Method and device for detecting infant sleep based on multi-mode signal and dynamic threshold value

Abstract

The invention provides a method for detecting infant sleep based on a multi-mode signal and a dynamic threshold value. The method comprises the steps that a photoelectric volume pulse wave signal and an acceleration signal of a collection area are obtained; optimizing the photoelectric volume pulse wave signal and the acceleration signal to obtain an optimized photoelectric volume pulse wave signal and an optimized acceleration signal; wherein the optimized photoelectric volume pulse wave signal and the optimized acceleration signal are optimized signals screened through filtering and multi-mode signal quality evaluation; determining the posture of the baby according to the optimized acceleration signal; and determining a monitoring risk level according to the baby posture and the optimized photoelectric volume pulse wave signal. The optimized photoelectric volume pulse wave signal and the optimized acceleration signal are obtained through optimization, the monitoring signal is screened and optimized, the noise interference resistance is improved, the accuracy of the monitoring result is ensured, meanwhile, the threshold value can be dynamically adjusted, and the applicability is improved.

Description

Technical Field

[0001] This application relates to the field of medical monitoring and identification technology, and in particular to a method and apparatus for detecting infant sleep based on multimodal signals and dynamic thresholds. Background Technology

[0002] Infant sleep-related respiratory safety monitoring is a crucial means of preventing Sudden Infant Death Syndrome (SIDS) and respiratory-related risks. Current wearable monitoring technologies primarily use photoplethysmography (PPG) to collect physiological parameters such as blood oxygen saturation (SpO2) and pulse rate (PR), triggering alarms for abnormal readings based on preset thresholds. However, this type of technology faces the following key technical bottlenecks:

[0003] The lack of a signal quality assessment mechanism and the high misjudgment rate due to noise interference:

[0004] Traditional infant sleep safety monitoring systems lack a real-time assessment mechanism for the quality of raw signals and fail to distinguish between valid physiological signals and noise interference (such as motion artifacts and electromagnetic interference). Clinical studies have shown that the decline in PPG signal quality caused by infant movement during sleep can lead to blood oxygenation measurement errors exceeding 10%, and current technologies have not established a correlation model between signal quality and the confidence level of monitoring results, resulting in a high false alarm rate.

[0005] Static threshold strategies lack adaptability:

[0006] Current algorithms mostly use fixed thresholds for signal segmentation and feature extraction. However, the abrupt changes in sleep breathing vary among infants of different ages, and even among premature infants, and their body movements also have inconsistent effects on signal features. Therefore, the static threshold setting strategy of traditional algorithms does not meet the needs of infant sleep safety monitoring. Summary of the Invention

[0007] In view of the aforementioned problems, this application is made to provide a method and apparatus for detecting infant sleep based on multimodal signals and dynamic thresholds, which overcomes or at least partially solves the aforementioned problems, comprising:

[0008] A method for detecting infant sleep based on multimodal signals and dynamic thresholds, the method acquiring signals through a dual-wavelength photoplethysmography (PPG) sensor and a triaxial accelerometer sensor worn on the human body, including the following steps:

[0009] Acquire photoplethysmography (PPG) signals and acceleration signals from the acquisition area;

[0010] The photoplethysmography (PPG) signal and the acceleration signal are optimized to obtain optimized PPG signal and optimized acceleration signal; wherein, the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment;

[0011] The infant's posture is determined based on the optimized acceleration signal;

[0012] The monitoring risk level is determined based on the infant's body position and the optimized photoplethysmography signal.

[0013] Further, the step of optimizing the photoplethysmography (PPG) signal and the acceleration signal to obtain optimized PPG signal and optimized acceleration signal; wherein the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment, includes:

[0014] The photoplethysmography (PPG) signal and the acceleration signal are filtered to obtain a filtered PPG signal and a filtered acceleration signal.

[0015] Generate the corresponding standard deviation based on the filtered acceleration signal;

[0016] The weights of the filtered acceleration signal are determined based on a preset standard deviation threshold and the standard layer.

[0017] The kurtosis of the filtered acceleration signal is calculated to obtain a statistical score for the filtered acceleration signal.

[0018] Furthermore, it also includes:

[0019] The frequency domain information of the filtered photoplethysmography (PPG) signal and the filtered acceleration signal is determined based on the filtered PPG signal and the filtered acceleration signal.

[0020] The frequency aliasing of the filtered photoplethysmography pulse wave signal and the filtered acceleration signal is determined based on the frequency domain information.

[0021] Frequency domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal are generated based on the frequency aliasing.

[0022] Furthermore, it also includes:

[0023] The quality of the filtered photoplethysmography (PPG) signal is evaluated by cross-correlation based on the filtered acceleration signal, and the normalized values ​​of the filtered PPG signal and the filtered acceleration signal are obtained.

[0024] The time-domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal are determined based on the normalized value.

[0025] Based on the weights, the frequency domain score, the time domain score, and the statistical score, a quality assessment score corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal is obtained.

[0026] The filtered photoplethysmography (PPG) signal and the filtered acceleration signal are screened based on a preset signal quality threshold and the quality evaluation score to obtain optimized PPG signal and optimized acceleration signal.

[0027] Furthermore, the step of determining the infant's posture based on the optimized acceleration signal includes:

[0028] The optimized acceleration signal is processed by calculating the elevation angle of the triaxial acceleration signal to obtain the corresponding sleep posture data;

[0029] Based on the sleep posture data and sleep thresholds, the infant's posture is determined; wherein, the infant's posture is either prone sleeping or non-prone sleeping.

[0030] Furthermore, the step of determining the monitoring risk level based on the infant's body posture and the optimized photoplethysmography signal includes:

[0031] When the infant is in a prone position, the blood oxygen saturation value is determined based on the optimized photoplethysmography signal.

[0032] The envelope curve of the optimized photoplethysmography (PPG) signal is obtained by performing a Hilbert transform on the optimized PPG signal, and the respiratory rate value is obtained through the envelope curve.

[0033] Based on the optimized acceleration signal and the preset infant age, the dynamic threshold learning rate, dynamic threshold time window, and minimum breathing judgment threshold of the acceleration signal are determined, and the dynamic amplitude of breathing is determined through the optimized acceleration signal, the dynamic threshold learning rate of the acceleration signal, the dynamic threshold time window, and the minimum breathing judgment threshold.

[0034] The standard deviation of the optimized acceleration is determined based on the optimized acceleration signal, and the respiratory history amplitude is determined by the standard deviation and a preset infant sleep judgment threshold.

[0035] The respiratory state is determined based on the minimum respiratory judgment threshold, the dynamic respiratory amplitude, and the historical respiratory amplitude.

[0036] The monitoring risk level is determined based on the blood oxygen saturation value and the respiratory status.

[0037] Furthermore, the step of determining the monitoring risk level based on the infant's body posture and the optimized photoplethysmography signal includes...

[0038] When the infant is not in a prone sleeping position, the blood oxygen saturation value and blood oxygen perfusion are determined based on the optimized photoplethysmography signal.

