Microfluidic anesthetic bullet control method based on pressure feedback

Abstract

This application provides a pressure-feedback-based microfluidic anesthetic bullet control method, which relates to the field of anesthetic bullet control. The method comprises: obtaining a two-layer structural model of the microfluidic anesthetic bullet, the two-layer structural model comprising an inner anesthetic storage cavity model and an outer elastically adjustable layer material model; during the anesthetic release process, obtaining the volume data of the released anesthetic droplets monitored in real time by a photoelectric sensor, and calculating the released dose based on a pre-established functional relationship between the droplet volume and the anesthetic dose; and predicting the anesthetic dose released by the anesthetic bullet under current pressure conditions based on the final, verified and optimized functional relationship model. The method utilizes online support vector regression as an incremental learning algorithm to dynamically update model parameters and adaptively adjust them based on the deviation between the baseline value and the actual dose. This method effectively improves the accuracy and controllability of anesthetic release, providing reliable technical support for clinical anesthesia administration.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a microfluidic anesthetic bullet control method based on pressure feedback. Background Art

[0002] In practical applications, microfluidic anesthetic cartridges face the technical challenge of precisely controlling anesthetic dosage. These cartridges release the anesthetic agent through a pressure feedback mechanism, but the functional relationship between pressure and the released anesthetic dose remains unclear, making precise dosage control difficult. Furthermore, the elastic modulus of a single material limits the adjustable range of pressure feedback intensity, further increasing the uncertainty of dosage control.

[0003] The structural design of the anesthetic cartridge also presents challenges. The anesthetic agent stored in the inner layer needs to be isolated from the outer layer material to prevent dose loss due to diffusion and leakage. Furthermore, the selection of the outer layer material must balance elastic modulus adjustment and biocompatibility, and changes in thickness and composition can affect the overall mechanical properties and stability of the anesthetic cartridge.

[0004] Establishing a pressure feedback control loop is another key technical challenge. The anesthetic delivery process is dynamic, requiring real-time monitoring of pressure changes and timely adjustment of material parameters to maintain the dosage output within the target range. This requires integrating pressure sensors, control algorithms, and actuators to create a real-time closed-loop control system. The system's response speed, stability, and robustness directly impact the accuracy of anesthetic dosage control.

[0005] In summary, precise control of the dosage of microfluidic anesthetic cartridges involves multiple technical areas, including materials, structure, and control. It requires systematic analysis of the pressure feedback mechanism, establishing a quantitative relationship between material parameters and dosage output, optimizing the structural design of the anesthetic cartridge, and constructing a real-time closed-loop control system to ultimately achieve precise regulation of the anesthesia process. This is of great significance for improving anesthetic efficacy and reducing medication risks, but many technical challenges remain to be overcome. Summary of the Invention

[0006] The present invention provides a microfluidic anesthetic bullet control method based on pressure feedback, which mainly includes:

[0007] Obtaining a double-layer structure model of a microfluidic anesthetic bomb, wherein the double-layer structure model includes an inner layer anesthetic storage cavity model and an outer layer elastic adjustable layer material model;

[0008] For the material model of the outer elastically adjustable layer, a finite element analysis method is used to simulate the deformation and stress distribution of the anesthetic cavity under a preset pressure when the thickness of the outer layer material is discretely sampled within a preset range, thereby obtaining a set of elastic modulus parameters of the outer layer material, wherein the parameter set includes elastic modulus values ​​at different thicknesses;

[0009] Obtaining a pressure value of the microfluidic anesthetic bomb monitored in real time by a pressure sensor, and determining whether the current pressure reaches the preset pressure threshold based on a preset pressure threshold. If so, controlling the valve connecting the inner anesthetic cavity and the outer adjustable layer to open, and squeezing the inner layer anesthetic under the action of pressure based on the elastic modulus parameter set of the outer layer material to release the anesthetic;

[0010] During the anesthetic release process, the volume data of the released anesthetic droplets monitored in real time by the photoelectric sensor is obtained, and the released dose is calculated according to the pre-established functional relationship between the droplet volume and the anesthetic dose;

[0011] Based on the pressure change data monitored by the pressure sensor, the elastic modulus parameters of the outer material, and the calculated anesthetic release dose data, a support vector regression algorithm was used to establish a functional relationship model between material parameters, pressure, and dose output. The kernel function and parameters were selected through cross-validation, and the model parameters were trained and optimized.

[0012] The performance of the functional relationship model is evaluated using a test data set, and the root mean square error and the coefficient of determination are calculated. If the root mean square error is greater than a preset error threshold or the coefficient of determination is less than a preset coefficient threshold, the support vector regression algorithm hyperparameters are adjusted and the model is retrained until a final model that meets the accuracy requirements is obtained;

[0013] Based on the final functional relationship model after verification and optimization, the anesthetic dose released by the anesthetic bullet with specific outer layer material parameters under the current pressure conditions is predicted. By adjusting the thickness of the outer layer material to control the elastic modulus of the anesthetic bullet, precise control of the released dose is achieved. The predicted dose result is used as the benchmark value for adaptive control;

[0014] Real-time data on pressure and released dose are continuously collected, and online support vector regression is used as an incremental learning algorithm to merge the newly collected data with the original training data set. The parameters of the functional relationship model are dynamically updated and optimized. The model parameters are gradually adjusted according to the deviation between the baseline value and the actual dose, thereby improving the model generalization ability and prediction accuracy, and realizing adaptive control of anesthetic dose.

