3D hepatotoxicity test monodisperse microgel preparation method and system, and medium

Abstract

The invention discloses a 3D hepatotoxicity test monodisperse microgel preparation method, a 3D hepatotoxicity test monodisperse microgel preparation system and a medium, and belongs to the technical field of 3D cell culture and drug screening. The system comprises a preparation module, a statistical analysis module, a material processing module and a control system, the method comprises the following steps: collecting historical data to train an adaptive neural network, and establishing a mapping relation between control parameters and microgel quality indexes; a neural network is used as a prediction model, and optimal control input is solved through a model prediction controller; performing microgel preparation; acquiring images in real time to obtain and feed back actual particle sizes and variation coefficients; applying the new data to incremental updating of the neural network; and performing iteration to form a closed-loop optimization process. According to the method, multi-parameter self-adaptive optimization is realized through an artificial intelligence algorithm, the problem of manual exploration of parameters of different hydrogel materials is solved, the monodispersity and consistency of microgel are improved, and full-process automation is realized.

Description

Technical Field

[0001] This invention relates to the field of 3D cell culture and drug screening technology, and in particular to a method, system and medium for preparing monodisperse microgels for 3D hepatotoxicity testing. Background Technology

[0002] Three-dimensional (3D) cell culture technology has shown broad application prospects in drug screening, toxicity evaluation, and tissue engineering due to its ability to better simulate the in vivo cell growth microenvironment and maintain cell-cell and cell-extracellular matrix (ECM) interactions. Compared with traditional two-dimensional (2D) monolayer cell culture, 3D cell culture models more closely resemble the in vivo physiological state in terms of gene expression, protein expression, and drug response, significantly improving the accuracy of drug efficacy evaluation and toxicity testing. Statistics show that the clinical failure rate of traditional 2D cell screening models is as high as approximately 95%, mainly because 2D models cannot accurately reflect the metabolic processes and toxic reactions of drugs in vivo.

[0003] Existing 3D cell culture technologies can be categorized into scaffold-free and scaffold-based methods based on whether or not scaffold materials are used. Scaffold-free 3D culture methods include hanging drop culture, low-adhesion plate culture, and magnetized cell culture; scaffold-based 3D culture methods include nanosieve methods and hydrogel methods. Among these, hydrogels are considered one of the most promising scaffold materials for 3D cell culture due to their excellent biocompatibility, tunable physicochemical properties, and ability to simulate the microenvironment of the natural extracellular matrix.

[0004] 3D cell culture in hydrogel form, particularly methods involving encapsulating cells in microgel spheres for three-dimensional culture, has attracted widespread attention in recent years. The technique of preparing monodisperse microgel spheres using a high-voltage electrostatic field can achieve uniform distribution and growth of cells in three-dimensional space. However, existing hydrogel microsphere preparation techniques and their applications in 3D hepatotoxicity testing still have the following technical problems and shortcomings: 1. The size of hydrogel microspheres significantly impacts their internal mass transfer processes. When microspheres are too large, the limited diffusion of oxygen, nutrients, and metabolites leads to significant differences in the microenvironment between cells in the central and peripheral regions. Cells in the central region may experience decreased viability or even death due to hypoxia or nutrient deficiency, while cells in the peripheral regions grow well. This inconsistency in cell growth status within the same microsphere severely affects the scientific validity and reliability of 3D cell culture models, making it difficult for drug screening results based on such models to accurately reflect the true toxicity of drugs. Therefore, preparing smaller hydrogel microspheres with better monodispersity is crucial for solving mass transfer problems and improving the scientific validity of models.

[0005] 2. Hydrogel materials developed by different research teams exhibit varying physicochemical properties and electrohydrodynamic behaviors due to differences in parameters such as composition, crosslinking method, molecular weight, and concentration. When preparing microgel spheres using high-voltage electrostatic methods, a series of process parameters need to be optimized for each hydrogel system, including electric field strength, working distance, propulsion speed, pulse frequency, and duty cycle. These parameters are interconnected, exerting complex influences on the particle size, morphology, and monodispersity of the microspheres. Traditional parameter optimization methods rely on extensive manual experimentation, which is not only time-consuming and labor-intensive but also makes it difficult to obtain globally optimal parameter combinations. This results in inconsistent monodispersity of the microspheres and poor quality consistency between different batches, failing to meet the repeatability requirements of high-throughput drug screening.

[0006] 3. Currently, the main methods for preparing hydrogel microspheres include microfluidic methods, emulsion crosslinking methods, and electrohydrodynamic methods, but each of these methods has its own drawbacks: Microfluidic methods typically require the introduction of an oil phase as a continuous phase, involving two-phase fluid operations of water and oil. The system is complex and the operation is cumbersome. Furthermore, the use of organic solvents may adversely affect cell viability. At the same time, the processing cost of microfluidic chips is high, making it difficult to achieve large-scale, high-throughput fabrication.

[0007] Emulsification and cross-linking methods typically involve forming an emulsion through mechanical stirring or ultrasonic emulsification, followed by reaction with a cross-linking agent under shear force to form microspheres. Microspheres prepared by this method have a wide particle size distribution and poor monodispersity, and the shear force during stirring may damage cells, making it unsuitable for shear-sensitive cell types.

[0008] Traditional electrohydrodynamic methods for generating high-voltage electrostatic microspheres often employ static high-voltage electric fields or simple pulsed high voltages. Multiple parameters, such as voltage, frequency, duty cycle, working distance, and propulsion speed, require manual setting and adjustment. Different hydrogel materials necessitate re-exploration of these parameters, lacking self-adaptive capabilities. Furthermore, traditional devices are often open structures, susceptible to external electromagnetic interference, and lack real-time quality monitoring and feedback control mechanisms, making closed-loop optimization of the fabrication process impossible.

