Method and system for offset analysis based on bio-wave resonance

Abstract

The application relates to the technical field of biological wave resonance, and provides a deviation analysis method and system based on biological wave resonance, so as to solve the problems of low precision and poor anti-interference capability of biological wave resonance abnormality detection in the prior art. The method comprises the following steps: collecting electrocardiosignal, electroencephalogram signal and micro-deformation data on the surface of a living body, and jointly aligning the signal data in a time-frequency domain; dynamically suppressing interference frequency bands of the electrocardiosignal and the electroencephalogram signal after joint alignment respectively; fusing the electrocardiosignal, the electroencephalogram signal after suppression and the micro-deformation data after joint alignment, to generate a fusion result; determining a deviation feature of the fusion result relative to a preset reference state by using convolution operation; and generating a biological wave resonance deviation analysis report of whether the biological wave resonance is abnormal according to the continuous length and the spatial distribution range of the deviation feature in a time domain dimension. The application improves the precision and the anti-interference capability of biological wave resonance abnormality detection.

Description

Technical Field

[0001] This application relates to the field of biowave resonance technology, and in particular to a migration analysis method and system based on biowave resonance. Background Technology

[0002] In the field of biomedical detection under complex electromagnetic environments, new technical requirements have been placed on the accurate analysis of bio-wave resonance signals. Medical diagnosis needs to simultaneously capture the coupling relationship between electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformations of the body surface. Traditional single-mode signal processing methods are insufficient to meet the collaborative analysis needs of multi-source heterogeneous data. Especially in scenarios with strong electromagnetic interference, maintaining signal fidelity and extracting effective bio-wave resonance features has become a key challenge for improving diagnostic accuracy.

[0003] Current solutions employ a joint noise reduction method based on wavelet transform. This method first performs wavelet decomposition on ECG and EEG signals separately, then prunes high-frequency noise components using a preset threshold, and finally performs a simple superposition analysis with surface strain sensor data. This solution achieves the suppression of noise in a single signal through time-frequency analysis and attempts to establish a correlation model for multi-source signals.

[0004] Existing solutions are limited in the face of strong electromagnetic interference. The fixed basis functions of wavelet decomposition are difficult to adapt to the non-stationary characteristics of bio-wave resonance signals, resulting in the loss of effective frequency band information. The signal superposition process lacks a deep coupling mechanism for cross-modal features and is insensitive to the dynamic correlation between micro-deformation data and electrophysiological signals. The separation of electromagnetic interference suppression and feature extraction processes affects the spatiotemporal consistency of the final analysis results. Summary of the Invention

[0005] This application provides a migration analysis method and system based on bio-wave resonance to solve the problems of low accuracy and poor anti-interference ability in the detection of bio-wave resonance anomalies in the prior art.

[0006] In a first aspect, this application provides a migration analysis method based on bio-wave resonance, comprising:

[0007] Collect electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformation data on the surface of an organism, and then perform joint alignment of the ECG signals, EEG signals, and micro-deformation data in the time-frequency domain.

[0008] The interference frequency bands of the jointly aligned ECG signal and the jointly aligned EEG signal are dynamically suppressed respectively. The suppressed ECG signal, the suppressed EEG signal and the jointly aligned micro-deformation data are then fused to generate a fusion result.

[0009] Convolution operations are used to determine the offset features of the fusion result relative to a preset baseline state;

[0010] Based on the duration and spatial distribution range of the aforementioned offset features in the time domain, a biowave resonance offset analysis report is generated to determine whether the biowave resonance is abnormal.

[0011] Optionally, determining the offset features of the fusion result relative to a preset reference state using convolution operations includes:

[0012] The fusion result is dimensionally aligned with the benchmark multidimensional feature set corresponding to the preset benchmark state.

[0013] By combining features of different dimensions in the aligned fusion result through multiple convolutional kernels, a multi-scale convolutional response feature set is generated.

[0014] Key features are extracted from the multi-scale convolution response feature set and used as offset features.

[0015] Optionally, the step of combining features of different dimensions in the aligned fusion result through multiple convolutional kernels to generate a multi-scale convolutional response feature set includes:

[0016] The first-layer convolutional kernel is used to slide and combine the temporal dimension features in the aligned fusion result to generate a primary response feature set.

[0017] A second-layer convolutional kernel is used to combine the spatial dimension features in the aligned fusion result across regions to generate a set of secondary response features;

[0018] The primary response feature set and the secondary response feature set are superimposed to form a multi-scale convolutional response feature set.

[0019] Optionally, the step of superimposing the primary response feature set and the secondary response feature set to form a multi-scale convolutional response feature set includes:

[0020] Based on the numerical fluctuation range of each time-domain feature in the primary response feature set, the numerical values ​​of each time-domain feature in the primary response feature set are adjusted.

[0021] Based on the distribution density of each spatial dimension feature in the secondary response feature set, the numerical values ​​of each spatial dimension feature in the secondary response feature set are adjusted.

[0022] Determine the first similarity between the numerically adjusted primary response feature set and the benchmark time-domain feature set corresponding to the preset benchmark state, and determine the second similarity between the numerically adjusted secondary response feature set and the benchmark spatial-dimensional feature set corresponding to the preset benchmark state;

[0023] Based on the first similarity and the second similarity, the numerically adjusted primary response feature set and the numerically adjusted secondary response feature set are linearly combined to form a multi-scale convolutional response feature set.

[0024] Optionally, the dynamic suppression of interference frequency bands in the jointly aligned ECG signal and the jointly aligned EEG signal includes:

[0025] Establish a first time-frequency distribution model of the jointly aligned electrocardiogram signal and a second time-frequency distribution model of the jointly aligned electroencephalogram signal;

[0026] Extract a first target frequency band related to electromagnetic interference from the first time-frequency distribution model, and extract a second target frequency band related to electromagnetic interference from the second time-frequency distribution model;

[0027] An adaptive filtering algorithm is used to dynamically suppress the electrocardiogram (ECG) signal in the first target frequency band to form a suppressed ECG signal, and the electroencephalogram (EEG) signal in the second target frequency band is also dynamically suppressed to form a suppressed EEG signal.

[0028] Optionally, the process of fusing the suppressed electrocardiogram signal, the suppressed electroencephalogram signal, and the jointly aligned micro-deformation data to generate a fusion result includes:

[0029] The suppressed electrocardiogram signal is mapped to the first multidimensional feature vector, the suppressed electroencephalogram signal is mapped to the second multidimensional feature vector, and the jointly aligned micro-deformation data is mapped to the third multidimensional feature vector.

[0030] Based on the signal intensity change pattern in the first multidimensional feature vector, the phase continuity pattern in the second multidimensional feature vector, and the deformation propagation pattern in the third multidimensional feature vector, a cross-modal association rule is constructed.

[0031] Based on the cross-modal association rule, the first multidimensional feature vector, the second multidimensional feature vector, and the third multidimensional feature vector are superimposed to generate a multidimensional feature set, which is the fusion result.

[0032] Optionally, the step of generating a biowave resonance migration analysis report to determine whether the biowave resonance is abnormal, based on the duration and spatial distribution range of the migration features in the time domain, includes:

[0033] The cumulative time length of the N consecutive occurrences of the offset feature in the time domain dimension, and the ratio of the area of ​​the organism surface covered to the whole organism surface in the spatial dimension are calculated, where N is greater than or equal to 3.

[0034] Based on the first comparison result between the cumulative time length and the preset length threshold, and the second comparison result between the ratio and the preset ratio threshold, a determination result of whether the bio-wave resonance is abnormally offset is generated.

[0035] Based on the judgment results, a bio-wave resonance migration analysis report is generated.

