Method for analyzing heart rate variability, apparatus and use thereof

Abstract

A method for analyzing heart rate variability, and an apparatus and use thereof, the method for analyzing heart rate variability including collecting ECG data in vitro; digitizing and denoising the ECG data; forming the processed ECG data into a sinus NN interval sequence; selecting sinus NN interval data of 4 hours in an awake state; performing MSE calculation on the sinus NN interval sequence of 4 hours in an awake state; and extracting parameters representing the complexity of a heart rate by using MSE curves. The present invention may provide accurate and efficient screening for drug-refractory epilepsy patients who are suitable for vagus nerve stimulation surgery, thus avoiding unnecessary expenses, and avoiding missing the most opportune moment for treatment. At the same time, patients suitable for VNS surgery are selected by using MSE complexity feature parameters of ECG, thus improving the overall efficacy of VNS therapy.

Claims

1. A computer-implemented method for analyzing heart rate to identify a patient with drug-resistant epilepsy suitable for Vagus Nerve Stimulation (VNS) surgery comprising: collecting electrocardiography (ECG) data in vitro for the patient with drug-resistant epilepsy; digitizing and denoising the ECG data; forming the digitized and denoised ECG data into a sinus NN interval data; selecting a portion of sinus NN interval data from the formed sinus NN interval data corresponding to a first period of time, the first period of time being while the first patient was in an awake state; performing a multiscale entropy (MSE) calculation on the selected sinus NN interval data in the awake state, comprising: performing coarse grained processing on the selected portion of sinus NN interval data {X.sub.1, . . . , x.sub.i, . . . , x.sub.N} and obtaining therefrom reconstructed sequences using a plurality of different scale factors and an equation defined as $y_j^{\tau }=1/\tau \underset{i=\left(j-1\right)\tau +1}{\overset{j\tau }{.Math.}}{x}_i,1\le j\le N/\tau ,$ with τ as a given scale factor of the plurality of different scale factors, and the plurality of different scale factors including scales 1-20; calculating a sample entropy for each reconstructed sequence with a different scale factor τ; drawing a MSE curve of the sample entropy with respect to the plurality of different scale factors with the plurality of different scale factors being along a horizontal axis and the sample entropy as a vertical axis, the curve having a plurality of segments corresponding to a respective one of the plurality of different scale factors; 1) extracting parameters representing a heart rate complexity based on the MSE curve, wherein extracting parameters representing the heart rate complexity based on the MSE curve comprises: for the curve of the sample entropy, obtaining a slope parameter by linearly fitting points of the curve that correspond with the scale 1 to scale 5; dividing the curve from the scale 1 to the scale 20 into a plurality of segments; and calculating Area parameters for each area encompassed by each segment of the plurality of segments of the curve to obtain the parameters representing heart rate complexity; 2) determining thresholds for the parameters representing heart rate complexity, the thresholds to identify patients with drug-resistant epilepsy suitable for VNS surgery, and identifying a second patient with drug-resistant epilepsy suitable for VNS surgery based on the thresholds; wherein the Area parameters are Area 1-5 which is based on scale 1 to scale 5, Area 6-15 which is based on scale 6 to scale 15, and Area 6-20 which is based on scale 6 to scale 20; wherein the threshold of the slope parameter is 0.071±0.002, the threshold of Area 1-5 is 4.32±0.04, the threshold of Area 6-15 is 10.57±0.2 and the threshold of Area 6-20 is 15.85±0.3.

2. The method of claim 1, wherein identifying the second patient with a drug-resistant epilepsy suitable for VNS surgery is further based on performing a multiscale entropy (MSE) calculation on the ECG data of the second patient, the MSE calculation to generate a corresponding curve; and wherein the method further comprises: extracting the Slope parameter and Area parameters for the second patient; and determining the second patient is eligible for VNS surgery based on the one or more parameters having an associated value that exceeds a corresponding threshold value of the thresholds for the extracted parameters representing heart rate complexity.