[0015] As a preferred embodiment of the microfluidic anesthetic bomb control method based on pressure feedback of the present invention, the method of obtaining a double-layer structure model of the microfluidic anesthetic bomb, wherein the double-layer structure model includes an inner layer anesthetic storage cavity model and an outer layer elastic adjustable layer material model, includes:

[0016] Acquiring three-dimensional data of the inner layer anesthetic storage cavity model;

[0017] According to the three-dimensional data, the inner layer anesthetic storage cavity model is parametrically modeled using SolidWorks software to obtain a three-dimensional solid model of the inner layer anesthetic storage cavity;

[0018] Determining the three-dimensional size parameters of the material model of the outer elastic adjustable layer according to the three-dimensional solid model of the inner layer anesthetic storage cavity;

[0019] According to the three-dimensional size parameters, the material model of the outer elastic adjustable layer is parametrically modeled using SolidWorks software to obtain a three-dimensional solid model of the outer elastic adjustable layer material;

[0020] Assembling the three-dimensional solid model of the inner anesthetic storage cavity and the three-dimensional solid model of the outer elastic adjustable layer material to obtain a three-dimensional assembly model of the double-layer structure of the microfluidic anesthetic bomb;

[0021] Using ANSYS software to perform structural statics analysis on the three-dimensional assembly model to obtain stress distribution and deformation;

[0022] If the stress distribution and deformation conditions do not meet the preset conditions, the size parameters of the inner anesthetic storage cavity model and the outer elastic adjustable layer material model are adjusted according to the stress distribution and deformation conditions until the preset conditions are met.

[0023] As a preferred embodiment of the pressure feedback-based microfluidic anesthetic bomb control method of the present invention, a finite element analysis method is used for the material model of the outer elastic adjustable layer to simulate the deformation and stress distribution of the anesthetic cavity under a preset pressure when the thickness of the outer layer material is discretely sampled within a preset range, thereby obtaining a set of elastic modulus parameters of the outer layer material. The parameter set includes elastic modulus values ​​at different thicknesses, including:

[0024] Obtaining the physical properties of the material of the outer elastic adjustable layer, establishing a constitutive model of the material, and determining the parameters required for the material model;

[0025] Constructing a three-dimensional geometric model of the anesthetic cavity and meshing the geometric model using finite element analysis software;

[0026] For a preset thickness range, a series of thickness values ​​are obtained, outer layer material parameters at different thicknesses are assigned to the geometric model, a preset pressure is set as a load condition, and a finite element solution is performed;

[0027] Based on the finite element analysis results, the deformation and stress distribution data under different thicknesses and preset pressures are obtained;

[0028] The elastic modulus values ​​of the outer layer material at different thicknesses are calculated based on the deformation and stress distribution data to obtain an elastic modulus parameter set.

[0029] As a preferred embodiment of the pressure feedback-based microfluidic anesthetic bullet control method of the present invention, the method includes: obtaining a pressure value of the microfluidic anesthetic bullet monitored in real time by a pressure sensor, judging whether the current pressure reaches the preset pressure threshold according to a preset pressure threshold; if so, controlling the valve connecting the inner anesthetic cavity and the outer adjustable layer to open, and squeezing the inner layer anesthetic to release the anesthetic under the action of pressure according to the elastic modulus parameter set of the outer layer material, including:

[0030] Obtaining a force value of the anesthetic bullet monitored in real time by the pressure sensor, comparing the value with a preset threshold, and determining whether the current pressure reaches the preset threshold;

[0031] If the current pressure reaches a preset threshold, the inner and outer layer connecting valves are controlled to open;

[0032] According to the anesthetic release rate predicted by the support vector regression model, a PID control algorithm is used to adjust the PWM duty cycle of the inner and outer layer connecting valves in real time to control the valve opening;

[0033] The parameters of the PID controller are optimized using a genetic algorithm so that the valve control strategy can be adaptively adjusted according to the pressure change trend.

[0034] As a preferred embodiment of the pressure feedback-based microfluidic anesthetic bomb control method of the present invention, the method includes: obtaining the volume data of the released anesthetic droplets monitored in real time by the photoelectric sensor during the anesthetic release process, and calculating the released dose based on a pre-established functional relationship between the droplet volume and the anesthetic dose, including:

[0035] Obtain the values ​​of the proportional term, integral term, and differential term, perform weighted summation on the three values, and obtain the control quantity;

[0036] adjusting the operating parameters of the anesthetic release device according to the control amount to change the anesthetic release rate;

[0037] A support vector machine algorithm is used to update the preset functional relationship model between the droplet volume and the anesthetic dose in real time. If the functional relationship model is updated, the dose calculation accuracy is improved.

[0038] As a preferred embodiment of the pressure feedback-based microfluidic anesthetic bomb control method of the present invention, the method comprises: monitoring pressure change data of the pressure sensor, elastic modulus parameters of the outer material, and calculating the anesthetic release dose data; using a support vector regression algorithm to establish a functional relationship model between material parameters, pressure, and dose output; selecting kernel functions and parameters through cross-validation; and training and optimizing model parameters, including:

[0039] Acquire pressure change data monitored in real time by a pressure sensor, perform denoising and smoothing on the pressure change data, and extract at least one statistical feature of the pressure change as a model input;

[0040] Obtaining elastic modulus parameters of the outer layer material;

[0041] Calculating a target anesthetic release dose under a current state based on a preset pharmacokinetic mathematical model, in combination with the pressure change statistical characteristics and the elastic modulus parameter;

[0042] The pressure change statistical characteristics, the elastic modulus parameters and the target anesthetic release dose are divided into a training set, a validation set and a test set, and a support vector regression model is constructed.

