Intelligent heart electrophysiological signal analysis and Jiangshi connecting line optimization system and method

Abstract

The invention relates to an intelligent heart electrophysiological signal analysis and Jiangshi connecting line optimization system and method, and belongs to the technical field of electrocardio systems, the system comprises a plurality of modules for cooperative work, a data acquisition module is used for acquiring human body 12-lead electrocardiogram and intracardiac electrogram signals and executing a pace-making stimulation function; the signal preprocessing module filters received signals, the feature extraction module extracts time domain, frequency domain and nonlinear features, the deep learning module identifies abnormal heart rhythms by using a pre-training model, the parameter tuning module dynamically adjusts model parameters and optimizes and analyzes the parameters, and the Jiangshi connecting line optimization module improves signal quality. The intelligent monitoring module automatically monitors multiple indexes, the display interface module visually displays an electrocardiogram, the system compatible module is connected with multiple devices, and the control module coordinates all the modules to ensure efficient operation of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of electrocardiographic systems, and in particular to an intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization system and method. Background Art

[0002] In recent years, left bundle branch pacing (LBBP), a novel physiological pacing technique, has emerged as a promising candidate for maintaining electrical and mechanical synchrony, mitigating ventricular remodeling and worsening cardiac function compared to traditional right ventricular pacing. LBBP has demonstrated advantages in clinical practice, including a low threshold, excellent long-term sensing, and relatively simple operation, and has gradually replaced His bundle pacing. However, LBBP requires a multi-channel electrophysiological recorder, which is lacking in many primary care hospitals, hindering its widespread implementation.

[0003] Some researchers have demonstrated the ability to achieve left bundle branch pacing by combining multi-lead or 12-lead ECG monitoring with intracardiac electrocardiograms (ICGs) from a pacemaker programmer. However, this approach is difficult and cannot be used routinely. Traditional ECG analysis methods rely primarily on physician experience, which can be subjective and inefficient. Although some computer-assisted analysis systems have recently emerged, they still suffer from numerous deficiencies in signal acquisition quality, analysis accuracy, and intelligence. Existing ECG signal acquisition equipment typically utilizes fixed-parameter sampling methods, making it difficult to adapt to the varying signal characteristics of different patients and pathological conditions. This is particularly true when acquiring specialized signals such as intracardiac electrocardiograms, often leading to issues such as low signal-to-noise ratio and loss of detail. Furthermore, the Jiang's cable, a crucial signal transmission medium, has a significant impact on signal quality due to its connection method and parameter settings, but existing technologies lack effective means for intelligent optimization.

[0004] In terms of signal analysis, existing automated analysis algorithms are mostly based on traditional statistical methods, which struggle to effectively process complex, nonlinear ECG signals. While some systems are beginning to incorporate machine learning techniques, they are often limited to a single model and struggle to address diverse arrhythmia types. Furthermore, these systems generally lack adaptive learning capabilities and are unable to dynamically adjust analysis parameters based on individual patient characteristics, resulting in inconsistent accuracy in practical applications.

[0005] Another prominent issue is the lack of compatibility in existing systems. Medical institutions often use a variety of ECG devices from different manufacturers, but existing analysis systems often only support specific models, resulting in wasted resources and low efficiency. Furthermore, existing systems generally suffer from inadequate security and poor real-time performance when it comes to data storage and transmission, making them unable to meet the demands of modern medical information technology.

[0006] Given these challenges, a system and method are urgently needed that can intelligently acquire and analyze cardiac electrophysiological signals and optimize the performance of Jiang's cable. Such a system should be able to adaptively adjust acquisition parameters to improve signal quality, employ advanced deep learning algorithms to enhance analysis accuracy, possess good compatibility with a variety of devices, and ensure secure data storage and efficient transmission. Summary of the Invention

[0007] The present invention addresses these issues with existing technologies. It provides an intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization system and method, aiming to comprehensively improve ECG signal acquisition quality, analysis accuracy, and system compatibility, providing more reliable and efficient technical support for the diagnosis and treatment of heart disease.

[0008] The present invention proposes an intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization system, comprising:

[0009] Data acquisition module for:

[0010] Collect 12-lead electrocardiogram signals from the human body;

[0011] Collect intracardiac electrogram signals;

[0012] Perform pacing stimulation functions and collect corresponding signals, including sensing, threshold, and impedance;

[0013] The signal preprocessing module is electrically connected to the data acquisition module and is used to:

[0014] Receiving the electrocardiogram signal and intracardiac electrocardiogram signal sent by the data acquisition module;

[0015] Performing filtering on the electrocardiogram signal and the intracardiac electrogram signal to generate a preprocessed signal;

[0016] A feature extraction module is electrically connected to the signal preprocessing module and is used to:

[0017] Receiving the preprocessed signal;

[0018] Extracting time domain features, frequency domain features and nonlinear features from the preprocessed signal to generate a feature data set;

[0019] A deep learning module is electrically connected to the feature extraction module and is used to:

[0020] Receiving the feature data set;

[0021] Analyzing the feature dataset using a pre-trained deep neural network model to identify abnormal heart rhythms;

[0022] A parameter tuning module is electrically connected to the deep learning module and is used to:

[0023] Based on the abnormal heart rhythm recognition results, dynamically adjusting the parameters of the deep neural network model;

[0024] Adaptively optimize analysis parameters based on the ECG characteristics of different patients;

[0025] The Jiang's connecting line optimization module is electrically connected to the data acquisition module and is used to:

[0026] Based on the signal quality collected by the data acquisition module, the connection method of the Jiang's connecting line is optimized;

[0027] Adjust signal transmission parameters to improve signal stability and reliability;

[0028] An intelligent monitoring module is electrically connected to the deep learning module and the parameter tuning module, and is used to:

[0029] Automatically monitor QRS wave width;

[0030] Automatically monitor the damage current;

[0031] During pacing, the time interval from the pacing pin to the top of the QRS wave in leads V4-6 is automatically measured;

[0032] The display interface module is electrically connected to the intelligent monitoring module and is used to:

[0033] Simultaneously display surface ECG leads and intracardiac electrocardiogram on the same interface;

[0034] Use dual channels to display the intracavity potential under two filtering conditions;

[0035] A system compatible module is electrically connected to the data acquisition module and is used to:

[0036] Achieve compatible connection with a variety of ECG devices;

[0037] Connect the pacing analyzer and multi-conductivity physiology meter simultaneously through a Y-type converter or jumper and bridge wire;

[0038] The control module is electrically connected to all the above modules and is used to coordinate the work of each module and control the operation of the entire system.