[0009] 4. Current microgel preparation, cleaning, culture medium replacement, and drug addition are mostly performed manually in separate steps, which is not only inefficient but also prone to operational errors and contamination risks. This is especially problematic for high-throughput drug screening applications that require processing large numbers of samples; manual operations cannot meet the requirements for experimental throughput and consistency. Furthermore, there is currently a lack of automated systems that integrate microgel preparation, quality control, parameter optimization, cell culture, and drug testing, making it impossible to achieve closed-loop control throughout the entire process from microsphere preparation to hepatotoxicity testing.

[0010] 5. In terms of the control of microgel preparation equipment, traditional control methods such as PID control and fuzzy control rely on fixed control models and parameters, making it difficult to adapt to the control requirements of different hydrogel materials and nonlinear time-varying systems. For complex multiple-input multiple-output (MIMO) systems, traditional control methods often fail to achieve multi-objective synergistic optimization, resulting in insufficient control accuracy and adaptability.

[0011] In summary, there is an urgent need for a monodisperse microgel preparation system that can adapt to different hydrogel materials, automatically optimize preparation parameters, detect microsphere quality in real time, and achieve full-process automation, in order to meet the requirements of 3D hepatotoxicity testing for high-quality, high-consistency, and high-throughput cell culture models. Summary of the Invention

[0012] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide a method for preparing monodisperse microgels for 3D hepatotoxicity testing, comprising the following steps: S1. Collect at least one set of control parameters and their corresponding microgel ball quality indicators from the historical preparation process, and train an adaptive neural network using the control parameters and their corresponding quality indicators to establish a mapping relationship between the control parameters and the microgel ball quality indicators; wherein, the control parameters include at least electric field parameters, working distance and propulsion speed, and the quality indicators include at least the particle size and coefficient of variation of the microgel balls; S2. Using a trained adaptive neural network as a prediction model, and through a model prediction controller, under the condition of satisfying the preset physical constraints, solve for the optimal control input sequence that minimizes the error between the predicted microgel ball quality index and the target quality index in the future finite time domain. S3. Apply the first control input in the optimal control input sequence to the electrohydrodynamic microsphere generator to prepare microgel spheres from the gel suspension containing cells. S4. Acquire images of the microgel spheres prepared in the current batch, obtain their actual particle size and coefficient of variation through image analysis, and feed this actual quality index back to the control system. S5. Using the control parameters and actual quality indicators of the current batch as new training data, incrementally learn the adaptive neural network and update the network parameters in real time. S6. Repeat steps S2 to S5 to achieve closed-loop iterative control of the microgel sphere preparation process.

[0013] Furthermore, the architecture of the adaptive neural network includes: The input layer contains neuron nodes corresponding to the control parameters; Multiple hidden layers are used, employing a fully connected structure and a non-linear activation function to learn the non-linear mapping relationship between control parameters and quality indicators; The output layer contains two neuron nodes, corresponding to the predicted microgel ball size and coefficient of variation, respectively.

[0014] Furthermore, the training loss function of the adaptive neural network is: in, and The actual measured diameter and coefficient of variation of the microgel spheres. and The diameter and coefficient of variation predicted by the neural network. The weighting hyperparameters are used to balance the diameter error and the coefficient of variation error.

[0015] Furthermore, the optimization objective function of the model predictive controller is: in, For at any time The predicted output, For the desired output, To control the input, and Let N be the weighted matrix, and N be the prediction range.

[0016] Furthermore, the physical constraints include: Control input constraints: ; System output constraints: ; in, and The physical adjustable range of electric field parameters, working distance, and propulsion speed. and The allowable range for the particle size and coefficient of variation of the target microgel spheres.

[0017] Furthermore, the electrohydrodynamic microsphere generator employs a high-voltage power supply; the method further includes: By adjusting the voltage amplitude of the high-voltage power supply, the number of microgel balls in each spraying cycle can be precisely controlled, thereby precisely controlling the number of microgel balls entering each well of the multi-well plate or culture dish to ensure the consistency of the number of 3D cultured cells in each well.

[0018] Furthermore, the step of obtaining the actual particle size and coefficient of variation through image analysis includes: Images of microgel spheres were acquired using a microscopic imaging system; Image recognition algorithms are used to identify individual microgel spheres in an image and the diameter of each microgel sphere is calculated. The particle size distribution of all microgel spheres was statistically analyzed, and the average diameter and coefficient of variation were calculated as the actual quality indicators.

[0019] Furthermore, it also includes S7 and fully automated process steps: After the preparation in step S3 is completed, the culture container containing microgel spheres is automatically transferred to the medium exchange area through an automated material transfer and liquid handling device. The gel forming solution is removed, the washing solution is added and removed, and the cell culture medium is added and replaced in sequence. Then the culture container is transferred to the culture area for three-dimensional culture.

[0020] A second objective of this invention is to provide a 3D hepatotoxicity testing monodisperse microgel preparation system, comprising: Automation platform; The monodisperse microgel preparation module, set on the automated platform, is based on the principle of electrohydrodynamics and uses adjustable electric field parameters, working distance and propulsion speed to prepare cell-containing gel suspensions into microgel spheres. The gel microsphere statistical analysis module, set on the automated platform, is used to image the prepared microgel spheres and statistically analyze their particle size distribution and coefficient of variation. The material transfer and liquid handling module is set on the automated platform and is used to automatically transfer culture containers within the automated platform and perform the addition and replacement of gel forming solution, washing solution and cell culture medium; The control system is communicatively connected to the monodisperse microgel preparation module, the gel microsphere statistical analysis module, and the material transfer and liquid handling module, and is configured to execute the above-described method steps.