[0036] Secondly, this application provides a migration analysis system based on bio-wave resonance, comprising:

[0037] The acquisition module is used to acquire electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformation data on the surface of the organism, and to jointly align the ECG signals, EEG signals, and micro-deformation data in the time-frequency domain.

[0038] The fusion module is used to dynamically suppress the interference frequency bands of the jointly aligned ECG signal and the jointly aligned EEG signal, respectively, and to fuse the suppressed ECG signal, the suppressed EEG signal and the jointly aligned micro-deformation data to generate a fusion result.

[0039] The determination module is used to determine the offset features of the fusion result relative to a preset baseline state using convolution operations;

[0040] The generation module is used to generate a biowave resonance migration analysis report to determine whether the biowave resonance is abnormal, based on the duration and spatial distribution range of the migration features in the time domain.

[0041] Thirdly, this application provides a computing device including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform a migration analysis method based on bio-wave resonance as described in any of the first aspects.

[0042] Fourthly, this application provides a computer storage medium storing computer program instructions thereon, which, when executed by a processor, implement a migration analysis method based on bio-wave resonance as described in any one of the first aspects.

[0043] This application provides a migration analysis method based on bio-wave resonance. The method includes: acquiring electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformation data of the surface of a living organism; jointly aligning the ECG signals, EEG signals, and micro-deformation data in the time-frequency domain; dynamically suppressing interference frequency bands of the jointly aligned ECG signals and EEG signals; fusing the suppressed ECG signals, suppressed EEG signals, and jointly aligned micro-deformation data to generate a fusion result; determining the migration characteristics of the fusion result relative to a preset baseline state using convolution operations; and generating a bio-wave resonance migration analysis report indicating whether the bio-wave resonance is abnormal based on the duration and spatial distribution range of the migration characteristics in the time domain.

[0044] The technical solution provided in this application has the following beneficial effects:

[0045] This application simultaneously acquires electrocardiogram (ECG), electroencephalogram (EEG), and micro-deformation data, aligning them in the time-frequency domain to ensure the spatiotemporal consistency of multi-source signals, providing a foundation for subsequent fusion analysis. Interference frequency bands are suppressed for both ECG and EEG signals to retain effective bio-wave resonance characteristics and improve the signal-to-noise ratio. The suppressed electrical signals are deeply fused with micro-deformation data to generate a fusion result containing multi-dimensional spatiotemporal features, enhancing the completeness of feature expression. Convolutional operations are used to compare the fusion result with the baseline state, accurately identifying abnormal offset features and improving detection sensitivity. Based on the temporal persistence and spatial distribution range of the offset features, the degree of bio-wave resonance abnormality is objectively quantified, assisting in medical diagnostic decisions.

[0046] Furthermore, this application also aligns the fusion result with the baseline multidimensional feature set in terms of dimensions, uses multi-layer convolutional kernels to combine features of different dimensions to generate a multi-scale convolutional response feature set, and extracts key features that differ from the baseline state as offset features.

[0047] Furthermore, this method achieves deep feature mining of the fusion results through multi-scale convolution operations, which can simultaneously capture local details and global pattern differences, improve the recognition accuracy and robustness of offset features, and is especially suitable for detecting weak bio-wave resonance anomalies in complex electromagnetic environments.

[0048] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating a migration analysis method based on bio-wave resonance provided in this application embodiment;

[0051] Figure 2 A schematic diagram of a migration analysis system based on bio-wave resonance provided in this application embodiment;

[0052] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0053] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0054] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0055] Existing wavelet transform-based methods for analyzing bio-wave resonance have limitations: their time-frequency decomposition model with fixed basis functions is ill-suited to the non-stationary characteristics of biological signals, leading to the loss of effective frequency band information; the multi-source signal processing stages are fragmented, achieving data fusion only through simple superposition, failing to capture the dynamic coupling relationship between ECG, EEG, and micro-deformation data; and the step-by-step design of electromagnetic interference suppression and feature extraction results in decreased signal fidelity and spatiotemporal consistency, making them prone to misjudgment in complex electromagnetic environments. These shortcomings essentially stem from the decoupling processing mode of traditional methods for the multi-dimensional and nonlinear characteristics of bio-wave resonance.

[0056] To address the aforementioned issues, this application proposes a migration analysis method based on bio-wave resonance. It constructs a unified time-frequency reference by jointly aligning multi-lead signals and micro-deformation data, employs frequency band dynamic suppression technology to simultaneously eliminate electromagnetic interference while preserving effective resonance features, and innovatively introduces a cross-modal convolutional fusion mechanism to deeply couple and analyze ECG signal intensity changes, EEG phase continuity, and micro-deformation spatial propagation patterns, generating a multi-scale feature set with spatiotemporal correlation. Finally, it identifies abnormal migration features through comparison with a preset reference state. This method overcomes the limitations of traditional step-by-step processing, achieving integrated collaborative processing of interference suppression, feature fusion, and migration detection. It can accurately extract abnormal bio-wave resonance features even under strong electromagnetic interference environments, providing reliable technical support for the diagnosis of cardiocerebrovascular dysfunction.

[0057] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0058] Figure 1 A flowchart of a migration analysis method based on bio-wave resonance provided for embodiments of this application is shown below. Figure 1 As shown, the method includes:

[0059] Step 101: Collect electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformation data of the organism's surface, and perform joint alignment of the ECG signals, EEG signals, and micro-deformation data in the time-frequency domain.

[0060] In step 101, the electrocardiogram (ECG) signal represents a periodic electrical signal reflecting the electrical activity of the heart, including characteristic waveforms such as the P wave, QRS complex, and T wave. The electroencephalogram (EEG) signal represents a signal recording the electrical activity of neurons in the brain, exhibiting rhythmic fluctuations at different frequency bands (such as alpha and beta waves). The microdeformation data represents information on minute deformations on the surface of a biological organism acquired through a distributed fiber optic sensor array, reflecting muscle or tissue vibration. The time-frequency domain joint alignment represents the unified mapping of ECG, EEG, and microdeformation data at different sampling rates onto a time-frequency analysis framework, ensuring consistent timestamp and frequency band correspondences.

[0061] In this embodiment, electrocardiogram (ECG) and electroencephalogram (EEG) signals are first acquired simultaneously using multi-lead electrodes, while a distributed fiber optic sensor array is used to monitor micro-deformation data on the surface of the organism. Then, time-frequency transformation is performed on the three types of data to map the RR interval of the ECG signal, the frequency band energy distribution of the EEG, and the spatial fluctuation pattern of the micro-deformation to a unified time-frequency coordinate system. Finally, an interpolation method is used to align signals with different sampling rates to ensure the comparability of multimodal data at the same time point.

[0062] For example, at a hospital's cardio-brain integrated diagnosis and treatment center, data was collected from a subject. The sampling rate for electrocardiogram (ECG) signals was 500 Hz, for electroencephalogram (EEG) signals was 1000 Hz, and for micro-deformation data was 200 Hz. Short-time Fourier transform was used to convert the three types of signals to the time-frequency domain, and cubic spline interpolation was employed to align the time axis, ensuring that the T-wave peak of the ECG signal, the alpha band energy peak of the EEG, and the maximum strain point of the micro-deformation data matched within the same time window.

[0063] Step 102: Dynamically suppress the interference frequency bands of the jointly aligned ECG signal and the jointly aligned EEG signal respectively, and fuse the suppressed ECG signal, the suppressed EEG signal and the jointly aligned micro-deformation data to generate a fusion result.