[0039] Preferably, the feature extraction module comprises:

[0040] A time domain feature extraction unit, configured to extract a maximum value, a minimum value, a mean value, and a variance from the preprocessed signal;

[0041] A frequency domain feature extraction unit, configured to extract average power and average heart rate from the preprocessed signal;

[0042] A nonlinear feature extraction unit is used to extract approximate entropy and sample entropy from the preprocessed signal.

[0043] Preferably, the deep learning module includes:

[0044] Convolutional neural network (CNN) unit, used to process spatial features;

[0045] Recurrent neural network RNN ​​unit, used to process time series features;

[0046] Long short-term memory (LSTM) units, used to capture long-term dependencies;

[0047] The model integration unit is used to integrate the output results of multiple sub-models to improve recognition accuracy.

[0048] Preferably, the parameter tuning module includes:

[0049] Patient characteristic analysis unit, used to analyze the ECG characteristics of different patients;

[0050] A parameter adaptation unit, used to dynamically adjust the parameters of the deep neural network model based on patient characteristics;

[0051] Performance evaluation unit, used to evaluate the performance of the model after parameter adjustment;

[0052] Feedback optimization unit, used to further optimize parameters based on performance evaluation results.

[0053] Preferably, the Jiang's connection line optimization module includes:

[0054] A signal quality evaluation unit, used to evaluate the quality of the collected signal;

[0055] A connection mode optimization unit, used to optimize the connection mode of Jiang's connection line according to the signal quality evaluation result;

[0056] A transmission parameter adjustment unit, used to dynamically adjust signal transmission parameters;

[0057] Hardware adapter unit, used to optimize the matching between high-precision sensors and Jiang's connecting cables.

[0058] Preferably, the intelligent monitoring module comprises:

[0059] QRS wave analysis unit, used to automatically detect and measure the QRS wave width;

[0060] Damage current monitoring unit, used to monitor and quantify the damage current in real time;

[0061] The pacing response analysis unit is used to automatically measure and analyze the time interval from the pacing pin to the top of the QRS wave of leads V4-6 during pacing.

[0062] Preferably, the display interface module includes:

[0063] Multi-lead display unit, used to simultaneously display 12-lead electrocardiogram and intracardiac electrogram;

[0064] Dual-channel potential display unit, used to display low-frequency potential of 0.5~500.0 Hz and high-frequency potential of 30.0~500.0 Hz;

[0065] Dynamic update unit, used to update the displayed electrophysiological signals in real time;

[0066] Interactive operation unit, used to allow users to customize display content and format.

[0067] Preferably, the system compatibility module includes:

[0068] Device identification unit, used to automatically identify the type of connected ECG device;

[0069] Protocol conversion unit, used to convert between different communication protocols;

[0070] Data format adaptation unit, used to unify the data formats of different devices;

[0071] Real-time synchronization unit, used to ensure real-time synchronization of data between multiple devices.

[0072] As an advantage, it also includes:

[0073] Data storage module, used to store the collected original signals, processed data and analysis results;

[0074] Remote transmission module, used to securely transmit system data to a remote server or other medical devices;

[0075] AI-assisted diagnosis module, used to provide preliminary diagnostic suggestions based on deep learning results;

[0076] The system self-check module is used to regularly check the working status and performance of each module to ensure the reliable operation of the system.

[0077] The intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization method based on the system includes the following steps:

[0078] S1. Acquire cardiac electrophysiological signals, including 12-lead electrocardiogram (ECG) signals and intracardiac electrocardiogram (ICG) signals, and perform pacing stimulation and acquire corresponding signals;

[0079] S2. Preprocessing the collected signal, including filtering, to generate a preprocessed signal;

[0080] S3. Extract time domain features, frequency domain features and nonlinear features from the preprocessed signal to generate a feature data set;

[0081] S4. Use the pre-trained deep neural network model to analyze the feature dataset and identify abnormal heart rhythms;

[0082] S5. Based on the abnormal heart rhythm recognition results, dynamically adjust the parameters of the deep neural network model and adaptively optimize the analysis parameters according to the ECG characteristics of different patients;

[0083] S6. Based on the collected signal quality, optimize the connection method of Jiang's connecting line and adjust the signal transmission parameters;

[0084] S7. Automatically monitors QRS wave width and injury current, and automatically measures the time interval from the pacing pin to the top of the QRS wave in leads V4-6 during pacing;

[0085] S8. Display surface ECG leads and intracardiac electrocardiogram simultaneously on the same display interface, using dual channels to display intracardiac potentials under two filtering conditions;

[0086] S9. Achieve compatible connection with various ECG devices, and connect to pacemaker analyzers and multi-conductor physiology instruments simultaneously through Y-type converters or jumpers and bridge cables;

[0087] S10. Coordinate the execution of each step and control the progress of the entire analysis and optimization process.

[0088] The beneficial effects of the present invention are mainly reflected in the following aspects:

[0089] Through modular design and the application of deep learning technology, the system of the present invention achieves multiple innovations and breakthroughs. First, regarding signal acquisition, the data acquisition module of the present invention employs an adaptive sampling strategy, dynamically adjusting sampling parameters based on the physiological characteristics and pathological conditions of individual patients. This not only improves the signal-to-noise ratio but also captures more subtle pathological changes. In particular, for specialized signals such as intracardiac electrograms, the present invention utilizes a sampling rate of up to 4000Hz, significantly enhancing the fidelity of signal detail.

