A radial artery three-part pulse acquisition system based on artificial intelligence

Abstract

The present invention discloses an artificial intelligence-based radial artery three-part pulse acquisition system, relating to the field of pulse detection technology. The system comprises an MCU control unit, an air pump and airbag unit, a stepper motor unit, and a pulse acquisition unit, each connected to the MCU control unit. The MCU control unit controls the air pump and airbag unit to inflate and deflate, while also controlling the upward and downward movement of the stepper motor unit to further fine-tune the applied pressure. The pulse acquisition unit collects and records pulse signal data based on changes in pressure. This system solves the problem of existing pulse acquisition devices that cannot precisely or independently control pulse pressure.

Description

Technical Field

[0001] The present invention relates to the technical field of pulse detection, and more particularly to an artificial intelligence-based radial artery three-part pulse acquisition system. Background Art

[0002] Pulse diagnosis, also known as "feeling the pulse," "feeling the pulse," or "holding the pulse," involves a doctor's fingertips on the patient's radial artery to detect changes in the condition. The most commonly used pulse site today is the "Cunkou" (Cunkou), located superficially behind the wrist. The "Cunkou" pulse is divided into three parts: the Cun, the Guan, and the Chi. Regarding the classification of these three pulses for internal organs, the current common clinical classification is: the right Cun pulse indicates the lungs, the right Guan pulse indicates the spleen and stomach, and the right Chi pulse indicates the kidneys (Mingmen); the left Cun pulse indicates the heart, the left Guan pulse indicates the liver, and the left Chi pulse indicates the kidneys.

[0003] Pulse diagnosis often uses three different levels of force: light pressure on the skin is called "lifting"; heavy pressure reaching the tendons and bones is called "sinking"; and moderate pressure reaching the muscles is called "seeking." The Cun, Guan, and Chi parts each have three types of signs: floating, medium, and sinking, collectively known as the "Three Parts and Nine Signs."

[0004] Pulse diagnosis is a real-time, accurate, and personalized diagnostic method that helps doctors determine symptoms and treatment, and predict changes in patients' condition. However, pulse diagnosis is subject to numerous factors. Pulse diagnosis involves feeling the patient's pulse for its distinct characteristics of floating, medium, and sinking pulses. The judgment process is primarily subjective and empirical, lacking objective quantitative standards. Furthermore, the scarcity of experienced Traditional Chinese Medicine practitioners has hindered the development of pulse diagnosis in Traditional Chinese Medicine.

[0005] The pulse acquisition devices on the market have the following shortcomings: First, there is no corresponding automatic pressure increase or decrease mode, making it difficult to obtain pulse information under different pressures of "floating, middle, and sinking" advocated by traditional Chinese medicine; second, this diagnostic method still requires users to find the best pulse taking position and pressure by themselves, which may affect the accuracy of pulse information and the promotion of the device. Summary of the Invention

[0006] In view of this, the present invention provides a radial artery three-part pulse acquisition system based on artificial intelligence to solve the problem that the pulse pressure of the existing pulse acquisition device cannot be accurately controlled or independently controlled.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] An artificial intelligence-based radial artery three-part pulse acquisition system comprises: an MCU control unit, and an air pump air bag unit, a stepper motor unit, and a pulse acquisition unit respectively connected to the MCU control unit; the MCU control unit controls the air pump air bag unit to add / deflat, and simultaneously controls the stepper motor unit to move up and down, further fine-tuning the applied pressure value; the pulse acquisition unit acquires and records pulse signal data based on changes in the pressure value.

[0009] Optionally, the pulse signal data collected by the pulse acquisition unit is transmitted to the host computer via a Type-c unit or a wireless communication unit.

[0010] Optionally, it also includes a multi-parameter acquisition unit connected to the MCU control unit, and the multi-parameter acquisition unit includes a PPG photoplethysmography sensor, an ECG electrocardiogram sensor, a body temperature sensor and an EMG electromyography sensor.