[0064] In step 102, the dynamic suppression of interference frequency bands means adaptively filtering out electromagnetic noise based on the time-frequency characteristics of the signal, while retaining effective biometric frequency bands. The fusion result includes the integrated and suppressed ECG, EEG, and micro-deformation data, forming a dataset containing multi-dimensional time-frequency-spatial features.

[0065] In this embodiment, the high-frequency noise components that overlap with electromagnetic interference in the spectrum of the jointly aligned ECG signal are analyzed, and an adaptive band-stop filter is used to dynamically suppress the interference frequency band; for the EEG signal, abnormal jump frequency bands are identified and compensated and corrected based on phase continuity detection; the processed ECG and EEG signals and micro-deformation data are spliced ​​together according to time windows, and a fusion feature matrix is ​​generated after dimensionality reduction by principal component analysis.

[0066] For example, based on the above subject data, power frequency interference was detected in the 50-60Hz frequency band of the electrocardiogram (ECG) signal, which was suppressed using an adaptive notch filter; phase discontinuity was found in the 30-40Hz frequency band of the electroencephalogram (EEG) signal, which was repaired using a phase reconstruction algorithm; finally, the denoised ECG and EEG signals and micro-deformation data were merged with each 5-second time window to generate a 200-dimensional fusion feature vector.

[0067] Step 103: Use convolution operations to determine the offset features of the fusion result relative to the preset baseline state.

[0068] In step 103, the baseline state refers to the stable correlation pattern formed between electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformation data under normal physiological conditions in an organism. This pattern is established through statistical analysis of healthy samples or individual historical data and serves as a reference standard for judging whether the current signal has experienced abnormal deviation. The deviation feature represents the difference between the fusion result and the baseline state, reflecting abnormal bio-wave resonance.

[0069] In this embodiment, the fused feature matrix is ​​compared dimension-by-dimensionally with the standard feature matrix in the baseline state database; multi-dimensional difference responses are extracted through a three-layer convolutional kernel group (small-scale local feature kernel, medium-scale rhythmic feature kernel, and large-scale trend feature kernel); non-linear activation is applied to the convolutional output, and features exceeding twice the standard deviation are selected as offset features.

[0070] For example, a baseline database is constructed using data from 100 healthy subjects, and the mean μ and standard deviation σ of each dimension feature are calculated. After convolution operation, the current subject fusion data is output in the third layer, and the time dimension feature value is detected as μ+2.3σ and the spatial dimension feature value is μ+1.8σ, which are determined to be offset features.

[0071] Step 104: Based on the duration and spatial distribution range of the offset features in the time domain, generate a biowave resonance offset analysis report to determine whether the biowave resonance is abnormal.

[0072] In step 104, the duration of the offset feature in the time domain refers to the total duration of the offset feature appearing continuously on the time axis. This value is calculated by multiplying the number of frames in which the offset feature is detected consecutively in the multi-frame time-frequency analysis results by the corresponding time resolution of each frame. For example, if each frame is 0.1 seconds and appears consecutively for 15 frames, the duration is 1.5 seconds. The spatial distribution range of the offset feature refers to the proportion of the area covered by the offset feature on the surface of the organism. This value is obtained by statistically analyzing the ratio of the number of distributed fiber optic sensor nodes that trigger anomaly detection to the total number of nodes. For example, if 5 out of 8 nodes detect anomalies, the spatial distribution range is 62.5%. Biowave resonance refers to the coordinated vibration mode formed by the energy coupling of electrocardiogram signals, electroencephalogram signals, and microdeformations on the body surface through a specific frequency band. Its physical essence is the interaction phenomenon of cardiac electrical activity, brain nerve discharge, and muscle mechanical vibration at a specific frequency. The biowave resonance migration analysis report is a diagnostic conclusion document generated based on the spatiotemporal distribution characteristics of the migration features. It includes three core contents: abnormality level determination, location of the main affected area, and recommended follow-up cycle. Its generation is based on the mapping relationship between preset spatiotemporal threshold conditions and clinical diagnostic standards.

[0073] In this embodiment, the number of times the statistical offset feature appears within a continuous time window is converted into a standardized duration; the proportion of sensor nodes that activate anomaly detection to the total number of nodes in the array is calculated; and a structured report containing anomaly level and core area location is generated by combining the duration and spatial proportion.

[0074] For example, the deviation characteristics of the above subjects lasted for 12 ECG cycles (approximately 9.6 seconds) and covered 5 of the 8 sensor nodes on the back (62.5%). According to the judgment rule γ = 0.6 × 12 + 0.4 × 62.5 = 31.8, a "Level 2 Abnormality" report was generated, indicating that the abnormality was concentrated in the T5-T7 vertebral region.

[0075] This method addresses data heterogeneity through multimodal signal joint alignment, effectively eliminates complex electromagnetic interference using dynamic suppression technology, and precisely captures abnormal bio-wave resonance features through cross-modal convolutional fusion. The resulting offset analysis report provides objective evidence for the diagnosis of cardiocerebral synergistic dysfunction. Examples show that this method can accurately identify abnormal bio-wave resonance in specific areas of the subject's back, and the analysis results are consistent with clinical symptoms.

[0076] To address the accuracy issue in bio-wave resonance feature extraction, in some embodiments, step 103: determining the offset features of the fusion result relative to a preset reference state using convolution operations includes:

[0077] Step 201: Align the fusion result with the benchmark multidimensional feature set corresponding to the preset benchmark state in terms of dimensions.

[0078] In step 201, the benchmark multidimensional feature set specifically includes the standard electrocardiogram (ECG) signal time-frequency distribution pattern, the standard electroencephalogram (EEG) signal phase coupling pattern, and the standard micro-deformation data spatial propagation pattern of a healthy organism. These patterns are constructed by collecting multimodal signal data from at least 100 healthy subjects in a resting state, removing outliers, and taking the mean and variance range of each dimension feature. The representation is a structured feature matrix containing temporal stability indices, frequency domain energy distribution proportions, and spatial correlation coefficients. Dimensional alignment refers to establishing a one-to-one mapping relationship between each feature dimension of the fusion result to be tested and the benchmark template in terms of numerical range and physical meaning.

[0079] In this embodiment, the ECG feature dimension in the fusion result is first aligned with the standard ECG waveform in the benchmark template by amplitude normalization, so that the peak amplitude of the R wave matches the standard value. Next, the frequency band division of the EEG feature dimension is adjusted to be completely consistent with the frequency band division of the benchmark template. Finally, spatial interpolation is performed on the micro-deformation features to unify their sensor node distribution density with the benchmark template. After completing the parameter matching of these three dimensions, standard aligned data that can be directly compared is formed.

[0080] Step 202: Combine the features of different dimensions in the aligned fusion result by using multi-layer convolutional kernels to generate a multi-scale convolutional response feature set.

[0081] In step 202, a multi-layer convolutional kernel refers to a group of convolutional operation units with different sensing ranges, including local detail sensing kernels, regional feature sensing kernels, and global trend sensing kernels. A multi-scale convolutional response feature set refers to a combination of features extracted by convolutional kernels of different scales that can reflect local anomalies and overall shift patterns in the signal.

[0082] In this embodiment, a local detail-aware kernel is first used to scan the aligned electrocardiogram (ECG) signal point by point to capture subtle waveform distortion features. Next, a regional feature-aware kernel is used to analyze energy transfer patterns between EEG frequency bands. Finally, a global trend-aware kernel is used to assess the propagation path offset of the micro-deformation data on the body surface. The convolutional outputs of the three scales are concatenated according to feature dimensions to form a multi-scale response set containing detail features, regional features, and global features.

[0083] Step 203: Extract key features from the multi-scale convolutional response feature set and use the key features as offset features.