[0039] The pulse rate value is obtained by effectively monitoring the peak pulse wave of the optimized photoplethysmography pulse wave signal using the differential threshold method.

[0040] The envelope curve of the optimized photoplethysmography (PPG) signal is obtained by performing a Hilbert transform on the optimized PPG signal, and the respiratory rate value is obtained through the envelope curve.

[0041] The monitoring risk level is determined based on the blood oxygen saturation value, the blood oxygen perfusion, the pulse rate value, and the respiratory rate value.

[0042] A device for detecting infant sleep based on multimodal signals and dynamic thresholds, comprising:

[0043] The acquisition module is used to acquire the photoplethysmography (PPG) signal and acceleration signal of the acquisition area;

[0044] An optimization module is used to optimize the photoplethysmography (PPG) signal and the acceleration signal to obtain optimized PPG signal and optimized acceleration signal.

[0045] A posture determination module is used to determine the infant's posture based on the optimized acceleration signal;

[0046] The monitoring module is used to determine the monitoring risk level based on the infant's body posture and the optimized photoplethysmography signal.

[0047] A device for detecting infant sleep based on multimodal signals and dynamic thresholds includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, it implements the method for detecting infant sleep based on multimodal signals and dynamic thresholds as described above.

[0048] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for detecting infant sleep based on multimodal signals and dynamic thresholds as described above.

[0049] This application has the following advantages:

[0050] In the embodiments of this application, compared with the shortcomings of the prior art, which does not evaluate physiological signals, has large noise interference, and uses only static thresholds with low adaptability, this application provides a solution for evaluating infant sleep by acquiring photoplethysmography (PPG) signals and acceleration signals and processing and optimizing the signals. Specifically, it is a method for detecting infant sleep based on multimodal signals and dynamic thresholds. The method acquires signals through a dual-wavelength PPG sensor and a triaxial accelerometer sensor worn on the human body, including the steps of: acquiring PPG signals and acceleration signals in the acquisition area; optimizing the PPG signals and acceleration signals to obtain optimized PPG signals and optimized acceleration signals; wherein the optimized PPG signals and optimized acceleration signals are optimized signals screened through filtering and multimodal signal quality assessment; determining the infant's body position based on the optimized acceleration signals; and determining the monitoring risk level based on the infant's body position and the optimized PPG signals. Optimized photoplethysmography (PPG) and acceleration signals are obtained through optimization. The monitoring signals are then filtered and optimized to improve the ability to resist noise interference and ensure the accuracy of the monitoring results. At the same time, the threshold can be dynamically adjusted to improve applicability. Attached Figure Description

[0051] To more clearly illustrate the technical solution of this application, the drawings used in the description of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating the steps of a method for detecting infant sleep based on multimodal signals and dynamic thresholds, as provided in an embodiment of this application.

[0053] Figure 2 This is a flowchart illustrating a method for detecting infant sleep based on multimodal signals and dynamic thresholds, provided in an embodiment of this application.

[0054] Figure 3 This is a flowchart illustrating a method for multidimensional signal quality assessment based on multimodal signals and dynamic threshold detection of infant sleep, provided in one embodiment of this application.

[0055] Figure 4 This is a flowchart illustrating the dynamic respiratory threshold assessment of a method for detecting infant sleep based on multimodal signals and dynamic thresholds, provided in one embodiment of this application.

[0056] Figure 5This is a flowchart illustrating the physiological indicator monitoring and alerting process of a method for detecting infant sleep based on multimodal signals and dynamic thresholds, provided in an embodiment of this application.

[0057] Figure 6 This is a flowchart illustrating a method for monitoring and alerting sleep breathing status in infants based on multimodal signals and dynamic threshold detection, provided in one embodiment of this application.

[0058] Figure 7 This is a structural block diagram of a device for detecting infant sleep based on multimodal signals and dynamic thresholds, provided in one embodiment of this application.

[0059] Figure 8 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0060] To make the objectives, features, and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0061] The inventors, through analysis of existing technologies, discovered that current infant sleep safety monitoring systems lack a real-time assessment mechanism for the quality of raw signals and fail to distinguish between valid physiological signals and noise interference (such as motion artifacts and electromagnetic interference). Clinical studies have shown that the decrease in PPG signal quality caused by infant movement during sleep can lead to blood oxygenation measurement errors exceeding 10%, and existing technologies have not established a correlation model between signal quality and the confidence level of monitoring results, resulting in a high false alarm rate.

[0062] Reference Figure 1 This application illustrates a method and apparatus for detecting infant sleep based on multimodal signals and dynamic thresholds, comprising:

[0063] A method for detecting infant sleep based on multimodal signals and dynamic thresholds, the method acquiring signals through a dual-wavelength photoplethysmography (PPG) sensor and a triaxial accelerometer sensor worn on the human body, including:

[0064] S110. Acquire the photoplethysmography (PPG) signal and acceleration signal of the acquisition area;

[0065] S120. The photoplethysmography (PPG) signal and the acceleration signal are optimized to obtain optimized PPG signal and optimized acceleration signal; wherein, the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment.

[0066] S130. Determine the infant's posture based on the optimized acceleration signal;

[0067] S140. Determine the monitoring risk level based on the infant's body position and the optimized photoplethysmography signal.

[0068] In the embodiments of this application, compared with the shortcomings of the prior art, which does not evaluate physiological signals, has large noise interference, and uses only static thresholds with low adaptability, this application provides a solution for evaluating infant sleep by acquiring photoplethysmography (PPG) signals and acceleration signals and processing and optimizing the signals. Specifically, it is a method for detecting infant sleep based on multimodal signals and dynamic thresholds. The method acquires signals through a dual-wavelength PPG sensor and a triaxial accelerometer sensor worn on the human body, including the steps of: acquiring PPG signals and acceleration signals in the acquisition area; optimizing the PPG signals and acceleration signals to obtain optimized PPG signals and optimized acceleration signals; wherein the optimized PPG signals and optimized acceleration signals are optimized signals screened through filtering and multimodal signal quality assessment; determining the infant's body position based on the optimized acceleration signals; and determining the monitoring risk level based on the infant's body position and the optimized PPG signals. Optimized photoplethysmography (PPG) and acceleration signals are obtained through optimization. The monitoring signals are then filtered and optimized to improve the ability to resist noise interference and ensure the accuracy of the monitoring results. At the same time, the threshold can be dynamically adjusted to improve applicability.

[0069] The following will further describe a method for detecting infant sleep based on multimodal signals and dynamic thresholds in this exemplary embodiment.

[0070] As described in step S110, the photoplethysmography (PPG) signal and acceleration signal of the acquisition area are obtained.

[0071] It should be noted that the PPG signal, i.e., the photoplethysmography signal, of the infant's wearing area is obtained through a dual-wavelength PPG (Photoplethysmography) sensor, and the ACC signal, i.e., the acceleration signal, of the infant's wearing area is obtained through a three-axis ACC (Accelerometer) sensor.