[0043] As a preferred embodiment of the pressure feedback-based microfluidic anesthetic bullet control method of the present invention, the method includes: using a test data set to evaluate the performance of the functional relationship model, calculating the root mean square error and the coefficient of determination; if the root mean square error is greater than a preset error threshold or the coefficient of determination is less than a preset coefficient threshold, adjusting the support vector regression algorithm hyperparameters and retraining the model until a final model that meets the accuracy requirements is obtained, including:

[0044] Obtain a test data set and use the root mean square error (RMSE) and the coefficient of determination (R²) as evaluation indicators to evaluate the performance of the support vector regression model;

[0045] If the RMSE is greater than the preset error threshold and the R² is less than the preset coefficient threshold, then according to the preset grid search rules, traverse the penalty coefficient C, kernel function type and kernel function parameter γ hyperparameter combination;

[0046] For each hyperparameter combination, a cross-validation method is used to evaluate and obtain the hyperparameter combination with the smallest RMSE on the validation set as the optimal hyperparameter;

[0047] From the model set, the model with the smallest RMSE and the largest R² is selected and determined as the final functional relationship model.

[0048] As a preferred embodiment of the pressure feedback-based microfluidic anesthetic bullet control method of the present invention, the method includes: predicting the anesthetic dose released by the anesthetic bullet with specific outer layer material parameters under current pressure conditions based on the verified and optimized final functional relationship model; controlling the elastic modulus of the anesthetic bullet by adjusting the outer layer material thickness to achieve precise control of the released dose; and using the predicted dose result as a reference value for adaptive control, including:

[0049] According to the outer material parameters of the anesthetic bullet and the current pressure conditions, the optimized functional relationship model is used to calculate the predicted anesthetic release dose;

[0050] By establishing a mapping relationship between material thickness and elastic modulus, the influence of adjusting material thickness on elastic modulus is determined;

[0051] According to the predicted release dose, the elastic modulus of the material required to achieve the target release dose is calculated using the mapping relationship;

[0052] The predicted release dose is used as the reference value for adaptive control, and a feedback control system based on the PID control algorithm is established.

[0053] The release rate and amount of the anesthetic bullet are monitored in real time using a sensor, the monitored value is compared with the reference value, and the deviation value is calculated;

[0054] According to the deviation value, the PID control algorithm is used to calculate the adjustment amount of the material thickness, and the material thickness is dynamically adjusted through the actuator to ensure that the release dose continues to meet the requirements of precise control.

[0055] As a preferred embodiment of the pressure feedback-based microfluidic anesthetic bomb control method of the present invention, the method continuously collects real-time data of pressure and released dose, uses online support vector regression as an incremental learning algorithm, merges the newly collected data with the original training data set, dynamically updates and optimizes the parameters of the functional relationship model, and gradually adjusts the model parameters according to the deviation between the baseline value and the actual dose, thereby improving the model generalization ability and prediction accuracy, and realizing adaptive control of the anesthetic dose, including:

[0056] Acquire data collected in real time by a pressure sensor and an anesthesia pump, and associate the pressure data with the dosage data and store them in a time series database;

[0057] Using the gradient boosting decision tree algorithm, the support vector regression model is retrained using the merged data set to dynamically update the parameters of the model;

[0058] The optimized support vector regression model is applied to the adaptive control of the anesthesia pump. According to the patient's real-time physiological parameters and surgical requirements, the output dose of the anesthesia pump is dynamically adjusted through the PID control algorithm to achieve precise anesthesia management.

[0059] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0060] The present invention discloses a method for adaptive control of the dose of a microfluidic anesthetic bomb. In order to solve the problem of accurate release of anesthetics, a double-layer structure design is adopted, including an inner cavity for storing anesthetics and an outer elastic adjustable layer. The deformation and stress distribution of the outer material under different thicknesses and pressures are simulated by the finite element analysis method to obtain a set of elastic modulus parameters. Combined with the real-time monitoring data of the pressure sensor and the photoelectric sensor, the support vector regression algorithm is used to establish a functional relationship model of material parameters, pressure and dose output. The model parameters are optimized through cross-validation to achieve accurate prediction of the anesthetic release dose. Online support vector regression is used as an incremental learning algorithm to dynamically update the model parameters, and adaptive adjustments are made according to the deviation between the baseline value and the actual dose, which effectively improves the accuracy and controllability of anesthetic release and provides reliable technical support for clinical anesthesia administration. Description of the figures

[0061] Figure 1 The figure is a flow chart of the microfluidic anesthetic bullet control method based on pressure feedback of the present invention.

[0062] Figure 2 Schematic diagram of the microfluidic anesthetic bullet control method based on pressure feedback of the present invention.

[0063] Figure 3 This is another schematic diagram of the microfluidic anesthetic bullet control method based on pressure feedback of the present invention.

[0064] The following will describe the technical solutions in the embodiments of the present invention in detail with reference to the accompanying drawings. The described embodiments are only a part of the embodiments of the present invention.

[0065] Example 1:

[0066] like Figure 1-3 The microfluidic anesthetic bullet control method based on pressure feedback in this embodiment may specifically include:

[0067] S101. Obtain a double-layer structure model of a microfluidic anesthetic bomb, wherein the double-layer structure model includes an inner layer anesthetic storage cavity model and an outer layer elastic adjustable layer material model.

[0068] Acquiring three-dimensional data of the inner layer anesthetic storage cavity model;

[0069] According to the three-dimensional data, the inner layer anesthetic storage cavity model is parametrically modeled using SolidWorks software to obtain a three-dimensional solid model of the inner layer anesthetic storage cavity;

[0070] Determining the three-dimensional size parameters of the material model of the outer elastic adjustable layer according to the three-dimensional solid model of the inner layer anesthetic storage cavity;

[0071] According to the three-dimensional size parameters, the material model of the outer elastic adjustable layer is parametrically modeled using SolidWorks software to obtain a three-dimensional solid model of the outer elastic adjustable layer material;

[0072] Assembling the three-dimensional solid model of the inner anesthetic storage cavity and the three-dimensional solid model of the outer elastic adjustable layer material to obtain a three-dimensional assembly model of the double-layer structure of the microfluidic anesthetic bomb;

[0073] Using ANSYS software to perform structural statics analysis on the three-dimensional assembly model to obtain stress distribution and deformation;

[0074] If the stress distribution and deformation conditions do not meet the preset conditions, the size parameters of the inner anesthetic storage cavity model and the outer elastic adjustable layer material model are adjusted according to the stress distribution and deformation conditions until the preset conditions are met.

