Control method and system for an arterial compression device

Abstract

The application relates to a control method and system of an arterial compression device, and solves the problem that neither unified parameter setting of a mechanical fixed mode nor simple feedback logic of a primary electronic regulation mode establishes deep connection with blood vessel characteristics and individual conditions, leading to mismatching of compression parameters and actual hemostasis requirements, the method comprising the following steps: real-time acquisition of current pressure of a compression unit and cumulative duration during timing, when the cumulative duration reaches a time interval of a preset curve, a decompression operation is performed according to a corresponding decompression value of a personalized decompression curve; if the current pressure deviates from a theoretical value of the curve by more than a preset deviation value, the decompression amplitude is simultaneously corrected according to the deviation value, and the correction amount is positively correlated with a blood vessel elasticity parameter; when the cumulative duration reaches a total compression time set by the curve, the termination pressure set by the curve is regulated and maintained for a preset duration. The application has the following effects: individualized precise compression is realized, and hemostasis safety and patient comfort are improved.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, and in particular to a control method and system for an arterial compression device. Background Technology

[0002] Arterial compression devices are the core equipment for achieving hemostasis at the puncture site after interventional diagnosis and treatment. By applying controllable pressure to the arterial puncture site to block blood flow, they provide a stable environment for vascular healing. The scientific nature of their control method is directly related to hemostasis efficiency, patient safety, and postoperative comfort. They have irreplaceable clinical value in the fields of cardiovascular intervention and neurointervention.

[0003] Currently, the control methods of arterial compression devices mainly include mechanical fixation mode and primary electronic adjustment mode. Mechanical fixation mode sets the compression force through preset mechanical structures (such as screw knobs and springs) and completes decompression at fixed time intervals, which cannot respond to individual differences such as vascular elasticity and arterial type. Although primary electronic adjustment mode introduces pressure sensors and simple timing logic, it can adjust the decompression rhythm based on real-time pressure feedback, but it mostly relies on a single pressure parameter or fixed algorithm.

[0004] Regarding the aforementioned technologies, the inventors have discovered the following drawbacks: neither the unified parameter settings of the mechanical fixation mode nor the simple feedback logic of the primary electronic adjustment mode have established a deep correlation with vascular characteristics and individual conditions, resulting in a mismatch between the compression parameters and the actual hemostasis requirements. This can cause patient discomfort in mild cases, and in severe cases, lead to clinical risks due to excessive or insufficient compression, thus affecting the safety of treatment. Summary of the Invention

[0005] To achieve personalized and precise compression, and improve hemostasis safety and patient comfort, this application provides a control method and system for an arterial compression device.

[0006] In a first aspect, this application provides a control method for an arterial compression device, employing the following technical solution:

[0007] A method for controlling an arterial compression device, comprising:

[0008] It receives the artery type and vascular elasticity parameters at the puncture site, automatically matches the baseline range of initial pressure, total compression time and termination pressure, and generates a personalized decompression curve with preset time intervals and corresponding decompression values ​​that are differentiated according to artery type after adjustment based on patient characteristics.

[0009] Based on a personalized decompression curve, the preset pressure regulation module is controlled to inflate the preset compression unit to the initial pressure, and timing is synchronized.

[0010] During the timing period, the current pressure and cumulative duration of the compression unit are collected in real time. When the preset interval is reached, the pressure is reduced according to the personalized decompression curve. If the current pressure deviates from the theoretical value of the curve by more than the preset value, the decompression amplitude is adjusted according to the deviation and the vascular elasticity parameters.

[0011] When the cumulative duration reaches the set total pressure time, the pressure is controlled according to the set termination pressure and maintained for the preset duration. After the pressure sensor continuously monitors and confirms that the pressure fluctuation is less than the preset pressure fluctuation, the pressure is released and a prompt is given.

[0012] By adopting the above technical solution, this method adapts to individual differences through personalized decompression curves, dynamically corrects the decompression amplitude, and combines fluctuation monitoring to accurately release pressure, thereby improving compression accuracy, reducing the risk of bleeding or ischemia, and ensuring postoperative hemostasis safety and patient comfort.

[0013] Secondly, this application provides a control system for an arterial compression device, which adopts the following technical solution:

[0014] A control system for an arterial compression device includes a memory, a processor, and a program stored in the memory and executable on the processor, the program being loaded and executed by the processor to implement the control method of the arterial compression device as described in the first aspect. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of a control method for an arterial compression device according to an embodiment of this application.

[0016] Figure 2 This is a schematic diagram of the process of decompression according to a personalized decompression curve when a preset interval is reached, according to another embodiment of this application. Detailed Implementation

[0017] The present application will be further described in detail below with reference to the accompanying drawings.

[0018] Reference Figure 1 The present application discloses a method for controlling an arterial compression device, comprising:

[0019] Step S100: Receive the arterial type and vascular elasticity parameters of the puncture site, automatically match the baseline range of initial pressure, total compression time and termination pressure, and generate a personalized decompression curve with preset time intervals and corresponding decompression values ​​that are differentiated according to arterial type after adjustment based on patient characteristics.

[0020] Among these, artery type refers to the type of artery at the puncture site, such as the femoral artery or radial artery. Different arteries require different compression strategies. Artery type can be automatically identified using technologies such as ultrasound imaging.

[0021] Vascular elasticity parameters: These are parameters reflecting vascular elasticity, such as elastic modulus. They affect the intensity and duration of compression. Vascular elasticity parameters can be obtained through automatic measurement or database matching. Automatic measurement involves the device's built-in sensors collecting vascular elasticity parameters in real time. Database matching involves matching reference values ​​from a database based on the patient's basic information.

[0022] Initial pressure: The pressure value at the start of compression, which needs to be set according to artery type and vascular elasticity parameters. Total compression time: The total duration from the start to the end of compression, determined based on artery type and patient condition. Termination pressure: The pressure value maintained at the end of compression, usually lower to prevent rebound bleeding. Personalized decompression curve: A decompression strategy developed based on patient characteristics and artery type. Preset time interval: Time points set in the decompression curve for phased decompression. Corresponding decompression value: The decompression amplitude performed at each preset time interval.

[0023] The necessary process is described below:

[0024] 1. Receiving input information: The system receives artery type and vascular elasticity parameters, which can be obtained through manual input, automatic measurement, or database matching.

[0025] 2. Matching the baseline range: Based on the artery type and vascular elasticity parameters, the system matches the baseline range of initial pressure, total compression time, and termination pressure from a preset database. For example, for the femoral artery, the baseline range might be an initial pressure of 80-120 mmHg, a total compression time of 30-60 minutes, and a termination pressure of 20-30 mmHg.

[0026] 3. Adjust to a personalized range: Adjust the baseline range based on individual patient characteristics (such as age, gender, and weight). For example, elderly patients may require a higher initial pressure and a shorter total compression time.

[0027] 4. Generate personalized decompression curves: Based on the adjusted parameters and the differences in arterial type, the system sets preset time intervals and corresponding decompression values. For example, for the femoral artery, a total time of 40 minutes can be set, with decompression every 5 minutes and the decompression value gradually decreasing until the termination pressure is reached.

[0028] Step S200: Based on the personalized decompression curve, control the preset pressure adjustment module to inflate the preset compression unit to the initial pressure, and time the process simultaneously.

[0029] The system includes: a pressure regulation module (controlling the inflation and deflation of the compression unit, typically comprising an electric air pump and a solenoid valve); a compression unit (the component that contacts and applies pressure to the patient's puncture site, such as an inflatable cuff); an initial pressure (the pressure value set at the start of compression, determined based on a personalized decompression curve); and a preset timing module (recording the cumulative compression duration, usually linked to the control module).

[0030] The necessary process is described below:

[0031] 1. Receive personalized decompression curve: The system receives the personalized decompression curve generated in step S100 and obtains the initial pressure, total compression time and other relevant parameters.

[0032] 2. Controlling the inflation of the pressure regulation module: Based on the initial pressure value in the personalized decompression curve, the system controls the pressure regulation module to inflate the compression unit until the set initial pressure is reached. For example, if the initial pressure is 100 mmHg, the system will control the electric air pump to inflate the airbag until the pressure sensor detects that the pressure inside the airbag has reached 100 mmHg.

[0033] 3. Start the timing module: As soon as the compression unit reaches the initial pressure, the system simultaneously starts the timing module to begin recording the cumulative compression time. The timing module will continuously monitor and record the compression process time, providing a time reference for subsequent decompression operations.

[0034] In step S300, the current pressure and cumulative duration of the compression unit are collected in real time during the timing period. When the preset interval is reached, the pressure is reduced according to the personalized decompression curve. If the current pressure deviates from the theoretical value of the curve by more than the preset value, the decompression amplitude is adjusted according to the deviation and the vascular elasticity parameters.

[0035] Among them, current pressure: the pressure value recorded by the compression unit in real-time monitoring. Cumulative duration: the total time from the start of compression to the current moment. Preset interval: the time node set in the personalized decompression curve, used for staged decompression operations. Preset value: the maximum allowable error range between the current pressure and the theoretical value.

[0036] The necessary process is described below:

[0037] 1. Real-time monitoring: During the compression process, the system collects the current pressure and cumulative duration of the compression unit in real time. For example, the system collects the current pressure and cumulative duration data once per second.

[0038] 2. Time Interval Determination: When the accumulated time reaches the preset time interval in the personalized stress reduction curve, the system prepares to perform the stress reduction operation. For example, if the preset time interval is 5 minutes, the system will prepare to perform the stress reduction operation when the accumulated time reaches 5 minutes.

[0039] 3. Perform pressure reduction operation: The system performs pressure reduction operation according to the corresponding pressure reduction value in the personalized pressure reduction curve. For example, if the pressure reduction value corresponding to the current time interval is 10 mmHg, the system will reduce the pressure of the compression unit by 10 mmHg.

[0040] 4. Determine pressure deviation: If the current pressure deviates from the theoretical value of the personalized pressure reduction curve by more than the preset deviation value, the system will adjust the pressure reduction amplitude. For example, if the preset deviation value is 5 mmHg, and the current pressure deviates from the theoretical value by 8 mmHg, then the pressure reduction amplitude needs to be adjusted.

[0041] 5. Correction for decompression: The correction amount is positively correlated with the vascular elasticity parameter; that is, the correction amount is larger when the vascular elasticity is poor. For example, if the vascular elasticity parameter is low, the correction amount may be 1.2 times the deviation value; if the vascular elasticity parameter is high, the correction amount may be 0.8 times the deviation value. The specific correction formula can be: Correction amount = Deviation value × Correction coefficient (the correction coefficient is adjusted according to the vascular elasticity parameter).

