Pasting type electrocardiogram lead wire wireless head end and electrocardiogram management system and method

Abstract

The invention provides a paste type electrocardiogram lead wire wireless head end and an electrocardiogram management system and method, and belongs to the technical field of electrocardiograms, and the paste type electrocardiogram lead wire wireless head end comprises a shell; the display layer and the identification layer are arranged at the top of the shell; the electrode layer is arranged in the middle of the bottom of the shell; and the bonding layer is arranged at the bottom of the shell and surrounds the electrode layer. Patient experience is optimized, suction cup falling and skin injury are avoided through the pasting type design, and the suction cup is particularly suitable for pediatric patients, elderly patients and critical patients.

Description

Technical Field

[0001] The present invention relates to the technical field of electrocardiograms, and in particular to an adhesive electrocardiogram lead wire wireless head end, an electrocardiogram management system and a method. Background Art

[0002] Clinical pain points of traditional electrocardiographs:

[0003] Lead wire tip design flaws: Traditional ECG machines use suction cup lead wire tips that rely on physical adsorption to secure them to the skin. For thin patients, those with dry skin, or those with low platelets, the suction cups can easily fall off, resulting in signal interruption. Repeated adsorption can also cause skin damage (such as bruising and allergies), making them unsuitable for long-term monitoring.

[0004] Complex lead wire management: A multi-lead ECG machine requires 10-12 lead wires to be connected, which are prone to tangling and knotting during clinical operations, increasing the preparation time of medical staff. In addition, the wired connection limits the patient's range of movement, making it difficult to meet the needs of dynamic ECG monitoring.

[0005] Signal transmission and processing limitations: Traditional wired transmission is subject to signal delays and noise interference (such as power frequency interference and motion artifacts). Furthermore, data must be transmitted back to the ECG machine for processing in real time, making it unsuitable for independent operation. This also lacks portability and flexibility for emergency or mobile scenarios (such as ambulances and home monitoring).

[0006] Single feature analysis: Traditional algorithms only extract time domain features (such as RR interval and QT interval) based on single-lead signals, ignoring the spatial correlation of multi-lead signals. The detection accuracy of complex arrhythmias (such as multi-lead ST segment abnormalities caused by myocardial ischemia) is limited. Summary of the Invention

[0007] The present invention provides an adhesive electrocardiogram lead wire wireless head end, an electrocardiogram management system and a method to solve one or more of the above problems.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] A wireless headend for an adhesive electrocardiogram lead wire, comprising:

[0010] case;

[0011] The display layer and the logo layer are both arranged on the top of the shell;

[0012] The electrode layer is provided at the middle position of the bottom of the shell;

[0013] The adhesive layer is arranged at the bottom of the shell and surrounds the electrode layer.

[0014] In this specification, the display layer, the identification layer, the electrode layer and the adhesive layer can be detachably mounted on the housing.

[0015] An electrocardiogram management system, comprising:

[0016] The adhesive ECG lead wire wireless head end as described in any one of the above;

[0017] a microprocessor, disposed in the housing;

[0018] An electrode sheet, provided on the electrode layer;

[0019] A display module, provided in the display layer and connected to the microprocessor;

[0020] A signal amplification and filtering module is provided in the housing and connected to the electrode sheet;

[0021] an analog-to-digital converter, disposed in the housing and connected to the signal amplification and filtering module and the microprocessor;

[0022] The wireless communication module is arranged in the housing and connected to the microprocessor to communicate wirelessly with the electrocardiograph.

[0023] In this specification, the microprocessor is connected to a signal encryption module.

[0024] In this specification, the microprocessor is connected to a storage unit.

[0025] In this specification, the microprocessor is connected to a power module, and the display module, signal amplification and filtering module, analog-to-digital converter, wireless communication module, signal encryption module and storage unit are all connected to the power module.

[0026] A method for applying a wireless head end of an adhesive electrocardiogram lead wire comprises:

[0027] S1. Using 12 of the above-mentioned adhesive ECG lead wire wireless head ends, 12 leads are collected to collect 12-lead ECG signals, and preprocessing is performed to obtain a short window moving average signal;

[0028] S2. Short window moving average signal Mapped to a space-time coordinate system, where: the lead axis [0,11] corresponds to 12 lead positions, , is the lead number; time axis [-100,100] corresponds to the time window based on the peak value of the R wave. , is the sampling point number, is the sampling point number of the peak value of the R wave in lead i; forming a set of discrete points in time and space ( ),in ,Right now The signal value of the short window moving average signal; using the quadratic polynomial Fitting a set of discrete points in space and time, solving the coefficients by the least squares method ; is the output value of the surface model, which represents the ECG signal in the space-time coordinate system ( , ) at the position;