[0072] In a specific implementation, Figure 2In the "ACC & PPG Acquisition" section, the default example is to acquire PPG signals from the infant's ankle and ACC signals from the corresponding wrist. However, the two signal acquisition modules can also be located at different parts of the infant's body. For example, acquiring PPG signals at the ankle and ACC signals at the chest is also applicable to the monitoring method proposed in this application. The PPG signal acquisition module includes red and infrared light emitting and receiving sensors to obtain the absorption of red and infrared light by the blood after it passes through the infant's acquisition site. The ACC acquisition module uses a triaxial accelerometer to acquire the direction of motion or force in three-dimensional space at the infant's measurement site, to obtain the infant's body movement and body movements caused by respiration.

[0073] The example provided in this application is to collect photoplethysmography (PPG) signals from the infant's ankle and acceleration signals from the corresponding wrist. However, the two signal acquisition modules can also be located at different parts of the infant's body. For example, collecting PPG signals from the ankle and acceleration signals from the chest are also applicable to the monitoring method proposed in this application.

[0074] As described in step S120, the photoplethysmography (PPG) signal and the acceleration signal are optimized to obtain optimized PPG signal and optimized acceleration signal; wherein, the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment.

[0075] It should be noted that the acquired acceleration signal and photoplethysmography (PPG) signal are filtered to remove baseline interference and high-frequency noise. The filtered signal is then framed with a fixed window length.

[0076] Multimodal signal quality estimation is performed on the processed acceleration signal and photoplethysmography signal to eliminate poor signal quality caused by infant movement, sensor detachment, sensor failure, etc., and to avoid false detections in infant sleep safety monitoring.

[0077] In one embodiment of the present invention, the specific process of step S120, which involves "optimizing the photoplethysmography (PPG) signal and the acceleration signal to obtain an optimized PPG signal and an optimized acceleration signal, wherein the optimized PPG signal and the optimized acceleration signal are optimized signals selected through filtering and multimodal signal quality assessment," can be further explained in conjunction with the following description.

[0078] As described in the following steps, the photoplethysmography (PPG) signal and the acceleration signal are filtered to obtain a filtered PPG signal and a filtered acceleration signal.

[0079] It should be noted that after smoothing filtering of the sampling rate window length for both the acceleration signal and the photoplethysmography (PPG) signal to remove some baseline interference, FIR (Finite Impulse Response) low-pass filtering is performed. Specifically, the passband for the acceleration signal is 10Hz, and the passband for the PPG signal is 5Hz, to remove high-frequency noise from both signals. After filtering, the two signals are framed according to the set time window and step size to obtain the data segments required for subsequent operations, namely the filtered PPG signal and the filtered acceleration signal.

[0080] In a specific implementation, Figure 2 In the "Signal Preprocessing" section, the specific implementation involves filtering and framing the acquired PPG and ACC signals. A preset example uses a sampling frequency of 100Hz for both physiological signals. Smoothing filtering of the PPG and ACC signals is performed within a preset 2-second window. This operation can be expressed as:

[0081]

[0082] Among them, X no_DC X represents the data after smoothing and filtering. i The current data is represented by L, which is the preset time window length. Baseline interference in the ACC and PPG signals is filtered out separately using a smoothing filter.

[0083] After smoothing and filtering, the ACC and PPG signals are respectively subjected to FIR low-pass filtering. Specifically, the passband of the ACC signal is 10Hz, and the passband of the PPG signal is 5Hz to filter out high-frequency noise in both signals. After filtering, the two data are segmented according to a preset time window (5 seconds), with a window sliding step of 2 seconds. In the following content, the filtered and framed ACC and PPG signals will be referred to as ACC segments and PPG segments, respectively.

[0084] As described in the following steps, a corresponding standard deviation is generated based on the filtered acceleration signal;

[0085] As described in the following steps, the weights of the filtered acceleration signal are determined based on a preset standard deviation threshold and the standard layer;

[0086] It should be noted that the standard deviation of the processed ACC data segment is calculated, and the weights Q0, Q1, and Q2 for multi-dimensional signal quality assessment are determined based on the set ACC motion standard deviation threshold.

[0087] As described in the following steps, the kurtosis of the filtered acceleration signal is calculated to obtain a statistical score for the filtered acceleration signal.

[0088] It should be noted that the kurtosis of the processed acceleration signal segment is calculated, and the resulting statistical score, Kur_scole, is calculated for that data segment.

[0089] In one embodiment of the present invention, it further includes:

[0090] As described in the following steps, the frequency domain information of the filtered photoplethysmography (PPG) signal and the filtered acceleration signal are determined based on the PPG signal and the filtered acceleration signal.

[0091] As described in the following steps, the frequency aliasing of the filtered photoplethysmography pulse wave signal and the filtered acceleration signal is determined based on the frequency domain information;

[0092] As described in the following steps, frequency domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal are generated based on the frequency aliasing.

[0093] As an example, an FFT (Fast Fourier Transform) is performed on the processed acceleration signal and the photoplethysmography pulse wave signal to obtain information about the two signals in the frequency domain, and the frequency aliasing of the two signals in the range of 1Hz to 5Hz is calculated, and their frequency domain score Fre_scole is calculated.

[0094] In one embodiment of the present invention, it further includes:

[0095] As described in the following steps, the quality of the filtered photoplethysmography (PPG) signal is evaluated by cross-correlation based on the filtered acceleration signal to obtain the normalized values ​​of the filtered PPG signal and the filtered acceleration signal.

[0096] As described in the following steps, the time-domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal are determined based on the normalized value;

[0097] It should be noted that a time-domain signal quality assessment based on cross-correlation is performed on the processed acceleration signal and photoplethysmography (PPG) signal, where the signal to be evaluated is the PPG signal and the reference signal is the acceleration signal. The cross-correlation coefficient between the two is calculated, and their peak values ​​are located using the differential thresholding method. The normalized value of the cross-correlation peak value between the acceleration signal and the PPG signal is calculated and used as the time-domain score (Time_scole) for this data segment.

[0098] As described in the following steps, a quality assessment score corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal is obtained based on the weight, the frequency domain score, the time domain score and the statistical score;

[0099] As described in the following steps, the filtered photoplethysmography (PPG) signal and the filtered acceleration signal are screened according to a preset signal quality threshold and the quality evaluation score to obtain optimized PPG signal and optimized acceleration signal.

[0100] It should be noted that, by combining the weights Q0, Q1, Q2 obtained above with the statistical score Kur_scole, the frequency domain signal quality score Fre_scole, and the time domain signal quality score Time_scole, the final signal quality assessment score SQI of the data segment is calculated, and compared with the preset signal quality threshold to confirm the subsequent use of the data segment.

[0101] In a specific implementation, Figure 2 The specific implementation method in the "Multi-dimensional Signal Quality Assessment" section is as follows: Figure 3 The diagram shown is a schematic representation of a method for multimodal signal quality assessment provided in this application. In this step, the standard deviation of the preprocessed ACC segment is first calculated. Based on a set ACC motion standard deviation threshold, multi-dimensional signal quality assessment weights Q0, Q1, and Q2 are assigned. These weights are determined by the infant's motion state. If the ACC segment standard deviation is greater than the set motion threshold, the infant is considered to be in a non-stationary state, and Q0 is increased while the weights of Q1 and Q2 are decreased. The standard deviation calculation formula is as follows:

[0102]

[0103] Here, x i For each ACC data point within a data segment, mean(x) is the mean of the ACC data within that segment. Specifically, Q0, Q1, and Q2 satisfy the condition that the sum of these three values ​​is 1.