[0075] For example, based on the double-layer structure model of the microfluidic anesthetic bomb, the three-dimensional data of the inner layer anesthetic storage cavity model is obtained. Using the obtained three-dimensional data, the inner layer anesthetic storage cavity model is parametrically modeled in the SolidWorks software to obtain a three-dimensional solid model of the inner layer anesthetic storage cavity. Based on the three-dimensional solid model of the inner layer anesthetic storage cavity model, the three-dimensional size parameters of the outer elastic adjustable layer material model are determined, including the thickness, length and width of the outer layer material. In the SolidWorks software, the outer layer elastic adjustable layer material model is parametrically modeled using the determined three-dimensional size parameters to obtain a three-dimensional solid model of the outer layer elastic adjustable layer material. In the assembly module of the SolidWorks software, the three-dimensional solid model of the inner layer anesthetic storage cavity and the three-dimensional solid model of the outer layer elastic adjustable layer material are coaxially assembled, and an interference check is performed to ensure that there is no interference in the assembled microfluidic anesthetic bomb double-layer structure model. ANSYS finite element analysis software was used to perform a structural static analysis on the three-dimensional assembly model of the double-layer structure of the microfluidic anesthetic cartridge. The stress distribution and deformation of the microfluidic anesthetic cartridge under external forces were determined. Based on the analysis results, the dimensional parameters of the inner anesthetic storage cavity model and the outer elastic adjustable layer material model were optimized to meet strength and deformation requirements. The optimized inner anesthetic storage cavity model and the outer elastic adjustable layer material model were converted into STL format 3D printing files in SolidWorks software for subsequent 3D printing of microfluidic anesthetic cartridge prototypes.

[0076] S102. For the material model of the outer elastically adjustable layer, a finite element analysis method is used to simulate the deformation and stress distribution of the anesthetic cavity under a preset pressure when the thickness of the outer layer material is discretely sampled within a preset range, and obtain a set of elastic modulus parameters of the outer layer material, wherein the parameter set includes elastic modulus values ​​at different thicknesses.

[0077] The physical properties of the outer elastic adjustable layer material are obtained, a constitutive model of the material is established, and the parameters required for the material model are determined; a three-dimensional geometric model of the anesthetic cavity is constructed, and the geometric model is meshed using finite element analysis software; for a preset thickness range, a series of thickness values ​​are obtained, the outer layer material parameters at different thicknesses are assigned to the geometric model, a preset pressure is set as a load condition, and a finite element solution is performed; based on the finite element analysis results, deformation and stress distribution data at different thicknesses and preset pressures are obtained; based on the deformation and stress distribution data, the elastic modulus values ​​of the outer layer material at different thicknesses are calculated to obtain a set of elastic modulus parameters.

[0078] For example, based on the physical properties of the outer elastic adjustable layer material, a corresponding material constitutive model is established to determine the parameters required for the material model. ANSYS finite element analysis software is used to construct a three-dimensional geometric model of the anesthetic cavity and divide the grid. For a preset thickness range, a discrete sampling method is used to obtain a series of thickness values, and the outer layer material parameters at different thicknesses are assigned to the geometric model. The preset pressure is set as the load condition, and a finite element solution is performed. The deformation and stress distribution data of the anesthetic cavity at different thicknesses and preset pressures are extracted from the finite element analysis results. Based on the deformation and stress distribution data, MATLAB is used to calculate the elastic modulus values ​​of the outer layer material at different thicknesses to form an elastic modulus parameter set. The elastic modulus parameter set is used as the basis for subsequent performance regulation of the outer layer material. By selecting the appropriate material thickness, the performance optimization of the outer elastic adjustable layer is achieved.

[0079] S103. Obtain the pressure value of the microfluidic anesthetic bullet monitored in real time by the pressure sensor, and determine whether the current pressure reaches the preset pressure threshold based on the preset pressure threshold; if so, control the valve connecting the inner anesthetic cavity and the outer adjustable layer to open, and squeeze the inner anesthetic to release it under the action of pressure based on the elastic modulus parameter set of the outer layer material.

[0080] The force value of the anesthetic bullet monitored in real time by the pressure sensor is obtained, and the value is compared with a preset threshold to determine whether the current pressure reaches the preset threshold; if the current pressure reaches the preset threshold, the inner and outer layer connecting valves are controlled to open; based on the anesthetic release rate predicted by the support vector regression model, a PID control algorithm is used to adjust the PWM duty cycle of the inner and outer layer connecting valves in real time to control the valve opening; and a genetic algorithm is used to optimize the parameters of the PID controller so that the valve control strategy can be adaptively adjusted according to the pressure change trend.