[0084] In step 203, the key features refer to the feature components in the multi-scale convolution response feature set that differ from the preset baseline state by more than two standard deviations. Their physical meaning reflects the degree of rhythm disorder of the body's electrocardiogram signal, the intensity of phase desynchronization of the brain wave, and the spatial range of the abnormal area of ​​micro-deformation propagation. These feature components are manifested as periodic signal destruction in the time domain, abnormal energy accumulation in the characteristic frequency band in the frequency domain, and spatially show that the deformation conduction path deviates from the normal pattern.

[0085] In this embodiment, the Euclidean distance between each dimension of the multi-scale response set and the baseline template is first calculated; then, feature dimensions with distance values ​​exceeding twice the standard deviation are selected; finally, these abnormal features are clustered into three categories: electrocardiogram, electroencephalogram, and microdeformation, and the feature with the largest distance value in each category is extracted as the final offset feature.

[0086] Here is a specific example:

[0087] In a practical application at a hospital's cardio-brain collaborative diagnosis and treatment center, for the aforementioned subject case where signal acquisition and fusion had been completed, the 200-dimensional fusion feature vector generated by the subject was first aligned dimensionally with the standard feature template in the benchmark library. Specifically, the amplitude of the ECG feature dimension was scaled to the mean level of the benchmark library, ensuring that the frequency band division of the EEG feature dimension was completely consistent with the benchmark library, and spatial standardization was performed on the micro-deformation feature dimension. Next, a three-layer convolutional kernel group was used to analyze the aligned feature vector. The first layer used a small-scale convolutional kernel to extract local distortion features of the ECG waveform; the second layer used a medium-scale convolutional kernel to capture energy transfer patterns between EEG frequency bands; and the third layer used a large-scale convolutional kernel to evaluate the overall offset of the micro-deformation conduction path, ultimately generating a multi-scale convolutional response set containing spatiotemporal multi-dimensional features. When extracting feature values ​​from this set, it was found that the ECG ST segment feature value reached the baseline mean plus 2.3 standard deviations. This value was obtained by calculating the difference between the current subject's ST segment amplitude and the mean of 100 healthy data in the baseline database, and then dividing by the baseline database standard deviation. The EEG α-β energy ratio feature value exceeded the baseline mean by 1.8 standard deviations, and the micro-deformation conduction delay feature value exceeded the baseline mean by 2.1 standard deviations. These key features were identified as deviation features.

[0088] In this embodiment, the proposed method ensures data comparability through precise alignment of multi-dimensional features, comprehensively captures anomalous patterns of bio-wave resonance using multi-scale convolution, and ultimately extracts key offset features with clear physiological significance and clinical diagnostic value, thereby improving the reliability and practicality of bio-wave resonance analysis.

[0089] To further improve the comprehensiveness of multi-scale feature extraction, in some embodiments, step 202: combining the features of different dimensions in the aligned fusion result through multi-layer convolutional kernels to generate a multi-scale convolutional response feature set, including:

[0090] Step 301: Use the first-layer convolutional kernel to slide and combine the temporal dimension features in the aligned fusion result to generate a primary response feature set.

[0091] In step 301, the first convolutional kernel refers to a convolutional operation unit focused on extracting temporal features, and its sensing range is set to cover a complete electrocardiogram cycle. The primary response feature set refers to the feature set extracted by this convolutional kernel that reflects the temporal variation law of the signal, including two subclasses: waveform morphology features and rhythm features.

[0092] In this embodiment, the aligned fusion result is first input into the first convolutional kernel, which slides along the time axis to scan the ECG signal dimension and extracts three key parameters—the waveform's rising slope, falling slope, and extreme point interval—within each scanning window. Simultaneously, the EEG signal dimension is scanned synchronously to extract the phase consistency features of energy changes in each frequency band. Finally, the feature parameters of the two types of signals are arranged in chronological order to form a primary response feature set that reflects the temporal evolution of the signal.

[0093] Step 302: Use a second-layer convolutional kernel to combine the spatial dimension features in the aligned fusion result across regions to generate a set of secondary response features.

[0094] In step 302, the second-layer convolutional kernel refers to a convolutional operation unit designed for spatial dimension features, whose sensing range covers at least three adjacent micro-deformation sensor nodes. The secondary response feature set refers to the feature set extracted by this convolutional kernel that reflects the deformation propagation law, including two subclasses: transmission velocity features and spatial cooperative features.

[0095] In this embodiment, the aligned micro-deformation data is input into the second convolutional kernel, which slides within the spatial range of the body surface sensor network to calculate the time difference of deformation peak transmission between adjacent nodes as a transmission velocity feature; at the same time, the coordination of deformation amplitude among multiple nodes is analyzed as a spatial coordination feature; finally, the two types of features are arranged according to spatial topological relationships to form a set of secondary response features that reflect the spatial distribution law of deformation.

[0096] Step 303: Superimpose the primary response feature set with the secondary response feature set to form a multi-scale convolutional response feature set.

[0097] In step 303, superposition refers to the operation of fusing temporal and spatial features according to biomechanical correlation rules. Its physical essence is to establish a coupling relationship between electrophysiological activity and mechanical vibration.

[0098] In this embodiment, the ECG waveform features in the primary response feature set are first paired with the conduction velocity features in the secondary response feature set to establish a correlation model between cardiac electrical activity and body surface vibration conduction; then, the EEG phase features are paired with deformation synergy features to construct a synchronization model between nerve discharge and muscle vibration; finally, the output features of the two correlation models are merged according to weights to form a multi-scale feature set with spatiotemporal dual representation capabilities.

[0099] Here is a specific example:

[0100] In the multi-scale feature extraction process of the aforementioned subjects at a hospital's Cardio-Brain Collaborative Diagnosis and Treatment Center, the first layer of convolutional kernels was used to process the aligned 200-dimensional fused feature vectors. This convolutional kernel slid along the time axis with a window size covering two ECG cycles, extracting the feature where the R-wave rise slope in the ECG signal decreased to 0.35 mV / s. This value was calculated by comparing it with the average rise slope of 0.65 mV / s in healthy subjects in the benchmark database. At the same time, the phase consistency coefficient of the alpha wave in the EEG signal was extracted to be 0.72, which was 37% lower than the benchmark average of 1.15. Next, the second layer of convolutional kernels was used to analyze the spatial features of micro-deformation. This convolutional kernel covered the region of five adjacent sensor nodes and detected a deformation conduction velocity of 1.2 m / s in the T5-T7 vertebral region, which was lower than the benchmark average of 1.8 m / s. The velocity difference was obtained by calculating the ratio of the transmission time difference of the deformation peak between nodes to the node spacing. Finally, the temporal and spatial features are superimposed. Specifically, the fusion weight is calculated using the formula ω = 0.7 × temporal feature difference + 0.3 × spatial feature difference. The temporal feature difference is the mean difference between the ECG slope and the EEG phase coefficient, and the spatial feature difference is the difference in deformation conduction velocity. The resulting multi-scale response feature set marks two key patterns: "ECG-EEG synergistic weakening" and "vertebral conduction block".

[0101] In this embodiment of the application, the step scheme achieves the collaborative extraction of fine temporal features and macroscopic spatial features through hierarchical convolution. The established spatiotemporal correlation model can accurately capture the multidimensional manifestations of biological wave resonance anomalies, providing feature basis for clinical diagnosis that combines local details and overall trends.

[0102] To further improve the accuracy of multi-scale feature fusion, in some embodiments, step 303: superimposing the primary response feature set with the secondary response feature set to form a multi-scale convolutional response feature set includes:

[0103] Step 401: Adjust the values ​​of each time-domain feature in the primary response feature set according to the numerical fluctuation range of each time-domain feature in the primary response feature set.