[0104] In "Multi-Dimensional Signal Quality Assessment," the kurtosis of the ACC segment is calculated. Kurtosis is an important statistic describing the shape of data distribution, primarily used to measure the "peaking" and "tail thickness" of the data distribution. The formula for calculating kurtosis is:

[0105]

[0106] Here, Kur represents the calculated kurtosis, and x i For each ACC data point within a data segment, mean(x) is the mean of the ACC data within that segment.

[0107] The kurtosis of the ACC segment is calculated, and based on the historically recorded maximum kurtosis of the ACC segment, the statistical signal quality score is calculated using the following formula:

[0108] Kur_scole = Kur / max(Kur) (4)

[0109] Here, Kur_scole is the calculated statistical score, and max(Kur) is the maximum kurtosis of the historical ACC segment. The score reflects the variability of the ACC data; for drastic variables, the closer the calculated kurtosis is to the maximum kurtosis, the poorer the data quality.

[0110] In the "Multi-Dimensional Signal Quality Assessment," the ACC and PPG segments undergo FFT, a classic method for signal frequency domain analysis. The FFT obtains the frequency domain information of the two signals, calculates the frequency aliasing of the two signal segments within the 1Hz to 5Hz range, and then calculates their frequency domain signal quality score. The formula is as follows:

[0111]

[0112] Here, PPG refers to the PPG segment of this data segment, ACC refers to the ACC segment of this data segment, FFT refers to Fast Fourier Transform, and "∩" represents the product of the aliasing regions. The calculated frequency domain signal quality score reflects the overlap between the ACC and PPG data. A higher overlap ratio indicates stronger interference from ACC data, suggesting poor data quality.

[0113] In the "Multi-Dimensional Signal Quality Assessment," the ACC segment is used as the reference signal to calculate the cross-correlation coefficient of the PPG signal. The cross-correlation coefficient is an important indicator that measures the degree of linear correlation and time delay relationship between two signals or variables, and is widely used in signal processing, statistics, engineering, and other fields. Here, the calculation formula is:

[0114]

[0115] Here, R xy (τ) is the calculated cross-correlation coefficient, N is the length of the two signal segments used for calculation, and τ is the preset delay.

[0116] After calculating the cross-correlation coefficients of the two signal segments, the maximum peak value of the cross-correlation coefficients is located using a thresholding method, and the normalized matching degree of the signals is calculated to obtain the time-domain signal quality score. The formula is as follows:

[0117]

[0118] Here, Time_scole is the calculated time-domain signal quality score, τ peak R is the location of the maximum peak value of the cross-correlation coefficient obtained from the location calculation. yy (0) represents the value of the autocorrelation of the reference signal, i.e., the ACC segment, at a delay of 0. The autocorrelation calculation can be viewed as calculating the cross-correlation coefficient of the ACC signal with the ACC segment as the reference signal. The time-domain signal quality score reflects the similarity between the ACC data and the PPG data. If the ratio is larger, it is judged that the PPG data is suffering from strong ACC data interference, and the data quality is considered poor.

[0119] In the "Multi-Dimensional Signal Quality Assessment," the signal quality scores obtained from three different calculation dimensions are combined with the scoring weights derived from the standard deviation of the ACC signal to obtain the multi-dimensional signal quality assessment score. The calculation formula is as follows:

[0120] SQI_scole=Q0*Kur_scole+Q1*Fre_scole+Q2*Time_scole(8)

[0121] Here, SQI_scole is the calculated multi-dimensional signal quality assessment score.

[0122] like Figure 2 As shown, the data segments undergo multi-dimensional signal quality evaluation. Data segments with signal quality scores exceeding the set quality threshold are sent to subsequent processing, while data segments that fail are discarded.

[0123] In one embodiment of the present invention, the specific process of "determining the infant's posture based on the optimized acceleration signal" in step S130 can be further described in conjunction with the following description.

[0124] As described in the following steps, the optimized acceleration signal is processed by calculating the elevation angle of the triaxial acceleration signal to obtain the corresponding sleep posture data;

[0125] As described in the following steps, the infant's body position is determined based on the sleep posture data and sleep thresholds; wherein the infant's body position is either prone sleeping or non-prone sleeping.

[0126] It should be noted that the elevation angle of the three-axis acceleration signal is calculated on the optimized acceleration signal to obtain the infant's sleeping posture within the data segment. Based on the preset posture threshold, it is determined whether the infant is in a prone sleeping position.

[0127] In a specific implementation, Figure 2In the "ACC prone sleeping recognition" section, the specific implementation steps are as follows: using the triaxial acceleration signal of the ACC data segment, the infant's tilt angle is calculated, and the calculated tilt angle is compared with a preset prone sleeping angle threshold to determine whether the infant is in a prone sleeping position. The formula for calculating the tilt angle using the triaxial acceleration signal is:

[0128]

[0129] Here, θ is the calculated elevation angle, arctan2(x,y) is the four-quadrant arctangent function, which can determine the angle from the positive x-axis to the point (x,y) based on the input signals x,y, with a range between [-π,π]; accx, accy, and accz are the data of the three axes in the input ACC segment, respectively.

[0130] In one embodiment of the present invention, the specific process of "determining the monitoring risk level based on the infant's body position and the optimized photoplethysmography signal" in step S140 can be further explained in conjunction with the following description.

[0131] As described in the following steps, when the infant is in a prone sleeping position, the blood oxygen saturation value is determined based on the optimized photoplethysmography signal;

[0132] As described in the following steps, a Hilbert transform is performed on the optimized photoplethysmography (PPG) signal to obtain the envelope curve of the optimized PPG signal, and the respiratory rate value is obtained through the envelope curve;

[0133] As described in the following steps, the dynamic threshold learning rate, dynamic threshold time window, and minimum breathing judgment threshold of the acceleration signal are determined based on the optimized acceleration signal and the preset infant age in months, and the dynamic amplitude of breathing is determined through the optimized acceleration signal, the dynamic threshold learning rate of the acceleration signal, the dynamic threshold time window, and the minimum breathing judgment threshold;

[0134] As described in the following steps, the standard deviation of the optimized acceleration is determined based on the optimized acceleration signal, and the respiratory history amplitude is determined by the standard deviation and a preset infant sleep judgment threshold.

[0135] As described in the following steps, the respiratory state is determined based on the minimum respiratory judgment threshold, the dynamic respiratory amplitude, and the historical respiratory amplitude;

[0136] As described in the following steps, the monitoring risk level is determined based on the blood oxygen saturation value and the respiratory status.

