CN100342820C - Method and apparatus for detecting, and analysing heart rate variation predication degree index - Google Patents

Abstract

The present invention relates to a method and an apparatus for detecting and analyzing a heart rate variation predication degree index. Predication degree PI is used as an analyzing and detecting index of heart rate variation by the present invention. The analyzing and detecting method comprises a calculation method which is characterized in that an electrocardio signal is recorded by a dynamic electrocardiograph; a dynamic electrocardio RR interval is automatically identified under manual assistance to form a time sequence RRi; the nonlinear prediction predication degree PI is given by the recorded time sequence. The detecting index can well react the kinetic behavior performance and the physiological and pathological characteristics of the heart rate variation. The index has strong separating capacity and high significance and has wide clinical application value.

Description

A kind of premeasure index check and analysis method and instrument of heart rate variability

Technical field

The present invention is the check and analysis method and the gauge of heart rate variability premeasure index in a kind of ambulatory electrocardiogram.

Background technology

But the clinical indices of heart rate variability is a kind of index of quantitative repeated application, and it is the quantitative target to the degree of arrhythmia, also is the important quantitative target of estimating the autonomic nerve regulatory function.No matter correct all indexs of measuring heart rate variability are to physiology, pathological study, still to clinical diagnosis and control, all have the meaning of directiveness.Especially to analyze and observe sympathetic nerve and vagal adjusting function, to the risk factor that defines myocardial infarction patient prognosis, the diagnosis of diabetes, paroxysmal arrhythmia, hypertension, sleep apnea syndrome etc. is had important clinical application value.

At present commonly used or common method is: the time domain index that (1) is linear mainly comprises the average of the SDANN/SDANN of the per 5 minute interval of standard deviation SDNN, short distance that detects the omnidistance Cardiac RR of dynamic electrocardiogram interval, the parameter S DSD/NN of adjacent interval₅₀/ PNN₉₀, differential/logarithm index etc.; (2) Xian Xing frequency-domain index mainly comprises general power, extremely low frequency, low frequency, high frequency, low frequency high frequency normal state and the low frequency high frequency ratio of short distance (per 5 minutes); The linear interpolation slope of the frequency spectrum in intrasonic, extremely low frequency, low frequency, high frequency and the logarithmic coordinates of long-range (24 hours); (3) nonlinear analysis method, using maximum at present is Poincare (Poincar é) scatterplot of RR interval, RR interval difference, mainly relies on naked eyes visual picture feature to distinguish.Further clinical application for existing time domain, frequency domain linear method, its main crux is to be difficult to accurately portrayal and to analyze that changes in heart rate---this is subjected to hemodynamics, the very complicated process of multinomial factor affecting such as the variation of electric physiology and hormone and autonomic nerve and cental system.So existing linear method is too coarse, noise jamming is relatively more responsive to external world, and its mathematical model does not also meet this basic fact that the heart rate regulator control system itself is a nonlinear system.For existing non-linear scatterplot analytical method, main still by means of the observation of naked eyes and the analysis of some simple numerical value quantitative targets, and this is far from being enough, does not also extremely meet the basic fact of this high dimensional nonlinear system of changes in heart rate.So no matter be those existing class methods, its significant limitations is all arranged, the clinical application and the popularization of heart rate variability metrics brought obstruction.

Summary of the invention

The objective of the invention is to propose a kind of more can objective reaction heart rate variability process in ambulatory electrocardiogram and the heart rate variability check and analysis method of nonlinear characteristic, and provide a kind of detecting instrument corresponding to the principle of the said method non-linear index of heart rate variability clear, simple in structure.

