Blood pressure pulse wave interference identification method

Abstract

According to the blood pressure pulse wave interference recognition method, a trained interference wave recognition model is arranged in blood pressure detection equipment, pulse wave data collected by the equipment are obtained in real time and sent to the interference wave recognition model for recognition, and once the pulse wave data with interference data are recognized, an alarm is given out in real time. Detectors are reminded to collect blood pressure pulse wave data again, it is ensured that whether measurement results receive external interference or not is found in time, and the blood pressure detection efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method for identifying blood pressure pulse wave interference. Background Art

[0002] Blood pressure measuring equipment based on the oscillometric method, such as an oscillometric sphygmomanometer, includes a controller, a cuff wrapped around the arm, and an inflation device, and a pressure sensor is arranged on the cuff. For example, the patent with publication number CN116509354A is a blood pressure measuring equipment currently used by our company. When used, the cuff is inflated by the inflation device and then wrapped around the arm of the person being tested. The oscillometric sphygmomanometer automatically adjusts the inflation volume of the cuff wrapped around the upper arm during a complete measurement process to change the pressure. The blood flow through the blood vessels has a certain oscillation wave, that is, the blood pressure pulse wave, which is received by the pressure sensor and gradually deflated. According to the change of the pulse wave, the pressure and fluctuation detected by the pressure sensor also change accordingly. The moment with the largest fluctuation is selected as the average pressure. Based on the average pressure, a certain value is found forward as the systolic pressure, and a certain value is found backward as the diastolic pressure. The oscillometric sphygmomanometer collects two sets of data in chronological order during a complete measurement process, namely the blood pressure pulse wave and the sphygmomanometer pressure curve. Specifically, Figure 1 This is an example of standard pulse wave data collected by an oscillometric sphygmomanometer. The blue solid line is the blood pressure pulse wave, and the yellow dotted line is the pressure curve. The horizontal axis is the time unit in milliseconds, and the vertical axis is millimeters of mercury.

[0003] However, in actual applications, it is found that oscillometric sphygmomanometers often generate interference due to some reasons, resulting in measurement failures. For example, environmental noise and external environment noise may be picked up by the sphygmomanometer's sensor, interfering with the measurement signal and causing inaccurate results; arm movement, arm movement or muscle tension will cause cuff pressure fluctuations, resulting in unstable measurement signals; electromagnetic interference, surrounding electromagnetic equipment affects the normal operation of the sphygmomanometer's electronic components and sensors, resulting in measurement errors; incorrect cuff position will lead to inaccurate measurement signals; the electronic components and signal processing circuits inside the sphygmomanometer may generate internal electromagnetic interference when working.

[0004] However, because the oscillometric sphygmomanometer does not provide a visual interface for measurement personnel to confirm the waveform, the measurement personnel on site can only see one measurement result. Experienced measurement personnel can find the measurement error caused by some interference from the measurement results and immediately ask the person being tested to remeasure. However, if it is not possible to find out that the measurement process has been interfered with on the spot, but later when the doctor exports the waveform of the blood pressure data from the oscillometric sphygmomanometer and confirms it on the computer, it is found from the waveform that the measurement result is the result of interference, and it is necessary to find the patient again to measure again, which wastes a lot of manpower and material resources. Summary of the invention

[0005] In order to solve the problem that the existing portable oscillometric sphygmomanometer cannot timely detect the interfered blood pressure pulse wave, the present invention provides a blood pressure pulse wave interference identification method, which can timely detect whether the measurement result is subject to external interference, thereby effectively improving the measurement result accuracy of the oscillometric blood pressure measurement device.

[0006] The technical solution of the present invention is as follows: a method for identifying blood pressure pulse wave interference, characterized in that it comprises the following steps:

[0007] S1: Obtaining blood pressure pulse wave data collected by a blood pressure detection device, recorded as: original pulse wave data;

[0008] Obtaining pressure curve data and blood pressure pulse wave corresponding to the original pulse wave data, which are respectively recorded as: original pressure curve and original pulse wave curve;

[0009] S2: Calculate the peaks and the number of peaks on the original pressure curve;

[0010] Assume that the number of peaks in the original pressure curve is n;

[0011] Taking the number of peaks n as a distinguishing characteristic value, determining the relationship between the number of peaks n included in the original pressure curve and a preset threshold value PE;

[0012] If n>PE, it is directly determined that there is interference in the original pulse wave;