[0021] Furthermore, the monodisperse microgel preparation module includes: Syringe clamp, used to hold a syringe containing a gel suspension; A syringe plunger drive mechanism, connected to the syringe plunger, is used to push the syringe plunger at an adjustable advance speed to extrude the gel suspension; A high-voltage power supply is used to generate an adjustable high-voltage electrical signal, the voltage amplitude of which is adjustable. The injection needle is connected to the outlet of the syringe; A ring electrode, encapsulated and disposed near the injection needle and electrically connected to the high-voltage power supply, is used to create an electric field between the injection needle and the culture container containing the gel-forming solution.

[0022] Furthermore, the statistical analysis module for the gel microspheres includes: The imaging unit, including a surface light source, a microscope objective, and an image sensor, is used to acquire high-resolution images of microgel spheres inside the culture container; The image processing unit is configured to identify, segment, and measure the acquired images, calculate the particle size distribution and coefficient of variation of the microgel spheres, and send the statistical results to the control system.

[0023] Furthermore, the material transfer and liquid handling module includes: At least one pipetting module with liquid level detection function is used for quantitative transfer, addition and aspiration of gel forming solution, washing solution and cell culture medium; At least one rotating gripper module is used to grip and transfer culture containers between different workstations on the automated platform.

[0024] Furthermore, the automated platform includes a HEPA filtration system and / or ultraviolet sterilization device to provide a sterile environment for cell manipulation and three-dimensional culture.

[0025] A third object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0026] Compared with the prior art, the beneficial effects of the present invention are: This invention encapsulates the high-voltage generation module and the ring electrode within the electrohydrodynamic microsphere generation module, effectively isolating external electromagnetic interference and avoiding the risk of high-voltage exposure. This significantly improves the safety and stability of the equipment operation, providing a reliable guarantee for long-term, high-throughput automated operation.

[0027] This invention integrates a high-precision imaging and detection module to acquire images of microgel spheres in real time and statistically analyze particle size distribution and coefficient of variation (CV). Combined with artificial intelligence algorithms, the system can continuously accumulate experimental data during operation. Through the synergistic optimization of adaptive neural networks and model predictive control, it dynamically adjusts electric field parameters, working distance, and propulsion speed to achieve adaptive optimization of preparation parameters. With an increase in the number of experiments, the system's understanding of material properties deepens, and the monodispersity of the prepared microgel spheres continuously improves, eventually stabilizing at an optimal level, effectively solving the technical challenge of ensuring monodispersity in traditional methods.

[0028] This invention integrates microgel preparation, gel-forming solution addition, washing solution replacement, and culture medium addition into a single automated platform, achieving fully unmanned operation through material transfer and liquid handling modules. This not only significantly improves experimental throughput and operational efficiency, reducing operational errors and contamination risks caused by human intervention, but also ensures consistency of operating conditions across different batches, providing a standardized experimental basis for high-throughput drug screening.

[0029] This invention utilizes a neural network model to learn the behavioral characteristics of different hydrogel materials during electrohydrodynamic processes, and combines this with model predictive control to achieve rapid adaptive optimization of parameters. Whether for single-component hydrogels or composite material systems, the system can quickly converge to the optimal parameter combination based on limited experimental data, producing microgel spheres with excellent monodispersity. This characteristic significantly simplifies the process optimization work in new material development, reducing the traditionally weeks- or even months-long manual trial-and-error cycle to hours, greatly improving development efficiency and experimental flexibility.

[0030] The microgel spheres prepared by this invention exhibit controllable size and good monodispersity, effectively solving the problem of inconsistent cell growth states within the microspheres due to mass transfer limitations. This ensures that cells within the same microsphere and across different microspheres are in a similar culture microenvironment. Simultaneously, precise control of the number of microspheres per well guarantees consistency in cell counts across different culture wells. These factors work together to significantly improve the scientific rigor and experimental repeatability of the 3D hepatotoxicity testing model, providing a more reliable evaluation tool for drug screening.

[0031] This invention automates the complex parameter optimization process using artificial intelligence algorithms, allowing operators to obtain high-quality microgel spheres simply by inputting basic parameters such as the target microsphere size. This feature reduces the need for specialized operator experience, enabling more research teams to easily apply hydrogel-based 3D cell culture technology and promoting its widespread application in drug screening.

[0032] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0033] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 Diagram of the system architecture for preparing monodisperse microgels for 3D hepatotoxicity testing; Figure 2 This is a control process diagram of a monodisperse microgel preparation system for 3D hepatotoxicity testing. Figure 3 Schematic diagram of a monodisperse microgel preparation system for 3D hepatotoxicity testing; Figure 4 Internal structure diagram of a monodisperse microgel preparation system for 3D hepatotoxicity testing; Figure 5This is a structural diagram of the material transfer and liquid handling module; Figure 6 A schematic diagram of the layout of the core modules of the automation platform; Figure 7 This is a structural diagram of the gel microsphere statistical analysis module; Figure 8 This is a diagram of the imaging unit structure. Figure 9 Schematic diagram of the module for preparing monodisperse microgels; Figure 10 Structural diagram of syringe clamp, drive mechanism and electric carrier for culture dish; Figure 11 This is a structural diagram of the injection needle and annular electrode; Figure 12 Flowchart of a method for preparing monodisperse microgels for 3D hepatotoxicity testing; Figure 13 This is a diagram of a neural network architecture. Figure 14 A flowchart for obtaining the actual particle size and coefficient of variation through image analysis; Figure 15 Development diagram of 3D hydrogel materials for hepatotoxicity testing; Figure 16 Figure showing the effect of different preparation parameters on the particle size and monodispersity of microgel spheres; Figure 17 A comparative graph showing the changes in hepatocyte function over time in different culture models; Figure 18 A schematic diagram of computer equipment; Figure 19 This is a schematic diagram of a computer-readable storage medium.

