Method for evaluating cardiac autonomic nervous function based on bispectrum nonlinear coupling analysis

Abstract

The application discloses a cardiac autonomic nerve function evaluation method based on bispectrum nonlinear coupling analysis, comprising the following steps: converting the heart beat interval sequence of a first to-be-detected object from a time domain to a frequency domain to obtain a heart beat spectrum sequence, wherein the heart beat spectrum sequence comprises a plurality of frequency domain components, and each frequency domain component comprises a frequency value, an amplitude value corresponding to the frequency value, and a phase value corresponding to the frequency value; dividing a low frequency band and a high frequency band based on a preset standard, and determining at least one group of first frequency domain component pairs from the plurality of frequency domain components, wherein two frequency values in the first frequency domain component pair are adjacent and respectively located in the high frequency band and the low frequency band; determining cross coupling power of the high frequency band and the low frequency band based on bispectrum estimation values of the at least one group of first frequency domain component pairs; and performing cardiac autonomic nerve function evaluation on the first to-be-detected object based on the cross coupling power. The application can calculate the cross coupling power of the low frequency band and the high frequency band to perform cardiac autonomic nerve function evaluation.

Description

Technical Field

[0001] This application relates to the field of electrocardiogram signal analysis technology, and in particular to a method for assessing cardiac autonomic nerve function based on bispectral nonlinear coupling analysis. Background Technology

[0002] In the field of electrocardiogram (ECG) signal analysis, heart rate variability is an important window for assessing the autonomic nervous system's regulatory function. Estimating power spectral density based on inter-beat sequences and extracting frequency domain indices has become a common method for obtaining information on autonomic nervous activity in this field.

[0003] In existing technologies, according to international standards, the low-frequency band of the power spectrum is defined as 0.04–0.15 Hz, and the high-frequency band as 0.15–0.40 Hz. Based on this, the power spectrum index obtained by dividing the low-frequency power by the high-frequency power is widely used to infer the balance between the sympathetic and vagus nerves. This calculation process only involves a simple ratio of power amplitudes within each frequency band; the power spectrum itself is obtained by squared amplitudes after the Fourier transform of the signal, and thus describes a second-order statistical measure of the signal.

[0004] However, the generation process of power spectrum indices discards phase information, so these indices cannot detect the temporal coordination between frequency components. Furthermore, the phase coupling phenomenon between different frequency bands in cardiac autonomic regulation may reflect a nonlinear interaction between the sympathetic and vagus nerves, information that cannot be captured by traditional power spectrum indices. Therefore, existing power spectrum indices cannot detect potential phase relationship changes in cardiac autonomic regulation, such as phase coupling between different frequency bands. Summary of the Invention

[0005] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a method for assessing cardiac autonomic function based on bispectral nonlinear coupling analysis. This method can obtain multiple frequency components by performing frequency domain transformation on the intercardiac interval sequence, determine the bispectral estimates of adjacent frequency components in different frequency bands, calculate the cross-coupling power between the low-frequency band and the high-frequency band, and assess cardiac autonomic function based on the cross-coupling power.

[0006] In a first aspect, this application provides a method for assessing cardiac autonomic nerve function based on bispectral nonlinear coupling analysis, the method comprising: The heartbeat interval sequence of the first subject to be tested is transformed from the time domain to the frequency domain to obtain the heartbeat spectrum sequence. The heartbeat spectrum sequence includes multiple frequency domain components, and each frequency domain component includes a frequency value and the amplitude value and phase value corresponding to the frequency value, respectively. Based on a preset standard, the low-frequency band and the high-frequency band are divided, and at least one first frequency domain component pair is determined from multiple frequency domain components. The two frequency values ​​in the first frequency domain component pair are adjacent and are respectively in the high-frequency band and the low-frequency band. The cross-coupling power of the high-frequency and low-frequency bands is determined based on the bispectral estimates of at least one set of first frequency domain component pairs. Cardiac autonomic nerve function was assessed in the first subject based on cross-coupling power.

[0007] In conjunction with the first aspect, in one possible implementation, the cross-coupling power of the high-frequency and low-frequency bands is determined based on bispectral estimates of at least one set of first frequency domain component pairs, specifically as follows:

[0008] In the formula, P LF-HF For cross-coupling power; k 1 and k 2 represents the discrete index value corresponding to the frequency value in its respective frequency domain component. k 1 and k 2 adjacent; and This is the low-frequency band marking function. The low-frequency band marking function takes the value 1 when the frequency value corresponding to the discrete index value belongs to the low-frequency band, and takes the value 0 otherwise. and This is the high-frequency band marking function; the high-frequency band marking function takes the value 1 when the frequency value corresponding to the discrete index value belongs to the high-frequency band, and takes the value 0 otherwise. for( k 1, k 2) The bispectral estimates of the corresponding frequency component pairs.

[0009] In conjunction with the first aspect, in one possible implementation, the cardiac autonomic nervous function of the first subject is assessed based on cross-coupling power, specifically as follows: The cross-coupling power is normalized to obtain the first evaluation index of the first test subject, and the cardiac autonomic nerve function of the first test subject is evaluated based on the first evaluation index of the first test subject.