[0013] Otherwise, execute step S3;

[0014] S3: Obtain the original pulse wave curve; calculate the positions and heights of all peaks of the original pulse wave curve;

[0015] At the same time, the distance between each pair of adjacent peaks is calculated, and the standard deviation of all peak distances is calculated, which is recorded as: PeakLocationSTD;

[0016] S4: Calculate the positions and heights of all troughs of the original pulse wave curve;

[0017] At the same time, the distance between each pair of adjacent troughs is calculated, and the standard deviation of all the obtained trough distances is calculated, which is recorded as: TroughLocationSTD;

[0018] S5: Calculating the change trend data of the peak of the original pulse wave curve;

[0019] The peak change trend data include: the standard deviation of the peak height difference PeakHeightSTD and the change rate of the peak height difference PeakHeightDrop;

[0020] The peak height standard deviation PeakHeightSTD is: the standard deviation of the height difference between the height of all peaks and the height of the next adjacent peak is calculated;

[0021] The change rate of the peak height difference PeakHeightDrop describes the difference between two adjacent peak height differences PeakHeight;

[0022] S6: Calculating the change trend data of the trough of the original pulse wave curve;

[0023] The trough change trend data include: Trough Height STD, the standard deviation of trough height, and the change rate of trough height difference, Trough Height Drop;

[0024] The trough height standard deviation TroughHeightSTD is: calculate the standard deviation of the height difference between the height of all troughs and the height of the next adjacent trough;

[0025] The change rate of the trough height difference TroughHeightDrop describes the difference between two adjacent trough height differences TroughHeight;

[0026] S7: Calculate the peak angle difference PeakAngle and the trough angle difference TroughAngle of the original pulse wave curve;

[0027] The peak angle difference PeakAngle is the difference between the angles of two adjacent peaks;

[0028] The trough angle difference TroughAngle is the difference between the angles of two adjacent troughs;

[0029] S8: Subtract the trough envelope from the peak envelope of the original pulse wave data to obtain an amplitude curve.

[0030] Calculate the number of peaks AmplitudePeaks included in the amplitude curve;

[0031] S9: Based on the distinguishing feature value, construct a distinguishing feature vector describing the original pulse wave data:

[0032] The following corresponding to the original pulse wave curve: peak distance standard deviation PeakLocationSTD, trough distance standard deviation TroughLocationSTD, peak height standard deviation PeakHeightSTD, peak height difference change rate PeakHeightDrop, trough height standard deviation TroughHeightSTD, trough height difference change rate TroughHeightDrop, peak angle difference PeakAngle, trough angle difference TroughAngle, peak number AmplitudePeaks are used as distinguishing feature values ​​to construct the distinguishing feature vector:

[0033] [PeakLocationSTD,TroughLocationSTD,PeakHeightSTD,TroughHeightSTD,PeakHeightDrop,

[0034] TroughHeightDrop,PeakAngle,TroughAngle,AmplitudePeaks];

[0035] S10: Construct interference wave recognition model based on machine learning algorithm model;

[0036] Constructing a training set and a validation set based on the distinguishing feature vectors of the normal waveform and the interference waveform corresponding to the pulse wave data in the historical data;

[0037] Using the training set to train the interference wave recognition model to obtain the trained interference wave recognition model;

[0038] The output result of the interference wave identification model includes: a normal waveform and an interference waveform;

[0039] S11: setting the trained interference wave recognition model in a blood pressure detection device;

[0040] S12: acquiring blood pressure pulse wave data collected by the device in real time, recorded as pulse wave data to be analyzed;

[0041] Calculating and obtaining a pressure curve to be analyzed corresponding to the pulse wave data to be analyzed;

[0042] If it is determined that the number n of peaks of the pressure curve to be analyzed is greater than the preset threshold PE, then step S14 is directly executed;

[0043] Otherwise, if n≤PE, the distinguishing characteristic value corresponding to the blood pressure pulse wave curve to be analyzed is calculated, and the distinguishing characteristic vector corresponding to the blood pressure pulse wave curve to be analyzed is constructed, which is recorded as: the distinguishing characteristic vector to be analyzed;

[0044] S13: sending the distinguishing feature vector to be analyzed into the trained interference wave recognition model;

[0045] Obtaining the classification result output by the interference wave identification model, and if the output result is an interference waveform, executing step S14;

[0046] Otherwise, if the distinguishing characteristic vector to be analyzed is a normal waveform, then the judgment is terminated;

[0047] S14: A warning sound is issued to indicate that the collected blood pressure pulse wave data contains interference data, and the judgment is terminated.