[0042] The process of simultaneously adjusting the decompression amplitude based on the deviation value, with the adjustment amount being positively correlated with vascular elasticity parameters, includes the following steps:

[0043] Step 1: Collect the current pressure, process it through a preset filter, and compare it with the theoretical value of the corresponding duration of the personalized decompression curve to obtain the deviation ΔP. Judge according to the preset threshold (set according to artery type). If the deviation exceeds the limit, start the correction.

[0044] Preset filtering: The Kalman filter algorithm (preset Q=0.02, R=0.6) is used to eliminate environmental vibration and sensor noise. Q is the process noise covariance and R is the measurement noise covariance.

[0045] Personalized decompression curve: A pressure-time change curve pre-generated based on the artery type at the puncture site (e.g., femoral artery, radial artery), vascular elasticity parameters, and individual patient characteristics (e.g., weight, blood pressure);

[0046] Theoretical value: The pressure setting value corresponding to the current cumulative compression duration on the personalized decompression curve;

[0047] Deviation ΔP: The difference between the current pressure and the theoretical value, i.e., ΔP = current pressure - theoretical value;

[0048] Preset thresholds: set differently according to artery type, with a preset value of ±5 mmHg for the femoral artery and ±3 mmHg for the radial artery.

[0049] The general processing procedure is as follows: The current pressure is collected by the piezoresistive pressure sensor (sampling frequency preset to 10Hz) built into the compression unit. After the above Kalman filtering process, the theoretical value that matches the current cumulative duration in the personalized decompression curve is extracted, and ΔP is calculated. If |ΔP| > the preset threshold of the corresponding artery type, and this state lasts for 2 sampling cycles (preset 0.2 seconds), the correction logic is triggered.

[0050] Step 2: Use a preset weighted algorithm to calculate the basic correction amount for the cumulative deviation within the preset window, thereby reducing the impact of single fluctuations.

[0051] Among them, the preset weighted algorithm refers to the calculation rule that assigns differentiated weights to the deviations at different time points. Here, the exponentially weighted moving average (EWMA) algorithm is used, with a higher weight for recent deviations; the preset window refers to the time range used for accumulating deviations, which is preset to 3 consecutive sampling periods (0.2 seconds per period, for a total of 0.6 seconds); the cumulative deviation refers to the set of deviation values ​​ΔP of each sampling period within the window in a time series; the basic correction amount is the preliminary adjustment value calculated based on the cumulative deviation, providing a benchmark for subsequent adjustments in conjunction with vascular elasticity parameters.

[0052] Calculation process: Extract the deviation value sequence {ΔP1 (current), ΔP2 (previous period), ΔP3 (previous two periods)} within the preset window, and calculate the basic correction amount using preset weighting coefficients (0.6, 0.3, 0.1): 0.6 × ΔP1 + 0.3 × ΔP2 + 0.1 × ΔP3. If there is a sudden change in the sign of the deviation within the window (e.g., ΔP1 is positive, ΔP2 is negative), the weight of the sudden deviation is automatically reduced (multiplied by a coefficient of 0.5) to further weaken the interference of instantaneous fluctuations.

[0053] Step 3: Call up the vascular elasticity parameters and adjust the basic correction amount according to the preset positive correlation formula. The higher the vascular elasticity, the greater the correction.

[0054] Preset positive correlation formula: refers to the mathematical relationship in which the correction amount increases with the increase of vascular elasticity parameters, and is preset as a linear proportional formula;

[0055] Adjusted correction amount: The final adjusted value after weighting the baseline correction amount by vascular elasticity parameters.

[0056] Acquisition and calculation process: The elastic modulus E of the puncture site obtained by ultrasound elastography before the operation is called (e.g., femoral artery E=120kPa, radial artery E=90kPa). The preset standard reference value E0=100kPa is used to calculate the following formula: Adjusted correction amount = basic correction amount × (E / E0).

[0057] Example: If the base correction is 10 mmHg and E = 150 kPa, then the adjusted correction is 10 × (150 / 100) = 15 mmHg, which shows that the higher the vascular elasticity, the greater the correction range; if E = 80 kPa, then the adjusted correction is 8 mmHg, which weakens the over-adjustment of low-elasticity blood vessels.

[0058] Step 4: Combine the preset upper limit constraint correction amount of the current decompression stage with the hardware adjustment capability.

[0059] Among them, the current decompression stage refers to the pressure regulation stage divided according to the personalized decompression curve, including the rapid regulation period (initial decompression stage), the fine regulation period (intermediate stable stage), and the closing regulation period (near the termination stage); the preset upper limit is the maximum allowable correction amount set for each stage, based on the hardware regulation accuracy and vascular safety preset; the constraint correction amount refers to the final execution regulation value after being limited by the stage upper limit; the hardware regulation capability refers to the maximum rate and accuracy of the pressure regulation module to drive the deflation through PWM pulses (e.g., the maximum single deflation volume corresponds to 15 mmHg).

[0060] Constraint Process: Extract the current decompression stage and call the preset upper limits for each stage: rapid adjustment phase ≤ 15 mmHg, fine adjustment phase ≤ 10 mmHg, and final adjustment phase ≤ 5 mmHg. Compare the adjusted correction amount obtained in step 3 with the corresponding stage upper limit, and take the minimum of the two as the constraint correction amount. Example: If the adjusted correction amount is 12 mmHg, and the current stage is fine adjustment phase (upper limit 10 mmHg), then the constraint correction amount = 10 mmHg; if the current stage is rapid adjustment phase, then execute according to 12 mmHg, ensuring that the correction amount does not exceed the hardware safety adjustment range and vascular tolerance threshold.

[0061] Step 5: After correction, monitor the deviation. If it still exceeds the preset ratio, extend the current stage by the preset duration and repeat the correction. Simultaneously, feed the data back to the curve library to optimize subsequent preset curves.

[0062] Execution Process: After correction, pressure is continuously monitored at a preset frequency of 10Hz, and the post-correction deviation is calculated. If |post-correction deviation| > original trigger threshold × 60% (e.g., if the original threshold for the femoral artery is 5 mmHg, then 3 mmHg is the critical value), the current decompression phase is automatically extended by 60 seconds, and the correction process of steps 1-4 is repeated; if the deviation meets the target, the original curve continues to be executed. Simultaneously, the vascular elasticity parameters, deviation values, correction amounts, and patient characteristics (such as age and artery type) involved in this correction are written into the curve library. The data of patients with similar conditions are grouped using a preset K-means clustering algorithm, and the preset parameters of the subsequent initial curves for the corresponding groups are optimized (e.g., improving the initial correction sensitivity for patients with highly elastic blood vessels).

[0063] Step S400: When the cumulative duration reaches the set total pressure time, the pressure is controlled according to the set termination pressure and maintained for the preset duration. After the pressure sensor continuously monitors and confirms that the pressure fluctuation is less than the preset pressure fluctuation, the pressure is released and a prompt is given.

[0064] The parameters are as follows: Total Compression Time: The total duration from the start to the end of compression, set by the personalized decompression curve. Termination Pressure: The pressure value maintained at the end of compression, typically lower for a smooth transition. Preset Duration: The duration of the termination pressure to ensure stability at the end of compression. Preset Pressure Fluctuation: The allowable pressure fluctuation range used to determine if the compression process is stable. Notification: A system-issued notification signal informing medical staff that the compression process has ended.

[0065] The necessary process is described below:

[0066] 1. Determine the total compression time: When the cumulative compression time reaches the total compression time set by the personalized decompression curve, the system prepares to enter the termination phase. For example, if the total compression time is 40 minutes, the system will enter the termination phase when the cumulative time reaches 40 minutes.

[0067] 2. Adjusting the termination pressure: The system adjusts the pressure of the compression unit to the termination pressure set by the curve and maintains that pressure for a period of time (preset duration). For example, if the termination pressure is 25 mmHg and the preset duration is 5 minutes, the system will adjust the pressure to 25 mmHg and maintain it for 5 minutes.

[0068] 3. Monitoring pressure fluctuations: During the maintenance of the final pressure, the system continuously monitors pressure fluctuations via a pressure sensor. If the pressure fluctuation is less than the preset pressure fluctuation (e.g., less than 3 mmHg), the compression process is considered stable.

[0069] 4. Release pressure and issue a notification: Once the pressure fluctuation stabilizes, the system releases the pressure from the compression unit and issues a termination notification. For example, the system may use a buzzer to sound an alarm, informing medical staff that the compression process is complete.

[0070] 5. Data Storage: The system stores the correlation data between pressure and time throughout the procedure, as well as characteristic parameters of the puncture site. For example, the stored data includes pressure values ​​per second, cumulative duration, artery type, and vascular elasticity parameters, facilitating subsequent analysis and recording.

[0071] Reference Figure 2 When the preset interval is reached, pressure reduction is performed according to the personalized pressure reduction curve, including:

[0072] Step S310: Based on the correspondence between the current cumulative duration and the personalized decompression curve, the target decompression value is calculated by calling the pre-stored pressure-time correlation model.

[0073] Among them, the current cumulative duration is the total time from the start of compression to the current moment. The personalized decompression curve is a decompression strategy tailored to the patient's individual characteristics and artery type, including preset time intervals and corresponding decompression values. The pressure-time correlation model is a pre-stored mathematical model used to calculate the target decompression value based on the current cumulative duration and the personalized decompression curve. The target decompression value is the decompression level to be implemented based on the personalized decompression curve and the pressure-time correlation model at the current cumulative duration.

[0074] The necessary process is described below:

[0075] 1. Get Current Cumulative Duration: The system retrieves the current cumulative duration from the timing module. This time is the total time from the start of the compression to the current moment. For example, the current cumulative duration is 15 minutes.

[0076] 2. Matching Personalized Stress Reduction Curves: The system searches for the corresponding stress reduction stage and value in the personalized stress reduction curve based on the current cumulative duration. For example, if the personalized stress reduction curve is set to reduce stress every 5 minutes, and the current cumulative duration is 15 minutes, it corresponds to the 3rd stress reduction stage.

[0077] 3. Invoke the pressure-time correlation model: The system invokes a pre-stored pressure-time correlation model, inputting the current cumulative duration and relevant parameters of the personalized decompression curve. The pressure-time correlation model is a mathematical model, typically built based on clinical data and experimental results, capable of calculating the target decompression value based on time and curve parameters.

[0078] 4. Calculate the target decompression value: The pressure-time correlation model calculates the target decompression value based on the input current cumulative duration and personalized decompression curve parameters. For example, the model calculates the target decompression value to be 8 mmHg at 15 minutes.

[0079] In step S320, the pressure data collected in real time by the preset pressure sensor is processed by Kalman filtering, and then fused with the three-dimensional activity data collected by the preset acceleration sensor through an attention mechanism network. The pressure data weight is dynamically reduced to eliminate false pressure fluctuations caused by limb swinging.

[0080] Among them, the pressure sensor is used to monitor the current pressure of the compression unit in real time. The Kalman filter is a highly efficient self-recursive filter used to estimate the dynamic state of the system from a series of noisy measurements, effectively reducing measurement noise. The accelerometer is used to monitor the patient's limb movements, capable of acquiring three-dimensional activity data (such as acceleration and direction).