[0010] In conjunction with the first aspect, one possible implementation involves normalizing the cross-coupling power, specifically as follows: Determine at least one pair of second frequency domain components from multiple frequency domain components, wherein the two frequency values ​​in the pair of second frequency domain components are adjacent and both are in the high frequency band; The self-coupling power in the high-frequency band is determined based on the bispectral estimates of each pair of second frequency domain components. The formula for calculating the self-coupling power is as follows:

[0011] In the formula, This is the self-coupling power; I HF ( k 3) and I HF ( k 4) All are high-frequency band marking functions; k 3 and k 4 represents the discrete index value corresponding to the frequency value in its respective frequency domain component, and k 3 and k 4 adjacent; for( k 3, k 4) The bispectral estimates of the corresponding frequency component pairs; The cross-coupling power is normalized based on its own coupling power.

[0012] In conjunction with the first aspect, one possible implementation involves normalizing the cross-coupling power based on its own coupling power, specifically as follows:

[0013] In the formula, The primary evaluation indicator for the first test subject; For cross-coupling power; This is the self-coupling power.

[0014] In conjunction with the first aspect, in one possible implementation, prior to assessing the cardiac autonomic nervous function of the first test subject based on a first evaluation index, the following is also included: Power spectrum indices were determined based on the intercardia sequence of the second test subject. Determine the first evaluation index of the second test object, and evaluate the correlation between the first evaluation index of the second test object and the power spectrum index; The effectiveness of the first evaluation indicator was verified based on its relevance.

[0015] In conjunction with the first aspect, in one possible implementation, the cardiac autonomic nervous function of the first test subject is assessed based on a first evaluation index of the first test subject, specifically as follows: The lower the value of the first assessment indicator for the first subject, the higher the risk of cardiac autonomic dysfunction in the first subject.

[0016] In conjunction with the first aspect, in one possible implementation, prior to assessing the cardiac autonomic nervous function of the first test subject based on a first evaluation index, the following is also included: Multiple third test subjects were grouped to obtain multiple test groups. The multiple test groups included a healthy group and at least two abnormal groups. The cardiac autonomic nerve function status of each third test subject was known, and the abnormality of each abnormal group was different. For each test group, determine the primary evaluation metric for each third test subject within the test group; Based on the first evaluation index of all test groups, the ratio of the mean square between groups to the mean square within groups is determined, the F-statistic is obtained, and the overall difference between groups is judged based on the associated probability corresponding to the F-statistic. Thus, determine the difference between the arithmetic means of each pair of test groups and the two-sided probability value; Based on the difference and bilateral probability values, the effectiveness of the first assessment index in distinguishing different cardiac autonomic nervous system functional states was verified.

[0017] In conjunction with the first aspect, in one possible implementation, the effectiveness of the first assessment index in distinguishing different cardiac autonomic functional states is verified based on the difference and bilateral probability values, specifically as follows: The pre-defined significance threshold is corrected based on the Bonferroni correction method; Each probability value is compared with the corrected significance threshold to obtain the corresponding comparison result; Based on the difference in the arithmetic mean between groups and the corresponding comparison results, the effectiveness of the first assessment indicator in distinguishing different cardiac autonomic nervous system functional states was verified.

[0018] In conjunction with the first aspect, one possible implementation method also includes: Collect the electrocardiogram (ECG) signal of the first subject to be tested within a preset time period, and preprocess the ECG signal, including at least one of filtering and noise reduction processing; The preprocessed electrocardiogram signal was subjected to QRS wave detection to obtain the heartbeat interval sequence of the first subject.

[0019] Compared with the prior art, this application has the following beneficial effects: This application filters adjacent frequency domain components in different frequency bands and determines their bispectral estimates. Then, based on the bispectral estimates, it calculates the cross-coupling power of the high-frequency and low-frequency bands. This cross-coupling power can quantify the secondary phase coupling strength caused by the nonlinear interaction between the low-frequency and high-frequency bands, so as to realize the detection and evaluation of the cardiac autonomic nerve function status. Attached Figure Description

[0020] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1This is a flowchart illustrating the method of this application in one embodiment; Figure 2 This is another flowchart illustrating the method of this application in one embodiment; Figure 3 In Example 1, Scatter plot of correlation with LF / HF; Figure 4 For example, in Example 2, there are four test groups. Distribution of violin diagrams; Figure 5 For example, in Example 2, there are four test groups. Distribution of violin diagrams; Figure 6 In Example 3, Comparison of receiver operating characteristic (ROC) curves for distinguishing between healthy individuals and CAD patients, and between LF / HF and healthy individuals and CAD patients; Figure 7 In Example 3, Comparison of ROC curves for distinguishing between healthy individuals and AMI patients, and between LF / HF and healthy individuals and AMI patients; Figure 8 In Example 3, Comparison of ROC curves for distinguishing between healthy individuals and ESRD patients, and between LF / HF and healthy individuals and ESRD patients. Detailed Implementation

[0021] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present application will now be described in detail with reference to the accompanying drawings and embodiments. Furthermore, the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The terms "first" and "second," etc., in the specification and claims of the embodiments of this application are used to distinguish different objects, not to describe a specific order of objects.

[0023] The following explains the terms used in this application: Secondary phase coupling: This is a typical manifestation of nonlinear interaction in the frequency domain. When the sympathetic and vagal regulation of the cardiac autonomic nervous system interacts through a nonlinear mechanism, energy transfer and phase locking will occur between low-frequency and high-frequency components. That is, a certain frequency is exactly the sum of the other two frequencies, and the phases of the three maintain a constant and stable relationship.