[0090] Secondly, the Jiang's cable optimization module significantly improves the stability and reliability of signal transmission by monitoring signal quality in real time and dynamically adjusting connection parameters. This innovation not only enhances the overall quality of signal acquisition but also reduces the risk of misdiagnosis due to poor connections. For example, in clinical practice, even with slight changes in patient movement or electrode position, the system can quickly adjust parameters to maintain signal stability, which is particularly important for long-term monitoring and Holter analysis.

[0091] In terms of signal analysis, the deep learning module of the present invention integrates multiple advanced neural network structures, including CNN, RNN, and LSTM, capable of comprehensively capturing the time, frequency, and nonlinear characteristics of ECG signals. This multi-model fusion approach significantly improves the system's ability to identify various complex arrhythmias. Of particular note, the attention mechanism introduced in the present invention enables the model to automatically focus on key components of the signal, further enhancing the accuracy and interpretability of the analysis.

[0092] The parameter tuning module of this invention utilizes a Bayesian optimization strategy, dynamically adjusting model parameters based on each patient's specific circumstances. This personalized analysis approach significantly improves the system's applicability and accuracy across diverse patient populations. For example, for elderly patients, the system might prioritize the detection of atrial fibrillation; for young athletes, it might prioritize the identification of exercise-induced arrhythmias.

[0093] The present invention's innovation is particularly significant in terms of system compatibility. The system compatibility module automatically identifies and adapts to a variety of ECG devices, enabling seamless connectivity between devices from different manufacturers and models. This significantly improves the utilization of medical resources and reduces equipment investment costs for hospitals. Furthermore, the present invention utilizes standardized data formats and communication protocols, facilitating future system expansion and upgrades.

[0094] Furthermore, this invention also achieves significant breakthroughs in data security and remote transmission. The data storage module utilizes a tiered storage and incremental backup strategy, ensuring both data security and improved storage efficiency. The remote transmission module utilizes end-to-end encryption technology, ensuring patient privacy and data security. It also supports resumable transmission and adapts to various network environments.

[0095] In summary, the intelligent cardiac electrophysiological signal analysis and Jiang's cable optimization system and method of this invention, through the synergistic effect of multiple innovative modules, optimizes the entire process from signal acquisition to analysis, from local processing to remote transmission. This not only significantly improves the accuracy and efficiency of ECG signal analysis, but also provides strong technical support for early diagnosis, precision treatment, and telemedicine of heart disease. The application of this invention is expected to significantly improve the diagnosis and treatment of cardiovascular diseases, reduce the medical burden, and ultimately provide patients with a better medical experience and health outcomes. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 It is a logic block diagram of the entire system of the present invention.

[0097] Figure 2 This is a logic block diagram of the feature extraction module of the present invention.

[0098] Figure 3This is a logical block diagram of the deep learning module of the present invention.

[0099] Figure 4 This is a logic block diagram of the parameter tuning module of the present invention.

[0100] Figure 5 This is a logic block diagram of the Jiang's connecting line optimization module of the present invention. DETAILED DESCRIPTION

[0101] See Figure 1-5 The present invention provides an intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization system and method. The system includes multiple functional modules that work together to achieve intelligent analysis of cardiac electrophysiological signals and optimization of Jiang's connection lines.

[0102] The system of the present invention includes a data acquisition module 1, a signal preprocessing module 2, a feature extraction module 3, a deep learning module 4, a parameter tuning module 5, a Jiang's connection line optimization module 6, an intelligent monitoring module 7, a display interface module 8, a system compatibility module 9, and a control module 10. These modules are electrically connected to each other to achieve data transmission and information exchange.

[0103] Data acquisition module 1 is primarily used to collect 12-lead ECG and intracardiac electrocardiogram (ICG) signals, as well as to perform pacing stimulation and collect corresponding signals. Preferably, the present invention utilizes high-precision sensors capable of capturing weak cardiac electrical signals. For example, for 12-lead ECG signal acquisition, an analog-to-digital converter with a sampling rate of 1000 Hz can be used to ensure capture of high-frequency components. For intracardiac electrocardiogram (ICG) signals, given their specific characteristics, the present invention utilizes a higher sampling rate, such as 4000 Hz, to capture more detailed information.

[0104] In the data acquisition module 1, the three parameters of sensing, threshold, and impedance are the core parameters of the pacemaker's operation. In the field of cardiac pacing technology, these are standard clinical parameters: sensing refers to the pacemaker's ability to detect the heart's spontaneous electrical activity; threshold refers to the minimum energy required to effectively stimulate the heart; impedance refers to the resistance between the electrode and the myocardium;

[0105] These parameters are basic functional indicators of the pacing system and are used to evaluate whether the electrode positioning is appropriate and the pacing effect is good or bad.

[0106] Signal preprocessing module 2 is electrically connected to data acquisition module 1 and is used to receive and preprocess various collected signals. The primary purpose of preprocessing is to remove noise and improve signal quality. The present invention employs a multi-stage filtering strategy, including a notch filter to remove power frequency interference and a bandpass filter to preserve useful frequency bands. For example, for electrocardiogram signals, a typical bandpass filter frequency range is 0.5 Hz to 150 Hz. Furthermore, the present invention incorporates wavelet transform denoising technology, which effectively removes high-frequency noise while preserving signal characteristics.

[0107] Feature extraction module 3 receives the preprocessed signal and extracts time-domain, frequency-domain, and nonlinear features from it to generate a feature dataset. Time-domain features include, but are not limited to, QRS duration and ST segment height; frequency-domain features, such as power spectral density, are primarily obtained through fast Fourier transforms (FFTs); and nonlinear features, such as sample entropy and approximate entropy, are used to capture the complex dynamics of the ECG signal.

[0108] Deep learning module 4 is a core component of the present invention. It uses a pretrained deep neural network model to analyze the feature dataset and identify abnormal heart rhythms. This invention utilizes an improved convolutional neural network (CNN) architecture, comprising multiple convolutional, pooling, and fully connected layers. The size of the convolution kernel is optimized based on the function of each layer. For example, the first layer uses a 5x5 kernel with a stride of 1 to capture local features; subsequent layers gradually reduce the kernel size to 3x3 to extract higher-level features. Furthermore, this invention incorporates an attention mechanism, enabling the model to automatically focus on key components of the signal.

