A method, device, terminal device and storage medium for processing electrocardiogram data for heart failure assessment

Abstract

The present invention discloses a method, apparatus, terminal device and storage medium for processing ECG data for heart failure assessment, which obtains ECG data of a patient; performs peak screening processing on an ECG in the ECG data to obtain a plurality of ECG waves; determines R waves, P waves, Q waves, S waves and T waves in the plurality of ECG waves; extracts ECG acquisition parameters and ECG waveform parameters corresponding to each ECG wave in the ECG data based on generation time; establishes a coordinate system for each ECG wave with the ECG waveform parameter type of the ECG wave as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate, determines the ECG waveform type of the ECG wave based on the parameter value range of historical ECG waveform types, and summarizes the ECG waveform type and the ECG acquisition parameters to generate ECG analysis data of the patient; performs heart failure assessment based on the obtained accurate ECG analysis data in combination with the user's body impedance, which is beneficial to improving the accuracy of heart failure assessment.

Description

Technical Field

[0001] The present invention relates to the field of power control technology, and in particular to a method, device, terminal equipment and storage medium for processing electrocardiogram data for heart failure assessment. Background Art

[0002] Electrocardiogram (ECG) data is essential for doctors to diagnose heart failure, a cardiovascular disease typically caused by impaired systolic and / or diastolic function, which results in impaired blood flow, leading to systemic blood congestion and insufficient arterial blood supply. The prognosis for heart failure patients depends on the cause, severity, and treatment options. While the prognosis for heart failure is generally serious, timely and effective treatment and lifestyle changes can improve patients' quality of life and survival rate.

[0003] In existing technologies, there is little analysis involved in ECG data, and the analysis of ECG waveforms is not accurate enough, which in turn affects the accuracy of heart failure assessment.

[0004] Therefore, there is an urgent need for a processing strategy for ECG data for heart failure assessment to solve the problem of low accuracy in heart failure assessment. Summary of the Invention

[0005] Embodiments of the present invention provide a method, apparatus, terminal device, and storage medium for processing electrocardiogram (ECG) data for heart failure assessment, to solve the problem of low accuracy in heart failure assessment.

[0006] To solve the above problem, an embodiment of the present invention provides a method for processing electrocardiogram data for heart failure assessment, comprising:

[0007] Obtain the patient's ECG data;

[0008] Performing peak screening on the electrocardiogram in the electrocardiogram data to obtain a plurality of electrocardiogram waves;

[0009] Determine the R wave among the plurality of electrocardiogram waves, and determine the P wave, Q wave, S wave, and T wave based on the position of the R wave;

[0010] Based on the generation time of R wave, P wave, Q wave, S wave and T wave, the ECG acquisition parameters and ECG waveform parameters corresponding to R wave, P wave, Q wave, S wave and T wave are extracted from the ECG data respectively;

[0011] For each ECG wave, a coordinate system is established with the ECG waveform parameter type of the ECG wave as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate. Based on the parameter value range of the historical ECG waveform type corresponding to the ECG wave, the historical type area is confirmed in the coordinate system, and the waveform type corresponding to the historical type area where the ECG waveform parameter of the ECG wave is located is used as the ECG waveform type of the ECG wave. The ECG waveform type and the ECG acquisition parameters are summarized to generate the ECG analysis data of the patient, and then heart failure assessment is performed based on the ECG analysis data.

[0012] As an improvement to the above solution, the method of determining the R wave among the plurality of electrocardiographic waves and determining the P wave, Q wave, S wave, and T wave based on the position of the R wave includes:

[0013] Based on the peak value of the electrocardiogram, screening electrocardiogram waves with peak values ​​greater than a peak threshold and a sawtooth waveform, and determining a number of initial waves;

[0014] Among all initial waves, the distance between each initial wave is calculated, and based on a preset distance difference threshold, the initial waves that meet the preset conditions are screened out to determine a number of R waves; wherein the preset conditions are specifically: the distance value to the left initial wave is not within the distance difference threshold range, and the distance value to the right initial wave is not within the distance difference threshold range;

[0015] The electrocardiographic wave connected to the left side of the R wave is determined as the Q wave, and the electrocardiographic wave connected to the right side of the R wave is determined as the S wave. In the area where the current R wave corresponds to the S wave and the next R wave corresponds to the Q wave, the P wave and the T wave are determined.

[0016] As an improvement to the above solution, determining the P wave and the T wave in the region where the current R wave corresponds to the S wave and the next R wave corresponds to the Q wave includes:

[0017] In the area where the current R wave corresponds to the S wave and the next R wave corresponds to the Q wave, the appearing ECG wave is determined as the wave to be determined;

[0018] Determine the distance of the wave to be determined;

[0019] If it is greater than or equal to the wave distance threshold, the wave to be determined is a T wave;

[0020] If it is less than the wave distance threshold, the wave to be determined is a P wave.

[0021] As an improvement to the above solution, the coordinate system is established with the ECG waveform parameter type as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate, including:

[0022] For the R wave, a plane coordinate system is established with amplitude, direction and symmetry as the horizontal coordinates and amplitude value, direction value and symmetry value as the vertical coordinates; wherein the direction values ​​include: 0, 1 and -1, upward is 1, downward is -1, and no waveform is 0;

[0023] For Q wave, a plane coordinate system is established with amplitude, time limit and overall curve smoothness as the horizontal coordinates and amplitude value, time limit value and overall curve smoothness value as the vertical coordinates;

[0024] For the R wave, a plane coordinate system is established with amplitude, direction and symmetry as the horizontal coordinates and amplitude value, direction value and symmetry value as the vertical coordinates;

[0025] For S waves, a plane coordinate system is established with amplitude, direction and width as the horizontal coordinates and amplitude value, direction value and width value as the vertical coordinates;

[0026] For P waves, a plane coordinate system is established with amplitude, direction and overall curve smoothness as the horizontal coordinates and amplitude value, direction value and overall curve smoothness value as the vertical coordinates;

[0027] For the T wave, a plane coordinate system is established with amplitude, direction and overall curve smoothness as the horizontal coordinates and amplitude value, direction value and overall curve smoothness value as the vertical coordinates.

