Heart disease risk early warning system based on multi-dimensional magnetocardiogram characteristic analysis

Abstract

The invention provides a heart disease risk early warning system based on multi-dimensional magnetocardiogram characteristic analysis. The heart disease risk early warning system comprises a magnetocardiogram signal sensor array, a signal conditioning module, a data acquisition module, a magnetic field correction module, a central processing module and an early warning output module. The sensor array collects an original signal of a heart magnetic field of a measured object, and the original signal is converted into a digital signal by the collection module after being subjected to noise reduction and amplification by the conditioning module. The central processing module is used for setting multi-dimensional magnetocardiogram characteristic parameters through time domain, frequency domain and space domain analysis, and calculating a heart disease risk index and regulating and controlling a dynamic threshold by combining the magnetocardiogram characteristic parameters compensated by the magnetic field correction module and matching deviation degree analysis on the basis of a dynamic risk assessment model constructed by machine learning. When the risk index exceeds a self-adaptive adjustment threshold value, the system triggers a multi-mode early warning response including abnormal feature marking, graded voice alarm and encrypted medical report automatic transmission. According to the invention, non-invasive and high-precision early warning of heart diseases can be realized, and sensitivity and accuracy of heart disease monitoring are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical disease monitoring, and more specifically, to a heart disease risk early warning system based on magnetocardiographic signal analysis. Background Art

[0002] Magnetocardiographic signals, biomagnetic field signals generated by cardiac electrical activity, contain a wealth of information about cardiac physiology and pathology. Their detection offers advantages such as non-invasiveness, high sensitivity, and high spatial resolution, demonstrating unique value in the field of heart disease monitoring. However, current heart disease risk assessment relies primarily on traditional techniques such as electrocardiograms (ECGs). These methods, limited by tissue impedance, struggle to capture subtle changes in cardiac function, particularly in the areas of early lesion identification and risk warning.

[0003] Existing monitoring systems based on magnetic cardiograms (MCGs) face multiple technical bottlenecks in practical applications: In signal acquisition, traditional sensor arrays are susceptible to interference from ambient magnetic fields, resulting in a low signal-to-noise ratio (SNR) in the original signal. In signal processing, conventional algorithms struggle to effectively integrate multi-scale frequency domain features with time domain morphological parameters, and they also struggle to adequately analyze the dynamic characteristics of two-dimensional MCG distributions. In risk assessment, static threshold models are unable to adapt to measurement biases caused by individual differences and changes in body position. At the system interaction level, the lack of real-time visualization and a graded warning mechanism compromises clinical practical value. These issues have severely restricted the widespread application of MCG technology in heart disease risk assessment.

[0004] In view of the above problems, the existing technology is in urgent need of improvement. Summary of the Invention

[0005] The present invention provides a heart disease risk warning system based on multi-dimensional magnetic heart feature analysis, comprising:

[0006] Magnetocardiographic signal sensor array, signal conditioning module, data acquisition module, central processing module, and early warning output module;

[0007] The magnetic heart signal sensor array is used to collect the original signal of the cardiac magnetic field, and the object to be measured is located in the detection area of ​​the magnetic heart signal sensor array; the signal conditioning module is used to reduce noise and amplify the original signal; the data acquisition module is used to convert the analog signal into a digital magnetic heart signal; the central processing module is used to set multi-dimensional magnetic heart characteristic parameters according to the time domain characteristics, frequency domain characteristics and spatial domain characteristics of the magnetic heart signal, extract each magnetic heart characteristic parameter through the magnetic heart signal processing algorithm, and realize the calculation of the heart disease risk index and dynamic threshold control based on the dynamic risk assessment model constructed by machine learning, combined with the magnetic heart characteristic parameters compensated by the magnetic field correction module and the matching deviation analysis; when the risk index exceeds the dynamic risk threshold, the multimodal early warning output module is triggered.

[0008] Furthermore, in the extraction of cardiac magnetic characteristic parameters, the central processing module is used to pre-process and recognize the waveform of the cardiac magnetic signal, extract time domain morphological indicators including RR interval variation coefficient, QT dispersion, ST segment offset, T wave asymmetry, R peak time difference, etc., to achieve a preliminary assessment of the stability and conduction characteristics of cardiac electrical activity; on this basis, the cardiac magnetic signal is further decomposed by wavelet, the multi-scale energy distribution characteristics within the QRS complex time window are extracted, and the low-frequency, high-frequency and ultra-high-frequency energy proportions and spectral entropy in the P wave, QRS complex and T wave components are calculated. , center frequency and other frequency domain parameters, thereby completing the detailed frequency domain characterization of the cardiac electromagnetic activity; at the same time, based on the two-dimensional current density distribution map of the cardiac magnetostrictive signal, the dynamic characteristics such as the total current vector, the maximum current vector and its spatial coordinates, current angle, current vector rotation and divergence in different phases of the cardiac cycle are analyzed to obtain the cardiac magnetostrictive source activity information with spatial positioning capability; finally, the above-mentioned time domain indicators, frequency domain characteristics and the derived parameters of the two-dimensional current density map are combined to form a multi-dimensional cardiac magnetostrictive feature set covering the time domain, frequency domain and space domain, and are sent to the risk assessment model as the core input parameters.

[0009] Furthermore, in the construction of the risk assessment model, the central processing module is used to receive the cardiac magnetic characteristic parameters and corresponding clinical diagnosis labels in the historical cardiac magnetic database, and train the risk classifier through a machine learning algorithm; the central processing module is also used to calculate in real time the deviation between the cardiac magnetic characteristic parameters of the subject and the matching degree of the risk classifier, and generate a primary risk flag when the deviation continues to exceed a preset time.

[0010] Furthermore, in the dynamic risk threshold update, the central processing module is used to automatically activate the magnetic field correction module according to the body position change signal of the subject; the magnetic field correction module is used to reconstruct the spatial posture matrix of the magnetocardiographic signal sensor array based on the acceleration sensor data, and eliminate the drift error in the two-dimensional plane caused by the body position change through orthogonal projection compensation; the central processing module dynamically adjusts the risk threshold according to the compensated magnetocardiographic characteristic parameters.

