Information processing method and apparatus for sleep apnea detection, device, and medium

Abstract

This application discloses an information processing method, apparatus, device, and medium for sleep apnea detection, relating to the field of health risk assessment technology and applied to a computer device. The method includes: time-aligning respiratory signals, blood oxygen saturation signals, and sleep noise signals collected during a target monitoring period to obtain a set of sleep physiological time slices; identifying a set of apnea segment ranges based on respiratory decline coefficients, performing physiological time delay calibration on blood oxygen, and extracting a set of apnea event features according to a paired proportional coefficient; connecting blood oxygen fluctuation segments within the target segments with increasing apnea duration to obtain an abnormal apnea structure sequence; identifying associated fluctuation segments of respiration and blood oxygen, connecting the blood oxygen change trend sequence with the fluctuation segments to form a risk adjustment basis set; comparing the blood oxygen change performance and trend of the segment distribution sequence, combining the change amplitude with the distribution span of apnea events, and mapping it to a preset grading interval to obtain a risk level. This effectively improves the accuracy and efficiency of sleep apnea detection.

Description

Technical Field

[0001] This invention relates to the field of health risk assessment technology, and in particular to information processing methods, devices, equipment and media for sleep apnea detection. Background Technology

[0002] Sleep apnea is a common sleep disorder characterized by repeated brief cessation of breathing during sleep, leading to decreased blood oxygen saturation, disrupted sleep structure, and other problems. Long-term episodes can easily induce serious complications such as cardiovascular and cerebrovascular diseases and cognitive impairment, posing a significant threat to human health. Therefore, accurate detection and risk classification of sleep apnea have important clinical significance and application value.

[0003] Current methods for detecting sleep apnea mostly rely on complex medical equipment such as polysomnography systems. These methods suffer from drawbacks such as high testing costs, cumbersome operation, the need for professional personnel on duty, and unsuitability for long-term home monitoring. Furthermore, some simplified testing methods only analyze a single physiological signal, lacking the synergistic utilization of multiple physiological signals. This results in poor signal synchronization, low accuracy in identifying pause segments, and a high risk of false positives and false negatives. Additionally, they struggle to capture the progressive pathological characteristics of sleep apnea and cannot achieve precise grading of the severity of the condition, thus failing to meet the practical needs of early clinical screening, home monitoring, and personalized intervention.

[0004] In summary, improving the accuracy and efficiency of sleep apnea detection is a problem that needs to be solved in this field. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide an information processing method, apparatus, device, and medium for sleep apnea detection, thereby improving the accuracy and efficiency of sleep apnea detection. The specific solution is as follows: In a first aspect, this application discloses an information processing method for sleep apnea detection, applied to a computer device, comprising: The respiratory signals, blood oxygen saturation signals, and sleep noise signals collected during the target monitoring period are time-aligned to obtain a set of sleep physiological time slices; The pause segment range set in the sleep physiological time slice set is identified based on the decrease coefficient of the respiratory signal in the sleep physiological time slice set, and the blood oxygen saturation in the pause segment range set is calibrated by physiological time delay to obtain the blood oxygen sequence set. The pairing ratio coefficient is determined based on the range length of the pause segment range set and the decrease delay of the blood oxygen sequence set. The pause event feature set is obtained from the pause segment range set based on the pairing ratio coefficient. The target segments with increasing pause duration are identified from the pause event feature set, and the fluctuation segments of blood oxygen saturation in the target segments are connected by the morphological gradient operator to obtain the pause abnormal structure sequence. Identify the correlation fluctuation segments between respiration and blood oxygen in the paused abnormal structure sequence, and connect the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set; Based on the risk adjustment criteria set, the temporal distribution characteristics of blood oxygen fluctuations are analyzed to form a fragmented distribution sequence. The blood oxygen change performance of the fragmented distribution sequence is compared with the fragmented distribution sequence to obtain a trend comparison sequence. The variation amplitude of the trend comparison sequence is compared with the distribution span of the pause event feature set. The comparison result is mapped to a preset grading interval to obtain a sleep apnea risk grading result set.

[0006] Optionally, the step of time-aligning the respiratory signals, blood oxygen saturation signals, and sleep noise signals collected during the target monitoring period to obtain a set of sleep physiological time slices includes: The respiratory motion amplitude sequence collected by a wearable piezoelectric ceramic pressure sensor, the blood oxygen saturation sequence collected by a finger clip infrared photoplethysmography pulse wave sensor, and the sleep noise sequence collected by a high-sensitivity capacitive microphone during the target monitoring period will be identified as respiratory signal, blood oxygen saturation signal, and sleep noise signal. The time sampling indexes of respiratory signals, blood oxygen saturation signals, and sleep noise signals are monitored, and the respiratory signals and blood oxygen saturation signals are interpolated using a linear interpolation algorithm. Then, the respiratory signals, blood oxygen saturation signals, and sleep noise signals with the same time sampling index are placed into a preset data frame container, and the respiratory signals, blood oxygen saturation signals, and sleep noise signals are dimensionally calibrated to obtain a time index aligned data frame set. The amplitude and blood oxygen saturation signal of the respiratory signal in the time-indexed aligned data frame set are examined to check the trend of change over time. The apnea synchronization segment is determined from the time-indexed aligned data frame set based on the amplitude gradient of the respiratory signal and the decreasing trend of the blood oxygen saturation signal. The sleep noise signal in the time-indexed aligned data frame set is compared with the apnea synchronization segment to obtain the physiological noise associated segment set. Based on the parametric variation relationship and pathological causal mechanism of the respiratory signal, blood oxygen saturation signal and sleep noise signal in the physiological noise-related segment, a set of sleep physiological time slices is selected from the physiological noise-related segment.

[0007] Optionally, identifying the range of pause segments in the sleep physiological time slice set based on the decrease coefficient of the respiratory signals in the sleep physiological time slice set includes: The amplitude changes of respiratory signals in each time segment of the sleep physiological time slice set are detected to determine the amplitude trend and decay coefficient of the respiratory signals. The time segments in which the amplitude trend represents the peak amplitude showing a monotonically decreasing trend within a consecutive preset number of respiratory cycles and the decay coefficient is greater than a preset attenuation weight are determined as pause segment ranges to obtain a pause segment range set.

[0008] Optionally, the step of performing physiological time delay calibration on the blood oxygen saturation within the range of the pause segments to obtain a blood oxygen sequence set includes: The blood oxygen saturation within each pause segment range of the pause segment range is sampled to form an initial blood oxygen time series containing multiple blood oxygen sampling points. Based on the concentration change between adjacent blood oxygen sampling points in the initial blood oxygen time series, a target blood oxygen time series containing an effective decrease action is identified from the initial blood oxygen time series. Based on the range of the pause segment, the corresponding target blood oxygen time sequence is shifted backward to perform physiological time delay calibration, thereby obtaining a blood oxygen sequence set; Accordingly, the step of determining a pairing ratio coefficient based on the length of the pause segment range set and the descent delay of the blood oxygen sequence set, and obtaining a pause event feature set from the pause segment range set based on the pairing ratio coefficient, includes: Determine the range length of each of the pause segments in the pause segment range set, and determine the time delay required for the blood oxygen trough in the blood oxygen sequence set to decrease as the decrease delay; The ratio of the range length to the corresponding descent delay is determined as the pairing ratio coefficient; The pause events with the matching ratio coefficient within a preset normal range are selected from the range of pause segments, and the characteristic parameters of the pause events are determined; wherein, the characteristic parameters include the pause duration, blood oxygen saturation drop depth, and blood oxygen recovery slope of the pause event; The pause event is encapsulated with the feature parameters to obtain a pause event feature set.

[0009] Optionally, the step of connecting the fluctuation segments of blood oxygen saturation in the target segment using a morphological gradient operator to obtain a paused abnormal structure sequence includes: A short-time Fourier transform is performed on the blood oxygen saturation in the target segment to extract the fluctuation signal within a preset frequency range, and fluctuation segments of blood oxygen saturation that meet the preset physiological structural fluctuation conditions are selected from the fluctuation signal to obtain a set of fluctuation segment sequences. The symmetry and steepness of the blood oxygen saturation waveform in the set of fluctuation segments are analyzed by morphological gradient operators to determine the sharpness coefficient of the troughs in the blood oxygen saturation waveform. Fluctuation segments with sharpness coefficients greater than a preset threshold and blood oxygen saturation waveforms that are sawtooth-shaped are connected to obtain a paused abnormal structure sequence.

