Disease prediction method and system based on left and right radial artery pulse wave analysis
By employing dual-channel signal acquisition and multi-dimensional feature extraction, and utilizing left and right radial artery pulse wave analysis, the problem of insufficient prediction accuracy caused by unilateral pulse wave features was solved, achieving higher accuracy in disease risk assessment.
Patent Information
- Application Number
- CN202511527695.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-24
- Publication Date
- 2026-01-23
AI Technical Summary
Existing disease prediction methods rely on unilateral radial artery pulse wave characteristics, which cannot capture the spatiotemporal differences in bilateral pulse waves, resulting in low prediction accuracy, especially insufficient sensitivity in early disease prediction.
Employing dual-channel signal acquisition technology, the pulse waves of the left and right radial arteries are collected by placing sensors in different arms. Combined with adaptive signal preprocessing and multi-dimensional feature engineering, including time domain, frequency domain and nonlinear feature extraction, intelligent risk assessment is performed using ensemble learning.
It improves the accuracy of disease prediction, supports more comprehensive prediction models, reduces individual errors, significantly improves the quality of pulse wave signals, and provides a high-fidelity data foundation for subsequent bilateral feature analysis.
Smart Images
Figure CN121393879A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of medical devices and data analysis, and in particular to a disease prediction method and system based on left and right radial artery pulse wave analysis. BACKGROUND
[0002] Due to the rise of data analysis, in the medical field, disease prediction according to physiological signals has been widely used in practical scenarios. However, existing pulse wave collection mostly uses unilateral radial artery measurement, which cannot capture the spatiotemporal difference information of bilateral pulse waves. Existing models rely on unilateral features, and the generalization ability is limited. Most AI prediction models (such as SVM, neural network) use unilateral pulse wave time domain (peak time, amplitude), frequency domain (harmonic component) or nonlinear feature (sample entropy) as input. Unilateral data cannot reflect key pathological indicators such as vascular symmetry and autonomic nervous regulation balance, and cannot reflect the overall state, resulting in insufficient sensitivity of the model in early disease prediction. SUMMARY
[0003] Therefore, in order to solve the technical problem that the existing disease prediction method mostly relies on unilateral pulse wave features, thereby resulting in low prediction accuracy, in a first aspect, the present application provides a disease prediction method based on left and right radial artery pulse wave analysis, which comprises the following steps: Dual-channel signal acquisition, sensors are arranged at corresponding positions of different arms, and left and right radial artery pulse waves are collected according to a preset rule to obtain signal data; Adaptive signal preprocessing, denoising the signal data, including synchronous wavelet decomposition, common mode noise component calculation and subtraction denoising; Multi-dimensional feature engineering, feature extraction is performed according to the denoised signal data, including time domain features, frequency domain features and nonlinear features; specifically including peak time difference, amplitude ratio, main frequency energy ratio, amplitude square coherence, sample entropy difference and phase synchronization index.
[0004] Intelligent risk assessment, using an integrated learning and model training framework, classifying according to the extracted features, and outputting a risk rating.
[0005] In a second aspect, the present application also provides a disease prediction system, which comprises the following modules: A signal acquisition unit for performing dual-channel signal acquisition; A signal preprocessing unit for performing adaptive signal preprocessing; A feature extraction unit for multi-dimensional feature engineering; A classification unit for intelligent risk assessment.
[0006] Based on the above scheme, the application provides a disease prediction method and system based on left and right radial artery pulse wave analysis, which uses bilateral features for disease risk prediction, helps to improve the accuracy of prediction, supports more comprehensive prediction model and reduces individual error; further, through double-channel joint denoising, the quality of the pulse wave signal is significantly improved, providing a high-fidelity data basis for subsequent bilateral feature analysis. BRIEF DESCRIPTION OF DRAWINGS
[0007] Figure 1 is a step flow chart of a disease prediction method based on left and right radial artery pulse wave analysis of the application; Figure 2 is a structural block diagram of a disease prediction system based on left and right radial artery pulse wave analysis of the application. DETAILED DESCRIPTION
[0008] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.
[0009] It should be noted that, for the convenience of description, only the parts related to the application are shown in the drawings. The embodiments in the application and the features in the embodiments can be combined with each other without conflict.
[0010] It should be understood that the "system", "device", "unit" and / or "module" used in the application is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.
[0011] As shown in the application and claims, unless the context clearly indicates otherwise, "one", "a", "an" and / or "the" do not refer to a single number, but also include plural. Generally speaking, the terms "include" and "contain" only indicate the inclusion of the steps and elements clearly identified, and these steps and elements do not constitute an exclusive list, and the method or device can also include other steps or elements. The element defined by the statement "including a" does not exclude the presence of another identical element in the process, method, product or device including the element.
[0012] In addition, flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or subsequent operations are not necessarily performed in sequence. Instead, the steps can be processed in reverse order or simultaneously. Meanwhile, other operations can be added to these processes, or one or more steps can be removed from these processes.