[0028] As an improvement to the above solution, after generating the electrocardiogram analysis data of the patient, the method further includes:

[0029] collecting the patient's body impedance value and medical history data;

[0030] generating body composition data based on the body impedance value;

[0031] The electrocardiogram analysis data, the body composition data and the medical record data are input into a heart failure assessment model to generate the patient's heart failure grade; wherein, the gradient enhancement model is trained with original training samples marked with heart failure grade labels to obtain a heart failure assessment model.

[0032] As an improvement to the above solution, generating human body composition data based on the human body impedance value includes:

[0033] Obtaining calculation parameters of extracellular fluid volume, intracellular fluid volume and body fat content;

[0034] Based on a preset impedance calculation formula, human body impedance value, extracellular fluid volume calculation parameters, intracellular fluid volume calculation parameters, and body fat content calculation parameters, human body composition data is calculated; wherein the human body composition data includes: extracellular fluid volume, intracellular fluid volume, body water content, and body fat content; the impedance calculation formula is specifically:

[0035] WY=a*Zb

[0036] NY=c*Zd

[0037] WR=WY+NY

[0038] FT=(e / Zf)*g

[0039] Where Z is the human body impedance value, a and b are the calculation parameters of extracellular fluid volume, c and d are the calculation parameters of intracellular fluid volume, e, f, and g are the calculation parameters of body fat content, WY is the extracellular fluid volume, NY is the intracellular fluid volume, WR is the body water content, and FT is the body fat content.

[0040] As an improvement to the above solution, the acquisition of medical record data includes:

[0041] obtaining the patient's medical history;

[0042] Determine the type of the historical medical record;

[0043] If it is a picture, the patient's name, doctor's name and consultation time are extracted through image text recognition technology, several corresponding work logs are retrieved according to the doctor's name, and a target log is retrieved from the several work logs according to the consultation time, and the patient's first historical disease is confirmed in the target log according to the patient's name;

[0044] If it is electronic text, the second historical disease of the patient is directly extracted from the electronic text data.

[0045] Accordingly, an embodiment of the present invention further provides an electrocardiogram data processing device for heart failure assessment, comprising: a data acquisition module, a waveform screening module, a waveform positioning module, a data extraction module, and a result generation module;

[0046] The data acquisition module is used to acquire the patient's electrocardiogram data;

[0047] The waveform screening module is used to perform peak screening on the electrocardiogram in the electrocardiogram data to obtain a plurality of electrocardiogram waves;

[0048] The waveform positioning module is used to determine the R wave among the multiple electrocardiogram waves, and determine the P wave, Q wave, S wave and T wave based on the position of the R wave;

[0049] The data extraction module is used to extract the ECG acquisition parameters and ECG waveform parameters corresponding to the R wave, P wave, Q wave, S wave and T wave from the ECG data based on the generation time of the R wave, P wave, Q wave, S wave and T wave;

[0050] The result generation module is used to establish a coordinate system for each ECG wave, using the ECG waveform parameter type of the ECG wave as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate, confirm the historical type area in the coordinate system based on the parameter value range of the historical ECG waveform type corresponding to the ECG wave, use the waveform type corresponding to the historical type area where the ECG waveform parameter of the ECG wave is located as the ECG waveform type of the ECG wave, and summarize the ECG waveform type and ECG acquisition parameters to generate the patient's ECG analysis data, and then perform heart failure assessment based on the ECG analysis data.

[0051] Correspondingly, an embodiment of the present invention also provides a computer terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements a method for processing electrocardiogram data for heart failure assessment as described in the present invention.

[0052] Correspondingly, an embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute a method for processing electrocardiogram data for heart failure assessment as described in the present invention.

[0053] As can be seen from the above, the present invention has the following beneficial effects:

[0054] The present invention provides a method for processing electrocardiogram data for heart failure assessment, which comprises the following steps: obtaining electrocardiogram data of a patient; performing peak screening processing on an electrocardiogram in the electrocardiogram data to obtain a plurality of electrocardiogram waves; determining an R wave in the plurality of electrocardiogram waves, and determining a P wave, a Q wave, an S wave and a T wave based on the position of the R wave; extracting electrocardiogram acquisition parameters and electrocardiogram waveform parameters corresponding to the R wave, the P wave, the Q wave, the S wave and the T wave from the electrocardiogram data based on the generation time of the R wave, the P wave, the Q wave, the S wave and the T wave; for each electrocardiogram wave, establishing a coordinate system with the type of electrocardiogram waveform parameter of the electrocardiogram wave as the horizontal coordinate and the value of the electrocardiogram waveform parameter as the vertical coordinate, confirming a historical type area in the coordinate system based on the parameter value range of the historical electrocardiogram waveform type corresponding to the electrocardiogram wave, taking the waveform type corresponding to the historical type area where the electrocardiogram waveform parameter of the electrocardiogram wave is located as the electrocardiogram waveform type of the electrocardiogram wave, and summarizing the electrocardiogram waveform type and the electrocardiogram acquisition parameters to generate electrocardiogram analysis data of the patient. The present invention classifies the electrocardiogram in the electrocardiogram data into R waves, P waves, Q waves, S waves and T waves, and based on the classification results, establishes a coordinate system and a historical category area determined by the parameter value range of the historical electrocardiogram waveform type, and identifies the electrocardiogram waveform type in the form of coordinates, thereby greatly improving the accuracy of electrocardiogram waveform type recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 1 is a flow chart of a method for processing electrocardiogram data for heart failure assessment provided by one embodiment of the present invention;

[0056] Figure 2 is a structural diagram of a system for processing electrocardiogram data for heart failure assessment provided by an embodiment of the present invention;

[0057] Figure 3 This is a schematic diagram of the structure of a terminal device provided by one embodiment of the present invention;

[0058] Figure 4 This is a Self-Training data labeling process provided by an embodiment of the present invention;

[0059] Figure 5 This is the Self-Training data labeling principle process provided by one embodiment of the present invention;

[0060] Figure 6 This is a schematic diagram of the result of the first regression tree provided by one embodiment of the present invention;

[0061] Figure 7 is a schematic diagram of the result of the second regression tree provided by one embodiment of the present invention;

[0062] Figure 8 1 is a schematic diagram of the results of the third regression tree provided by one embodiment of the present invention;

[0063] Figure 9 1 is a schematic diagram of the result of the fourth regression tree provided by one embodiment of the present invention;

[0064] Figure 10 1 is a schematic diagram of the result of the fifth regression tree provided by one embodiment of the present invention;

[0065] Figure 11 4 is a result diagram of the final regression tree provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0067] Example 1

[0068] See also Figure 1 , Figure 1 FIG. 1 is a flow chart of a method for processing electrocardiogram data for heart failure assessment provided by an embodiment of the present invention. Figure 1 As shown, this embodiment includes steps 101 to 105, and each step is specifically as follows:

[0069] Step 101: Obtain the patient's electrocardiogram data.