[0029] S3. At the reference point of the space-time coordinate system ( =5.5, =0) to calculate the Gaussian curvature K, mean curvature H and peak offset of the surface ; Detect the peak value of R wave in each lead, calculate RR interval, QT interval and their correction values, and calculate the average RR interval , RR interval standard deviation , maximum RR interval difference , the average value of the 12-lead corrected QT interval, the maximum value of the 12-lead corrected QT interval;

[0030] S4. Fitting the surface to features , Gaussian curvature K, mean curvature H, peak offset Average RR interval with time domain characteristics , RR interval standard deviation , maximum RR interval difference , the average value of the 12-lead corrected QT interval, and the maximum value of the 12-lead corrected QT interval are fused into a 15-dimensional feature vector; the vector is input into the random forest model for training to obtain a random forest model;

[0031] The real-time collected 12-lead ECG signals are input into the random forest model for real-time classification, and arrhythmia warning is triggered based on the classification results.

[0032] In this manual, the calculation process of Gaussian curvature K and mean curvature H is as follows:

[0033] Calculate the reference point of the surface in the space-time coordinate system ( =5.5, =0) at first and second order partial derivatives:

[0034] First-order partial derivatives: ;

[0035] ;

[0036] Second-order partial derivatives: ;

[0037] Substitute into the Gaussian curvature and mean curvature formulas:

[0038] ;

[0039] in, express about The first-order partial derivative of express about The first-order partial derivative of express about The second-order partial derivative of express about The second-order partial derivative of express First about Find the partial derivative, then about Find the mixed second-order partial derivatives of the partial derivatives.

[0040] In this manual, peak shift The calculation process is as follows:

[0041] By solving the equations for the surface gradient to be zero, the coordinates of the peak point of the surface are determined ( ): ;

[0042] After obtaining the coordinates of the peak point, calculate the offset relative to the reference point: .

[0043] In this manual, the process of R wave peak detection is as follows:

[0044] when Exceeding the threshold When it is a local maximum, it is determined to be the R wave peak value and its sampling point number is recorded. , threshold The formula is as follows:

[0045] ;

[0046] and is the empirical coefficient, The long window moving average signal is obtained by performing mean filtering on the short window moving average signal.

[0047] In summary, the present invention has at least the following beneficial effects:

[0048] Improved patient experience: The adhesive design prevents the suction cup from falling off and causing skin damage, making it particularly suitable for pediatric, geriatric, and critically ill patients. Wireless operation eliminates lead wire entanglement, allowing patients to move more freely and for use in Holter monitoring, rehabilitation training, and other scenarios.

[0049] Improved medical care efficiency: The electrodes can be attached with one click, and the operation time is shortened from 5-10 minutes with traditional methods to 1-2 minutes.

[0050] Real-time wireless transmission: supports remote consultation, such as transmitting ECG data from an ambulance to the hospital in advance, shortening emergency diagnosis time.

[0051] Through surface fitting and random forest, multimodal feature analysis (multi-lead synchronous analysis) is performed to effectively improve detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 The figure is a structural diagram of the wireless head end of the adhesive electrocardiogram lead wire involved in the present invention.

[0054] Figure 2 This is a schematic diagram of the composition of the electrocardiogram management system involved in the present invention.

[0055] Figure 3 The figure is a flow chart of the method for the wireless head end of the adhesive electrocardiogram lead wire involved in the present invention.

[0056] Figure 4 It is a flow chart of the surface fitting and spatial feature extraction involved in the present invention.

[0057] Reference numerals:

[0058] 1. Shell; 2. Display layer; 3. Logo layer; 4. Electrode layer; 5. Adhesive layer. DETAILED DESCRIPTION

[0059] Hereinafter, only certain exemplary embodiments are briefly described. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the embodiments of the present invention. Therefore, the drawings and description are to be regarded as illustrative in nature and not restrictive.

[0060] The disclosure below provides many different embodiments or examples for implementing different structures of the embodiments of the present invention. In order to simplify the disclosure of the embodiments of the present invention, the components and configurations of specific examples are described below. Of course, these are merely examples and are not intended to limit the embodiments of the present invention. In addition, the embodiments of the present invention may repeat reference numerals and / or reference letters in different examples. Such repetition is for the purpose of simplicity and clarity and does not in itself indicate the relationship between the various embodiments and / or configurations discussed.

[0061] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0062] like Figure 1 As shown, this embodiment provides an adhesive electrocardiogram lead wire wireless head end, including:

[0063] Shell 1;

[0064] The display layer 2 and the logo layer 3 are both arranged on the top of the housing 1;

[0065] The electrode layer 4 is provided at the middle position of the bottom of the shell 1;

[0066] The adhesive layer 5 is disposed at the bottom of the housing 1 and surrounds the electrode layer 4 .

[0067] In some embodiments, the display layer 2 , the identification layer 3 , the electrode layer 4 and the adhesive layer 5 can all be detachably mounted on the housing 1 .