[0081] For example, the pressure sensor monitors the force applied to the anesthetic cartridge in real time, compares this value with a preset threshold, and determines whether the current pressure has reached the preset threshold. The pressure sensor is a piezoresistive pressure sensor with a measurement range of 0-10 MPa, installed between the outer and inner layers of the anesthetic cartridge, with a sampling frequency of 100 Hz. If the current pressure reaches the preset threshold, the inner-outer layer connecting valve is controlled to open, establishing a communication channel between the inner anesthetic chamber and the outer adjustable layer. The inner-outer layer connecting valve is a solenoid valve, whose opening is controlled by a PWM signal. Based on the elastic modulus parameters of the outer layer material, a finite element analysis method is used to develop a numerical model of the outer layer material's stress and deformation. The deformation of the outer layer material and the pressure exerted on the inner anesthetic chamber under the current pressure are calculated. A support vector regression (SVR) algorithm is used to establish a mathematical model between the outer layer material's elastic modulus and pressure as inputs and the inner layer anesthetic release as output. The SVR model parameters are obtained by training with existing experimental data to predict the anesthetic release rate under the current pressure. Based on the anesthetic release rate predicted by the SVR model, a PID control algorithm is used to adjust the PWM duty cycle of the inner and outer connecting valves in real time to control the valve opening, ensuring stable and continuous anesthetic release under pressure. The pressure sensor's value is continuously monitored to determine whether the pressure effect is weakening. When the pressure drops below the preset threshold, the inner and outer connecting valves are controlled to close, halting anesthetic release. During the use of the anesthetic bomb, pressure changes and valve control parameters are continuously recorded to generate feedback data. A genetic algorithm is used to optimize the PID controller's parameters, allowing the valve control strategy to adaptively adjust according to pressure trends, improving the accuracy and stability of anesthetic release.

[0082] S104 . During the anesthetic release process, obtain the released anesthetic droplet volume data monitored in real time by the photoelectric sensor, and calculate the released dose according to a pre-established functional relationship between the droplet volume and the anesthetic dose.

[0083] The values ​​of the proportional term, the integral term, and the differential term are obtained, and a weighted sum of the three values ​​is performed to obtain a control quantity; based on the control quantity, the operating parameters of the anesthetic release device are adjusted to change the anesthetic release rate; and a support vector machine algorithm is used to update a preset functional relationship model between droplet volume and anesthetic dose in real time. If the functional relationship model is updated, the dosage calculation accuracy is improved.

[0084] For example, (1) the values ​​of the proportional term, the integral term, and the differential term are calculated; (2) the three values ​​are weighted and summed to obtain the control quantity; and (3) the operating parameters of the anesthetic release device, such as the motor speed and valve opening, are adjusted according to the control quantity to change the anesthetic release rate. The adjusted anesthetic release rate is monitored, the error between the actual release dose and the target dose is calculated, and the error value is fed back to the PID control algorithm for the next round of adjustment until the actual release dose stabilizes near the target dose. The support vector machine (SVM) algorithm is used to update and optimize the functional relationship model between the droplet volume and the anesthetic dose in real time to improve the dose calculation accuracy. The SVM algorithm is trained on the newly collected droplet volume and anesthetic dose data and continuously adjusts the model parameters so that the model can more accurately reflect the relationship between the two. The monitoring data and calculation results of the anesthetic release process are transmitted to the central control system in real time for unified management and analysis.

[0085] S105. Based on the pressure change data monitored by the pressure sensor, the elastic modulus parameters of the outer material and the calculated anesthetic release dose data, a support vector regression algorithm is used to establish a functional relationship model between material parameters, pressure and dose output, and the kernel function and parameters are selected through cross-validation to train and optimize the model parameters.

[0086] The method comprises obtaining pressure change data monitored in real time by a pressure sensor, performing denoising and smoothing on the pressure change data, and extracting at least one statistical feature of the pressure change as a model input; obtaining the elastic modulus parameter of the outer layer material; calculating the target anesthetic release dose under the current state based on a preset pharmacokinetic mathematical model and combining the statistical features of the pressure change and the elastic modulus parameter; dividing the statistical features of the pressure change, the elastic modulus parameter, and the target anesthetic release dose into a training set, a validation set, and a test set, and constructing a support vector regression model.

[0087] For example, real-time pressure change data monitored by a pressure sensor is acquired and preprocessed, including denoising and smoothing, to extract statistical features of the pressure change, such as mean, variance, and peak value, as one input to the model. The elastic modulus parameters of the outer layer material are obtained and used as another input to the model. Based on an established pharmacokinetic mathematical model, the theoretical anesthetic release dose under the current conditions is calculated by combining the pressure characteristics and material parameters. This mathematical model is typically based on the law of conservation of mass and the law of diffusion, describing the release kinetics of the drug within the material. The pressure characteristics, material parameters, and calculated release dose are randomly divided into training, validation, and test sets in an 8:1:1 ratio to construct a support vector regression model. A grid search and 5-fold cross-validation method are used, with mean squared error and coefficient of determination used as evaluation metrics to optimize the kernel function type (e.g., linear kernel, Gaussian kernel) and kernel function parameters (e.g., regularization coefficient, kernel width) for the support vector regression model. Using optimized kernel functions and parameters, a support vector regression model was trained on the training set and then fine-tuned on the validation set to establish a nonlinear functional mapping between pressure characteristics, material parameters, and released dose. The model's generalization performance was evaluated on the test set to ensure its predictive accuracy and robustness. Considering the computing resources and real-time requirements of the embedded system, the trained support vector regression model was optimized, including model compression and fixed-point quantization, to reduce model size and computational complexity. The optimized model was then ported to the embedded system, where it predicted the drug release dose based on real-time pressure data and known material parameters. The drug delivery system was then adjusted in real time based on the predicted values, achieving closed-loop control.

[0088] S106. Use a test data set to evaluate the performance of the functional relationship model and calculate the root mean square error and the coefficient of determination; if the root mean square error is greater than a preset error threshold or the coefficient of determination is less than a preset coefficient threshold, adjust the support vector regression algorithm hyperparameters and retrain the model until a final model that meets the accuracy requirements is obtained.