[0011] Furthermore, the magnetocardiographic signal sensor array is composed of multiple superconducting quantum interference devices or optically pumped magnetometers, and each channel is arranged in a matrix; the signal conditioning module includes a bandpass filter and a phase-locked amplifier.

[0012] Furthermore, the risk assessment model uses the cardiomagnetic risk index calculation formula: ; in: is the heart disease risk index; 、 、 They constitute the linear combination of time domain, frequency domain and space domain features respectively; is the deviation variable based on the classifier matching degree; is the weight coefficient of each feature and satisfies .

[0015] Furthermore, the weight coefficient is determined by clinical data regression. For any weight coefficient λ∈{α,β,γ,δ}, the calculation formula is:

[0016]

[0017] Where: N is the total number of clinical samples; F s,λ Indicates the feature combination value of the sth sample in the corresponding dimension (S t,s 、S f,s 、S s,s or D s );R s is the risk level value of the sth sample confirmed by the gold standard; and are the sample means of the corresponding features.

[0018] Furthermore, the posture compensation formula of the magnetic field correction module is:

[0019] B cor =T(θ x ,θ y )·B raw +ΔB

[0020] Among them: B cor is the magnetic field vector after correction; B raw is the original magnetic field vector; T is the rotation compensation matrix calculated from the acceleration sensor data; θ x ,θ y is the rotation angle around the x-axis and y-axis; ΔB is the ambient magnetic field reference offset.

[0021] Furthermore, the system also includes a user interaction module; the user interaction module includes a touch screen and a voice prompt unit; the touch screen is used to display the heart disease risk index and the magnetic heart characteristic parameter curve in real time; the voice prompt unit is used to broadcast the risk level information when the early warning output module is triggered.

[0022] Furthermore, the central processing module and the user interaction module are linked to control: when the heart disease risk index exceeds the dynamic risk threshold, the following operations are synchronously performed:

[0023] Freeze the magnetic cardiology characteristic parameter curve on the touch screen and mark the abnormal timestamp;

[0024] Drive the voice prompt unit to broadcast the third-level risk warning voice;

[0025] Generate an encrypted risk report and send it to the designated medical terminal via the wireless transmission module.

[0026] The above embodiments of the present invention have at least the following beneficial effects:

[0027] 1. By combining the cardiac magnetoelectric signal sensor array with signal conditioning and data acquisition modules, high-sensitivity acquisition and high-fidelity conversion of cardiac magnetic field signals are achieved, solving the problems of high tissue impedance and susceptibility to interference in traditional ECG detection, and providing a more comprehensive bio-electromagnetic information basis for heart disease risk warning.

[0028] 2. The use of time-frequency domain fusion analysis to extract multi-dimensional features and combine it with two-dimensional current density distribution can accurately capture the time-frequency-space dynamic characteristics of cardiac electrical activity, overcome the defect of insufficient sensitivity of single time domain or frequency domain analysis methods in detecting early cardiac abnormalities, and improve the ability to identify latent heart diseases.

[0029] 3. A dynamic risk assessment model is constructed based on machine learning and a posture adaptive correction mechanism is introduced. By real-time analysis of the matching deviation between the magnetic cardiogram characteristic parameters and the risk classifier, the false alarm and missed alarm problems caused by individual differences and posture interference in the traditional static threshold warning model are solved, and personalized and highly reliable risk warnings are achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily apparent by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present invention are shown by way of example and not limitation, in which:

[0031] Figure 1 This is a structural diagram of a heart disease risk warning system based on multi-dimensional magnetic heart feature analysis provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0032] The technical solutions of this application will be described clearly and completely below, in conjunction with the accompanying drawings. It should be understood that the described embodiments represent only a portion of the embodiments of this application, and not all of them. The components of this application, generally described and illustrated in the drawings herein, may be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but rather merely represents selected embodiments of this application. All other embodiments derived by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. Furthermore, in the description of this application, the terms "first," "second," etc., are used solely to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0033] In traditional existing cardiac risk warning systems, the acquisition of cardiac magnetic signals is susceptible to environmental electromagnetic interference, resulting in signal baseline drift. The multi-channel sensor array lacks a spatial posture calibration mechanism, which causes errors in magnetic field gradient calculation. The time-frequency domain feature extraction method in the signal processing process fails to effectively integrate multi-dimensional parameter features. The static risk assessment model cannot dynamically adjust the judgment threshold according to the individual's physiological state. The human-computer interaction interface lacks a real-time visual feedback channel, resulting in delayed response to abnormal conditions.

[0034] If the above problems are not solved, the quantitative analysis of cardiac current distribution will produce systematic errors, reducing the sensitivity of early myocardial ischemia detection; static risk assessment strategies cannot identify normal fluctuations in physiological parameters caused by changes in body position, resulting in decreased specificity; and delays in human-computer interaction will prolong clinical decision-making response time and increase the risk of acute cardiovascular event treatment.

[0035] In the face of the above problems, this application first addresses the problem of magnetic field gradient calculation errors caused by the spatial posture offset of the multi-channel sensor array, and explores the elimination of the influence of body position changes by establishing a dynamic posture compensation mechanism. Taking into account that the traditional fixed threshold strategy cannot adapt to individual differences, this application attempts to build a risk assessment model with real-time feedback capabilities, and improves the accuracy of judgment by introducing dynamic calibration of physiological parameters. In response to the defect of insufficient feature extraction in the signal processing process, the study uses multi-dimensional parameter fusion technology to extract the time-frequency domain distribution characteristics of the cardiomagnetic signal, and combines two-dimensional current density parameters to enhance the feature characterization capability. It was finally determined to build a closed-loop system by integrating sensor arrays, signal conditioning, data acquisition, processing core and early warning modules to form a complete technical chain from signal acquisition to risk assessment.

