences and facilitating subsequent analysis. The constructed multi-dimensional physiological parameter matrix includes the time-domain features of the pulse wave (such as peak interval, rising slope, and trough area), heart rate variability features, and body movement features (such as body movement intensity, number of body position flips, and proportion of rest periods). The calculation of the correlation coefficient matrix is through statistical analysis methods to evaluate the correlation between the features of each dimension in the matrix. For example, when the absolute value of the correlation coefficient between two features is greater than 0.7, it indicates a strong correlation between them, and these strongly correlated feature groups will be used for subsequent sleep stage division and abnormal respiratory rhythm detection. In this way, the most valuable feature combinations for sleep monitoring can be screened out, improving the accuracy and efficiency of monitoring.

[0105] Preferably, in order to further optimize the fusion processing process, the normalization method and correlation analysis can be refined. For example, in the normalization process, the Z-score normalization method can be adopted, that is, normalization is achieved by subtracting the mean of the features and dividing by their standard deviation. This method can better preserve the distribution characteristics of the original data. In the correlation analysis, in addition to calculating the correlation coefficients between pairwise features, multivariate correlation analysis can also be introduced to evaluate the joint correlation between multiple features, so as to more comprehensively reflect the interactions between different physiological parameters.

[0106] Furthermore, for the selected strongly correlated feature groups, their performance differences in different sleep stages and breathing states can be further analyzed to provide a more accurate feature basis for sleep stage classification and detection of abnormal breathing rhythms. For example, feature models can be established for deep sleep stage and light sleep stage respectively to improve the accuracy of sleep stage classification.

[0107] In some embodiments, the step S5 includes:

[0108] S51. Classify sleep stages according to the strongly correlated feature groups, including wakefulness stage, light sleep stage, deep sleep stage, and REM stage;

[0109] S52. Detect abnormal breathing rhythms. When the standard deviation of the respiratory interval σ b >σ th and the sudden drop in HRV ΔHRV > 30%, it is determined as an abnormal breathing rhythm, where σ b is the standard deviation of the respiratory interval, σ th is the dynamic threshold, and ΔHRV is the decline amplitude of heart rate variability;

[0110] S53. Update the decision threshold σ th = μσ hist +(1 - μ)σ curr , where μ = 0.6 is the forgetting factor, σ hist is the historical threshold mean, and σ curr is the calculated value of the current window.

[0111] It should be noted that the spatio-temporal correlation analysis mentioned in the present invention is based on a multi-dimensional physiological parameter matrix. Its purpose is to classify sleep stages and detect abnormal breathing rhythms by analyzing the spatio-temporal relationships between features. Sleep stage classification is to distinguish the wakefulness stage, light sleep stage, deep sleep stage, and REM stage according to the dynamic changes of the strongly correlated feature groups. Detection of abnormal breathing rhythms is to judge whether there is abnormal breathing by analyzing features such as the standard deviation of the respiratory interval and the sudden drop in heart rate variability. The dynamic adjustment of the abnormal breathing rhythm decision threshold is achieved through a sliding window mechanism. This method can adjust the threshold in real time according to the user's physiological state and environmental changes, thereby improving the accuracy and adaptability of monitoring.

[0112] Specifically, the sleep stage division is based on a strongly correlated feature group in a multi-dimensional physiological parameter matrix. For example, heart rate variability (HRV) is usually higher during deep sleep and lower during wakefulness or light sleep; body movement intensity is usually lower during deep sleep and REM sleep and higher during wakefulness or light sleep. By analyzing the dynamic changes of these features, the sleep process can be divided into different stages. The abnormal respiratory rhythm detection is achieved by monitoring the standard deviation of the respiratory interval and the sudden drop in heart rate variability. For example, when the standard deviation of the respiratory interval exceeds a set dynamic threshold and the sudden drop in heart rate variability exceeds 30%, it is determined as an abnormal respiratory rhythm. The adjustment of the dynamic threshold is achieved through a sliding window mechanism, that is, the mean value and the current value of the threshold are calculated based on historical data and current window data, and weighted by a forgetting factor to achieve the dynamic update of the threshold. This method can adapt to the physiological characteristics of different users and environmental changes, improving the accuracy and reliability of monitoring.