[0024] Independent predictor: refers to a factor whose changes in the current assessment indicator itself can independently explain the risk of disease, without relying on other indicators.

[0025] Suppressed variables: When one variable is included in the model, it eliminates or corrects the spurious relationship between another variable and the outcome, thereby allowing the true effect of the latter to be revealed.

[0026] like Figure 1 , Figure 2 As shown, this application provides a method for assessing cardiac autonomic nerve function based on bispectral nonlinear coupling analysis, the method comprising: S1. The heartbeat interval sequence of the first test subject is converted from the time domain to the frequency domain to obtain the heartbeat spectrum sequence. The heartbeat spectrum sequence includes multiple frequency domain components, and each frequency domain component includes a frequency value and the amplitude value and phase value corresponding to the frequency value, respectively. Specifically, the heartbeat interval sequence of the first subject was obtained based on the following method: Collect the electrocardiogram (ECG) signal of the first subject to be tested within a preset time period, and preprocess the ECG signal; One possible implementation involves preprocessing including at least one of filtering and denoising to eliminate power frequency interference and electromyography artifacts.

[0027] The preprocessed electrocardiogram signal was subjected to QRS wave detection to identify the R wave peak point, and the cardiac interval sequence of the first subject was obtained by differential calculation.

[0028] In one possible implementation, this application further performs at least one of outlier detection and replacement, and detrending processing on the cardiac interbeat sequence to obtain a clean cardiac interbeat sequence.

[0029] For example, the heartbeat interval sequence is , It is a finite-length discrete random signal; For length of Heartbeat interval sequence First, calculate its Fourier transform to obtain a length of cardiac spectral sequence X ( k ):

[0030] in, yes exist k The Fourier transform result at the point, k It is a discrete index value corresponding to a certain frequency value, with a value range from 0 to... N -1, corresponding to The first in k One frequency domain component; It is a length of N The intercardiac interval sequence n It is a time-domain index, with values ​​ranging from 0 to... N -1, Representing the n Heartbeat interval values ​​at each sampling point; e It is the base of the natural logarithm. j It is the imaginary unit (satisfying) j ²=-1); It is the phase angle.

[0031] It is worth mentioning that discrete index values k Compared with the actual frequency value f To facilitate calculation, a one-to-one correspondence is typically established, and the actual frequency value is usually converted into a discrete index value by combining the resampling frequency. k Actual frequency value f With discrete index value k The conversion relationship is as follows: f = k Δ f In the formula, Δ f Frequency resolution: Δ f = f s / N , f s The sampling frequency (in Hz) for the cardiac interval sequence is a preset value.

[0032] S2. Divide the low-frequency band and the high-frequency band based on the preset standard, and determine at least one first frequency domain component pair from multiple frequency domain components. The two frequency values ​​in the first frequency domain component pair are adjacent and are respectively in the high-frequency band and the low-frequency band. One possible implementation is to divide the frequency band into a low-frequency band (0.04-0.15 Hz) and a high-frequency band (0.15-0.40 Hz) according to the HRV standard frequency band definition.

[0033] S3. Determine the cross-coupling power of the high-frequency band and the low-frequency band based on the bispectral estimates of at least one set of first frequency domain component pairs; One possible implementation involves determining at least one first frequency domain component pair from multiple sets of frequency domain components, and determining the cross-coupling power of the high-frequency and low-frequency bands based on the bispectral estimates of each first frequency domain component pair, specifically as follows:

[0034] In the formula, P LF-HF For cross-coupling power; k 1 and k 2 represents the discrete index value corresponding to each frequency value in a group of adjacent frequency components; and This is the low-frequency band marking function. The low-frequency band marking function takes the value 1 when the frequency value corresponding to the discrete index value belongs to the low-frequency band, and takes the value 0 otherwise. and This is the high-frequency band marking function; the high-frequency band marking function takes the value 1 when the frequency value corresponding to the discrete index value belongs to the high-frequency band, and takes the value 0 otherwise. for( k 1, k 2) The bispectral estimates of the corresponding frequency component pairs.

[0035] It is worth mentioning that, I LF ( k 1) I HF ( k 2)+ I HF ( k 1) I LF ( k 2)] The function is to filter out all frequency pairs where one frequency is in the low-frequency range and the other is in the high-frequency range, because only when... k 1 and k The absolute value of this term is 1 only when the frequencies belong to different frequency bands; otherwise, it is 0. It is worth mentioning that this application sets the summation range as follows: k 1=0 to N -2, k 2= k 1+1 arrive N-1 This is to avoid repeatedly calculating the same frequency domain component pair; multiplying by 1 / 2 is because k 1 belongs to low frequency, k 2 belongs to high frequency and k 1 belongs to high frequency, k 2. The low-frequency components are symmetrical during the summation process to avoid double counting.

[0036] One possible implementation is a low-frequency band labeling function. and high-frequency band labeling function Calculated based on the following formulas respectively:

[0037]

[0038] in, f The actual continuous frequency value (unit: Hz), based on f and k From the relationship, we can know that I LF ( f )= I LF ( k Δ f )= I LF ( k f s / N ), I HF ( f )= I HF ( k Δ f ) = I HF ( k f s / N ),Depend on f and k The conversion relationship can be used to calculate the corresponding values ​​for the low-frequency and high-frequency bands respectively. k Value range.