[0068] like Figure 2 As shown, an electrocardiogram management system includes:

[0069] The adhesive ECG lead wire wireless head end as described in any one of the above;

[0070] A microprocessor is provided in the housing 1;

[0071] An electrode sheet, provided on the electrode layer 4;

[0072] A display module, provided in the display layer 2 and connected to the microprocessor;

[0073] A signal amplification and filtering module is provided in the housing 1 and connected to the electrode sheet;

[0074] An analog-to-digital converter, disposed in the housing 1 and connected to the signal amplification and filtering module and the microprocessor;

[0075] The wireless communication module is disposed in the housing 1 and is connected to the microprocessor to communicate wirelessly with the electrocardiograph.

[0076] In some embodiments, the microprocessor is connected to a signal encryption module.

[0077] In some embodiments, the microprocessor is connected to a storage unit.

[0078] In some embodiments, the microprocessor is connected to a power module, and the display module, signal amplification and filtering module, analog-to-digital converter, wireless communication module, signal encryption module and storage unit are all connected to the power module.

[0079] like Figure 3 As shown, a method for applying a wireless head end of an adhesive electrocardiogram lead wire includes:

[0080] S1. Using 12 of the above-mentioned adhesive ECG lead wire wireless head ends, 12 leads are collected to collect 12-lead ECG signals, and preprocessing is performed to obtain a short window moving average signal;

[0081] S2. Short window moving average signal Mapped to a space-time coordinate system, where: the lead axis [0,11] corresponds to 12 lead positions, , is the lead number; time axis [-100,100] corresponds to the time window based on the peak value of the R wave. , is the sampling point number, is the sampling point number of the peak value of the R wave in lead i; forming a set of discrete points in time and space ( ),in ,Right now The signal value of the short window moving average signal; using the quadratic polynomial Fitting a set of discrete points in space and time, solving the coefficients by the least squares method ; is the output value of the surface model, which represents the ECG signal in the space-time coordinate system ( , ) at the position;

[0082] S3. At the reference point of the space-time coordinate system ( =5.5, =0) to calculate the Gaussian curvature K, mean curvature H and peak offset of the surface ; Detect the peak value of R wave in each lead, calculate RR interval, QT interval and their correction values, and calculate the average RR interval , RR interval standard deviation , maximum RR interval difference , the average value of the 12-lead corrected QT interval, the maximum value of the 12-lead corrected QT interval;

[0083] S4. Fitting the surface to features , Gaussian curvature K, mean curvature H, peak offset Average RR interval with time domain characteristics , RR interval standard deviation , maximum RR interval difference , the average value of the 12-lead corrected QT interval, and the maximum value of the 12-lead corrected QT interval are fused into a 15-dimensional feature vector; the vector is input into the random forest model for training to obtain a random forest model;

[0084] The real-time collected 12-lead ECG signals are input into the random forest model for real-time classification, and arrhythmia warning is triggered based on the classification results.

[0085] The technical ideas of the present invention are as follows:

[0086] 1. Adhesive ECG lead wire wireless headend and its system

[0087] Core design:

[0088] Adhesive electrodes: They use medical-grade breathable tape (adhesive layer 5) + dry / wet electrodes (such as Ag / AgCl electrodes or pre-coated conductive gel), fixed to the skin through an adhesive backing, replacing traditional suction cups to avoid mechanical damage and are suitable for all types of patients.

[0089] Dry electrode solution:

[0090] Metal electrodes: Use silver (Ag) or silver chloride (Ag / AgCl) electrode sheets, such as 3MRedDot2233 (medical-grade disposable electrodes with adhesive backing and customizable lead markings).

[0091] Flexible electrodes: Flexible printed circuit (FPC) electrode sheets, such as the Epson RCX series (ultra-thin flexible substrate, suitable for lamination on complex body surfaces, and can be integrated with dry electrode arrays).

[0092] Wet electrode solution:

[0093] Conductive gel electrodes: such as Hollister7325 (disposable electrode sheets pre-coated with conductive gel, strong adhesion, suitable for long-term monitoring).

[0094] Lead identification: Directly screen print or laser engrave on the electrode surface or identification layer 3, such as marking "V1" and "RA", refer to the identification specifications of the GE Marquette 12SL lead system.

[0095] Wireless communication module: Integrates Bluetooth 5.1 / Wi-Fi 6 chip (such as Nordic nRF52840), supports real-time communication with ECG machines or mobile terminals, with a transmission distance of ≥10 meters, and has a built-in AES-256 hardware encryption chip (such as Microchip ATECC608A) to ensure ECG data security.

[0096] Independent processing unit (microprocessor): Equipped with a low-power MCU (such as the STM32L476RG, Cortex-M4 core, floating-point support, suitable for real-time signal processing), and built-in signal amplification and filtering circuits (such as the TIA DS1298), it independently performs ECG signal acquisition, analog-to-digital conversion (24-bit accuracy), and preliminary analysis (such as heart rate calculation), and displays the waveform in real time on the OLED screen. Integrated algorithm chip: Texas Instruments MSP430FR6989 (FRAM memory technology, suitable for low-power data storage and simple algorithm execution).

