Remote traditional Chinese medicine pulse condition recognition system based on pressure-isolation-electricity coupling flexible sensor
By using a multimodal pulse recognition system based on a piezo-ion-electro-coupled flexible sensor array and a deep learning model, the problems of subjectivity and sensor sensitivity in traditional Chinese medicine pulse diagnosis have been solved. This system achieves highly sensitive pulse acquisition and remote intelligent interpretation, thereby improving the objectivity and interpretability of pulse diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Tianfu Jincheng Laboratory (Frontier Medical Center)
- Filing Date
- 2026-03-11
- Publication Date
- 2026-04-24
AI Technical Summary
Existing TCM pulse diagnosis techniques are highly subjective, rely on physician experience, are difficult to standardize and remotely utilize, and traditional sensors have low sensitivity, poor signal stability, and limited ability to analyze TCM pulse characteristics.
A multimodal pulse recognition system is constructed by using a piezo-ion-electro-coupled flexible sensor array, combined with adaptive filtering and deep learning models, to achieve highly sensitive pulse acquisition, standardized feature extraction, and remote monitoring.
It achieves high sensitivity and stability of flexible sensors under low-frequency weak signals, can accurately extract the characteristics of traditional Chinese medicine pulse diagnosis, supports remote real-time data upload and intelligent interpretation, and improves the objectivity and interpretability of pulse diagnosis.
Smart Images
Figure CN121910337A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of pulse diagnosis technology, and in particular to a remote TCM pulse diagnosis system based on a piezoelectric coupling flexible sensor. Background Technology
[0002] Pulse diagnosis in Traditional Chinese Medicine (TCM), as one of the core methods of syndrome differentiation and treatment, has a long history and rich theoretical connotations. However, its strong subjectivity and reliance on the physician's personal experience limit its standardized promotion and application in the modern medical system. To achieve objectification, digitization, and remote pulse diagnosis, scholars both domestically and internationally have conducted research for many years, but still face numerous technical bottlenecks and practical challenges.
[0003] Research on the objectification of pulse diagnosis began in the 1950s. Early methods mainly used rigid sensors such as piezoelectric ceramics, PVDF, and MEMS to construct pulse imagers, which suffered from large size, poor skin adhesion, and low sensitivity (e.g., d33 < 30 pC N). -1 Traditional pulse diagnosis methods often suffer from several limitations, including insufficient sensitivity, poor signal stability, and reliance on external power. Furthermore, the development of flexible electronics technology has led to the introduction of flexible sensor materials such as graphene, carbon nanotubes, and ion gels into pulse acquisition, improving comfort to some extent. However, key issues remain, including insufficient sensitivity, poor signal stability, and dependence on external power. In terms of signal processing, traditional methods rely heavily on shallow feature extraction in the time and frequency domains, limiting their ability to analyze the multi-dimensional characteristics of traditional Chinese medicine pulse diagnosis, such as "floating, sinking, slow, rapid, weak, strong, slippery, and hesitant." Currently, the industry lacks a widely accepted, objective pulse diagnosis system that can fully reproduce the sensations felt by traditional Chinese medicine practitioners.
[0004] Therefore, developing a technology and system capable of remotely and with high fidelity to collect pulse information and possessing intelligent interpretation capabilities has become a key technical problem that urgently needs to be solved in current TCM healthcare work. Summary of the Invention
[0005] The purpose of this invention is to use piezo-ion-electro-coupling technology as the core to prepare a sensor array, collect multimodal pulse signals, construct an intelligent pulse recognition model, and output pulse data, thereby providing a remote TCM pulse recognition system based on a piezo-ion-electro-coupling flexible sensor.
[0006] To achieve the above-mentioned objectives, the embodiments of the present invention provide the following technical solutions:
[0007] A remote TCM pulse diagnosis system based on a piezoelectric coupling flexible sensor includes:
[0008] The pulse sensor module is designed to be attached to the wrist of the human body in a three-finger synchronized pulse-taking manner.
[0009] The multimodal signal acquisition and preprocessing module is used to acquire multimodal pulse signals through the pulse sensor module, and extract feature parameters after preprocessing.
[0010] The pulse recognition module is used to build an intelligent pulse recognition model. After inputting multimodal pulse signals and feature parameters into the pulse recognition module, the user's pulse data is obtained.
[0011] The remote monitoring module is used to enable real-time uploading, analysis, and feedback of pulse data.
[0012] Furthermore, the pulse sensor module includes a sensor array based on a piezoelectric coupling film, used to reflect the three pulse-taking pressures: superficial, middle, and deep.