[0104] In step 401, the numerical fluctuation range refers to the normal variation range of the time-domain feature under healthy conditions, obtained through statistics from the benchmark library. Numerical adjustment refers to scaling the amplitude of the feature to be tested to the same dimension range as the benchmark feature, which facilitates subsequent similarity calculation.

[0105] In this embodiment, the normal fluctuation limits of each feature point of the electrocardiogram (ECG) waveform are first extracted from the benchmark library. The parameters such as the R-wave amplitude and T-wave height of the ECG signal to be tested are then proportionally adjusted to the benchmark dimensions. Simultaneously, the energy values ​​of each frequency band of the electroencephalogram (EEG) signal are standardized to ensure their distribution within a reasonable range defined by the benchmark library. The adjusted time-domain features retain the original trend of change and are suitable for direct comparison with benchmark features.

[0106] Step 402: Adjust the values ​​of each spatial dimension feature in the secondary response feature set according to the distribution density of each spatial dimension feature in the secondary response feature set.

[0107] In step 402, the distribution density refers to the frequency of spatial features appearing in the body surface sensor network. The features involved in this application include the range of numerical fluctuations of temporal dimension features, spatial dimension feature distribution density, numerically adjusted primary response feature set, numerically adjusted secondary response feature set, baseline temporal dimension feature set, baseline spatial dimension feature set, first similarity, second similarity, and multi-scale convolutional response feature set formed by linear combination. These features together constitute a complete technology chain from signal processing to feature fusion.

[0108] In this embodiment, the distribution density of sensor nodes for micro-deformation features in each region of the back is first statistically analyzed. Lower weights are assigned to the deformation conduction velocity features in high-density regions, while higher weights are assigned to the same features in low-density regions. Simultaneously, the deformation amplitude coordination features are region-normalized to eliminate biases caused by uneven sensor distribution. The adjusted spatial features more objectively reflect the true biomechanical state.

[0109] Step 403: Determine the first similarity between the numerically adjusted primary response feature set and the benchmark time-domain dimension feature set corresponding to the preset benchmark state, and determine the second similarity between the numerically adjusted secondary response feature set and the benchmark spatial dimension feature set corresponding to the preset benchmark state.

[0110] In step 403, the first similarity is the degree of matching between the temporal features and the reference temporal features, obtained through a comprehensive evaluation of the feature waveform morphology and rhythmic characteristics. The second similarity is the degree of matching between the spatial features and the reference spatial features, obtained through a comparative analysis of deformation transmission paths and regional coordination.

[0111] In this embodiment, the waveform correlation coefficient between the adjusted ECG features and the baseline ECG template is calculated as the first principal component of similarity, while the overlap area of ​​the energy distribution of the EEG frequency bands is calculated as an auxiliary indicator. The path conformity between the adjusted micro-deformation features and the baseline spatial pattern is evaluated as the second core parameter of similarity, and correction is made in conjunction with the regional coordination difference. The two types of similarity quantify the degree of deviation between the current state and the health baseline from the temporal and spatial dimensions, respectively.

[0112] Step 404: Based on the first similarity and the second similarity, the numerically adjusted primary response feature set and the numerically adjusted secondary response feature set are linearly combined to form a multi-scale convolutional response feature set.

[0113] In step 404, linear combination refers to the process of weighted fusion of two types of features based on the relative importance of temporal and spatial similarities.

[0114] In this embodiment, when the first similarity indicates a significant anomaly in the temporal features, the weight of the temporal features in the combination is increased; when the second similarity reflects a severe deviation in the spatial features, the contribution ratio of the spatial features is increased. In the final multi-scale feature set, each feature dimension contains evaluation results from both temporal and spatial perspectives, and the weight allocation is automatically adapted to the abnormal performance.

[0115] Here is a specific example:

[0116] In the process of multi-scale feature fusion of the aforementioned subjects at a cardiocerebrovascular collaborative diagnosis and treatment center of a certain hospital, the extracted primary response features were first numerically adjusted. The ECG R-wave rise slope of 0.35 mV / s was linearly normalized to a per-unit value of 0.25 according to the baseline healthy range of 0.4-0.9 mV / s. This value was calculated using the formula (measured value - lower limit) / (upper limit - lower limit). At the same time, the EEG alpha wave phase consistency coefficient of 0.72 was adjusted to a per-unit value of 0.4 according to the baseline range of 0.8-1.5. Next, the deformation conduction velocity of 1.2 m / s in the secondary response features was spatially density-weighted. Considering the dense sensor distribution in the T5-T7 region, a density coefficient of 0.8 was introduced to obtain a weighted velocity value of 0.96, which was then normalized to a per-unit value of 0.32 according to the baseline range of 1.0-2.5 m / s. Next, the similarity between the adjusted temporal features and the baseline temporal template was calculated. The correlation coefficient of the ECG waveform was 0.68, and the overlap area of ​​the EEG frequency bands was 0.71. The average of the two, 0.695, was taken as the first similarity. The path fit of the adjusted spatial features and the baseline spatial pattern was calculated to be 0.55, and the coordination matching degree was 0.62. The average of the two, 0.585, was taken as the second similarity. Finally, based on the similarity results, according to the principle that the temporal feature weight = 1 - the first similarity and the spatial feature weight = 1 - the second similarity, the temporal weight was determined to be 0.305, and the spatial weight was determined to be 0.415. The final multi-scale response feature set was generated by using the linear combination formula γ = 0.305 × temporal feature + 0.415 × spatial feature.

[0117] In this embodiment of the application, the step scheme realizes the intelligent fusion of temporal and spatial features through a dynamic weight allocation mechanism. The generated multi-scale feature set can not only comprehensively reflect the biological wave resonance state, but also highlight the main abnormal patterns, providing a hierarchical feature basis for clinical diagnosis.

[0118] To further improve the anti-interference capability of biological signals, in some embodiments, step 102: the dynamic suppression of interference frequency bands of the jointly aligned ECG signal and the jointly aligned EEG signal includes:

[0119] Step 501: Establish the first time-frequency distribution model of the jointly aligned electrocardiogram signal and the second time-frequency distribution model of the jointly aligned electroencephalogram signal.

[0120] In step 501, the first time-frequency distribution model refers to the energy distribution representation of the electrocardiogram signal in the time-frequency two-dimensional plane, including two elements: waveform feature points and rhythm feature bands. The second time-frequency distribution model refers to the energy concentration characteristics of the electroencephalogram signal in the time-frequency domain, which is reflected in the phase coupling relationship of different rhythm bands.

[0121] In this embodiment, the jointly aligned electrocardiogram (ECG) signal is first subjected to time-frequency transformation to extract the energy distribution patterns of feature points such as R waves and T waves on the time-frequency plane, while simultaneously analyzing the distribution of rhythmic feature bands corresponding to the RR interval. Synchronous time-frequency analysis is then performed on the EEG signal to delineate the phase coupling regions of feature bands such as α waves and β waves on the time-frequency plane. By establishing these two time-frequency distribution models, an analytical basis is provided for subsequent interference frequency band identification.

[0122] Step 502: Extract the first target frequency band related to electromagnetic interference from the first time-frequency distribution model, and extract the second target frequency band related to electromagnetic interference from the second time-frequency distribution model.

[0123] In step 502, the first target frequency band refers to the characteristic region in the electrocardiogram signal that overlaps with the electromagnetic interference spectrum, which is identified through interference feature matching. The second target frequency band refers to the characteristic waveband in the electroencephalogram signal affected by electromagnetic interference, which is determined through abnormal phase change detection.