[0089] A test data set is obtained, and the root mean square error (RMSE) and coefficient of determination (R²) are used as evaluation indicators to evaluate the performance of the support vector regression model. If the RMSE is greater than the preset error threshold, and the R² is less than the preset coefficient threshold, then according to the preset grid search rules, the hyperparameter combinations of penalty coefficient C, kernel function type, and kernel function parameter γ are traversed. For each hyperparameter combination, a cross-validation method is used for evaluation, and the hyperparameter combination with the smallest RMSE on the validation set is obtained as the optimal hyperparameter. From the model set, the model with the smallest RMSE and the largest R² is selected and determined as the final functional relationship model.

[0090] For example, a test dataset is obtained and the performance of the established support vector regression model is evaluated. The root mean square error (RMSE) and coefficient of determination (R²) are used as evaluation metrics. RMSE measures the deviation between the predicted value and the true value, while R² measures the model's fit to the data. The RMSE and R² of the current model are calculated to determine whether the RMSE is greater than a preset error threshold and whether the R² is less than a preset coefficient threshold. If these conditions are met, the model performance is suboptimal and hyperparameter tuning is required. Otherwise, the current model has achieved the desired performance level and can be used as the final functional relationship model, skipping to step 6. Based on the preset grid search rules, different hyperparameter combinations are tested, including the penalty coefficient C, kernel function type (e.g., linear kernel, Gaussian kernel), and kernel function parameter γ. Each hyperparameter combination is evaluated using cross-validation, and the one with the lowest RMSE on the validation set is selected as the optimal hyperparameter combination. Using the optimal hyperparameter combination, the support vector regression model is retrained on the training set to obtain a new candidate model. This model is then added to the model collection. Repeat steps 2 through 4 until the model performance requirements are met or the preset number of iterations is reached. From the model set, select the model with the lowest RMSE and highest R² as the final functional relationship model. Deploy the final support vector regression model to the production environment. When new input data arrives, use the model to predict it and output the corresponding target value. At the same time, continuously monitor the model's predictive performance in actual business scenarios and update and optimize the model as necessary.

[0091] S107. Based on the final functional relationship model after verification and optimization, predict the anesthetic dose released by the anesthetic bullet with specific outer layer material parameters under the current pressure conditions, and control the elastic modulus of the anesthetic bullet by adjusting the thickness of the outer layer material to achieve precise control of the released dose. The predicted dose result is used as a benchmark value for adaptive control.

[0092] According to the outer material parameters of the anesthetic bullet and the current pressure conditions, the optimized functional relationship model is used to calculate the predicted anesthetic release dose. By establishing a mapping relationship between material thickness and elastic modulus, the influence of adjusting the material thickness on the elastic modulus is determined. According to the predicted release dose, the mapping relationship is used to calculate the material elastic modulus required to achieve the target release dose. The predicted release dose is used as the benchmark value for adaptive control, and a feedback control system based on the PID control algorithm is established. The release rate and release amount of the anesthetic bullet are monitored in real time by a sensor, and the monitored value is compared with the benchmark value to calculate the deviation value. According to the deviation value, the PID control algorithm is used to calculate the adjustment amount of the material thickness, and the material thickness is dynamically adjusted through the actuator to ensure that the release dose continues to meet the requirements of precise control.

[0093] For example, an optimized functional relationship model is used to calculate a predicted anesthetic release dose based on the outer layer material parameters and current pressure conditions. A mapping relationship between material thickness and elastic modulus is established to determine the effect of adjusting material thickness on the elastic modulus. This mapping relationship is then used to calculate the material elastic modulus required to achieve the target release dose based on the predicted release dose. The calculated target elastic modulus is compared with the material's current elastic modulus to determine whether the outer layer material thickness needs to be adjusted. If adjustment is necessary, the material thickness corresponding to the target elastic modulus is calculated based on the material thickness-elastic modulus mapping relationship and used as a reference for adjustment. The outer layer material is then precision-machined to achieve the calculated target thickness, thereby achieving the desired target elastic modulus. The adjusted outer layer material is then used in the manufacture of the anesthetic bolus, and precise control of the release dose is achieved by precisely controlling the material thickness during the molding process. A feedback control system based on a PID control algorithm is established, using the predicted release dose as a baseline for adaptive control. During use, sensors monitor the release rate and release amount of the anesthetic bolus in real time. The monitored values ​​are compared with the baseline values ​​to calculate a deviation. According to the deviation value, the PID control algorithm is used to calculate the adjustment amount of the material thickness, and the material thickness is dynamically adjusted through the actuator to ensure that the release dose continues to meet the requirements of precise control.

[0094] S108. Continuously collect real-time data on pressure and released dose, use online support vector regression as an incremental learning algorithm, merge the newly collected data with the original training data set, dynamically update and optimize the function relationship model parameters, gradually adjust the model parameters according to the deviation between the baseline value and the actual dose, improve the model generalization ability and prediction accuracy, and realize adaptive control of anesthetic dose.

[0095] Data collected in real time by the pressure sensor and the anesthesia pump are acquired, and the pressure data and dosage data are associated and stored in a time series database. A gradient boosting decision tree algorithm is used to retrain the support vector regression model using the merged data set, and the parameters of the model are dynamically updated. The optimized support vector regression model is applied to the adaptive control of the anesthesia pump. According to the patient's real-time physiological parameters and surgical requirements, the output dosage of the anesthesia pump is dynamically adjusted through a PID control algorithm to achieve precise anesthesia management.