[0109] The parameter tuning module 5 works closely with the deep learning module 4 to dynamically adjust the parameters of the deep neural network model based on the abnormal heart rhythm recognition results. This invention employs an adaptive learning rate strategy, with the initial learning rate set to 0.001 and then dynamically adjusted based on validation set performance. For example, if validation set accuracy does not improve over five consecutive epochs, the learning rate is reduced to 0.1 times the original value. Furthermore, this invention incorporates regularization techniques, such as L2 regularization with a coefficient set to 0.0001, to prevent model overfitting.

[0110] Another innovative feature of this invention is the Jiang's cable optimization module 6. Based on the signal quality collected by the data acquisition module 1, it optimizes the Jiang's cable connection and adjusts signal transmission parameters. This invention utilizes an adaptive impedance matching algorithm to adjust the cable impedance in real time to maximize signal transmission efficiency. Specifically, the system measures the signal-to-noise ratio (SNR) of the signal. When the SNR falls below 20dB, the impedance adjustment process is triggered. The adjustment step size is initially set to 0.1Ω and then dynamically adjusted based on the degree of SNR improvement.

[0111] Intelligent monitoring module 7 implements multiple automatic monitoring functions. For QRS width monitoring, the present invention utilizes a modified Pan-Tompkins algorithm, accurately locating the onset and endpoint of the QRS complex. Injury current monitoring is achieved by analyzing ST segment excursion, with the system automatically displaying the magnitude of the injury current. During pacing, the system automatically calculates the time interval by precisely locating the pacing pulse and the QRS complex apex.

[0112] The innovation of the display interface module 8 lies in its highly integrated display. On the same interface, the system simultaneously displays both surface ECG leads and intracardiac electrograms, using dual channels to display intracardiac potentials under two different filtering conditions. This design significantly improves physicians' work efficiency. For example, the low-frequency channel (0.5-500.0 Hz) can display the complete potential waveform, while the high-frequency channel (30.0-500.0 Hz) is more suitable for observing rapid potential changes.

[0113] System compatibility module 9 ensures compatibility with a variety of ECG devices. It intelligently identifies the type of connected device and automatically adjusts the communication protocol and data format. For example, for common 12-lead ECG machines, the system uses the standard HL7 format for data exchange; however, for specialized intracardiac electrocardiographic devices, proprietary protocols may be required. Furthermore, the design of a Y-type converter or patch cord allows the system to connect simultaneously to a pacemaker analyzer and a multi-conductor physiology meter, significantly improving device utilization.

[0114] Control module 10 serves as the system's core, coordinating the operations of various modules. It employs a state-machine-based control strategy, determining the next action based on the system's current state and external inputs. For example, if signal quality deteriorates during data acquisition, the control module immediately triggers Jiang's Link Optimization module 6 to improve signal quality.

[0115] The feature extraction module 3 of the present invention includes a time-domain feature extraction unit 31, a frequency-domain feature extraction unit 32, and a nonlinear feature extraction unit 33. The time-domain feature extraction unit 31 primarily focuses on signal characteristics in the time dimension, such as maximum, minimum, mean, and variance. These features can intuitively reflect the basic characteristics of the ECG signal. For example, the maximum value of the QRS wave is typically between 1 and 2 mV; if it exceeds this range, it may indicate myocardial hypertrophy or ventricular hypertrophy. The frequency-domain feature extraction unit 32 converts the signal to the frequency domain using a fast Fourier transform (FFT) and extracts features such as average power and average heart rate. These features can reveal periodic components in the signal and are very useful for identifying certain arrhythmias. The nonlinear feature extraction unit 33 calculates approximate entropy and sample entropy. These features can reflect the complexity and irregularity of the ECG signal and are important for identifying complex arrhythmias such as atrial fibrillation.

[0116] The present invention discloses the structure of a deep learning module 4, comprising a convolutional neural network (CNN) unit 41, a recurrent neural network (RNN) unit 42, a long short-term memory (LSTM) unit 43, and a model integration unit 44. CNN unit 41 primarily processes spatial features and comprises multiple convolutional and pooling layers. For example, the first convolutional layer can use 32 3x3 convolution kernels with a stride of 1, "same" padding, and a Reluctant Unit (ReLU) activation function. This configuration effectively captures local features. RNN unit 42 and LSTM unit 43 focus on processing temporal features. RNN is suitable for processing short-term dependencies, while LSTM is capable of capturing long-term dependencies. For example, an LSTM unit can have 128 hidden units to balance computational complexity and model performance. Model integration unit 44 employs a weighted averaging strategy, combining the outputs of the CNN, RNN, and LSTM to produce the final prediction result. The initial weights can be set to (0.4, 0.3, 0.3), and then optimized using a validation set.

[0117] The detailed description above demonstrates the innovative and advanced nature of the system presented in this paper in numerous aspects. It not only enables high-precision acquisition and intelligent analysis of cardiac electrophysiological signals, but also adaptively optimizes hardware connections to enhance signal quality. This comprehensive solution is crucial for improving diagnostic accuracy and treatment outcomes for heart disease.

[0118] The present invention discloses the structure and functions of parameter tuning module 5. Parameter tuning module 5 includes patient characteristic analysis unit 51, parameter adaptation unit 52, performance evaluation unit 53, and feedback optimization unit 54. These units work together to achieve adaptive optimization of the system for different patients.

[0119] The primary task of the patient feature analysis unit 51 is to analyze the ECG characteristics of different patients. The present invention utilizes a multidimensional feature extraction method that not only considers traditional ECG features such as QRS morphology and ST segment changes, but also incorporates personal information such as age, gender, and medical history as auxiliary features. Preferably, the present invention uses principal component analysis (PCA) to reduce the dimensionality of these features to extract the most representative feature combinations. For example, the system may retain the principal components that explain 95% of the variance, which typically corresponds to 3-5 principal components.