Heart rate variability check and analysis method is called premeasure PI (Predictability Index) detection method in the ambulatory electrocardiogram that the present invention proposes, promptly the analyzing and testing index of premeasure PI as heart rate variability.The check and analysis method of premeasure PI is as follows: (1) utilizes K hour core signal ECG of dynamic electrocardiogram instrument record; (2) with core signal ECG playback in computer of gathering, wherein consider links such as filtering, digital signal compression, thereby show K hour dynamic electrocardiogram waveform; (3) based on artificial auxiliary automatic identification dynamic electrocardiogram RR interval down, formation time series RR_i={ ri₁, ri₂..., ri_j... ri_n, i=1 wherein, 2 ..., K; (4) based on the jerk (being as the criterion) at the automatic identification dynamic electrocardiogram R peak under artificial the assisting, constitute time series RL with horizontal base line_i={ li₁, li₂..., li_j... li_n, i=1 wherein, 2 ..., K; (5) time series by above-mentioned record, the value of the nonlinear prediction degree PI of the corresponding heart rate variability of analysis to measure.Above-mentioned this nonlinear prediction degree index is based on fully sets up suitable theoretical model and strict computational methods.Below we provide the correlation model and the computational methods of this index.

The horal RR interval time series RR that we will obtain_i(i=1,2 ..., K) front and back are divided into two sections, i.e. RR_i=LRR_In/2∪ TRR_In/2, and the order of sequence remains unchanged in every section.Wherein, LRR_In/2Be used to construct nonlinear model as guidance section (Supervised-Part), and TRR_In/2(Test-Part) is used to calculate premeasure as detection segment, and K is dynamic electrocardiogram writing time, and the general value of K is the natural number between the 20-32, gets K=24 usually, is diel.After the Nonlinear Mechanism in fully taking into account dynamic electrocardiogram RR interval time series, we can set up following nonlinear mapping (function) relation:

Φ：RR_imLRR_in/2R^m→R

Portray dynamic electrocardiogram RR interval seasonal effect in time series front and back iterative relation.Wherein, from the natural number of the general value of spatial dimension m between 10-50 under the variable vector.In order to construct this many-to-one nonlinear function Φ, utilize the function approximation principle, we have provided following algorithm, are limited in dynamic heart rate interval time series LRR in the hope of obtaining nonlinear function Φ_In/2On approximate expression ξ.

So we suppose that at first approximate expression ξ has following form:

Wherein, symbol " ο " representative function compound, for any j=1,2 ..., h, each function$f_j=(z_{j1},z_{j2},\·\·\·z_{{\mathrm{jm}}_j}),$Z especially_0p=x_p, p=1,2 ..., m; And

$z_{\mathrm{jk}}=f(\underset{p=1}{\overset{m_{j-1}}{\mathrm{\Σ}}}{w}_{\mathrm{jkp}}{z}_{j-1,p}-{\mathrm{\δ}}_{\mathrm{jk}})\·f\left(\underset{p=0}{\overset{m_{j-1}}{\mathrm{\Σ}}}{w}_{\mathrm{jkp}}{z}_{j-1,p}\right),---\left(1.2\right)$

k＝1，2，…，m_j，j＝1，2，…，h，

$f_{h+1}=f(\underset{p=1}{\overset{m_h}{\mathrm{\Σ}}}{w}_{h+\mathrm{1,1},p}{z}_{\mathrm{hp}}-{\mathrm{\δ}}_{h+\mathrm{1,1}})\·f\left(\underset{p=0}{\overset{m_h}{\mathrm{\Σ}}}{w}_{h+\mathrm{1,1},p}{z}_{\mathrm{hp}}\right),---\left(1.3\right)$

Wherein, w_JkpBe weight coefficient, δ_JkBe the threshold value coefficient, be designated as weight coefficient w equally_Jk0, constant-1 is designated as normal value input z_J-1,0Function f is the Sigmoid class function, for example can get work: f (x)=tanh (θ x),$f\left(x\right)=\frac{1}{1+e^{-\mathrm{\θx}}},$θ be value in interval (0,1] on attenuation parameter.If can determine weight coefficient w_JkpValue, we have also just constructed approximate function ξ come out so.

Below, we will be according to the dynamic electrocardiogram RR time series LRR of interval_In/2Provide and determine weight coefficient w_JkpA kind of method, thereby determine the expression formula of function ξ, construct the nonlinear mapping model.