[0070] In this embodiment, the electrocardiogram data of the patient before taking the medicine is collected as the first electrocardiogram collected data.

[0071] In a specific embodiment, the first ECG data is input into a baseline filter and a low-pass filter for filtering to obtain filtered data;

[0072] Performing peak transformation on the filtered data to obtain peak transformation data;

[0073] Determining, based on the peak conversion data, electrocardiographic characteristic parameters, heart rate time domain analysis parameters, heart rate frequency domain analysis parameters, and heart rate nonlinear analysis parameters corresponding to the first electrocardiographic acquisition data;

[0074] The first electrocardiogram analysis data is generated according to the electrocardiogram characteristic parameters, the heart rate time domain analysis parameters, the heart rate frequency domain analysis parameters and the heart rate nonlinear analysis parameters.

[0075] In a specific embodiment, the first ECG data collected is data of more than 5 minutes, and the data is first preprocessed by a baseline removal filter and a low-pass filter: the low-pass filter is mainly used to filter out interference signals such as power frequency interference and machine interference. The filter adopts FIR ripple design, in which the passband cutoff frequency is 35Hz, the stopband cutoff frequency is 45Hz, the passband attenuation coefficient is 0.2dB, the stopband attenuation coefficient is 55dB, and the filter order is 59.

[0076] In one possible implementation, the calculation formula of the baseline filter is:

[0077]

[0078] Where x(i) is the first ECG data, f(i) is the filtered data, and n is the filtering window.

[0079] Furthermore, the specific formula for the peak conversion is:

[0080]

[0081] Where x′(i), x′(ij), and x′(i+j) are filtered data, y(i) is the peak transformation data, and w is the peak transformation window.

[0082] In a specific embodiment, peak transformation is performed on the filtered data. The peak transformation mainly emphasizes the R wave of the electrocardiogram waveform and suppresses other waveforms.

[0083] In a specific embodiment, the target person is given an electrocardiogram (ECG) acquisition device to collect impedance data and personal information, wherein the sampling rate of the first ECG data is 250 Hz.

[0084] Step 102: performing peak screening processing on the electrocardiogram in the electrocardiogram data to obtain a plurality of electrocardiogram waves.

[0085] In this embodiment,

[0086] Step 103: Determine the R wave among the plurality of electrocardiogram waves, and determine the P wave, Q wave, S wave, and T wave based on the position of the R wave.

[0087] In this embodiment, determining the R wave among the plurality of electrocardiogram waves and determining the P wave, Q wave, S wave, and T wave based on the position of the R wave includes:

[0088] Based on the peak value of the electrocardiogram, screening electrocardiogram waves with peak values ​​greater than a peak threshold and a sawtooth waveform, and determining a number of initial waves;

[0089] Among all initial waves, the distance between each initial wave is calculated, and based on a preset distance difference threshold, the initial waves that meet the preset conditions are screened out to determine a number of R waves; wherein the preset conditions are specifically: the distance value to the left initial wave is not within the distance difference threshold range, and the distance value to the right initial wave is not within the distance difference threshold range;

[0090] The electrocardiographic wave connected to the left side of the R wave is determined as the Q wave, and the electrocardiographic wave connected to the right side of the R wave is determined as the S wave. In the area where the current R wave corresponds to the S wave and the next R wave corresponds to the Q wave, the P wave and the T wave are determined.

[0091] In this embodiment, determining the P wave and the T wave in the region where the current R wave corresponds to the S wave and the next R wave corresponds to the Q wave includes:

[0092] In the area where the current R wave corresponds to the S wave and the next R wave corresponds to the Q wave, the appearing ECG wave is determined as the wave to be determined;

[0093] Determine the distance of the wave to be determined;

[0094] If it is greater than or equal to the wave distance threshold, the wave to be determined is a T wave;

[0095] If it is less than the wave distance threshold, the wave to be determined is a P wave.

[0096] Step 104: Based on the generation time of the R wave, P wave, Q wave, S wave and T wave, ECG acquisition parameters and ECG waveform parameters corresponding to the R wave, P wave, Q wave, S wave and T wave are extracted from the ECG data.

[0097] In this embodiment, the ECG acquisition parameters include: ECG characteristic parameters, heart rate time domain analysis parameters, heart rate frequency domain analysis parameters and heart rate nonlinear analysis parameters;

[0098] ECG characteristic parameters, including: QT_list: QT interval parameter array, which refers to the time from the start of cardiac depolarization to the end of repolarization, that is, the time from the start of the QRS wave to the end of the T wave; RR_list: RR interval parameter array, which refers to the time between two R waves; HR_list: heart rate trend chart.

[0099] The calculation of heart rate variability parameters is mainly divided into time domain analysis, frequency domain analysis, and nonlinear analysis, so as to obtain the corresponding heart rate time domain analysis parameters, heart rate frequency domain analysis parameters, and heart rate nonlinear analysis parameters:

[0100] i) Heart rate time domain analysis parameters:

[0101] SDNN: standard deviation of normal RR interval;

[0102] RMSSD: root mean square difference between adjacent normal RR intervals;

[0103] NN50: the number of heartbeats with a difference of more than 50 milliseconds between adjacent normal RR intervals in the full recording;

[0104] PNN50: NN50 divided by the percentage of the total number of normal RR intervals.

[0105] ii) Heart rate frequency domain analysis parameters:

[0106] VLF: power in the frequency range of 0.003Hz to 0.04Hz;

[0107] LF: power in the frequency range of 0.04Hz to 0.15Hz;

[0108] HF: power in the frequency range of 0.15 Hz to 0.4 Hz;

[0109] LF_HF: ratio of LF to HF;

[0110] W: Total power, the power of variation of all normal RR intervals with a frequency less than 0.4 Hz.