[0039] One possible implementation, the bispectral estimate is calculated as follows:

[0040] in, yes( k 1, k The bispectral estimate at point 2) k 1, k 2 is a discrete index value, ranging from 0, 1, ... N 1; when k 1+ k 2≥ N Points outside the range are either not calculated or are processed as periodic expansions. for The complex conjugate of is a complex number containing both amplitude and phase values; X ( k 1) is the Fourier transform value at the discrete index k1. X ( k 2) These are the discrete index values. k Fourier transform value at position 2; It is a discrete index value k 1+ k Fourier transform value at position 2.

[0041] It is worth mentioning that the bispectral estimate is a two-dimensional Fourier transform of the third-order cumulant of the heartbeat signal, specifically used to detect secondary phase coupling in the heartbeat signal. Specifically, the amplitude of the bispectral estimate... | Directly reflects k 1. k 2 and k 1+ k 2. The strength of the secondary phase coupling between these three frequency components, and the phase of the bispectral estimate reflects the phase relationship between them.

[0042] It is worth mentioning that when Gaussian noise introduced by the environment is mixed into the ECG signal, regardless of the noise power, it will be automatically reduced to zero in the calculation of the third-order cumulant. Based on this mechanism, the bispectral estimation can extract the coupled signal and remove Gaussian noise at the same time, ensuring the accuracy of the index calculation without relying on additional noise reduction algorithms, thereby enhancing the stability and repeatability of the evaluation results under different acquisition conditions.

[0043] S4. The cardiac autonomic nervous function of the first subject is assessed based on the cross-coupling power.

[0044] One possible implementation, the method further includes: S5. Normalize the cross-coupling power to obtain the first evaluation index, and evaluate the cardiac autonomic nerve function of the first test subject based on the first evaluation index of the first test subject.

[0045] Specifically, the lower the value of the first assessment indicator for the first subject, the higher the risk of cardiac autonomic dysfunction in the first subject.

[0046] One possible implementation involves normalizing the cross-coupling power, specifically as follows: Determine at least one pair of second frequency domain components from multiple frequency domain components, wherein the two frequency values ​​in the pair of second frequency domain components are adjacent and both are in the high frequency band; The self-coupling power in the high-frequency band is determined based on the bispectral estimate of each second frequency domain component pair; One possible implementation method, the formula for calculating self-coupling power is:

[0047] In the formula, This is the self-coupling power; I HF ( k 3) and I HF ( k 4) All are 1; k 3 and k 4 represents the discrete index value corresponding to the frequency value in its respective frequency domain component, and k 3 and k 4 adjacent; for( k 3, k 4) The bispectral estimates of the corresponding frequency component pairs.

[0048] The cross-coupling power is normalized based on its own coupling power.

[0049] One possible implementation is to normalize the cross-coupling power based on its own coupling power, specifically as follows:

[0050] In the formula, The primary evaluation indicator for the first test subject; For cross-coupling power; This is the self-coupling power.

[0051] It is worth mentioning that nCP HF nCP represents the ratio of the strength of the secondary phase coupling between the low-frequency and high-frequency bands to the coupling strength within the high-frequency band itself, reflecting the relative contribution of the nonlinear coupling between the low-frequency and high-frequency bands to the nonlinear dynamics of the high-frequency band itself. HF The lower the value, the weaker the nonlinear coupling strength between the low-frequency band and the high-frequency band, and the higher the risk of cardiac autonomic dysfunction.

[0052] This application aims to achieve a quantitative assessment of nonlinear phase coupling, thereby compensating for the information gaps in traditional power spectrum analysis for autonomic nervous function assessment. Specifically, traditional power spectrum indices discard phase information during calculation, making them unable to detect potential secondary phase coupling phenomena. The bispectral estimation value in this application belongs to the third-order spectrum. Its calculation process directly uses the Fourier transform results containing phase information, multiplying the three frequency components—low-frequency component, high-frequency component, and their sum-frequency component—in complex form and averaging them. The existence of secondary phase coupling is explored by detecting whether the phases of the three components cancel each other out in the product. This allows this application to capture and quantify the cross-frequency phase coupling strength generated by nonlinear interactions, ultimately extracting non-Gaussian information and nonlinear coupling characteristics that power spectra cannot present, providing a more multidimensional assessment basis for cardiac autonomic nervous function.

[0053] Furthermore, this application also calculates the ratio of low-frequency-high-frequency cross-coupling power to high-frequency self-coupling power, removes the individual's high-frequency background coupling baseline from the overall nonlinear activity, and extracts a normalized index that purely reflects the intensity of nonlinear interaction between low and high frequencies, so that the assessment results of cardiac autonomic nerve function are not affected by the inherent nonlinear activity differences in the high-frequency segment between individuals.

[0054] One possible implementation, prior to assessing the cardiac autonomic nervous function of the first test subject based on a first evaluation index, further includes: Power spectrum indices were determined based on the intercardia sequence of the second test subject. Determine the first evaluation index of the second test object, and evaluate the correlation between the first evaluation index of the second test object and the power spectrum index; The effectiveness of the first evaluation index for the first test object is verified based on the correlation.