[0036] In this regard, the present application proposes a heart disease risk warning system based on multi-dimensional cardiac magnetic feature analysis, comprising a cardiac magnetic signal sensor array, a signal conditioning module, a data acquisition module, a magnetic field correction module, a central processing module, and a warning output module; the cardiac magnetic signal sensor array is used to collect the original cardiac magnetic field signal, and the object to be measured is located in the detection area of ​​the cardiac magnetic signal sensor array; the signal conditioning module is used to reduce noise and amplify the original signal; the data acquisition module is used to convert the analog signal into a digital cardiac magnetic signal; the central processing module is used to set multi-dimensional cardiac magnetic feature parameters according to the time domain characteristics, frequency domain characteristics and spatial domain characteristics of the cardiac magnetic signal, extract each cardiac magnetic feature parameter through the cardiac magnetic signal processing algorithm, and realize the calculation of the heart disease risk index and dynamic threshold control based on the dynamic risk assessment model constructed by machine learning, combined with the cardiac magnetic feature parameters compensated by the magnetic field correction module and the matching deviation analysis; when the risk index exceeds the dynamic risk threshold, the multimodal warning output module is triggered.

[0037] The magnetic cardiotoc sensor array refers to a device used to collect raw cardiac magnetic field signals. Specifically, it can be constructed using multiple sensors arranged in a matrix. This spatial distribution across the heart area enables high-sensitivity detection, addressing the issue of tissue impedance interference that traditional electrical signal acquisition is susceptible to. The signal conditioning module is a circuit unit that reduces noise and amplifies the raw signal. Specifically, it can be implemented using a combination of a bandpass filter and a lock-in amplifier to suppress ambient noise and improve the signal-to-noise ratio, ensuring signal quality for subsequent processing. The data acquisition module is a hardware component that converts analog signals into digital signals. Specifically, it can be implemented using a high-precision analog-to-digital converter, providing digitized magnetic cardiotoc signal input to the central processing module. The central processing module is the computational unit that performs signal analysis and risk assessment. Specifically, it can use an embedded processor running an algorithm to extract time-frequency domain feature parameters. This algorithm, combined with a machine learning model, dynamically calculates a risk index, enabling personalized heart disease detection. The early warning output module triggers an alarm based on the risk index. Specifically, it can use audio and visual prompts or a data communication interface to provide real-time risk information when an anomaly is detected. Among them, the object being measured is located in the detection area, which means that the position of the human heart needs to be aligned with the spatial coverage range of the sensor array. Positioning calibration is used to ensure the accuracy of signal acquisition and eliminate the influence of body position deviation on the measurement results.

[0038] The core innovation of this application lies in realizing the whole process processing from cardiomagnetic signal acquisition to risk warning through multi-module collaboration, using sensor array to enhance signal sensitivity, combining digital processing with dynamic risk assessment model, breaking through the limitations of traditional methods in feature extraction and adaptability, and finally forming a closed-loop heart disease monitoring and early warning system.

[0039] The operating process and principles of this application are as follows: a magnetic cardiogram (MCG) sensor array acquires raw cardiac magnetic field signals. A signal conditioning module performs noise reduction and amplification on the raw signals. A data acquisition module converts the processed analog signals into digital MCG signals. A central processing module sets multiple MCG characteristic parameters based on the time and frequency domain characteristics of the MCG signals and extracts these characteristic parameters using a MCG signal processing algorithm. Based on a pre-set risk assessment model, the central processing module uses the extracted MCG characteristic parameters to calculate a heart disease risk index. When the calculated heart disease risk index exceeds a dynamic risk threshold, the early warning output module is triggered. The MCG signal sensor array, signal conditioning module, data acquisition module, central processing module, and early warning output module are electrically connected in sequence, forming a closed-loop processing chain from signal acquisition to risk warning. The magnetic field correction module interacts with the central processing module via an independent control bus, achieving dynamic posture compensation while maintaining the integrity of the main signal chain. The subject is located within the detection area of ​​the MCG signal sensor array, ensuring reliable acquisition of the MCG signals. Setting a dynamic risk threshold enables the system to adapt to individual physiological differences and improve the accuracy of early warnings.

[0040] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0041] The MCG sensor array utilizes optically pumped magnetometer technology and consists of 64 sensor units arranged in an 8×8 matrix. The signal conditioning module includes a low-noise preamplifier and a bandpass filter with a filtering band of 2-50 Hz. The data acquisition module utilizes a 24-bit analog-to-digital converter with a sampling rate of 1 kHz. The central processing module utilizes an embedded processor running a real-time operating system. MCG signal processing algorithms include wavelet transform denoising, independent component analysis, and adaptive filtering. The risk assessment model is constructed based on a support vector machine algorithm, with input features including time-frequency domain wave groups, energy characteristics, and current density map-related indicators. Dynamic risk thresholds are personalized based on factors such as the subject's age, gender, and medical history. The early warning output module includes an audible and visual alarm device and a wireless communication module, which transmits warning information to medical staff. The system is also equipped with a touchscreen display for real-time display of MCG waveforms and risk index curves.

[0042] Through the above scheme, this application realizes heart disease risk warning based on cardiac magnetostrictive signals. The cardiac magnetostrictive signal sensor array provides high-sensitivity and high-spatial-resolution signal acquisition capabilities, overcoming the limitations of traditional electrocardiogram detection affected by tissue impedance. Multi-channel signal processing and feature extraction algorithms can comprehensively capture subtle changes in cardiac electrical activity. The dynamic risk assessment model adapts to individual differences and improves the accuracy and reliability of warnings. The system integrates complete functions from signal acquisition to risk warning, providing a more accurate and practical technical means for heart disease monitoring.

[0043] In some of the above-mentioned schemes of the present application, the central processing module sets the cardiac magnetic characteristic parameters through time domain characteristics and frequency domain characteristics. In actual applications, the existing technology has technical defects such as a single dimension of cardiac magnetic feature extraction, insufficient fusion of time and space characteristics, and difficulty in capturing the energy distribution characteristics of different frequency components in the QRS complex through single-scale frequency domain analysis. At the same time, there is a lack of quantitative description of the spatial distribution of the cardiac current density map, resulting in insufficient information dimension of the characteristic parameter set, which affects the discrimination accuracy of the risk assessment model.

[0044] This application further proposes a multi-dimensional collaborative analysis method for extracting features from magnetic cardiogram signals. The central processing module first uses an adaptive window function to accurately segment the pre-processed magnetic cardiogram signals into P waves, QRS complexes, and T waves, and then constructs a feature extraction system that integrates the time domain, frequency domain, and space domain.