[0081] 3D Activity Data: Three-dimensional data on limb movement collected by accelerometers, including information such as acceleration, velocity, and direction. Attention Mechanism Network: A deep learning model that automatically learns important features in the data and assigns different weights to different features, thereby improving the accuracy of data fusion. Dynamic Down-Adjustment of Pressure Data Weights: During the fusion process, the weights of pressure data are adjusted based on the limb activity data to reduce the influence of limb movement on pressure measurement.

[0082] The necessary process is described below:

[0083] 1. Real-time pressure data acquisition: The system acquires the current pressure data of the compression unit in real time through a preset pressure sensor. For example, the pressure sensor acquires pressure data once per second.

[0084] 2. Kalman Filtering: The acquired pressure data is processed using Kalman filtering to reduce measurement noise and improve data accuracy. The Kalman filter dynamically adjusts the filtering parameters through prediction and update steps, thereby effectively removing noise.

[0085] 3. Acquisition of 3D Activity Data: The system acquires the patient's 3D activity data in real time through a pre-set accelerometer, including information such as acceleration, velocity, and direction. For example, the accelerometer acquires 3D activity data once per second.

[0086] 4. Data Fusion: An attention mechanism network is used to fuse the filtered pressure data with the 3D activity data. This network automatically learns important features from the data and assigns different weights to different features. During the fusion process, the weights of the pressure data are dynamically adjusted based on the limb activity data to eliminate spurious pressure fluctuations caused by limb movements.

[0087] 5. Dynamically adjust weights: When limb activity is significant, increase the weight of activity data and decrease the weight of pressure data to reduce the impact of spurious fluctuations. When limb activity is minimal, maintain the weight of pressure data to ensure the accuracy of pressure monitoring.

[0088] Step S330: After confirming the effective pressure reduction requirement, a preset segmented PWM pulse algorithm is used to drive the preset pressure regulation module. The gas release rate is controlled by the built-in current limiting structure of the pressure regulation module to ensure a smooth pressure reduction process.

[0089] The effective decompression requirement refers to the decompression operation confirmed after step S320. The segmented PWM pulse algorithm is an algorithm that controls the gas release rate through pulse width modulation (PWM) to execute the decompression operation in segments. The pressure regulation module is a device used to control the inflation and deflation of the compression unit, typically including an electric air pump and a solenoid valve. The flow-limiting structure is a structure built into the pressure regulation module used to limit the maximum gas release rate, ensuring a smooth decompression process. The deflation rate is the speed at which gas is released from the compression unit during the decompression process.

[0090] The necessary process is described below:

[0091] 1. Confirm valid pressure reduction requirement: Based on the data processed in step S320, the system confirms whether a pressure reduction operation is needed. For example, if the pressure data after filtering and fusion indicates a pressure reduction of 8 mmHg, then it is confirmed as a valid pressure reduction requirement.

[0092] 2. Invoking the Segmented PWM Pulse Algorithm: The system invokes a preset segmented PWM pulse algorithm, which divides the decompression process into multiple stages, with each stage controlling gas release through pulse width modulation. For example, the decompression requirement of 8 mmHg can be divided into 4 stages, with each stage reducing the pressure by 2 mmHg.

[0093] 3. Perform segmented pressure reduction: In each stage, the system controls the opening and closing of the solenoid valve through PWM pulses to release a certain amount of gas. For example, each stage lasts for 5 seconds, and the solenoid valve opens in a pulse manner, opening for 0.5 seconds and closing for 0.5 seconds each time, cycling 10 times to complete a pressure reduction of 2 mmHg.

[0094] 4. Flow-limiting structure controls the gas release rate: The flow-limiting structure built into the pressure regulating module limits the maximum rate of gas release, ensuring a smooth decompression process. For example, the flow-limiting structure can be a small orifice that limits the maximum gas flow rate, preventing instability caused by excessively rapid decompression.

[0095] 5. Monitoring the pressure reduction process: During the pressure reduction process, the system monitors pressure changes in real time to ensure that the pressure reduction process follows the predetermined segmented PWM pulse algorithm. If the system detects that the pressure change is too rapid or unstable, it will automatically adjust the PWM pulse parameters to slow down the pressure reduction rate.

[0096] The data is fused with 3D activity data collected by a pre-set accelerometer through an attention mechanism network. This process dynamically down-weights the pressure data to eliminate spurious pressure fluctuations caused by limb movement, including:

[0097] Step S321: Activate the accelerometer, adjust the sampling frequency according to the characteristics of the puncture site, collect three-dimensional activity data, and perform data preprocessing.

[0098] Accelerometer: A sensor capable of detecting the acceleration of an object, typically used to monitor limb movement. Sampling frequency: The frequency at which the sensor collects data, measured in Hertz (Hz), representing the number of data acquisitions per second. Puncture site characteristics: The physiological characteristics and common activity intensities of the puncture site; for example, femoral artery puncture sites typically exhibit higher activity intensities, while radial artery puncture sites show lower activity intensities. Three-dimensional activity data: Three-dimensional data on limb movement collected by the accelerometer, including the magnitude and direction of acceleration.

[0099] The necessary process is described below:

[0100] 1. Activate the accelerometer: The system activates the preset accelerometer to prepare for collecting limb movement data. For example, the system activates a triaxial accelerometer to monitor limb movement at the puncture site.

[0101] 2. Adjust the sampling frequency: Adjust the sampling frequency of the accelerometer according to the characteristics of the puncture site to match the common activity intensities at that site. For example, for femoral artery puncture sites, due to the higher activity intensity, a higher sampling frequency (e.g., 100Hz) may be needed; while for radial artery puncture sites, the activity intensity is lower, and the sampling frequency can be appropriately reduced (e.g., 50Hz). The adjustment of the sampling frequency can be based on clinical experience and experimental data to ensure accurate capture of key features of limb movement.

[0102] 3. Synchronous acquisition of 3D activity data: The accelerometer synchronously acquires 3D activity data, including the magnitude and direction of acceleration, at an adjusted sampling frequency. For example, the sensor acquires data 100 times per second, recording the acceleration value and direction at each sampling moment.

[0103] Step S322: Based on the preprocessed data, axial weights are assigned according to the location features to filter irrelevant signals and extract key activity features, including duration, acceleration peak threshold, and frequency.

[0104] The data includes: Axial weights: Weights assigned to each axis (X, Y, Z) based on the activity characteristics of the puncture site, used to highlight the main activity direction. Preprocessing: Preliminary processing of the raw data, including filtering, denoising, and normalization. Distal limb-irrelevant activity signals: Signals unrelated to the activity at the puncture site, such as minute movements from the distal end of the limb (e.g., fingers or toes), which may interfere with the monitoring of the main activity. Preprocessed data: Three-dimensional activity data after filtering, denoising, and normalization. Key activity features: Important features extracted from the preprocessed data that characterize limb activity. Duration: The duration of limb activity. Peak acceleration threshold: A threshold for peak acceleration set according to the characteristics of the puncture site, used to distinguish between normal and abnormal activity. Frequency: The frequency of limb activity, i.e., the number of movements per unit time.

[0105] The necessary process is described below:

[0106] 1. Receiving 3D Activity Data: The system receives 3D activity data collected by the accelerometer, including acceleration values ​​along the X, Y, and Z axes. For example, the sensor may collect data as follows: X-axis acceleration 0.2 m / s², Y-axis acceleration 0.1 m / s², and Z-axis acceleration 0.3 m / s².

[0107] 2. Data Preprocessing: The acquired 3D activity data is preprocessed, including filtering, denoising, and normalization, to improve data quality. For example, a low-pass filter is used to remove high-frequency noise, making the data smoother.

[0108] 3. Assign weights to axes: Based on the activity characteristics of the puncture site, assign weights to each axis to highlight the main direction of activity. For example, for the femoral artery puncture site, the main direction of activity may be the Z-axis (vertical direction), so a higher weight can be assigned to the Z-axis (e.g., 0.6), while the X-axis and Y-axis can be assigned lower weights (e.g., 0.2 and 0.2, respectively).

[0109] 4. Filter irrelevant activity signals: Identify and filter signals that are unrelated to the activity at the puncture site, such as minute movements from the distal extremities. For example, by setting a threshold (e.g., acceleration less than 0.1 m / s²), signals below this threshold can be filtered out. These signals typically originate from minute movements in the distal extremities and have little impact on the monitoring of major activities.

[0110] 5. Extracting Duration: The system analyzes the preprocessed data to extract the duration of limb activities. For example, by detecting the start and end points of acceleration data, the duration of the activity is calculated. Assuming the start point of an activity is time point t1 and the end point is time point t2, then the duration is t2−t1.

[0111] 6. Set a peak acceleration threshold: Based on the characteristics of the puncture site, set a threshold for the peak acceleration. For example, for a femoral artery puncture site, set the peak acceleration threshold to 0.5 m / s². Any acceleration value exceeding this threshold is considered a significant activity signal.

[0112] 7. Extracting peak acceleration values: The system analyzes the preprocessed data and extracts peak acceleration values ​​that exceed a threshold. For example, the peak acceleration value detected from the preprocessed data is 0.6 m / s², which exceeds the set threshold of 0.5 m / s².

[0113] 8. Calculate Activity Frequency: The system analyzes preprocessed data and calculates the frequency of limb movements per unit time. For example, the activity frequency is calculated by counting the number of activities exceeding a threshold within a certain time window. Assuming three activities exceeding the threshold are detected within 10 seconds, the activity frequency is 0.3 Hz.

[0114] Step S323: Input the filtered pressure data and activity features into the attention mechanism network. The network dynamically allocates weights according to the activity features, and the weights of the pressure data are downgraded in stages when the limbs swing.

[0115] The data includes: Filtered pressure data: Pressure data processed by Kalman filtering to reduce measurement noise and improve data accuracy. Activity features: Key activity features extracted from the preprocessed data, including duration, peak acceleration, and activity frequency. Attention mechanism network: A deep learning model that automatically learns important features in the data and assigns different weights to different features, thereby improving the accuracy of data fusion. Dynamic weight allocation: The weights of each feature are dynamically adjusted according to its importance to highlight important features. Limb swing judgment criteria: Preset conditions used to determine whether limb activity constitutes swinging behavior, such as peak acceleration exceeding a threshold and high frequency. Swing intensity grading: Limb swing intensity is categorized into different levels to adjust the weights of the pressure data.

[0116] The necessary process is described below:

[0117] 1. Input Data Preparation: Input the pressure data processed by Kalman filtering and the extracted activity features (duration, peak acceleration, activity frequency) into the attention mechanism network. For example, the pressure data is 99.8 mmHg, and the activity features are a duration of 5 seconds, a peak acceleration of 0.6 m / s², and an activity frequency of 0.2 Hz.