(1) initialize: utilize random function to generate weight coefficient w at random_JkpInitial value (initial value that comprises the threshold parameter that is designated as weight coefficient), attenuation parameter θ elects the real number on the interval (0,1) as;

(2) with LRR_In/2In electrocardio sequence ri_P+1, ri_P+2..., r_P+m(p=0,1,2 ... n/2-m-1) as the initial value input of m the independent variable of function ξ, so, can obtain (n/2-m) individual value: y of function ξ according to formula (1.1), (1.2) and (1.3)_P+m+1(p=0,1,2 ... n/2-m-1);

(3) be calculated as follows error function:

$\mathrm{Olderror}=\underset{p=0}{\overset{n/2-m-1}{\mathrm{\Σ}}}{(y_{p+m+1}-{y^{*}}_{p+m+m})}^2,$

Y wherein^*_P+m+1=ri_P+m+1

(4) for weight coefficient w_JkpGiven iterative increment Δ w_Jkp, and put w*_Jkp=w_Jkp+ Δ w_Jkp, calculate (n/2-m) individual value and the error function of corresponding function ξ by step (2), (3), and remember that the value that error function thus provides is Newerror;

(5) if Newerror＜Olderror puts w so_Jkp=w*_Jkp, Olderror=Newerror, and to turning to step (4); Otherwise turn to step (6);

(6) put w*_Jkp=w_Jkp-Δ w_Jkp, calculate (n/2-m) individual value and the error function of corresponding function ξ by step (2), (3), and remember that the value that error function thus provides is Newerror;

(7) if Newerror＜Olderror puts w so earlier_Jkp=w*_Jkp, Olderror=Newerror.And then turn to step (4); Otherwise, directly turn to step (4);

(8), obtain the optimum relatively { w of gang through the multiple cycles iteration_Jkp.We just can be limited in dynamic heart rate interval time series LRR by constructed fuction Φ like this_In/2On approximate expression ξ.

We know, the regulating system of cardiac rhythm itself is a nonlinear system, and normal person's autonomic nervous system (comprising sympathetic nerve and vagus nerve) to external world variation and the stimulation of environment be relative responsive, so the function phi that constructs is limited in dynamic heart rate interval time series LRR_In/2On approximate expression ξ can accurately reflect is LRR_In/2System features on time period, and be limited for the sign ability that section At All Other Times goes up Nonlinear Mechanism.Therefore, for the normal people of autonomic nervous system function, the function that corresponding function phi becomes when in fact being can be designated as Φ (t).This just is the specific algorithm of our premeasure given below, has provided the physiological explanation of medical science.Therefore, detect TRR with regard to the available above-mentioned approximate expression ξ that obtains below_In/2Electro-cardio interval time series on time period, thus the specific algorithm of premeasure provided.

(9) utilize the expression formula (1.1), (1.2) and (1.3) of the function ξ that provides in the step (8), ri_P+1, ri_P+2..., r_P+m(p=n/2 ..., n-m-1) calculate y_P+m+1(p=n/2 ..., n-m-1);

(10), can calculate RR interval series { y according to the formula of canonical correlation coefficient_P+m+1And { ri_P+m+1Correlation coefficient CCi-1,

$\mathrm{CCi}-1=\frac{\underset{p=n/2}{\overset{n-m-1}{\mathrm{\Σ}}}(y_{p+m+1}-\stackrel{\‾}{y})({\mathrm{ri}}_{p+m+1}-\stackrel{\‾}{\mathrm{ri}})}{\sqrt{\underset{p=n/2}{\overset{n-m-1}{\mathrm{\Σ}}}{(y_{p+m+1}-\stackrel{\‾}{y})}^2\underset{p=n/2}{\overset{n-m-1}{\mathrm{\Σ}}}{({\mathrm{ri}}_{p+m+1}-\stackrel{\‾}{\mathrm{ri}})}^2}},$

Wherein, y and ri are respectively sequence { y_P+m+1And { ri_P+m+1Average, the numerical value that we provide CCi-1 calls the heart beating interval at i hour one-step prediction degree;