[0124] In this embodiment, by comparing the similarity between the time-frequency distribution of the electrocardiogram (ECG) signal and the known electromagnetic interference spectrum, the high-frequency region of the QRS complex is located as the first target frequency band; by analyzing the time-frequency phase continuity of the electroencephalogram (EEG), the β-band phase jump region is identified as the second target frequency band. The determination of these two target frequency bands comprehensively considers the coupling relationship between interference characteristics and biological signal features.

[0125] Step 503: Using an adaptive filtering algorithm, the electrocardiogram (ECG) signal in the first target frequency band is dynamically suppressed to form a suppressed ECG signal, and the electroencephalogram (EEG) signal in the second target frequency band is dynamically suppressed to form a suppressed EEG signal.

[0126] In step 503, dynamic suppression processing refers to a filtering method that adjusts in real time according to the intensity of interference. Its core is to eliminate interference components while preserving effective biological characteristics.

[0127] In this embodiment, an adaptive notch filtering technique based on reference noise is used for the first target frequency band of the electrocardiogram (ECG) signal to dynamically track and precisely suppress interference frequency changes; a phase-preserving filter is applied to the second target frequency band of the electroencephalogram (EEG) signal to eliminate interference while maintaining the phase continuity of the normal EEG rhythm. The processed signal retains the original waveform morphology and rhythm characteristics, with only the identified interference components removed.

[0128] Here is a specific example:

[0129] In the signal processing of the aforementioned subjects at a hospital's Cardio-Brain Collaborative Diagnosis and Treatment Center, a short-time Fourier transform was first performed on the time-frequency aligned electrocardiogram (ECG) signals. When establishing the first time-frequency distribution model, an energy peak was found in the 49-51 Hz frequency band, with an amplitude eight times the baseline noise level. This value was obtained by calculating the ratio of the average power of the interference band to the baseline noise power. Simultaneously, a second time-frequency distribution model was established for the EEG signals, showing a phase abrupt change point in the 32-38 Hz frequency band, with a phase difference standard deviation reaching 0.35 radians, significantly larger than the healthy baseline mean of 0.12 radians. To address the power frequency interference in the ECG signals, an adaptive notch filter was used for dynamic suppression. The notch center frequency was automatically adjusted within the 49.8-50.2 Hz range based on real-time spectrum analysis results, the notch bandwidth was set to 2 Hz, and the notch depth was dynamically adjusted according to the interference intensity, achieving a maximum attenuation of -40 dB. For the phase discontinuity problem in the EEG signals, a reconstruction algorithm based on minimum phase error was used to repair the phase difference standard deviation in the 32-38 Hz frequency band to the 0.15 radian level. After processing, the signal-to-noise ratio of the ECG signal in the 50Hz band improved from 8:1 to 32:1, which was obtained by calculating the ratio of effective signal power to residual noise power; the phase continuity index of the EEG signal in the 35Hz band recovered from 0.65 to 0.92.

[0130] In this embodiment of the application, the proposed method achieves intelligent identification of interference frequency bands by establishing an accurate time-frequency distribution model, effectively removes electromagnetic interference by adopting a differentiated dynamic suppression strategy, and at the same time protects the effective characteristics of biological signals to the maximum extent, thereby improving the reliability of biological wave resonance analysis in complex electromagnetic environments.

[0131] To further improve the accuracy of multimodal data fusion, in some embodiments, step 102: fusing the suppressed ECG signal, the suppressed EEG signal, and the jointly aligned micro-deformation data to generate a fusion result includes:

[0132] Step 601: Map the suppressed ECG signal to a first multidimensional feature vector, map the suppressed EEG signal to a second multidimensional feature vector, and map the jointly aligned micro-deformation data to a third multidimensional feature vector.

[0133] In step 601, the first multidimensional feature vector refers to the structured representation formed after feature extraction of the electrocardiogram signal, which includes three dimensions: waveform amplitude features, interval features, and spectral features. The second multidimensional feature vector refers to the feature expression of the electroencephalogram signal, which consists of rhythm energy distribution, phase coupling strength, and frequency band transition probability. The third multidimensional feature vector refers to the feature encoding of micro-deformation data, which includes three components: deformation amplitude, conduction velocity, and spatial coordination.

[0134] In this embodiment, the R-wave amplitude, ST-segment slope, and T-wave area are first extracted from the suppressed ECG signal as waveform features, the RR interval and QT interval are calculated as rhythm features, and the energy proportion of the 0-40Hz frequency band is analyzed as spectral features, combining them to form a first multidimensional feature vector. For the suppressed EEG signal, the relative energy ratios of the α, β, and θ bands are calculated as rhythm features, and the cross-band phase-locking value is extracted as a coupling feature, constructing a second multidimensional feature vector. For micro-deformation data, the maximum deformation of each sensor node, the deformation wave propagation time difference, and the deformation correlation between adjacent nodes are measured to form a third multidimensional feature vector.

[0135] Step 602: Construct cross-modal association rules based on the signal intensity change pattern in the first multidimensional feature vector, the phase continuity pattern in the second multidimensional feature vector, and the deformation propagation pattern in the third multidimensional feature vector.

[0136] In step 602, the signal intensity variation pattern refers to the periodic fluctuation characteristics of the ECG signal in the time-frequency domain, which is identified by extracting the extreme point distribution interval and amplitude variation gradient of the signal envelope. The phase continuity pattern refers to the smooth transition characteristics of the phase angle of the EEG signal between different frequency bands, which is identified by calculating that the standard deviation of the phase difference within adjacent time windows is less than a set threshold. The deformation propagation pattern refers to the temporal transmission characteristics of micro-deformation data between sensor nodes on the surface of a biological organism, which is identified by analyzing the spatial progression relationship of the signal peak occurrence time at each node. The cross-modal association rule refers to establishing a mathematical model of the relationship between ECG, EEG, and micro-deformation features, which is achieved through temporal synchronization analysis and frequency coupling evaluation.

[0137] In this embodiment, the temporal correlation between the ECG R-wave amplitude variation in the first multidimensional feature vector and the α-wave energy fluctuation in the second multidimensional feature vector is analyzed to establish a cardiac-brain electrocoupling rule. The frequency domain coherence between the deformation conduction velocity in the third multidimensional feature vector and the ECG ST-segment slope in the first multidimensional feature vector is studied to construct an electrocardiogram-mechanical correlation rule. The correspondence between the β-wave phase in the second multidimensional feature vector and the spatial consistency in the third multidimensional feature vector is explored to form an electroencephalogram-mechanical correlation rule. These three types of rules together constitute a cross-modal correlation system.

[0138] Step 603: Based on the cross-modal association rule, the first multidimensional feature vector, the second multidimensional feature vector, and the third multidimensional feature vector are superimposed to generate a multidimensional feature set, which is the fusion result.

[0139] In step 603, overlay refers to the process of organically integrating feature vectors from different modalities according to the weights determined by the association rules. The multidimensional feature set refers to the final generated feature matrix that contains the fusion results of multi-source information.

[0140] In this embodiment, the fusion weights of ECG and EEG features are determined according to the ECG-EMG coupling rule. The combination ratio of ECG and micro-deformation features is adjusted according to the ECG-mechanical association rule. The contribution of EEG and micro-deformation features is balanced according to the EEG-mechanical association rule. The three types of feature vectors are linearly superimposed according to the optimized weights to generate a multi-dimensional feature set that simultaneously contains electrophysiological and mechanical features.