[0011] Optionally, the air pump airbag unit includes an ultra-quiet pump, an airbag, an air pressure sensor and an electromagnetic valve; the ultra-quiet pump is connected to the airbag, the air pressure sensor and the electromagnetic valve respectively through an air pipe; there are multiple airbags, and limit sensors are placed between the airbags to prevent over-inflation.

[0012] Optionally, the stepper motor unit includes a stepper motor, and the stepper motor is provided with an upper and lower travel limit switch for applying pressure to the wrist.

[0013] Optionally, the pulse acquisition unit is divided into a Cun pulse acquisition part, a Guan pulse acquisition part and a Chi pulse acquisition part; the Cun pulse acquisition part is further divided into a Cun pulse channel center sensor module, a Cun upper pulse position sensor module and a Cun lower pulse position sensor module; the Guan pulse acquisition part is further divided into a Guan pulse channel center sensor module, a Guanshang pulse position sensor module and a Guanxia pulse position sensor module; the Chi pulse acquisition part is further divided into a Chi pulse channel center sensor module, a Chishang pulse position sensor module and a Chixia pulse position sensor module; the pulse acquisition unit has a total of 9 stress partitions, and the 9 stress partitions are respectively subjected to action pressure combination to obtain the specific information conduction mechanism and markers of the organ corresponding to each partition; and use neural network learning; and feed back the collected pulse signal data to the MCU single-chip computer.

[0014] Optionally, the PPG photoplethysmography sensor is attached to the wrist of the person being measured to obtain heart rate, blood oxygen saturation, blood pressure and respiratory rate based on changes in blood volume; the ECG electrocardiogram sensor is attached to the wrist of the person being measured to obtain electrocardiogram data; the body temperature sensor is attached to the wrist of the person being measured to obtain body temperature; and the EMG electromyography sensor is attached to the wrist of the person being measured to obtain electromyography data.

[0015] Optionally, the MCU control unit is also connected to a stress finger motion transmission mechanism, which is divided into three stress fingertips, which apply pressure toward the pulse acquisition unit respectively. The stress fingertips have built-in pressure feedback sensors, which automatically adjust the applied pressure according to the feedback pressure.

[0016] Optionally, the pulse acquisition unit is packaged on an FPC soft board, and three FPC soft boards are combined together to simulate three fingers contacting the radial artery at the wrist, so as to traverse the lateral position of the wrist.

[0017] It can be seen from the above technical solution that, compared with the prior art, the present invention provides a radial artery three-part pulse acquisition system based on artificial intelligence, which has the following beneficial effects:

[0018] 1. The present invention uses a new pressure increase and decompression system combining an airbag and a stepping motor to control the pulse pressure. The pressure is increased and decreased continuously at the three parts of the wrist, namely, Cun, Guan and Chi, separately or simultaneously, simulating the collection method of the three parts and nine lords. It can collect more comprehensive pulse information and find the optimal pulse pressure more quickly. Compared with traditional airbag overall pressurization or mechanical pressurization, it is more stable and accurate.

[0019] 2. The MCU control unit module adopts a low-power, highly integrated microcontroller, which reduces the power consumption of the entire system.

[0020] 3. The FPC soft board is used to contact the radial artery at the wrist, which can traverse the lateral position of the wrist, so there is no need to worry about the user's inaccurate positioning.

[0021] 4. The user only needs to place the wrist at the position designated by the arrow on the strap and tie the strap to complete the accurate collection of pulse information. No special knowledge of traditional Chinese medicine is required, the operation is easy, and it can be widely used in personal and family health care.