(11) similarly, our RR interval forecasting sequence { y to obtaining_P+m+1(p=n/2 ..., n-m-1) can take out m value equally successively, and the method for utilization step (9) generates new RR interval forecasting sequence { y as independent variable_J+m+1(j=n/2+m ..., n-m-1), thereby can obtain new correlation coefficient CCi-2, so the numerical value that we provide CCi-2 calls the heart beating interval at i hour two step premeasures; By that analogy, we can calculate any rational i hour N step premeasure, and the general value of natural number N is 10 ~ 20.Referring to the ten six step premeasures of Fig. 1 about a certain hour of interval of RR aroused in interest;

(12) because a certain hour the premeasure that we often provide is multiple step-length, so we utilize regressive method, provide 1 to N step premeasure CCi-1, CCi-2 ..., the slope S C of CCi-N_Ni=CCN_i/ N_i,

Wherein,${\mathrm{CCN}}_i=\underset{j=1}{\overset{N}{\mathrm{\Σ}}}j\·\mathrm{CCi}-j-N(N+1)\·\stackrel{\‾}{\mathrm{CCi}}/2,$$N_i=\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{j}^2-\frac{N{(N+1)}^2}{4},$

$\stackrel{\‾}{\mathrm{CCi}}=\frac{1}{N}\underset{j=1}{\overset{N}{\mathrm{\Σ}}}\mathrm{CCi}-j.$

We claim SC_NiBe heart beating interval premeasure PI at i hour_Ni

So far, we utilize the model for heart beating interval structure specifically to calculate the non-linear index of premeasure PI.This index can reflect the autonomic nervous system function quality of regulating cardiac rhythm, strong and weak non-linear index.Just when a certain hour premeasure value relatively hour, show the regulatory function more weak or complete disorder of reflection relatively of autonomic nerve in this hour, opposite then show that the regulatory function of autonomic nerve is stronger in this hour.

The detection method of the non-linear index of heart rate variability---premeasure---of quantitative assessment autonomic nervous function and the basic ideas of model construction have below promptly been provided.Certainly for providing the corresponding nonlinear prediction degree of R peak jerk seasonal effect in time series index, can the clinical medicine physiological mechanism be described with above similar approach.

According to the detection method of above-mentioned heart rate variability nonlinear prediction degree index, the present invention has designed the relevant detection device especially, and this device comprises dynamic electrocardiogram recorder, electrocardiosignal pretreatment system and nonlinear prediction degree index detection system.

The dynamic electrocardiogram device is divided into two kinds, and a kind of is the magnetic tape type recorder, and a kind of is flash of light cassette recorder.All possess 24-30 hour dynamic electrocardiogram and gather memory function.

In the electrocardiosignal pretreatment system, our employing of data that provides for the magnetic tape type recorder is a core with the single-chip microcomputer, is furnished with read-write memory (RAM), read only memory (ROM), A/D converter and computer and connects the parallel communication interface; The data that provide for flash of light cassette recorder can be directly by Flash card USB reader and be equipped with the corresponding driving program, decoding software gets final product.In addition, solidify by software, in read only memory, also comprise the electrocardiosignal automatic filter, discern functions such as R peak automatically at pretreatment system.Computer can obtain and show the PQRST wave group (see figure 2) that automatic identification R peak is later by parallel communication interface or USB interface.In addition, on computers, can do further artificial the correction in the R peak of identification automatically to machine, and can reject the ARR waveform of for example premature beat and so on, thereby obtain to be used to calculate the electro-cardio interval time series RR of heart rate variability nonlinear prediction degree index_iAnd RL_iCalculation procedure (1)-(12) compiled program is formed nonlinear prediction degree index detection system.Processing procedure is seen Fig. 2 and shown in Figure 3.