[0013] Furthermore, the multimodal signal acquisition and preprocessing module includes:
[0014] A multimodal signal acquisition unit is used to acquire multimodal pulse signals through a pulse sensor module. The multimodal pulse signals include voltage amplitude, current phase, and waveform timing information output by the sensor array.
[0015] The preprocessing unit is used to preprocess the multimodal pulse signal using adaptive filtering and wavelet threshold denoising algorithms to eliminate interference from baseline drift, power frequency interference and motion noise in the original signal.
[0016] The feature extraction unit is used to extract feature parameters of the preprocessed multimodal pulse signal. The extracted feature parameters include pulse position, pulse frequency, pulse shape, and pulse momentum.
[0017] Furthermore, the pulse image intelligent recognition model includes a first branch, a second branch, a feature fusion unit, a hybrid unit, and an attention mechanism unit;
[0018] The first branch is a CNN, which is used to process multimodal pulse signals to obtain a two-dimensional spatiotemporal image;
[0019] The second branch is a fully connected network used to process the feature vectors of the feature parameters;
[0020] The feature fusion unit fuses the two-dimensional spatiotemporal image and feature vectors to obtain fused features;
[0021] The hybrid unit includes CNN and LSTM. CNN is used to extract local waveform features of the fused features, and LSTM is used to capture the long-term temporal sequence of the pulse sequence.
[0022] The attention mechanism is used to integrate local waveform features and long-term temporal data of the pulse sequence, ultimately outputting pulse image data.
[0023] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0024] (1) Sensitivity and stability of flexible sensor array under weak and low frequency pulse signals: Traditional piezoelectric devices are not responsive enough to low frequency and weak signals such as human pulse. This solves the problem of signal attenuation and baseline drift of PNP devices under dynamic pressure.
[0025] (2) Quantification and standardization of pulse characteristics: Traditional Chinese medicine pulse has multi-dimensional characteristics such as "position, number, shape and momentum". The ability to accurately and repeatedly extract these characteristics from the original signal output by the sensor array is the core difficulty of objectiveing pulse.
[0026] (3) Generalization and interpretability of pulse intelligent recognition model: The pulse characteristics of different human bodies and different physiological states are significantly different. A recognition model with strong generalization ability was constructed and its interpretability was enhanced to conform to the diagnostic logic of traditional Chinese medicine.
[0027] (4) Balancing data security, real-time performance and user experience in remote systems: Under the premise of ensuring data privacy and stable transmission, achieve low-latency uploading and user-friendly interaction of pulse data to improve the system's usability and acceptance. Attached Figure Description
[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0029] Figure 1 This is a block diagram of the system modules of the present invention;
[0030] Figure 2 This is a schematic diagram of the pulse intelligent recognition model structure of the present invention. Detailed Implementation
[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0032] It should be noted that similar reference numerals and letters in the following figures denote similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, the terms "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance, or suggesting any such actual relationship or order between these entities or operations. Additionally, the terms "connected," "linked," etc., can refer to a direct connection between elements, components, modules, etc., or an indirect connection via other elements, components, modules, etc.
[0033] Piezoelectric coupling, as an emerging force-to-electric conversion mechanism, has shown great potential in the field of flexible sensing due to its advantages such as high output, self-polarization, and full flexibility. The research results published in *Nature Communications* by Yang Weiqing's team demonstrate that they have successfully pioneered a new concept of piezoelectric coupling devices and developed a PVDF / Nafion / PVDF all-polymer thin film with an ordered ion-electron interface, achieving a piezoelectric coefficient as high as 80.70 pC / N. -1 This further demonstrates the versatile application of this novel piezo-ion-electro-coupled device in energy harvesting, physiological signal monitoring, and vibration state detection.
[0034] This solution is based on a piezo-ion-electro-coupled all-polymer flexible sensing film (PNP-Film) with zero power consumption and ultra-high sensitivity (80.70 pC N). -1 Leveraging its technological advantages, the system develops flexible pulse patch and constructs a pulse recognition model to achieve a standardized closed-loop system for pulse diagnosis, encompassing "collection-transmission-recognition." This system can be deployed in power-deficient environments such as homes, communities, and outdoors, filling a gap in the medical field of objective pulse diagnosis in terms of integrated "high sensitivity + self-powered + remote" capabilities.
[0035] This invention is achieved through the following technical solutions, such as... Figure 1 As shown, a remote TCM pulse identification system based on a piezo-ion-electro-coupled flexible sensor is proposed, including: a pulse sensor module, a multimodal signal acquisition and preprocessing module, a pulse identification module, and a remote monitoring module.