[0055] Specifically, this application designates the first evaluation indicator as The power spectrum index is the ratio of low-frequency power to high-frequency power (LF / HF) obtained from power spectrum density (PSD) analysis. This application determines the correlation coefficient between the two based on Spearman correlation. When the correlation coefficient is greater than the preset value and the two are positively correlated, the first evaluation index passes the validity verification. The preset value is set by those skilled in the art.

[0056] The following is based on Example 1. Correlation analysis with LF / HF: Example 1: Data was sourced from the AutonomicAging database within the PhysioNet (https: / / www.physionet.org / ) database, encompassing 1121 healthy volunteers (secondary subjects) aged 18-92 years, divided into 15 groups at 5-year intervals. ECG (electrocardiogram) signals were recorded under standard resting conditions for an average duration of 19 minutes (range 8-45 minutes), monitored by an operator throughout. After removing records with incomplete data, incomplete age group and gender information, or poor signal quality, 1081 records were obtained. For the short-term ECG records of these 1081 healthy volunteers, a 5-minute stable segment of the ECG signal was used for subsequent preprocessing. The primary evaluation metric and HRV (Heart Rate Variability) were calculated based on the entire 5-minute standard duration data segment, and were obtained separately. And the LF / HF index.

[0057] Figure 3 for A scatter plot of the correlation between LF / HF and the linear fit trend line is shown, with the solid green line representing the trend line and the orange shaded area representing the 95% confidence interval of the fitted line. This plot helps to illustrate the overall trend of the data. Because LF / HF exhibits a skewed distribution (obtained by both the Kolmogorov-Smirnov test and the Shapiro-Wilk test),... P <0.001, P The correlation coefficient (value for significance test) does not meet the criteria for Pearson correlation analysis; therefore, Spearman correlation analysis is used for evaluation. Correlation with LF / HF. Results show: It showed a moderate positive correlation with LF / HF (correlation coefficient of 0.604). P <0.001). The above results indicate that... While LF / HF reflects shared autonomic regulatory information to some extent, the correlation coefficient did not reach a strong level (correlation coefficient > 0.8), indicating that... It also carries independent information that LF / HF cannot interpret, making it a powerful supplement to the traditional HRV index (i.e., power spectrum index), confirming... It carries autonomic information that is complementary to LF / HF and can be used to assess cardiac autonomic function.

[0058] The method of this application can complete the calculation based on a short 5-minute ECG signal, which is consistent with the recording duration requirements of the current international standard for HRV analysis. It does not require an additional extension of the data acquisition time, making it easy to promote and apply in clinical ECG machines, dynamic ECG analysis systems and wearable devices.

[0059] Another possible implementation includes, before assessing the cardiac autonomic nervous function of the first test subject based on the first assessment index, the following: Multiple third test subjects were grouped to obtain multiple test groups. The multiple test groups included a healthy group and at least two abnormal groups. The cardiac autonomic nerve function status of each third test subject was known, and the abnormality of each abnormal group was different. For each test group, determine the primary evaluation metric for each third test subject within the test group; Based on the first evaluation index of all test groups, the ratio of the mean square between groups to the mean square within groups is determined, the F-statistic is obtained, and the overall difference between groups is judged based on the associated probability corresponding to the F-statistic. Specifically, the associated probability is compared with a standard threshold (usually 0.05). If the associated probability is less than 0.05, the overall difference between groups is considered significant; otherwise, it is not significant.

[0060] Thus, determine the difference between the arithmetic means of each pair of test groups and the two-sided probability value; Based on the difference and bilateral probability values, the effectiveness of the first assessment index in distinguishing different cardiac autonomic nervous system functional states was verified.

[0061] One possible implementation is that, for each pair of groups, the corresponding two-sided probability value is obtained based on the arithmetic mean, standard deviation, and independent samples t-test of the two groups.

[0062] One possible implementation method, based on the difference and bilateral probability values, is to verify the effectiveness of the first assessment index in distinguishing different cardiac autonomic functional states, specifically as follows: The pre-defined significance threshold is corrected based on the Bonferroni correction method; Each two-sided probability value is compared with the corrected significance threshold to obtain the corresponding comparison result; The effectiveness of the first assessment index in distinguishing different cardiac autonomic nervous system functional states was verified based on the difference between the arithmetic means of each group and the corresponding comparison of the two-sided probability values.

[0063] Specifically, for each pair of groups, the arithmetic mean is primarily used to directly calculate the difference between the two groups; while the two-tailed probability value is used to assess the probability that this difference is entirely due to random sampling error. In other words, in this implementation, validity verification first calculates the magnitude of the difference using the mean, and then... P The statistical reliability of this difference is verified only when the difference is greater than a preset difference threshold and POnly when the value is less than the preset significance threshold is it finally determined to be a significant difference. When there is a significant difference, the first evaluation index can effectively distinguish different cardiac autonomic nerve function states; otherwise, it cannot be effectively distinguished.