The method that the non-linear index prediction degree of the heart rate variability that the present invention provides detects is to be based upon on the basis that strict theory of nonlinear dynamic system and autonomic nerve regulating system have Nonlinear Mechanism.The index system of being set up be based upon the science modeling and 300 the example above clinical sample controlled trials the basis on (this sample is from First People's Hospital, Shanghai, Wuhan hospital of Tongji University and international standard MIT data base etc.).Therefore, this cover index can reflect the dynamic behavior feature of changes in heart rate and physiology, pathological characters preferably, and this index separating capacity is strong, and the significance height has wide clinical value.The fact shows, the present invention and corresponding instrument can be widely used in the clinical evaluation for the autonomic nerve regulatory function, for the evaluation of the risk factor of myocardial infarction, for the early diagnosis of cardiovascular disease, diagnosis in time and accurately diagnosis useful appraisement system is provided; Simultaneously, its objective evaluation screening and further developing of promotion cardiovascular research field to the pharmaceutical properties of cardiovascular disease has great importance.

Description of drawings

Fig. 1 be certain sample 16 step of certain hour premeasure CCi-j (j=1,2 ..., 16) junction curve.

Fig. 2 is PQRST wave group and R peak identification diagram.

Fig. 3 calculates nonlinear prediction degree index flow chart for dynamic electrocardiogram (ECG) data.

The specific embodiment

Below provide some samples about nonlinear prediction degree index S C_Ni(i=24 ..., 28) The actual calculation and certain sample a certain hour 16 step premeasure CCi-j (j=1,2 ..., 16) the junction curve (see figure 1).

0-normal person's sample, 1-has patient's sample

In this computational process, specifically get as minor function and parameter:

$f\left(x\right)=\frac{1}{1+e^{-\mathrm{\θx}}},$θ＝0.25，

K=24, h=3, n is by per hour being obtained issue between sample RR (because of sample and time variation difference), m=20, N=16.

Sample institute number	Health status	24:00	25:00	26:00	27:00	28:00	24 hours
								3724 5001 5012 5019 5021 6023 6024 6037 6040 6080 6089 6405 6406 6419 6426	1 1 1 1 1 0 0 0 0 0 0 0 0 0 0	0.005426 -0.01632 0.009304 -0.04537 -0.04654 -0.00488 -0.03817 -0.00319 -0.00688 -0.00584 -0.02833 -0.00141 -0.01157 -0.03699 -0.03773	-0.00535 -0.03791 -0.025 0.008217 -0.01927 -5.4E-05 -0.0401 0.002094 -0.00344 -0.01413 -0.05925 0.005804 -0.0041 -0.00744 -0.01816	-0.0204 -0.04427 -0.0172 -0.02478 0.000594 -0.00387 -0.02547 0.002925 -0.00997 -0.02278 -0.03126 -0.01571 -0.01509 -0.04603 -0.01669	-0.01685 0.010266 -0.03344 -0.03098 -0.03533 -0.0232 -0.0308 -0.00212 -0.02001 -0.01202 -0.00755 -0.0019 -0.01361 -0.00592 -0.02787	-0.00089 -0.0087 -0.05236 -0.02518 -0.038 -0.01474 -0.01859 0.002342 -0.00503 -0.02209 -0.01393 -0.02048 -0.0143 -0.01725	-0.01435 -0.01663 0.006625 -0.02532 -0.00185 -0.00868 -0.02172 -0.01327 -0.02087 -0.02277 -0.03768 -0.00504 -0.02761 -0.04548 -0.006

6427 6517 6542 6573

0 0 0 0

-0.01881 -0.05287 -0.04327 -0.00949

-0.02464 -0.04735 -0.02492 -0.00539

-0.00694 -0.0408 -0.02947 -0.01027

-0.01206 -0.02645 -0.00437 -0.01086

-0.02685 -0.05165 -0.02846 -0.00586

0.00252 -0.02801 -0.02353 -0.00561

Claims (5)

1, a kind of premeasure index check and analysis method of heart rate variability is characterized in that (1) utilizes K hour core signal ECG of dynamic electrocardiogram instrument record; (2) with core signal ECG playback in computer of gathering, shows K hour dynamic electrocardiogram waveform: (3) are based on artificial automatic identification dynamic electrocardiogram RR interval under auxiliary, formation time series RR_i={ ri₁, ri₂..., ri_j... ri_n), i=1,2 ..., K; (4) based on the jerk at the automatic identification dynamic electrocardiogram R peak under artificial the assisting, constitute time series RL_i={ li₁, li₂..., li_j... li_n, i=1,2 ..., K; (5) time series by above-mentioned record, the value of the nonlinear prediction degree PI of the corresponding heart rate variability of analysis to measure, K is dynamic electrocardiogram writing time here, the K value is the natural number between the 20-32.