[0036] The pulse sensor module is designed and optimized using a highly sensitive flexible pulse sensor array. Based on a piezo-ion-electro-coupled thin film, a flexible sensor array suitable for the radial artery region of the human body is designed and fabricated. The focus is on studying the sensor's response characteristics under low-frequency, weak pulse signals, optimizing its structural layout and packaging to achieve adaptive sensing of three pulse-taking pressures: superficial, middle, and deep. This is achieved through the following three aspects:
[0037] (1) Material and structure optimization: Based on PVDF / Nafion / PVDF all-polymer film, the flexibility, stability and biocompatibility of the film were further optimized by adjusting the thickness ratio of each layer, heat treatment process and hot pressing parameters. The piezoelectric-piezoelectric ion coupling response characteristics of the film under micro-pressure (corresponding to the pulse "floating", medium pressure (corresponding to the pulse "medium"), and high pressure (corresponding to the pulse "deep") were the focus.
[0038] (2) Sensor array electrode design: High-precision, multi-channel interdigital electrode arrays are fabricated on PVDF / Nafion / PVDF all-polymer films using microfabrication techniques such as photolithography or laser etching. The array design will simulate the pulse-taking method of traditional Chinese medicine practitioners using three fingers simultaneously (cun, guan, chi), with each part containing 3×3 sensing units to achieve high-resolution capture of the spatial distribution of pulse signals.
[0039] (3) Encapsulation and bonding technology: Develop flexible encapsulation layer materials suitable for human wrists (such as medical-grade Ecoflex silicone), design a wristband-type fixation device with a biomimetic structure to ensure that the sensor fits tightly and comfortably against the skin surface, and can effectively isolate the interference of environmental temperature and humidity and motion artifacts.
[0040] The multimodal signal acquisition and preprocessing module is used to acquire multimodal pulse signals, simultaneously obtaining multidimensional information such as pressure, frequency, and waveform to form multimodal pulse signals. Combining traditional Chinese medicine pulse theory, key feature parameters including pulse position, pulse frequency, pulse shape, and pulse strength are extracted, and a standardized pulse signal preprocessing workflow is established to improve the signal-to-noise ratio and feature recognizability. This is achieved through the following three aspects:
[0041] (1) Synchronous acquisition of multi-modal signals: Build a hardware platform for synchronous acquisition of multi-channel signals based on a high-precision data acquisition card (DAQ) and an embedded system (such as an STM32 series MCU). Synchronously acquire the voltage (reflecting voltage amplitude), current phase (reflecting dynamic response), and waveform timing information output by each sensing unit on the sensor array.
[0042] (2) Signal preprocessing and enhancement: For baseline drift, power frequency interference and motion noise in the original signal, adaptive filtering (LMS) and wavelet threshold denoising (Wavelet Transform) algorithms are used for preprocessing to preserve key physiological information in the signal.
[0043] (3) TCM-oriented feature extraction: From the preprocessed multi-channel signal, the set of quantitative feature parameters corresponding to the "position, number, shape and momentum" of TCM pulse diagnosis theory is extracted as shown in Table 1.
[0044] Table 1. Extracted feature parameter set
[0045]
[0046] The pulse recognition module integrates traditional Chinese medicine pulse classification knowledge with modern machine learning methods (such as CNN, LSTM, and graph neural networks) to construct an intelligent pulse recognition model based on multimodal feature parameters. The accuracy and generalization ability of the intelligent pulse recognition model are verified by comparing it with the diagnostic results of clinical TCM physicians. This is achieved through the following three aspects:
[0047] (1) Multimodal data fusion: A two-branch deep learning network is constructed. One branch (CNN) processes the two-dimensional spatiotemporal image formed by multi-channel pulse waves; the other branch (fully connected network) processes the structured feature vectors of the feature parameters extracted above. Feature fusion of the two branches is performed at the back end of the model to make full use of the spatial, temporal and statistical information of the data.
[0048] (2) Model Architecture and Training: The pulse intelligent recognition model will employ an attention mechanism to enhance the hybrid model of convolutional neural network-long short-term memory network (CNN-LSTM). CNN is used to extract local waveform features, LSTM is used to capture the long-term temporal dependence of the pulse sequence, and the attention mechanism allows the model to focus on the signal segments and channels most relevant to diagnosis, such as... Figure 2 As shown.
[0049] (3) Model interpretability study: Gradient weighted class activation mapping (Grad0-CAM) and other techniques are used to visualize the input signal region of the model when making classification decisions, enhance the credibility of the model, and attempt to associate and translate with traditional Chinese medicine theory.
[0050] The remote monitoring module develops an integrated hardware and software system supporting wireless transmission, cloud storage, and remote diagnosis. This system enables real-time uploading, analysis, and feedback of pulse data, constructing a digital platform for traditional Chinese medicine pulse diagnosis, and demonstrating its application in community hospitals or family health management scenarios. This is achieved through the following three aspects:
[0051] (1) Hardware and software system integration: Develop low-power, miniaturized embedded signal processing and wireless transmission modules (such as those based on the ESP32 series) to achieve local preprocessing of pulse data and real-time uploading via WiFi / Bluetooth 5.0. Develop cloud servers and terminal applications to achieve data storage, management, intelligent analysis, and visualization report generation.