[0111] iii) Heart rate nonlinear analysis parameters:

[0112] Lorenz_plot: Poincare scatter plot, also known as Lorenz scatter plot, is a scatter plot formed by plotting all normal cardiac cycles in sequence with the first of two adjacent cardiac cycles as the horizontal coordinate and the second cardiac cycle as the vertical coordinate as a point;

[0113] SD1: short axis of the scatter plot, representing the magnitude of the difference between adjacent NN intervals;

[0114] SD2: The long axis of the scatter plot, representing the magnitude of changes in heart rate over a long period of time;

[0115] MP_plot: A scatter plot of the difference between two consecutive cardiac cycles. The first difference is plotted as the horizontal axis, and the second difference as the vertical axis. This plot is repeated for all cardiac cycles. The points are distributed in four quadrants of the rectangular ordinate, depending on the sign of the difference, reflecting the dispersion of the RR difference.

[0116] Step 105: For each ECG wave, a coordinate system is established with the ECG waveform parameter type of the ECG wave as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate. Based on the parameter value range of the historical ECG waveform type corresponding to the ECG wave, the historical type area is confirmed in the coordinate system. The waveform type corresponding to the historical type area where the ECG waveform parameter of the ECG wave is located is used as the ECG waveform type of the ECG wave. The ECG waveform type and the ECG acquisition parameters are summarized to generate the ECG analysis data of the patient, and then heart failure assessment is performed based on the ECG analysis data.

[0117] In this embodiment, the coordinate system is established by taking the ECG waveform parameter type as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate, including:

[0118] For the R wave, a plane coordinate system is established with amplitude, direction and symmetry as the horizontal coordinates and amplitude value, direction value and symmetry value as the vertical coordinates; wherein the direction values ​​include: 0, 1 and -1, upward is 1, downward is -1, and no waveform is 0;

[0119] For Q wave, a plane coordinate system is established with amplitude, time limit and overall curve smoothness as the horizontal coordinates and amplitude value, time limit value and overall curve smoothness value as the vertical coordinates;

[0120] For the R wave, a plane coordinate system is established with amplitude, direction and symmetry as the horizontal coordinates and amplitude value, direction value and symmetry value as the vertical coordinates;

[0121] For S waves, a plane coordinate system is established with amplitude, direction and width as the horizontal coordinates and amplitude value, direction value and width value as the vertical coordinates;

[0122] For P waves, a plane coordinate system is established with amplitude, direction and overall curve smoothness as the horizontal coordinates and amplitude value, direction value and overall curve smoothness value as the vertical coordinates;

[0123] For the T wave, a plane coordinate system is established with amplitude, direction and overall curve smoothness as the horizontal coordinates and amplitude value, direction value and overall curve smoothness value as the vertical coordinates.

[0124] In one specific embodiment, Q waves are classified as normal Q waves and abnormal Q waves. A normal Q wave has an amplitude less than 1 / 4 of the R wave amplitude, a duration less than 0.03, and an overall curve smoothness less than 0.1. An abnormal Q wave has an amplitude greater than or equal to 1 / 4 of the R wave amplitude, a duration greater than or equal to 0.03, and an overall curve smoothness greater than or equal to 0.1. Historical category regions include normal Q wave regions (amplitude less than 1 / 4 of the R wave amplitude and greater than 0, duration less than 0.03 and greater than 0, overall curve smoothness less than 0.1 and greater than 0) and abnormal Q wave regions (amplitude greater than or equal to 1 / 4 of the R wave amplitude, duration greater than or equal to 0.03, and overall curve smoothness greater than or equal to 0.1).

[0125] In one specific embodiment, R waves are divided into normal R waves and abnormal R waves. A normal R wave has an amplitude value less than 2.5 and greater than 0.5, a direction value of 1, and a symmetry value greater than or equal to 0.85. An abnormal R wave has an amplitude value greater than 2.5 and less than 0.5, a direction value of 0 or -1, and a symmetry value less than 0.85. Historical category regions include normal R wave regions (amplitude values ​​less than 2.5 and greater than 0.5, direction value of 1, and symmetry value greater than or equal to 0.85) and abnormal R wave regions (amplitude values ​​greater than 2.5 and less than 0.5, direction values ​​of 0 or -1, and symmetry value less than 0.85).

[0126] In one specific embodiment, S waves are divided into normal S waves and abnormal S waves. A normal S wave has an amplitude less than 0.3 and greater than 0.1, a direction value of -1, and a width greater than 0.04 and less than 0.12. An abnormal S wave has an amplitude greater than 0.3 and less than 0.1, a direction value of 0 or 1, and a width less than 0.04 and greater than 0.12. Historical category regions include a normal S wave region (amplitude less than 0.3 and greater than 0.1, direction value of -1, width greater than 0.04 and less than 0.12) and an abnormal S wave region (amplitude greater than 0.3 and less than 0.1, direction values ​​of 0 or 1, and width less than 0.04 and greater than 0.12).

[0127] In a specific embodiment, P waves are divided into normal P waves, missing P waves, inverted P waves, high-peaked P waves and abnormal P waves; the amplitude value of normal P waves is less than or equal to 0.25, the direction value is 1, and the overall curve smoothness is less than 0.1; the direction value of missing P waves is 0, the amplitude value and the overall curve smoothness can be any values; the direction value of inverted P waves is -1, the amplitude value and the overall curve smoothness can be any values; the amplitude value of high-peaked P waves is greater than 0.25, the direction value is 1, and the overall curve smoothness can be any value; the overall curve smoothness of abnormal P waves is greater than or equal to 0.1, the direction value is 1, and the amplitude value can be any value. Historical type areas include: normal P wave area (amplitude value less than or equal to 0.25, direction value 1, overall curve smoothness less than 0.1), missing P wave area (direction value 0, amplitude value and overall curve smoothness can be any value), inverted P wave area (direction value -1, amplitude value and overall curve smoothness can be any value), high-peaked P wave area (amplitude value greater than 0.25, direction value 1, overall curve smoothness can be any value) and abnormal P wave area (overall curve smoothness greater than or equal to 0.1, direction value 1, amplitude value can be any value).