[0113] Preferably, in order to further improve the accuracy of sleep stage division and abnormal respiratory rhythm detection, the analysis of relevant features and the threshold adjustment mechanism can be optimized. For example, in sleep stage division, more physiological parameters such as skin electrical activity or body temperature change can be introduced to more comprehensively reflect the sleep state. For abnormal respiratory rhythm detection, the change in respiratory rate can be combined for comprehensive analysis to improve the detection sensitivity. In terms of dynamic threshold adjustment, a compensation mechanism for environmental factors can be introduced. For example, the determination threshold for abnormal respiratory rhythm can be adjusted according to the environmental temperature and humidity.

[0114] Furthermore, the size of the sliding window and the selection of the forgetting factor can also be adjusted personalized according to the user's sleep data. For example, for users with poor sleep quality, the size of the sliding window can be appropriately reduced to improve the real-time monitoring; for users with good sleep quality, the value of the forgetting factor can be appropriately increased to improve the stability of the threshold. These optimization measures can further enhance the performance of the monitoring system and provide more accurate sleep and respiratory health monitoring for users.

[0115] In some embodiments, step S6 includes:

[0116] S61. Adjust the respiratory threshold in real time according to the environmental temperature and humidity, and the correction formula is

[0117] σ th ′ =σ th ×[1 + α(T - T0)]

[0118] where T is the environmental temperature, T0 is the reference temperature, and α is the temperature compensation coefficient; α

[0119] S62. Output the sleep quality score S q =ω1Ds + ω2N a + ω3R e , where D s is the deep sleep duration, N a is the number of awakenings, R e is the number of abnormal breathing times, and ω1, ω2, and ω3 are weight coefficients;

[0120] S63. When three consecutive cycles of S q < S min are met, the device triggers a vibration alarm, where S min is the lowest scoring threshold.

[0121] It should be noted that the steps of dynamically adjusting the abnormal breathing rhythm determination threshold, outputting the sleep quality assessment result, and breathing event alarm in the present invention are based on correcting the breathing threshold according to the environmental temperature and humidity, calculating the sleep quality score by integrating multiple physiological parameters, and triggering an alarm when an abnormality is detected. This process takes into account the influence of environmental factors on the breathing pattern, and provides personalized health monitoring and warning services for users by adjusting the threshold and evaluating the sleep quality in real time.

[0122] Specifically, the dynamic adjustment of the abnormal breathing rhythm determination threshold is achieved by correcting the environmental temperature and humidity. For example, when the environmental temperature rises, the breathing frequency of the human body may naturally increase, so it is necessary to appropriately adjust the breathing threshold to avoid misjudgment. The correction formula dynamically adjusts the breathing threshold by introducing a temperature compensation coefficient and combining the difference between the environmental temperature and the reference temperature. The sleep quality score is calculated by integrating physiological parameters such as deep sleep duration, number of awakenings, and number of abnormal breathing times. For example, the higher the proportion of deep sleep duration, the fewer the number of awakenings, and the fewer the number of abnormal breathing times, the higher the sleep quality score. In addition, when the sleep quality score for multiple consecutive cycles is lower than the set lowest threshold, the device will trigger a vibration alarm to remind the user that there may be a sleep problem.

[0123] Preferably, in order to further improve the accuracy of abnormal breathing rhythm determination and the reliability of sleep quality assessment, the environmental temperature and humidity correction mechanism and the sleep quality scoring model can be optimized. For example, in the environmental temperature and humidity correction, a humidity compensation coefficient can be introduced to further refine the adjustment of the breathing threshold to adapt to breathing changes under different humidity conditions. For the sleep quality scoring model, more physiological parameters such as body movement frequency or heart rate variability can be introduced to more comprehensively reflect the sleep quality.