[0010] Optionally, the step of identifying the correlation fluctuation segments between respiration and blood oxygen in the paused abnormal structure sequence, and connecting the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set, includes: Determine the first center point of the fluctuation segment of blood oxygen saturation in the paused abnormal structure sequence, and determine the offset time of the first center point relative to the second center point of the pause segment range in the paused event feature set; Based on the fragment distribution of the pause anomaly structure sequence, the coverage rate of the blood oxygen decline phase on the pause event feature set is determined. Based on the bias time and the coverage rate, the correlation fluctuation fragments between respiration and blood oxygen in the pause anomaly structure sequence are identified to obtain the fluctuation fragment range set. A first-order difference operation is performed on the blood oxygen saturation in the associated fluctuation segment to obtain the blood oxygen change rate, and the blood oxygen change trend sequence is analyzed based on the blood oxygen change rate. Based on the blood oxygen change trend sequence, the starting point of blood oxygen decline is determined, and according to the phase difference between the starting point of blood oxygen decline and the starting point of apnea in the pause event, the blood oxygen change trend sequence and the associated fluctuation segment are connected in time index order based on the phase difference to obtain the risk adjustment basis set.

[0011] Optionally, the blood oxygenation changes in the fragment distribution sequence are compared with the fragment distribution sequence to obtain a trend comparison sequence. The variation amplitude of the trend comparison sequence is then compared with the distribution span of the sleep apnea event feature set. The comparison results are mapped to a preset grading interval to obtain a sleep apnea risk grading result set, including: Determine the first time point in the fragment distribution sequence where blood oxygen saturation recovers to the target value and the second time point corresponding to the respiratory recovery point, and determine the time difference between the first time point and the second time point; wherein, the target value is a preset multiple of the benchmark value; Based on the time difference, the blood oxygenation change performance of the fragment distribution sequence is determined, and the blood oxygenation change performance is compared with the fragment distribution sequence to obtain a trend comparison sequence. The variation range of the trend comparison sequence is compared with the distribution span of the pause event feature set. The comparison results are quantified into target indices, and the target indices are weighted and summed to obtain a comprehensive risk index. The target indices are the average slope of blood oxygen decline, the average pause duration, and the normalized apnea index. The comprehensive risk index is mapped to a preset grading interval to obtain a set of sleep apnea risk grading results.

[0012] Secondly, this application discloses an information processing device for sleep apnea detection, applied to a computer device, comprising: The signal alignment module is used to perform time alignment on the respiratory signals, blood oxygen saturation signals and sleep noise signals collected within the target monitoring period to obtain a set of sleep physiological time slices; The first screening module is used to identify the range of pause segments in the sleep physiological time slice set according to the decrease coefficient of the respiratory signal in the sleep physiological time slice set, and to perform physiological time delay calibration on the blood oxygen saturation in the range of pause segments to obtain a blood oxygen sequence set. The module determines the pairing ratio coefficient according to the range length of the pause segment set and the decrease delay of the blood oxygen sequence set, and obtains the pause event feature set from the pause segment set based on the pairing ratio coefficient. The second filtering module is used to identify target segments with increasing pause duration from the pause event feature set, and to connect the fluctuating segments of blood oxygen saturation in the target segments through the morphological gradient operator to obtain the pause abnormal structure sequence. The signal correlation module is used to identify the correlation fluctuation segments between respiration and blood oxygen in the paused abnormal structure sequence, and to connect the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set; The pause detection module is used to analyze the time distribution characteristics of blood oxygen fluctuations based on the risk adjustment criteria set, form a fragment distribution sequence, compare the blood oxygen change performance of the fragment distribution sequence with the fragment distribution sequence to obtain a trend comparison sequence, compare the change amplitude of the trend comparison sequence with the distribution span of the pause event feature set, and map the comparison result to a preset grading interval to obtain a sleep apnea risk grading result set.

[0013] Thirdly, this application discloses an electronic device, including: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the aforementioned disclosed information processing method for sleep apnea detection.

[0014] Fourthly, this application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed information processing method for sleep apnea detection.

[0015] The beneficial effects of this application are as follows: This application is applied to a computer device, comprising: time-aligning respiratory signals, blood oxygen saturation signals, and sleep noise signals collected within a target monitoring time period to obtain a set of sleep physiological time slices; identifying a set of pause segment ranges in the set of sleep physiological time slices based on the decrease coefficient of the respiratory signals in the set of sleep physiological time slices, and performing physiological delay calibration on the blood oxygen saturation in the set of pause segment ranges to obtain a set of blood oxygen sequences; determining a pairing ratio coefficient based on the range length of the set of pause segment ranges and the decrease delay of the blood oxygen sequence set; obtaining a set of pause event features from the set of pause segment ranges based on the pairing ratio coefficient; identifying target segments with increasing pause durations from the set of pause event features, and using morphological laddering... The degree operator connects the fluctuation segments of blood oxygen saturation in the target segment to obtain the pause abnormal structure sequence; identifies the correlation fluctuation segments between respiration and blood oxygen in the pause abnormal structure sequence, and connects the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain the risk adjustment basis set; analyzes the time distribution characteristics of blood oxygen fluctuations based on the risk adjustment basis set to form a segment distribution sequence, compares the blood oxygen change performance of the segment distribution sequence with the segment distribution sequence to obtain a trend comparison sequence, compares the change amplitude of the trend comparison sequence with the distribution span of the pause event feature set, and maps the comparison result to a preset classification interval to obtain the sleep apnea risk classification result set.Therefore, this application achieves precise synchronization and effective noise reduction of different physiological signals in the temporal dimension by aligning and filtering multi-source heterogeneous respiratory signals, blood oxygen saturation signals, and sleep noise signals to obtain a set of sleep physiological time slices. It accurately identifies the range of pause segments based on the respiratory signal decrement coefficient and performs physiological time delay calibration on the blood oxygen signal, improving the accuracy of blood oxygen correlation matching in accordance with the physiological laws of human blood circulation. Furthermore, it constructs a pause event feature set by filtering through a pairing ratio coefficient, eliminating invalid matching data to ensure the pathological validity of the features. Subsequently, it identifies target segments with increasing pause duration and combines them with morphological gradient operators to connect blood oxygen fluctuation segments to generate a pause abnormal structure sequence, strengthening the overall characterization ability of progressively deteriorating pathological features. Finally, it extracts the correlation fluctuation segments between respiration and blood oxygen and connects them with the blood oxygen change trend sequence to form a risk adjustment basis set. This approach achieves a deep integration of physiological causal relationships and temporal variation patterns. Finally, based on the risk adjustment basis set analysis of blood oxygen fluctuation distribution characteristics, it generates a fragment distribution sequence. Through trend comparison, it obtains a trend comparison sequence characterizing changes in the body's compensatory ability. Combining the magnitude of indicator changes and the span of segment distribution, it comprehensively judges and maps grading intervals. It can significantly reduce the probability of false positives and false negatives from multiple dimensions, including signal synchronization accuracy, event recognition precision, pathological evolution continuity, correlation matching rationality, and risk assessment comprehensiveness. It completely depicts the occurrence, development, and deterioration process of sleep apnea. It can complete automated, refined, and highly robust stratified risk assessment without relying on complex hardware, effectively improving detection efficiency and clinical adaptability, enhancing the fit between detection results and actual pathological states, and providing reliable, coherent, and quantitative technical support for the early screening and severity determination of sleep apnea. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0017] Figure 1 This is a flowchart of an information processing method for sleep apnea detection disclosed in this application; Figure 2 This application discloses a specific flowchart for obtaining a set of sleep physiological time slices. Figure 3 This application discloses a specific flowchart for obtaining a feature set of a pause event; Figure 4 This application discloses a specific flowchart for obtaining a paused abnormal structure sequence; Figure 5This application discloses a flowchart for obtaining a specific risk adjustment basis set; Figure 6 This application discloses a specific flowchart for obtaining a sleep apnea risk grading result set; Figure 7 This is a schematic diagram of an information processing device for sleep apnea detection disclosed in this application. Figure 8 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] The field of health risk assessment technology identifies potential risks by analyzing individual health data, providing decision support for medical intervention and health management. Its core encompasses technologies such as physiological data collection, health indicator calculation, and risk assessment model construction, and is widely applied in chronic disease prevention and personalized health management scenarios. Sleep apnea risk stratification assessment is an important branch of this field. Sleep apnea is a common sleep disorder characterized by repeated, brief cessations of breathing during sleep, leading to decreased blood oxygen saturation, disrupted sleep structure, and other problems. Long-term episodes can easily induce serious complications such as cardiovascular and cerebrovascular diseases and cognitive impairment, posing a significant threat to human health. Therefore, accurate detection and risk stratification of sleep apnea have significant clinical importance and application value.