[0045] Specifically, at the time domain analysis level, the system synchronously calculates five core time domain parameters, including the RR interval variation coefficient, QT dispersion, and ST segment slope. In the frequency domain feature extraction stage, the Daubechies 5 wavelet basis function is used to perform differential decomposition of the segmented waveform components: a 5-layer wavelet decomposition is performed on the QRS complex, and the energy proportion of the detail coefficients from D2 to D5 is extracted as a multi-scale distribution feature; for the P wave and T wave, the energy weights of their low-frequency, mid-frequency, and high-frequency bands are calculated respectively, supplemented by spectral entropy and center frequency calculations to form a complete frequency domain feature set containing nonlinear characteristics. In terms of spatial domain feature extraction, based on 64-channel magnetic field gradient data, the current density distribution maps of the P wave peak, QRS peak, and T wave peak in the key phases of the cardiac cycle are obtained through regularized least squares inversion.

[0046] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0047] The system achieves precise feature extraction and risk assessment of magnetic cardiogram (MC) signals through a multi-stage signal processing pipeline. The central processing module first performs waveform recognition and segmentation on the preprocessed digital MC signals. It uses an improved Pan-Tompkins algorithm to detect the QRS complex and locate the R wave peak. Template matching is used to identify the P wave onset and T wave endpoint, establishing a complete cardiac cycle time window encompassing the P wave, QRS complex, and T wave.

[0048] Furthermore, during the feature extraction phase, the system performs multidimensional feature extraction in the time, frequency, and spatial domains. This includes calculating time-domain parameters such as the coefficient of variation of the RR interval and the dispersion of the QT interval across consecutive cardiac cycles, and analyzing ST segment morphology. The QRS complex is decomposed into five layers using the Daubechies 5 wavelet, and the energy contribution of each decomposition layer is calculated. The 2-15 Hz band is defined as the low-frequency band for analyzing myocardial repolarization; the 15-35 Hz band is defined as the mid-frequency band, reflecting the main ventricular depolarization activity; and the 35-50 Hz band is defined as the high-frequency band for detecting microvolt-level abnormal discharges. The QRS complex is decomposed into five layers using the Daubechies 5 wavelet, and the energy contribution of each decomposition layer is calculated. The energy components and spectral entropy values ​​of different frequency bands are extracted for the P and T waves, respectively. For spatial feature extraction, the current density distribution in the cardiac region is reconstructed using an electromagnetic field inversion algorithm based on multi-channel magnetic field gradient data. The system analyzes current density characteristics at different phases of the cardiac cycle, including the maximum current density value and its spatial location, as well as the dynamic characteristics of the current vector field. Using spatiotemporal registration technology, these characteristics are aligned with a standard heart model to accurately locate abnormal electromagnetic activity.

[0049] Through the above technical solution, this application achieves multi-dimensional feature extraction of cardiac magnetoencephalogram signals, which includes both time-frequency domain information and spatial current distribution information, providing comprehensive feature parameters for heart disease risk assessment. This improves the accuracy and reliability of heart disease risk warnings and effectively reduces the false positive rate. Furthermore, this method can detect subtle abnormalities in cardiac electrical activity that are difficult to detect with conventional electrocardiograms, facilitating early detection of potential heart disease risks.

[0050] In some of the above-mentioned solutions in this application, the risk assessment model lacks a dynamic adjustment mechanism and is difficult to adaptively update according to individual differences and real-time data changes, resulting in a decrease in the matching accuracy of the risk classifier and an inability to effectively identify potential risks that continue to deviate from the normal pattern.

[0051] This application further proposes a central processing module for receiving cardiac magnetic characteristic parameters and corresponding clinical diagnostic labels in a historical cardiac magnetic database, and training a risk classifier through a machine learning algorithm; the central processing module is also used to calculate in real time the deviation between the cardiac magnetic characteristic parameters of the subject and the matching degree of the risk classifier, and generate a primary risk flag when the deviation continues for more than a preset time.

[0052] The process of training the risk classifier using a machine learning algorithm involves using historical CMT characteristic parameters as input features and corresponding clinical diagnostic labels as supervisory signals. Risk discrimination rules are then derived through classification model training. Deviation calculation is achieved by comparing the real-time CMT characteristic parameters with the predicted values ​​output by the risk classifier. This difference can be measured using Euclidean distance, cosine similarity, or a probability distribution difference algorithm. A preset duration is set based on clinical experience to filter out transient noise interference and ensure statistical significance when triggering primary risk markers.

[0053] Specifically, the historical magnetic cardio database stores the magnetic cardio characteristic parameters and medically verified heart disease status labels of multiple subjects. The central processing module uses support vector machines, random forests or neural network algorithms to learn the mapping relationship between features and risk categories from labeled data to form a risk classifier. In real-time monitoring, the magnetic cardio characteristic parameters of the subject are input into the trained classifier, and the predicted risk probability or category is output. The deviation is obtained by calculating the distance between the real-time feature vector and the decision boundary of the classifier or the degree of consistency with the distribution of historical data. When the deviation exceeds the threshold for a continuous period of time reaching a preset length of time, the system determines that there is a persistent abnormality and generates a primary risk flag. Therefore, by dynamically evaluating the matching status of the feature parameters and the classifier, the sensitivity of identifying potential risks is improved while avoiding false alarms.

[0054] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0055] In the construction of the risk assessment model, the central processing module receives the cardiac magnetic characteristic parameters and corresponding clinical diagnosis labels from the historical cardiac magnetic database and trains the risk classifier through a machine learning algorithm. Specifically, cardiac magnetic characteristic parameters including QRS complex amplitude, energy distribution, maximum current density value and spatial coordinates are first extracted from the historical cardiac magnetic database, and the corresponding clinical diagnosis labels are obtained. Then, the support vector machine (SVM) algorithm is used to construct the risk classifier model. The SVM model maps the feature space to a high-dimensional space through a kernel function to find the optimal classification hyperplane. During the training process, the grid search method is used to optimize the kernel function parameters and penalty factor of the SVM to improve the classification accuracy.