3. A device comprising: a monitoring apparatus configured to: collect electrocardiography (ECG) data in vitro for a first patient with drug-resistant epilepsy; digitize and denoising the ECG data; format the digitized and denoised ECG data into a sinus NN interval data; select a portion of sinus NN interval data from the formatted sinus NN interval data corresponding to a first period of time, the first period of time being while the first patient was in an awake state; perform a multiscale entropy (MSE) calculation on the selected sinus NN interval data in the awake state based on: performing coarse grained processing on the selected portion of sinus NN interval data {X.sub.1, . . . , x.sub.i, . . . , x.sub.N} and obtaining therefrom reconstructed sequences using a plurality of different scale factors and an equation defined as $y_j^{\tau }=1/\tau \underset{i=\left(j-1\right)\tau +1}{\overset{j\tau }{.Math.}}{x}_i,1\le j\le N/\tau ,$ with τ as a given scale factor and the plurality of different scale factors including scales 1-20; calculating a sample entropy for each reconstructed sequence with a different scale factor τ; drawing an MSE curve of the sample entropy with respect to the plurality of different scale factors with the plurality of different scale factors being along a horizontal axis and the sample entropy along a vertical axis and the curve having a plurality of segments corresponding to a respective one of the plurality of different scale factors; extracting parameters representing a heart rate complexity based on the MSE curve, wherein extracting parameters representing the heart rate complexity based on the MSE curve comprises: for the MSE curve of the sample entropy, obtaining a slope parameter by linearly fitting points of the MSE curve that correspond with the scale 1 to scale 5; dividing the curve from the scale 1 to the scale 20 into a plurality of segments; calculating Area parameters for each area encompassed by each segment of the plurality of segments of the curve to obtain the parameters representing heart rate complexity; determining thresholds for the parameters representing heart rate complexity, the thresholds for identification of a second patient with drug-resistant epilepsy suitable for VNS surgery, the second patient different than the first patient; identifying the second patient with drug-resistant epilepsy suitable for VNS surgery based on the thresholds; wherein the Area parameters are Area 1-5 which is based on scale 1 to scale 5, Area 6-15 which is based on scale 6 to scale 15, and Area 6-20 which is based on scale 6 to scale 20; wherein the threshold of the slope parameter is 0.071±0.002, the threshold of Area 1-5 is 4.32±0.04, the threshold of Area 6-15 is 10.57±0.2 and the threshold of Area 6-20 is 15.85±0.3.

4. The device of claim 3, wherein the monitoring apparatus is wearable.

Description

BRIEF DESCRIPTION OF DRAWINGS

(1) FIG. 1 is a flow chart of the prior art for screening patients with VNS indications.

(2) FIG. 2 is a diagram of 12-lead ECG acquisition.

(3) FIG. 3 is a flow chart of ECG signal processing.

(4) FIG. 4 is a flow chart of MSE analysis method.

(5) FIG. 5 is a diagram of extracting indicators of MSE Complexity.

(6) FIG. 6 is a threshold selection ROC curve

(7) FIG. 7 is a flow chart for judgment.

(8) FIG. 8 shows MSE curves of the effective group and the ineffective group

EMBODIMENTS

Example 1

(9) As shown in FIG. 2, a standard 12-lead ECG acquisition for 24 hours before surgery requires: sampling frequency of the ECG acquisition device is greater than or equal to 500 Hz; during ECG recording, activities such as strenuous exercises and medications that could affect cardiac functions should be avoided; the recording period is 24 hours. Recording environment and conditions of subjects should be basically similar. The data used for HRV analysis should be ensured to be normal sinus NN intervals. During HRV analysis, normal sinus NN intervals of 4 hours are selected for MSE analysis from the 24-hour long-term ECG records during which the subject is in an awake state. The specific processing flow of ECG signals is shown in FIG. 3.

(10) 1) collecting and digitizing ECG signals;

(11) 2) denoising and de-articulating digital signals;

(12) 3) automatically detecting QRS waves thereof;

(13) 4) manually inspecting QRS waves of the detected signals;

(14) 5) removing ectopic exciting QRS wave signals;

(15) 6) forming a sinus NN interval sequence;

(16) 7) selecting 4-hour sinus NN interval sequences in the case of a subject in an awake state;

(17) 8) calculating MSE based on the 4-hour sinus NN interval sequences;

(18) 9) drawing a MSE curving, wherein the scale factor is as the horizontal and the entropy value corresponding to the scale factor is as the ordinate;

(19) 10) extracting characteristic parameters representing heart rate complexity, according to the MSE curve.