[0141] Here is a specific example:

[0142] In the process of multimodal data fusion of the aforementioned subjects at a hospital's cardio-brain collaborative diagnosis and treatment center, the suppressed electrocardiogram (ECG) signals were first processed by extracting features in 5-second time windows to obtain a first multidimensional feature vector containing 30 parameters, including R-wave amplitude (1.85 mV), ST-segment slope (-0.06 mV / s), and T-wave area (1.2 mV / s). The R-wave amplitude was obtained by measuring the maximum peak value of the waveform, and the ST-segment slope was calculated using linear regression. Simultaneously, a second multidimensional feature vector was constructed from the suppressed EEG signals, containing 25 parameters, including α-wave energy percentage (38%), β-wave energy ratio (1.25), and α-β phase coupling (0.68). The phase coupling was calculated using Hilbert transform to determine the standard deviation of the instantaneous phase difference. Microdeformation data generated a third multidimensional feature vector containing 20 parameters, including T5 node strain (0.15%), T6-T7 conduction time difference (120 ms), and spatial coordination coefficient (0.72). The coordination coefficient was obtained by calculating the correlation between strain waveforms of adjacent nodes using the Pearson correlation coefficient formula.

[0143] Next, cross-modal association rules were constructed. Calculations revealed a correlation coefficient of 0.81 between ECG R-wave amplitude and EEG alpha wave energy, obtained by standardizing the two sets of feature vectors and calculating their covariance. The mutual information between ECG ST-segment slope and micro-deformation conduction time was 0.58, calculated using probability distribution statistics. The coherence function value between EEG beta wave phase and micro-deformation coordination coefficient was 0.63, obtained through spectral analysis based on Fourier transform. Based on these association metrics, the fusion weight allocation scheme was determined as follows: ECG feature weight 0.45, EEG feature weight 0.35, and micro-deformation feature weight 0.2.

[0144] Finally, a multidimensional feature set containing 75 effective features was generated according to the linear combination formula γ = 0.45 × ECG feature + 0.35 × EEG feature + 0.2 × microdeformation feature. Two key patterns were marked in this set: "ECG ST segment abnormality - microdeformation delay" and "EEG α-β imbalance".

[0145] In this embodiment of the application, the step scheme realizes the intelligent fusion of multi-source features by establishing cross-modal association rules. The generated feature set not only retains the independent characteristics of each modal signal, but also reflects the synergistic effect of bio-wave resonance, providing a comprehensive and reliable data foundation for subsequent migration analysis.

[0146] To further improve the accuracy of bio-wave resonance anomaly determination, in some embodiments, step 104: generating a bio-wave resonance migration analysis report on whether bio-wave resonance is abnormal based on the duration and spatial distribution range of the migration feature in the time domain dimension, includes:

[0147] Step 701: Calculate the cumulative time length of the N consecutive occurrences of the offset feature in the time domain dimension, and the ratio of the area of ​​the organism surface covered to the whole organism surface in the spatial dimension, where N is greater than or equal to 3.

[0148] In step 701, "N consecutive occurrences" means that the offset feature continuously appears for at least three complete ECG cycles in the time domain without interruption, ensuring the reliability of the abnormal feature. The cumulative duration refers to the total duration of the abnormal feature's existence, measured in ECG cycles. The proportion refers to the ratio of the body surface area covered by the abnormal feature to the total monitored area, calculated using the number of activated sensor nodes.

[0149] In this embodiment, the system first detects whether the offset features appear consecutively in three or more adjacent analysis windows in the time domain, and then calculates the total time span of these windows as the cumulative length. Simultaneously, the system counts the number of sensor nodes that trigger anomaly detection in the spatial dimension, and divides this number by the total number of nodes to obtain the distribution ratio. These two parameters quantify the degree of diffusion of the anomaly features in both time and spatial dimensions.

[0150] Step 702: Based on the first comparison result of the cumulative time length and the preset length threshold, and the second comparison result of the ratio and the preset ratio threshold, generate a determination result of whether the bio-wave resonance is abnormally offset.

[0151] In step 702, the preset length threshold is the upper limit of normal fluctuations statistically derived from baseline data of healthy individuals. The preset proportion threshold is the minimum range standard for abnormal areas determined clinically. The judgment result refers to the grading conclusion given by comprehensively considering the degree of abnormality in both spatiotemporal dimensions.

[0152] In this embodiment, the cumulative length is compared with a preset threshold of three ECG cycles; if it exceeds the threshold, a temporal abnormality is determined. Simultaneously, the distribution ratio is compared with a preset standard of 20%; if it exceeds the standard, a spatial abnormality is determined. A positive result is determined when both temporal and spatial dimensions are abnormal; a suspicious result is determined when only one dimension is abnormal; and a negative result is determined when neither exceeds the limit.

[0153] Step 703: Generate a bio-wave resonance migration analysis report based on the judgment results.

[0154] In this embodiment, a corresponding report is generated based on the level of the judgment result: a positive result generates a Level 1 report containing a heatmap of the abnormal area, a suspicious result generates a Level 2 report marking the suspicious area, and a negative result generates a Level 3 report with a list of basic parameters. All reports include a recommended follow-up period based on the duration of the abnormality. The specific process for generating a bio-wave resonance analysis report based on the judgment results includes: First, converting the cumulative time length of the offset feature in the time domain into standard time units, where the standard time unit is determined by the average fluctuation period of normal bio-wave resonance signals in the healthy sample library, specifically taking a value of 0.5 seconds; Second, calculating the proportion of the biological surface area covered by the offset feature, which is obtained by the ratio of the number of sensor nodes covered by the abnormal signal to the total number of nodes; Then, determining the temporal anomaly coefficient α based on the ratio of the cumulative time length to the standard time unit, and determining the spatial anomaly coefficient β based on the area proportion, where α = cumulative time length / standard time unit, and β = area proportion × 100; Finally, substituting the temporal anomaly coefficient α and the spatial anomaly coefficient β into the grading judgment formula γ = 0.6α + 0.4β, generating a level 1 anomaly report when γ is greater than the preset threshold of 15, a level 2 anomaly report when γ is between 10 and 15, and a normal report when γ is less than 10. The report content includes the anomaly level, a map of the main affected area location, and a recommended re-examination time. Specific Implementation: In an application scenario at a tertiary hospital's cardiocerebrovascular integrated diagnosis and treatment center, this method was implemented on a 45-year-old subject, collecting multi-lead electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and back micro-deformation data over 30 minutes. Analysis revealed a sustained 25 standard time unit shift in the ECG signal within the 0.8-1.2Hz frequency band, covering 7 out of 12 sensor nodes on the back. Calculations yielded α = 25, β = 58.33, and γ = 0.6 × 25 + 0.4 × 58.33 = 37.33. Since 37.33 exceeded the threshold of 15, the system automatically generated a Level 1 abnormality report, indicating that the abnormality was concentrated in the T4-T6 vertebral region of the back, and recommending a follow-up MRI within 24 hours. The standard time unit of 0.5 seconds in this report was derived from the average fluctuation cycle of 1000 samples statistically analyzed from the hospital's health database, and the threshold of 15 was determined through a retrospective analysis of 200 clinically diagnosed abnormal cases.