[0118] 2. Attention Mechanism Network Processing: Attention mechanism networks dynamically assign weights based on the activity characteristics of the input, highlighting important features. For example, the network may assign a high weight (e.g., 0.6) to the acceleration peak, a medium weight (e.g., 0.3) to the duration, and a low weight (e.g., 0.1) to the activity frequency.

[0119] 3. Determine limb swing: Based on preset limb swing determination criteria, determine whether the current activity constitutes swinging behavior. For example, preset criteria include a peak acceleration exceeding 0.5 m / s² and an activity frequency exceeding 0.1 Hz. If the current activity characteristics meet these conditions, it is determined to be limb swinging.

[0120] 4. Dynamically reduce the weight of pressure data: When the activity characteristics meet the criteria for limb swing determination, the weight of pressure data is dynamically reduced according to the swing intensity level. For example, if the peak acceleration is 0.6 m / s², it belongs to medium-intensity swing, and the weight of pressure data is reduced from 1.0 to 0.7; if the peak acceleration is 0.8 m / s², it belongs to high-intensity swing, and the weight of pressure data is reduced from 1.0 to 0.5.

[0121] In step S324, the attention mechanism network performs fusion calculation on pressure data and activity characteristics through dynamic weights to generate an effective pressure signal and calculates the correlation quantification index between its fluctuation and activity characteristics. The correlation quantification index includes the fluctuation start time difference and intensity change correlation coefficient.

[0122] The data includes: Pressure data: Pressure data after Kalman filtering. Activity characteristics: Key activity characteristics extracted from the preprocessed data, including duration, peak acceleration, and activity frequency. Fusion calculation: Combining the pressure data and activity characteristics, and generating the final effective pressure signal through weighted summation or other mathematical methods. Effective pressure signal: The true pressure signal after fusion calculation, after eliminating spurious fluctuations, used for subsequent decompression decisions. Effective pressure signal fluctuation: The fluctuation of the pressure signal generated after fusion calculation. Activity characteristics: Key activity characteristics extracted from the preprocessed data, including duration, peak acceleration, and activity frequency. Correlation quantification index: A quantitative index used to measure the relationship between effective pressure signal fluctuation and activity characteristics. Fluctuation start time difference: The difference between the start time of effective pressure signal fluctuation and the start time of activity characteristic change. Intensity change correlation coefficient: A statistical index measuring the correlation between the intensity of effective pressure signal fluctuation and the intensity change of activity characteristics.

[0123] The necessary process is described below:

[0124] 1. Receiving Input Data: The attention mechanism network receives pressure data and extracted activity features after Kalman filtering. For example, the pressure data is 99.8 mmHg, and the activity features include a duration of 5 seconds, a peak acceleration of 0.6 m / s², and an activity frequency of 0.2 Hz.

[0125] 2. Application of Dynamic Weights: The attention mechanism network weights the pressure data and activity features according to the dynamic weights of the activity features. For example, the dynamic weight allocation is as follows: pressure data weight: 0.7; duration weight: 0.3; peak acceleration weight: 0.6; activity frequency weight: 0.1.

[0126] 3. Fusion Calculation: The weighted pressure data and activity characteristics are fused and calculated to generate an effective pressure signal. The fusion calculation formula can be expressed as: Effective pressure signal = Pressure data × Pressure data weight + ∑(Activity characteristics × Activity characteristic weight). For example: Effective pressure signal = 99.8 × 0.7 + (5 × 0.3 + 0.6 × 0.6 + 0.2 × 0.1).

[0127] 4. Generate an effective pressure signal: The calculation result is the effective pressure signal, which eliminates spurious pressure fluctuations caused by limb movement. For example, the calculation result is:

[0128] Effective pressure signal = 99.8 × 0.7 + (5 × 0.3 + 0.6 × 0.6 + 0.2 × 0.1) = 71.74 mmHg.

[0129] 5. Extract fluctuation characteristics: Extract fluctuation characteristics from the effective pressure signal, including the start time, duration, and amplitude of the fluctuation. For example, suppose the effective pressure signal drops from 100 mmHg to 95 mmHg, with a start time of t1 and a duration of 5 seconds.

[0130] 6. Extract activity feature changes: Extract change features from the activity features, including the start time, duration, and magnitude of the change. For example, suppose the peak acceleration increases from 0.2 m / s² to 0.6 m / s², with an start time of t2 and a duration of 5 seconds.

[0131] 7. Calculate the fluctuation start time difference: Calculate the difference between the start time of the effective pressure signal fluctuation and the start time of the change in activity characteristics. For example, the fluctuation start time difference is Δt = t1 − t2.

[0132] 8. Calculate the correlation coefficient of intensity change: Use statistical methods to calculate the correlation coefficient between the intensity of effective pressure signal fluctuation and the intensity change of activity characteristics.

[0133] The formula for calculating the correlation coefficient is:

[0134] ;

[0135] in, and These represent the changes in the effective pressure signal fluctuation intensity and the activity characteristic intensity, respectively. and These are their average values.

[0136] Step S325: When the fluctuation start time difference is less than or equal to the preset time difference threshold and the intensity change correlation coefficient is greater than or equal to the preset correlation coefficient threshold, it is determined to be a false fluctuation caused by limb swinging, and the current pressure is maintained for a preset duration appropriate to the puncture site.

[0137] The necessary process is described below:

[0138] 1. Result Confirmation: The system confirms whether a strong correlation is determined. If so, proceed to the next step. For example, according to the determination result in step S327, the fluctuation start time difference is 1.5 seconds, and the correlation coefficient of intensity change is 0.98, which meets the condition for strong correlation.

[0139] 2. Identify False Pressure Fluctuations: The system identifies current valid pressure signal fluctuations as false pressure fluctuations caused by limb movement. For example, the system confirms that current pressure fluctuations are caused by limb movement rather than actual pressure changes.

[0140] 3. Maintain current pressure: The system selects an appropriate preset duration to maintain the current pressure based on the characteristics of the puncture site. For example, the preset duration may be 30 seconds for femoral artery puncture sites and 20 seconds for radial artery puncture sites.

[0141] 4. Perform pressure maintenance operation: The system maintains the current pressure unchanged for a preset time to avoid unnecessary pressure reduction operations caused by false pressure fluctuations. For example, after confirming a false pressure fluctuation, the system maintains the current pressure at 95 mmHg for 30 seconds.

[0142] Step S326: Otherwise, it is considered a true pressure decay, triggering the preset pressure reduction operation logic.

[0143] The pressure reduction operation logic includes a preset pressure reduction operation procedure, comprising parameters such as the pressure reduction rate, amplitude, and time. For example, the pressure reduction operation logic may include: pressure reduction rate: 5 mmHg per minute; pressure reduction amplitude: 10 mmHg per reduction; pressure reduction time: continuous pressure reduction until the target pressure is reached.

[0144] The preset pressure regulation module, driven by a preset segmented PWM pulse algorithm, includes:

[0145] Step S331: Initialize pulse parameters based on puncture site and personalized decompression curve. Specifically, this includes setting preset base frequency and initial duty cycle according to site, and correcting preset segmented thresholds in combination with individual patient characteristics.

[0146] Among them, the preset base frequency is the fundamental frequency of the PWM pulse, measured in Hertz (Hz), representing the number of pulses per second. The initial duty cycle is the initial duty cycle of the PWM pulse, representing the ratio of the pulse's high-level time to the total cycle time. Segmented thresholds are thresholds that divide the decompression process into different stages, used to distinguish between rapid adjustment, fine adjustment, and termination adjustment. Individual patient characteristics include the patient's age, gender, weight, and vascular elasticity, which affect parameter adjustments during the decompression process.

[0147] The necessary process is described below:

[0148] 1. Determine the puncture site: The system determines the type of artery being operated on based on the input puncture site information. For example, the puncture site is the femoral artery.

[0149] 2. Initialize pulse parameters: Set the preset base frequency and initial duty cycle according to the characteristics of the puncture site. For example, for the femoral artery, the preset base frequency is 100Hz and the initial duty cycle is 50%.

[0150] 3. Adjust segmentation thresholds based on individual patient characteristics: Adjust the preset segmentation thresholds according to the patient's individual characteristics (such as age, gender, weight, vascular elasticity, etc.). For example, for elderly patients, the segmentation thresholds may need to be adjusted to accommodate their poor vascular elasticity. The adjusted segmentation thresholds might be as follows: Rapid adjustment phase: pressure deviation > 10 mmHg; Fine adjustment phase: pressure deviation between 5 mmHg and 10 mmHg; Final adjustment phase: pressure deviation ≤ 5 mmHg.

[0151] 4. Generate initial pulse parameters: The system generates initial pulse parameters based on the above information, including the fundamental frequency, initial duty cycle, and corrected segmented thresholds. For example, the generated initial pulse parameters are: fundamental frequency: 100Hz; initial duty cycle: 50%.

[0152] The segmented thresholds are as follows: 1. Rapid adjustment stage: pressure deviation > 10 mmHg; 2. Fine adjustment stage: pressure deviation between 5 mmHg and 10 mmHg; 3. Final adjustment stage: pressure deviation ≤ 5 mmHg.

[0153] Step S332: Based on the effective pressure signal, the current pressure is divided into three stages according to the deviation between the current pressure and the target pressure reduction value, based on the preset segmented threshold: rapid adjustment, fine adjustment, and final adjustment. Each stage adopts a preset combination of frequency and duty cycle.

[0154] The parameters include: Effective pressure signal: The actual pressure signal generated after fusion calculation, used for subsequent pressure reduction operations. Current pressure: The current pressure value of the compression unit monitored in real time. Target pressure reduction value: The pressure value to be achieved calculated based on the personalized pressure reduction curve. Deviation: The difference between the current pressure and the target pressure reduction value. Preset segmented thresholds: Thresholds that divide the pressure reduction process into different stages, used to distinguish between rapid adjustment, fine adjustment, and final adjustment. Rapid adjustment stage: The stage where pressure is rapidly reduced when the deviation is large. Fine adjustment stage: The stage where pressure is precisely adjusted when the deviation is moderate. Final adjustment stage: The stage where final adjustments are made to achieve the target pressure when the deviation is small. Frequency and duty cycle combination: The preset PWM pulse frequency and duty cycle for each stage, used to control the pressure reduction rate.

[0155] The necessary process is described below:

[0156] 1. Deviation Calculation: The system calculates the deviation between the current pressure and the target decompression value based on the effective pressure signal. For example, if the current pressure is 95 mmHg and the target decompression value is 85 mmHg, the deviation is 10 mmHg.

[0157] 2. Determining the Adjustment Stage: Based on the deviation value and preset segmented thresholds, the decompression process is divided into three stages: rapid adjustment, fine adjustment, and final adjustment. For example, the preset segmented thresholds are as follows: Rapid adjustment stage: deviation > 10 mmHg; Fine adjustment stage: deviation between 5 mmHg and 10 mmHg; Final adjustment stage: deviation ≤ 5 mmHg.