[0020] Current methods for detecting sleep apnea mostly rely on complex medical equipment such as polysomnography systems. These methods suffer from drawbacks such as high testing costs, cumbersome operation, the need for professional personnel on duty, and unsuitability for long-term home monitoring. Furthermore, some simplified testing methods only analyze a single physiological signal, lacking the synergistic utilization of multiple physiological signals. This results in poor signal synchronization, low accuracy in identifying pause segments, and a high risk of false positives and false negatives. Additionally, they struggle to capture the progressive pathological characteristics of sleep apnea and cannot achieve precise grading of the severity of the condition, thus failing to meet the practical needs of early clinical screening, home monitoring, and personalized intervention.

[0021] Therefore, this application provides an information processing scheme for sleep apnea detection, which improves the accuracy and efficiency of sleep apnea detection.

[0022] See Figure 1As shown in the figure, this application discloses an information processing method for sleep apnea detection, applied to a computer device, including: Step S11: Time-align the respiratory signals, blood oxygen saturation signals, and sleep noise signals collected during the target monitoring period to obtain a set of sleep physiological time slices.

[0023] In this embodiment, the time alignment of the respiratory signal, blood oxygen saturation signal, and sleep noise signal collected within the target monitoring time period to obtain a set of sleep physiological time slices includes: determining the respiratory motion amplitude sequence collected using a wearable piezoelectric ceramic pressure sensor, the blood oxygen saturation sequence collected using a finger-clip infrared photoplethysmography (FPPM) sensor, and the sleep noise sequence collected using a high-sensitivity capacitive microphone within the target monitoring time period as the respiratory signal, blood oxygen saturation signal, and sleep noise signal, respectively; monitoring the time sampling index of the respiratory signal, blood oxygen saturation signal, and sleep noise signal, and interpolating the respiratory signal and blood oxygen saturation signal using a linear interpolation algorithm; and then placing the respiratory signal, blood oxygen saturation signal, and sleep noise signal with the same time sampling index into... A pre-defined data frame container is used, and the respiratory signal, blood oxygen saturation signal, and sleep noise signal are dimensionally calibrated to obtain a time-indexed aligned data frame set. The amplitude and blood oxygen saturation signal of the respiratory signal in the time-indexed aligned data frame set are examined to observe their changing trends over time. Based on the amplitude gradient of the respiratory signal and the decreasing trend of the blood oxygen saturation signal, apnea synchronization segments are determined from the time-indexed aligned data frame set. The sleep noise signal in the time-indexed aligned data frame set is compared with the apnea synchronization segments to obtain a set of physiological noise-related segments. Based on the parametric changes and pathological causal mechanisms of the respiratory signal, blood oxygen saturation signal, and sleep noise signal in the set of physiological noise-related segments, a set of sleep physiological time slices is selected from the set of physiological noise-related segments.

[0024] For example Figure 2As shown, firstly, respiratory motion amplitude sequence, blood oxygen saturation sequence and sleep noise sequence are acquired. The time sampling index of the three sets of sequences is monitored, and the same index values ​​are put into the data frame. The dimensions are calibrated according to the unit difference within the frame to obtain the time index aligned data frame set. The study acquires electrical signals of mechanical deformation in the chest and abdomen from a wearable piezoelectric ceramic pressure sensor, setting the sampling frequency to 50 Hz to obtain a respiratory motion amplitude sequence. Simultaneously, a finger-clip infrared photoplethysmography (FP-P) sensor is used to collect infrared light absorptivity at wavelengths of 660 nm and 940 nm. Based on Beer-Lambert's law, the ratio of oxyhemoglobin concentration to total hemoglobin concentration is calculated to obtain a blood oxygen saturation sequence with a sampling frequency of 1 Hz. A high-sensitivity capacitive microphone is used to record ambient and nasal / oral airflow audio data. By extracting the sound envelope and power spectral density features, a sleep noise sequence with a sampling frequency of 16000 Hz is obtained. The time sampling indexes of the three sequences are monitored. In other words, the respiratory motion amplitude sequence acquired by the wearable piezoelectric ceramic pressure sensor, the blood oxygen saturation sequence acquired by the finger-clip infrared FP-P, and the high-sensitivity capacitive microphone are combined to obtain a comprehensive analysis of the data. The sleep noise sequence collected by the condenser microphone is used as the respiratory signal, blood oxygen saturation signal, and sleep noise signal, respectively. The time sampling index of the three signals is monitored. Due to the physical differences in the original sampling frequencies of respiration, blood oxygen, and noise, the respiratory amplitude sequence and blood oxygen saturation sequence are aligned to a high-frequency time axis of 16,000 Hz using a linear interpolation algorithm. Alternatively, a cubic spline interpolation algorithm is used to complete the time axis alignment, that is, each original sampling point of the respiratory sequence is expanded to 320 points, and each original sampling point of the blood oxygen sequence is expanded to 16,000 points. The linear interpolation algorithm has high computational efficiency and short processing time, making it suitable for automated rapid assessment. The signal interpolated by the cubic spline interpolation algorithm is smoother and has lower distortion, making it suitable for clinical precision diagnosis scenarios with extremely high signal accuracy requirements. The two methods differ only in computational efficiency and signal smoothness, and the final time alignment effect is consistent. The same index values ​​are placed into a preset data frame container. Dimensional calibration is performed based on the differences in the millimeter-level units of respiratory motion amplitude, the percentage units of blood oxygen saturation, and the decibel units of sleep noise within the frame. Specifically, each dimension is mapped to a dimensionless space between 0 and 1. The baseline value for respiratory motion amplitude is set to 25 mm, the baseline value for blood oxygen saturation is set to 100%, and the baseline value for sleep noise is set to 100 dB. Through this mapping process, the heterogeneous features are transformed into comparable scalars, resulting in a time-indexed aligned data frame set.

[0025] Secondly, based on the time-indexed aligned data frame set, the respiratory amplitude and blood oxygen saturation within the frame are retrieved to examine their changing trends over time. Synchronization judgment is performed based on the amplitude gradient and saturation decreasing trend. The sleep noise value within the frame is compared with the synchronization segment, and a set of physiological noise-related segments is obtained based on the noise trend. Based on the temporal changing trends of respiratory amplitude and blood oxygen saturation in the time-indexed aligned data frame set, combined with the respiratory amplitude gradient and blood oxygen saturation decreasing trend, apnea synchronization segments are identified and determined. Specifically, a moving average window is used to calculate the gradient change of respiratory amplitude, with a window length set to 80,000 sampling points, corresponding to a 5-second duration. When the respiratory amplitude gradient is continuously less than a preset gradient threshold of 0.12 mm / s for 3 consecutive seconds, and simultaneously, the absolute value of the decreasing slope of blood oxygen saturation within the same time index segment is greater than the decreasing trend threshold of 0.06% / s, a synchronization judgment is performed, and this segment is marked as an apnea synchronization segment. Then, the sleep noise signal within the frame is compared with each apnea synchronization segment. Based on the correspondence between noise characteristics and physiological events, a set of physiological noise-related segments is obtained. In other words, the intra-frame sleep noise value is compared with the synchronization segment, and the energy performance of the noise power spectrum in the 200 Hz to 500 Hz frequency band is retrieved. According to the noise trend, if a surge in noise decibel value occurs at the end of the suspected synchronization segment, and the increase exceeds the average background noise by 18 decibels, it is determined to be physiological noise triggered by breathing effort. For example, when the respiratory amplitude gradient is 0.07 mm / s and the blood oxygen decrease trend is 0.08% / s, the synchronization condition is met. At this time, the noise is detected to jump from 32 decibels to 55 decibels, an increase of 23 decibels, which is consistent with the physiological noise correlation characteristics, and the set of physiological noise-related segments is obtained.