[0022] 5. The mobile phone platform or PC can realize the analysis and processing of pulse data, and store it in real time, establish a personal database to provide reference for expert diagnosis, and realize the sharing of precious medical resources. At the same time, a large amount of pulse data provides a valuable pulse training set, paving the way for the realization of smart medical care. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0024] Figure 1 It is a schematic diagram of the internal structure principle of the present invention;

[0025] Figure 2 It is an exploded view of the physical structure of the present invention;

[0026] Figure 3 It is an overall diagram of the entity structure of the present invention;

[0027] Among them, 1-housing, 2-MCU control unit, 3-air pressure sensor, 4-ultra-quiet pump, 5-solenoid valve, 6-large air bag, 7-small air bag, 8-limit sensor, 9-stepping motor, 10-upper pulse position sensor, 11-pulse center sensor, 12-lower pulse position sensor, 13-PPG photoelectric capacitance measurement sensor, 14-ECG electrocardiogram sensor, 15-body temperature sensor, 16-EMG electromyography sensor, 17-stress finger motion transmission mechanism, 18-pressure feedback sensor, 19-wireless communication unit, 20-Type-c unit, 21-trachea, 22-sheath. DETAILED DESCRIPTION

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0029] Example 1

[0030] The embodiment of the present invention discloses a radial artery three-part pulse acquisition system based on artificial intelligence, such as Figure 1-Figure 3 As shown, it includes: an MCU control unit 2, and an air pump airbag unit, a stepper motor unit and a pulse acquisition unit respectively connected to the MCU control unit 2; the MCU control unit 2 controls the air pump airbag unit to add / deflat, and at the same time controls the stepper motor 9 unit to move up and down, and further fine-tune the applied pressure value; the pulse acquisition unit collects pulse signal data based on the change of pressure value and records it.

[0031] Specifically, in this embodiment, the MCU control unit 2 controls the air pump and airbag unit to inflate, the solenoid valve 5 to deflate, and the motor unit to apply pressure. The pulse signal collected by the pulse acquisition unit and the heart rate, blood oxygen saturation, blood pressure, respiratory rate, electrocardiogram, body temperature, and electromyographic signals collected by the multi-parameter acquisition unit are transmitted to the host computer via the Type-C unit 20 and the wireless communication unit 19. The air pump and airbag unit comprises four ultra-quiet pumps 4, four airbags, four air pressure sensors 3, and four solenoid valves 5. The ultra-quiet pumps 4 generate gas for the airbags and air pressure sensors 3. The airbags inflate and apply pressure to the wrist's Cun, Guan, and Chi points. Built-in limit sensors 8 protect the subject's wrist from excessive pressure and pain. When the set pressure is exceeded, the solenoid valve 5 deflates. The stepper motor unit comprises four stepper motors 9, each with upper and lower travel limit switches. The stepper motors 9 apply pressure to the wrist's Cun, Guan, and Chi points. The pulse acquisition unit is divided into three acquisition sections and nine acquisition modules to acquire pulse signals from the wrist's Cun, Guan, and Chi points. The multi-parameter acquisition unit includes a PPG photoplethysmography sensor 13, an ECG sensor 14, a temperature sensor 15, and an EMG sensor 16. It acquires heart rate, blood oxygen saturation, blood pressure, respiratory rate, ECG, body temperature, and EMG signals from the wrist. The force finger motion transmission mechanism 17 (servo motor and muscle hydraulic transmission mechanism) consists of three force fingertips and houses three built-in pressure feedback sensors 18. It automatically adjusts the pressure applied to the Cun, Guan, and Chi points based on the feedback pressure.

[0032] The MCU control unit 2 uses the pressure value fed back by the air pressure sensor 3 to control the ultra-quiet pump 4 to add air to the airbag and the solenoid valve 5 to deflate, ensuring that the airbag pressure value remains constant. Simultaneously, the MCU control unit 2 controls the up and down movement of the stepper motor unit to further fine-tune the applied pressure value. This allows for more accurate and standardized recording of the pulse signal data collected by the pulse acquisition unit when different pressures are applied to the Cun, Guan, and Chi points. The pulse acquisition unit 2 uses the pressure value fed back by the air pressure sensor 3 to control the ultra-quiet pump 4 to add air to the airbag and the solenoid valve 5 to deflate, ensuring that the airbag pressure value remains constant. Simultaneously, the MCU control unit 2 controls the up and down movement of the stepper motor unit to further fine-tune the applied pressure value. This allows for more accurate and standardized recording of the pulse signal data collected by the pulse acquisition unit when different pressures are applied to the Cun, Guan, and Chi points. The pulse signal data collected by the pulse acquisition unit is transmitted to the host computer via the Type-C unit 20 and the wireless communication unit 19. The host computer analyzes and compares the pulse signal data to obtain a pulse condition conclusion.