[0034] In the diagram: 1. HEPA filtration system; 2. Petri dish adapter holder; 3. Pipe tip holder; 4. Gel microsphere statistical analysis module; 5. Waste station; 6. Circuit board frame; 7. Pipetting Y-axis; 8. X-axis; 9. Single-channel pipetting module; 10. Electrical control system; 11. Gripper Z-axis; 12. Gripper Y-axis; 13. Rotary gripper module; 16. Monodisperse microgel preparation module; 17. Surface light source; 18. X-axis displacement stage; 19. Y-axis displacement stage; 20. Imaging system base; 21. 22. CCD image sensor; 23. Microscope objective lens; 24. Right-angle mirror; 25. Syringe clamp; 26. Electric petri dish carrier; 27. Syringe plunger; 28. Petri dish adapter; 29. ​​Sensor; 30. Cross roller guide; 31. Stepper motor; 32. Syringe clamp groove; 33. Carrier support rod; 34. Electric petri dish carrier motor; 35. Worm gear; 36. Electrode connection; 37. Feed line; 38. Injection needle; 39. Ring electrode.

[0035] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0036] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0037] The drawing numbers in this application are only used to distinguish the steps in the scheme and are not used to limit the execution order of the steps. The specific execution order is as described in the specification.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0039] In the field of 3D cell culture technology, based on whether or not scaffold materials are used, it can be divided into two categories: scaffold-free and scaffold-based. Scaffold-free 3D culture includes methods such as hanging drop culture, low-viscosity plates, and magnetized cell culture. For example, hanging drop culture plates and low-viscosity plates such as Corning are scaffold-free 3D cell culture models. Scaffold-based 3D culture includes methods such as nanosieves and hydrogel methods. Among these, hydrogels are widely used in the field of 3D cell culture because they can simulate the microenvironment of the natural extracellular matrix. However, traditional hydrogel-based 3D culture models often struggle to achieve automated scientific experiments due to their viscous state and mass transfer problems after gel formation, and the monodispersity of microspheres is difficult to guarantee.

[0040] Currently, the main methods for preparing hydrogel microspheres include electrohydrodynamics, microfluidics, and emulsion crosslinking. Microfluidics often requires multiphase fluid manipulation (such as aqueous or oil phases), involving multiple solutions and involving complex steps. Emulsion crosslinking typically uses stirring emulsification, forming microspheres under shear force, resulting in poor monodispersity. Traditional electrohydrodynamic equipment often employs static high-voltage electric fields, involving multiple parameters such as voltage, working distance, and propulsion speed, all of which significantly affect microsphere size and monodispersity. In particular, hydrogel materials developed by different research teams exhibit varying behavior in electrohydrodynamic processes due to differences in composition and crosslinking methods, requiring extensive exploration and optimization of electrical parameters to achieve good preparation results; this process is complex and inefficient.

[0041] To address the aforementioned problems, this invention develops a bioactive hydrogel and designs an adaptive, autonomous learning intelligent monodisperse hydrogel preparation system for the hydrogel system. This system employs artificial intelligence combined with automated preparation and detection to form a closed-loop system, capable of autonomously learning based on the characteristics of different hydrogel systems to rapidly prepare hydrogel microspheres with good monodispersity for drug screening applications. The specific scheme is as follows: Example 1

[0042] A 3D hepatotoxicity testing monodisperse microgel preparation system is provided for drug screening applications in 3D cell culture. This system rapidly establishes a well-dispersed, three-dimensional cell culture hepatotoxicity model for high-throughput drug screening. Figure 1 , Figure 2 As shown, the system includes: Automation platform; The monodisperse microgel preparation module, set on the automated platform, is based on the principle of electrohydrodynamics and uses adjustable electric field parameters, working distance and propulsion speed to prepare cell-containing gel suspensions into microgel spheres. The gel microsphere statistical analysis module, set on the automated platform, is used to image the prepared microgel spheres and statistically analyze their particle size distribution and coefficient of variation. The material transfer and liquid handling module is set on the automated platform and is used to automatically transfer culture containers within the automated platform and perform the addition and replacement of gel forming solution, washing solution and cell culture medium; The control system is communicatively connected to the monodisperse microgel preparation module, the gel microsphere statistical analysis module, and the material transfer and liquid handling module. The control system is configured to execute the steps of the 3D hepatotoxicity test monodisperse microgel preparation method. A detailed description of the method can be found in the corresponding descriptions in the following method embodiments, and will not be repeated here.

[0043] The monodisperse microgel preparation module can prepare hydrogel microspheres by combining the action of a high-voltage electrostatic field with multi-parameter adjustment of electrohydrodynamics. In some embodiments, such as Figure 4 , Figure 6 , Figures 9-12 As shown, the monodisperse microgel preparation module 16 includes: The syringe clamp 24 is used to fix a syringe containing a gel suspension; the syringe clamp is provided with a syringe clamp groove 31 to accommodate syringes of different sizes.

[0044] A syringe plunger drive mechanism, connected to the syringe plunger, is used to push the syringe plunger at an adjustable speed to extrude the gel suspension. Specifically, the syringe plunger drive mechanism includes a stepper motor 30, a cross roller guide 29, and a sensor 28. The stepper motor 30 drives the syringe plunger 26 to push the syringe plunger at an adjustable speed to extrude the gel suspension. The sensor 28 is used to monitor the plunger position to achieve precise control.

[0045] The high-voltage power supply is integrated into the electronic control system 10 to generate an adjustable high-voltage electrical signal with adjustable voltage amplitude. The injection needle 37 is connected to the outlet of the syringe via the feed line 36; A ring electrode 38, encapsulated and disposed near the injection needle 37, is electrically connected to the high-voltage power supply via electrode connection 35, and is used to generate an electric field between the injection needle 37 and the culture container containing the gel-forming solution. The encapsulation design of the ring electrode improves anti-interference performance while ensuring operational safety.