[0128] In a specific embodiment, T waves are divided into normal T waves, dome-shaped T waves, flat T waves, inverted T waves and tall T waves; the amplitude of normal T waves is greater than 0.1 and less than 0.3, the direction is 1, and the overall curve smoothness is less than 0.1; the amplitude of dome-shaped T waves is greater than 0.1 and less than 0.3, the direction is 1, and the overall curve smoothness is greater than or equal to 0.1; the amplitude of flat T waves is less than 0.1, the direction is 1, and the overall curve smoothness is an arbitrary value; the direction of inverted T waves is -1, and the overall curve smoothness and amplitude value are arbitrary values; the amplitude of tall T waves is greater than 1.5, the direction is 1, and the overall curve is Line smoothness is arbitrary; historical type regions include: normal T wave region (amplitude greater than 0.1 and less than 0.3, orientation of 1, and overall curve smoothness less than 0.1), dome-angle T wave region (amplitude greater than 0.1 and less than 0.3, orientation of 1, and overall curve smoothness greater than or equal to 0.1), flat T wave region (amplitude less than 0.1, orientation of 1, and overall curve smoothness arbitrary), inverted T wave region (orientation of -1, overall curve smoothness and amplitude arbitrary), and towering T wave region (amplitude greater than 1.5, orientation of 1, and overall curve smoothness arbitrary).

[0129] In this embodiment, after generating the electrocardiogram analysis data of the patient, the method further includes:

[0130] collecting the patient's body impedance value and medical history data;

[0131] generating body composition data based on the body impedance value;

[0132] The electrocardiogram analysis data, the body composition data and the medical record data are input into a heart failure assessment model to generate the patient's heart failure grade; wherein, the gradient enhancement model is trained with original training samples marked with heart failure grade labels to obtain a heart failure assessment model.

[0133] In a specific embodiment, the training of the heart failure assessment model includes:

[0134] Acquire sample data of several patients diagnosed with heart failure; wherein the sample data includes: electrocardiogram (ECG) sample data, human body impedance sample values, and medical record sample data;

[0135] Sort each data point in the sample data in chronological order, and perform a data supplement operation on the sample data according to a preset time granularity by interpolation to obtain complete sample data;

[0136] The complete sample data is labeled with the heart failure grade to obtain original training samples; wherein the original training samples include: electrocardiogram acquisition training samples, human body impedance value training samples and medical record data training samples;

[0137] The original training samples are input into the gradient enhancement model for training through the gradient enhancement algorithm to obtain a heart failure assessment model.

[0138] Furthermore, the data supplementation operation is specifically as follows:

[0139] Performing a missing identification operation on the time corresponding to each data point in the sample data according to a preset time granularity;

[0140] If the time of a data point is missing, the data is supplemented by interpolation. After the data supplementation of the missing data point is completed, the data supplementation operation is performed on the next data point until the operation of all data points is completed;

[0141] If the time of the data point is not missing, the data supplement operation is directly performed on the next data point until the operation of all data points is completed.

[0142] In a specific embodiment, in each sample data, the data points are sorted by time. Interpolation is used to complete the missing time data according to the time granularity. If the time data of the data points in the training set are complete, there is no need for interpolation to complete them and they can be used directly for training.

[0143] The following example is provided for illustration: Assume that the time granularity is day. There are currently 3 data in the data set, namely (20220103, 30), (20220105, 50), and (20220107, 70). The former represents the date and the latter represents the data value. Using the interpolation method, we can expand to obtain two points (20220104, 40) and (20220105, 50) according to the time granularity of day. A total of 5 points constitute a sample data set.

[0144] In one possible implementation, the original training samples are input into a gradient boosting model for training using a gradient boosting algorithm to obtain a heart failure assessment model, including:

[0145] Obtaining an average value according to the heart failure grades annotated by all the original training samples, and determining a base learner of a gradient boosting model based on the average value and a preset loss function;

[0146] The iterative training operation is repeatedly performed to obtain a heart failure assessment model.

[0147] Furthermore, the iterative training operation is specifically as follows:

[0148] Inputting each of the original training samples into the current base learner, and calculating the negative gradient data of each of the original training samples according to the prediction result of each of the original training samples output by the current base learner and the labeling result of each of the original training samples;

[0149] Constructing a regression tree model based on the negative gradient data; wherein the regression tree model is used to fit the residual corresponding to the negative gradient data;

[0150] Add the regression tree model multiplied by a preset step size to the current base learner to obtain the next base learner and record the current number of iterations;

[0151] The current number of iterations is judged: if it is less than the preset number of iterations, the next iterative training operation is performed through the next base learner; if it is equal to the preset number of iterations, the current base learner is output as the heart failure assessment model.

[0152] In one possible implementation, the heart failure assessment model satisfies the following conditions:

[0153]

[0154] Where, is the output of the heart failure assessment model in the last training round, f0(x) is the output of the base learner, I is the preset step size, L is the loss function, and y i is the true label, γ is a constant, is a linear model, f m-1 (x i ) is the output of the heart failure assessment model in the previous training round, r im is the negative gradient value.

[0155] In this embodiment, the gradient boosting algorithm (GBDT) is used for the heart failure assessment model. GBDT is an integrated learning method based on gradient boosting. It uses multiple regression trees as base learners to improve the predictive ability of the model by iteratively fitting the negative gradient of the loss function (also called residual or pseudo residual).

[0156] To better illustrate, the basic process of the GBDT regression algorithm is as follows:

[0157] Initialize a constant as the first base learner, usually the mean of the training data labels or the value that minimizes the loss function.

[0158]

[0159] Among them, L is the loss function, y i is the true label, and γ is a constant.

[0160] The loss function L is the multiple linear models learned Converted to the probability of the corresponding classification, each Corresponding to a classification category, the specific calculation formula is:

[0161]

[0162] For each round of iteration, the negative gradient (residual) of each sample is calculated as the new response variable. The number of iterations is m = 1, 2, 3, ..., M. The specific process is as follows:

[0163] i) For sample i=1,2,3,...,N, calculate the negative gradient:

[0164]

[0165] ii) Use the regression tree to fit the new response variable and obtain a regression tree whose leaf nodes are divided into several regions γ jm ,j=1,2,3,...,J m

[0166] iii) For j = 1, 2, 3, ..., J m Calculate the best fit value between the negative gradient, usually the residual mean of all samples in the region or the value that minimizes the loss function:

[0167]

[0168] iv) Update the current model by multiplying the regression tree obtained in this round by a step size (learning rate) and adding it to the previous model:

[0169]

[0170] v) Repeat steps i-iv until the preset number of iterations is reached or the model performance no longer improves.