[0158] 3. Rapid Adjustment Phase: When the deviation exceeds 10 mmHg, the rapid adjustment phase begins. A preset combination of high frequency and high duty cycle is used to quickly reduce the pressure. For example, the preset frequency is 100 Hz and the duty cycle is 70%.

[0159] 4. Fine-tuning stage: When the deviation is between 5 mmHg and 10 mmHg, the fine-tuning stage begins. Precisely adjust the pressure using a preset combination of intermediate frequency and duty cycle. For example, the preset frequency is 50 Hz and the duty cycle is 50%.

[0160] 5. Final Adjustment Stage: When the deviation is less than or equal to 5 mmHg, the final adjustment stage begins. A preset combination of low frequency and low duty cycle is used for final adjustments to achieve the target pressure. For example, the preset frequency is 20 Hz and the duty cycle is 30%.

[0161] In step S333, pressure feedback is collected synchronously during the adjustment process. If the rate of change exceeds the preset stable standard or an activity interference signal is received, the preset parameter correction mechanism is triggered, including temporarily adjusting the duty cycle and extending the current stage. After stabilization, the original adjustment logic is restored.

[0162] Among them, the interference signal is caused by limb movement, which may affect the accuracy of pressure measurement. The parameter correction mechanism is a mechanism for temporarily adjusting PWM pulse parameters (such as duty cycle) when pressure changes are unstable or subject to interference. Duty cycle is the ratio of the high-level time to the total cycle time in a PWM pulse. The current stage is the current adjustment stage (rapid adjustment, fine adjustment, or final adjustment).

[0163] The necessary process is described below:

[0164] 1. Synchronous pressure feedback acquisition: During the adjustment process, the system acquires the current pressure data of the compression unit in real time. For example, pressure data is acquired once per second.

[0165] 2. Calculate the rate of change of pressure: The system calculates the rate of change of pressure over time to determine whether the pressure is stable. For example, it calculates the ratio of the pressure difference between two adjacent measurements to the time interval.

[0166] 3. Determine if the preset stability standard is exceeded: Compare the calculated rate of change with the preset stability standard. For example, the preset stability standard is a change of no more than 2 mmHg per second.

[0167] 4. Determine if interference signals are received: The system detects whether interference signals are received, such as limb movement signals from the accelerometer. For example, the accelerometer detects limb movement and emits interference signals.

[0168] 5. Triggering parameter correction mechanism: If the pressure change rate exceeds the preset stability standard or an activity disturbance signal is received, the parameter correction mechanism is triggered. For example, if the change rate is 3 mmHg / s, it exceeds the preset stability standard of 2 mmHg / s.

[0169] 6. Temporarily adjust the duty cycle: The system temporarily adjusts the duty cycle of the PWM pulse to slow down the rate of pressure change. For example, the duty cycle can be temporarily adjusted from 70% to 50%.

[0170] 7. Extend the current phase: The system extends the duration of the current adjustment phase until the pressure change stabilizes. For example, extend the duration of the current rapid adjustment phase until the rate of change drops below the preset stability standard.

[0171] 8. Restore original regulation logic: After the pressure change stabilizes, the system restores the original regulation logic and continues to reduce pressure according to the preset frequency and duty cycle combination. For example, when the rate of change drops to 1 mmHg / s, the parameters of the original rapid regulation phase (frequency 100Hz, duty cycle 70%) are restored.

[0172] Pressure feedback is collected synchronously during the adjustment process. If the rate of change exceeds the preset stability standard or an activity disturbance signal is received, the preset parameter correction mechanism is triggered, including:

[0173] Step S3331: Simultaneously collect multi-source data, including pressure feedback, pulse parameters, activity interference signals, hardware status, and environmental parameters, and generate a data sequence with timestamps according to a preset sampling frequency.

[0174] Among them, the preset sampling frequency is the frequency at which the system collects data. The timestamp is a specific time marker for data collection, used for data synchronization and analysis.

[0175] Step S3332: Extract and quantify features based on the collected data, including pressure change rate, interference intensity, hardware attenuation coefficient and environmental impact coefficient converted by a preset model, to form a comprehensive feature vector.

[0176] Among them, the hardware attenuation coefficient is a hardware performance attenuation coefficient calculated through a preset model, reflecting the impact of hardware status on pressure regulation. The environmental impact coefficient is the influence coefficient of environmental parameters on pressure measurement calculated through a preset model. The comprehensive feature vector combines the extracted and quantified features into a single vector for subsequent analysis and decision-making.

[0177] The necessary process is described below:

[0178] 1. Receiving Acquired Data: The system receives the multi-source data sequence with timestamps generated in step S3331. For example, the received data sequence is as follows: Timestamp 1: Pressure 95 mmHg, frequency 100 Hz, duty cycle 70%, interference signal strength 0.2, hardware status normal, ambient temperature 25°C. Timestamp 2: Pressure 94.5 mmHg, frequency 100 Hz, duty cycle 70%, interference signal strength 0.1, hardware status normal, ambient temperature 25°C. Timestamp 3: Pressure 94 mmHg, frequency 100 Hz, duty cycle 70%, interference signal strength 0.1, hardware status normal, ambient temperature 25°C.

[0179] 2. Calculate the rate of pressure change: The system calculates the rate of pressure change over time to assess pressure stability. For example, it calculates the ratio of the pressure difference between adjacent time stamps to the time interval.

[0180] .

[0181] 3. Quantify Interference Intensity: The system quantifies the intensity of the interference signal, reflecting the impact of limb activity on pressure measurement. For example, the interference intensity is directly taken as the output value of the accelerometer: Interference Intensity = 0.2 (unit: m / s²).

[0182] 4. Calculate the hardware attenuation coefficient: The system calculates the hardware performance attenuation coefficient through a preset model, reflecting the impact of hardware status on pressure regulation.

[0183] For example, the preset model calculates the attenuation coefficient based on the hardware status (such as the working status of the solenoid valve, the speed of the air pump, etc.): Hardware attenuation coefficient = f(hardware status) = 0.95 (assuming the attenuation coefficient under normal conditions is 0.95).

[0184] 5. Calculate the environmental impact coefficient: The system calculates the impact coefficient of environmental parameters on pressure measurement using a preset model. For example, the preset model calculates the impact coefficient based on ambient temperature: Environmental impact coefficient = g(ambient temperature) = 1.02 (assuming the impact coefficient is 1.02 at 25°C).

[0185] 6. Forming a comprehensive feature vector: The system combines the extracted and quantified features into a comprehensive feature vector for subsequent analysis and decision-making.

[0186] For example, the comprehensive feature vector is: F = [pressure change rate, interference intensity, hardware attenuation coefficient, environmental impact coefficient]. F = [50mmHg / s, 0.2m / s², 0.95, 1.02].

[0187] Step S3333: Compare the comprehensive feature vector with the multidimensional threshold matrix and match the correction strategy according to the trigger type.

[0188] The multi-dimensional threshold matrix is ​​a matrix containing threshold values ​​across multiple dimensions, used to determine whether each feature in the comprehensive feature vector exceeds the normal range. The trigger type is the correction strategy type determined based on the comparison between the comprehensive feature vector and the multi-dimensional threshold matrix, such as "pressure change too rapid" or "interference intensity too high." The correction strategy is a preset parameter adjustment scheme for different trigger types, used to stabilize the pressure regulation process.

[0189] Matching Correction Strategy: Based on the trigger type, the system matches the corresponding correction strategy from the preset correction strategy library. For example, for trigger types such as "pressure change too fast" and "interference intensity too high", the matched correction strategies may include: temporarily reducing the duty cycle; increasing the sampling frequency; and extending the duration of the current adjustment phase.

[0190] Step S3334: Based on the matching strategy and the characteristics of the current adjustment stage, generate the duty cycle adjustment range, stage duration extension value and frequency fine-tuning coefficient, and ensure that the hardware safety range is not exceeded through a preset constraint algorithm.

[0191] Among them, the stage duration extension value is the extension duration of the current adjustment stage generated according to the matching strategy. The frequency fine-tuning coefficient is the frequency adjustment coefficient generated according to the matching strategy. The preset constraint algorithm is an algorithm used to ensure that the adjusted parameters do not exceed the hardware safety range.

[0192] The necessary process is described below:

[0193] 1. Receiving Matching Strategy: The system receives the matching correction strategy in step S3333. For example, if the matching strategy is "pressure change is too rapid" and "interference intensity is too high", the corresponding correction measures include temporarily reducing the duty cycle, increasing the sampling frequency, and extending the time of the current adjustment phase.

[0194] 2. Determine the characteristics of the current adjustment phase: The system determines the current adjustment phase and its characteristics. For example, it is currently in the fast adjustment phase, with a preset frequency of 100Hz and a duty cycle of 70%.

[0195] 3. Generate Adjustment Parameters: Based on the matching strategy and the characteristics of the current adjustment phase, generate the duty cycle adjustment range, phase duration extension value, and frequency fine-tuning coefficient. For example: Duty cycle adjustment range: adjust from 70% to 50%; Phase duration extension value: extend by 10 seconds; Frequency fine-tuning coefficient: adjust from 100Hz to 80Hz.

[0196] 4. Application of preset constraint algorithms: The system uses preset constraint algorithms to ensure that the adjusted parameters do not exceed the hardware safety range. For example, the preset constraint algorithm checks whether the adjusted duty cycle is within the hardware's allowed range (e.g., 10% to 90%) and whether the frequency is within the safe range (e.g., 20Hz to 200Hz).

[0197] 5. Output Adjustment Parameters: The system outputs the adjusted duty cycle, stage duration, and frequency for subsequent decompression operations. For example, the output adjustment parameters are: adjusted duty cycle: 50%; adjusted stage duration: extended by 10 seconds; adjusted frequency: 80Hz.

[0198] Step S3335: Perform pressure reduction operation according to the adjusted parameters, simultaneously increase the sampling frequency to the preset high-frequency mode, calculate the deviation between the pressure trajectory and the theoretical curve in real time, and continuously evaluate the correction effect.

[0199] The preset high-frequency mode is used for more accurate monitoring of pressure changes. The pressure trajectory is the curve showing the actual pressure changing over time. The theoretical curve is the curve showing the target pressure changing over time, calculated based on the personalized decompression curve. The deviation value is the difference between the actual pressure trajectory and the theoretical curve. The correction effect is the degree of closeness between the pressure trajectory and the theoretical curve after parameter adjustments.

[0200] The necessary process is described below:

[0201] 1. Perform decompression operation: The system performs a decompression operation based on the adjustment parameters generated in step S3334. For example, the adjusted parameters are: duty cycle: 50%; phase duration: 30 seconds; frequency: 80Hz;

[0202] 2. Increase sampling frequency to preset high-frequency mode: The system increases the sampling frequency to a preset high-frequency mode to more accurately monitor pressure changes. For example, the preset high-frequency mode is 200Hz.