[0048] It is further characterized by:

[0049] The calculation method of the peak distance standard deviation PeakLocationSTD and the trough distance standard deviation TroughLocationSTD is:

[0050] PeakLocationSTD=std(PeakLocation1,PeakLocation2,PeakLocation3...PeakLocationx);

[0051] TroughLocationSTD=std(TroughLocation1,TroughLocation2,TroughLocation3...TroughLocationy);

[0052] Where, PeakLocation1, PeakLocation2, PeakLocation3…PeakLocationx are the locations of the peaks;

[0053] TroughLocation1, TroughLocation2, TroughLocation3...TroughLocationy is the location of the trough;

[0054] x is the total number of peaks, y is the total number of troughs;

[0055] The std() function is used to calculate the standard deviation along the specified axis;

[0056] The calculation method of the peak height standard deviation PeakHeightSTD, the peak height difference change rate PeakHeightDrop, the trough standard deviation TroughHeightSTD and the trough height difference change rate TroughHeightDrop is:

[0057] PeakHeightSTD=std(abs(PeakHeight1-PeakHeight2), abs(PeakHeight2-PeakHeight3), abs(PeakHeight3-Peak Height4)...abs(PeakHeight(x-1)-PeakHeightx));

[0058] TroughHeightSTD=std(abs(TroughHeight1-TroughHeight2),abs(TroughHeight2-TroughHeight3),abs(Trough Height3-TroughHeight4)...abs(TroughHeight(x-1)-TroughHeightx));

[0059]

[0060]

[0061] Wherein, PeakHeight1, PeakHeight2, PeakHeight3…PeakHeightx are the heights of x peaks;

[0062] TroughHeight1, TroughHeight2, TroughHeight3…TroughHeighty is the height of y troughs;

[0063] x is the total number of peaks, y is the total number of troughs;

[0064] The std() function is used to calculate the standard deviation along the specified axis; the abs() function is the absolute value function;

[0065] The calculation method of the peak angle difference PeakAngle and the trough angle difference TroughAngle is:

[0066]

[0067]

[0068] Where PeakAngle i is the angle of the i-th peak Peak, TroughAngle i represents the angle of the i-th trough;

[0069] x is the total number of peaks, y is the total number of troughs;

[0070] The angle PeakAngle i and TroughAngle i The calculation method is:

[0071] The angle of the i-th peak PeakAngle i The calculation method is: draw tangents from the highest point of the i-th peak Peak to the peak curves on both sides, and the angle between the two tangents is the angle PeakAngle i ;

[0072] TroughAngle i The calculation method is: draw tangents from the highest point of the ith trough Through to the peak curves on both sides, and the angle between the two tangents is the angle TroughAngle i ;

[0073] In step S2, the method for calculating the peak of the original pressure curve includes the following steps:

[0074] a1: Calculate all local maxima existing on the original pressure curve,

[0075] a2: read each of the maximum values ​​one by one, and record them as: the maximum value to be judged;

[0076] a3: Compare the maximum value to be judged with the preset peak judgment threshold;

[0077] When the maximum value to be determined is greater than the peak determination threshold, the maximum value to be determined is recorded as a peak; and step a4 is executed;

[0078] Otherwise, execute step a4;

[0079] a4: Loop through steps a2 to a3 until all local maxima have been calculated, and then stop the calculation.

[0080] The present application provides a blood pressure pulse wave interference recognition method, which sets a trained interference wave recognition model in a blood pressure detection device, obtains pulse wave data collected by the device in real time and sends it to the interference wave recognition model for recognition. Once pulse wave data with interference data is recognized, a warning is issued in real time to remind the detection personnel to re-collect the blood pressure pulse wave data, so as to ensure that whether the measurement result is subject to external interference is discovered in time, thereby improving the efficiency of blood pressure detection. In this method, the normal waveform and the interference waveform are distinguished based on the pressure curve of the blood pressure pulse wave data and the overall curve distinguishing features of the pulse wave curve. When there are multiple peak values ​​in the pressure curve, the pressure peak value is The number of peaks in the force curve is used as a distinguishing feature to determine whether there is interference data in the pulse wave data to be analyzed; if there is only one peak in the pressure curve, a distinguishing feature vector containing 9 distinguishing feature values ​​is constructed based on the curve characteristics of the pulse wave curve to determine whether there is interference data in the pulse wave data to be analyzed; this method does not require image recognition of the specific waveforms of the pressure curve and the pulse wave curve, but only obtains the distinguishing feature values ​​based on simple calculations to determine the type of waveform, with small calculation amount and high performance efficiency; and the normal waveform and the interference waveform are distinguished based on the distinguishing feature values ​​at multiple angles to ensure the accuracy of the results. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 This is a schematic diagram of a normal pulse wave;