[0026] Finally, based on the set of physiological noise-related segments, the numerical groups of respiratory amplitude, blood oxygen saturation, and sleep noise within each segment were examined synchronously. Segments were filtered according to the parameter variation relationships, and the filtered segments were connected into continuous time slices by time index to obtain a set of sleep physiological time slices. Based on the parameter variation relationships of respiratory signals, blood oxygen saturation signals, and sleep noise signals within the physiological noise-related segments, the pathological causal mechanism and judgment threshold of sleep apnea were matched to filter the segments for effectiveness. Segments meeting the pathological characteristics were connected in time index order to finally obtain a set of sleep physiological time slices that can be used for subsequent analysis. Specifically, the parameter variation relationships within the set of physiological noise-related segments were used for segment filtering. The specific criteria were that the respiratory amplitude reduction ratio must be greater than 50% of the normal respiratory average, and the cumulative decrease in blood oxygen saturation must exceed 3%. If the value change satisfies this logic, the segment is retained. The filtered segments are then connected in the order of the original time index. For adjacent segments with an interval of less than 5 seconds, a closure operation is performed to form a continuous time slice. For example, if the respiratory amplitude of segment one decreases from 16 mm to 6.4 mm, a reduction of 60%, and blood oxygen decreases from 98% to 93%, a cumulative decrease of 5%, it meets the screening criteria. It is then connected with segment two, which meets the criteria. The index range is expanded from 80,000 to 1,600,000 to form a complete sleep physiological time slice, resulting in a set of sleep physiological time slices.

[0027] Step S12: Identify the range set of pause segments in the sleep physiological time slice set according to the decrease coefficient of the respiratory signal in the sleep physiological time slice set, and perform physiological time delay calibration on the blood oxygen saturation in the range set of pause segments to obtain the blood oxygen sequence set. Determine the pairing ratio coefficient according to the range length of the pause segment set and the decrease delay of the blood oxygen sequence set, and obtain the pause event feature set from the pause segment set based on the pairing ratio coefficient.

[0028] In this embodiment, identifying the pause range set in the sleep physiological time slice set based on the decrease coefficient of the respiratory signal in the sleep physiological time slice set includes: detecting the amplitude change of the respiratory signal in each time segment of the sleep physiological time slice set to determine the amplitude trend and decrease coefficient of the respiratory signal, and determining the time segment in which the amplitude trend represents the peak amplitude showing a monotonically decreasing trend within a consecutive preset number of respiratory cycles and the decrease coefficient is greater than a preset attenuation weight as the pause range, so as to obtain the pause range set.

[0029] For example Figure 3As shown, based on a set of sleep physiological time slices, the respiratory amplitude content within each time slice is monitored, the continuity of the respiratory amplitude content is examined, and segment actions are identified based on the hourly decrease in amplitude. The start and end time indices of the segments are converted into segment ranges, and the ranges are divided according to the baseline that the amplitude changes slowly at the end of the segment, resulting in a set of pause segment ranges. Specifically, the changes in respiratory signal amplitude in each time slice of the sleep physiological time slice set are detected. The amplitude change trend is determined by analyzing the interval between respiratory peaks and the amplitude changes, and a decrease coefficient is calculated. The decrease coefficient is obtained by taking the natural logarithm of the ratio of the peak amplitude of the initial reference cycle to the peak amplitude of the current cycle. Time slices in which the peak amplitude shows a monotonically decreasing trend within a preset number of consecutive respiratory cycles and the decrease coefficient is greater than a preset attenuation weight are identified as pause segments. At the same time, the plateau characteristic that the amplitude tends to flatten at the end of the segment is combined to define the boundary. Finally, all pause segments that meet the conditions are integrated to form a set of pause segment ranges.

[0030] In other words, by detecting the change in distance between respiratory peaks, the system identifies the hourly decrease in amplitude. When the peak amplitude of three consecutive respiratory cycles shows a monotonically decreasing trend and the decrease coefficient is greater than the preset attenuation weight of 0.25, the system identifies a segment action. The decrease coefficient is obtained by calculating the ratio of the peak amplitude of the initial reference cycle to the peak amplitude of the current cycle and taking the natural logarithm of this ratio. The starting peak index and ending trough index of the segment are converted into the segment range. The range is divided according to the benchmark that the amplitude changes slowly at the end of the segment. That is, when the absolute value of the amplitude change between adjacent sampling points is less than 0.03 mm for more than 3 seconds, it is determined to enter the pause plateau period, and this boundary is defined as the pause segment. For example, if the initial respiratory amplitude is 15 mm, and the subsequent three cycles are 11 mm, 7 mm, and 4 mm respectively, the decrease coefficient calculation result is approximately 1.32, which is greater than 0.25. Combined with the plateau period detection, the pause segment range set is obtained.

[0031] In this embodiment, the step of performing physiological time delay calibration on the blood oxygen saturation within the pause segment range set to obtain a blood oxygen sequence set includes: sampling the blood oxygen saturation within each pause segment range of the pause segment range set to construct an initial blood oxygen time series containing multiple blood oxygen sampling points; identifying a target blood oxygen time series containing an effective decrease action from the initial blood oxygen time series based on the concentration change between adjacent blood oxygen sampling points in the initial blood oxygen time series; and shifting the corresponding target blood oxygen time series backward based on the pause segment range to perform physiological time delay calibration, thereby obtaining a blood oxygen sequence set.

[0032] Based on the pause segment range set, the blood oxygen saturation content within the range of the sleep physiological time slice set is called, the time series of blood oxygen saturation within the range is examined, the sequence action is extracted according to the changes between adjacent sampling points, and the blood oxygen saturation sequence is located corresponding to the pause segment range to obtain the blood oxygen sequence set corresponding to the range.

[0033] The blood oxygen saturation within each pause segment of the pause range is sampled point by point to construct an initial blood oxygen time series composed of multiple blood oxygen sampling points. Based on the blood oxygen concentration change vector of adjacent sampling points in the initial blood oxygen time series, sequence change features are extracted. The state where the values ​​of adjacent sampling points decrease in the same direction and last for more than 6 seconds is determined as an effective decrease action. Based on this, the target blood oxygen time series is obtained by screening from the initial blood oxygen time series. Combining the physiological time delay law of human blood circulation from the lungs to the fingertips, the target blood oxygen time series corresponding to the pause segment is shifted backward by a dynamic delay constant of 15 to 30 seconds (usually with an average of 20 seconds) to complete the physiological time delay calibration, so that the blood oxygen time series and the pause segment correspond precisely in time and space. Finally, the calibrated blood oxygen sequence set is obtained. For example, in the pause range of index 200000 to 520000, the blood oxygen value sequence within the corresponding lag time period is obtained. Its decrease vector remains downward and lasts for more than 6 seconds. The correspondence is established, and the blood oxygen sequence set corresponding to the range is obtained.

[0034] In this embodiment, the step of determining a pairing ratio coefficient based on the range length of the pause segment range set and the decrease delay of the blood oxygen sequence set, and obtaining a pause event feature set from the pause segment range set based on the pairing ratio coefficient, includes: determining the range length of each pause segment range in the pause segment range set; determining the time delay required for the blood oxygen in the blood oxygen sequence set to reach its trough as the decrease delay; determining the ratio of the range length to the corresponding decrease delay as the pairing ratio coefficient; filtering pause events from the pause segment range set whose pairing ratio coefficient is within a preset normal range, and determining the feature parameters of the pause events; wherein, the feature parameters include the pause duration, blood oxygen decrease depth, and blood oxygen recovery slope of the pause event; and encapsulating the pause event and the feature parameters to obtain a pause event feature set.

[0035] Based on the range corresponding to the blood oxygen sequence set, the pause segment range set is called, the range is associated with the blood oxygen saturation content, and the pairing and inspection are performed according to the range length and the variation relationship within the blood oxygen sequence. The pairing results are then connected in time index order to obtain the pause event feature set. First, the length of each pause segment in the pause range set is calculated. The time required for blood oxygen to drop from the initial level to the trough in the blood oxygen sequence set is recorded as the drop delay. Then, the pause segment length is divided by the corresponding drop delay to obtain the pairing ratio coefficient, and the normal range of this coefficient is set to 0.5 to 1.5. Subsequently, valid pause events with pairing ratio coefficients within this normal range are selected. Key feature parameters such as pause duration, blood oxygen drop depth, and blood oxygen recovery slope are extracted for each pause event. The pause duration is determined by the pause start and end time markers. Finally, the pause events that meet the conditions and their corresponding feature parameters are encapsulated and integrated in time index order to form a pause event feature set that can be used for subsequent structured analysis. For example, if the pause range length is 24 seconds, the time for blood oxygen to drop from the initial level to the trough is 28 seconds, and the pairing ratio coefficient is approximately 0.86, which is within a reasonable range, this set of data and feature parameters is used to obtain the pause event feature set.