2, check and analysis method according to claim 1 is characterized in that horal RR interval time series RR_iFront and back are divided into two sections: RR_i=LRR_In/2∪ TRR_In/2, and the order of sequence remains unchanged in every section, wherein, and LRR_In/2As instructing section to be used to construct nonlinear model, TRR_In/2Be used to calculate premeasure as detection segment; Portray iterative relation before and after the dynamic electrocardiogram RR interval seasonal effect in time series if concern with following nonlinear mapping:

Φ：RR_imLRR_in/2R^m→R，

Wherein, m is a spatial dimension under variable vector, and m gets the natural number between the 10-50;

If ξ is the approximate expression of nonlinear function on dynamic heart rate interval time series, form is as follows:

Wherein, symbol " ° " representative function compound, for any j=1,2 ..., h, each function$f_j=(z_{j1},z_{j2},\·\·\·z_{{\mathrm{jm}}_j}),$z_0p=x_p, p=1,2 ..., m; And

$z_{\mathrm{jk}}=f(\underset{p=1}{\overset{m_{j-1}}{\mathrm{\Σ}}}{w}_{\mathrm{jkp}}{z}_{j-1,p}-{\mathrm{\δ}}_{\mathrm{jk}})\·f\left(\underset{p=0}{\overset{m_{j-1}}{\mathrm{\Σ}}}{w}_{\mathrm{jkp}}{z}_{j-1,p}\right),k=\mathrm{1,2},\·\·\·,m_j,j=\mathrm{1,2},\·\·\·,h,---\left(1.2\right)$

$f_{h+1}=f(\underset{p=1}{\overset{m_h}{\mathrm{\Σ}}}{w}_{h+\mathrm{1,1},p}{z}_{\mathrm{hp}}-{\mathrm{\δ}}_{h+\mathrm{1,1}})\·f\left(\underset{p=0}{\overset{m_h}{\mathrm{\Σ}}}{w}_{h+\mathrm{1,1},p}{z}_{\mathrm{hp}}\right),---\left(1.3\right)$

Wherein, w_JkpBe weight coefficient, δ_JkBe the threshold value coefficient, be designated as weight coefficient w equally_Jk0, constant-1 is designated as normal value input z_J-1,0Function f is the Sigmoid class function.

3, check and analysis method according to claim 2 is characterized in that definite step of function ξ is as follows:

(1) initialize: utilize random function to generate weight coefficient w at random_JkpInitial value, attenuation parameter θ elects the real number on the interval (0,1) as;

(2) with LRR_In/2In electrocardio sequence ri_P+1, ri_P+2..., r_P+m, p=0,1,2 ... n/2-m-1 as the initial value input of m the independent variable of function ξ, according to formula (1.1), (1.2) and (1.3), obtains (n/2-m) individual value: y of function ξ_P+m+1, p=0,1,2 ... n/2-m-1;

(3) be calculated as follows error function:

$\mathrm{Olderror}=\underset{p=0}{\overset{n/2-m-1}{\mathrm{\Σ}}}{(y_{p+m+1}-{y^{*}}_{p+m+1})}^2,$

Y wherein^*_P+m+1=ri_P+m+1

(4) for weight coefficient w_JkpGiven iterative increment Δ w_Jkp, and put w*_Jkp=w_Jkp+ Δ w_Jkp, calculate (n/2-m) individual value and the error function of corresponding function ξ by step (2), (3), and remember that the value that error function thus provides is Newerror;

(5) if Newerror＜Olderror puts w so_Jkp=w*_Jkp, O1derror=Newerror, and to turning to step (4); Otherwise turn to step (6);

(6) put w*_Jkp=w_Jkp-Δ w_Jkp, calculate (n/2-m) individual value and the error function of corresponding function ξ by step (2), (3), and remember that the value that error function thus provides is Newerror;

(7) if Newerror＜Olderror puts w so earlier_Jkp=w*_Jkp, Olderror=Newerror, and then turn to step (4); Otherwise, directly turn to step (4);

(8), obtain the optimum relatively { w of gang through the multiple cycles iteration_Jkp.