[0046] The electric culture dish carrier 25 is used to place culture dishes containing gel forming solution. The bottom of the electric culture dish carrier 25 is equipped with a culture dish electric carrier motor 33, which is driven by a worm gear 34 and can adjust the working distance between the culture dish and the injection needle 37. The culture dish adapter 27 is used to adapt to culture dishes of different sizes. The carrier support rod 32 is used to support the entire electric carrier structure.

[0047] The gel microsphere statistical analysis module is equipped with image processing and particle size statistics functions, which can quickly calculate the diameter distribution of gel microspheres. In some embodiments, such as Figure 3 , Figure 4 , Figure 7 , Figure 8 As shown, the gel microsphere statistical analysis module 4 includes: The imaging unit, including a surface light source, a microscope objective, and a CCD image sensor, is used to acquire high-resolution images of microgel spheres within the culture vessel; specifically, the imaging unit includes: The imaging system base frame 20 supports the entire imaging system; A surface light source 17 is positioned below the petri dish to provide uniform transmitted illumination; The X-axis stage 18 and the Y-axis stage 19 are used to drive the culture dish to move in the horizontal plane, so as to realize scanning imaging of different areas inside the culture dish. Microscope objective 22 is used to magnify images of microgel spheres; CCD image sensor 21 is used to acquire magnified images of microgel spheres; The right-angle mirror 23 is used to change the direction of the light path, making the imaging system structure more compact; The image processing unit is configured to identify, segment, and measure the acquired images, calculate the particle size distribution and coefficient of variation (CV value) of the microgel spheres, and send the statistical results to the control system.

[0048] In some embodiments, such as Figure 4 , Figure 5 As shown, the material transfer and liquid handling module includes: At least one pipetting module 9 has a liquid level detection function for quantitative transfer, addition and aspiration of gel forming solution, washing solution and cell culture medium; At least one rotating gripper module 13 is used to grip and transfer culture containers between different workstations on the automated platform.

[0049] In some embodiments, such as Figure 3 As shown, the automated platform includes a HEPA filtration system 1 and / or an ultraviolet sterilization device to provide a sterile environment for cell manipulation and three-dimensional culture.

[0050] In addition, the automated platform is also equipped with auxiliary mechanisms such as a petri dish adapter rack 2, a pipette tip carrier rack 3, a waste station 5, a circuit board rack 6, a pipetting Y-axis 7, an X-axis 8, a gripper Z-axis 11, a gripper Y-axis 12, and an electrical control system 10, which together constitute a complete automated work platform.

[0051] Combination Figures 1-11 The workflow of the system of this invention is described in detail below. First, initial setup is performed: Before the experiment begins, the operator draws the cell and hydrogel suspension into a disposable syringe and attaches the syringe to the syringe holder 24 of the monodisperse microgel preparation module 16. The gel-forming solution (such as calcium chloride solution) is added to the culture dish, and the culture dish is placed on the electric culture dish carrier 25. The size of the target microgel spheres, such as 500 μm, is input through the control system. Then, initial model construction is performed: For systems used for the first time or new hydrogel materials, the system needs to construct an initial neural network model. The specific implementation steps are as follows: Step 1: The control system generates a set of initial experimental parameter combinations (e.g., using orthogonal experimental design or uniform design) based on the preset parameter range (voltage: 1-10kV, working distance: 5-20mm, propulsion speed: 10-100mm / h), covering the main area of ​​the parameter space.

[0052] Step 2: The control system sequentially drives the monodisperse microgel preparation module 16 to prepare microgels according to the generated initial parameter combination. The specific process is as follows: the stepper motor 30 drives the syringe plunger 26 to extrude the gel suspension at a set advance speed; the high-voltage power supply generates a set high-voltage electrical signal, which forms an electric field between the injection needle 37 and the gel forming liquid through the annular electrode 38; under the action of the electric field force, the gel suspension at the needle overcomes the surface tension to form tiny droplets, which drip into the gel forming liquid and crosslink with calcium ions to form microgel spheres.

[0053] Step 3: After preparation, the material transfer and liquid handling module (rotary gripper module 13 and single-channel pipetting module 9) transfers the culture dish containing microgel spheres to the gel microsphere statistical analysis module 4. The culture dish is moved by the X-axis stage 18 and Y-axis stage 19, illumination is provided by the surface light source 17, the microscope objective 22 magnifies the image, and the CCD image sensor 21 acquires images of the microgel spheres. The image processing unit analyzes the acquired images and calculates the average diameter and CV value of the microgel spheres for each parameter combination.

[0054] Step 4: The control system takes the control parameters (voltage, working distance, propulsion speed) from Step 2 as input and the average diameter and CV value obtained in Step 3 as output to train the initial neural network model and establish the mapping relationship between the control parameters and the quality indicators.

[0055] Then, formal fabrication and closed-loop optimization are performed: After the initial model is established, the system enters the formal fabrication and closed-loop optimization process. This process includes the following steps: S1. Use the neural network trained in step 4 as the initial system dynamic model.

[0056] S2. The control system uses a trained neural network as a predictive model. Through the model predictive controller, under the condition of satisfying physical constraints, it solves for the optimal control input sequence that minimizes the error between the predicted mass index of the microgel spheres and the target mass index (e.g., diameter 500μm, CV value <5%) within a finite time domain. For example, the MPC may calculate the current optimal parameters as: voltage 3.5kV, working distance 12mm, and propulsion speed 45mm / h.

[0057] S3. The control system applies the first control input (voltage 3.5kV, working distance 12mm, propulsion speed 45mm / h) from the optimal control input sequence to the monodisperse microgel preparation module 16, preparing the cell-containing gel suspension into microgel spheres. During the preparation process, the number of microgel spheres in each spray cycle is precisely controlled by adjusting the voltage amplitude of the high-voltage power supply, thereby precisely controlling the number of microgel spheres entering each well of the multi-well plate or culture dish to ensure the consistency of the number of 3D cultured cells in each well.