[0082] Figure 2 Schematic diagram of pulse wave with interference;

[0083] Figure 3 This is an example of an amplitude curve of a normal waveform;

[0084] Figure 4 is an example of the amplitude curve of the interference waveform;

[0085] Figure 5 This is a flow chart of the blood pressure pulse wave interference identification method of this application. DETAILED DESCRIPTION

[0086] like Figure 1 and Figure 2 As shown, Figure 1 In a normal pulse wave, the blue blood pressure pulse wave should be spindle-shaped, that is, it oscillates slowly up first, and then oscillates slowly down; the yellow pressure curve, that is, the pressure in the cuff should show a slowly decreasing trend, because the cuff gradually deflates during the measurement process, so the pressure slowly decreases. Figure 2In the disturbed pulse wave shown, the blood pressure pulse wave presents an irregular and chaotic waveform, and due to the compression of the cuff during measurement or the incorrect placement of the cuff when wearing it, there may be a tendency for the pressure curve to rise. This method identifies and distinguishes the two waveforms based on the distinguishing features in the normal waveform and the interference waveform.

[0087] like Figure 5 As shown, a method for identifying blood pressure pulse wave interference includes the following steps.

[0088] S1: Obtaining blood pressure pulse wave data collected by a blood pressure detection device, recorded as: original pulse wave data;

[0089] The pressure curve data and the blood pressure pulse wave corresponding to the original pulse wave data are obtained and recorded as: original pressure curve and original pulse wave curve respectively.

[0090] S2: Calculate the peaks and the number of peaks on the original pressure curve;

[0091] Assume that the number of peaks in the original pressure curve is n;

[0092] The number of peaks n is used as a distinguishing characteristic value to determine the relationship between the number of peaks n included in the original pressure curve and the preset threshold PE;

[0093] If n>PE, it is directly judged that there is interference in the original pulse wave;

[0094] Otherwise, execute step S3.

[0095] The value of n is set according to the calculation accuracy requirement. In this embodiment, the value of n is 2.

[0096] In step S2, the method for calculating the peak of the original pressure curve includes the following steps:

[0097] a1: Calculate all local maxima existing on the original pressure curve,

[0098] a2: Read each maximum value one by one, recorded as: maximum value to be judged;

[0099] a3: Compare the maximum value to be judged with the preset peak judgment threshold;

[0100] When the maximum value to be determined is greater than the peak determination threshold, the maximum value to be determined is recorded as a peak; and step a4 is executed;

[0101] Otherwise, execute step a4;

[0102] a4: Loop through steps a2 to a3 until all local maxima have been calculated, and then stop the calculation.

[0103] In a normal pulse wave without any interference, the pressure curve should be a straight line. However, in actual applications, it is impossible to be completely free of any interference during the measurement process, so there will be slight changes in the pressure curve, and these changes can also be calculated as a peak through local maximum value calculation. Depending on the application scenario, some degree of curve change is acceptable. Therefore, a peak judgment threshold is set in this method, and all local maximum values ​​below the peak judgment threshold are regarded as acceptable data. Figure 1 In the embodiment shown, there is no peak in the pressure curve, but Figure 2 There are 8 peaks in the curve shown that are greater than the peak judgment threshold, specifically refer to the 8 marked with red numbers in the curve. The setting of the peak judgment threshold ensures that the method is more practical. The specific local maximum algorithm is implemented based on the existing technology.

[0104] If there are more than PE peaks in the pressure curve, it can be determined that the pulse wave has interference. By setting the distinguishing characteristic value of the pressure curve, a part of the interference that can be reflected by the pressure can be quickly eliminated, reducing the subsequent calculation amount and improving the detection efficiency. If there is no interference in the pressure curve, the pulse wave curve is judged again.