[0203] 3. Real-time calculation of the deviation between the pressure trajectory and the theoretical curve: The system collects pressure data in real time and generates the actual pressure trajectory. The system calculates the theoretical curve based on the personalized decompression curve. The system calculates the deviation between the actual pressure trajectory and the theoretical curve. For example, the theoretical curve is: Time 0 seconds: 100 mmHg; Time 10 seconds: 90 mmHg; Time 20 seconds: 85 mmHg; Time 30 seconds: 80 mmHg; the actual pressure trajectory is: Time 0 seconds: 100 mmHg; Time 10 seconds: 92 mmHg; Time 20 seconds: 87 mmHg; Time 30 seconds: 82 mmHg; the deviation is: Time 10 seconds: 2 mmHg; Time 20 seconds: 2 mmHg; Time 30 seconds: 2 mmHg.

[0204] 4. Continuous Evaluation of Correction Effectiveness: The system continuously evaluates the correction effectiveness, determining whether the deviation value is within the preset accuracy range. For example, the preset accuracy threshold is 2 mmHg. If the deviation value remains within 2 mmHg, the correction effect is considered good; if the deviation value exceeds 2 mmHg, the system will re-enter the correction mechanism.

[0205] Step S3336: When the deviation value is less than or equal to the preset accuracy threshold and the external interference factors are normal, the preset smooth transition algorithm is started to gradually restore the original adjustment parameters in an exponential decay manner until the correction mechanism ends after connecting to the original logic.

[0206] Among them, the smooth transition algorithm is used to smoothly restore the original adjustment parameter after the correction mechanism ends. The exponential decay method is a mathematical method used to gradually reduce the adjustment amplitude, ensuring a smooth parameter transition. The original adjustment parameter is the initial adjustment parameter before the correction mechanism begins.

[0207] The necessary process is described below:

[0208] 1. Determine Deviation Value and External Interference Factors: The system determines whether the current deviation value is less than or equal to the preset accuracy threshold. The system checks whether external interference factors have returned to normal. For example, if the preset accuracy threshold is 2 mmHg, the current deviation value is 1.5 mmHg, and external interference factors (such as limb movement) have returned to normal.

[0209] 2. Activate the smooth transition algorithm: When both the deviation value and external disturbance factors meet the conditions, the system activates the preset smooth transition algorithm. For example, the system activates the smooth transition algorithm and begins to gradually restore the original adjustment parameters.

[0210] 3. Adjust parameters using an exponential decay method: The system gradually adjusts parameters such as duty cycle and frequency using an exponential decay method to ensure a smooth transition. For example, the exponential decay formula is:

[0211] ;in, It is the attenuation coefficient. It's time.

[0212] 4. Restore original adjustment parameters: The system gradually restores to the original adjustment parameters until it is fully connected to the original logic. For example, the original adjustment parameters are: duty cycle: 70%; frequency: 100Hz; the system gradually adjusts the parameters until it is restored to the above original adjustment parameters.

[0213] 5. End correction mechanism: When the parameters are completely restored to the original adjustment parameters, the system ends the correction mechanism and continues to perform the pressure reduction operation according to the original logic.

[0214] In the step of synchronously acquiring multi-source data, a sub-step for signal acquisition and preprocessing using a miniature pulse sensor is added, specifically including:

[0215] Step SA00: Synchronously acquire the real-time pulse signal of the miniature pulse sensor. The miniature pulse sensor is based on the preset optimal attachment area of ​​the puncture site, and the attachment status is confirmed by the preset position recognition module.

[0216] The components include: a miniaturized pulse sensor for real-time monitoring of pulse signals at the puncture site; a real-time pulse signal reflecting blood flow at the puncture site; an optimal attachment area predetermined based on the anatomical structure and physiological characteristics of the puncture site; a position recognition module to confirm whether the miniature pulse sensor is correctly attached to the predetermined optimal area; and an attachment status indicating whether the sensor is correctly attached to the predetermined position.

[0217] The necessary process is described below:

[0218] 1. Activate the miniature pulse sensor: The system activates the miniature pulse sensor to prepare for collecting pulse signals at the puncture site. For example, the miniature pulse sensor is installed near the puncture site to monitor blood flow.

[0219] 2. Determine the optimal attachment area: Based on the anatomical structure and physiological characteristics of the puncture site, preset the optimal attachment area for the miniature pulse sensor. For example, for femoral artery puncture, the optimal attachment area may be 2 cm above the puncture point.

[0220] 3. Confirm Attachment Status: The system uses a preset position recognition module to confirm whether the miniature pulse sensor is correctly attached to the preset optimal area. For example, the position recognition module uses optical or electromagnetic induction technology to confirm the sensor's attachment position. If the sensor is not correctly attached, the system issues an alarm and prompts medical staff to adjust the sensor's position.

[0221] 4. Synchronous pulse signal acquisition: The system synchronously acquires real-time pulse signals from the miniature pulse sensor to ensure data real-time performance and accuracy. For example, the miniature pulse sensor acquires pulse signals 100 times per second, generating a high-frequency pulse data stream.

[0222] Step SB00 involves preprocessing the pulse signal, including position offset compensation and environmental interference filtering.

[0223] The process includes: Position offset compensation: correcting signal deviations caused by sensor position offsets. Environmental interference filtering: filtering out signal interference caused by external environmental factors (such as electromagnetic interference, noise, etc.). Preprocessing: preliminary processing of the raw signal to improve signal quality and the accuracy of subsequent analysis.

[0224] Position offset compensation: The system detects whether the sensor has shifted position and compensates for the signal deviation caused by the shift. For example, if the position recognition module detects a sensor shift of 0.5 cm, the system adjusts the signal strength according to a preset compensation model to correct the impact of the shift. The compensation model can be a simple linear model or a complex nonlinear model, depending on the characteristics of the sensor and the degree of impact of the shift.

[0225] Environmental interference filtering: The system filters the pulse signal to remove the influence of external environmental factors (such as electromagnetic interference, noise, etc.). For example, a low-pass filter is used to remove high-frequency noise, or a band-pass filter is used to retain signals within a specific frequency range. The filter design can be optimized based on the characteristics of the pulse signal and the frequency range of the main interference sources.

[0226] Step SC00: Extract and quantize features from the preprocessed pulse signal.

[0227] The process includes: Preprocessed pulse signal: The pulse signal after position offset compensation and environmental interference filtering. Feature extraction: Extracting key information that characterizes the signal properties from the signal. Quantized features: Converting the extracted features into numerical form for easier subsequent processing and analysis. Feature vector: Combining multiple quantized features into a single vector for subsequent analysis and decision-making.

[0228] Key features are extracted from the preprocessed pulse signal. Common features include: pulse rate: the number of pulses per unit time; pulse intensity: the amplitude of the pulse signal; pulse waveform: the shape characteristics of the pulse signal, such as rise time, fall time, and peak value; and pulse variability: the degree of variation in the pulse signal, reflecting the stability of blood flow.

[0229] Quantitative Features: The extracted features are converted into numerical form for easier subsequent processing and analysis. For example: pulse rate: 70 beats / minute; pulse intensity: 1.2 units; pulse waveform characteristics: rise time 0.5 seconds, fall time 0.3 seconds, peak value 1.5 units; pulse variability: 0.1 units.

[0230] Step SD00 involves fusing the pulse signal features with the original quantized features to form an extended comprehensive feature vector.

[0231] The extended comprehensive feature vector includes:

[0232] Step SD10: Based on the results of environmental interference filtering and the position attachment status, a preset reliable quantitative analysis method is used to determine the reliable quantitative value of the pulse signal.

[0233] Among them, the environmental interference filtering result is as follows: the pulse signal after filtering removes the influence of external environmental factors (such as electromagnetic interference, noise, etc.). The position attachment status refers to whether the miniature pulse sensor is correctly attached to the preset optimal area. The reliability quantification analysis method is a preset method used to evaluate the reliability and accuracy of the pulse signal. The reliability quantification value is a numerical value representing the reliability of the pulse signal, typically between 0 and 1, where 1 represents complete reliability and 0 represents complete unreliability.

[0234] The necessary process is described below:

[0235] 1. Receiving the Environmental Interference Filtering Result: The system receives the pulse signal after environmental interference filtering generated in step SB00. For example, the filtered pulse signal data stream is as follows: Timestamp 1: Pulse signal strength 1.1; Timestamp 2: Pulse signal strength 1.2; Timestamp 3: Pulse signal strength 1.1;

[0236] 2. Confirm Position Attachment Status: The system confirms the position attachment status of the miniature pulse sensor. For example, the position recognition module confirms that the sensor is correctly attached to the preset optimal area.

[0237] 3. Select a credibility quantification analysis method: The system selects a preset credibility quantification analysis method, which can evaluate credibility based on the stability and consistency of the signal. For example, the credibility quantification analysis method can be a comprehensive evaluation model based on signal variability and noise level.

[0238] Assuming a reliable metric The calculation formula is:

[0239] ;

[0240] in, It represents the signal variability, ranging from [0,1], where 0 indicates no variation and 1 indicates maximum variation. It represents the noise level, ranging from [0,1], where 0 indicates no noise and 1 indicates maximum noise. It indicates the attachment status, ranging from [0,1], where 1 indicates correct attachment and 0 indicates incorrect attachment. and These are weighting coefficients used to adjust the contribution of each factor, satisfying... .

[0241] 4. Calculate the quantified reliability value: The system calculates the quantified reliability value of the pulse signal based on the environmental interference filtering results and the position attachment status. For example, the calculation method is as follows: Signal variability: 0.1 unit; Noise level: 0.05 unit; Position attachment status: Correctly attached; Quantified reliability value: Based on the above parameters, the calculated quantified reliability value is 0.9 (indicating high reliability).

[0242] Step SD20: Assign dynamic weights to pulse features based on the credible quantification value.

[0243] Among them, dynamic weights are weights that are dynamically adjusted based on the reliability of the pulse signal, and are used to reflect the reliability of features during feature fusion.

[0244] The system uses a preset dynamic weight calculation method, which typically adjusts the weights based on the credibility quantification value. For example, the dynamic weight calculation formula is: Dynamic Weight = Credibility quantification value × Base Weight; where the base weight is a preset weight value used when the credibility is 1.

[0245] Calculating dynamic weights: The system calculates the dynamic weight of each pulse feature based on the confidence quantification value and the base weight. For example, if the base weight is 0.5 and the confidence quantification value is 0.945, then the dynamic weight is: Dynamic weight = 0.945 × 0.5 = 0.4725.

[0246] Assigning dynamic weights: The system assigns the calculated dynamic weights to each pulse feature.