[0033] The air pump and airbag unit includes four ultra-quiet pumps 4, four airbags, four air pressure sensors 3, and four solenoid valves 5. One ultra-quiet pump 4 is connected to one airbag, one air pressure sensor 3, and one solenoid valve 5 through an air tube 21. The difference is that among the four airbags, three small airbags 7 are 1.4 cm wide and 4 cm long, corresponding to the three collection positions of Cun, Guan, and Chi on the wrist. One large airbag 6 is 6 cm wide and 4 cm long. It covers the entire wrist collection position. This allows pressure to be applied to the Cun, Guan, and Chi positions of the wrist respectively, and to the Cun, Guan, and Chi positions simultaneously in the floating, middle, and sinking pressures. Limit sensors 8 are placed between the airbags to prevent over-inflation and discomfort in the wrist of the person being measured. The ultra-quiet pump 4 inflates the airbag. After inflation, the airbag applies pressure to the pulse collection unit. The air pressure sensor 3 collects the pressure applied to the pulse collection unit after inflation and the pressure feedback from the Cun, Guan, and Chi positions. The solenoid valve 5 is used to deflate the airbag and to quickly deflate it in the event of a fault. The device can set the applied force in the range of 0 to 250 mmHg, with a maximum allowable error of ±15%. The device can display the applied force in the range of 0 to 250 mmHg, with a maximum allowable error of ±15%.

[0034] The stepper motor unit contains four stepper motors 9, each with an upper and lower travel limit switch. The slides on these four stepper motors 9 are fixed behind the four airbags. When the airbag pressure is constant during pulse acquisition, micro-movements are performed. If the measured pressure is less than the set pressure, the slide moves downward to equalize the measured pressure; if the measured pressure is greater than the set pressure, the slide moves upward to equalize the measured pressure. This ensures consistent pressure during repeated pulse acquisition, effectively preventing unstable pulse acquisition caused by pressure fluctuations, improving pulse acquisition stability, and providing quantitative data for accurate pulse sensor results.

[0035] The pulse acquisition unit is comprised of 144 pressure sensors (which can be a combination of liquid sensors, flexible sensors, and piezoelectric bridge sensors). These sensors utilize an upper pulse position sensor 10, a pulse channel center sensor 11, and a lower pulse position sensor 12 to collect pulse signal data from the Cun, Guan, and Chi points at the wrist. The Cun pulse acquisition section is further divided into the Cun channel center sensor module, the Cun upper pulse position sensor module, and the Cun lower pulse position sensor module. The Guan pulse acquisition section is further divided into the Guan channel center sensor module, the Guan upper pulse position sensor module, and the Guan lower pulse position sensor module. The Chi pulse acquisition section is further divided into the Chi channel center sensor module, the Chi upper pulse position sensor module, and the Chi lower pulse position sensor module. There are nine stress zones in total, each containing 4 x 4 = 16 pressure sensors. Each of the nine stress zones is individually combined with action pressure to acquire specific information transmission mechanisms and markers corresponding to the five internal organs (10 organs) in each zone. This is accomplished using neural network learning.

[0036] The organ-specific neural network architecture based on the dynamic pressure combination of the nine stress zones is designed as follows, combining biomechanical conduction characteristics with deep learning algorithms:

[0037] Multi-pressure spatiotemporal feature extraction: Input layer: Assume that the three-dimensional biological signal of the i-th stress partition (i=1~9) under k pressure gradients (floating / medium / sinking) is:

[0038] ;

[0039] is the pressure component, is the displacement component, is the shear stress component and R is the spatial component.