[0097] Housing 1: Medical-grade materials can be used, such as ABS plastic (Chimei PA-757), PC / ABS alloy (Covestro Bayblend®), medical-grade POM (DuPont Delrin® 100P), and flexible silicone (Dow Corning Sylgard® 184). Housing 1 can be cylindrical, quasi-circular (elliptical), or rectangular, and can be designed based on actual needs.

[0098] Signal amplification and filtering module:

[0099] Integrated chip: Use TIADS1298 (8-channel bioelectric signal acquisition chip with built-in programmable gain amplifier (PGA) and 50Hz / 60Hz notch filter).

[0100] Discrete solution: Use ADI AD8232 (single-channel ECG signal conditioning module with integrated amplification, filtering and baseline drift correction circuits).

[0101] Analog-to-digital converter (ADC):

[0102] High-precision ADC: TIA DS1299 (24-bit resolution, supports multi-channel simultaneous sampling, suitable for medical-grade signal accuracy requirements).

[0103] Wireless communication module:

[0104] Bluetooth module: Nordic nRF52832 (low-power Bluetooth 5.1 chip, supports long-distance transmission and multi-device networking).

[0105] Wi-Fi module: ESP32-C3 (supports Wi-Fi 4 and Bluetooth 5, has high integration, and is suitable for scenarios requiring access to hospital LANs).

[0106] Signal encryption module:

[0107] Hardware encryption chip: Microchip ATECC608A (compliant with EAL6+ security certification, supports AES-256 encryption, and can be directly integrated into the circuit board).

[0108] Software encryption algorithm: integrated mbedTLS library (lightweight encryption protocol, supporting RSA / ECC key exchange and AES data encryption).

[0109] Storage Unit:

[0110] Flash memory chip: SpansionS25FL256S (256Mb serial flash memory, supports fast read and write, used for offline data caching).

[0111] EEPROM: Atmel AT24C256 (256Kb electrically erasable memory, used to store device configuration and calibration parameters).

[0112] Power module:

[0113] Button battery: Maxell CR2032 (3V lithium manganese battery, capacity about 225mAh, suitable for miniaturized design).

[0114] Wireless charging module: STMicroelectronics ST25R3911 (supports Qi protocol wireless charging and enables contactless power supply).

[0115] Display module (display layer 2):

[0116] OLED screen: 0.96-inch OLED module driven by SH1106 (resolution 128×64, supports real-time waveform display, low power consumption).

[0117] LCD screen: 2.2-inch TFT LCD screen driven by ILI9341 (resolution 320×240, color display, suitable for scenarios requiring detailed parameter display).

[0118] Structural innovation:

[0119] Modular design: The electrodes, processing module, and power supply (such as a CR2032 button battery with a battery life of ≥24 hours) can be quickly disassembled and replaced, supporting single-use or repeated disinfection.

[0120] Multi-lead synchronous acquisition: supports 12-lead synchronous sampling (sampling rate 500Hz), and the lead identification (such as RA, V1) is silk-screened on the surface of the electrode to avoid position misjudgment.

[0121] 2. ECG feature analysis algorithm (method for wireless headend of adhesive ECG lead wires)

[0122] Multi-dimensional feature extraction:

[0123] Space-time surface fitting:

[0124] Map the 12-lead signal to the time-space coordinate system (lead axis , Timeline ), by fitting a quadratic polynomial surface ( ) Model the spatial distribution of multi-lead signals and extract geometric features such as Gaussian curvature K, mean curvature H, and peak offset to reflect the morphological differences of the waveforms in the time and space domains.

[0125] Traditional time domain features:

[0126] Parameters such as the RR interval mean / standard deviation, QT interval correction (QTc), and ST segment deviation are calculated, and multi-lead synchronous analysis is combined to improve feature robustness.

[0127] Random Forest Classification Model:

[0128] It integrates 15-dimensional feature vectors (10-dimensional surface features + 5-dimensional time domain features) and uses a random forest algorithm (100 trees, 3 node splitting features) to classify arrhythmias. It supports multi-category recognition, including normal, ventricular premature beats, and atrial premature beats. It is trained on the MIT-BIH database and has a high accuracy rate.

[0129] In some embodiments, S1. Multi-lead ECG signal acquisition and preprocessing:

[0130] Objective: To synchronously collect 12-lead ECG signals using adhesive electrodes, perform noise reduction and feature enhancement, and provide high-quality data for subsequent analysis.