[0124] Furthermore, dynamic weights can be set for different physiological parameters. For example, when the weight of deep sleep duration is relatively high, the weight of the number of awakenings can be appropriately reduced to better reflect the impact of different sleep stages on sleep quality. These optimization measures can further improve the adaptability and accuracy of the monitoring system, providing users with more accurate health monitoring and early warning services.

[0125] In some embodiments, the calculation method of the frequency-domain heart rate variability in step S23 is as follows:

[0126] Perform Hilbert transform on the PPG signal, extract the envelope signal e(t), and calculate its power spectral density:

[0127] P HR (f) = |FFT[e(t)]| 2 / T w

[0128] where T w is the analysis window duration, FFT is the fast Fourier transform, and e(t) is the envelope signal.

[0129] It should be noted that the calculation method of the frequency-domain heart rate variability mentioned in the present invention is realized by extracting the envelope signal of the PPG signal through Hilbert transform and calculating its power spectral density. Hilbert transform is a signal processing technique used to extract the instantaneous amplitude and phase information from the original signal, thereby obtaining the envelope signal. The power spectral density describes the energy distribution of the signal in the frequency domain. By analyzing the heart rate power spectral density, the activity state of the autonomic nervous system can be effectively reflected, providing important physiological indicators for sleep and respiration monitoring.

[0130] Specifically, the role of Hilbert transform is to convert the PPG signal into an analytic signal, thereby extracting its envelope signal. The envelope signal can reflect the amplitude change of the PPG signal, and the power spectral density is calculated by performing fast Fourier transform (FFT) on the envelope signal. During the calculation process, the analysis window duration is a key parameter, which determines the resolution and accuracy of the frequency-domain analysis. For example, choosing a shorter analysis window can improve the time resolution but may reduce the frequency resolution; while a longer analysis window has the opposite effect. By calculating the power spectral density, the frequency-domain characteristics of heart rate variability (HRV) can be obtained, such as the power ratio of the low-frequency (LF) and high-frequency (HF) components. These characteristics can reflect the activity balance of the sympathetic and parasympathetic nerves, providing an important basis for sleep stage classification and detection of abnormal respiratory rhythms.

[0131] Preferably, in order to further improve the calculation accuracy and reliability of the heart rate variability in the frequency domain, the Hilbert transform and the power spectral density calculation process can be optimized. For example, in the Hilbert transform, an improved algorithm can be adopted to reduce the calculation complexity while improving the extraction accuracy of the envelope signal. In the power spectral density calculation, a window function (such as a Hanning window or a Hamming window) can be introduced to reduce spectral leakage, thereby improving the accuracy of the frequency domain analysis.

[0132] Furthermore, a multi-window analysis method can also be combined. By calculating the power spectral density within different time windows and taking the average value, the stability of the heart rate variability characteristics can be further improved. These optimization measures can better reflect the dynamic changes of the autonomic nervous system and provide more reliable physiological indicators for sleep and respiration monitoring.

[0133] In some embodiments, the calculation formula of the correlation coefficient matrix in step S43 is:

[0134]

[0135] where x ik is the i-th dimensional eigenvalue of the k-th sample, μ i is the i-th dimensional feature mean, σ i is the standard deviation, and n is the total number of samples.

[0136] It should be noted that the correlation coefficient matrix calculation method mentioned in the present invention is used to evaluate the correlation between each feature dimension in the multi-dimensional physiological parameter matrix. The correlation coefficient matrix can help screen out strongly correlated feature groups that are of great significance for sleep and respiration monitoring, thereby improving the accuracy and efficiency of the monitoring system. The correlation coefficient is a statistical indicator that measures the linear relationship between two variables, and its value ranges from -1 to 1. The closer the absolute value is to 1, the stronger the correlation between the variables.