[0120] The parameter adaptation unit 52 dynamically adjusts the parameters of the deep neural network model based on patient characteristics. The present invention employs a hyperparameter tuning strategy based on Bayesian optimization. Specifically, the system defines a prior distribution for each key hyperparameter. For example, the learning rate might follow a log-uniform distribution (log-uniform(1e-4, 1e-1)), and the batch size might follow a discrete uniform distribution (discrete_uniform(16, 128)). The system then continuously updates the posterior distribution of these hyperparameters through an iterative optimization process to find the optimal parameter combination.

[0121] The performance evaluation unit 53 is responsible for evaluating the performance of the model after parameter adjustment. This invention utilizes a multi-metric comprehensive evaluation method, including accuracy, sensitivity, specificity, and F1 score. Preferably, the system assigns different weights to these metrics. For example, the accuracy weight can be set to 0.4, sensitivity and specificity to 0.3 each, and the F1 score to 0.2. This weighting strategy can more comprehensively reflect the model's performance in all aspects.

[0122] Feedback optimization unit 54 further optimizes parameters based on the performance evaluation results. This invention incorporates a progressive optimization strategy: after each optimization, the system compares the performance of the new and old models. If the new model's overall score improves by less than 1%, the system reduces the parameter adjustment step size. If there is no significant improvement after three consecutive optimizations, the system considers early stopping to prevent overfitting.

[0123] The present invention discloses the structure of the Jiang's connection line optimization module 6, which includes a signal quality assessment unit 61, a connection mode optimization unit 62, a transmission parameter adjustment unit 63, and a hardware adaptation unit 64. These units work together to ensure high-quality acquisition and transmission of ECG signals.

[0124] Signal quality assessment unit 61 utilizes a multi-dimensional assessment method. This method not only considers the traditional signal-to-noise ratio (SNR) metric, but also incorporates factors such as baseline drift and myoelectric interference intensity. Preferably, the system uses fuzzy logic to integrate these metrics and derive a signal quality score from 0 to 100. For example, when the SNR is above 40dB, the baseline drift is less than 0.1mV, and there is no significant myoelectric interference, the signal quality score may be above 90.

[0125] Regarding the scoring system of the signal quality assessment unit, the 0-100 point scoring system is actually based on the following specific criteria:

[0126] 0-20 points: The signal quality is extremely poor and the connection needs to be adjusted or the electrode needs to be replaced immediately (SNR < 10dB, baseline drift > 0.5mV);

[0127] 21-40 points: poor signal quality, which may lead to misdiagnosis (SNR is 10-20dB, baseline drift is 0.3-0.5mV);

[0128] 41-60 points: Average signal quality, suitable for preliminary analysis (SNR 20-30dB, baseline drift 0.2-0.3mV);

[0129] 61-80 points: Good signal quality, suitable for accurate analysis (SNR 30-40dB, baseline drift 0.1-0.2mV);

[0130] 81-100 points: Excellent signal quality, suitable for high-precision analysis (SNR>40dB, baseline drift<0.1mV);

[0131] This hierarchical scoring system allows the system to automatically decide whether to trigger Jiang's connection line optimization, for example, automatically starting the optimization process when the score is below 40 points.

[0132] The connection optimization unit 62 optimizes the connection method of the Jiang's cable based on the signal quality assessment results. The present invention employs an adaptive connection strategy that automatically adjusts electrode position and pressure based on different situations. For example, if the signal quality of a particular lead is significantly lower than that of other leads, the system will prompt an adjustment of the electrode position for that lead. Preferably, the system is also equipped with a pressure sensor that can monitor the contact pressure between the electrode and the skin in real time and provide adjustment recommendations.

[0133] The transmission parameter adjustment unit 63 is responsible for dynamically adjusting signal transmission parameters. This invention introduces an adaptive sampling rate adjustment algorithm that dynamically adjusts the sampling rate based on the signal's spectral characteristics. For example, when an increase in high-frequency components is detected (such as during ventricular tachycardia), the system automatically increases the sampling rate to 2000Hz or higher to capture more detail. When the signal is relatively stable, the sampling rate may be reduced to 500Hz to conserve storage space.

[0134] The primary task of hardware adaptation unit 64 is to optimize the matching between the high-precision sensor and the Jiang's cable. This invention utilizes intelligent impedance matching technology that adjusts the sensor's input impedance in real time to maximize signal transmission efficiency. Preferably, the system adjusts the input impedance within a range of 0.1-10 MΩ, in 0.1 MΩ increments, until the setting that optimizes signal quality is found.

[0135] The present invention discloses the structure of an intelligent monitoring module 7, which includes a QRS wave analysis unit 71, an injury current monitoring unit 72, and a pacing response analysis unit 73. These units together realize intelligent monitoring of key electrocardiogram indicators.

[0136] The QRS analysis unit 71 uses a modified Pan-Tompkins algorithm to automatically detect and measure QRS width. This invention incorporates a wavelet transform preprocessing step based on the traditional algorithm, improving the algorithm's robustness in noisy environments. Preferably, the system calculates the QRS width for 100 consecutive heartbeats and reports the median and interquartile range to provide more stable and representative measurement results.

[0137] The injury current monitoring unit 72 enables real-time monitoring and quantification of the injury current. The present invention employs an injury current assessment method based on ST segment analysis. The system calculates the ST segment offset relative to the baseline in real time and determines the severity of the injury based on the offset.

[0138] During pacing, the pacing response analysis unit 73 automatically measures and analyzes the time interval from the pacing pin to the top of the QRS wave in leads V4-6. This invention introduces a high-precision time measurement algorithm that combines threshold detection and interpolation techniques to improve time measurement accuracy to 0.1 ms. Preferably, the system measures 100 pacing responses continuously and reports the mean and standard deviation to provide a more comprehensive assessment of pacing effectiveness.

[0139] The present invention discloses the detailed structure of the display interface module 8, which includes a multi-lead display unit 81, a dual-channel potential display unit 82, a dynamic update unit 83, and an interactive operation unit 84. These units together constitute a powerful and user-friendly display interface.