[0036] Step S13: Identify target segments with increasing pause duration from the pause event feature set, and connect the fluctuating segments of blood oxygen saturation in the target segments using a morphological gradient operator to obtain a pause abnormal structure sequence.

[0037] For example Figure 4 As shown, based on the pause segment range content in the pause event feature set, the duration of the pause segment range is monitored. The duration performance is examined according to the incremental change benchmark of the duration, and segments with extended duration are identified to obtain a segment range set. First, the duration in seconds of each pause segment is obtained by reading the time index difference, i.e., the pause duration. The duration performance is examined according to the incremental change benchmark of the duration, and it is analyzed whether the pause event shows a worsening pattern with the sleep process. Segments with extended duration are identified. If the duration of the subsequent pause segment increases by more than a preset growth coefficient of 2 seconds compared to the previous adjacent segment, or the duration of a single pause exceeds 30 seconds, it is identified as a severely extended segment, i.e., the target segment. For example, if a set of sequence lengths of 18 seconds, 22 seconds, and 26 seconds are detected, and the increase reaches 4 seconds, which is greater than the threshold of 2 seconds, the sequence interval is marked to obtain a segment range set containing each target segment.

[0038] In this embodiment, the step of connecting the fluctuation segments of blood oxygen saturation in the target segment using a morphological gradient operator to obtain a paused abnormal structure sequence includes: performing a short-time Fourier transform on the blood oxygen saturation in the target segment to extract fluctuation signals within a preset frequency range, and selecting fluctuation segments of blood oxygen saturation that meet preset physiological structural fluctuation conditions from the fluctuation signals to obtain a fluctuation segment sequence set; analyzing the symmetry and steepness of the blood oxygen saturation waveform in the fluctuation segment sequence set using a morphological gradient operator to determine the sharpness coefficient of the troughs in the blood oxygen saturation waveform, and connecting the fluctuation segments with sharpness coefficients greater than a preset threshold and blood oxygen saturation waveforms that are sawtooth waveforms to obtain a paused abnormal structure sequence.

[0039] Based on the fragment range set, the blood oxygen saturation content within the fragment range of the pause event feature set is called to examine the temporal distribution of blood oxygen saturation within the fragment range. Periodic inspection actions are performed based on the changes between adjacent sampling points to extract continuous fluctuation fragments from the blood oxygen saturation sequence. The extracted content is matched with the fragment range to obtain a set of fluctuation fragment sequences. Based on the set of fluctuation fragment sequences, the fragment range is associated with the blood oxygen saturation sequence. The structure is examined based on the relationship between the range span and the fluctuation within blood oxygen saturation. The examined sequences are connected into a continuous structure sequence according to the time index order to obtain the pause abnormal structure sequence.

[0040] First, a periodic inspection is performed to examine the changes between blood oxygen sampling points using short-time Fourier transform. The focus is on extracting low-frequency fluctuation signals within a preset frequency range (e.g., 0.01 Hz to 0.05 Hz). From the fluctuation signals, fluctuation segments of blood oxygen saturation that meet preset physiological structural fluctuation conditions are selected. For example, when the fluctuation amplitude (peak-to-peak value) exceeds 3% and the number of cycles is greater than 2, it is determined to be a physiological structural fluctuation. The extracted content is then matched with the segment range. For instance, within a 45-second segment, blood oxygen exhibits periodic fluctuations from 97% to 92% and then to 96%, with an amplitude of 5%, which meets the fluctuation extraction conditions, resulting in a set of fluctuation segment sequences.

[0041] The symmetry and steepness of the blood oxygen waveform are analyzed using morphological gradient operators to calculate the sharpness coefficient of the troughs. This coefficient is calculated by dividing the absolute value of the difference between the blood oxygen minimum and the previous sampling point by the amount of time change. If the sharpness coefficient is greater than the preset coefficient threshold of 0.15 and the waveform exhibits a typical sawtooth shape, then through structural inspection, the inspected sequence is connected into a continuous structural sequence according to the time index order. If the trough depth of adjacent fluctuation segments shows an increasing trend, they are merged into a whole structured sequence. For example, if the trough values ​​of multiple fluctuation segments monotonically decrease from 92% to 88%, they are connected to form a structural curve reflecting the deepening of hypoxia, resulting in a pause abnormal structural sequence. The pause abnormal structural sequence includes the abnormal duration level, abnormal fluctuation frequency index, and abnormal structural morphology category.

[0042] Step S14: Identify the correlation fluctuation segments between respiration and blood oxygen in the paused abnormal structure sequence, and connect the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set.

[0043] In this embodiment, identifying the correlation fluctuation segments between respiration and blood oxygen in the pause anomaly structure sequence, and connecting the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set, includes: determining the first center point of the blood oxygen saturation fluctuation segment in the pause anomaly structure sequence, and determining the offset time of the first center point relative to the second center point of the pause segment range in the pause event feature set; determining the coverage of the blood oxygen decline phase on the pause events in the pause event feature set based on the segment distribution of the pause anomaly structure sequence, and based on the... The bias time and coverage are used to identify the associated fluctuation segments between respiration and blood oxygen in the pause anomaly structure sequence to obtain a set of fluctuation segment ranges; a first-order difference operation is performed on the blood oxygen saturation in the associated fluctuation segments to obtain the blood oxygen change rate, and a blood oxygen change trend sequence is analyzed based on the blood oxygen change rate; the starting point of blood oxygen decrease is determined based on the blood oxygen change trend sequence, and based on the phase difference between the starting point of blood oxygen decrease and the starting point of respiratory arrest in the pause event, the blood oxygen change trend sequence and the associated fluctuation segments are connected in time index order based on the phase difference to obtain a risk adjustment basis set.

[0044] like Figure 5As shown, based on the pause anomaly structure sequence, the number and distribution of segments within the sequence are monitored. The distribution of segments within paused regions is examined based on their extension along the time axis. The location and span of blood oxygen saturation fluctuation segments are identified from the distribution, resulting in a set of fluctuation segment ranges. Based on this range set, blood oxygen saturation data within the range is retrieved, and the temporal trend of blood oxygen saturation within that range is examined. The trend is analyzed based on the rising and falling relationships between adjacent sampling points, resulting in a blood oxygen change trend sequence. Based on this trend sequence, the fluctuation segment ranges are correlated with the corresponding blood oxygen change trends. Synchronicity is analyzed based on the rising and falling rhythms within the trends and the segment spans. The correspondence between segments and trends is then connected in chronological order to form a continuous basis sequence, resulting in a risk adjustment basis set.

[0045] Specifically, firstly, based on the abnormal structure sequence of pauses, the number and distribution of segments within the sequence are monitored. The density is assessed by calculating the apnea-hypopnea index, i.e., the frequency of abnormal segments per hour. The distribution of segments in the pause segment is examined based on the extent of their extension on the time axis. The offset time of the center point of each blood oxygen fluctuation segment (i.e., the first center point) relative to the center point of the apnea segment (i.e., the second center point) is calculated. The position and span of the blood oxygen saturation fluctuation segments are identified from the distribution, and the completeness of the coverage of the respiratory event during the blood oxygen decline phase is judged. If the offset time is less than 15 seconds and the span coverage rate, i.e., the fluctuation duration divided by the pause duration, is greater than 80%, it is defined as a strongly correlated fluctuation region. For example, if a segment spans 28 seconds, covers 87.5% of the corresponding 32-second pause segment, and the offset time is within the circulatory physiological delay, the fluctuation segment range set is obtained.