[0013] Referring to Figure 1 The flowchart of an optional example of the disease prediction method based on left and right radial artery pulse wave analysis proposed in the present application can be applied to a computer device. The disease prediction method proposed in the present application can include but is not limited to the following steps: Step S1, arranging a sensor and collecting left and right radial artery pulse waves to obtain signal data; Step S2, performing denoising processing on the signal data to obtain denoised signal data; Step S3, performing feature extraction according to the denoised signal data to obtain bilateral features; Step S4, constructing a classification model and classifying according to the bilateral features to output a risk rating.
[0014] In some feasible embodiments, step S1 specifically includes: Arranging a pulse wave signal acquisition module at a corresponding position of different arms respectively; For the measurement process: first, stop at a specified pressure (such as 80 mmHg), and then measure the pulse wave; then, stop at a specified pressure of 100, and then measure the pulse wave; finally, stop at 120 mmHg and then measure.
[0015] In some feasible embodiments, step S2 specifically includes: For the denoising algorithm, the traditional wavelet threshold method uses a fixed threshold, which can cause signal distortion.
[0016] S2.1, synchronously decomposing left and right signals and to the same scale to obtain and respectively; S2.2, calculating the common-mode noise component: wherein, Dil(t) represents the i-th layer of detail coefficient of the left channel at time t, Dir(t) represents the i-th layer of detail coefficient of the right channel at time t; S2.3, subtracting the common-mode noise from the original detail coefficient: wherein, a represents a decay factor, and the default value is 0.7.
[0017] S2.4, using mathematical morphology filtering (opening operation + closing operation) to smooth the baseline drift of the denoised signal, and the structural element length is set to 10% of the sampling rate (such as 100 Hz sampling, element length = 10 points).
[0018] In some feasible embodiments, further comprising: Before threshold processing, cross-correlation analysis is performed on the detail coefficients of the left and right signals to mark the high correlation frequency band (reflecting the true pulse wave component) to avoid phase deviation caused by excessive denoising.
[0019] In some feasible embodiments, step S3 specifically comprises: S3.1, extracting time domain features.
[0020] S3.1.1, calculation of peak time difference: 1) using adaptive threshold peak detection (based on moving average + standard deviation), identifying main wave (P1), tidal wave (P2) and re-beat wave (P3); 2) calculating the time difference of each peak: ΔT1 (P1 left - P1 right), ΔT2 (P2 left - P2 right), taking the weighted average value (weight determined by peak amplitude); 3) median filtering (window = 5 cycles) is performed on the ΔT sequence to eliminate outliers.
[0021] S3.1.2, calculation of amplitude ratio: 1) baseline correction: using morphological filtering (top-hat transformation) to eliminate baseline shift introduced by uneven sensor contact pressure; 2) energy normalization: calculating amplitude ratio , using the signal energy in the period instead of single-point peak value, enhancing stability; 3) dynamic range compression: taking logarithmic processing (log10R) of R value, reducing the numerical sensitivity of high amplitude difference.
[0022] S3.2, frequency domain feature extraction.
[0023] S3.2.1, calculation of main frequency energy ratio (FER): 1) using synchronous short-time Fourier transform (STFT) (window length = 200 ms, overlap = 75%), generating time-frequency spectrum of left and right signals; 2) extracting main frequency band (1~10Hz) energy ratio: , wherein, ; 3) harmonic compensation: for pulse wave harmonic components (2f0, 3f0), calculating the standard deviation of 1-5th harmonic energy ratio as a frequency domain asymmetry supplement index.
[0024] S3.2.2, two-channel coherence analysis: calculating the amplitude square coherence (MSC) of left and right signals in the main frequency band: , wherein, represents the self-power spectral density of the left channel signal, represents the self-power spectral density of the right channel signal, Cross-power spectral density of left and right channel signals.
[0025] Through MSC analysis, the frequency domain synergy of bilateral pulse waves can be objectively quantified, providing a new dimension for vascular function assessment.
[0026] S3.3, extraction of nonlinear features.
[0027] S3.3.1, sample entropy difference : 1) Coarse-grained processing of signals (scale factor τ = 1~5), calculate left and right (embedding dimension m = 2, tolerance r = 0.2×SD); 2) define as the entropy difference under scale 3: ; 3) According to the local complexity of the signal, adjust value adaptively (such as high-frequency noise area decrease).
[0028] S3.3.2, phase synchronization index: based on Hilbert transform to extract instantaneous phase difference , calculate the phase synchronization index: ; Suggesting a decrease in aortic elasticity.
[0029] This scheme realizes end-to-end prediction from pulse wave to disease risk through bilateral feature difference modeling and interpretable scoring system, combining algorithm performance and clinical practicability.
[0030] In some feasible embodiments, step S4 specifically includes: Dataset construction: Sample source: Healthy control group (N = 200, age / gender matched) Disease group (such as atherosclerosis N = 100, diabetic neuropathy N = 80, aortic dissection N = 30) Data labeling: Label: healthy (0), disease (1) or sub-level (such as atherosclerosis degree 1-3) Based on the feature matrix construction of the dataset.