[0140] The multi-lead display unit 81 enables simultaneous display of a 12-lead electrocardiogram (ECG) and an intracardiac electrogram (ICG). The present invention utilizes an intelligent layout algorithm that automatically adjusts the size and position of each lead graphic based on screen resolution and user preferences. Preferably, the system provides multiple preset layout options, such as "3x4+1" (displaying the 12-lead ECG in 3 rows and 4 columns, with the ICG displayed separately) and "6+6+1" (displaying the first 6 leads and the last 6 leads in two groups, with the ICG displayed separately).

[0141] The dual-channel potential display unit 82 is used to display low-frequency potentials from 0.5 to 500 Hz and high-frequency potentials from 30.0 to 500 Hz. This invention introduces a color coding technique that uses different colors to distinguish low-frequency and high-frequency components, improving information readability. Preferably, the system also provides a spectrum overlay display mode that can intuitively display the energy distribution of the signal in different frequency bands.

[0142] The filtering process in the system is for specific analysis purposes, rather than a permanent change to all data. The actual implementation is as follows:

[0143] The bandpass filter (0.5Hz-150Hz) mentioned in the signal preprocessing module is mainly used in the abnormal rhythm analysis and processing path of the 12-lead ECG signal;

[0144] For intracardiac electrogram signals, the system retains complete original data (up to 4000Hz sampling rate);

[0145] The dual-channel display function of the display interface module actually accesses the raw data, allowing users to view the signal characteristics of different frequency bands:

[0146] The low-frequency channel (0.5-500.0 Hz) is used to observe the basic waveforms of cardiac depolarization and repolarization;

[0147] The high-frequency channel (30.0~500.0Hz) is used to observe special high-frequency components such as His bundle potential and left bundle branch potential;

[0148] This design enables physicians to view information from different frequency bands simultaneously, especially in left bundle branch pacing, where high-frequency components are crucial for identifying bundle branch potentials.

[0149] Dynamic update unit 83 is responsible for updating the displayed electrophysiological signals in real time. The present invention utilizes an efficient data stream processing algorithm to ensure smooth display even at high sampling rates (e.g., 1000 Hz). Preferably, the system provides an adjustable update frequency setting, allowing users to select a refresh rate of 20 Hz, 30 Hz, or 60 Hz as needed.

[0150] Interactive operation unit 84 allows users to customize the display content and format. This invention features an intuitive interactive interface that allows users to freely adjust the display position and size of each lead through operations such as dragging, zooming, and scaling. Preferably, the system also provides a custom markup function, allowing users to add annotations or measurement marks to waveform locations of interest.

[0151] The detailed description above demonstrates the innovative and advanced nature of the system in various aspects, including parameter tuning, signal optimization, intelligent monitoring, and user interface. These innovations not only enhance the accuracy and efficiency of ECG signal analysis but also significantly improve the user experience, providing strong technical support for the diagnosis and treatment of heart disease.

[0152] The system compatibility module 9 of the present invention comprises a device identification unit 91, a protocol conversion unit 92, a data format adaptation unit 93, and a real-time synchronization unit 94. These units work together to ensure seamless compatibility between the system of the present invention and a variety of ECG devices.

[0153] Device identification unit 91 utilizes an intelligent device identification algorithm. This invention automatically identifies the device type by analyzing the electrical characteristics and communication protocols of connected devices. Preferably, the system establishes a device feature database containing characteristic information for common ECG devices on the market. When a new device is connected, the system matches its features with the information in the database for rapid and accurate identification. For example, for a common 12-lead ECG machine, the system might detect its unique 9600 baud serial communication port and specific data frame format, allowing rapid identification of the device model.

[0154] Protocol conversion unit 92 implements conversion between different communication protocols. The present invention utilizes a modular protocol conversion architecture that can flexibly handle a variety of communication protocols. Preferably, the system supports multiple standard medical data exchange protocols, including HL7, DICOM, and SCP-ECG, as well as some vendor-proprietary protocols. For example, when the system receives data in HL7 format, it can convert it to DICOM format as needed to facilitate data exchange with other medical imaging devices. During the conversion process, the system automatically handles data field mapping and unit conversion to ensure information integrity and consistency.

[0155] The data format adaptation unit 93 is responsible for unifying the data formats of different devices. The present invention designs a universal internal data representation format that is compatible with various external data formats. Preferably, the system uses an XML-based data structure with good scalability and readability. For example, for electrocardiogram data, the system may use the following structure:

[0156] ```xml

[0157] <ecgdata>

[0158] <patientinfo>

[0159] <id> 12345< / id>

[0160] <name> John Doe< / name>

[0161] <age> 45< / age>

[0162] < / patientinfo>

[0163] <recordinginfo>

[0164] <date> 2025-01-11< / date>

[0165] <time>14:30:00< / time>

[0166] <devicetype> 12-Lead ECG< / devicetype>

[0167] < / recordinginfo>

[0168] <waveforms>

[0169] <lead name="I">

[0170] <datapoint> 0.1< / datapoint>

[0171] <datapoint> 0.2< / datapoint>

[0172] <!-- More data points -->

[0173] < / lead>

[0174] <!-- More leads -->

[0175] < / waveforms>

[0176] < / ecgdata>

[0177] ```

[0178] Real-time synchronization unit 94 ensures real-time data synchronization between multiple devices. This invention utilizes a timestamp-based synchronization mechanism that precisely aligns data streams from different devices. Preferably, the system uses the Network Time Protocol (NTP) to synchronize with a standard time server, achieving millisecond-level synchronization accuracy. For example, when an electrocardiograph and a pacemaker are used simultaneously, the system accurately aligns the pacing pulse with the corresponding electrocardiogram waveform based on timestamp information, facilitating comprehensive analysis by the physician.

[0179] The present invention further expands the functionality of the system by adding a data storage module 11, a remote transmission module 12, an artificial intelligence-assisted diagnosis module 13, and a system self-test module 14. The addition of these modules greatly enhances the overall performance and practicality of the system.