[0040] It includes pressure, displacement, shear stress components, and sampling time T.

[0041] Time-frequency-space joint coding: using the improved Wigner-Ville distribution for time-frequency analysis:

[0042] ;

[0043] Extracting spatiotemporal features through 3D convolution:

[0044] ;

[0045] in It is a learnable three-dimensional convolution kernel (size 5×5×3); ReLU is a linear algebra operation that takes the real part;

[0046] Organ-specific attention mechanism: Dynamic weight allocation: Define the pressure sensitivity coefficient of organ j (such as heart, liver, kidney) to partition i:

[0047] ;

[0048] in is the query vector of the organ; LSTM is the long short-term memory network

[0049] Pressure-Organ Correlation Matrix: Constructing Organ Feature Maps:

[0050]

[0051] is the organ-specific projection matrix;

[0052] Bidirectional Gated Graph Conduction Network: Modeling Biological Conduction Graphs: Constructing Adjacency Matrix Based on the Internal and External Relationships of Zang-Fu Organs in Traditional Chinese Medicine , define graph convolution:

[0053]

[0054] in

[0055] Bidirectional gated update:

[0056]

[0057]

[0058] Multi-task dynamic decision making: Organ state prediction:

[0059]

[0060] in represents vector concatenation, For cross-partition aggregation;

[0061] Loss function design:

[0062]

[0063] BCE is binary cross entropy; is the weight coefficient; It is the outer product operation.

[0064] Contains multi-task cross entropy and TCM relationship constraints.

[0065] Pressure-Organ Response Surface Optimization: Define the Organ Diagnostic Efficacy Surface:

[0066]

[0067] W is the variance measure.

[0068] Optimize the pressure combination through Monte Carlo policy gradient:

[0069]

[0070] For parameters The gradient of ; IE is the sampling integral.

[0071] The total effective sensor surface, perpendicular to the artery, measures 53mm ± 10% wide by 30mm ± 10% long. The collected pulse signal data is fed back to the MCU. Each of the 48 pressure sensors is packaged on a flexible printed circuit board (FPC). Three FPC boards are combined to simulate three fingers contacting the radial artery at the wrist, allowing for accurate positioning across the wrist. Simultaneously, pulse data collected by the Cun, Guan, and Chi pulse center sensor modules, as well as the upper and lower pulse position sensor modules, are fine-tuned to ensure that the pulse center sensor module remains positioned over the center of the wrist artery. The pulse pressure collection range is 0 mmHg to 250 mmHg, with a maximum allowable error of ±10%. The pulse rate display range is 40 to 200 beats / min, with a resolution of 1 beat / min and a maximum allowable error of ±3 beats / min.

[0072] The multi-parameter acquisition unit includes a PPG photoplethysmography sensor 13 , an ECG electrocardiogram sensor 14 , a body temperature sensor 15 and an EMG electromyography sensor 16 .

[0073] A PPG sensor 13 is attached to the wrist of the subject to monitor blood volume changes and obtain heart rate, oxygen saturation, blood pressure, and respiratory rate. An ECG sensor 14 is attached to the wrist of the subject to obtain electrocardiographic data. A temperature sensor 15 is attached to the wrist of the subject to obtain body temperature. An EMG sensor 16 is attached to the wrist of the subject to obtain electromyographic data.

[0074] The wireless communication unit 19 includes a wireless Bluetooth transmission module and / or a wireless WIFI transmission module.

[0075] The airbag, the stepping motor 9 , the pulse acquisition unit and the multi-parameter acquisition unit are enclosed in the sheath 22 .

[0076] The stress finger motion transmission mechanism 17 (servo motor and muscle hydraulic transmission mechanism) is divided into three stress finger tips, each with a transmission mechanism for applying pressure toward the pulse acquisition unit. The stress finger tips are equipped with built-in pressure feedback sensors 18, which automatically adjust the applied pressure based on the feedback pressure.