[0031] S4.1, construct a random forest (Random Forest) model: use the dataset for training to generate a classification model for multiple disease classification or risk grading.
[0032] S4.2, input the bilateral features to be tested into the trained classification model, and output the risk rating.
[0033] For example Figure 2As shown, a disease prediction system based on left and right radial artery pulse wave analysis comprises the following modules: a signal acquisition unit for performing step S1; a signal preprocessing unit for performing step S2; a feature extraction unit for performing step S3; a classification unit for performing step S4.
[0034] The contents in the above method embodiments are all applicable to the present system embodiment, the present system embodiment specifically realizes the same functions as the above method embodiments, and achieves the same beneficial effects as the above method embodiments.
[0035] A disease prediction device based on left and right radial artery pulse wave analysis: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned disease prediction method based on left and right radial artery pulse wave analysis.
[0036] The contents in the above method embodiments are all applicable to the present device embodiment, the present device embodiment specifically realizes the same functions as the above method embodiments, and achieves the same beneficial effects as the above method embodiments.
[0037] A storage medium having processor-executable instructions stored therein, the processor-executable instructions, when executed by a processor, are used to implement the above-mentioned disease prediction method based on left and right radial artery pulse wave analysis.
[0038] The contents in the above method embodiments are all applicable to the present storage medium embodiment, the present storage medium embodiment specifically realizes the same functions as the above method embodiments, and achieves the same beneficial effects as the above method embodiments.
[0039] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above-mentioned embodiments, and those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the present application, and these equivalent modifications or replacements are all included in the scope defined by the claims of the present application.
Claims
1. A disease prediction method based on left and right radial artery pulse wave analysis, characterized by, The method comprises the following steps: arranging sensors and collecting left and right radial artery pulse waves to obtain signal data; performing denoising processing on the signal data to obtain denoised signal data; performing feature extraction according to the denoised signal data to obtain bilateral features; constructing a classification model and classifying according to the bilateral features to output a risk rating.
2. The disease prediction method based on left and right radial pulse wave analysis according to claim 1, characterized in that, The step of arranging sensors and collecting left and right radial artery pulse waves to obtain signal data specifically comprises: arranging pulse wave signal acquisition modules at corresponding positions of different arms, respectively; setting a pressure value and performing pulse wave measurement based on the pulse wave signal acquisition modules to obtain signal data.
3. The disease prediction method based on left and right radial pulse wave analysis according to claim 2, characterized in that, The step of performing denoising processing on the signal data to obtain denoised signal data specifically comprises: performing synchronous wavelet decomposition of signal data of different channels to the same scale to obtain decomposed signal data; calculating a common mode noise component according to the decomposed signal data; subtracting the decomposed signal data from the common mode noise component to obtain denoised signal data.
4. The disease prediction method based on left and right radial pulse wave analysis according to claim 2, characterized in that, Further comprising: performing cross-correlation analysis on the decomposed signal data to mark a high correlation frequency band.
5. The disease prediction method based on left and right radial pulse wave analysis according to claim 4, characterized in that, The calculation formula of the common mode noise component is as follows: wherein denotes a common mode noise component at the time instant denotes a left channel a detail coefficient of the i-th layer at the time instant denotes a right channel a detail coefficient of the i-th layer at the time instant 6. The disease prediction method based on left and right radial pulse wave analysis according to claim 5, characterized in that, Further comprising: performing smoothing on the denoised signal data based on mathematical morphological filtering.
7. The disease prediction method based on left and right radial pulse wave analysis according to claim 6, characterized in that, The step of performing feature extraction according to the denoised signal data to obtain bilateral features specifically comprises: extracting time domain features from the denoised signal data, the time domain features including a peak time difference and an amplitude ratio; extracting frequency domain features from the denoised signal data, the frequency domain features including a main frequency energy ratio and an amplitude square coherence; extracting nonlinear features from the denoised signal data, the nonlinear features including a sample entropy difference and a phase synchronization index; organizing the time domain features, the frequency domain features and the nonlinear features to obtain final features corresponding to both sides.
8. The disease prediction method based on left and right radial artery pulse wave analysis according to claim 7, wherein the calculation formula of the amplitude square coherence is represented as follows: wherein, a self-power spectral density representing the left channel signal, a self-power spectral density representing the right channel signal, a cross-power spectral density representing the left and right channel signals.
9. The disease prediction method based on left and right radial pulse wave analysis according to claim 1, characterized in that, The step of constructing a classification model and classifying according to the bilateral features to output a risk rating specifically comprises: constructing a random forest model and training using a pre-labeled data set to obtain a trained classification model; inputting the bilateral features into the trained classification model to output a risk rating.
10. A disease prediction system based on left and right radial artery pulse wave analysis, characterized by, The method comprises: a signal acquisition unit for arranging sensors and collecting left and right radial artery pulse waves to obtain signal data; a signal preprocessing unit for performing denoising processing on the signal data to obtain denoised signal data; a feature extraction unit for performing feature extraction according to the denoised signal data to obtain bilateral features; a classification unit for constructing a classification model and classifying according to the bilateral features to output a risk rating.