[0137] Specifically, the calculation of the correlation coefficient matrix is achieved by pairwise comparison of each feature dimension in the multi-dimensional physiological parameter matrix. For example, for the i-th and j-th dimensional features in the matrix, the calculation formula of the correlation coefficient involves statistical quantities such as the mean, standard deviation, and total number of samples of the samples. The sample mean is the average value of all samples in a certain feature dimension, reflecting the average level of this feature; the standard deviation measures the degree of dispersion of the sample data. By calculating the correlation coefficient between each pair of features, a matrix can be obtained, where the elements on the diagonal are 1 (because each feature is completely correlated with itself), and the other elements represent the correlation between different features. For example, when the absolute value of the correlation coefficient is greater than 0.7, it can be considered that there is a strong correlation between these two features, and these strongly correlated feature groups will be used for subsequent sleep stage classification and abnormal respiratory rhythm detection.

[0138] Preferably, in order to further improve the accuracy and efficiency of the calculation of the correlation coefficient matrix, the calculation process can be optimized. For example, when calculating the sample mean and standard deviation, an online algorithm can be adopted to update these statistics in real time in the data stream, thereby reducing the computational complexity and improving the response speed. In addition, in order to more comprehensively evaluate the relationship between features, partial correlation coefficient analysis can be introduced to exclude the interference of other variables, so as to more accurately reflect the direct correlation between two features. When screening strong correlation feature groups, in addition to setting a fixed correlation coefficient threshold (such as 0.7), the threshold can also be dynamically adjusted according to the specific application scenario. For example, when the feature dimension is large, the threshold can be appropriately reduced to avoid missing important features. These optimization measures can further improve the reliability of the correlation analysis and provide a more effective feature selection method for sleep and breathing monitoring.

[0139] The above-mentioned various embodiments of the present invention have the following beneficial effects: By synchronously collecting the hand PPG signal and the three-dimensional gravitational acceleration signal through the smart ring, the present invention can accurately monitor the sleep and breathing states. In the signal acquisition stage, the anti-motion interference algorithm is used to perform baseline calibration on the original signal, which can effectively improve the signal-to-noise ratio and ensure the reliability of the collected signal quality. The application of the adaptive filtering technology can further optimize the PPG signal processing, extract accurate time-domain pulse waves and frequency-domain heart rate variability features, and provide high-quality data for subsequent analysis. At the same time, by calculating the body movement intensity and the body position change frequency based on the gravitational acceleration signal and generating a body movement feature vector, the body activity situation during sleep can be comprehensively reflected. By fusing the pulse wave features, heart rate variability features and body movement features to construct a multi-dimensional physiological parameter matrix and screening strong correlation feature groups, the dimension difference can be effectively eliminated, and the efficiency and accuracy of feature fusion can be improved, providing strong support for sleep stage division and abnormal breathing rhythm detection.

[0140] In terms of sleep stage division and abnormal breathing rhythm detection, based on the spatio-temporal correlation analysis of the multi-dimensional matrix, the present invention can accurately divide the wakefulness period, light sleep period, deep sleep period and REM period, and at the same time detect abnormal conditions such as the standard deviation of the respiratory interval and the sudden drop of heart rate variability, and timely discover potential breathing problems. In addition, the output of the sleep quality score and the breathing abnormality warning mechanism can provide real-time health feedback for users and help users better manage their own sleep health.

[0141] As Figure 2 shown, a PPG optical monitoring sleep and respiratory system 200 based on a gravity sensing smart ring according to some embodiments, the system 200 includes:

[0142] An acquisition module 201 for synchronously collecting the hand PPG signal and the three-dimensional gravitational acceleration signal of a user through the smart ring;

[0143] The feature extraction module 202 is used to perform adaptive filtering on the PPG signal and extract time-domain pulse waves and frequency-domain heart rate variability features;

[0144] The vector generation module 203 is used to calculate the body movement intensity and the body position change frequency according to the gravity acceleration signal and generate a body movement feature vector;

[0145] The matrix construction module 204 is used to fuse the pulse wave features, heart rate variability features and body movement features to construct a multi-dimensional physiological parameter matrix;

[0146] The division module 205 is used to divide the sleep stages and detect abnormal respiratory rhythms based on the spatio-temporal correlation analysis of the multi-dimensional matrix;

[0147] The dynamic adjustment module 206 is used to dynamically adjust the determination threshold of abnormal respiratory rhythms and output the sleep quality assessment result and respiratory event alarm.