[0131] S11.Synchronous acquisition of multiple lead signals:

[0132] Twelve adhesive ECG lead wires are used for 12-lead ECG wireless headends to acquire 12-lead ECG signals. An acquisition module with an integrated 24-bit analog-to-digital converter (such as the TIA DS1298) is used to synchronously acquire 12-lead ECG signals via adhesive electrodes. The 12-lead ECG signals include the right arm lead (RA), left arm lead (LA), left leg lead (LL), precordial leads V1, V2, V3, V4, V5, and V6, as well as three augmented limb leads (aVR, aVL, and aVF).

[0133] During the acquisition process, the signal is sampled at a sampling frequency of 500 Hz to convert the continuous analog signal into a discrete digital signal. , where i represents the lead number (i=1 to i=12), and n represents the sampling point number (n=1 to n=N, where N is the total number of sampling points).

[0134] S12. Signal preprocessing (based on improved Pan-Tompkins algorithm):

[0135] Discrete digital signal for each lead The following processes are performed in sequence:

[0136] 1. First-order difference calculation: The differential signal is obtained by calculating the difference between the current sampling point and the previous sampling point. , to highlight the slope characteristics of the QRS complex.

[0137] 2. Square operation: Square the differential signal to obtain the signal , in order to enhance the energy characteristics of the QRS complex and facilitate subsequent detection.

[0138] 3. Double window moving average filter:

[0139] Short window filtering: Use a sliding window with a width of 30 sampling points (corresponding to 60 milliseconds) to perform mean filtering on the squared signal to obtain a short window moving average signal , used to filter out high-frequency noise.

[0140] Long window filtering: Use a sliding window with a width of 150 sampling points (corresponding to 300 milliseconds) to perform mean filtering on the short window moving average signal to obtain a long window moving average signal , used to estimate signal baseline and dynamic threshold calculation.

[0141] In some embodiments, as Figure 4 As shown, S2. Surface fitting and spatial feature extraction:

[0142] Objective: To map the short-window moving average signal into a spatiotemporal coordinate system, construct a spatial distribution model of the signal through quadratic polynomial surface fitting, and extract geometric features that reflect the waveform morphology.

[0143] S21. Construction of space-time coordinate system:

[0144] Establish a two-dimensional coordinate system, where:

[0145] Lead axis u: evenly maps 12 leads to continuous intervals Specifically, the right arm lead (RA, i=1) corresponds to =0, left arm lead (LA, i=2) corresponds to =1, left leg lead (LL, i=3) corresponds to =2, pressurized limb lead aVR (i=4) corresponds to =3, pressurized limb lead aVL (i=5) corresponds to =4, pressurized limb lead aVF (i=6) corresponds to =5, precordial lead V1 (i=7) corresponds to =6, precordial lead V2 (i=8) corresponds to =7, precordial lead V3 (i=9) corresponds to =8, precordial lead V4 (i=10) corresponds to =9, precordial lead V5 (i=11) corresponds to =10, precordial lead V6 (i=12) corresponds to =11. is the lead axis coordinate, indicating the lead position.

[0146] Time axis v: Take the peak value of the R wave of each cardiac cycle as the reference point (v=0), and take 100 sampling points before and after the reference point (corresponding to ±200 milliseconds), that is, the time axis range is v is the time axis coordinate, which represents the time offset based on the peak value of the R wave.

[0147] The preprocessed short window moving average signal Mapped into the space-time coordinate system to form a discrete point set ,in ( is the sampling point number of the peak value of the R wave in lead i). ,Right now The signal value is the short window moving average signal.

[0148] S22. Quadratic polynomial surface fitting:

[0149] A quadratic polynomial model is used to perform surface fitting on a set of discrete points in space and time. The model expression is: ;

[0150] Among them, z is the output value of the surface model, which represents the ECG signal in the space-time coordinate system ( , ) (corresponding to the short window moving average signal ), is a constant term, which represents the reference amplitude of the surface at the origin of the space-time coordinate system (u=0,v=0); is the linear coefficient of the lead axis, reflecting the linear change trend of the signal amplitude along the lead position u direction (such as the increase / decrease of signal strength between leads); is the time axis linear term coefficient, reflecting the linear trend of the signal amplitude changing with time v (such as the overall rise or fall of the waveform); The quadratic coefficient of the lead axis reflects the curvature of the signal amplitude distribution among the leads (such as the convexity and concavity of the multi-lead waveform); is the cross-term coefficient, which reflects the interaction between the lead axis u and the time axis v (such as the rate of change of the waveform difference between leads over time); is the quadratic term coefficient of the time axis, reflecting the curvature of the signal amplitude over time (such as the steepness of the rising / falling edge of the QRS complex). The surface of a normal ECG signal usually has a symmetrical time-space distribution (such as Close to 0, Close to 0); Arrhythmias (such as myocardial ischemia) may cause abnormal surface curvature (K or H deviates from the normal range) or peak shift ( or ).