[0064] Example 2 The process of effectively differentiating different cardiac autonomic nervous system functional states is explained: In Example 2, the test group included a healthy group and three disease groups. The first evaluation indicator for the third test subject in each group was... ; Example 2: Using 24-hour long-term recordings, healthy individuals were obtained from the PhysioNet RR interval time series from healthy subjects database, the Normal Sinus Rhythm RR Interval Database, the MIT-BIH Normal Sinus Rhythm Database, and the Normal database (E-HOL-03-0202-003) from THEW (http: / / thew-project.org / databases.htm). The disease group included three patient categories: 1) Coronary Artery Disease (CAD) database (E-HOL-03-0271-002); 2) Acute Myocardial Infarction (AMI) database (E-HOL-03-0160-001); and 3) End-Stage Renal Disease (ESRD) database (E-HOL-12-0051-016). All three disease groups are high-risk groups for autonomic nervous system dysfunction, especially the ESRD group, where the autonomic nervous system dysfunction is most pronounced due to the combined effects of multiple factors such as uremic toxins, electrolyte imbalance, and excessive activation of the sympathetic nervous system.

[0065] To match the age of the disease population, data from healthy individuals under 35 years of age were excluded. Furthermore, after excluding records with incomplete information, missing age and gender details, and poor ECG quality, the final included data consisted of: 170 healthy individuals, 64 AMI patients, 231 CAD patients, and 51 ESRD patients. For the 24-hour long-term recordings, non-overlapping 5-minute sliding windows were used for calculating both the primary assessment indicator and the HRV indicator, and the mean value of the indicator across all windows was taken as the representative value.

[0066] In Example 2, for ANOVA (Analysis of Variance) was used to calculate... The ratio of the between-group mean square to the within-group mean square is used to obtain the corresponding F-statistic. The associated probability corresponding to the F-statistic (i.e., in Table 1) P The 1-value indicates whether the overall difference between groups is significant. Specifically, it will... P The value of 1 was compared with the standard threshold of 0.05 (corrected by Bonferroni). The comparison results are shown in Table 1. of P 1 < 0.0001; Due to significant overall differences between groups, pairwise comparisons were made between each test group: nCP of the CAD group, AMI group, and ESRD group. HF All values ​​were significantly lower than those in the normal group (-0.235±0.292). vs. -0.476±0.280, -0.481±0.294, and -0.719±0.305; all P 2 < 0.001, where, P 2 represents the two-sided probability; the preset significance threshold, after Bonferroni correction, is set to 0.05. Another possible implementation, in Example 2, is to use one-way ANOVA to calculate the power spectral density index LF / HF. The ratio of the between-group mean square to the within-group mean square is obtained. The corresponding F-statistic; as shown in Table 1, LF / HF... P 1 = 0.018, which is also less than 0.05, indicating that Both the LF / HF index and the LF / HF index showed significant differences between groups overall. F=45.456, corresponding to P 1 < 0.0001; LF / HF: F = 3.388, corresponding to P 1 = 0.018).

[0067] Furthermore, regarding LF / HF, since the overall differences between groups were also significant, this application also compared each test group corresponding to LF / HF: The LF / HF of the ESRD group was significantly lower than that of the normal group (3.399±2.109). vs. 2.420±2.008; P 2<0.05); comparisons between disease groups showed that the ESRD group Significantly lower than the CAD group and AMI group (-0.719±0.305) vs.-0.476±0.280 and -0.481±0.294). There was no significant difference in LF / HF between any two disease groups. P 2>0.05).

[0068] Table 1. Comparison between different test groups Comparison with LF / HF

[0069] In Table 1, the arithmetic mean within each test group is expressed as mean ± standard deviation; a: compared with the Normal group, b: compared with the CAD group, c: compared with the AMI group; P 2 indicates a two-sided probability; express P 2<0.05, express P 2<0.01, express P 2 < 0.001, express P 2 < 0.0001.

[0070] Based on the above results, it can be concluded that Compared to the traditional LF / HF index, it has the following two significant advantages: First, it has higher sensitivity and a wider detection range. (For the three disease groups (CAD, AMI, ESRD)) All were significantly lower than the healthy control group (all P 2 < 0.001), while LF / HF showed a significant difference only in the ESRD group ( P (2 < 0.05), there were no significant differences between the CAD and AMI groups and the healthy group. This result indicates that... It can detect autonomic dysfunction that LF / HF cannot identify, is more sensitive to early or mild pathological changes, and has a wider spectrum of disease detection.

[0071] Second, it can quantify the damage gradient between different diseases. Figure 4 Four test groups Distribution of violin diagrams; Figure 5 The violin plots of the LF / HF distributions for the four test groups are presented by... Figure 4 and Figure 5 It can be seen that the ESRD group Significantly lower than the CAD group and AMI group (both) P2 < 0.001), showing a gradient change pattern of "ESRD > CAD ≈ AMI", which is highly consistent with the pathophysiological characteristics of the most severe autonomic dysfunction in ESRD patients; while LF / HF showed no significant difference between any two disease groups (both P (2>0.05), making it impossible to distinguish the degree of impact of different diseases on autonomic nerve function. In summary, Compared with the traditional LF / HF index, it has higher sensitivity, stronger disease identification ability and better gradient quantification ability in detecting autonomic dysfunction related to cardiovascular disease.

[0072] Another possible implementation, prior to S5, is based on Embodiment 3. The effectiveness in diagnosing autonomic neuropathy was validated: Example 3: Combination Figure 6 It can be seen that, in distinguishing between healthy individuals and CAD patients, nCP HF It exhibited the best diagnostic performance, with an AUC (Area Under the Curve) of (95% confidence interval: 0.667–0.763). P <0.001, where, P The AUC significance test value is 3, with an optimal cutoff value of -0.262, a sensitivity of 81.0%, a specificity of 51.2%, and a Youden index of 0.321. The AUC of LF / HF is 0.552. P (3=0.071), which did not show significant diagnostic value.