4, check and analysis method according to claim 3 is characterized in that further utilizing expression formula ξ to come electro-cardio interval time series on section detection time, and calculates premeasure, and concrete steps are as follows:

(1) utilizes the expression formula (1.1), (1.2) and (1.3) of the function ξ that provides in the step (8), ri_P+1, ri_P+2..., r_P+m, p=n/2 ..., n-m-1 calculates y_P+m+1, p=n/2 ..., n-m-1;

(2), calculate RR interval series { y according to the formula of canonical correlation coefficient_P+m+1And { ri_P+m+1Correlation coefficient CCi-1:

$\mathrm{CCi}-1=\frac{\underset{p=n/2}{\overset{n-m-1}{\mathrm{\Σ}}}(y_{p+m+1}-\stackrel{\‾}{y})({\mathrm{ri}}_{p+m+1}-\stackrel{\‾}{\mathrm{ri}})}{\sqrt{\underset{p=n/2}{\overset{n-m-1}{\mathrm{\Σ}}}{(y_{p+m+1}-\stackrel{\‾}{y})}^2\underset{p=n/2}{\overset{n-m-1}{\mathrm{\Σ}}}{({\mathrm{ri}}_{p+m+1}-\stackrel{\‾}{\mathrm{ri}})}^2}},$

Wherein, y and ri are respectively sequence { y_P+m+1And { ri_P+m+1Average, the numerical value that we provide CCi-1 calls the heart beating interval at i hour one-step prediction degree;

(3) similarly, the RR interval forecasting sequence { y to obtaining_P+m+1, p=n/2 ..., n-m-1 takes out m value equally successively as independent variable, and the method for utilization step (1) generates new RR interval forecasting sequence { y_J+m+1, j=n/2+m ..., n-m-1, thus new correlation coefficient CCi-2 obtained, and the numerical value that CCi-2 is provided calls the heart beating interval at i hour two step premeasures; By that analogy, calculate any rational i hour N step premeasure, natural number N value is 10 ~ 20;

(4) utilize regressive method, provide 1 to N step premeasure CCi-1, CCi-2 ..., the slope S C of CCi-N_Ni=CCN_i/ N_i,

Wherein,${\mathrm{CCN}}_i=\underset{j=1}{\overset{N}{\mathrm{\Σ}}}j\·\mathrm{CCi}-j-N(N+1)\·\stackrel{\‾}{\mathrm{CCi}}/2{,N}_i=\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{j}^2-\frac{N{\left(N+1\right)}^2}{4},$

$\stackrel{\‾}{\mathrm{CCi}}=\frac{1}{N}\underset{j=1}{\overset{N}{\mathrm{\Σ}}}\mathrm{CCi}-j.$

Claim SC_NiBe heart beating interval premeasure PI at i hour_Ni

5, a kind of realization instrument of premeasure index check and analysis method of heart rate variability as claimed in claim 4 is characterized in that being made up of dynamic electrocardiogram recorder, electrocardiosignal pretreatment system and nonlinear prediction degree index detection system; Wherein, the dynamic electrocardiogram recorder adopts magnetic tape type recorder or flash of light cassette recorder; In the electrocardiosignal pretreatment system, the The data that provides for the magnetic tape type recorder is a core with the single-chip microcomputer, is furnished with read-write memory, read only memory, A/D converter and computer and connects the parallel communication interface; The data that provide for flash of light cassette recorder directly by Flash card USB reader and be equipped with the corresponding driving program, decoding software gets final product; Nonlinear prediction degree index detection system is made up of the computer program of describing described each calculation procedure of nonlinear prediction degree index.

Images