[0151] Solving for the coefficient vector by the least squares method , T is the matrix transpose, which minimizes the sum of the squares of the vertical distances from the discrete points to the surface, that is, minimizes the objective function: ; or the coefficients can be solved through matrix inversion or iterative optimization algorithm (such as gradient descent method); finally, a fitting surface is obtained, and the coefficients and geometric characteristics (such as curvature) of the fitting surface can be used to analyze the spatiotemporal distribution of multi-lead ECG signals, providing key features for arrhythmia detection.

[0152] S23. Calculation of surface geometric features:

[0153] At the reference point (u=5.5,v=0) of the space-time coordinate system (i.e., the midpoint of the 12 leads), the following surface geometric features are calculated:

[0154] 1. Gaussian curvature K and mean curvature H: First calculate the first-order partial derivative and second-order partial derivative of the surface at this point:

[0155] First-order partial derivatives: ;

[0156] ;

[0157] Second-order partial derivatives: ;

[0158] Substitute into the Gaussian curvature and mean curvature formulas: ; express about The first-order partial derivative of express about The first-order partial derivative of express about The second-order partial derivative of express about The second-order partial derivative of express First about Find the partial derivative, then about Find the mixed second-order partial derivatives of the partial derivatives.

[0159] 2. Peak offset: Determine the coordinates of the peak point of the surface by solving the equations for the surface gradient to be zero ( ): ;

[0160] After obtaining the coordinates of the peak point, calculate the offset relative to the reference point: .

[0161] In some embodiments, S3. Traditional time domain feature extraction:

[0162] Objective: To extract classic time-domain ECG parameters from the signal of each lead, including R wave position, RR interval, QT interval, etc., and calculate global statistical features.

[0163] Single lead feature calculation, for the original signal of each lead Do the following:

[0164] 1. R wave peak detection: based on short window moving average signal , the dynamic threshold method is used to detect the peak position of the R wave. The dynamic threshold calculation formula is: ; and is the empirical coefficient (such as =0.1, =0.6), dynamically adjusting the detection threshold by using the baseline mean and the maximum value of the short window signal to adapt to the signal amplitude differences of different patients.

[0165] when Exceeding the threshold When it is a local maximum, it is determined to be the R wave peak value and its sampling point number is recorded. .

[0166] 2. RR interval calculation: For continuous cardiac cycles, calculate the sampling point interval between adjacent R wave peaks (j=2 to j=M, M is the total number of cardiac cycles), and converted to time intervals: (Unit: seconds).

[0167] 3. QT interval measurement: The sampling interval from the start of the Q wave to the end of the T wave is , converted to milliseconds: The QT interval was corrected using the Fridericia formula: (in is the RR interval corresponding to the cardiac cycle)

[0168] Global statistical features:

[0169] Average RR interval: ;

[0170] Interval standard deviation: ;

[0171] Maximum RR interval difference: .

[0172] In some embodiments, S4. Feature fusion and random forest classification:

[0173] Objective: To integrate spatial features obtained by surface fitting with traditional time domain features, construct a random forest classification model, and realize automatic identification of arrhythmias.

[0174] S41. Feature vector construction combines the following two types of features to form a 15-dimensional feature vector :

[0175] 1. Surface fitting features (10 dimensions):

[0176] Quadratic polynomial coefficients: ;

[0177] Geometric characteristics: Gaussian curvature K, mean curvature H, peak offset

[0178] 2. Time domain features (5 dimensions):

[0179] Mean RR interval , standard deviation of RR interval , maximum RR interval difference ;

[0180] The average value of the corrected QT interval in 12 leads ;

[0181] The maximum corrected QT interval in 12 leads .

[0182] S42. Random Forest Model Training:

[0183] Use a labeled arrhythmia dataset (such as the MIT-BIH arrhythmia database) to train a random forest model. The specific steps are as follows:

[0184] 1. Data preparation: transform the feature vector The training samples are composed of the corresponding category labels y (such as normal, ventricular premature beats, atrial premature beats, etc.) ,in is the total number of training set samples.

[0185] 2. Feature standardization: Standardize the feature vector so that the mean of each feature is 0 and the standard deviation is 1.

[0186] 3. Model parameter settings:

[0187] Number of decision trees: 100;

[0188] The number of features randomly selected when each node splits: 3 (i.e. );

[0189] Minimum number of leaf node samples: 5;

[0190] 4. Model construction:

[0191] 100 bootstrap samples are randomly selected from the training set using the bootstrap sampling method, each sample contains samples (sampling with replacement).

[0192] For each bootstrap sample, a decision tree is constructed. When splitting each node, the feature that reduces the Gini impurity the most is selected from the three randomly selected features as the splitting feature. The Gini impurity calculation formula is: Where C is the total number of arrhythmia categories, is the proportion of class c samples in the node.

[0193] After all decision trees are constructed, the final category of the sample is determined by majority voting to obtain a trained random forest model.