[0077] Furthermore, in this embodiment, the relationship between blood oxygen saturation and pulse pressure is expressed as follows:

[0078] The average value of pulse pressure P(t) can be expressed as

[0079] ;

[0080] The average value of volume pulse blood flow Q(t) is expressed as

[0081] ;

[0082] Where P(t) is the pulse pressure curve, Q(t) is the volume pulse blood flow curve, and p s is systolic blood pressure, p d is diastolic pressure, K=(p m -p d ) / (p s -p d ) is the waveform of pulse pressure, Q max is the maximum value of the blood flow waveform, Q min is the minimum value of the blood flow waveform, K'=(Q m -Q min ) / Q max -Q min ) is the waveform coefficient of volume pulse blood flow.

[0083] P m and Q m The two can be linked by the peripheral vascular resistance R (light absorption ratio), that is,

[0084] R=P m / Q m ;

[0085] Therefore, the relationship between volume pulse blood flow and pulse pressure can be expressed as

[0086] ;

[0087] According to the Lambert-Beer law and a large number of studies, there is a negative linear relationship between R and blood oxygen saturation. The lower the R, the greater the blood oxygen saturation.

[0088] SpO2=a+bR;

[0089] Therefore, the relationship between blood oxygen saturation and pulse pressure can be obtained as follows:

[0090] .

[0091] Example 2

[0092] The difference between this embodiment and embodiment 1 is that:

[0093] The device of this embodiment is provided with a housing 1, which allows for setting an applied force within a range of 0 to 300 mmHg, with a maximum allowable error of ±10%. The applied force can be displayed continuously or in steps within the range of 0 to 50 mmHg; 50 to 150 mmHg; and 150 to 260 mmHg, with a maximum allowable error of ±10%. The direct pressure measurement range for radial artery contact is 0 to 260 mmHg, with a maximum allowable error of ±5%. The pulse frequency feedback display range is 40 to 200 beats / min, with a resolution of less than 1 beat / min, and a maximum allowable error of ±1 beat / min. The effective surface of the multi-target sensor for radial artery contact, perpendicular to the wrist skin section, must not exceed 55 mm ±10% width by 35 mm ±10% length. The noise level of the device during normal operation should be no greater than 60 dB(A).

[0094] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0095] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A radial artery three-part pulse acquisition system based on artificial intelligence, characterized in that: include: An MCU control unit, and an air pump air bag unit, a stepper motor unit and a pulse acquisition unit respectively connected to the MCU control unit; the MCU control unit controls the air pump air bag unit to add / deflat, and simultaneously controls the stepper motor unit to move up and down, and further fine-tune the applied pressure value; the pulse acquisition unit collects pulse signal data based on the change of the pressure value and records it; The pulse acquisition unit is divided into a Cun pulse acquisition part, a Guan pulse acquisition part and a Chi pulse acquisition part; the Cun pulse acquisition part is further divided into a Cun pulse channel center sensor module, a Cun upper pulse position sensor module and a Cun lower pulse position sensor module; the Guan pulse acquisition part is further divided into a Guan pulse channel center sensor module, a Guan upper pulse position sensor module and a Guan lower pulse position sensor module; the Chi pulse acquisition part is further divided into a Chi pulse channel center sensor module, a Chi upper pulse position sensor module and a Chi lower pulse position sensor module; the pulse acquisition unit has a total of 9 stress zones, and the 9 stress zones are respectively subjected to action pressure combination to obtain the specific information conduction mechanism and markers of the corresponding organs of each zone; and use neural network learning; and feed back the collected pulse signal data to the MCU single chip microcomputer; The organ-specific neural network architecture based on the dynamic pressure combination of the nine stress zones is designed as follows, combining biomechanical conduction characteristics with deep learning algorithms: Multi-pressure spatiotemporal feature extraction: Input layer: Assume the i-th stress partition, i = 1 to 9, and the three-dimensional biological signal under k pressure gradients is: F p is the pressure component, Δx is the displacement component, σ v is the shear stress component, and R is the spatial component; It includes pressure, displacement, and shear stress components, and the sampling time is T; Time-frequency-space joint coding: using the improved Wigner-Ville distribution for time-frequency analysis: Extracting spatiotemporal features through 3D convolution: where Θ space It is a learnable three-dimensional convolution kernel; ReLU is a linear algebra operation that takes the real part; Organ-specific attention mechanism: Dynamic weight allocation: Define the pressure sensitivity coefficient of organ j to partition i: where q j is the query vector of the organ; LSTM is the long short-term memory network; Pressure-Organ Correlation Matrix: Constructing Organ Feature Maps: W j is the organ-specific projection matrix; Bidirectional gated graph conduction network: Biological conduction graph modeling: Constructing the adjacency matrix A∈0,1 based on the internal and external relationship of TCM viscera 9 ×9 , define graph convolution: G (l+1) =σ((D -1 / 2 AD -1 / 2 )G (l) Θ g Among them G (0) =[O1; ...; O9] T Bidirectional gated update: Multi-task dynamic decision making: Organ state prediction: in represents vector concatenation, For cross-partition aggregation; Loss function design: BCE is binary cross entropy; β is the weight coefficient; Θ is the outer product operation; Contains multi-task cross entropy and TCM relationship constraints; Pressure-Organ Response Surface Optimization: Define the Organ Diagnostic Efficacy Surface: Optimize the pressure combination through Monte Carlo policy gradient: is the gradient of the parameter θ; IE is the sampling integral.