(20) The MSE calculation method in the HRV analysis adopted in the present invention extracts characteristic parameters of Slope5, Area1-5, Area6-15, and Area6-20 to represent the heart rate complexity.

(21) The MSE method has the following steps (see FIG. 4):

(22) (1) performing coarse grained processing on the 4-hour normal sinus NN interval sequence {X.sub.1, . . . , X.sub.i, . . . , X.sub.N} to obtain reconstructed sequences

(23) $y_j^{\tau }=1/\tau \underset{i=\left(j-1\right)\tau +1}{\overset{j\tau }{.Math.}}{x}_i,1\le j\le N/\tau$
with different scales, τ as a scale factor;

(24) (2) calculating a sample entropy, for each scale's sequence

(25) $y_j^{\tau }=1/\tau \underset{i=\left(j-1\right)\tau +1}{\overset{j\tau }{.Math.}}{x}_i;$

(26) (3) drawing a curve of the sample entropies with respect to the different scale factors, shown in FIG. 5; linearly fitting points of scales 1-5 to obtain slope 5; then calculating Area1-5, Area 6-15, and Area 6-20 encompassed by scale 1-5, scale 6-15, scale 6-20 curve and the horizontal axis, wherein, the above four parameters are characteristic parameters representing the heart rate complexity.

(27) For patients with drug-resistant epilepsy, 24-hour electrocardiogram acquisition was performed before surgery. The 24-hour electrocardiogram data collected was processed, according to the above-mentioned method, to obtain normal sinus NN interval sequences in 4 hours during which a subject is in an awake state. The MSE analysis was performed on the above-mentioned 4-hour NN interval sequences according to the above-mentioned method. Characteristic parameters such as Slope5, Area1-5, Area6-15, Area6-20, etc., which represent heart rate complexity, were extracted. Then comprehensive judgment selection was performed through corresponding threshold judgment (as shown in FIG. 6). The VNS surgical patients in the training set were classified according to follow-up efficacy after a certain period of time (effective group and ineffective group). The above-mentioned heart rate complexity indicators of the effective group and the ineffective group were statistically analyzed, and Receiver Operating Characteristic (ROC) curves were drawn for Slope5, Area 1-5, Area 6-15, and Area 6-20. The threshold of each indicator (Youden index) is a point in each curve which has the shortest distance to the top left corner (that is, the coordinates (1, 1)). Finally, patients who are eligible for VNS surgery and patients who are not eligible for VNS surgery are distinguished based on the corresponding threshold (as shown in FIG. 7). When the four characteristic parameters Slope5, Area1-5, Area6-15 and Area6-20, representing heart rate complexity, are used to distinguish patients eligible for VNS surgery from patients ineligible for VNS surgery respectively, their corresponding threshold value selection and their corresponding screening accuracy are as follows:

(28) When Slope5=0.071, patients with a value higher than said value were considered eligible for VNS surgery. The screening accuracy was 67.9%.

(29) When Area 1-5=4.32, patients with a value higher than said value were considered eligible for VNS surgery. The screening accuracy was 71.4%.

(30) When Area6-15=10.57, patients with a value higher than said value were considered eligible for VNS surgery. The screening accuracy was 92.9%.

(31) When Area6-20=15.85, patients with a value higher than said value were considered eligible for VNS surgery. The screening accuracy was 96.4%.

Example 2

(32) The complexity indicator Area 6-n, when the scale factor in the MES analysis method of Example 1 is expanded to n, could also be used for screening VNS patients as described above.

(33) In the present invention, for patients with drug refractory epilepsy, preoperative electrocardiogram acquisition 24 hours before surgery and the MSE analysis of HRV were performed. In this way, patients with drug refractory epilepsy could be screened before surgery, thereby guiding patients who are not eligible for VNS therapy not to receive the surgery and choose other therapies, which could save unnecessary expenditures and avoiding delaying the optimal timing for treatment. Meanwhile, patients with VNS surgical indications were clearly selected by extracting characteristic parameters representing heart rate complexity through ECG's MSE curve, which could improve overall VNS therapeutic efficacy.