[0180] The data storage module 11 employs an efficient tiered storage strategy. This invention divides data into hot data (recently collected data) and cold data (historical data), storing them in high-speed solid-state drives and high-capacity mechanical hard drives, respectively. Preferably, the system employs incremental backup and data compression technologies to ensure data security while improving storage efficiency. For example, for electrocardiogram data, the system may employ lossless compression algorithms, such as Huffman coding or run-length encoding, to increase data compression rates to over 50% without any information loss.

[0181] The remote transmission module 12 enables secure remote transmission of system data. This invention utilizes end-to-end encryption technology to ensure data security during transmission. Preferably, the system uses the AES-256 encryption algorithm for data encryption and the RSA-2048 algorithm for key exchange. Furthermore, the system implements a resumable transmission function, ensuring complete data transmission even in unstable network conditions. For example, when transmitting large amounts of ECG data, if the network is interrupted, the system automatically records the amount of data transmitted and resumes transmission from the point of interruption when the network is restored.

[0182] The AI-assisted diagnosis module 13 provides preliminary diagnostic recommendations based on deep learning results. This invention utilizes a multi-model integration approach, comprehensively considering the diagnostic results of multiple specialized models. Preferably, the system integrates specialized models for arrhythmia identification, right ventricular septal pacing, deep septal pacing, non-selective left bundle branch pacemaker, left bundle branch pacing, His bundle / left bundle branch potentials, and their lesion currents. The outputs of these models are weighted and voted together to produce the final diagnostic recommendation.

[0183] The system self-test module 14 regularly checks the operating status and performance of each module. This invention incorporates a comprehensive self-test mechanism, including hardware status checks, software functionality testing, and performance evaluation. Preferably, the system performs a comprehensive self-test daily and a quick self-test upon each power-up. Self-test results are logged, and prompts are issued if an anomaly is detected. For example, if the self-test detects that the signal quality of a particular lead is consistently below a threshold, the system will indicate that the electrode may need to be replaced or the lead connection checked.

[0184] Finally, the present invention also discloses an intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization method. This method includes a series of steps, from signal acquisition to data analysis, to result display and device compatibility processing, forming a complete workflow.

[0185] In step S1, the method of the present invention acquires multiple cardiac electrophysiological signals. Preferably, the system uses high-precision sampling technology, employing a 1000Hz sampling rate for 12-lead electrocardiogram signals and a higher 4000Hz sampling rate for intracardiac electrocardiogram signals to capture more detailed information. When performing pacing stimulation, the system precisely controls the amplitude and duration of the stimulation pulses. For example, a stimulation intensity of 0.5-10mA and a pulse width of 0.1-2ms can be set.

[0186] Regarding the sampling rate and filtering issues, this involves the multi-path data processing design of the system:

[0187] The high sampling rate of 4000Hz is specifically designed to capture high-frequency potential components in intracardiac electrograms, especially the His bundle potential and left bundle branch potential, which have high-frequency characteristics (up to 1-2kHz).

[0188] The system adopts a multi-channel parallel processing architecture:

[0189] One processing channel uses bandpass filtering (0.5 Hz-150 Hz) to handle routine heart rhythm analysis;

[0190] Another dedicated channel retains high frequency components (up to 2kHz) for special potential analysis

[0191] The third channel holds the raw data for display and later analysis;

[0192] This multi-channel design ensures that the system can perform both conventional heart rhythm analysis and capture and analyze high-frequency potential characteristics without losing important information due to filtering processing in one channel.

[0193] Through this design, the system achieves multifunctional integration without technical contradictions, meeting the needs of different clinical scenarios, especially for left bundle branch pacing, which requires attention to both routine ECG activity and special high-frequency potentials.

[0194] Steps S2-S4 involve signal preprocessing, feature extraction, and deep learning analysis. The method of the present invention employs a multi-stage filtering strategy, including notch filtering, bandpass filtering, and wavelet denoising. During the feature extraction stage, the system not only extracts traditional time-domain and frequency-domain features but also calculates a series of nonlinear features, such as approximate entropy and sample entropy. Deep learning analysis utilizes an improved convolutional neural network architecture and introduces an attention mechanism, increasing the model's sensitivity to key features.

[0195] Steps S5-S7 implement parameter tuning, Jiang's patch cord optimization, and intelligent monitoring. The method of the present invention uses a Bayesian optimization strategy to dynamically adjust model parameters and employs adaptive impedance matching technology to optimize the Jiang's patch cord connection. Regarding intelligent monitoring, the system can automatically detect QRS wave width, monitor injury current magnitude, and accurately measure time intervals during pacing.

[0196] Steps S8-S10 involve display interface design, device compatibility, and overall control. The method of this invention designs an intuitive multi-lead display interface that supports dual-channel potential display. Regarding device compatibility, the system can automatically identify and adapt to different types of ECG devices. Overall control utilizes a state machine-based strategy to ensure coordinated operation of each step.

[0197] The detailed description above demonstrates the innovative and advanced nature of the system and method of this invention in numerous aspects. From device compatibility to data processing and AI-assisted diagnosis, this invention provides a comprehensive, intelligent, and efficient solution for cardiac electrophysiological signal analysis. This not only improves the accuracy and efficiency of ECG signal analysis but also provides strong technical support for the early diagnosis and precise treatment of heart disease.