[0052] (2) Clinical validation protocol: Recruit healthy volunteers and subjects with typical pulse patterns (such as flat, wiry, slippery, thin, rapid, slow pulse, etc.) to evaluate the accuracy, sensitivity and specificity of pulse pattern recognition of this system.
[0053] (3) System performance evaluation: Evaluate the system's signal quality, transmission stability, response latency, and battery life from a technical perspective. Evaluate its ease of use, comfort, and report understandability from a user experience perspective.
[0054] Based on the above system, this solution also proposes a remote TCM pulse diagnosis method based on a piezo-iono-electro-coupled flexible sensor, including the following steps:
[0055] Step 1: Construct and encapsulate a pulse sensor module using a sensor array based on a piezo-ion-electro-coupling thin film; attach the pulse sensor module to the human wrist using a three-finger synchronous pulse retrieval method.
[0056] Step 2: The multimodal signal acquisition and preprocessing module acquires multimodal pulse signals through the pulse sensor module. The multimodal pulse signals include voltage amplitude, current phase, and waveform timing information output by the sensor array. Adaptive filtering and wavelet threshold denoising algorithms are used to preprocess the multimodal pulse signals to eliminate interference from baseline drift, power frequency interference, and motion noise in the original signal, thereby preserving key physiological information in the signal. Then, feature parameters are extracted, including pulse position, pulse frequency, pulse shape, and pulse potential.
[0057] Step 3: Input the collected multimodal pulse signals and extracted feature parameters into the pulse recognition module to train the pulse intelligent recognition model.
[0058] In the application phase, the pulse sensor module is attached to the patient's wrist. The multimodal signal acquisition and preprocessing module acquires multimodal pulse signals and extracts feature parameters after preprocessing. The multimodal pulse signals and feature parameters are then input into the pulse recognition module, and the pulse intelligent recognition model outputs the patient's pulse.
[0059] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A remote TCM pulse diagnosis system based on a piezo-iono-electro-coupled flexible sensor, characterized in that, include: The pulse sensor module is designed to be attached to the wrist of the human body in a three-finger synchronized pulse-taking manner. The multimodal signal acquisition and preprocessing module is used to acquire multimodal pulse signals through the pulse sensor module, and extract feature parameters after preprocessing. The pulse recognition module is used to build an intelligent pulse recognition model. After inputting multimodal pulse signals and feature parameters into the pulse recognition module, the user's pulse data is obtained. The remote monitoring module is used to enable real-time uploading, analysis, and feedback of pulse data.
2. The remote TCM pulse diagnosis system based on a piezo-iono-electro-coupled flexible sensor according to claim 1, characterized in that, The pulse sensor module includes a sensor array based on a piezoelectric coupling film, used to reflect the three pulse-taking pressures: superficial, middle, and deep.
3. The remote TCM pulse diagnosis system based on a piezo-iono-electro-coupled flexible sensor according to claim 1, characterized in that, The multimodal signal acquisition and preprocessing module includes: A multimodal signal acquisition unit is used to acquire multimodal pulse signals through a pulse sensor module. The multimodal pulse signals include voltage amplitude, current phase, and waveform timing information output by the sensor array. The preprocessing unit is used to preprocess the multimodal pulse signal using adaptive filtering and wavelet threshold denoising algorithms to eliminate interference from baseline drift, power frequency interference and motion noise in the original signal. The feature extraction unit is used to extract feature parameters of the preprocessed multimodal pulse signal. The extracted feature parameters include pulse position, pulse frequency, pulse shape, and pulse momentum.
4. The remote TCM pulse diagnosis system based on a piezo-iono-electro-coupled flexible sensor according to claim 3, characterized in that, The pulse intelligent recognition model includes a first branch, a second branch, a feature fusion unit, a hybrid unit, and an attention mechanism unit; The first branch is a CNN, which is used to process multimodal pulse signals to obtain a two-dimensional spatiotemporal image; The second branch is a fully connected network used to process the feature vectors of the feature parameters; The feature fusion unit fuses the two-dimensional spatiotemporal image and feature vectors to obtain fused features; The hybrid unit includes CNN and LSTM. CNN is used to extract local waveform features of the fused features, and LSTM is used to capture the long-term temporal sequence of the pulse sequence. The attention mechanism is used to integrate local waveform features and long-term temporal data of the pulse sequence, and finally outputs pulse data.