2. The artificial intelligence-based radial artery three-part pulse acquisition system according to claim 1, characterized in that: The pulse signal data collected by the pulse acquisition unit is transmitted to the host computer via the Type-C unit or the wireless communication unit.

3. The artificial intelligence-based radial artery three-part pulse acquisition system according to claim 1, characterized in that: It also includes a multi-parameter acquisition unit connected to the MCU control unit, and the multi-parameter acquisition unit includes a PPG photoplethysmography sensor, an ECG electrocardiogram sensor, a body temperature sensor and an EMG electromyography sensor.

4. The artificial intelligence-based radial artery three-part pulse acquisition system according to claim 1, characterized in that: The air pump airbag unit includes an ultra-quiet pump, an airbag, an air pressure sensor and an electromagnetic valve; the ultra-quiet pump is connected to the airbag, the air pressure sensor and the electromagnetic valve respectively through an air pipe; there are multiple airbags, and limit sensors are placed between the airbags to prevent over-inflation.

5. The artificial intelligence-based radial artery three-part pulse acquisition system according to claim 1, characterized in that: The stepper motor unit includes a stepper motor, and the stepper motor is provided with an upper and lower travel limit switch for applying pressure to the wrist.

6. The artificial intelligence-based radial artery three-part pulse acquisition system according to claim 3, characterized in that: The PPG photoplethysmography sensor is attached to the wrist of the person being measured to obtain heart rate, blood oxygen saturation, blood pressure and respiratory rate based on changes in blood volume; the ECG electrocardiogram sensor is attached to the wrist of the person being measured to obtain electrocardiogram data; the body temperature sensor is attached to the wrist of the person being measured to obtain body temperature; and the EMG electromyography sensor is attached to the wrist of the person being measured to obtain electromyography data.

7. The artificial intelligence-based radial artery three-part pulse acquisition system according to claim 1, characterized in that: The MCU control unit is also connected to the stress finger motion transmission mechanism, which is divided into three stress finger ends, which apply pressure toward the pulse acquisition unit respectively. The stress finger ends have built-in pressure feedback sensors, which automatically adjust the applied pressure according to the feedback pressure.

8. The artificial intelligence-based radial artery three-part pulse acquisition system according to claim 1, characterized in that: The pulse acquisition unit is packaged on an FPC soft board, and three FPC soft boards are combined together to simulate three fingers contacting the radial artery at the wrist, so as to traverse the lateral position of the wrist.

Images