[0096] For example, based on real-time data collected by a pressure sensor and an anesthesia pump, pressure and dose data are associated and stored in a time series database. At preset time intervals, the latest batch of pressure-dose data is retrieved from the time series database and merged with the original training dataset of the support vector regression model. A gradient boosting decision tree algorithm is used to retrain the support vector regression model using the merged dataset, dynamically updating and optimizing the model parameters. The patient's current physiological parameters, including heart rate, blood pressure, and blood oxygen saturation, are acquired in real time via medical monitoring equipment. The optimal baseline anesthetic dose for the patient is calculated using the support vector regression model. The actual output dose of the anesthesia pump is obtained in real time via the anesthesia pump's data interface, and the deviation from the baseline is calculated to determine whether the deviation exceeds a preset threshold. If the deviation exceeds the threshold, the relevant parameters of the support vector regression model are automatically adjusted based on the direction and magnitude of the deviation, improving the model's generalization and prediction accuracy. The optimized support vector regression model is applied to the adaptive control of the anesthesia pump. Based on the patient's real-time physiological parameters and surgical requirements, the output dose of the anesthesia pump is dynamically adjusted using a PID control algorithm, achieving precise anesthesia management. At the same time, a decision tree model is introduced to conduct real-time assessment of the patient's physiological state, and to assist the support vector regression model in adjusting the anesthetic dose to improve the safety and effectiveness of anesthesia management.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A microfluidic anesthetic bullet control method based on pressure feedback, characterized in that: The method comprises: Obtaining a double-layer structure model of a microfluidic anesthetic bomb, wherein the double-layer structure model includes an inner layer anesthetic storage cavity model and an outer layer elastic adjustable layer material model; For the material model of the outer elastically adjustable layer, a finite element analysis method is used to simulate the deformation and stress distribution of the anesthetic cavity under a preset pressure when the thickness of the outer layer material is discretely sampled within a preset range, thereby obtaining a set of elastic modulus parameters of the outer layer material, wherein the parameter set includes elastic modulus values ​​at different thicknesses; Obtaining a pressure value of the microfluidic anesthetic bomb monitored in real time by a pressure sensor, and determining whether the current pressure reaches the preset pressure threshold based on a preset pressure threshold. If so, controlling the valve connecting the inner anesthetic cavity and the outer adjustable layer to open, and squeezing the inner layer anesthetic under the action of pressure based on the elastic modulus parameter set of the outer layer material to release the anesthetic; During the anesthetic release process, the volume data of the released anesthetic droplets monitored in real time by the photoelectric sensor is obtained, and the released dose is calculated according to the pre-established functional relationship between the droplet volume and the anesthetic dose; Based on the pressure change data monitored by the pressure sensor, the elastic modulus parameters of the outer material, and the calculated anesthetic release dose data, a support vector regression algorithm was used to establish a functional relationship model between material parameters, pressure, and dose output. The kernel function and parameters were selected through cross-validation, and the model parameters were trained and optimized. The performance of the functional relationship model is evaluated using a test data set, and the root mean square error and the coefficient of determination are calculated. If the root mean square error is greater than a preset error threshold or the coefficient of determination is less than a preset coefficient threshold, the support vector regression algorithm hyperparameters are adjusted and the model is retrained until a final model that meets the accuracy requirements is obtained; Based on the final functional relationship model after verification and optimization, the anesthetic dose released by the anesthetic bullet with specific outer layer material parameters under the current pressure conditions is predicted. By adjusting the thickness of the outer layer material to control the elastic modulus of the anesthetic bullet, precise control of the released dose is achieved. The predicted dose result is used as the benchmark value for adaptive control; Real-time data on pressure and released dose are continuously collected, and online support vector regression is used as an incremental learning algorithm to merge the newly collected data with the original training data set. The parameters of the functional relationship model are dynamically updated and optimized. The model parameters are gradually adjusted according to the deviation between the baseline value and the actual dose, thereby improving the model generalization ability and prediction accuracy, and realizing adaptive control of anesthetic dose.

2. The method according to claim 1, characterized in that The double-layer structure model of the microfluidic anesthetic bomb is obtained, and the double-layer structure model includes an inner layer anesthetic storage cavity model and an outer layer elastic adjustable layer material model, including: Acquiring three-dimensional data of the inner layer anesthetic storage cavity model; According to the three-dimensional data, the inner layer anesthetic storage cavity model is parametrically modeled using SolidWorks software to obtain a three-dimensional solid model of the inner layer anesthetic storage cavity; Determining the three-dimensional size parameters of the material model of the outer elastic adjustable layer according to the three-dimensional solid model of the inner layer anesthetic storage cavity; According to the three-dimensional size parameters, the material model of the outer elastic adjustable layer is parametrically modeled using SolidWorks software to obtain a three-dimensional solid model of the outer elastic adjustable layer material; Assembling the three-dimensional solid model of the inner anesthetic storage cavity and the three-dimensional solid model of the outer elastic adjustable layer material to obtain a three-dimensional assembly model of the double-layer structure of the microfluidic anesthetic bomb; Using ANSYS software to perform structural statics analysis on the three-dimensional assembly model to obtain stress distribution and deformation; If the stress distribution and deformation conditions do not meet the preset conditions, the size parameters of the inner anesthetic storage cavity model and the outer elastic adjustable layer material model are adjusted according to the stress distribution and deformation conditions until the preset conditions are met.

3. The method according to claim 2, characterized in that For the material model of the outer elastically adjustable layer, a finite element analysis method is used to simulate the deformation and stress distribution of the anesthetic cavity under a preset pressure when the thickness of the outer layer material is discretely sampled within a preset range, thereby obtaining a set of elastic modulus parameters of the outer layer material. The parameter set includes elastic modulus values ​​at different thicknesses, including: Obtaining the physical properties of the material of the outer elastic adjustable layer, establishing a constitutive model of the material, and determining the parameters required for the material model; Constructing a three-dimensional geometric model of the anesthetic cavity and meshing the geometric model using finite element analysis software; For a preset thickness range, a series of thickness values ​​are obtained, outer layer material parameters at different thicknesses are assigned to the geometric model, a preset pressure is set as a load condition, and a finite element solution is performed; Based on the finite element analysis results, the deformation and stress distribution data under different thicknesses and preset pressures are obtained; The elastic modulus values ​​of the outer layer material at different thicknesses are calculated based on the deformation and stress distribution data to obtain an elastic modulus parameter set.