[0194] S43. Real-time classification and early warning:

[0195] For the real-time collected ECG signal, extract the feature vector according to steps S1-S3 , input into the trained random forest model for prediction. If the prediction result is an abnormal category (such as ventricular premature beats, atrial premature beats, etc.), an early warning signal is triggered, prompting medical staff to intervene.

[0196] In some embodiments, S5. Model optimization and clinical feedback:

[0197] Objective: To evaluate model performance and adjust parameters based on real data from clinical applications to further improve classification accuracy and robustness.

[0198] S51. Model performance evaluation:

[0199] The model was evaluated using an independent test set, and the following performance metrics were calculated:

[0200] Accuracy: The number of correctly classified samples (true positive samples TP + true negative samples TN) accounts for the total number of samples in the test set The ratio, that is, the accuracy = .

[0201] Sensitivity: The ratio of the number of correctly identified abnormal samples (true positive TP) to the actual number of abnormal samples (TP + FN, FN is the number of false negative samples), that is, sensitivity = .

[0202] Specificity: The ratio of the number of correctly identified normal samples (true negative TN) to the actual number of normal samples (TN + FP, FP is the number of false positive samples), that is, specificity = .

[0203] S52. Parameter adjustment strategy:

[0204] Adjust model parameters based on performance evaluation results:

[0205] 1. If the sensitivity is low (many abnormal samples are missed):

[0206] Increase the number of decision trees to 150 to improve the model's ability to capture complex features.

[0207] Reducing the minimum number of leaf node samples to 3 makes it easier for the decision tree to fit abnormal patterns in the training data.

[0208] 2. If the specificity is low (normal samples are often misclassified as abnormal):

[0209] Increase the minimum number of leaf node samples to 10 to improve the strictness of node splitting and reduce overfitting.

[0210] Reduce the polynomial order of surface fitting (such as using a linear polynomial instead), simplify the spatial feature model, and improve generalization ability.

[0211] 3. Iterative optimization: Add new data collected from clinical applications (including new abnormal samples and normal samples) to the training set, re-execute step S42, and update the random forest model to ensure that the model performance continues to improve as data accumulates.

[0212] The above embodiments are intended to illustrate the present invention, not to limit the present invention. Therefore, changes in illustrative values ​​or substitutions of equivalent components should still fall within the scope of the present invention.

[0213] From the above detailed description, it will be clear to those skilled in the art that the present invention can indeed achieve the aforementioned objectives and is in compliance with the provisions of the Patent Law.

[0214] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as covering the preferred embodiments and all changes and modifications that fall within the scope of the invention. The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

[0215] It should be noted that the above description of the relevant processes is for illustration and purpose only and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to the processes under the guidance of this specification. However, such modifications and changes are still within the scope of this specification.

[0216] The basic concepts have been described above. It will be apparent to those skilled in the art after reading this application that the above disclosures are merely illustrative and do not constitute limitations on this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and amendments to this application. Such modifications, improvements, and amendments are suggested in this application and remain within the spirit and scope of the exemplary embodiments of this application.

[0217] At the same time, this application uses specific terms to describe the embodiments of this application. For example, "one embodiment," "an embodiment," and / or "some embodiments" refer to a certain feature, structure, or characteristic related to at least one embodiment of this application. Therefore, it should be emphasized and noted that "one embodiment," "an embodiment," or "an alternative embodiment" mentioned twice or more in different places in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of this application may be appropriately combined.

[0218] Furthermore, those skilled in the art will appreciate that various aspects of the present application may be illustrated and described in terms of a number of patentable categories or situations, including any new and useful process, machine, product, or combination of substances, or any new and useful improvement thereof. Thus, various aspects of the present application may be implemented entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software. Each of the above hardware and software may be referred to as a "unit," "module," or "system." Furthermore, various aspects of the present application may take the form of a computer program product embodied in one or more computer-readable media, with computer-readable program code embodied therein.

[0219] The computer program code required for the operation of each part of this application can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, etc., conventional procedural programming languages ​​such as C programming language, Visual Basic, Fortran2103, Perl, COBOL2102, PHP, ABAP, dynamic programming languages ​​such as Python, Ruby and Groovy, or other programming languages. The program code can be run entirely on the user's computer, or as a standalone software package on the user's computer, or partly on the user's computer and partly on a remote computer, or entirely on a remote computer or server. In the latter case, the remote computer can be connected to the user's computer via any network form, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or used as a service such as software as a service (SaaS).

[0220] In addition, unless expressly stated in the claims, the order of the processing elements and sequences described in this application, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this application. Although the above disclosure discusses some embodiments of the invention that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this application. For example, although the implementation of the various components described above can be embodied in a hardware device, it can also be implemented as a pure software solution, for example, installation on an existing server or mobile device.

[0221] Similarly, it should be noted that in order to simplify the presentation of this disclosure and thereby facilitate understanding of one or more of the invention's embodiments, the foregoing descriptions of the embodiments of this disclosure sometimes combine multiple features into a single embodiment, figure, or description thereof. However, this approach should not be interpreted as reflecting an intention that the claimed subject matter requires more features than expressly recited in each claim. Rather, the subject matter of the invention may possess fewer features than the single embodiment described above.