[0056] Furthermore, the central processing module calculates in real time the deviation between the subject's MCG characteristic parameters and the matching degree of the risk classifier. Thus, for each newly acquired MCG signal sample, its characteristic parameters are first extracted and then input into the trained SVM classifier. The classifier outputs the probability distribution of the sample belonging to each risk level. The deviation is defined as the KL divergence between the risk probability distribution of the sample and the probability distribution of a standard low-risk sample.

[0057] A primary risk flag is generated when the deviation exceeds a preset duration. For example, set the deviation threshold to 0.5 and the preset duration to 1 minute. If the deviation exceeds 0.5 for six consecutive samples (sampling frequency 0.1Hz), the primary risk flag is triggered. The primary risk flag can be implemented by changing the color of the system interface or a slight sound prompt to attract the operator's attention.

[0058] Through the above technical solution, this application realizes the automatic construction and dynamic update of the risk assessment model based on historical data. The risk classifier trained by the machine learning algorithm can make full use of the laws contained in a large amount of historical data, improving the accuracy and personalization of risk assessment. The mechanism of calculating the deviation in real time and setting the primary risk flag enables the system to capture subtle changes in the heart state in a timely manner, providing the possibility for early warning. This method avoids the limitations of the traditional fixed threshold method, improves the system's adaptability to individual differences and sensitivity to potential risks.

[0059] In some of the above-mentioned schemes of the present application, the adjustment of the dynamic risk threshold relies on the assumption of stable body posture. When the body posture of the subject changes, the spatial posture offset of the magnetic cardiotoc signal sensor array will cause drift errors in the magnetic field measurement values, thereby causing distortion of the magnetic cardiotoc characteristic parameters and misjudgment of the dynamic risk threshold.

[0060] This application further proposes to automatically activate a two-dimensional magnetic field correction module based on the posture change signal of the subject being measured. This module reconstructs the cardiomagnetic signal sensor array based on the acceleration sensor data, the posture matrix in the two-dimensional plane, and eliminates the posture drift error through orthogonal projection compensation. The central processing module dynamically adjusts the dynamic risk threshold according to the compensated cardiomagnetic characteristic parameters.

[0061] The detection of body position change signals is achieved through an accelerometer, which is fixedly connected to the magnetic cardiosignal sensor array to measure the motion acceleration data in a two-dimensional plane in real time. The calculation of the spatial posture matrix uses a simplified rotation model to integrate the x-axis and y-axis angular velocities obtained by the accelerometer to obtain the rotation angle θ. x ,θ y Construct a two-dimensional rotation transformation matrix T; orthogonal projection compensation is performed by converting the original magnetic field vector B raw Projection is performed in the inverse matrix direction of the rotational transformation matrix T to eliminate the deviation of the magnetic field component caused by body rotation. The ambient magnetic field baseline offset ΔB is obtained through zero-position calibration in the initialization phase to compensate for static magnetic field interference. The dynamic risk threshold is adjusted using a sliding window mechanism, with the mean of the cardiac magnetic characteristic parameters of 10 consecutive heart beats after compensation as the benchmark, and the threshold interval is updated within a range of ±3 times the standard deviation.

[0062] Specifically, when the accelerometer detects that the body position change exceeds the preset acceleration threshold, the central processing module automatically calls the two-dimensional magnetic field correction algorithm. The two-axis angular velocity data output by the accelerometer is numerically integrated to calculate the rotation angle of the magnetic cardiosignal sensor array relative to the initial coordinate system, generating a two-dimensional rotation transformation matrix T. The original magnetic field vector B raw Through the matrix operation T(θ x ,θ y )·B raw Perform spatial coordinate transformation and then superimpose the environmental magnetic field reference offset ΔB to obtain the corrected magnetic field vector B cor The compensated magnetic field data is input into the characteristic parameter extraction process. The central processing module uses the mean plus or minus three times the standard deviation as the upper and lower limits of the dynamic risk threshold based on the statistical distribution of the characteristic parameters of the latest 20 cardiac cycles. For example, when the subject changes position in the coronal plane, the acceleration sensor detects that the x-axis angular velocity changes by more than 0.3 rad / s, and the system automatically triggers posture compensation. The fluctuation amplitude of the maximum current density modulus after correction is reduced, which improves the adjustment accuracy of the dynamic risk threshold.

[0063] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0064] During the dynamic risk threshold update, the central processing module automatically activates the two-dimensional magnetic field correction module based on the subject's position change signals. This module reconstructs the planar pose matrix of the magnetocardiographic sensor array based on accelerometer data and eliminates position drift errors through orthogonal projection compensation. The central processing module dynamically adjusts the dynamic risk threshold based on the compensated magnetocardiographic characteristic parameters.

[0065] Specifically, the system first obtains the motion data of the object under test in a two-dimensional plane through an accelerometer, and obtains the rotation angle of the sensor array relative to the initial position after integration. The two-dimensional rotation transformation matrix is ​​constructed using this angle information to perform coordinate transformation and compensation on the original magnetic heart signal. The compensated signal is used to re-extract the magnetic heart characteristic parameters, and the risk threshold range is dynamically adjusted based on these updated parameters. This solution effectively reduces the interference of body position changes on risk assessment and improves the measurement accuracy of the system in dynamic environments. For example, if the corrected characteristic parameters show that the heart activity is relatively stable, the system may appropriately increase the risk threshold; conversely, if the corrected parameters show potential abnormalities, the system may lower the threshold to increase the alarm sensitivity.

[0066] Through the above technical solution, this application achieves effective compensation for the distortion of the cardiac magnetic signal caused by changes in body position. The dynamic risk threshold update mechanism improves the accuracy and reliability of heart disease risk warning. This solution can adapt to the changes in the body position of the subject during testing, reducing false positive alarms caused by body position changes, while ensuring sensitive detection of real heart abnormalities. In addition, by adjusting the risk threshold in real time, the system can better adapt to individual differences and environmental changes, providing users with more accurate and personalized heart disease monitoring services.