[0058] S4. After preparation, the material transfer and liquid processing module transfers the culture dish to the gel microsphere statistical analysis module 4, acquires images of the microgel spheres prepared in the current batch, obtains their actual particle size and coefficient of variation through image analysis, and feeds back the actual quality index to the control system.

[0059] S5. The control system uses the control parameters of the current batch (voltage 3.5kV, working distance 12mm, propulsion speed 45mm / h) and actual quality indicators (such as diameter 498μm, CV value 5.2%) as new training data to incrementally learn the adaptive neural network and update the network parameters in real time. This step enables the neural network to continuously adapt to changes in system characteristics (such as the effects of temperature and humidity on the hydrogel, or the differences between batches of hydrogel materials).

[0060] S6. The control system repeats steps S2 to S5, forming a fully automated closed-loop control process of preparation, detection, feedback, updating, and optimization. As the number of experiments increases, the neural network model describes the system characteristics more and more accurately, the optimization effect of MPC becomes better and better, and the monodispersity of the subsequently prepared microgel spheres continues to be optimized and approaches the preset target value.

[0061] Post-processing and cultivation: After the microgel spheres are prepared, the system automatically performs the post-processing steps: The material transfer and liquid handling module transfers the culture dish containing microgel spheres to the liquid exchange area. The single-channel pipetting module 9 aspirates the gel formation liquid, adds cleaning solution (such as PBS buffer) to clean the surface, removes residual cross-linking agent and uncross-linked gel material, and then aspirates the cleaning solution again.

[0062] The single-channel pipetting module 9 adds cell culture medium (such as DMEM medium containing 10% fetal bovine serum) to the culture dish and resuspends the microgel beads in the culture medium.

[0063] The culture dish is transferred to a culture zone (such as a CO2 incubator) for three-dimensional cell culture. During culture, the system can periodically change the culture medium or add drugs according to a preset protocol to perform hepatotoxicity testing.

[0064] Finally, hepatotoxicity testing is performed: Once the cells within the microgel spheres have reached a suitable culture stage (usually 3-7 days), the system can be used for hepatotoxicity testing. According to the experimental design, the control system adds different concentrations of the test drug to different culture wells through the single-channel pipetting module 9.

[0065] Continue culturing the culture dish for the predetermined time (e.g., 24-72 hours). During this period, the system can periodically acquire images of the microgel spheres to observe cell growth status and morphological changes.

[0066] After culture, the system can perform cell viability assays (such as CCK-8 assay, Live / Dead staining, etc.) or collect microgel spheres for biochemical analysis (such as detecting hepatotoxicity markers ALT, AST, etc.). Because the microgel spheres prepared by this invention have good monodispersity and consistency, with highly consistent cell numbers and states across different wells, the toxicity test results exhibit excellent repeatability and reliability.

[0067] In summary, this invention employs an electrohydrodynamic module packaging design, which improves the system's anti-interference performance and operational safety. Real-time detection of microsphere distribution via an image processing system, combined with artificial intelligence algorithms, continuously improves the monodispersity of the prepared microspheres through the accumulation of experimental data and autonomous learning. It achieves fully automated operations throughout the entire process, including crosslinking agent addition, cleaning solution replacement, and culture medium replacement, demonstrating a high degree of automation and realizing a complete encapsulation process from microsphere preparation to hepatotoxicity testing. It can be matched with hydrogel systems of different materials and composite materials, and through model optimization and autonomous learning, it can rapidly prepare hydrogel microspheres with good monodispersity, significantly simplifying the process optimization and parameter exploration.

[0068] This invention is based on a three-dimensional cell culture system in the form of hydrogels. It combines automation and artificial intelligence algorithms to establish a self-learning monodisperse 3D hydrogel system for hepatotoxicity testing. This allows for rapid process optimization, preparation of hydrogel microspheres with good monodispersity, and provides a 3D cell culture model for hepatotoxicity testing.

[0069] Example 2 In the microgel preparation process, traditional equipment control methods often rely on manual settings or simple control algorithms, leading to problems such as high errors, low efficiency, and high human intervention. To address the complex control tasks in microgel sphere preparation systems, this invention proposes an intelligent control algorithm combining adaptive neural networks and model predictive control. This method can quickly establish a dynamic model of the system using neural networks even with limited experimental data, and optimize the control strategy using model prediction, achieving intelligent and adaptive control of the equipment. Furthermore, it can dynamically adjust control parameters based on the quality of the microgels, thereby improving the preparation quality and production efficiency. This invention models the microgel preparation system as a multiple-input multiple-output (MIMO) system. Input control parameters include electric field magnitude (voltage), nozzle height (working distance), and preparation speed (propulsion speed). Outputs include the diameter and coefficient of variation (CV) of the microgels, and optimization is performed using adaptive deep learning and model predictive control. The overall architecture is as follows: Figure 1 As shown.

[0070] like Figure 1As shown, this invention employs an intelligent control framework combining adaptive neural networks and model predictive control. During microgel preparation, an adaptive neural network first learns a dynamic model of the system based on limited experimental data, establishing a mapping relationship between control parameters (voltage, working distance, propulsion speed) and microgel quality indicators (particle size, CV value). Then, this model is used to predict future system behavior, and optimization control is achieved through a model predictive control (MPC) module. MPC selects the optimal control input based on the predicted dynamic model, minimizing control error and satisfying preset constraints. Throughout the process, a real-time feedback mechanism continuously adjusts the control strategy, dynamically optimizing the equipment's control parameters, thereby achieving efficient and adaptive control of the microgel preparation process.