Example 3

(34) In accordance with the above screening method, 32 patients with medical refractory epilepsy, who had undergone VNS surgery at Beijing Tiantan Hospital from Aug. 13, 2014 to Dec. 31, 2014, were selected for test. Before VNS surgery, these 32 patients with medical refractory epilepsy were comprehensively evaluated (including demographic characteristics, clinical history, history of antiepileptic medication, 24-hours video-EEG, MRI, and 24-hour dynamic electrocardiogram etc.).

(35) According to the above ECG signal processing method, the MSE analysis was performed, based on 24-hour dynamic electrocardiographic data. The corresponding characteristic parameters Slope5, Area1-5, Area6-15, and Area6-20 were extracted based on each patient's MSE curve. At the end of 1-year follow-up, among 32 patients with drug refractory epilepsy who had undergone VNS treatment, 28 patients' seizure frequencies had been reduced to various degrees (seizures had been completely controlled in 6 patients), who were considered as the effective group, and the remaining 4 patients' seizure frequencies after VNS surgery hadn't changed compared with those before VNS surgery, who were considered as the ineffective group. The MSE curves of the effective group and the ineffective group differ greatly, which suggested that the MSE method could be adapted to screen patients with VNS indications. Furthermore, each patient's characteristic parameters Slope5, Area1-5, Area6-15, and Area6-20 before surgery could be adapted to predict efficacy. The results shown that, among the above four parameters, Area6-20 was the most accurate parameter: when its threshold was set to 15.85, only one patient's Area6-20 was 15.09 among the 28 effective patients, as shown in Table 1, and the rest patients' Area6-20 were greater than 15.85. The screening accuracy-rate exceeded 96%, which confirmed that the MSE method of the above HRV analysis could accurately and effectively screen patients with VNS indications.

(36) TABLE-US-00001 TABLE 1 Slope5 Area1-5 Area6-15 Area6-20 Effective patient1 0.141 5.651 14.074 21.871 group patient2 0.098 2.925 9.587 15.093 patient3 0.012 7.596 16.155 24.702 patient4 0.069 5.341 14.477 23.084 patient5 0.081 4.531 10.915 17.081 patient6 0.045 6.766 16.296 25.248 patient7 0.092 4.150 11.503 18.059 patient8 0.101 3.685 11.804 18.769 patient9 0.091 6.962 17.257 27.045 patient10 0.177 3.983 14.009 22.139 patient11 0.169 5.571 14.666 23.090 patient12 0.116 5.529 15.255 23.922 patient13 0.112 4.124 11.135 17.381 patient14 0.144 4.189 12.097 19.044 patient15 0.099 4.953 12.932 20.050 patient16 0.070 5.904 13.657 21.414 patient17 0.172 4.599 11.162 16.557 patient18 −0.024 5.839 13.736 21.497 patient19 0.098 6.115 15.443 23.849 patient20 −0.066 5.197 11.964 19.043 patient21 0.072 3.875 10.273 16.025 patient22 0.087 5.132 14.689 23.120 patient23 0.101 6.058 15.677 24.371 patient24 0.053 4.871 14.701 23.563 patient25 0.128 5.580 15.252 23.891 patient26 −0.041 5.299 13.201 20.832 patient27 0.107 4.417 13.854 21.959 patient28 0.030 3.322 10.847 17.468 Ineffective patient29 0.070 4.226 9.647 14.219 group patient30 0.093 5.740 12.949 20.118 patient31 0.033 2.257 7.488 12.139 patient32 0.068 4.043 10.298 15.680

(37) The above description is only preferred embodiments of the present invention. It should be noted that, those skilled in the art can make improvements and modifications, without departing from the principle of the present invention. These improvements and modifications should be regarded in the scope of the present invention. In addition, although specific terms are used in this description, these terms are merely for convenience of illustration and do not constitute any limitation to the present invention.