[0198] It should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. Intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization system, characterized by ,include: Data acquisition module for: Collect 12-lead electrocardiogram signals from the human body; Collect intracardiac electrogram signals; Perform pacing stimulation functions and collect corresponding signals, including sensing, threshold, and impedance; The signal preprocessing module is electrically connected to the data acquisition module and is used to: Receiving the electrocardiogram signal and intracardiac electrocardiogram signal sent by the data acquisition module; Performing filtering on the electrocardiogram signal and the intracardiac electrogram signal to generate a preprocessed signal; A feature extraction module is electrically connected to the signal preprocessing module and is used to: Receiving the preprocessed signal; Extracting time domain features, frequency domain features and nonlinear features from the preprocessed signal to generate a feature data set; A deep learning module is electrically connected to the feature extraction module and is used to: Receiving the feature data set; Analyzing the feature dataset using a pre-trained deep neural network model to identify abnormal heart rhythms; A parameter tuning module is electrically connected to the deep learning module and is used to: Based on the abnormal heart rhythm recognition results, dynamically adjusting the parameters of the deep neural network model; Adaptively optimize analysis parameters based on the ECG characteristics of different patients; The Jiang's connecting line optimization module is electrically connected to the data acquisition module and is used to: Based on the signal quality collected by the data acquisition module, the connection method of the Jiang's connecting line is optimized; Adjust signal transmission parameters to improve signal stability and reliability; An intelligent monitoring module is electrically connected to the deep learning module and the parameter tuning module, and is used to: Automatically monitor QRS wave width; Automatically monitor the damage current; During pacing, the time interval from the pacing pin to the top of the QRS wave in leads V4-6 is automatically measured; The display interface module is electrically connected to the intelligent monitoring module and is used to: Simultaneously display surface ECG leads and intracardiac electrocardiogram on the same interface; Use dual channels to display the intracavity potential under two filtering conditions; A system compatible module is electrically connected to the data acquisition module and is used to: Achieve compatible connection with a variety of ECG devices; Connect the pacemaker analyzer and multi-conductor physiology meter simultaneously through a Y-type converter or jumper and bridge wire; The control module is electrically connected to all the above modules and is used to coordinate the work of each module and control the operation of the entire system.

2. The system according to claim 1, characterized in that , the feature extraction module includes: A time domain feature extraction unit, configured to extract a maximum value, a minimum value, a mean value, and a variance from the preprocessed signal; A frequency domain feature extraction unit, configured to extract average power and average heart rate from the preprocessed signal; A nonlinear feature extraction unit is used to extract approximate entropy and sample entropy from the preprocessed signal.

3. The system according to claim 1, characterized in that , the deep learning module includes: Convolutional neural network (CNN) unit, used to process spatial features; Recurrent neural network RNN ​​unit, used to process time series features; Long short-term memory (LSTM) units, used to capture long-term dependencies; The model integration unit is used to integrate the output results of multiple sub-models to improve recognition accuracy.

4. The system according to claim 1, characterized in that , the parameter tuning module includes: Patient characteristic analysis unit, used to analyze the ECG characteristics of different patients; A parameter adaptation unit, used to dynamically adjust the parameters of the deep neural network model based on patient characteristics; Performance evaluation unit, used to evaluate the performance of the model after parameter adjustment; Feedback optimization unit, used to further optimize parameters based on performance evaluation results.

5. The system according to claim 1, characterized in that , the Jiang's connection line optimization module includes: A signal quality evaluation unit, used to evaluate the quality of the collected signal; A connection mode optimization unit, used to optimize the connection mode of Jiang's connection line according to the signal quality evaluation result; A transmission parameter adjustment unit, used to dynamically adjust signal transmission parameters; Hardware adapter unit, used to optimize the matching between high-precision sensors and Jiang's connecting cables.

6. The system according to claim 1, characterized in that , the intelligent monitoring module includes: QRS wave analysis unit, used to automatically detect and measure the QRS wave width; Damage current monitoring unit, used to monitor and quantify the damage current in real time; The pacing response analysis unit is used to automatically measure and analyze the time interval from the pacing pin to the top of the QRS wave of leads V4-6 during pacing.

7. The system according to claim 1, characterized in that , the display interface module includes: Multi-lead display unit, used to simultaneously display 12-lead electrocardiogram and intracardiac electrogram; Dual-channel potential display unit, used to display low-frequency potential of 0.5~500.0 Hz and high-frequency potential of 30.0~500.0 Hz; Dynamic update unit, used to update the displayed electrophysiological signals in real time; Interactive operation unit, used to allow users to customize display content and format.

8. The system according to claim 1, characterized in that , the system compatible module includes: Device identification unit, used to automatically identify the type of connected ECG device; Protocol conversion unit, used to convert between different communication protocols; Data format adaptation unit, used to unify the data formats of different devices; Real-time synchronization unit, used to ensure real-time synchronization of data between multiple devices.

9. The system according to claim 1, characterized in that , also includes: Data storage module, used to store the collected original signals, processed data and analysis results; Remote transmission module, used to securely transmit system data to a remote server or other medical devices; AI-assisted diagnosis module, used to provide preliminary diagnostic suggestions based on deep learning results; The system self-check module is used to regularly check the working status and performance of each module to ensure the reliable operation of the system.

10. An intelligent cardiac electrophysiological signal analysis and Jiang's connection line optimization method based on the system according to any one of claims 1 to 9, characterized in that , including the following steps: S1. Acquire cardiac electrophysiological signals, including 12-lead electrocardiogram (ECG) signals and intracardiac electrocardiogram (ICG) signals, and perform pacing stimulation and acquire corresponding signals; S2. Preprocessing the collected signal, including filtering, to generate a preprocessed signal; S3. Extract time domain features, frequency domain features and nonlinear features from the preprocessed signal to generate a feature data set; S4. Use the pre-trained deep neural network model to analyze the feature dataset and identify abnormal heart rhythms; S5. Based on the abnormal heart rhythm recognition results, dynamically adjust the parameters of the deep neural network model and adaptively optimize the analysis parameters according to the ECG characteristics of different patients; S6. Based on the collected signal quality, optimize the connection method of Jiang's connecting line and adjust the signal transmission parameters; S7. Automatically monitors QRS wave width and injury current, and automatically measures the time interval from the pacing pin to the top of the QRS wave in leads V4-6 during pacing; S8. Simultaneously display the surface ECG leads and the intracardiac electrogram on the same display interface, using dual channels to display the intracardiac potentials under two filtering conditions; S9. Achieve compatible connection with various ECG devices, and connect to pacemaker analyzers and multi-conductor physiology instruments simultaneously through Y-type converters or jumpers and bridge cables; S10. Coordinate the execution of each step and control the progress of the entire analysis and optimization process.

Images