[0071] Instrument control process as follows Figure 2 As shown in the diagram, in this process, the current output of the device (i.e., the actual particle size and CV value of the microgel spheres) is first measured using a sensor (such as a CCD image sensor 21), and this real-time data is input into an adaptive neural network. The neural network learns and updates the dynamic model of the system based on the experimental data. Next, the model predictive control module predicts the future system behavior based on the updated model and calculates optimized control inputs (such as electric field strength, working distance, preparation speed, etc.). The optimized control inputs are applied to the device, and the control process proceeds accordingly. Whenever the device output changes, the new measurement value is fed back to the neural network for further training and model updates, thereby improving control accuracy. The MPC adjusts the control strategy according to the latest model predictions, ensuring that the device output always approaches the desired target value while satisfying relevant constraints. Through this adaptive learning and feedback mechanism, the entire instrument control process can maintain efficient and stable performance when facing complex process requirements and dynamic environments.

[0072] A method for preparing monodisperse microgels for 3D hepatotoxicity testing is disclosed. Based on the aforementioned system, a detailed description of the system can be found in the corresponding descriptions in the system embodiments described above, and will not be repeated here. This invention employs an intelligent control algorithm combining adaptive neural networks and model predictive control. The overall architecture is as follows: Figure 1 As shown, this method can quickly establish a dynamic model of the system using neural networks when faced with limited experimental data, and use model prediction to optimize the control strategy, thereby achieving intelligent and adaptive control of the equipment. Furthermore, it can dynamically adjust the control parameters according to the quality of the microgel, thus improving the preparation quality and production efficiency of the microgel.

[0073] The microgel fabrication system is modeled as a multiple-input multiple-output (MIMO) system. Input control parameters include electric field magnitude (voltage), nozzle height (working distance), and fabrication speed (propellant speed). Outputs include the diameter and coefficient of variation (CV) of the microgel. Specifically, such as... Figure 12 As shown, the method includes the following steps: S1. Collect at least one set of control parameters and their corresponding microgel ball quality indicators from the historical preparation process, and train an adaptive neural network using the control parameters and their corresponding quality indicators to establish a mapping relationship between the control parameters and the microgel ball quality indicators; wherein, the control parameters include at least electric field parameters, working distance and propulsion speed, and the quality indicators include at least the particle size and coefficient of variation of the microgel balls; In this embodiment, a neural network model is first used to simulate the complex nonlinear relationship between the input and output in the microgel preparation system. For example... Figure 13 As shown, a deep neural network is used as the simulator, and the architecture of the adaptive neural network includes: The input layer contains neuron nodes corresponding to the control parameters; specifically, the input layer contains three neuron nodes, each corresponding to the magnitude of the electric field (…). ), Nozzle height ( ) and preparation speed ( The input variables are standardized at the input layer to ensure that their numerical range is suitable for network training.

[0074] Multiple hidden layers, employing a fully connected structure and using non-linear activation functions, are used to learn the non-linear mapping relationship between control parameters and quality indicators. Specifically, the hidden layers adopt a deep neural network structure, i.e., multiple hidden layers to learn complex mapping relationships. Each layer uses an activation function (such as ReLU) to introduce non-linearity and improve the network's expressive power. A multi-layer fully connected network is used, in which the number of neurons in each layer gradually decreases to increase the model's complexity and capture more high-order features.

[0075] The output layer contains two neuron nodes, each corresponding to the predicted microgel sphere size. ) and coefficient of variation ( value, ), which represents the stability of the microgel particle size distribution, i.e., the control objective of interest.

[0076] Using existing experimental data, a supervised learning method was employed to train a neural network capable of effectively simulating the behavior of the device. The training objective was to optimize the network weights using experimental data, enabling the model to accurately predict the microgel diameter and CV value under different control input conditions.

[0077] The mean squared error (MSE) is used as the loss function to measure the difference between the predicted result and the actual value: in, and The actual measured diameter and coefficient of variation of the microgel spheres. and The diameter and coefficient of variation predicted by the neural network. The weight hyperparameters are designed to balance diameter error and coefficient of variation error. During training, the network weights are optimized using backpropagation and gradient descent algorithms, with adjustments made based on error feedback.

[0078] S2. Using a trained adaptive neural network as a prediction model, and through a model prediction controller, under the condition of satisfying the preset physical constraints, solve for the optimal control input sequence that minimizes the error between the predicted microgel ball quality index and the target quality index in the future finite time domain. Model predictive control (MPC) is an advanced control strategy widely used in industrial control, autonomous driving, and robotics control. Combining MPC with neural networks can further improve control performance, especially in complex, nonlinear, or difficult-to-model systems.

[0079] The core advantages of MPC include online optimization, which calculates a series of control actions at each time step, rather than just a single control input, and can take future effects into account during the control process to better optimize system behavior; constraint handling, which can naturally handle constraints in the system, such as restrictions on control inputs and physical constraints on the state; and foresight, which predicts future states through system models and can take corresponding control measures in advance when the system undergoes dynamic changes or external disturbances occur.

[0080] The optimization problem is defined as obtaining the optimal control input at each control step t by optimizing the following cost function: in, For at any time The predicted output (generated by a neural network). For the desired output, To control the input, and is a weighted matrix used to adjust the error and control the input weights, and N is the prediction range, i.e., the number of future time points.

[0081] Because the system's input and output are physically limited, constraints need to be set: Control input constraints: ; System output constraints: ; in, and The physical adjustable range of electric field parameters, working distance, and propulsion speed. and The allowable range for the particle size and coefficient of variation of the target microgel spheres.

[0082] The optimal control input sequence is solved by optimization algorithms (such as sequential quadratic programming, interior point method, etc.), and the first control input in the sequence is applied to the system to update the controller input. Specifically, S3, the first control input in the optimal control input sequence is applied to the electrohydrodynamic microsphere generator to prepare microgel spheres from a gel suspension containing cells; S4. Acquire images of the microgel spheres prepared in the current batch, obtain their actual particle size and coefficient of variation through image analysis, and feed this actual quality index back to the control system. Among them, such as Figure 14 As shown, the steps for obtaining the actual particle size and coefficient of variation through image analysis include: S41. Acquire images of microgel spheres using a microscopic imaging system; S42. Use an image recognition algorithm to identify each microgel sphere in the image and calculate the diameter of each microgel sphere; S43. Statistically analyze the particle size distribution of all microgel spheres, calculate the average diameter and coefficient of variation, and use them as the actual quality indicators.