[0105] S3: Obtaining the original pulse wave curve; calculating the positions and heights of all peaks of the original pulse wave curve;

[0106] At the same time, the distance between each pair of adjacent peaks is calculated to obtain the standard deviation of the peak distance, which is recorded as: PeakLocationSTD.

[0107] S4: Calculate the positions and heights of all troughs of the original pulse wave curve;

[0108] At the same time, the distance between each pair of adjacent troughs is calculated, and the obtained trough distance standard deviation is recorded as: TroughLocationSTD.

[0109] The calculation method of the peak distance standard deviation PeakLocationSTD and the trough distance standard deviation TroughLocationSTD is:

[0110] PeakLocationSTD=std(PeakLocation1,PeakLocation2,PeakLocation3...PeakLocationx);

[0111] TroughLocationSTD=std(TroughLocation1,TroughLocation2,TroughLocation3...TroughLocationy);

[0112] Where, PeakLocation1, PeakLocation2, PeakLocation3…PeakLocationx are the locations of the peaks;

[0113] TroughLocation1, TroughLocation2, TroughLocation3...TroughLocationy is the location of the trough;

[0114] x is the total number of peaks, y is the total number of troughs;

[0115] The std() function is used to calculate the standard deviation along the specified axis.

[0116] A normal pulse wave oscillates regularly and presents a spindle shape, so the distance between each peak and each trough should be uniform. Therefore, PeakLocationSTD and TroughLocationSTD should be smaller in a normal waveform and larger in an interference waveform.

[0117] S5: Calculate the change trend data of the peak of the original pulse wave curve;

[0118] The peak change trend data include: peak height standard deviation PeakHeightSTD and peak height difference change rate PeakHeightDrop;

[0119] The peak height standard deviation PeakHeightSTD is: the height standard deviation of all peaks and the height of the next adjacent peak;

[0120] The rate of change of the peak height difference PeakHeightDrop describes the difference between the height differences PeakHeight of two adjacent peaks.

[0121] S6: Calculate the change trend data of the trough of the original pulse wave curve;

[0122] The trough change trend data include: trough height standard deviation TroughHeightSTD and trough height difference change rate PeakHeightDrop;

[0123] The trough height standard deviation TroughHeightSTD is: the height standard deviation of all trough heights and the height of the next adjacent trough;

[0124] The rate of change of the trough height difference TroughHeightDrop describes the difference between two adjacent trough height differences TroughHeight.

[0125] The calculation method of the peak height standard deviation PeakHeightSTD, the change rate of the peak height difference PeakHeightDrop, the trough height standard deviation TroughHeightSTD and the change rate of the trough height difference TroughHeightDrop is:

[0126] PeakHeightSTD=std(abs(PeakHeight1-PeakHeight2), abs(PeakHeight2-PeakHeight3), abs(PeakHeight3-Peak Height4)...abs(PeakHeight(x-1)-PeakHeightx));

[0127] TroughHeightSTD=std(abs(TroughHeight1-TroughHeight2),abs(TroughHeight2-TroughHeight3),abs(Trough Height3-TroughHeight4)...abs(TroughHeight(x-1)-TroughHeightx));

[0128]

[0129] Wherein, PeakHeight1, PeakHeight2, PeakHeight3…PeakHeightx are the heights of x peaks;

[0130] TroughHeight1, TroughHeight2, TroughHeight3…TroughHeighty is the height of y troughs;

[0131] x is the total number of peaks, y is the total number of troughs;

[0132] The std() function is used to calculate the standard deviation along the specified axis; the abs() function is the absolute value function.

[0133] In a normal waveform, the peak height shows a trend of gradually rising and then gradually falling; the trough height shows a trend of gradually falling and then gradually rising. In an interference waveform, it shows irregular oscillations. In a normal waveform, PeakHeightSTD and TroughHeightSTD should be smaller, while in an interference wave, they are larger. PeakHeightSTD and TroughHeightSTD describe the change between the height of the current peak or trough in the curve and the height of the next one, while PeakHeightDrop and PeakHeightDrop describe the difference between the current peak or trough height and the next one, and compare the difference between the next one and the next one, further describing the curve from the perspective of the law of change. PeakHeightSTD, TroughHeightSTD, PeakHeightDrop and PeakHeightDrop ensure accurate description of the changes and laws of the pulse wave curve.

[0134] S7: Calculate the peak angle difference PeakAngle and the trough angle difference TroughAngle of the original pulse wave curve; in a normal waveform, PeakAngle and TroughAngle should be smaller, and in an interference wave, they should be larger.