[0247] Step SD30: Calculate the temporal correlation coefficient between the original quantitative features and the pulse features.

[0248] Among them, the original quantized features are the features in the comprehensive feature vector generated in step S3332, such as pressure change rate, interference intensity, hardware attenuation coefficient, environmental influence coefficient, etc. The pulse features are the features extracted and quantized from the preprocessed pulse signal, such as pulse rate, pulse intensity, pulse waveform features, and pulse variability.

[0249] Time series correlation coefficient: A statistical indicator that measures the correlation between two time series, usually the Pearson correlation coefficient.

[0250] The formula for the Pearson correlation coefficient is: ;

[0251] in, and These represent the values ​​of the two time series, and These are their average values.

[0252] The system calculates the time-series correlation coefficient between the original quantitative characteristics and pulse characteristics. For example, it calculates the correlation coefficient between the rate of change of pressure and the pulse rate: rate of change of pressure: 50 mmHg / s; pulse rate: 70 beats / minute; assuming the time series data are as follows: time series of rate of change of pressure: 50, 52, 51, 53, 50; time series of pulse rate: 70, 72, 71, 73, 70.

[0253] Calculate the correlation coefficient:

[0254] ;

[0255] ;

[0256] Based on the above parameters, the Pearson correlation coefficient formula can be used to calculate... .

[0257] Step SD40: The dynamically weighted pulse features are integrated with the original features after weight adjustment using a preset fusion algorithm to generate an extended comprehensive feature vector.

[0258] Select a preset fusion algorithm: The system selects a preset fusion algorithm, such as weighted summation and principal component analysis (PCA). For example, weighted summation can be selected as the fusion algorithm.

[0259] Integrated features: The system integrates the dynamically weighted pulse features with the original features after weight adjustment to generate an extended comprehensive feature vector.

[0260] For example, using the weighted summation method: .

[0261] Step SD50: Verify the validity of the extended comprehensive feature vector and calculate the coherence of each feature in the vector.

[0262] Validity verification: The generated extended comprehensive feature vector is validated to ensure it meets preset standards and requirements. Coordination: A metric that measures the consistency and coordination among features in the feature vector, typically used to evaluate the validity of the feature vector.

[0263] Selecting a Collaboration Calculation Method: The system selects a preset collaboration calculation method, such as Pearson correlation coefficient or cosine similarity. For example, cosine similarity can be selected as the collaboration calculation method.

[0264] Computational Synergy: The system calculates the synergy between features in the expanded composite feature vector. For example, it calculates the cosine similarity between each pair of features in the feature vector.

[0265] Generate the synergy matrix:

[0266] The system generates a synergy matrix to record the synergy between features in the feature vector.

[0267] For example, the degree of synergy matrix is:

[0268] .

[0269] Evaluation of the synergy matrix: The system evaluates the synergy matrix to ensure that the synergy between each feature meets a preset threshold. For example, if the preset synergy threshold is 0.8, the system checks whether all off-diagonal elements in the synergy matrix are greater than or equal to 0.8.

[0270] Step SD60: If the degree of cooperation is greater than or equal to the preset verification threshold, continue with the subsequent steps.

[0271] Preset verification threshold: A preset threshold used to determine whether the degree of collaboration has reached an acceptable level. Subsequent steps: If the degree of collaboration reaches or exceeds the preset verification threshold, the system will continue to execute subsequent processing steps, such as feature fusion and decision-making.

[0272] In step SD70, if the coherence is less than the verification threshold, proceed to the step of preprocessing the pulse signal, adjust the key preprocessing parameters, and then repeat the pulse signal preprocessing and feature extraction quantization steps. Then, regenerate the extended comprehensive feature vector and verify it until the coherence meets the standard.

[0273] The necessary process is described below:

[0274] 1. Receive the coordination degree evaluation result: The system receives the coordination degree evaluation result from step SD60. For example, if some off-diagonal elements in the coordination degree matrix are less than the preset verification threshold of 0.8, the coordination degree does not meet the condition.

[0275] 2. Adjust key preprocessing parameters: Based on the synergy evaluation results, the system adjusts key parameters in the preprocessing steps to improve the quality of the pulse signal. For example, the filter cutoff frequency is increased from 10Hz to 15Hz to better remove high-frequency noise.

[0276] 3. Reprocess the pulse signal: The system reprocesses the pulse signal, including position offset compensation and environmental interference filtering. For example, the adjusted filter parameters are reapplied to filter the pulse signal.

[0277] 4. Re-extract and quantize features: The system re-extracts and quantizes features from the preprocessed pulse signal. For example, it recalculates pulse rate, pulse intensity, pulse waveform characteristics, and pulse variability.

[0278] 5. Regenerate the extended comprehensive feature vector: The system regenerates the extended comprehensive feature vector and performs validity verification. For example, it recalculates the dynamically weighted pulse features and the original features after adjusting the weights to generate a new extended comprehensive feature vector.

[0279] 6. Repeatedly verify the synergy: The system repeats steps SD50 and SD60 to evaluate the synergy of the new extended comprehensive feature vector. If the synergy still does not meet the verification threshold, the system continues to adjust the preprocessing parameters and repeat the above steps until the synergy meets the standard.

[0280] By integrating through a preset fusion algorithm, an extended comprehensive feature vector is generated, including:

[0281] Step SD41: Simultaneously acquire iPPG non-contact pulse signals and three-axis gyroscope attitude data, and align them with contact pulse features and existing features according to timestamps.

[0282] Among them, the iPPG non-contact pulse signal is a pulse signal acquired through non-contact photoplethysmography (iPPG), reflecting the blood flow at the puncture site. The three-axis gyroscope attitude data is limb attitude data acquired through a three-axis gyroscope, including rotation angle and angular velocity, used to monitor limb movement. Contact pulse characteristics are pulse characteristics acquired through contact sensors (such as miniature pulse sensors), such as pulse rate and pulse intensity. Existing characteristics are features in the comprehensive feature vector generated in step S3332, such as pressure change rate, interference intensity, hardware attenuation coefficient, and environmental influence coefficient.

[0283] Step SD42: Wavelet transform is used to remove motion artifacts from the iPPG signal, and waveform amplitude and main peak interval are extracted as quantization features; Kalman filtering is used to smooth the gyroscope data, and the peak value of rotation angle and the mean value of angular velocity are extracted as attitude features.

[0284] Wavelet transform is a signal processing technique used to remove noise and artifacts from a signal while preserving its main features. Motion artifacts are signal interference caused by limb movement, which can affect the accuracy of pulse signals. Waveform amplitude reflects the intensity of the pulse signal.

[0285] Interpeak Duration: The time interval between adjacent peaks in the pulse signal, reflecting the pulse frequency. Three-Axis Gyroscope Data: Limb posture data acquired through a three-axis gyroscope, including rotation angle and angular velocity. Kalman Filter: A highly efficient self-recursive filter used to estimate the dynamic state of a system from a series of noisy measurements. Peak Rotation Angle: The maximum value of the rotation angle in the gyroscope data, reflecting the maximum rotation angle of the limb. Mean Angular Velocity: The average value of the angular velocity in the gyroscope data, reflecting the average rotational speed of the limb.

[0286] Step SD43: Based on the physiological homology between contact pulse features and iPPG features, calculate the waveform similarity between the two and dynamically assign weights according to the similarity.

[0287] Physiological homology: refers to the similarity of two pulse signals in terms of physiological mechanism, that is, they reflect the same physiological phenomenon.

[0288] Waveform similarity: an index that measures the degree of similarity between two pulse signal waveforms, usually calculated using methods such as correlation coefficient or Euclidean distance.

[0289] Selecting a waveform similarity calculation method: The system selects a preset waveform similarity calculation method, such as Pearson correlation coefficient or Euclidean distance. For example, Pearson correlation coefficient can be selected as the calculation method.

[0290] Waveform similarity calculation: The system calculates the waveform similarity between contact pulse characteristics and iPPG characteristics. For example, it calculates the correlation coefficient between contact pulse intensity and iPPG waveform amplitude.

[0291] Dynamic weight allocation: The system dynamically allocates weights based on waveform similarity. The weight allocation formula can be: Weight = Similarity × Base weight.

[0292] Generate a weighted feature vector: The system dynamically assigns weights to the contact pulse feature and the iPPG feature to generate a weighted feature vector.

[0293] Step SD44 introduces an individualized arterial wall stress-strain model. Based on CT images of the puncture site and individual patient characteristics, the elastic modulus of the blood vessel is inferred from the pulse wave conduction velocity. The influence of vascular displacement on pressure distribution is corrected by combining gyroscope attitude data, generating quantitative features of vascular closure, which are then incorporated into the feature set.

[0294] The components include: Individualized arterial wall stress-strain model: An arterial wall mechanics model established based on individual patient characteristics (such as age, gender, blood pressure, etc.) to simulate arterial wall stress and strain. Puncture site CT images: Image data of the puncture site obtained through CT scans, used to determine the geometry and location of the blood vessel. Pulse wave velocity: The speed at which the pulse wave propagates in the blood vessel, related to the vascular elastic modulus. Vascular elastic modulus: A parameter reflecting the elasticity of the blood vessel wall, used to assess the health of the blood vessel. Gyroscope posture data: Limb posture data acquired through a three-axis gyroscope, including rotation angle and angular velocity. Quantitative characteristics of vascular closure: A quantitative indicator reflecting the degree of vascular closure, used to assess the compression effect on the blood vessel.

[0295] The necessary process is described below:

[0296] 1. Receiving CT images of the puncture site and individual patient characteristics: The system receives CT image data of the puncture site and individual patient characteristics (such as age, gender, blood pressure, etc.). For example, the CT image shows that the diameter of the blood vessel at the puncture site is 10mm, and the individual patient characteristics include age 60 years, gender male, and blood pressure 120 / 80mmHg.

[0297] 2. Establishing an individualized arterial wall stress-strain model: The system establishes an individualized arterial wall stress-strain model based on the patient's individual characteristics and CT image data. For example, the model considers parameters such as the geometric structure of the blood vessel, wall thickness, and elastic modulus.

[0298] 3. Calculation of pulse wave velocity: The system infers the vascular elastic modulus from the pulse wave velocity. For example, if the pulse wave velocity is 10 m / s, it can be calculated using the formula... Calculate the elastic modulus of blood vessels, where, It's blood density. It is the pulse wave conduction velocity.

[0299] Assuming blood density =1060kg / m³, pulse wave propagation velocity =10m / s, then the elastic modulus of blood vessels for: =1060×102=106000Pa.

[0300] 4. Correcting the impact of blood vessel displacement on pressure distribution: The system combines gyroscope attitude data to correct the impact of blood vessel displacement on pressure distribution. For example, if gyroscope data shows that the limb rotation angle is 10° and the angular velocity is 0.5 rad / s, the system corrects the impact of blood vessel displacement on pressure distribution based on this data.