[0137] It should be noted that when the infant is determined to be sleeping prone, an optimized photoplethysmography (PPG) signal is used, combined with the Lambert-Beer law, to calculate the infant's blood oxygen saturation value within that data segment. A Hilbert transform is performed on the PPG signal within the segment to obtain the PPG signal envelope curve, which serves as the blood oxygen-derived respiratory signal for that segment. Respiratory peak location is then performed to calculate the infant's respiratory rate within that data segment. Using the optimized acceleration signal within that segment, the dynamic threshold learning rate, dynamic threshold time window, and minimum respiratory judgment threshold are selected based on the user-input infant's age in months. The mean of the acceleration signal is calculated based on the matched time window, and the dynamic amplitude of the infant's breathing within that segment is calculated. Using the optimized acceleration signal within that segment, the standard deviation of the acceleration signal within that time period is calculated. Based on a preset infant sleep judgment threshold, it is determined whether the infant is asleep within that time period, and a historical amplitude attenuation coefficient α is selected to calculate the historical amplitude of the infant's breathing within that segment. Finally, the maximum value of the minimum respiratory threshold, dynamic respiratory amplitude, and historical respiratory amplitude is used as the threshold for judging the infant's respiratory peak in the acceleration signal within this stage, thereby determining the infant's respiratory status at this time. Combined with the obtained infant's blood oxygen saturation value, infant's respiratory rate value, and infant's current respiratory status, a risk classification for infant sleep apnea monitoring is performed, and a warning is issued.

[0138] like Figure 2 As shown, when the "ACC prone sleeping recognition" part identifies that the infant is in a prone sleeping position, the PPG data segment is used to calculate the infant's blood oxygen saturation value and blood oxygen-derived respiratory rate value in this situation. The calculation method is the same as the blood oxygen saturation value and blood oxygen-derived respiratory rate value mentioned above. The ACC data segment is used to determine the infant's ACC dynamic threshold respiratory status.

[0139] In the ACC dynamic threshold respiratory status assessment, please refer to Figure 4 The specific implementation plan is as follows: First, according to the plan, the infant's age input by the user is categorized into preterm infants, 0-6 months, 6-12 months, and over 12 months. The plan then selects the ACC signal dynamic threshold learning rate K, dynamic threshold time window length Win, and minimum respiratory judgment threshold Min required for ACC dynamic threshold respiratory status judgment. Specifically, as the infant's age increases, the infant's movement amplitude and respiratory arc will change; therefore, the higher the age, the greater the values ​​of K, Win, and Min.

[0140] In the dynamic threshold respiratory status judgment of ACC, the standard deviation σ of the ACC data in the selected dynamic threshold time window is calculated according to the selected dynamic threshold time window length Win. The calculation formula is the same as that of formula (1). The obtained standard deviation is compared with the preset motion threshold, the attenuation coefficient α is selected, and the historical amplitude threshold Hist of the ACC segment is calculated. The calculation formula is as follows:

[0141] Diff=K*σ(12)

[0142] Hist = α * mean(x) Win (13)

[0143] Here, Diff is the calculated dynamic amplitude threshold, Hist is the calculated historical amplitude threshold, and x Win For the ACC segment within this time window, mean(x) Win ) represents the average value of ACC segments within this time window.

[0144] In the dynamic threshold respiratory status assessment of ACC, the dynamic threshold for infant sleep apnea monitoring is confirmed based on the obtained minimum respiratory assessment threshold Min, historical amplitude threshold Hist, and dynamic amplitude threshold Diff.

[0145] Th resp =max(Min,Diff,Hist)(14)

[0146] Here, Th resp This is the final threshold for subsequently determining the infant's sleep breathing status.

[0147] Figure 2 The specific implementation method in the "Sleep Breathing Monitoring and Alert" section is as follows: Figure 6 The diagram shown is a risk classification and warning method for monitoring infant sleep breathing status provided in an embodiment of this application.

[0148] In the sleep apnea monitoring and alert section, monitoring and alerts are specifically based on the blood oxygen saturation value, blood oxygen-derived respiratory rate value calculated from the PPG segment, and the respiratory status obtained from the ACC segment at that time. Specifically, if the blood oxygen saturation is below 94%, further respiratory monitoring is performed; otherwise, no alert is output.

[0149] In the sleep apnea monitoring and alert section, specifically, dual-source respiratory monitoring is performed when blood oxygen saturation is below 94%. Using the blood oxygen-derived respiratory signal from the PPG segment, the respiratory peak is located using a differential thresholding method, and the energy of the respiratory wave is calculated. Invalid respiratory waves are counted according to a preset respiratory wave energy threshold. If consecutive invalid respiratory waves exceed a set time length (20 seconds) or the blood oxygen-derived respiratory rate is continuously lower than a set respiratory rate (10 breaths / minute) for more than a set time length (20 seconds), a PPG source respiratory asphyxia alert is output. For the ACC segment, a differential thresholding method is applied to determine if the peak value in the three-axis ACC data exceeds the dynamic threshold of the ACC respiratory state. If so, the infant is considered to be in a non-asphyxiation state; otherwise, an ACC respiratory asphyxia count is initiated. When an ACC respiratory asphyxia count is initiated, an ACC source infant asphyxia alert is output.

[0150] In the sleep apnea monitoring and alert section, specifically, when blood oxygen saturation is below 94%, if neither the PPG nor ACC source outputs a corresponding alert, and the duration of low blood oxygen saturation exceeds the set duration, a "low blood oxygen saturation" alert will be output. If only a respiratory asphyxia alert from the PPG source or an infant asphyxia alert from the ACC source is present, a moderate alert will be output; if both a respiratory asphyxia alert from the PPG source and an infant asphyxia alert from the ACC source are present simultaneously, a severe alert will be output.

[0151] Specifically, the warnings output in the sleep breathing monitoring and warning section and the physiological indicator monitoring and warning section are implemented by outputting to the locally integrated LED and buzzer for warning reminders. In particular, this function can also be used for remote reminders via Bluetooth or other transmission methods.

[0152] In one embodiment of the present invention, the specific process of "determining the monitoring risk level based on the infant's body position and the optimized photoplethysmography signal" in step S140 can be further explained in conjunction with the following description.

[0153] As described in the following steps, when the infant is not in a prone sleeping position, the blood oxygen saturation value and blood oxygen perfusion are determined based on the optimized photoplethysmography signal;

[0154] As described in the following steps, the pulse rate value is obtained by effectively monitoring the peak pulse of the optimized photoplethysmography pulse wave signal using the differential threshold method.

[0155] As described in the following steps, a Hilbert transform is performed on the optimized photoplethysmography (PPG) signal to obtain the envelope curve of the optimized PPG signal, and the respiratory rate value is obtained through the envelope curve;

[0156] As described in the following steps, the monitoring risk level is determined based on the blood oxygen saturation value, the blood oxygen perfusion, the pulse rate value, and the respiratory rate value.