[0046] Next, a first-order difference operation is performed on the blood oxygen saturation in the associated fluctuation segments to obtain the rate of change of blood oxygen at each moment. Alternatively, the rate of change of blood oxygen can be obtained by fitting the slope using the least squares method. When the rate of change of blood oxygen is negative and its absolute value continues to increase, it is determined to be an accelerating deterioration trend; when the difference value is positive, it is determined to be a recovery trend. The mean of the decreasing slope, the lowest blood oxygen value, and the recovery slope are recorded. For example, if the rate of decrease of blood oxygen at consecutive sampling points increases from -0.2% per second to -0.5% per second, it reflects an accelerated hypoxia process, resulting in a blood oxygen change trend sequence. The first-order difference operation can capture the real-time rate of change of blood oxygen and is suitable for analyzing short-term, rapid blood oxygen trends; the least squares method for fitting the slope can reflect the overall trend of blood oxygen changes and has stronger anti-interference ability, making it suitable for analyzing long-term, stable blood oxygen trends. The two methods differ only in the timeliness and anti-interference ability of trend capture; the final trend determination results are consistent.

[0047] Then, the phase difference between the starting point of blood oxygen decline and the starting point of apnea is calculated, and the time delay is mapped to the angular representation of the respiratory cycle. When the phase difference is within the physiological range corresponding to the delay of human blood circulation, it is determined to be highly synchronized. The correspondence between segments and trends is connected into a continuous basis sequence according to the time index order. The blood oxygen fluctuation response function triggered by each respiratory cycle is recorded. For example, if blood oxygen begins to decline 18 seconds after the start of apnea, it conforms to the circulation delay logic and is determined to be a valid basis, thus obtaining a risk adjustment basis set. The risk adjustment basis set includes the fluctuation trend direction category, risk weight coefficient, and segment distribution density index.

[0048] Step S15: Based on the risk adjustment criteria set, analyze the time distribution characteristics of blood oxygen fluctuations to form a fragment distribution sequence. Compare the blood oxygen change performance of the fragment distribution sequence with the fragment distribution sequence to obtain a trend comparison sequence. Compare the variation amplitude of the trend comparison sequence with the distribution span of the pause event feature set, and map the comparison result to a preset grading interval to obtain a sleep apnea risk grading result set.

[0049] For example Figure 6 As shown, based on the risk adjustment criteria set, the monitoring of blood oxygen saturation fluctuation segments and blood oxygen changes within the criteria set examines the time distribution of these fluctuation segments within the paused segment range of the paused event feature set. A comparison is made based on the correspondence between the fluctuation segment span and the paused segment range span to obtain the segment distribution sequence. Specifically, based on the risk adjustment criteria set, the monitoring of blood oxygen saturation fluctuation segments and blood oxygen changes examines the time distribution position and overall arrangement of these fluctuation segments within the corresponding paused segment range of the paused event feature set. The ratio of the fluctuation segment span to the paused segment range span is calculated to obtain the coverage weight coefficient. When this ratio is greater than 0.8, the paused event is assigned a risk gain weight of 1.2 times. Simultaneously, the relative percentage of time during which the fluctuation occurs within the paused segment is recorded. For example, if the fluctuation span is 26 seconds and the pause span is 30 seconds, the coverage weight coefficient is 0.87, and the starting position is 3 seconds after the pause begins. The coverage weight, risk gain weight, and relative time distribution information are integrated to form the segment distribution sequence.

[0050] In this embodiment, the blood oxygen change performance of the fragment distribution sequence is compared with the trend of the fragment distribution sequence to obtain a trend comparison sequence. The variation amplitude of the trend comparison sequence is compared with the distribution span of the sleep apnea event feature set. The comparison result is mapped to a preset grading interval to obtain a sleep apnea risk grading result set. This includes: determining the first time point in the fragment distribution sequence where blood oxygen saturation recovers to the target value and the second time point corresponding to the breathing recovery point, and determining the time difference between the first time point and the second time point; wherein, the target value is a preset multiple of the benchmark value; The blood oxygen saturation of the segment distribution sequence is determined based on the time difference, and the blood oxygen saturation is compared with the segment distribution sequence to obtain a trend comparison sequence. The variation amplitude of the trend comparison sequence is compared with the distribution span of the pause event feature set, and the comparison result is quantified into target indices. The target indices are then weighted and summed to obtain a comprehensive risk index. The target indices are the average slope of blood oxygen saturation decrease, the average pause duration, and the normalized apnea index. The comprehensive risk index is mapped to a preset grading interval to obtain a sleep apnea risk grading result set.

[0051] Based on the fragment distribution sequence, the blood oxygen saturation changes within the risk adjustment basis set are obtained. The temporal correspondence between the blood oxygen saturation changes and the fragment distribution sequence is examined. A trend comparison is performed based on the rise and fall of blood oxygen saturation changes and the differences between fragment positions to obtain a trend comparison sequence. Based on the trend comparison sequence, the overall distribution of pause ranges within the pause event feature set is obtained. The trend comparison sequence is compared with the overall distribution of pause ranges. Interval judgment is performed based on the magnitude of change in the trend comparison sequence and the span of pause range distribution. The comparison results are mapped to preset grading intervals to obtain the sleep apnea risk grading result set.

[0052] First, determine the first time point in the segment distribution sequence where blood oxygen saturation recovers to the target value and the second time point corresponding to the respiratory recovery point. Then, determine the time difference between the first time point and the second time point. The target value is a preset multiple of the benchmark value. For example, calculate the time difference between the time point where blood oxygen recovers to 90% of the benchmark value and the respiratory recovery point. If the difference gradually increases with the monitoring time, indicating a slower recovery, it indicates a decline in the body's compensatory ability or fatigue accumulation. For example, if the recovery delay is 6 seconds in the initial event and increases to 11 seconds in the later monitoring period, it reflects the decline in autonomic nervous system regulation ability. Based on the time difference, determine the blood oxygen change performance of the segment distribution sequence and compare the blood oxygen change performance with the segment distribution sequence to obtain a trend comparison sequence.

[0053] Then, the variation range of the trend comparison sequence is compared with the distribution span of the pause event feature set. The comparison results are quantified into target indices, namely the average slope of blood oxygen saturation decrease, the average pause duration, and the normalized apnea index. A comprehensive risk index is obtained by weighted summation of the average slope of blood oxygen saturation decrease, the average pause duration, and the normalized apnea index. The specific weights are allocated according to the preset contribution coefficients. Based on the calculation results, the interval is judged and the comparison results are mapped to the preset classification intervals. The interval of 0 to 0.35 is low risk, the interval of 0.35 to 0.65 is medium risk, and the interval of 0.65 to 1.0 is high risk. For example, the final risk index calculated by combining the weights of various physiological parameters is 0.6864, which falls in the interval of 0.65 to 1.0, and is judged as a severe sleep apnea risk. The sleep apnea risk classification result set is obtained, which includes preset low risk level, preset medium risk level, and preset high risk level.

[0054] The beneficial effects of this application are as follows: This application is applied to a computer device, comprising: time-aligning respiratory signals, blood oxygen saturation signals, and sleep noise signals collected within a target monitoring time period to obtain a set of sleep physiological time slices; identifying a set of pause segment ranges in the set of sleep physiological time slices based on the decrease coefficient of the respiratory signals in the set of sleep physiological time slices, and performing physiological delay calibration on the blood oxygen saturation in the set of pause segment ranges to obtain a set of blood oxygen sequences; determining a pairing ratio coefficient based on the range length of the set of pause segment ranges and the decrease delay of the blood oxygen sequence set; obtaining a set of pause event features from the set of pause segment ranges based on the pairing ratio coefficient; identifying target segments with increasing pause durations from the set of pause event features, and using morphological laddering... The degree operator connects the fluctuation segments of blood oxygen saturation in the target segment to obtain the pause abnormal structure sequence; identifies the correlation fluctuation segments between respiration and blood oxygen in the pause abnormal structure sequence, and connects the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain the risk adjustment basis set; analyzes the time distribution characteristics of blood oxygen fluctuations based on the risk adjustment basis set to form a segment distribution sequence, compares the blood oxygen change performance of the segment distribution sequence with the segment distribution sequence to obtain a trend comparison sequence, compares the change amplitude of the trend comparison sequence with the distribution span of the pause event feature set, and maps the comparison result to a preset classification interval to obtain the sleep apnea risk classification result set.Therefore, this application achieves precise synchronization and effective noise reduction of different physiological signals in the temporal dimension by aligning and filtering multi-source heterogeneous respiratory signals, blood oxygen saturation signals, and sleep noise signals to obtain a set of sleep physiological time slices. It accurately identifies the range of pause segments based on the respiratory signal decrement coefficient and performs physiological time delay calibration on the blood oxygen signal, improving the accuracy of blood oxygen correlation matching in accordance with the physiological laws of human blood circulation. Furthermore, it constructs a pause event feature set by filtering through a pairing ratio coefficient, eliminating invalid matching data to ensure the pathological validity of the features. Subsequently, it identifies target segments with increasing pause duration and combines them with morphological gradient operators to connect blood oxygen fluctuation segments to generate a pause abnormal structure sequence, strengthening the overall characterization ability of progressively deteriorating pathological features. Finally, it extracts the correlation fluctuation segments between respiration and blood oxygen and connects them with the blood oxygen change trend sequence to form a risk adjustment basis set. This approach achieves a deep integration of physiological causal relationships and temporal variation patterns. Finally, based on the risk adjustment basis set analysis of blood oxygen fluctuation distribution characteristics, it generates a fragment distribution sequence. Through trend comparison, it obtains a trend comparison sequence characterizing changes in the body's compensatory ability. Combining the magnitude of indicator changes and the span of segment distribution, it comprehensively judges and maps grading intervals. It can significantly reduce the probability of false positives and false negatives from multiple dimensions, including signal synchronization accuracy, event recognition precision, pathological evolution continuity, correlation matching rationality, and risk assessment comprehensiveness. It completely depicts the occurrence, development, and deterioration process of sleep apnea. It can complete automated, refined, and highly robust stratified risk assessment without relying on complex hardware, effectively improving detection efficiency and clinical adaptability, enhancing the fit between detection results and actual pathological states, and providing reliable, coherent, and quantitative technical support for the early screening and severity determination of sleep apnea.