[0067] In some of the above-mentioned solutions of this application, the structural design of the magnetic cardiosignal sensor array affects the signal acquisition quality. The number and arrangement of traditional sensor channels make it difficult to balance spatial resolution and signal coverage uniformity, resulting in blind spots or interference superposition in the collected original signals, affecting the accuracy of subsequent processing.

[0068] This application further proposes that the magnetocardiographic signal sensor array is composed of 64-channel optically pumped magnetometers, with each channel arranged in an 8×8 matrix; the spacing between adjacent channels is 25 mm; and the signal conditioning module includes a bandpass filter and a phase-locked amplifier.

[0069] The 64-channel optically pumped magnetometer utilizes a two-dimensional array topology, covering the cardiac magnetic field detection area. Each channel is arranged in an 8-row, 8-column matrix, forming 64 independent detection nodes. The center-point spacing between adjacent channels is set at 25 mm, ensuring accurate magnetic field gradient detection while preventing adjacent magnetic field coupling. A bandpass filter is set to a frequency range of 2 Hz to 50 Hz to filter out power frequency interference and baseline drift. A lock-in amplifier uses coherent detection technology to extract microvolt-level cardiac magnetic signals.

[0070] Specifically, a 64-channel optically pumped magnetometer covers the projected area of ​​the anterior thoracic surface, achieving two-dimensional spatial sampling of the cardiac magnetic field through an 8×8 matrix arrangement. The 25mm spacing design is based on the anatomical dimensions of the adult heart. A bandpass filter uses a Butterworth second-order active circuit to achieve a 2Hz high-pass cutoff and a 50Hz low-pass cutoff, suppressing myoelectric interference and power supply harmonics. A phase-locked amplifier uses reference frequency synchronous demodulation technology to amplify nanovolt-level cardiac magnetic signals to the volt level. This structure achieves full coverage detection of the spatial distribution of the magnetic field through a high-density sensor array. Combined with the frequency domain filtering and amplification functions of the signal conditioning module, it effectively improves the signal-to-noise ratio and stability of the original signal.

[0071] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0072] The magnetocardiographic sensor array consists of 64 superconducting quantum interference devices (SQIDs), arranged in an 8×8 matrix. Adjacent channels are spaced 25 mm apart. The signal conditioning module includes a bandpass filter and a lock-in amplifier. The bandpass filter has a passband of 2-50 Hz to filter out power frequency interference and baseline drift. The lock-in amplifier utilizes dual-phase lock-in detection technology, with the reference frequency set to the main frequency component of the magnetocardiographic signal to improve the signal-to-noise ratio.

[0073] Through the above technical solution, this application achieves the precise acquisition of high-density cardiac magnetometry signals. The 64-channel optically pumped magnetometer array provides a large spatial coverage range, which can comprehensively capture the magnetic field distribution in various areas of the heart. The 25mm channel spacing ensures high spatial resolution. Bandpass filtering and phase-locked amplification effectively suppress external electromagnetic interference and improve the detection sensitivity of weak cardiac magnetometry signals. As a result, the system obtains high-quality cardiac magnetometry raw data, laying the foundation for subsequent feature extraction and risk assessment.

[0074] In some of the above-mentioned schemes of this application, it is proposed to train a risk classifier through a machine learning algorithm to assess the risk of heart disease. However, in this process, there is a lack of quantitative basis for the contribution of different magnetic cardiotactic characteristic parameters to the risk index, resulting in deviations in the model's sensitivity to different pathological characteristics.

[0075] This application further proposes a risk assessment model using the formula for calculating the magnetic cardiorespiratory risk index: ; in: is the heart disease risk index; 、 、 They constitute the linear combination of time domain, frequency domain and space domain features respectively; is the deviation variable based on the classifier matching degree; is the weight coefficient of each feature and satisfies .

[0078] The weight coefficient is determined by clinical data regression. For any weight coefficient λ∈{α,β,γ,δ}, its calculation formula is:

[0079]

[0080] Where: N is the total number of clinical samples; F s,λ Indicates the feature combination value of the sth sample in the corresponding dimension (S t,s 、S f,s 、S s,s or D s );R sis the risk level value of the sth sample confirmed by the gold standard; and are the sample means of the corresponding features.

[0081] Specifically, the formula for calculating the CMI risk index constructs a multidimensional risk assessment system by linearly weighting the integration of time-frequency domain CMI features, current density features, and model matching deviation features. Clinical data regression methods are used to determine the objective weights of each feature. Based on covariance analysis of a large number of clinical samples, the correlation coefficients between each feature parameter and the gold standard risk level are calculated and normalized to serve as weighting coefficients.

[0082] Taking the application example of 500 clinical sample data sets as an example, we first calculate the S of each sample. Xi 、S Yi 、S Zi Measured value, model matching deviation D i and the corresponding R i Clinical diagnostic value, and then calculate each feature and R i The covariance and variance ratios of the features were calculated to obtain initial weights, which were then normalized to ensure that α + β + γ + δ = 1. This process uses mathematical modeling to convert the pathological contributions of different features into quantitative parameters, enabling the model to adaptively adjust the sensitivity of different signs, thereby eliminating feature sensitivity bias. In one calculation, the normalized values ​​were α = 0.38, β = 0.29, γ = 0.21, and δ = 0.12. This set of weight coefficients reflects the relative importance of time domain features, frequency domain features, current density features, and feature deviation for risk assessment in this clinical dataset.

[0083] Through the above technical solution, this application achieves adaptive optimization of weight coefficients. Due to the use of a large amount of clinical data for regression analysis, the weight coefficients are more in line with the actual situation, improving the accuracy and universality of the risk assessment model. At the same time, through clear mathematical formulas, the calculation process of the weight coefficients is more standardized and repeatable. This method also facilitates the dynamic updating of the weight coefficients based on newly added clinical data, enabling the model to continuously improve and evolve.

[0084] In some of the above-mentioned schemes of the present application, when reconstructing the spatial posture matrix of the magnetocardiographic signal sensor array based on the acceleration sensor data and eliminating the body position drift error through orthogonal projection compensation, since the body position change involves complex spatial rotational movement and the presence of baseline offset interference in the ambient magnetic field, it is difficult to completely eliminate the magnetic field measurement error caused by the posture change by relying solely on orthogonal projection, resulting in deviations in the compensated magnetocardiographic characteristic parameters, which affects the adjustment accuracy of the dynamic risk threshold.