[0155] Here is a specific example:

[0156] During the analysis of the aforementioned subjects at a hospital's Cardiovascular and Neurological Collaborative Diagnosis and Treatment Center, the system first counted the continuous appearance of the deviation feature across five consecutive analysis windows, with a total duration of approximately 9.6 seconds (12 ECG cycles). This value was calculated by multiplying the duration of each window (0.8 seconds) by the number of windows (5). Simultaneously, abnormal features were detected covering five of the eight sensor nodes on the back, with a spatial distribution ratio of 62.5%, calculated by dividing the number of abnormal nodes by the total number of nodes. The duration was compared to the clinically set threshold of three cycles, exceeding the standard by three times; the distribution ratio was compared to the threshold of 20%, exceeding it by 2.1 times. Based on the judgment rule γ = 0.6 × (actual number of cycles / threshold number of cycles) + 0.4 × (actual ratio / threshold ratio), the calculated γ = 0.6 × 12 / 3 + 0.4 × 62.5 / 20 = 2.4 + 1.25 = 3.65. According to the hospital's grading standards: γ greater than 3 indicates a Level 1 abnormality, between 2 and 3 a Level 2 abnormality, and less than 2 a Level 3 abnormality. The system automatically generated a Level 1 abnormality report.

[0157] In this embodiment of the application, the step scheme achieves objective grading of the biowave resonance state by quantitatively assessing the spatiotemporal distribution characteristics of abnormal features. The generated diagnostic report not only contains clear abnormal location but also provides personalized follow-up suggestions, thereby improving the accuracy and practicality of clinical diagnosis and treatment.

[0158] Figure 2 A schematic diagram of a migration analysis system based on bio-wave resonance provided in this application embodiment is shown below. Figure 2 As shown, the system includes:

[0159] The acquisition module 21 is used to acquire electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformation data on the surface of the organism, and to jointly align the ECG signals, EEG signals, and micro-deformation data in the time-frequency domain.

[0160] The fusion module 22 is used to dynamically suppress the interference frequency bands of the jointly aligned electrocardiogram signal and the jointly aligned electroencephalogram signal, respectively, and to fuse the suppressed electrocardiogram signal, the suppressed electroencephalogram signal, and the jointly aligned micro-deformation data to generate a fusion result.

[0161] The determination module 23 is used to determine the offset features of the fusion result relative to the preset reference state using convolution operations.

[0162] The generation module 24 is used to generate a biowave resonance migration analysis report on whether the biowave resonance is abnormal, based on the duration and spatial distribution range of the migration feature in the time domain dimension.

[0163] Figure 2 The aforementioned migration analysis system based on bio-wave resonance can perform... Figure 1The implementation principle and technical effects of the bio-wave resonance-based migration analysis method described in the illustrated embodiment will not be repeated here. The specific operation methods of each module and unit in the bio-wave resonance-based migration analysis system described in the above embodiments have been detailed in the embodiments related to this method, and will not be elaborated upon here.

[0164] In one possible design, Figure 2 The migration analysis system based on bio-wave resonance shown in the embodiment can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0165] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0166] The processing component 32 is used to perform the above. Figure 1 The embodiment describes a migration analysis method based on bio-wave resonance.

[0167] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0168] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0169] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0170] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0171] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0172] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0173] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown is a migration analysis method based on bio-wave resonance.

[0174] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0175] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0176] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A migration analysis system based on bio-wave resonance, characterized in that, include: The acquisition module is used to acquire electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and micro-deformation data on the surface of the organism, and to jointly align the ECG signals, EEG signals, and micro-deformation data in the time-frequency domain. The fusion module is used to dynamically suppress the interference frequency bands of the jointly aligned ECG signal and the jointly aligned EEG signal, respectively, and to fuse the suppressed ECG signal, the suppressed EEG signal and the jointly aligned micro-deformation data to generate a fusion result. The determination module is used to determine the offset features of the fusion result relative to a preset baseline state using convolution operations; The generation module is used to generate a biowave resonance migration analysis report on whether the biowave resonance is abnormal, based on the duration and spatial distribution range of the offset features in the time domain dimension. Biowave resonance refers to the coordinated vibration mode formed by the energy coupling between electrocardiogram signals, electroencephalogram signals and micro-deformations of the body surface through a specific frequency band. The method of determining the offset features of the fusion result relative to a preset baseline state using convolution operations includes: The fusion result is dimensionally aligned with the benchmark multidimensional feature set corresponding to the preset benchmark state. By combining features of different dimensions in the aligned fusion result through multiple convolutional kernels, a multi-scale convolutional response feature set is generated. Key features are extracted from the multi-scale convolutional response feature set, and these key features are used as offset features. The step involves combining features of different dimensions from the aligned fusion result using multiple convolutional kernels to generate a multi-scale convolutional response feature set, including: The first-layer convolutional kernel is used to slide and combine the temporal dimension features in the aligned fusion result to generate a primary response feature set. A second-layer convolutional kernel is used to combine the spatial dimension features in the aligned fusion result across regions to generate a set of secondary response features; The primary response feature set and the secondary response feature set are superimposed to form a multi-scale convolutional response feature set.

2. The system according to claim 1, characterized in that, The step of superimposing the primary response feature set with the secondary response feature set to form a multi-scale convolutional response feature set includes: Based on the numerical fluctuation range of each time-domain feature in the primary response feature set, the numerical values ​​of each time-domain feature in the primary response feature set are adjusted. Based on the distribution density of each spatial dimension feature in the secondary response feature set, the numerical values ​​of each spatial dimension feature in the secondary response feature set are adjusted. Determine the first similarity between the numerically adjusted primary response feature set and the benchmark time-domain feature set corresponding to the preset benchmark state, and determine the second similarity between the numerically adjusted secondary response feature set and the benchmark spatial-dimensional feature set corresponding to the preset benchmark state; Based on the first similarity and the second similarity, the numerically adjusted primary response feature set and the numerically adjusted secondary response feature set are linearly combined to form a multi-scale convolutional response feature set.

3. The system according to claim 1, characterized in that, The dynamic suppression of interference frequency bands in the jointly aligned ECG and EEG signals includes: Establish a first time-frequency distribution model of the jointly aligned electrocardiogram signal and a second time-frequency distribution model of the jointly aligned electroencephalogram signal; Extract a first target frequency band related to electromagnetic interference from the first time-frequency distribution model, and extract a second target frequency band related to electromagnetic interference from the second time-frequency distribution model; An adaptive filtering algorithm is used to dynamically suppress the electrocardiogram (ECG) signal in the first target frequency band to form a suppressed ECG signal, and the electroencephalogram (EEG) signal in the second target frequency band is also dynamically suppressed to form a suppressed EEG signal.

4. The system according to claim 1, characterized in that, The process of fusing the suppressed electrocardiogram signal, the suppressed electroencephalogram signal, and the jointly aligned micro-deformation data to generate a fusion result includes: The suppressed electrocardiogram signal is mapped to the first multidimensional feature vector, the suppressed electroencephalogram signal is mapped to the second multidimensional feature vector, and the jointly aligned micro-deformation data is mapped to the third multidimensional feature vector. Based on the signal intensity change pattern in the first multidimensional feature vector, the phase continuity pattern in the second multidimensional feature vector, and the deformation propagation pattern in the third multidimensional feature vector, a cross-modal association rule is constructed. Based on the cross-modal association rule, the first multidimensional feature vector, the second multidimensional feature vector, and the third multidimensional feature vector are superimposed to generate a multidimensional feature set, which is the fusion result.

5. The system according to claim 1, characterized in that, The process of generating a biowave resonance migration analysis report, based on the duration and spatial distribution range of the migration features in the time domain, to determine whether the biowave resonance is abnormal, includes: The cumulative time length of the N consecutive occurrences of the offset feature in the time domain dimension, and the ratio of the area of ​​the organism surface covered to the whole organism surface in the spatial dimension are calculated, where N is greater than or equal to 3. Based on the first comparison result between the cumulative time length and the preset length threshold, and the second comparison result between the ratio and the preset ratio threshold, a determination result of whether the bio-wave resonance is abnormally offset is generated. Based on the judgment results, a bio-wave resonance migration analysis report is generated.

Images