[0171] Output the final model expression:

[0172]

[0173] The model is trained using the final model expression obtained.

[0174] Furthermore, the step of inputting all the data collected by the data acquisition device and the original training samples as updated training samples into the gradient boosting model for training to update the heart failure assessment model includes:

[0175] Inputting all data collected by the data collection device into the current heart failure assessment model to obtain intermediate training samples with pseudo labels;

[0176] The intermediate training samples and the original training samples are respectively input into the gradient enhancement model for training, thereby updating the heart failure assessment model.

[0177] In a specific embodiment, in order to achieve a more optimized heart failure assessment model, the existing model needs to be continuously updated and iterated. This method adopts a semi-supervised learning method Self-Training.

[0178] To better explain, Self-Training is to first train a weak classifier with a small amount of labeled data, and then use this weak classifier to label unlabeled samples. When certain conditions are met (the predicted probability is greater than the threshold), the predicted result is used as the true label of the sample. Then, the model is trained again with the current labeled data and the unlabeled samples are labeled and iterated repeatedly until the stopping condition is reached.

[0179] like Figure 4As shown in the figure, the first step is to train a weak classifier with a small amount of labeled data; the second step is to use this weak classifier to predict unlabeled data samples, and according to a certain strategy, the predicted part of the label is used as the true label of the sample; the third step is to use all the current labeled samples (including the samples processed in the second step) to train the classifier. If the stopping condition is met, it will enter the fourth step, otherwise it will continue to enter the second step to predict the unlabeled samples; the fourth step is to use the classifier trained in the third step to evaluate on the test set.

[0180] The specific algorithm principle is as follows Figure 5 , including steps 601 to 604, the process is as follows:

[0181] Step 601: First, use some labeled samples to train a teacher model θ t , minimize the standard

[0182] Cross entropy loss of signed data;

[0183]

[0184] Where, f noised () is the expression of the base learner;

[0185] Step 602: Generate pseudo labels for unlabeled samples using the teacher model;

[0186]

[0187] Where f() is the expression of the base learner, is a pseudo label;

[0188] Step 603: Use the above labeled samples and pseudo-labeled samples to train the student model θ s ;

[0189]

[0190] During the training process, data enhancement and random depth operations are required to better train the model;

[0191] Step 604: Use the student model as the new teacher model and repeat steps 602, 603, and 604. Repeat 10 times or stop when convergence is reached.

[0192] In a possible implementation, various body characteristic data other than electrocardiogram data are acquired; and the body condition characteristics of the patient are analyzed based on the body characteristic data;

[0193] The side effect characteristics of the recommended drug are retrieved, and the data overlap between the side effect characteristics and the physical condition characteristics is calculated; when the data overlap is greater than the overlap threshold, a reminder message is issued to replace the recommended drug; when the data overlap is less than the overlap threshold, the recommended drug is maintained in use.

[0194] Furthermore, the side effect characteristics of the recommended drug are retrieved and the matching degree between the side effect characteristics and the physical condition characteristics is calculated, specifically:

[0195] Retrieving the side effect characteristics of the recommended drug, dividing them by organ type, and determining the side effect sub-characteristic range data for each organ type;

[0196] According to the body state characteristics, the body state characteristics are divided according to the organ type, and the state sub-characteristic data of each organ type is determined;

[0197] In each organ type, the data overlap corresponding to each organ type is calculated based on the side effect sub-feature range data and the status sub-feature data.

[0198] In a possible implementation, the data overlap corresponding to each organ type is calculated based on the side effect sub-feature range data and the state sub-feature data, specifically:

[0199] Construct a coordinate plane with data values ​​as ordinates and time values ​​as abscissas;

[0200] The side effect sub-feature range data and the state sub-feature data corresponding to each moment are mapped onto the coordinate plane, and the overlapping time intervals in which the state sub-feature data and the side effect sub-feature range data overlap are recorded. The data overlap degree corresponding to each organ type is calculated based on the overlapping time intervals and the total time interval.

[0201] In a specific embodiment, in addition to monitoring the patient's heart failure level before and after medication, the status characteristics of the patient's other organs are also required to avoid recommending drugs that cause serious side effects on other organs while treating the heart.

[0202] For better explanation, the status characteristics of each organ of the population who have side effects from using the recommended drugs are extracted to obtain the side effect sub-feature range data. By real-time monitoring the status sub-feature data of each organ of the patient, when it is found that the overlap between the status sub-feature data and the side effect sub-feature range data is very high, it indicates that the patient's organ has a problem and has caused side effects. This embodiment avoids the problem of side effects in patients after taking the recommended drugs by paying attention to the impact of drug side effects on body organs and monitoring the organ status of patients after taking the drugs in real time.

[0203] In this embodiment, generating human body composition data based on the human body impedance value includes:

[0204] Obtaining calculation parameters of extracellular fluid volume, intracellular fluid volume and body fat content;

[0205] Based on a preset impedance calculation formula, human body impedance value, extracellular fluid volume calculation parameters, intracellular fluid volume calculation parameters, and body fat content calculation parameters, human body composition data is calculated; wherein the human body composition data includes: extracellular fluid volume, intracellular fluid volume, body water content, and body fat content; the impedance calculation formula is specifically:

[0206] WY=a*Zb

[0207] NY=c*Zd

[0208] WR=WY+NY

[0209] FT=(e / Zf)*g

[0210] Where Z is the human body impedance value, a and b are the calculation parameters of extracellular fluid volume, c and d are the calculation parameters of intracellular fluid volume, e, f, and g are the calculation parameters of body fat content, WY is the extracellular fluid volume, NY is the intracellular fluid volume, WR is the body water content, and FT is the body fat content.

[0211] In a specific embodiment, a is 0.356, b is 0.118, c is 0.246, d is 0.138, e is 495, f is 450, and g is 0.36.

[0212] In this embodiment, the acquisition of medical record data includes:

[0213] obtaining the patient's medical history;

[0214] Determine the type of the historical medical record;

[0215] If it is a picture, the patient's name, doctor's name and consultation time are extracted through image text recognition technology, several corresponding work logs are retrieved according to the doctor's name, and a target log is retrieved from the several work logs according to the consultation time, and the patient's first historical disease is confirmed in the target log according to the patient's name;

[0216] If it is electronic text, the second historical disease of the patient is directly extracted from the electronic text data.