4. The method according to claim 3, characterized in that The pressure sensor is used to obtain the pressure value of the microfluidic anesthetic bullet monitored in real time, and the preset pressure threshold is used to determine whether the current pressure reaches the preset pressure threshold; If so, the valve connecting the inner anesthetic cavity and the outer adjustable layer is controlled to open, and the inner anesthetic is squeezed and released under pressure according to the elastic modulus parameter set of the outer layer material, including: Obtaining a force value of the anesthetic bullet monitored in real time by the pressure sensor, comparing the value with a preset threshold, and determining whether the current pressure reaches the preset threshold; If the current pressure reaches a preset threshold, the inner and outer layer connecting valves are controlled to open; According to the anesthetic release rate predicted by the support vector regression model, a PID control algorithm is used to adjust the PWM duty cycle of the inner and outer layer connecting valves in real time to control the valve opening; The parameters of the PID controller are optimized using a genetic algorithm so that the valve control strategy can be adaptively adjusted according to the pressure change trend.

5. The method according to claim 4, characterized in that During the anesthetic release process, obtaining the released anesthetic droplet volume data monitored in real time by the photoelectric sensor, and calculating the released dose according to a pre-established functional relationship between the droplet volume and the anesthetic dose, including: Obtain the values ​​of the proportional term, integral term, and differential term, perform weighted summation on the three values, and obtain the control quantity; adjusting the operating parameters of the anesthetic release device according to the control amount to change the anesthetic release rate; A support vector machine algorithm is used to update the preset functional relationship model between the droplet volume and the anesthetic dose in real time. If the functional relationship model is updated, the dose calculation accuracy is improved.

6. The method according to claim 5, characterized in that The method comprises: using the pressure change data monitored by the pressure sensor, the elastic modulus parameters of the outer layer material, and the calculated anesthetic release dose data, using a support vector regression algorithm to establish a functional relationship model between material parameters, pressure, and dose output, selecting kernel functions and parameters through cross-validation, and training and optimizing model parameters, including: Acquire pressure change data monitored in real time by a pressure sensor, perform denoising and smoothing on the pressure change data, and extract at least one statistical feature of the pressure change as a model input; Obtaining elastic modulus parameters of the outer layer material; Calculating a target anesthetic release dose under a current state based on a preset pharmacokinetic mathematical model, in combination with the pressure change statistical characteristics and the elastic modulus parameter; The pressure change statistical characteristics, the elastic modulus parameters and the target anesthetic release dose are divided into a training set, a validation set and a test set, and a support vector regression model is constructed.

7. The method according to claim 6, characterized in that The test data set is used to evaluate the performance of the functional relationship model, and the root mean square error and the coefficient of determination are calculated; if the root mean square error is greater than a preset error threshold or the coefficient of determination is less than a preset coefficient threshold, the support vector regression algorithm hyperparameters are adjusted and the model is retrained until a final model that meets the accuracy requirements is obtained, including: Obtain a test data set and use the root mean square error (RMSE) and the coefficient of determination (R²) as evaluation indicators to evaluate the performance of the support vector regression model; If the RMSE is greater than the preset error threshold and the R² is less than the preset coefficient threshold, then according to the preset grid search rules, traverse the penalty coefficient C, kernel function type and kernel function parameter γ hyperparameter combination; For each hyperparameter combination, a cross-validation method is used to evaluate and obtain the hyperparameter combination with the smallest RMSE on the validation set as the optimal hyperparameter; From the model set, the model with the smallest RMSE and the largest R² is selected and determined as the final functional relationship model.

8. The method according to claim 7, characterized in that The final functional relationship model after verification and optimization is used to predict the anesthetic dose released by the anesthetic bullet with specific outer layer material parameters under current pressure conditions, and the elastic modulus of the anesthetic bullet is controlled by adjusting the thickness of the outer layer material to achieve precise control of the released dose. The predicted dose result is used as a reference value for adaptive control, including: According to the outer material parameters of the anesthetic bullet and the current pressure conditions, the optimized functional relationship model is used to calculate the predicted anesthetic release dose; By establishing a mapping relationship between material thickness and elastic modulus, the influence of adjusting material thickness on elastic modulus is determined; According to the predicted release dose, the elastic modulus of the material required to achieve the target release dose is calculated using the mapping relationship; The predicted release dose is used as the reference value for adaptive control, and a feedback control system based on the PID control algorithm is established. The release rate and amount of the anesthetic bullet are monitored in real time using a sensor, the monitored value is compared with the reference value, and the deviation value is calculated; According to the deviation value, the PID control algorithm is used to calculate the adjustment amount of the material thickness, and the material thickness is dynamically adjusted through the actuator to ensure that the release dose continues to meet the requirements of precise control.

9. The method according to claim 8, characterized in that The method continuously collects real-time data on pressure and released dose, uses online support vector regression as an incremental learning algorithm, merges newly collected data with the original training data set, dynamically updates and optimizes the parameters of the functional relationship model, and gradually adjusts the model parameters based on the deviation between the baseline value and the actual dose, thereby improving the model's generalization ability and prediction accuracy, and achieving adaptive control of anesthetic dose, including: Acquire data collected in real time by a pressure sensor and an anesthesia pump, and associate the pressure data with the dosage data and store them in a time series database; Using the gradient boosting decision tree algorithm, the support vector regression model is retrained using the merged data set to dynamically update the parameters of the model; The optimized support vector regression model is applied to the adaptive control of the anesthesia pump. According to the patient's real-time physiological parameters and surgical requirements, the output dose of the anesthesia pump is dynamically adjusted through the PID control algorithm to achieve precise anesthesia management.

Images