[0135] PeakAngle is the difference between the angles of two adjacent peaks;

[0136] The trough angle difference TroughAngle is the difference between the angles of two adjacent troughs.

[0137] The calculation method of the peak angle difference PeakAngle and the trough angle difference TroughAngle is:

[0138]

[0139]

[0140] Where PeakAngle i is the angle of the i-th peak Peak, TroughAngle i represents the angle of the i-th trough;

[0141] x is the total number of peaks, and y is the total number of troughs.

[0142] PeakAngle i and TroughAngle i The calculation method is:

[0143] The angle of the i-th peak PeakAngle iThe calculation method is: draw tangents from the highest point of the i-th peak Peak to the peak curves on both sides, and the angle between the two tangents is the angle PeakAngle i ;

[0144] TroughAngle i The calculation method is: draw tangents from the highest point of the ith trough Through to the peak curves on both sides, and the angle between the two tangents is the angle TroughAngle i .

[0145] S8: Subtract the trough envelope from the peak envelope of the original pulse wave data to obtain an amplitude curve.

[0146] Calculates the number of peaks AmplitudePeaks included in the amplitude curve.

[0147] like Figure 3 and Figure 4 As shown, the horizontal axis is time, the light blue thin solid line is the original oscillation curve, that is, the pulse wave curve, the green dotted line is the peak envelope line, the dark blue dotted line is the trough envelope line, and the red thick solid line is the amplitude curve.

[0148] The amplitude curve in a normal waveform should be a parabola, and the amplitude curve of an interference waveform has a larger vibration interference. Therefore, AmplitudePeaks is calculated as the number of peaks in the amplitude curve. In a normal waveform, AmplitudePeaks should be smaller, and in an interference waveform, AmplitudePeaks should be larger.

[0149] S9: Based on the distinguishing eigenvalues, construct a distinguishing eigenvector describing the original pulse wave data:

[0150] The original pulse wave curve corresponding to: peak distance standard deviation PeakLocationSTD, trough distance standard deviation TroughLocationSTD, peak height standard deviation PeakHeightSTD, peak height difference change rate PeakHeightDrop, trough height standard deviation TroughHeightSTD, trough height difference change rate TroughHeightDrop, peak angle difference PeakAngle, trough angle difference TroughAngle, peak number AmplitudePeaks as distinguishing feature values ​​to construct a distinguishing feature vector:

[0151] [PeakLocationSTD,TroughLocationSTD,PeakHeightSTD,TroughHeightSTD,PeakHeightDrop,

[0152] TroughHeightDrop,PeakAngle,TroughAngle,AmplitudePeaks].

[0153] S10: Construct interference wave recognition model based on machine learning algorithm model;

[0154] Constructing a training set and a validation set based on distinguishing feature vectors of normal waveforms and interference waveforms corresponding to pulse wave data in historical data;

[0155] Use the training set to train the interference wave recognition model to obtain a trained interference wave recognition model;

[0156] The output results of the interference wave identification model include: normal waveform and interference waveform.

[0157] The interference wave recognition model is constructed based on an existing machine learning algorithm model. In this embodiment, the interference wave recognition model is constructed based on a random forest algorithm model. The specific construction method of the training data set and the verification set and the training method can be implemented based on existing technologies.

[0158] S11: Setting the trained interference wave recognition model in the blood pressure detection device.

[0159] S12: acquiring blood pressure pulse wave data collected by the device in real time, recorded as pulse wave data to be analyzed;

[0160] Calculate and obtain the pressure curve to be analyzed corresponding to the pulse wave data to be analyzed;

[0161] If it is determined that the number n of peaks of the pressure curve to be analyzed is greater than the preset threshold PE, then step S14 is directly executed;

[0162] Otherwise, if n≤PE, the distinguishing characteristic value corresponding to the blood pressure pulse wave curve to be analyzed is calculated, and a distinguishing characteristic vector corresponding to the blood pressure pulse wave curve to be analyzed is constructed, which is recorded as: distinguishing characteristic vector to be analyzed.

[0163] S13: sending the distinguishing feature vector to be analyzed into the trained interference wave recognition model to distinguish the normal waveform from the interference waveform;

[0164] Obtain the classification result output by the interference wave recognition model. If the output result is an interference waveform, execute step S14;

[0165] Otherwise, if the distinguishing characteristic vector to be analyzed is a normal waveform, the current judgment is terminated.