[0085] This application further proposes a posture compensation formula for the magnetic field correction module:

[0086] B cor =T(θ x ,θ y )·B raw +ΔB

[0087] Among them: B cor is the magnetic field vector after correction; B raw is the original magnetic field vector; T is the rotation transformation matrix θ calculated from the acceleration sensor data x ,θ y is the rotation angle around the x-axis and y-axis; ΔB is the ambient magnetic field reference offset.

[0088] The rotation transformation matrix T is used to calculate the spatial rotation angle θ caused by the body position change based on the acceleration data collected by the acceleration sensor. x ,θ y Generate and construct a two-dimensional rotation operation relationship through homogeneous coordinate transformation.

[0089] The ambient magnetic field baseline offset ΔB is determined by measuring the background magnetic field of the magnetocardiographic sensor array in the absence of a measured object, eliminating interference from the geomagnetic environment and the inherent magnetic field of the device. Matrix multiplication is used to correct for the magnetic field vector direction offset caused by body rotation, while scalar addition ΔB compensates for the constant offset component of the ambient magnetic field. This formula decomposes the spatial posture transformation into two independent computational steps: rotation and translation, achieving separate compensation for multi-degree-of-freedom errors.

[0090] In the specific implementation, when the position change of the measured object triggers the magnetic field correction, the acceleration sensor collects the three-axis acceleration data in real time and transmits it to the central processing module. The central processing module calculates the rotation angle θ around the x-axis and y-axis based on the acceleration data. x ',θ y 'Generate the corresponding rotation transformation matrix T. Original magnetic field vector θ x ,θ y After coordinate transformation by the transformation matrix T, the pre-calibrated environmental magnetic field reference offset ΔB is superimposed to obtain the corrected magnetic field vector B cor . This process eliminates the deviation in the direction of the magnetic field vector caused by changes in body position through rotation transformation, and compensates for the interference of the environmental magnetic field through the reference offset, so that the compensated magnetic field data accurately reflects the true magnetic field distribution of the heart. The corrected magnetic field data is input into the subsequent cardiac magnetic feature parameter extraction process to ensure the accuracy of the dynamic risk threshold adjustment. For example, the rotation transformation matrix is ​​constructed using the rotation order YX to avoid the universal joint lock problem; the environmental magnetic field reference offset is updated through daily system self-calibration to adapt to changes in the external magnetic field environment.

[0091] Through the above technical solution, this application effectively eliminates the offset error in the spatial coordinate system of the cardiomagnetic signal caused by body position movement, and solves the problem of magnetic field vector measurement distortion caused by sensor posture changes in traditional systems. By dynamically compensating for the impact of body position changes on magnetic field measurements, the accuracy of the calculation of cardiomagnetic characteristic parameters is ensured, enabling subsequent risk assessment models to determine disease status based on stable and reliable signal characteristics.

[0092] In some of the above-mentioned solutions of this application, although the central processing module can trigger the early warning output module, in actual applications, users may not be able to obtain specific information on the risk status in a timely manner, and lack effective interactive means to confirm key data when an anomaly occurs, resulting in response delays or operational errors.

[0093] The present application further proposes a system including a user interaction module, which includes a touch screen and a voice prompt unit; the touch screen is used to display the heart disease risk index and the magnetic heart characteristic parameter curve in real time; the voice prompt unit is used to broadcast the risk level information when the early warning output module is triggered; the central processing module and the user interaction module are linked for control: when the heart disease risk index exceeds the dynamic risk threshold, the following operations are performed synchronously: freeze the display of the magnetic heart characteristic parameter curve on the touch screen and mark the abnormal timestamp; drive the voice prompt unit to broadcast the third-level risk alarm voice; generate an encrypted risk report and send it to the designated medical terminal through the wireless transmission module.

[0094] The touchscreen displays real-time risk indexes and magnetic cardiogram (MCM) signal curves through a dynamically updated interface, using a high-refresh rate LCD panel to ensure data visibility. The voice prompt unit incorporates a built-in multi-level alarm library, calling corresponding audio files based on risk level. Interrupt-triggered control logic ensures the simultaneous execution of multiple operations. The central processing module generates encrypted risk reports using an asymmetric encryption algorithm and transmits them to a pre-defined medical terminal via a secure transmission protocol.

[0095] Specifically, when the heart disease risk index exceeds the dynamic threshold, the central processing module immediately sends a control instruction to the user interaction module. After receiving the instruction, the touch screen stops the dynamic refresh of the curve, locks the current display screen and superimposes the abnormal timestamp mark, retaining the characteristic parameter waveform at the triggering moment for subsequent analysis. The voice prompt unit plays the corresponding voice content according to the preset three-level alarm classification, such as "Level 1 Alarm: It is recommended to seek medical attention immediately." At the same time, the central processing module packages the current magnetic heart characteristic parameters, risk index and timestamp data into a structured report file, processes it using an encryption module based on the RSA algorithm, and transmits it to the hospital server or doctor terminal via a cellular network or Wi-Fi module. In this process, display freezing prevents the loss of coverage of key data, voice broadcasting enhances the user's perception of the emergency state, and encrypted transmission ensures the security of sensitive medical data.

[0096] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0097] When the heart disease risk index exceeds the dynamic risk threshold, the touch screen immediately stops refreshing the dynamic waveform of the magnetic heart characteristic parameter curve and generates a time axis positioning mark with a red mark on the display interface. The timestamp is accurate to the millisecond level and synchronized with the data acquisition time code; the voice prompt unit calls the pre-stored three-level alarm voice file and plays the synthesized voice containing the "three-level heart risk warning" content in a loop through the directional speaker; the encrypted risk report is transmitted to the preset hospital cardiology monitoring system server through the wireless communication module.