[0073] The results show that Its ability to recognize CAD is significantly better than LF / HF, and its sensitivity is particularly outstanding.

[0074] Combination Figure 7 It can be seen that nCP is effective in distinguishing between healthy individuals and AMI patients. HF The AUC was 0.716 (95% confidence interval: 0.643–0.789). P 3 < 0.001), the optimal cutoff value is -0.323, corresponding to a sensitivity of 73.4%, a specificity of 59.4%, and a Youden index of 0.328. The AUC of LF / HF is 0.540 ( P (3=0.373), which did not show significant diagnostic value.

[0075] The results show that Its ability to identify AMIs is significantly better than that of LF / HF.

[0076] Combination Figure 8 It can be seen that, in distinguishing between healthy individuals and ESRD patients, nCPHF The AUC reached 0.866 (95% confidence interval: 0.811–0.920). P The optimal cutoff value was -0.599 (3 < 0.001), with a sensitivity of 64.7%, a specificity of 91.8%, and a Youden index of 0.565. The AUC of LF / HF was 0.677 (95% confidence interval: 0.584–0.769). P (3 < 0.001), the optimal cutoff value was 1.807, the sensitivity was 51.0%, the specificity was 49.0%, and the Youden index was 0.322. The results indicate that... Its ability to distinguish ESRDs is significantly better than that of LF / HF, and its specificity is particularly outstanding.

[0077] In conclusion, The assay demonstrated superior diagnostic performance compared to the traditional LF / HF index in all three diseases: CAD, AMI, and ESRD. The advantage in identifying CAD and AMI was reflected in higher sensitivity (81.0% and 73.4%, respectively), while the advantage in identifying ESRD was reflected in higher specificity (91.8%). Furthermore, the AUC values ​​for all three diseases exceeded 0.715 (reaching 0.866 for ESRD), indicating… It is a robust, sensitive, and widely adaptable indicator for assessing cardiac autonomic dysfunction across a broad disease spectrum.

[0078] One possible implementation, prior to S5, is based on Embodiment 4. The independent predictive value was verified.

[0079] Example 4 As shown in Table 2, for each disease, multivariate logistic regression analysis was used to construct three progressively advancing models: Model 1 included age, male sex, and the traditional indicator LF / HF; Model 2, based on Model 1, used nCP... HF Instead of LF / HF, Model 3 incorporates age, male sex, LF / HF, and nCP. HF Four variables.

[0080] Table 2. Independent risk factor analysis for three diseases based on multivariate logistic regression.

[0081] In Table 2: OR < 1 indicates that the higher the indicator value, the lower the risk of disease; OR > 1 indicates that the higher the indicator value, the higher the risk of disease. The regression coefficient (OR) and significance test p-value (P4) are used... It means that, among them, <0.05, <0.01, <0.001.

[0082] As shown in Table 2, regardless of whether the traditional power spectral density index LF / HF is included in the statistical model, the new index nCP proposed in this invention... HF It was consistently a highly significant independent predictor of coronary heart disease, myocardial infarction, and end-stage renal disease (OR: 0.001-0.006, P4 < 0.001), i.e., the odds ratio (OR) was 0.001-0.006, representing nCP. HF For every unit increase, the risk of disease decreases by 99.9% to 99.4%; a P4 value less than 0.001 means that the probability of this conclusion being due to random factors is less than one in a thousand.

[0083] After controlling the new indicator nCP HF After considering the nonlinear coupling information, the correlation between the traditional indicator LF / HF and disease risk reversed, changing from a negative association to a positive association. This demonstrates that nCP... HF It is a strong independent predictor of these three cardiovascular diseases, and its predictive effect remains highly robust under different model settings, unaffected by whether other indicators are included.

[0084] It is worth mentioning that nCP HF It played the role of a "suppressive variable" in the regression model. In this application, when nCP was not included... HF At that time, the traditional indicator LF / HF showed an anomalous negative effect (odds ratio OR less than 1, meaning that the lower the LF / HF, the higher the risk of disease), which contradicts the understanding of clinical physiology; however, once it was included in nCP HF By controlling the nonlinear phase coupling information between low and high frequencies, the true and positive predictive effect of LF / HF (odds ratio OR greater than 1, i.e., the higher the LF / HF, the higher the risk of disease) can be observed. This result indicates that nCP HF The predictive value independent of traditional indicators further demonstrates that the true predictive effect of the linear index LF / HF is severely distorted when the confounding factor of nonlinear coupling is not corrected. This highlights the necessity of combining bispectral nonlinear analysis with traditional linear power spectral analysis in the assessment of cardiac autonomic function.

[0085] It should be noted that although the operations of the method of the present invention are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0086] On the other hand, this application also provides a computer-readable storage medium, which may be included in a computer device or exist independently without being assembled into the computer device. The aforementioned computer-readable storage medium stores one or more programs that, when used by one or more processors, execute the methods described in this application. For example, it may execute... Figure 1 The steps of the method shown are as follows.

[0087] This application provides a computer program product including instructions that, when executed, cause the method described in this application to be performed. For example, it can execute... Figure 1 The steps of the method shown are as follows.