[0166] S14: A warning sound is issued to indicate that the collected blood pressure pulse wave data contains interference data, and the judgment is terminated.

[0167] After using the technical solution of the present invention, after the measurement, the blood pressure detection personnel can judge in real time whether the collected blood pressure pulse wave data is interfered based on the collected blood pressure pulse wave data and the pressure curve and the blood pressure pulse wave curve, which greatly improves the work efficiency. The calculation method of the distinguishing characteristic value corresponding to the pressure curve and the blood pressure pulse wave curve in the method is simple and has a fast calculation speed, ensuring that an alarm can be issued in real time.

Claims

1. A method for identifying blood pressure pulse wave interference, It is characterized in that It includes the following steps: S1: Obtaining blood pressure pulse wave data collected by a blood pressure detection device, recorded as: original pulse wave data; Obtaining pressure curve data and blood pressure pulse wave corresponding to the original pulse wave data, which are respectively recorded as: original pressure curve and original pulse wave curve; S2: Calculate the peaks and the number of peaks on the original pressure curve; Assume that the number of peaks in the original pressure curve is n; Taking the number of peaks n as a distinguishing characteristic value, determining the relationship between the number of peaks n included in the original pressure curve and a preset threshold value PE; If n>PE, it is directly determined that there is interference in the original pulse wave; Otherwise, execute step S3; S3: Obtain the original pulse wave curve; calculate the positions and heights of all peaks of the original pulse wave curve; At the same time, the distance between each pair of adjacent peaks is calculated, and the standard deviation of all peak distances is calculated, which is recorded as: PeakLocationSTD; S4: Calculate the positions and heights of all troughs of the original pulse wave curve; At the same time, the distance between each pair of adjacent troughs is calculated, and the standard deviation of all the obtained trough distances is calculated, which is recorded as: TroughLocationSTD; S5: Calculating the change trend data of the peak of the original pulse wave curve; The peak change trend data include: the standard deviation of the peak height difference PeakHeightSTD and the change rate of the peak height difference PeakHeightDrop; The peak height standard deviation PeakHeightSTD is: the standard deviation of the height difference between the height of all peaks and the height of the next adjacent peak is calculated; The change rate of the peak height difference PeakHeightDrop describes the difference between two adjacent peak height differences PeakHeight; S6: Calculating the change trend data of the trough of the original pulse wave curve; The trough change trend data include: Trough Height STD, the standard deviation of trough height, and the change rate of trough height difference, Trough Height Drop; The trough height standard deviation TroughHeightSTD is: calculate the standard deviation of the height difference between the height of all troughs and the height of the next adjacent trough; The change rate of the trough height difference TroughHeightDrop describes the difference between two adjacent trough height differences TroughHeight; S7: Calculate the peak angle difference PeakAngle and the trough angle difference TroughAngle of the original pulse wave curve; The peak angle difference PeakAngle is the difference between the angles of two adjacent peaks; The trough angle difference TroughAngle is the difference between the angles of two adjacent troughs; S8: Subtract the trough envelope from the peak envelope of the original pulse wave data to obtain an amplitude curve. Calculate the number of peaks AmplitudePeaks included in the amplitude curve; S9: Based on the distinguishing feature value, construct a distinguishing feature vector describing the original pulse wave data: The following corresponding to the original pulse wave curve: peak distance standard deviation PeakLocationSTD, trough distance standard deviation TroughLocationSTD, peak height standard deviation PeakHeightSTD, peak height difference change rate PeakHeightDrop, trough height standard deviation TroughHeightSTD, trough height difference change rate TroughHeightDrop, peak angle difference PeakAngle, trough angle difference TroughAngle, peak number AmplitudePeaks are used as distinguishing feature values ​​to construct the distinguishing feature vector: [PeakLocationSTD,TroughLocationSTD,PeakHeightSTD,TroughHeightSTD,PeakHeightDrop, TroughHeightDrop,PeakAngle,TroughAngle,AmplitudePeaks]; S10: Construct interference wave recognition model based on machine learning algorithm model; Constructing a training set and a validation set based on the distinguishing feature vectors of the normal waveform and the interference waveform corresponding to the pulse wave data in the historical data; Using the training set to train the interference wave recognition model to obtain the trained interference wave recognition model; The output result of the interference wave identification model includes: a normal waveform and an interference waveform; S11: setting the trained interference wave recognition model in a blood pressure detection device; S12: acquiring blood pressure pulse wave data collected by the device in real time, recorded as pulse wave data to be analyzed; Calculating and obtaining a pressure curve to be analyzed corresponding to the pulse wave data to be analyzed; If it is determined that the number n of peaks of the pressure curve to be analyzed is greater than the preset threshold PE, then step S14 is directly executed; Otherwise, if n≤PE, the distinguishing characteristic value corresponding to the blood pressure pulse wave curve to be analyzed is calculated, and the distinguishing characteristic vector corresponding to the blood pressure pulse wave curve to be analyzed is constructed, which is recorded as: the distinguishing characteristic vector to be analyzed; S13: sending the distinguishing feature vector to be analyzed into the trained interference wave recognition model; Obtaining the classification result output by the interference wave identification model, and if the output result is an interference waveform, executing step S14; Otherwise, if the distinguishing characteristic vector to be analyzed is a normal waveform, then the judgment is terminated; S14: A warning sound is issued to indicate that the collected blood pressure pulse wave data contains interference data, and the judgment is terminated.