[0157] It should be noted that when the infant is identified as not in a prone sleeping stage, an optimized photoplethysmography (PPG) signal is used, combined with the Lambert-Beer law, to calculate the infant's blood oxygen saturation and blood oxygen perfusion within that data segment. The PPG signal within this stage is then monitored using a differential thresholding method to calculate the infant's pulse rate within that data segment. The PPG signal within this stage is then subjected to a Hilbert transform to obtain the envelope curve of the PPG signal, which serves as the blood oxygen-derived respiratory signal for that segment. The respiratory wave peak is then located to calculate the infant's respiratory rate within that data segment. Combining the calculated infant's blood oxygen saturation, blood oxygen perfusion, pulse rate, and respiratory rate, and based on pre-set moderate and severe warning thresholds for corresponding infant physiological indicators, risk grading and warnings for infant sleep respiratory physiological indicator monitoring are established.

[0158] like Figure 2 As shown, when the "ACC prone sleeping recognition" section identifies the infant as not in a prone sleeping position, it uses PPG segments to calculate the infant's blood oxygen saturation, pulse rate, and oxygen-derived respiratory rate at that stage. In the PPG segment, effective pulse waves are first detected. Differential operations are performed on the PPG signal to obtain the rising and falling edge slopes for effective pulse wave matching. Based on the located effective pulse waves, the average pulse interval is calculated, and the pulse rate is obtained. Based on the number of effective pulse waves calculated and after passing a set threshold, the PPG signal undergoes bandpass filtering and notch filtering. The DC and AC components of the PPG signal are separated according to signal composition. The blood perfusion rate is calculated using the ratio of the DC to AC components, and the blood oxygen saturation value is calculated according to Lambert-Beer's law.

[0159] In the PPG segment, the signal is subjected to a Hilbert transform, and the transform formula is as follows:

[0160] z[k]=x[k]+j*δ{x[k]}(10)

[0161] Where x[k] is the PPG signal, δ is the Hilbert operator, and z[k] is the Hilbert analytic signal. The instantaneous amplitude of the analytic signal is calculated to obtain the envelope curve, which is then used to derive the respiratory signal from the pulse wave.

[0162]

[0163] Differential operation is performed on the pulse wave-derived respiratory signal to obtain the rising slope and falling slope for effective respiratory wave matching. Based on the located effective respiratory wave, the average respiratory wave interval is calculated to obtain the blood oxygen-derived respiratory rate.

[0164] Figure 2 The specific implementation method in the "Physiological Indicator Monitoring and Warning" section is as follows: Figure 5 The diagram shown is a risk grading and warning method for monitoring infant physiological indicators provided in this application embodiment. Here, the previously calculated blood oxygen saturation value, pulse rate value, blood perfusion value, and blood oxygen-derived respiratory rate value are used to determine physiological thresholds.

[0165] Specifically, for blood oxygen saturation, if its value is lower than the set first alarm threshold and remains below it for more than a preset time (10 seconds), a moderate physiological alert for blood oxygen saturation will be issued. If its value is lower than the set second alarm threshold and remains below it for more than a preset time (10 seconds), a severe physiological alert for blood oxygen saturation will be issued. Otherwise, no blood oxygen saturation alert will be output.

[0166] Specifically, for other indicators, such as pulse rate, blood perfusion, and oxygen-derived respiratory rate, the alarm logic is similar to that of oxygen saturation, but the first and second alarm thresholds set for each physiological indicator are different.

[0167] As an example, a method for detecting infant sleep based on multimodal signals and dynamic thresholds proposed in this application includes the following steps:

[0168] PPG signals from the area where the baby wears the device are acquired using a dual-wavelength PPG sensor, and ACC signals from the same area are acquired using a triaxial ACC sensor. The acquired signals are then transmitted to the signal processing module via local transmission or Bluetooth.

[0169] The PPG and ACC signals are filtered separately. A smoothing filter is used to remove baseline interference, and an FIR low-pass filter is used to remove high-frequency interference. The filtered signals are then processed into frames.

[0170] The two signals are scored statistically, in the time domain, and in the frequency domain. The weights of the three signal quality scores are determined based on the standard deviation of the ACC signal. The final signal quality assessment is then performed, and low-quality signals are eliminated.

[0171] Using high-quality ACC signals to determine an infant's sleeping posture, the infant's sleeping supine angle is calculated by calculating the ACC signals of three axes to determine whether the infant is in the prone sleeping stage.

[0172] During the infant's prone sleeping phase, a high-quality PPG signal was used, combined with Beer-Lambert's law, to calculate the infant's blood oxygen saturation value within that data segment. A Hilbert transform was performed on the PPG signal within that segment to obtain its envelope curve, which served as the blood oxygen-derived respiratory signal for that segment. The respiratory wave peak was then located, and the infant's respiratory rate value within that data segment was calculated.

[0173] During the infant's prone sleeping phase, the relevant processing time window length and some thresholds for the ACC signal are selected based on the infant's age in months. Historical and dynamic amplitude thresholds of the ACC signal are also calculated to monitor and assess the infant's respiratory status. Furthermore, the infant's blood oxygen saturation, respiratory rate, and respiratory status are used in conjunction to classify the risk of infant sleep apnea and issue warnings.

[0174] During the non-prone sleeping stage of the infant, high-quality PPG signals were used, combined with Beer-Lambert's law, to calculate the infant's blood oxygen saturation and blood perfusion values ​​within the data segment. The differential thresholding method was then used to monitor the effective pulse wave peak value, and the infant's pulse rate within the data segment was calculated.

[0175] During the non-prone sleeping stage of the infant, the PPG signal within this stage is subjected to Hilbert transform to obtain the envelope curve of the PPG signal, which serves as the blood oxygen-derived respiratory signal for that segment. The peak value of the respiratory wave is then located, and the infant's respiratory rate value within this data segment is calculated. Combined with the calculated infant's blood oxygen saturation, blood oxygen perfusion, pulse rate, and respiratory rate values, and based on the set moderate and severe warning thresholds for corresponding infant physiological indicators, risk classification and warnings for infant sleep physiological indicator monitoring are established.

[0176] The innovation of this example lies in its rational algorithm design, which only requires the acquisition of PPG and ACC signals. Through multi-dimensional signal quality assessment, the validity of data in infant sleep safety monitoring is ensured, and monitoring accuracy is improved. Dynamic thresholding is used to reasonably modify sleep safety monitoring thresholds for different ages and sleep stages of infants, further enhancing monitoring accuracy. The PPG and ACC signals acquired in this application utilize sensor technology that has been clinically validated in existing mature solutions and is safe for measuring infant physiological signals. The part of the sensor that contacts the infant's skin is made of medical-grade flexible material, soft and skin-friendly, avoiding irritation or pressure on the infant's delicate skin. Simultaneously, the overall design of the acquisition device is lightweight, compact, and easy to wear, minimizing restrictions on infant movement while ensuring stable signal acquisition, achieving safe and harmless continuous signal acquisition. Compared to traditional infant sleep breathing monitoring methods, this solution significantly improves monitoring accuracy in complex scenarios while reducing the false positive rate. Non-invasive acquisition and low-power design ensure infant sleep comfort and device battery life, providing a safe, reliable, and efficient solution for infant sleep health monitoring.