[0217] Preferably, since patients with heart failure may have been ill for a long time, their medical records may exist in paper medical records and electronic medical records: for electronic medical records, electronic text data can be directly extracted to filter out historical diseases; for paper medical record images, since the doctor's handwriting in paper medical record images is difficult to recognize, and the handwriting naming methods of different doctors are different, the doctor's name, patient name and consultation time can be identified, and the doctor's work log can be retrieved, so that the patient's historical diseases can be obtained through the doctor's work log, thereby avoiding the problem of the patient's historical diseases being unable to be known due to the difficulty in recognizing handwriting.

[0218] Preferably, the present invention can also collect different types of doctor's handwritings, and annotate the doctor's handwritings with the true meaning of the fonts, and use the annotated doctor's handwritings as training samples, thereby inputting the training samples into the CNN neural network for training to obtain a paper medical record handwriting recognition model; by inputting the picture of the paper medical record into the paper medical record handwriting recognition model, the doctor's electronic medical record text is output, and the patient's historical disease is extracted from the electronic medical record text.

[0219] In a specific embodiment, a heart failure drug database is matched based on the first heart failure grade and the patient's medical records to obtain recommended drugs; wherein the heart failure drug database includes: a plurality of drug data marked with heart failure grades and prohibited population information;

[0220] collecting second electrocardiogram data and a second human body impedance value of the patient after the patient uses the recommended drug;

[0221] Preprocessing the second electrocardiogram data and the second human body impedance value to obtain second electrocardiogram analysis data and second human body composition data;

[0222] inputting the second electrocardiogram analysis data, the second body composition data, and the medical history data into a heart failure assessment model, so that the heart failure assessment model determines a second heart failure grade of the patient;

[0223] The first heart failure grade and the second heart failure grade are used to determine whether a medication needs to be changed for treatment.

[0224] It should be noted that, based on the first heart failure grade, therapeutic drugs corresponding to the heart failure grade are screened out from the heart failure drug database;

[0225] Based on the medical record data, recommended drugs that do not contain allergy drug information and interaction drug information are screened out from the therapeutic drugs.

[0226] It is understandable that the drug recommendation module first matches the therapeutic drugs corresponding to the heart failure level in the heart failure drug database based on the heart failure level, and based on the medical record data, screens out recommended drugs that do not contain allergy drug information and interaction drug information among the therapeutic drugs; thereby being able to recommend to the patient the most suitable drug for the current treatment of heart failure and avoid major side effects due to the patient's allergy history and previous medications.

[0227] In a specific embodiment, several drugs for treating heart failure are obtained as target drugs, each target drug is labeled, the level of heart failure that can be treated, the components of the target drug, and the drugs that interact with the target drug are annotated into the target drug, and a heart failure drug database is constructed based on the level of heart failure, the components of the target drug, the drugs that interact with the target drug, and the target drug.

[0228] Preferably, extracting the patient's historical disease and allergen information based on the medical record data;

[0229] Extracting historical medications corresponding to the historical diseases;

[0230] Determine allergy medication information based on the allergen information; determine interaction medication information based on the historical medication and the interaction relationship between the medications;

[0231] According to the allergy drug information and the interaction drug information, recommended drugs that do not contain the allergy drug information and the interaction drug information are screened out from the therapeutic drugs.

[0232] See also Figure 2 , Figure 2 2 is a schematic diagram of a structure of an electrocardiogram data processing device for heart failure assessment provided by one embodiment of the present invention, comprising: a data acquisition module 201, a waveform screening module 202, a waveform positioning module 203, a data extraction module 204, and a result generation module 205;

[0233] The data acquisition module is used to acquire the patient's electrocardiogram data;

[0234] The waveform screening module is used to perform peak screening on the electrocardiogram in the electrocardiogram data to obtain a plurality of electrocardiogram waves;

[0235] The waveform positioning module is used to determine the R wave among the multiple electrocardiogram waves, and determine the P wave, Q wave, S wave and T wave based on the position of the R wave;

[0236] The data extraction module is used to extract the ECG acquisition parameters and ECG waveform parameters corresponding to the R wave, P wave, Q wave, S wave and T wave from the ECG data based on the generation time of the R wave, P wave, Q wave, S wave and T wave;

[0237] The result generation module is used to establish a coordinate system for each ECG wave, using the ECG waveform parameter type of the ECG wave as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate, confirm the historical type area in the coordinate system based on the parameter value range of the historical ECG waveform type corresponding to the ECG wave, use the waveform type corresponding to the historical type area where the ECG waveform parameter of the ECG wave is located as the ECG waveform type of the ECG wave, and summarize the ECG waveform type and ECG acquisition parameters to generate the patient's ECG analysis data, and then perform heart failure assessment based on the ECG analysis data.

[0238] It can be understood that the above-mentioned system embodiment corresponds to the method embodiment of the present invention, which can implement the electrocardiogram data processing method for heart failure assessment provided by any of the above-mentioned method embodiments of the present invention.

[0239] This embodiment obtains the patient's ECG data; performs peak screening on the ECG in the ECG data to obtain several ECG waves; determines the R wave in the several ECG waves, and determines the P wave, Q wave, S wave and T wave based on the position of the R wave; extracts the ECG acquisition parameters and ECG waveform parameters corresponding to the R wave, P wave, Q wave, S wave and T wave from the ECG data based on the generation time of the R wave, P wave, Q wave, S wave and T wave; for each ECG wave, establishes a coordinate system with the ECG waveform parameter type of the ECG wave as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate, confirms the historical type area in the coordinate system based on the parameter value range of the historical ECG waveform type corresponding to the ECG wave, takes the waveform type corresponding to the historical type area where the ECG waveform parameter of the ECG wave is located as the ECG waveform type of the ECG wave, and summarizes the ECG waveform type and the ECG acquisition parameters to generate the ECG analysis data of the patient. The present invention classifies the electrocardiogram in the electrocardiogram data into R waves, P waves, Q waves, S waves and T waves, and based on the classification results, establishes a coordinate system and a historical category area determined by the parameter value range of the historical electrocardiogram waveform type, and identifies the electrocardiogram waveform type in the form of coordinates, thereby greatly improving the accuracy of electrocardiogram waveform type recognition.