2. A method for identifying blood pressure pulse wave interference according to claim 1, Features: The calculation method of the peak distance standard deviation PeakLocationSTD and the trough distance standard deviation TroughLocationSTD is: PeakLocationSTD=std(PeakLocation1,PeakLocation2,PeakLocation3...PeakLocationx); TroughLocationSTD=std(TroughLocation1,TroughLocation2,TroughLocation3...TroughLocationy); Where, PeakLocation1, PeakLocation2, PeakLocation3…PeakLocationx are the locations of the peaks; TroughLocation1, TroughLocation2, TroughLocation3...TroughLocationy is the location of the trough; x is the total number of peaks, y is the total number of troughs; The std() function is used to calculate the standard deviation along the specified axis.

3. A method for identifying blood pressure pulse wave interference according to claim 1, Features: The calculation method of the peak height standard deviation PeakHeightSTD, the peak height difference change rate PeakHeightDrop, the trough standard deviation TroughHeightSTD and the trough height difference change rate TroughHeightDrop is: PeakHeightSTD=std(abs(PeakHeight1-PeakHeight2), abs(PeakHeight2-PeakHeight3), abs(PeakHeight3-Peak Height4)...abs(PeakHeight(x-1)-PeakHeightx)); TroughHeightSTD=std(abs(TroughHeight1-TroughHeight2),abs(TroughHeight2-TroughHeight3),abs(Trough Height3-TroughHeight4)...abs(TroughHeight(x-1)-TroughHeightx)); Where, PeakHeight1, PeakHeight2, PeakHeight3…PeakHeightx are the heights of x peaks; TroughHeight1, TroughHeight2, TroughHeight3…TroughHeighty is the height of y troughs; x is the total number of peaks, y is the total number of troughs; The std() function is used to calculate the standard deviation along the specified axis; the abs() function is the absolute value function.

4. A method for identifying blood pressure pulse wave interference according to claim 1, Features: The calculation method of the peak angle difference PeakAngle and the trough angle difference TroughAngle is: Where PeakAngle i is the angle of the i-th peak Peak, TroughAngle i represents the angle of the i-th trough; x is the total number of peaks, and y is the total number of troughs.

5. A method for identifying blood pressure pulse wave interference according to claim 4, Features: The angle PeakAngle i and TroughAngle i The calculation method is: The angle of the i-th peak PeakAngle i The calculation method is: draw tangents from the highest point of the i-th peak Peak to the peak curves on both sides, and the angle between the two tangents is the angle PeakAngle i ; TroughAngle i The calculation method is: draw tangents from the highest point of the ith trough Through to the peak curves on both sides, and the angle between the two tangents is the angle TroughAngle i .

6. A method for identifying blood pressure pulse wave interference according to claim 1, Features: In step S2, the method for calculating the peak of the original pressure curve includes the following steps: a1: Calculate all local maxima existing on the original pressure curve, a2: read each of the maximum values ​​one by one, and record them as the maximum value to be judged; a3: Compare the maximum value to be judged with the preset peak judgment threshold; When the maximum value to be determined is greater than the peak determination threshold, the maximum value to be determined is recorded as a peak; and step a4 is executed; Otherwise, execute step a4; a4: Loop through steps a2 to a3 until all local maxima have been calculated, and then stop the calculation.

Images