[0177] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0178] Reference Figure 7 This application illustrates an embodiment of a device for detecting infant sleep based on multimodal signals and dynamic thresholds.

[0179] Specifically, it includes:

[0180] The acquisition module 710 is used to acquire the photoplethysmography (PPG) signal and acceleration signal of the acquisition area;

[0181] The optimization module 720 is used to optimize the photoplethysmography (PPG) signal and the acceleration signal to obtain optimized PPG signal and optimized acceleration signal; wherein the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment.

[0182] The posture determination module 730 is used to determine the infant's posture based on the optimized acceleration signal;

[0183] The monitoring module 740 is used to determine the monitoring risk level based on the infant's body posture and the optimized photoplethysmography signal.

[0184] In one embodiment of the present invention, the optimization module 720 includes:

[0185] The filtering submodule is used to filter the photoplethysmography (PPG) signal and the acceleration signal to obtain a filtered PPG signal and a filtered acceleration signal.

[0186] The standard layer submodule is used to generate a corresponding standard deviation based on the filtered acceleration signal;

[0187] The weighting submodule is used to determine the weights of the filtered acceleration signal based on a preset standard deviation threshold and the standard layer.

[0188] The statistical molecular module is used to calculate the kurtosis of the filtered acceleration signal and obtain a statistical score for the filtered acceleration signal.

[0189] In one embodiment of the present invention, it further includes:

[0190] The frequency domain information submodule is used to determine the frequency domain information of the filtered photoplethysmography (PPG) signal and the filtered acceleration signal based on the PPG signal and the filtered acceleration signal.

[0191] A frequency aliasing submodule is used to determine the frequency aliasing of the filtered photoplethysmography pulse wave signal and the filtered acceleration signal based on the frequency domain information.

[0192] The frequency domain molecular module is used to generate frequency domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal based on the frequency aliasing.

[0193] In one embodiment of the present invention, it further includes:

[0194] The normalization value submodule is used to perform cross-correlation quality assessment on the filtered photoplethysmography (PPG) signal based on the filtered acceleration signal to obtain the normalized values ​​of the filtered PPG signal and the filtered acceleration signal.

[0195] The time-domain numerator module is used to determine the time-domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal based on the normalized value.

[0196] The quality assessment molecular module is used to obtain a quality assessment score corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal based on the weight, the frequency domain score, the time domain score and the statistical score;

[0197] The optimization submodule is used to filter the filtered photoplethysmography (PPG) signal and the filtered acceleration signal based on a preset signal quality threshold and the quality evaluation score, so as to obtain optimized PPG signal and optimized acceleration signal.

[0198] In one embodiment of the present invention, the posture determination module 730 includes:

[0199] The body posture data submodule is used to process the optimized acceleration signal by calculating the elevation angle of the triaxial acceleration signal to obtain the corresponding sleep body posture data;

[0200] The prone sleeping determination submodule is used to determine the infant's body position based on the sleep posture data and sleep threshold; wherein, the infant's body position is prone sleeping or non-prone sleeping.

[0201] In one embodiment of the present invention, the monitoring module 740 includes:

[0202] The blood oxygen saturation submodule is used to determine the blood oxygen saturation value based on the optimized photoplethysmography signal when the infant is in a prone position.

[0203] The respiratory rate value submodule is used to perform Hilbert transform on the optimized photoplethysmography (PPG) signal to obtain the envelope curve of the optimized PPG signal, and obtain the respiratory rate value through the envelope curve.

[0204] The respiratory dynamic amplitude submodule is used to determine the dynamic threshold learning rate, dynamic threshold time window and minimum respiratory judgment threshold of the acceleration signal based on the optimized acceleration signal and the preset infant age in months, and to determine the respiratory dynamic amplitude through the optimized acceleration signal, the dynamic threshold learning rate of the acceleration signal, the dynamic threshold time window and the minimum respiratory judgment threshold;

[0205] The breathing history amplitude submodule is used to determine the standard deviation of the optimized acceleration based on the optimized acceleration signal, and to determine the breathing history amplitude through the standard deviation and a preset infant sleep judgment threshold.

[0206] The breathing state submodule is used to determine the breathing state based on the minimum breathing judgment threshold, the dynamic amplitude of breathing, and the historical amplitude of breathing.

[0207] The risk level submodule is used to determine the monitoring risk level based on the blood oxygen saturation value and the respiratory status.

[0208] In one embodiment of the present invention, the monitoring module 740 includes:

[0209] The blood oxygen perfusion submodule is used to determine the blood oxygen saturation value and blood oxygen perfusion based on the optimized photoplethysmography signal when the infant's body position is not prone.

[0210] The pulse rate value submodule is used to monitor the effective pulse wave peak value of the optimized photoplethysmography pulse wave signal using the differential threshold method to obtain the pulse rate value.

[0211] The respiratory rate value submodule is used to perform Hilbert transform on the optimized photoplethysmography (PPG) signal to obtain the envelope curve of the optimized PPG signal, and obtain the respiratory rate value through the envelope curve.

[0212] The risk level submodule is used to determine the monitoring risk level based on the blood oxygen saturation value, the blood oxygen perfusion, the pulse rate value, and the respiratory rate value.

[0213] Reference Figure 8 The present invention illustrates a computer device for detecting infant sleep based on multimodal signals and dynamic thresholds, which may specifically include the following:

[0214] The computer device 12 described above is in the form of a general-purpose computing device. The components of the computer device 12 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and bus 18 connecting different system components (including system memory 28 and processing unit 16).

[0215] Bus 18 refers to one or more of several types of bus 18 architectures, including memory bus 18 or memory controller, peripheral bus 18, graphics acceleration port, processor, or local bus 18 using any of the various bus 18 architectures. For example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus 18, Micro Channel Architecture (MAC) bus 18, Enhanced ISA bus 18, Audio / Video Electronics Standards Association (VESA) local bus 18, and Peripheral Component Interconnect (PCI) bus 18.

[0216] Computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0217] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as a "hard disk drive"). Figure 8Not shown, a disk drive for reading and writing to a removable non-volatile disk (such as a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules 42 configured to perform the functions of the embodiments of the present invention.

[0218] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory. Such program modules 42 include—but are not limited to—an operating system, one or more application programs, other program modules 42, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of the present invention.

[0219] Computer device 12 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, camera, etc.), and with one or more devices that enable a user to interact with the computer device 12, and / or with any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through input / output (I / O) interface 22. Furthermore, computer device 12 can also communicate with one or more networks (e.g., local area network (LAN)), wide area network (WAN), and / or public networks (e.g., the Internet) via network adapter 20. As shown, network adapter 20 communicates with other modules of computer device 12 via bus 18. It should be understood that, although... Figure 8 Not shown, it can be combined with computer device 12 to use other hardware and / or software modules, including but not limited to: microcode, device drivers, redundant processing unit 16, external disk drive array, RAID system, tape drive and data backup storage system 34, etc.

[0220] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing a method for detecting infant sleep based on multimodal signals and dynamic thresholds provided in this embodiment of the invention.