[0088] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0089] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for assessing cardiac autonomic nervous system function based on bispectral nonlinear coupling analysis, characterized in that, The method includes: The heartbeat interval sequence of the first subject to be tested is transformed from the time domain to the frequency domain to obtain the heartbeat spectrum sequence. The heartbeat spectrum sequence includes multiple frequency domain components, and each frequency domain component includes a frequency value and an amplitude value and a phase value corresponding to the frequency value, respectively. Based on a preset standard, the low-frequency band and the high-frequency band are divided, and at least one first frequency domain component pair is determined from multiple frequency domain components. The two frequency values ​​in the first frequency domain component pair are adjacent and are respectively in the high-frequency band and the low-frequency band. The cross-coupling power of the high-frequency band and the low-frequency band is determined based on the bispectral estimates of the at least one set of first frequency domain component pairs. The cardiac autonomic nervous function of the first test subject is assessed based on the cross-coupling power.

2. The method according to claim 1, characterized in that, The cross-coupling power of the high-frequency band and the low-frequency band is determined based on the bispectral estimates of the at least one set of first frequency domain component pairs, specifically as follows: In the formula, P LF-HF The cross-coupling power; k 1 and k 2 represents the discrete index value corresponding to the frequency value in its respective frequency domain component. k 1 and k 2 adjacent; and The function is a low-frequency band marking function, which takes the value 1 when the frequency value corresponding to the discrete index value belongs to the low-frequency band, and takes the value 0 otherwise. and This is a high-frequency band marking function; the high-frequency band marking function takes the value of 1 when the frequency value corresponding to the discrete index value belongs to the high-frequency band, and takes the value of 0 otherwise. for( k 1, k 2) The bispectral estimates of the corresponding frequency component pairs.

3. The method according to claim 1, characterized in that, The cardiac autonomic nervous function of the first test subject is assessed based on the cross-coupling power, specifically as follows: The cross-coupling power is normalized to obtain a first evaluation index for the first test subject, and cardiac autonomic nerve function is evaluated based on the first evaluation index of the first test subject.

4. The method according to claim 3, characterized in that, The cross-coupling power is normalized, specifically as follows: Determine at least one pair of second frequency domain components from multiple frequency domain components, wherein the two frequency values ​​in the pair of second frequency domain components are adjacent and both are in the high frequency band; The self-coupling power of the high-frequency band is determined based on the bispectral estimates of each pair of second frequency domain components. The formula for calculating the self-coupling power is as follows: In the formula, This refers to the self-coupling power; I HF ( k 3) and I HF ( k 4) All are high-frequency band marking functions; k 3 and k 4 represents the discrete index value corresponding to the frequency value in its respective frequency domain component, and k 3 and k 4 adjacent; for( k 3, k 4) The bispectral estimates of the corresponding frequency component pairs; The cross-coupling power is normalized based on the self-coupling power.

5. The method according to claim 4, characterized in that, The cross-coupling power is normalized based on the self-coupling power, specifically as follows: In the formula, The first evaluation index for the first test object; The cross-coupling power; The self-coupling power is denoted as .

6. The method according to claim 3, characterized in that, Before assessing the cardiac autonomic nervous function of the first subject based on the first assessment index, the process also includes: Power spectrum indices were determined based on the intercardia sequence of the second test subject. A first evaluation index is determined for the second test object, and the correlation between the first evaluation index of the second test object and the power spectrum index is evaluated. The effectiveness of the first evaluation index is verified based on the correlation.

7. The method according to claim 3, characterized in that, The cardiac autonomic nervous function of the first subject is assessed based on the first evaluation index of the first subject, specifically as follows: The lower the value of the first assessment indicator for the first subject, the higher the risk of cardiac autonomic dysfunction in the first subject.

8. The method according to claim 3, characterized in that, Before assessing the cardiac autonomic nervous function of the first subject based on the first assessment index, the process also includes: Multiple third test subjects are grouped to obtain multiple test groups. The multiple test groups include a healthy group and at least two abnormal groups. The cardiac autonomic nerve function status of each third test subject is known, and the abnormality of each abnormal group is different. For each test group, determine the first evaluation index for each third test subject within the test group; Based on the first evaluation index of all test groups, the ratio of the mean square between groups to the mean square within groups is determined to obtain the F-statistic, and the overall difference between groups is judged based on the associated probability corresponding to the F-statistic. Thus, determine the difference between the arithmetic means of each pair of test groups and the two-sided probability value; Based on the difference and the bilateral probability values, the effectiveness of the first evaluation index in distinguishing different cardiac autonomic nervous system functional states is verified.

9. The method according to claim 8, characterized in that, Based on the difference and the bilateral probability values, the effectiveness of the first evaluation index in distinguishing different cardiac autonomic nervous system functional states is verified, specifically as follows: The pre-defined significance threshold is corrected based on the Bonferroni correction method; Each probability value is compared with the corrected significance threshold to obtain the corresponding comparison result; Based on the differences in the arithmetic means among the groups and the corresponding comparison results, the effectiveness of the first evaluation index in distinguishing different cardiac autonomic nervous system functional states is verified.

10. The method according to claim 1, characterized in that, The method further includes: Collect the electrocardiogram (ECG) signal of the first subject to be tested within a preset time period, and preprocess the ECG signal, the preprocessing including at least one of filtering and noise reduction processing; The preprocessed electrocardiogram signal was subjected to QRS wave detection to obtain the cardiac interval sequence of the first subject.

Images