[0148] It can be understood that the various modules described in the PPG optical monitoring sleep and respiratory system 200 based on the gravity sensing smart ring correspond to the respective steps in the PPG optical monitoring sleep and respiratory method based on the gravity sensing smart ring described in the reference Figure 1 Therefore, the operations, features, and beneficial effects described above for the PPG optical monitoring sleep and respiratory method based on the gravity sensing smart ring also apply to the PPG optical monitoring sleep and respiratory system 200 based on the gravity sensing smart ring and the modules included therein, and will not be repeated here.

[0149] Next, refer to Figure 3 , which shows a schematic structural diagram of an electronic device 300 suitable for implementing some embodiments of the present invention. The electronic devices in some embodiments of the present invention may include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet Computers), PMPs (Portable Multimedia Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The terminal device shown is only an example and should not impose any limitations on the functions and usage scopes of the embodiments of the present invention.

[0150] As Figure 3As shown, the electronic device 300 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 301, which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0151] Generally, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 3 an electronic device 300 with various devices is shown, it should be understood that it is not required to implement or include all the shown devices. Instead, more or fewer devices may be implemented or included. Figure 3 Each block shown in may represent a device or, as needed, multiple devices.

[0152] Furthermore, the storage medium of the embodiments of the present application stores program instructions capable of implementing all the above methods. Among them, the program instructions may be stored in the above storage medium in the form of a software product, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: various media capable of storing program codes such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc, or a terminal device such as a computer, a server, a mobile phone, or a tablet.

[0153] The above description is only some preferred embodiments of the present invention and an explanation of the technical principles applied. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the technical features (but not limited to) having similar functions disclosed in the embodiments of the present invention.

Claims

1. A PPG optical monitoring sleep and breathing method based on a gravity sensing smart ring, characterized in that: The following steps are involved: S1. Synchronously collect the user's hand PPG signal and three-dimensional gravity acceleration signal through the smart ring; S2. Adaptively filter the PPG signal to extract the pulse wave in the time domain and the heart rate variability characteristics in the frequency domain; S3. Calculate the body motion intensity and body position change frequency according to the gravity acceleration signal to generate a body motion feature vector; S4. Integrate pulse wave features, heart rate variability features and body movement features to construct a multi-dimensional physiological parameter matrix; S5. Based on the spatiotemporal correlation analysis of multidimensional matrix, sleep stages are divided and abnormal breathing rhythm is detected; S6. Dynamically adjust the judgment threshold of abnormal respiratory rhythm, and output sleep quality assessment results and respiratory event alarms.

2. The method according to claim 1, characterized in that: The step S1 comprises: S11. Collecting the hand blood flow signal I(λ, t) through the multi-wavelength PPG sensor, where λ is the wavelength of the light source and I(λ, t) is the light intensity signal at the wavelength λ; S12. Obtain the gravitational acceleration component G through a three-axis accelerometer x (t), G y (t), G z (t), where G x (t), G y (t), G z (t) are the acceleration components in the X, Y, and Z axis directions respectively; S13. Use anti-motion interference algorithm to perform baseline calibration on the original signal to meet the signal-to-noise ratio SNR ≥ 20dB.

3. The method according to claim 1, characterized in that The specific steps of the adaptive filtering in step S2 include: S21. Construct a bandpass filter bank to suppress high frequency motion noise; S22. Calculate the pulse wave time domain characteristics, including the peak interval ΔT p , rising slope K r , trough area A t ; S23. Extract the frequency domain heart rate variability feature, as shown in the following formula: Among them, HRV is the heart rate variability characteristic, P HR (f) is the power spectrum density of heart rate, and f is the frequency variable.