[0221] That is, when the processing unit 16 executes the above program, it achieves the following: acquiring the photoplethysmography (PPG) signal and acceleration signal of the acquisition area;

[0222] The photoplethysmography (PPG) signal and the acceleration signal are optimized to obtain optimized PPG signal and optimized acceleration signal; wherein, the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment;

[0223] The infant's posture is determined based on the optimized acceleration signal;

[0224] The monitoring risk level is determined based on the infant's body position and the optimized photoplethysmography signal.

[0225] In this embodiment of the invention, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements a method for detecting infant sleep based on multimodal signals and dynamic thresholds as provided in all embodiments of this application:

[0226] That is, when the program is executed by the processor, it should achieve the following: acquire the photoplethysmography (PPG) wave signal and acceleration signal of the acquisition area;

[0227] The photoplethysmography (PPG) signal and the acceleration signal are optimized to obtain optimized PPG signal and optimized acceleration signal; wherein, the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment;

[0228] The infant's posture is determined based on the optimized acceleration signal;

[0229] The monitoring risk level is determined based on the infant's body position and the optimized photoplethysmography signal.

[0230] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-to-signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium that contains or stores a program that may be used by or in connection with an instruction execution system, apparatus, or device.

[0231] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0232] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider). The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments; similar or identical parts between embodiments can be referred to interchangeably.

[0233] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0234] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0235] The above provides a detailed description of the method and apparatus for detecting infant sleep based on multimodal signals and dynamic thresholds provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and its core ideas. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for detecting infant sleep based on multimodal signals and dynamic thresholds, characterized in that, The method acquires signals using a dual-wavelength photoplethysmography (PPG) sensor and a triaxial accelerometer sensor worn on the human body, and includes the following steps: Acquire photoplethysmography (PPG) signals and acceleration signals from the acquisition area; The photoplethysmography (PPG) signal and the acceleration signal are optimized to obtain optimized PPG signal and optimized acceleration signal; wherein, the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment; The infant's posture is determined based on the optimized acceleration signal; The monitoring risk level is determined based on the infant's body position and the optimized photoplethysmography signal.

2. The method according to claim 1, characterized in that, The step of optimizing the photoplethysmography (PPG) signal and the acceleration signal to obtain optimized PPG signal and optimized acceleration signal; wherein the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment, includes: The photoplethysmography (PPG) signal and the acceleration signal are filtered to obtain a filtered PPG signal and a filtered acceleration signal. Generate the corresponding standard deviation based on the filtered acceleration signal; The weights of the filtered acceleration signal are determined based on a preset standard deviation threshold and the standard layer. The kurtosis of the filtered acceleration signal is calculated to obtain a statistical score for the filtered acceleration signal.

3. The method according to claim 2, characterized in that, Also includes: The frequency domain information of the filtered photoplethysmography (PPG) signal and the filtered acceleration signal is determined based on the filtered PPG signal and the filtered acceleration signal. The frequency aliasing of the filtered photoplethysmography pulse wave signal and the filtered acceleration signal is determined based on the frequency domain information. Frequency domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal are generated based on the frequency aliasing.

4. The method according to claim 3, characterized in that, Also includes: The quality of the filtered photoplethysmography (PPG) signal is evaluated by cross-correlation based on the filtered acceleration signal, and the normalized values ​​of the filtered PPG signal and the filtered acceleration signal are obtained. The time-domain scores corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal are determined based on the normalized value. Based on the weights, the frequency domain score, the time domain score, and the statistical score, a quality assessment score corresponding to the filtered photoplethysmography pulse wave signal and the filtered acceleration signal is obtained. The filtered photoplethysmography (PPG) signal and the filtered acceleration signal are screened based on a preset signal quality threshold and the quality evaluation score to obtain optimized PPG signal and optimized acceleration signal.

5. The method according to claim 1, characterized in that, The step of determining the infant's posture based on the optimized acceleration signal includes: The optimized acceleration signal is processed by calculating the elevation angle of the triaxial acceleration signal to obtain the corresponding sleep posture data; Based on the sleep posture data and sleep thresholds, the infant's posture is determined; wherein, the infant's posture is either prone sleeping or non-prone sleeping.

6. The method according to claim 1, characterized in that, The step of determining the monitoring risk level based on the infant's body position and the optimized photoplethysmography signal includes: When the infant is in a prone position, the blood oxygen saturation value is determined based on the optimized photoplethysmography signal. The envelope curve of the optimized photoplethysmography (PPG) signal is obtained by performing a Hilbert transform on the optimized PPG signal, and the respiratory rate value is obtained through the envelope curve. Based on the optimized acceleration signal and the preset infant age, the dynamic threshold learning rate, dynamic threshold time window, and minimum breathing judgment threshold of the acceleration signal are determined, and the dynamic amplitude of breathing is determined through the optimized acceleration signal, the dynamic threshold learning rate of the acceleration signal, the dynamic threshold time window, and the minimum breathing judgment threshold. The standard deviation of the optimized acceleration is determined based on the optimized acceleration signal, and the respiratory history amplitude is determined by the standard deviation and a preset infant sleep judgment threshold. The respiratory state is determined based on the minimum respiratory judgment threshold, the dynamic respiratory amplitude, and the historical respiratory amplitude. The monitoring risk level is determined based on the blood oxygen saturation value and the respiratory status.

7. The method according to claim 1, characterized in that, The step of determining the monitoring risk level based on the infant's body position and the optimized photoplethysmography signal includes: When the infant is not in a prone sleeping position, the blood oxygen saturation value and blood oxygen perfusion are determined based on the optimized photoplethysmography signal. The pulse rate value is obtained by effectively monitoring the peak pulse wave of the optimized photoplethysmography pulse wave signal using the differential threshold method. The envelope curve of the optimized photoplethysmography (PPG) signal is obtained by performing a Hilbert transform on the optimized PPG signal, and the respiratory rate value is obtained through the envelope curve. The monitoring risk level is determined based on the blood oxygen saturation value, the blood oxygen perfusion, the pulse rate value, and the respiratory rate value.

8. A device for detecting infant sleep based on multimodal signals and dynamic thresholds, characterized in that, The device acquires signals through a dual-wavelength photoplethysmography (PPG) sensor and a triaxial accelerometer sensor worn on the human body, including: The acquisition module is used to acquire the photoplethysmography (PPG) signal and acceleration signal of the acquisition area; An optimization module is used to optimize the photoplethysmography (PPG) signal and the acceleration signal to obtain optimized PPG signal and optimized acceleration signal; wherein the optimized PPG signal and the optimized acceleration signal are optimized signals screened through filtering and multimodal signal quality assessment; A posture determination module is used to determine the infant's posture based on the optimized acceleration signal; The monitoring module is used to determine the monitoring risk level based on the infant's body posture and the optimized photoplethysmography signal.

9. A device for detecting infant sleep based on multimodal signals and dynamic thresholds, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the method for detecting infant sleep based on multimodal signals and dynamic thresholds as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the method for detecting infant sleep based on multimodal signals and dynamic thresholds as described in any one of claims 1 to 7.

Images