[0301] 5. Generate quantitative features of vascular closure: The system generates quantitative features of vascular closure based on the corrected pressure distribution. For example, a quantitative feature of vascular closure of 0.8 indicates a high degree of vascular closure.

[0302] 6. Incorporation into Feature Set: The system incorporates the quantitative features of blood vessel closure into the feature set for subsequent feature fusion and analysis. For example, the feature set is updated as follows:

[0303] .

[0304] Step SD45: The hierarchical attention fusion algorithm is used to integrate the weighted contact pulse features, iPPG features, posture features, original features and vascular closure features. The final weight matrix is ​​calculated through the interaction relationship between features. After weighted summation and dimensional normalization, an extended comprehensive feature vector is generated.

[0305] The algorithm includes: a hierarchical attention fusion algorithm, a multi-layer attention mechanism used to fuse features from different sources, dynamically adjusting weights by calculating the interaction relationships between features; weighted contact pulse features, which are contact pulse features after dynamic weight adjustment; iPPG features, which are non-contact pulse features after wavelet transform processing, including waveform amplitude and interpeak duration; and attitude features, which are three-axis gyroscope data after Kalman filtering, including peak rotation angle and mean angular velocity.

[0306] The necessary process is described below:

[0307] 1. Receive all features: The system receives the weighted contact pulse features, iPPG features, posture features, original features, and vascular closure features generated in step SD43.

[0308] 2. Initialize the hierarchical attention fusion algorithm: The system initializes the hierarchical attention fusion algorithm, setting initial weights and attention mechanism parameters. For example, initializing the weight matrix... and attention weight .

[0309] 3. Calculate the interaction relationships between features: The system calculates the interaction relationships between features through a hierarchical attention mechanism and dynamically adjusts the weights. For example, it calculates the similarity or correlation between features to generate an attention weight matrix. .

[0310] 4. Generate the final weight matrix: The system generates the final weight matrix based on the interaction relationships between features. .

[0311] For example, the final weight matrix as follows:

[0312] .

[0313] 5. Weighted summation: The system performs weighted summation on each feature to generate a fused feature vector.

[0314] For example, the fused feature vector as follows:

[0315] .

[0316] 6. Dimension Normalization: The system performs dimension normalization on the fused feature vectors to ensure that each feature is on the same scale.

[0317] For example, the normalized feature vector as follows:

[0318] .

[0319] 7. Generate extended comprehensive feature vectors:

[0320] The system generates extended comprehensive feature vectors for subsequent analysis and decision-making.

[0321] For example, the extended comprehensive feature vector as follows:

[0322] .

[0323] Based on the same inventive concept, embodiments of the present invention provide a control system for an arterial compression device, including a memory and a processor, wherein the memory stores a function that can run on the processor to implement the following... Figures 1 to 2 The procedure for any method.

[0324] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A control system for an arterial compression device, characterized in that, The device includes a processor and a memory storing a computer program. The processor is configured to execute the computer program to implement a control method for the arterial compression device, the specific steps of which are as follows: It receives the artery type and vascular elasticity parameters at the puncture site, automatically matches the baseline range of initial pressure, total compression time and termination pressure, and generates a personalized decompression curve with preset time intervals and corresponding decompression values ​​that are differentiated according to artery type after adjustment based on patient characteristics. Based on a personalized decompression curve, the preset pressure regulation module is controlled to inflate the preset compression unit to the initial pressure, and timing is synchronized. During the timing period, the current pressure and cumulative duration of the compression unit are collected in real time. When the preset interval is reached, the pressure is reduced according to the personalized decompression curve. If the current pressure deviates from the theoretical value of the curve by more than the preset value, the decompression amplitude is adjusted according to the deviation and the vascular elasticity parameters. When the cumulative duration reaches the set total pressure time, the pressure is controlled according to the set termination pressure and maintained for the preset duration. After the pressure sensor continuously monitors and confirms that the pressure fluctuation is less than the preset pressure fluctuation, the pressure is released and a prompt is given. When the preset interval is reached, pressure reduction is performed according to the personalized pressure reduction curve, including: Based on the correspondence between the current cumulative duration and the personalized stress reduction curve, the target stress reduction value is calculated by calling the pre-stored stress-time correlation model; Simultaneously, the pressure data collected in real time by the preset pressure sensor is processed by Kalman filtering, and then fused with the three-dimensional activity data collected by the preset acceleration sensor through an attention mechanism network. The pressure data weight is dynamically reduced to eliminate false pressure fluctuations caused by limb swinging. After confirming the effective pressure reduction requirement, a preset segmented PWM pulse algorithm is used to drive the preset pressure regulation module. The gas release rate is controlled by the built-in current limiting structure of the pressure regulation module to ensure a smooth pressure reduction process. The preset pressure regulation module, driven by a preset segmented PWM pulse algorithm, includes: The pulse parameters are initialized based on the puncture site and personalized decompression curve. Specifically, the preset base frequency and initial duty cycle are set according to the site, and the preset segmented thresholds are modified in combination with the individual characteristics of the patient. Based on the effective pressure signal, and according to the deviation between the current pressure and the target pressure reduction value, the system is divided into three stages: rapid adjustment, fine adjustment, and final adjustment, based on the preset segmented threshold. Each stage uses a preset combination of frequency and duty cycle. Pressure feedback is collected synchronously during the adjustment process. If the rate of change exceeds the preset stable standard or an activity interference signal is received, the preset parameter correction mechanism is triggered, including temporarily adjusting the duty cycle and extending the current stage. After stabilization, the original adjustment logic is restored.

2. The control system of the arterial compression device according to claim 1, characterized in that, The data is fused with 3D activity data collected by a pre-set accelerometer through an attention mechanism network. This process dynamically down-weights the pressure data to eliminate spurious pressure fluctuations caused by limb movement, including: The accelerometer is activated, the sampling frequency is adjusted according to the characteristics of the puncture site, and three-dimensional motion data is collected and preprocessed. Based on the preprocessed data, axial weights are assigned according to the location characteristics to filter irrelevant signals and extract key activity features; The filtered pressure data and activity features are input into the attention mechanism network. The network dynamically assigns weights according to the activity features, and the weights of the pressure data are downgraded in stages when the limbs swing. The attention mechanism network uses dynamic weights to fuse pressure data and activity characteristics to generate an effective pressure signal and calculate the quantitative index of its correlation with activity characteristics. When the fluctuation start time difference is less than or equal to the preset time difference threshold and the intensity change correlation coefficient is greater than or equal to the preset correlation coefficient threshold, it is determined to be a false fluctuation caused by limb swinging, and the current pressure is maintained for a preset duration appropriate to the puncture site. Otherwise, it is considered a genuine pressure decrease, triggering the preset pressure reduction operation logic.

3. The control system of the arterial compression device according to claim 1, characterized in that, The preset parameter correction mechanisms include: Simultaneously collect multi-source data, including pressure feedback, pulse parameters, activity interference signals, hardware status, and environmental parameters, and generate a time-stamped data sequence according to a preset sampling frequency; Based on the collected data, features are extracted and quantified, including pressure change rate, interference intensity, hardware attenuation coefficient and environmental impact coefficient converted by a preset model, to form a comprehensive feature vector; The comprehensive feature vector is compared with the multidimensional threshold matrix, and the correction strategy is matched according to the trigger type. Based on the matching strategy and the characteristics of the current adjustment phase, the duty cycle adjustment range, the phase duration extension value, and the frequency fine-tuning coefficient are generated, and the preset constraint algorithm is used to ensure that the hardware safety range is not exceeded. Perform pressure reduction operation according to the adjusted parameters, simultaneously increase the sampling frequency to the preset high-frequency mode, calculate the deviation between the pressure trajectory and the theoretical curve in real time, and continuously evaluate the correction effect; When the deviation value is less than or equal to the preset accuracy threshold and the external interference factors are normal, the preset smooth transition algorithm is activated to gradually restore the original adjustment parameters in an exponential decay manner until the correction mechanism ends after connecting to the original logic.

4. The control system of the arterial compression device according to claim 3, characterized in that, In the step of synchronously acquiring multi-source data, a sub-step for signal acquisition and preprocessing using a miniature pulse sensor is added, specifically including: The real-time pulse signal of the miniature pulse sensor is collected synchronously. The miniature pulse sensor is based on the preset optimal attachment area of ​​the puncture site, and the attachment status is confirmed by the preset position recognition module. The pulse signal is preprocessed, including position offset compensation and environmental interference filtering. Features are extracted and quantified from the preprocessed pulse signal; The pulse signal features are fused with the original quantized features to form an extended comprehensive feature vector.

5. The control system of the arterial compression device according to claim 4, characterized in that, The extended comprehensive feature vector includes: Based on the results of environmental interference filtering and the position attachment status, a preset reliable quantitative analysis method is used to determine the reliable quantitative value of the pulse signal. Dynamic weights are assigned to pulse features based on the credible quantification value; Calculate the temporal correlation coefficient between the original quantitative characteristics and the pulse characteristics; The dynamically weighted pulse features are combined with the original features after weight adjustment using a preset fusion algorithm to generate an extended comprehensive feature vector. The validity of the extended comprehensive feature vector is verified, and the coherence of each feature in the vector is calculated. If the degree of coordination is greater than or equal to the preset verification threshold, continue with the subsequent steps; If the degree of synergy is less than the verification threshold, proceed to the step of preprocessing the pulse signal, adjust the key parameters of preprocessing, repeat the pulse signal preprocessing and feature extraction quantization steps, regenerate the extended comprehensive feature vector and verify it, until the degree of synergy meets the standard.

6. The control system of the arterial compression device according to claim 5, characterized in that, The generation of extended comprehensive feature vectors includes: Simultaneously acquire iPPG non-contact pulse signals and three-axis gyroscope attitude data, and align them with contact pulse characteristics and existing characteristics according to timestamps; Wavelet transform was used to remove motion artifacts from the iPPG signal, and waveform amplitude and main peak interval were extracted as quantization features. The gyroscope data was smoothed by Kalman filtering, and the peak value of rotation angle and mean value of angular velocity were extracted as attitude features. Based on the physiological homology between contact pulse features and iPPG features, the waveform similarity between the two is calculated, and weights are dynamically assigned according to the similarity. An individualized arterial wall stress-strain model was introduced. Based on CT images of the puncture site and individual patient characteristics, the elastic modulus of the blood vessel was inferred by pulse wave propagation velocity. The influence of blood vessel displacement on pressure distribution was corrected by combining gyroscope attitude data, generating quantitative features of blood vessel closure, which were then incorporated into the feature set. A hierarchical attention fusion algorithm is used to integrate the weighted contact pulse features, iPPG features, posture features, original features and vascular closure features. The final weight matrix is ​​calculated through the interaction relationship between features. After weighted summation and dimensional normalization, an extended comprehensive feature vector is generated.

Images