4. The method according to claim 1, characterized in that The step S3 comprises: S31. Calculate body dynamic intensity E m The moving variance, where E m is the body motion energy intensity, as shown in the following formula; S32. Detect body position flip events, and mark as flip when the angle change Δθ>90° and the duration Δt<1s; S33. Generate a body motion feature vector, the body motion feature vector is: F m =[E m ,N f ,T s ], where N f is the number of body position turns, T s is the proportion of rest time.

5. The method according to claim 3, characterized in that: The fusion processing of step S4 includes: S41. Pulse wave time domain features, heart rate variability features HRV and body motion feature vector F m Perform normalization to eliminate dimension differences; S42. Based on the normalized pulse wave time domain features, heart rate variability features HRV and body motion feature vector F m Construct a multi-dimensional physiological parameter fusion matrix.

6. The method according to claim 5, characterized in that The step S5 comprises: S51. Divide the sleep stages according to the strongly correlated feature groups, including wakefulness, light sleep, deep sleep, and REM; S52. Detect abnormal respiratory rhythm, including: When the standard deviation of breathing interval σ b >σ th When ΔHRV>30%, it is judged as abnormal respiratory rhythm, where σ b is the standard deviation of the breathing interval, σ th is the dynamic threshold, ΔHRV is the decrease in heart rate variability; S53. Update the dynamic threshold using a sliding window mechanism; as shown in the following formula; σ th =μσ hist +(1-μ)σ curr , where μ = 0.6 is the forgetting factor, σ hist is the historical threshold mean, σ curr Calculates a value for the current window.

7. The method according to claim 6, characterized in that The step S6 comprises: S61. Adjust the breathing threshold in real time according to the ambient temperature and humidity. The correction formula is: s th ′ =s th ×[1+α(T-T0)] Where T is the ambient temperature, T0 is the reference temperature, α is the temperature compensation coefficient, σ th ′ is the respiratory threshold; S62. Output the sleep quality score, as shown in the following formula; S q =ω1D s +ω2N a +ω3R e , where D s is the deep sleep duration, N a is the number of awakenings, R e is the number of abnormal breathing, ω1, ω2, ω3 are weight coefficients; S63. When three consecutive cycles S q min When the device vibrates, an alarm is triggered, where S min is the minimum score threshold.​ 8. The method according to claim 3, characterized in that The method for calculating the frequency domain heart rate variability in step S23 is: Perform Hilbert transform on the PPG signal, extract the envelope signal e(t), and calculate its power spectrum density, as shown in the following formula; P HR (f)=|FFT[e(t)]| 2 / T w Among them, T w is the analysis window duration, FFT is fast Fourier transform, and e(t) is the envelope signal.

9. The method according to claim 5, characterized in that The calculation formula of the correlation coefficient matrix in step S43 is: Among them, x ik is the eigenvalue of the i-th dimension of the k-th sample, μ i is the feature mean of the i-th dimension, σ i is the standard deviation and n is the total number of samples.

10. A PPG optical monitoring sleep and breathing system based on a gravity sensing smart ring, characterized in that: include: The acquisition module is used to synchronously collect the user's hand PPG signal and three-dimensional gravity acceleration signal through the smart ring; The feature extraction module is used to adaptively filter the PPG signal and extract the time-domain pulse wave and frequency-domain heart rate variability features; A vector generation module is used to calculate the body motion intensity and body position change frequency according to the gravity acceleration signal to generate a body motion feature vector; Matrix construction module, used to fuse pulse wave features, heart rate variability features and body motion features to construct a multi-dimensional physiological parameter matrix; A partitioning module is used to divide sleep stages and detect abnormal breathing rhythm based on spatiotemporal correlation analysis of multidimensional matrices; The dynamic adjustment module is used to dynamically adjust the judgment threshold of abnormal respiratory rhythm and output sleep quality assessment results and respiratory event alarms.

Images