[0098] Through the above technical solution, this application achieves the simultaneous output of multi-dimensional information under risk warning conditions, solving the technical defect of the existing system's single interactive feedback. By freezing the display interface and marking the abnormal time point, it is possible to accurately locate the specific time period when the abnormal magnetic cardiogram signal occurs, avoiding the loss of key data caused by dynamic refresh. Through standardized voice broadcast content and encrypted data transmission, it not only ensures the timeliness and standardization of risk warnings, but also ensures the security of patients' privacy data during transmission.

[0099] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A heart disease risk warning system based on multi-dimensional magnetic heart feature analysis, characterized by: The system includes a magnetic cardiosignal sensor array, a signal conditioning module, a data acquisition module, a magnetic field correction module, a central processing module and an early warning output module; The magnetic heart signal sensor array is used to collect the original signal of the cardiac magnetic field, and the object to be measured is located in the detection area of ​​the magnetic heart signal sensor array; the signal conditioning module is used to reduce noise and amplify the original signal; the data acquisition module is used to convert the analog signal into a digital magnetic heart signal; the central processing module is used to set multi-dimensional magnetic heart characteristic parameters according to the time domain characteristics, frequency domain characteristics and spatial domain characteristics of the magnetic heart signal, extract each magnetic heart characteristic parameter through the magnetic heart signal processing algorithm, and realize the calculation of the heart disease risk index and dynamic threshold control based on the dynamic risk assessment model constructed by machine learning, combined with the magnetic heart characteristic parameters compensated by the magnetic field correction module and the matching deviation analysis; when the risk index exceeds the dynamic risk threshold, the multimodal early warning output module is triggered.

2. The heart disease risk warning system according to claim 1, characterized in that: In the extraction of cardiac magnetic characteristic parameters, the central processing module is used to pre-process and recognize the waveform of the cardiac magnetic signal, extract time domain morphological indicators including RR interval variation coefficient, QT dispersion, ST segment offset, T wave asymmetry, R peak time difference, etc., to achieve a preliminary assessment of the stability and conduction characteristics of cardiac electrical activity; on this basis, the cardiac magnetic signal is further decomposed by wavelet, and the multi-scale energy distribution characteristics within the QRS wave group time window are extracted, and the low-frequency, high-frequency and ultra-high-frequency energy proportions, spectral entropy, and central Frequency domain parameters such as frequency and frequency are obtained to complete the detailed frequency domain characterization of the cardiac electromagnetic activity; at the same time, based on the two-dimensional current density distribution map of the cardiac magnetostrictive signal, the dynamic characteristics such as the total current vector, the maximum current vector and its spatial coordinates, the current angle, the current vector rotation and divergence in different phases of the cardiac cycle are analyzed to obtain the cardiac magnetostrictive source activity information with spatial positioning capability; finally, the above-mentioned time domain indicators, frequency domain characteristics and the derived parameters of the two-dimensional current density map are combined to form a multi-dimensional cardiac magnetostrictive feature set covering the time domain, frequency domain and space domain, and are sent to the risk assessment model as the core input parameters.

3. The heart disease risk warning system according to claim 1, characterized in that: In the construction of the risk assessment model, the central processing module is used to receive the cardiac magnetic characteristic parameters and corresponding clinical diagnostic labels in the historical cardiac magnetic database, and train the risk classifier through a machine learning algorithm; the central processing module is also used to calculate in real time the deviation between the cardiac magnetic characteristic parameters of the subject and the matching degree of the risk classifier, and generate a primary risk flag when the deviation continues to exceed a preset time.

4. The heart disease risk warning system according to claim 1, characterized in that: In the dynamic risk threshold update, the central processing module is used to automatically activate the magnetic field correction module according to the body position change signal of the subject; the magnetic field correction module is used to reconstruct the spatial posture matrix of the magnetocardiographic signal sensor array based on the acceleration sensor data, and eliminate the drift error in the two-dimensional plane caused by body position change through orthogonal projection compensation; the central processing module dynamically adjusts the risk threshold according to the compensated magnetocardiographic characteristic parameters.

5. The heart disease risk warning system according to claim 1, characterized in that: The magnetocardiographic signal sensor array is composed of a multi-channel superconducting quantum interference device or an optically pumped magnetometer, and each channel is arranged in a matrix; the signal conditioning module includes a bandpass filter and a phase-locked amplifier.

6. The heart disease risk warning system according to claim 2 or 3, characterized in that: The risk assessment model uses the formula for calculating the magnetic cardiorespiratory risk index: ； in: is the heart disease risk index; 、 、 They constitute the linear combination of time domain, frequency domain and space domain features respectively; is the deviation variable based on the classifier matching degree; is the weight coefficient of each feature and satisfies .

7. The heart disease risk warning system according to claim 6, characterized in that: The weight coefficient is determined by clinical data regression. For any weight coefficient λ∈{α,β,γ,δ}, its calculation formula is: Where: N is the total number of clinical samples; F s,λ Indicates the feature combination value of the sth sample in the corresponding dimension (S t,s 、S f,s 、S s,s or D s );R s is the risk level value of the sth sample confirmed by the gold standard; and are the sample means of the corresponding features.

8. The heart disease risk warning system according to claim 4, characterized in that: The posture compensation formula of the magnetic field correction module is: B cor =T(θ x ,i y )·B raw +ΔB Among them: B cor is the magnetic field vector after correction; B raw is the original magnetic field vector; T is the rotation compensation matrix calculated from the acceleration sensor data; θ x ,θ y is the rotation angle around the x-axis and y-axis; ΔB is the ambient magnetic field reference offset.

9. The heart disease risk warning system according to claim 1, characterized in that: The system further includes a user interaction module; the user interaction module includes a touch screen and a voice prompt unit; the touch screen is used to display the heart disease risk index and the cardiomagnetic characteristic parameter curve in real time; The voice prompt unit is used to broadcast risk level information when the early warning output module is triggered.

10. The heart disease risk warning system according to claim 9, characterized in that: The central processing module and the user interaction module are linked to control: when the heart disease risk index exceeds the dynamic risk threshold, the following operations are performed synchronously: Freeze the magnetic cardiology characteristic parameter curve on the touch screen and mark the abnormal timestamp; Drive the voice prompt unit to broadcast the third-level risk warning voice; Generate an encrypted risk report and send it to the designated medical terminal via the wireless transmission module.

Images