[0240] Example 2

[0241] Heart failure grade assessment model training:

[0242] Input: training set sample D, maximum number of iterations M = 5;

[0243] Output: Strong Learner

[0244] GBDT is an ensemble learning method based on gradient boosting. It uses multiple regression trees as base learners and improves the predictive ability of the model by iteratively fitting the negative gradient of the loss function (also called residual or pseudo residual).

[0245] The basic process of the GBDT regression algorithm is as follows:

[0246] 1) Initialize a constant as the first base learner, usually the mean of the training data labels or the value that minimizes the loss function.

[0247]

[0248] Among them, L is the loss function, y i is the true label, and γ is a constant.

[0249] The loss function is square loss. Since the square loss function is a convex function, we can directly take the derivative and get c:

[0250]

[0251] Let the derivative be equal to 0:

[0252]

[0253] Therefore, the value of c is the mean of all sample labels.

[0254] The initial learner is obtained as:

[0255] For each round of iteration, the negative gradient (residual) of each sample is calculated as the new response variable. The number of iterations is m = 5. The specific process is as follows:

[0256] i) Calculate the negative gradient of the sample, y and the learner f obtained in the previous round m-1 Difference of (x):

[0257]

[0258] ii) Use the regression tree to fit the new response variable and obtain a regression tree whose leaf nodes are divided into several regions γ j1 ,j=1,2,3,...,J m

[0259] iii) For j = 1, 2, 3, ..., J m Calculate the best fit value, usually the mean of the residuals for all samples in the region or the value that minimizes the loss function:

[0260]

[0261] iv) Update the current model by multiplying the regression tree obtained in this round by a step size (learning rate) and adding it to the previous model:

[0262]

[0263] Thus, the first regression tree is obtained, such as Figure 6 As shown;

[0264] v) Repeat steps i-iv to get the 2nd, 3rd, 4th, and 5th trees, such as Figure 7 、 8 , 9, and 10.

[0265] 3) Output the final model expression:

[0266]

[0267] The final tree is Figure 11 shown.

[0268] 4) The model was validated using test data, and the validation results showed an accuracy of 87.9%

[0269] In a specific embodiment, after new ECG data is obtained, the heart failure grade assessment model is updated through the semi-supervised learning method Self-Training:

[0270] 1) Using existing models Label the data;

[0271] 2) Set the training termination condition to a 3% improvement in accuracy;

[0272] 3) Use all currently labeled samples (including original data samples and newly labeled data samples) to train the classifier. If the training termination condition is met or the training is repeated 10 times, the classifier is terminated.

[0273] 4) Select the model with the highest accuracy For the final model;

[0274] 5) Update the heart failure grade assessment model to

[0275] Example 3

[0276] See also Figure 3 , Figure 3 It is a schematic diagram of the terminal device structure provided by one embodiment of the present invention.

[0277] A terminal device of this embodiment includes: a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program, the steps of the above-mentioned method for processing electrocardiogram data for heart failure assessment in the embodiment are implemented, for example Figure 1Alternatively, when the processor executes the computer program, the functions of the modules in the above-mentioned system embodiments are realized, for example: Figure 2 All modules of the ECG data processing system for heart failure assessment are shown.

[0278] In addition, an embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for processing electrocardiogram data for heart failure assessment as described in any of the above embodiments.

[0279] Those skilled in the art will understand that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. The terminal device may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0280] The processor 301 may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor 301 is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0281] The memory 302 can be used to store the computer programs and / or modules. The processor 301 implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory 302. The memory 302 can mainly include a program storage area and a data storage area. The program storage area can store an operating system and at least one application required for a function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created based on the use of the mobile phone (such as audio data, a phone book, etc.). In addition, the memory can include a high-speed random access memory and can also include a non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0282] Wherein, if the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or system that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0283] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive work.

[0284] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A device for processing electrocardiogram data for heart failure assessment, characterized in that: include: Data acquisition module, waveform screening module, waveform positioning module, data extraction module and result generation module; The data acquisition module is used to acquire the patient's electrocardiogram data; The waveform screening module is used to perform peak screening on the electrocardiogram in the electrocardiogram data to obtain a plurality of electrocardiogram waves; The waveform positioning module is used to determine the R wave among the multiple electrocardiogram waves, and determine the P wave, Q wave, S wave and T wave based on the position of the R wave; The data extraction module is used to extract the ECG acquisition parameters and ECG waveform parameters corresponding to the R wave, P wave, Q wave, S wave and T wave from the ECG data based on the generation time of the R wave, P wave, Q wave, S wave and T wave; The result generating module is configured to establish a coordinate system for each ECG wave, using the ECG waveform parameter type of the ECG wave as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate; identify a historical type region in the coordinate system based on the parameter value range of the historical ECG waveform type corresponding to the ECG wave; use the waveform type corresponding to the historical type region in which the ECG waveform parameter of the ECG wave falls as the ECG waveform type of the ECG wave; and aggregate the ECG waveform type with the ECG acquisition parameters to generate ECG analysis data for the patient, thereby performing a heart failure assessment based on the ECG analysis data; The method of establishing a coordinate system with the ECG waveform parameter type as the horizontal coordinate and the ECG waveform parameter value as the vertical coordinate includes: for the R wave, using the amplitude, direction and symmetry as the horizontal coordinate, and using the amplitude value, direction value and symmetry value as the vertical coordinate to establish a plane coordinate system; wherein the direction value includes: 0, 1 and -1, upward is 1, downward is -1, and no waveform is 0; for the Q wave, using the amplitude, time limit and overall curve smoothness as the horizontal coordinate, and using the amplitude value, time limit and overall curve smoothness value as the vertical coordinate to establish a plane coordinate system; for the R wave, using the amplitude, direction and A plane coordinate system is established with symmetry as the horizontal coordinate and amplitude, orientation and symmetry as the vertical coordinates. For S waves, a plane coordinate system is established with amplitude, orientation and width as the horizontal coordinates and amplitude, orientation and width as the vertical coordinates. For P waves, a plane coordinate system is established with amplitude, orientation and overall curve smoothness as the horizontal coordinates and amplitude, orientation and overall curve smoothness as the vertical coordinates. For T waves, a plane coordinate system is established with amplitude, orientation and overall curve smoothness as the horizontal coordinates and amplitude, orientation and overall curve smoothness as the vertical coordinates.

Images