[0055] At the data level, this embodiment achieves unified time indexing and dimensional calibration of three types of multi-source physiological signals: respiration, blood oxygen, and sleep noise. It also completes the synchronous correlation of signals by combining physiological laws, solving the problem that heterogeneous data cannot be compared and analyzed in the existing technology. It fully preserves the pathological correlation characteristics between physiological indicators, making the analysis results more in line with human physiological laws.

[0056] At the segment identification level, this embodiment breaks through the single threshold judgment and introduces the decreasing coefficient and plateau phase characteristics to identify apnea segments. At the same time, it combines short-time Fourier transform to extract the periodic fluctuation structure of blood oxygen and identifies the increasing trend of apnea duration. This achieves a structured and multi-dimensional characterization of apnea events, accurately captures the deterioration pattern of apnea events with the sleep process, and reflects the complete process of pathological evolution.

[0057] At the level of hierarchical logic, this embodiment constructs a comprehensive risk index quantification model, which weights and integrates multiple indicators such as abnormal event statistics, distribution density of pause segments, blood oxygen change trend, and the body's compensatory capacity, rather than classifying based solely on the number of events. This allows the risk classification results to not only reflect the frequency of events, but also to reflect core pathological information such as the degree of hypoxia and the speed of pathological deterioration, which is highly consistent with the actual pathological evolution process.

[0058] At the execution level, this embodiment achieves full automation and standardization from data collection and feature extraction to risk grading, eliminating the need for manual image review and correction, removing the subjectivity of human judgment, improving the consistency and accuracy of assessment results, and significantly increasing assessment efficiency, thus meeting the needs of large-scale, batch clinical assessments.

[0059] See Figure 7 As shown in the figure, this application discloses an information processing device for sleep apnea detection, applied to a computer device, comprising: The signal alignment module 11 is used to perform time alignment on the respiratory signal, blood oxygen saturation signal and sleep noise signal collected during the target monitoring period to obtain a set of sleep physiological time slices; The first screening module 12 is used to identify the range set of pause segments in the sleep physiological time slice set according to the decrease coefficient of the respiratory signal in the sleep physiological time slice set, and to perform physiological time delay calibration on the blood oxygen saturation in the range set of pause segments to obtain a blood oxygen sequence set. It also determines a pairing ratio coefficient according to the range length of the pause segment set and the decrease delay of the blood oxygen sequence set, and obtains a pause event feature set from the pause segment set based on the pairing ratio coefficient. The second filtering module 13 is used to identify target segments with increasing pause duration from the pause event feature set, and to connect the fluctuating segments of blood oxygen saturation in the target segments through the morphological gradient operator to obtain a pause abnormal structure sequence. The signal correlation module 14 is used to identify the correlation fluctuation segments between respiration and blood oxygen in the pause abnormal structure sequence, and to connect the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set; The pause detection module 15 is used to analyze the time distribution characteristics of blood oxygen fluctuations based on the risk adjustment criteria set, form a segment distribution sequence, compare the blood oxygen change performance of the segment distribution sequence with the segment distribution sequence to obtain a trend comparison sequence, compare the change amplitude of the trend comparison sequence with the distribution span of the pause event feature set, and map the comparison result to a preset grading interval to obtain a sleep apnea risk grading result set.

[0060] Furthermore, embodiments of this application also provide an electronic device. Figure 8 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0061] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Specifically, it may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the information processing method for sleep apnea detection performed by the electronic device disclosed in any of the foregoing embodiments.

[0062] In this embodiment, the power supply 23 is used to provide operating voltage for various hardware devices on the electronic device; the communication interface 24 can create a data transmission channel between the electronic device and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0063] The processor 21 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 21 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 21 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 21 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 21 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0064] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored on it include operating system 221, computer program 222 and data 223, etc., and the storage method can be temporary storage or permanent storage.

[0065] The operating system 221 manages and controls the various hardware devices and computer programs 222 on the electronic device to enable the processor 21 to perform calculations and processing on the massive amounts of data 223 in the memory 22. The operating system can be Windows, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the information processing method for sleep apnea detection executed by the electronic device as disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the electronic device from external devices, as well as data collected by its own input / output interface 25.

[0066] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned information processing method for sleep apnea detection. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0067] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0068] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly in hardware, software modules executed by a processor, or a combination of both. The software module may be located in random access memory (RAM), memory, read-only memory (ROM), electrically programmable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), register, hard disk, removable disk, CD-ROM (Compact Disc Read-Only Memory), or any other form of storage medium known in the art.

[0069] 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 apparatus 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 apparatus. 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 apparatus that includes said element.

[0070] The above provides a detailed description of the information processing method, apparatus, device, and medium for sleep apnea detection provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only intended to help understand the method and core ideas of the present invention. 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 the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An information processing method for sleep apnea detection, characterized in that, Applied to computer devices, including: The respiratory signals, blood oxygen saturation signals, and sleep noise signals collected during the target monitoring period are time-aligned to obtain a set of sleep physiological time slices; The pause segment range set in the sleep physiological time slice set is identified based on the decrease coefficient of the respiratory signal in the sleep physiological time slice set, and the blood oxygen saturation in the pause segment range set is calibrated by physiological time delay to obtain the blood oxygen sequence set. The pairing ratio coefficient is determined based on the range length of the pause segment range set and the decrease delay of the blood oxygen sequence set. The pause event feature set is obtained from the pause segment range set based on the pairing ratio coefficient. The target segments with increasing pause duration are identified from the pause event feature set, and the fluctuation segments of blood oxygen saturation in the target segments are connected by the morphological gradient operator to obtain the pause abnormal structure sequence. Identify the correlation fluctuation segments between respiration and blood oxygen in the paused abnormal structure sequence, and connect the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set; Based on the risk adjustment criteria set, the temporal distribution characteristics of blood oxygen fluctuations are analyzed to form a fragmented distribution sequence. The blood oxygen change performance of the fragmented distribution sequence is compared with the fragmented distribution sequence to obtain a trend comparison sequence. The variation amplitude of the trend comparison sequence is compared with the distribution span of the pause event feature set. The comparison result is mapped to a preset grading interval to obtain a sleep apnea risk grading result set.