Claims

1. A wireless head end of an adhesive electrocardiogram lead wire, characterized in that: include: case; The display layer and the logo layer are both arranged on the top of the shell; The electrode layer is provided at the middle position of the bottom of the shell; The adhesive layer is arranged at the bottom of the shell and surrounds the electrode layer.

2. The adhesive electrocardiogram lead wire wireless head end according to claim 1, characterized in that: The display layer, the marking layer, the electrode layer and the adhesive layer can all be detachably mounted on the housing.

3. An electrocardiogram management system, characterized in that: include: The adhesive electrocardiogram lead wire wireless head end according to any one of claims 1-2; a microprocessor, disposed in the housing; An electrode sheet, provided on the electrode layer; A display module, provided in the display layer and connected to the microprocessor; A signal amplification and filtering module is provided in the housing and connected to the electrode sheet; an analog-to-digital converter, disposed in the housing and connected to the signal amplification and filtering module and the microprocessor; The wireless communication module is arranged in the housing and connected to the microprocessor to communicate wirelessly with the electrocardiograph.

4. The electrocardiogram management system according to claim 3, characterized in that: The microprocessor is connected to a signal encryption module.

5. The electrocardiogram management system according to claim 4, characterized in that: The microprocessor is connected to a storage unit.

6. The adhesive electrocardiogram lead wire wireless head end according to claim 5, characterized in that: The microprocessor is connected to a power module, and the display module, signal amplification and filtering module, analog-to-digital converter, wireless communication module, signal encryption module and storage unit are all connected to the power module.

7. An application method of a wireless head end of an adhesive electrocardiogram lead wire, characterized in that: include: S1. Using the adhesive ECG lead wire wireless head end of any one of claims 1-2 to perform 12-lead ECG to collect 12-lead ECG signals and obtain a short window moving average signal after preprocessing; S2. Short window moving average signal Mapped to a space-time coordinate system, where: the lead axis [0,11] corresponds to 12 lead positions, , is the lead number; time axis [-100,100] corresponds to the time window based on the peak value of the R wave. , is the sampling point number, is the sampling point number of the peak value of the R wave in lead i; forming a set of discrete points in time and space ( ),in ,Right now The signal value of the short window moving average signal; using the quadratic polynomial Fitting a set of discrete points in space and time, solving the coefficients by the least squares method ; is the output value of the surface model, which represents the ECG signal in the space-time coordinate system ( , ) at the position; S3. At the reference point of the space-time coordinate system ( =5.5, =0) to calculate the Gaussian curvature K, mean curvature H and peak offset of the surface ; Detect the peak value of R wave in each lead, calculate RR interval, QT interval and their correction values, and calculate the average RR interval , RR interval standard deviation , maximum RR interval difference , the average value of the 12-lead corrected QT interval, the maximum value of the 12-lead corrected QT interval; S4. Fitting the surface to features , Gaussian curvature K, mean curvature H, peak offset Average RR interval with time domain characteristics , RR interval standard deviation , maximum RR interval difference , the average value of the 12-lead corrected QT interval, and the maximum value of the 12-lead corrected QT interval are fused into a 15-dimensional feature vector; the vector is input into the random forest model for training to obtain a random forest model; The real-time collected 12-lead ECG signals are input into the random forest model for real-time classification, and arrhythmia warning is triggered based on the classification results.

8. The adhesive electrocardiogram lead wire wireless head end according to claim 7, characterized in that: The calculation process of Gaussian curvature K and mean curvature H is as follows: Calculate the reference point of the surface in the space-time coordinate system ( =5.5, =0) at first and second order partial derivatives: First-order partial derivatives: ; ； Second-order partial derivatives: ; Substitute into the Gaussian curvature and mean curvature formulas: ； in, express about The first-order partial derivative of express about The first-order partial derivative of express about The second-order partial derivative of express about The second-order partial derivative of express First about Find the partial derivative, then about Find the mixed second-order partial derivatives of the partial derivatives.

9. The adhesive electrocardiogram lead wire wireless head end according to claim 7, characterized in that: Peak offset The calculation process is as follows: By solving the equations for the surface gradient to be zero, the coordinates of the peak point of the surface are determined ( ): ; After obtaining the coordinates of the peak point, calculate the offset relative to the reference point: 。 10. The adhesive electrocardiogram lead wire wireless head end according to claim 7, characterized in that: The process of R wave peak detection is as follows: when Exceeding the threshold When it is a local maximum, it is determined to be the R wave peak value and its sampling point number is recorded. , threshold The formula is as follows: ； and is the empirical coefficient, The long window moving average signal is obtained by performing mean filtering on the short window moving average signal.

Images