2. The information processing method for sleep apnea detection according to claim 1, characterized in that, The process of time-aligning respiratory signals, blood oxygen saturation signals, and sleep noise signals collected within the target monitoring period to obtain a set of sleep physiological time slices includes: The respiratory motion amplitude sequence collected by a wearable piezoelectric ceramic pressure sensor, the blood oxygen saturation sequence collected by a finger clip infrared photoplethysmography pulse wave sensor, and the sleep noise sequence collected by a high-sensitivity capacitive microphone during the target monitoring period will be identified as respiratory signal, blood oxygen saturation signal, and sleep noise signal. The time sampling indexes of respiratory signals, blood oxygen saturation signals, and sleep noise signals are monitored, and the respiratory signals and blood oxygen saturation signals are interpolated using a linear interpolation algorithm. Then, the respiratory signals, blood oxygen saturation signals, and sleep noise signals with the same time sampling index are placed into a preset data frame container, and the respiratory signals, blood oxygen saturation signals, and sleep noise signals are dimensionally calibrated to obtain a time index aligned data frame set. The amplitude and blood oxygen saturation signal of the respiratory signal in the time-indexed aligned data frame set are examined to check the trend of change over time. The apnea synchronization segment is determined from the time-indexed aligned data frame set based on the amplitude gradient of the respiratory signal and the decreasing trend of the blood oxygen saturation signal. The sleep noise signal in the time-indexed aligned data frame set is compared with the apnea synchronization segment to obtain the physiological noise associated segment set. Based on the parametric variation relationship and pathological causal mechanism of the respiratory signal, blood oxygen saturation signal and sleep noise signal in the physiological noise-related segment, a set of sleep physiological time slices is selected from the physiological noise-related segment.

3. The information processing method for sleep apnea detection according to claim 1, characterized in that, The step of identifying the range of pause segments in the sleep physiological time slice set based on the decrease coefficient of the respiratory signals in the sleep physiological time slice set includes: The amplitude changes of respiratory signals in each time segment of the sleep physiological time slice set are detected to determine the amplitude trend and decay coefficient of the respiratory signals. The time segments in which the amplitude trend represents the peak amplitude showing a monotonically decreasing trend within a consecutive preset number of respiratory cycles and the decay coefficient is greater than a preset attenuation weight are determined as pause segment ranges to obtain a pause segment range set.

4. The information processing method for sleep apnea detection according to claim 1, characterized in that, The step of performing physiological time delay calibration on the blood oxygen saturation within the range of the paused segments to obtain a blood oxygen sequence set includes: The blood oxygen saturation within each pause segment range of the pause segment range is sampled to form an initial blood oxygen time series containing multiple blood oxygen sampling points. Based on the concentration change between adjacent blood oxygen sampling points in the initial blood oxygen time series, a target blood oxygen time series containing an effective decrease action is identified from the initial blood oxygen time series. Based on the range of the pause segment, the corresponding target blood oxygen time sequence is shifted backward to perform physiological time delay calibration, thereby obtaining a blood oxygen sequence set; Accordingly, the step of determining a pairing ratio coefficient based on the length of the pause segment range set and the descent delay of the blood oxygen sequence set, and obtaining a pause event feature set from the pause segment range set based on the pairing ratio coefficient, includes: Determine the range length of each of the pause segments in the pause segment range set, and determine the time delay required for the blood oxygen trough in the blood oxygen sequence set to decrease as the decrease delay; The ratio of the range length to the corresponding descent delay is determined as the pairing ratio coefficient; The pause events with the matching ratio coefficient within a preset normal range are selected from the range of pause segments, and the characteristic parameters of the pause events are determined; wherein, the characteristic parameters include the pause duration, blood oxygen saturation drop depth, and blood oxygen recovery slope of the pause event; The pause event is encapsulated with the feature parameters to obtain a pause event feature set.

5. The information processing method for sleep apnea detection according to claim 1, characterized in that, The step of connecting the fluctuating segments of blood oxygen saturation in the target segment using a morphological gradient operator to obtain a paused abnormal structure sequence includes: A short-time Fourier transform is performed on the blood oxygen saturation in the target segment to extract the fluctuation signal within a preset frequency range, and fluctuation segments of blood oxygen saturation that meet the preset physiological structural fluctuation conditions are selected from the fluctuation signal to obtain a set of fluctuation segment sequences. The symmetry and steepness of the blood oxygen saturation waveform in the set of fluctuation segments are analyzed by morphological gradient operators to determine the sharpness coefficient of the troughs in the blood oxygen saturation waveform. Fluctuation segments with sharpness coefficients greater than a preset threshold and blood oxygen saturation waveforms that are sawtooth-shaped are connected to obtain a paused abnormal structure sequence.

6. The information processing method for sleep apnea detection according to claim 1, characterized in that, The process involves identifying correlated fluctuation segments between respiration and blood oxygenation in the paused abnormal structural sequence, and then connecting the blood oxygenation change trend sequence analyzed based on these correlated fluctuation segments with the correlated fluctuation segments to obtain a risk adjustment basis set, including: Determine the first center point of the fluctuation segment of blood oxygen saturation in the paused abnormal structure sequence, and determine the offset time of the first center point relative to the second center point of the pause segment range in the paused event feature set; Based on the fragment distribution of the pause anomaly structure sequence, the coverage rate of the blood oxygen decline phase on the pause event feature set is determined. Based on the bias time and the coverage rate, the correlation fluctuation fragments between respiration and blood oxygen in the pause anomaly structure sequence are identified to obtain the fluctuation fragment range set. A first-order difference operation is performed on the blood oxygen saturation in the associated fluctuation segment to obtain the blood oxygen change rate, and the blood oxygen change trend sequence is analyzed based on the blood oxygen change rate. Based on the blood oxygen change trend sequence, the starting point of blood oxygen decline is determined, and according to the phase difference between the starting point of blood oxygen decline and the starting point of apnea in the pause event, the blood oxygen change trend sequence and the associated fluctuation segment are connected in time index order based on the phase difference to obtain the risk adjustment basis set.

7. The information processing method for sleep apnea detection according to any one of claims 1 to 6, characterized in that, The blood oxygen saturation changes of the fragment distribution sequence are compared with the trend of the fragment distribution sequence to obtain a trend comparison sequence. The variation amplitude of the trend comparison sequence is then compared with the distribution span of the sleep apnea event feature set. The comparison results are mapped to a preset grading interval to obtain a sleep apnea risk grading result set, including: Determine the first time point in the fragment distribution sequence where blood oxygen saturation recovers to the target value and the second time point corresponding to the respiratory recovery point, and determine the time difference between the first time point and the second time point; wherein, the target value is a preset multiple of the benchmark value; Based on the time difference, the blood oxygenation change performance of the fragment distribution sequence is determined, and the blood oxygenation change performance is compared with the fragment distribution sequence to obtain a trend comparison sequence. The variation range of the trend comparison sequence is compared with the distribution span of the pause event feature set. The comparison results are quantified into target indices, and the target indices are weighted and summed to obtain a comprehensive risk index. The target indices are the average slope of blood oxygen decline, the average pause duration, and the normalized apnea index. The comprehensive risk index is mapped to a preset grading interval to obtain a set of sleep apnea risk grading results.

8. An information processing device for detecting sleep apnea, characterized in that, Applied to computer devices, including: The signal alignment module is used to perform time alignment on the respiratory signals, blood oxygen saturation signals and sleep noise signals collected within the target monitoring period to obtain a set of sleep physiological time slices; The first screening module is used to identify the range of pause segments in the sleep physiological time slice set according to the decrease coefficient of the respiratory signal in the sleep physiological time slice set, and to perform physiological time delay calibration on the blood oxygen saturation in the range of pause segments to obtain a blood oxygen sequence set. The module determines the pairing ratio coefficient according to the range length of the pause segment set and the decrease delay of the blood oxygen sequence set, and obtains the pause event feature set from the pause segment set based on the pairing ratio coefficient. The second filtering module is used to identify target segments with increasing pause duration from the pause event feature set, and to connect the fluctuating segments of blood oxygen saturation in the target segments through the morphological gradient operator to obtain the pause abnormal structure sequence. The signal correlation module is used to identify the correlation fluctuation segments between respiration and blood oxygen in the paused abnormal structure sequence, and to connect the blood oxygen change trend sequence analyzed based on the correlation fluctuation segments with the correlation fluctuation segments to obtain a risk adjustment basis set; The pause detection module is used to analyze the time distribution characteristics of blood oxygen fluctuations based on the risk adjustment criteria set, form a fragment distribution sequence, compare the blood oxygen change performance of the fragment distribution sequence with the fragment distribution sequence to obtain a trend comparison sequence, compare the change amplitude of the trend comparison sequence with the distribution span of the pause event feature set, and map the comparison result to a preset grading interval to obtain a sleep apnea risk grading result set.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the information processing method for sleep apnea detection as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the information processing method for sleep apnea detection as described in any one of claims 1 to 7.

Images