[0083] Real-time feedback is a key factor in microgel preparation systems. The system can monitor key performance indicators such as particle size and CV value of microgels in real time during the preparation process, and adjust the control input based on real-time feedback to achieve optimal production.

[0084] By integrating high-precision imaging and detection equipment (gel microsphere statistical analysis module 4), the system can continuously monitor the key properties of the microgels during the preparation process. Microgel particle size is measured in real time using instruments such as microscopes, and this information is fed back to the control system via a data transmission interface.

[0085] Once real-time feedback data is received, the neural network model immediately makes predictions and determines whether the current preparation process meets the predetermined target. If the prediction results show large deviations in particle size or CV value, the system will adjust according to the model's predictive control algorithm, modifying control parameters such as electric field magnitude, nozzle height, and preparation speed.

[0086] Based on real-time feedback, the system continuously adjusts the control input through a model predictive control algorithm to minimize the error of the microgels. For example, when the microgel particle size is found to be large, the system may reduce the particle size by decreasing the electric field strength or adjusting the nozzle spacing; when the CV value is found to be large, the system will automatically adjust the preparation speed to improve the distribution uniformity. Specifically, S5, the control parameters and actual quality indicators of the current batch are used as new training data to incrementally learn the adaptive neural network and update the network parameters in real time; S6. Repeat steps S2 to S5 to achieve closed-loop iterative control of the microgel sphere preparation process.

[0087] Specifically, during the preparation of microgel spheres, the dynamic behavior of the system is approximated by an adaptive neural network model, whose input is the control input. The output is the system's state variable. The neural network model learns the mapping relationship between the input and output through training, denoted as: in, It is a moment The predicted output, It is a mapping function learned through an adaptive neural network, where θ is the network's parameters (weights and biases). It is the system's state variable.

[0088] The network training method involves incremental learning based on real-time data and adjusting the network parameters through feedback. This process includes acquiring the input-output pairs at the current moment using experimental or simulation data. And update the training set of the neural network; calculate the error between the neural network's predictions and the actual values. The network weights are adjusted using methods such as gradient descent to minimize prediction error; the weights and biases of the neural network are adjusted based on error feedback to gradually improve the model's prediction ability; and the optimal control input is solved using MPC and applied to the system.

[0089] In some embodiments, the process also includes S7, a fully automated step: After the preparation in step S3 is completed, the culture container containing microgel spheres is automatically transferred to the medium exchange area through an automated material transfer and liquid handling device. The gel forming solution is removed, the washing solution is added and removed, and the cell culture medium is added and replaced in sequence. Then the culture container is transferred to the culture area for three-dimensional culture.

[0090] This invention employs artificial intelligence to optimize electrical parameter control, enabling rapid adaptation to hydrogel systems with different properties and achieving the preparation of monodisperse hydrogel microspheres, effectively solving the technical challenge of microsphere consistency. Specifically, this invention has the following technical advantages: 3D cell culture using microgels ensures that the culture microenvironment is similar for cells at the center and at the edge of the microsphere due to the small size of the microspheres. This prevents mass transfer issues from affecting the scientific validity of the 3D culture model and ensures the consistency of cell growth status within the same microsphere. A bioactive hydrogel system was developed using chemical grafting and immobilization methods to improve the biomimeticity of the drug screening microenvironment simulation, and a complete in-situ characterization system was provided through pore-forming technology. Artificial intelligence algorithms are used to optimize the control of electrical parameters, enabling rapid adaptation to hydrogel systems with different properties and ensuring the monodispersity and consistency of microspheres; A ring electrode encapsulation design enhances anti-interference performance while ensuring operational safety. In high-throughput experiments with multi-well plates, precise control of the number of microspheres in each well of the multi-well plate and culture dish is achieved by adjusting parameters such as the voltage amplitude and working distance of the high-voltage power supply, combined with an on-demand spray control strategy. This ensures consistency in cell count across wells and guarantees the reliability of hepatotoxicity testing.

[0091] Furthermore, this invention achieves precise control over the size of microgel spheres through multi-parameter control (working distance, propulsion rate, and electrical parameters) of the electrohydrodynamic module. The high-voltage module is encapsulated inside the electrohydrodynamic microsphere generation module, unaffected by external interference, and the distance between the module and the gel-forming liquid is unrestricted. The use of a high-voltage power supply combined with a ring electrode encapsulation mode enables on-demand spraying, precisely controlling the number of microspheres in each well of the multi-well plate, thereby ensuring the consistency of the number of cells in 3D culture. Figure 16 This study demonstrates the influence of three key parameters—voltage, working distance, and preparation speed—on the particle size and monodispersity of microgel spheres. A single-factor design was employed, fixing two parameters while varying one, measuring the peak position and diameter of the microgel spheres, and evaluating monodispersity using the coefficient of variation (CV). The results show that optimal monodispersity (CV = 3.80%) is achieved at a preparation speed of 0.01 mm / s, providing fundamental data for subsequent intelligent control algorithms.

[0092] This invention also uses artificial intelligence image algorithms combined with optical imaging to quickly statistically analyze the diameter distribution of microspheres, and then evaluates and organizes the results.

[0093] A comprehensive solution encompassing automated microsphere preparation, sample transfer to microscopy, cross-linking agent, cleaning agent, and culture medium replacement and automated culture, as well as high-throughput drug addition; such as... Figure 15 As shown, an alginate-based composite hydrogel system was prepared using a freeze-drying process combined with partially oxidized alginate and peptide grafting technology for 3D cell culture in hepatotoxicity testing. The test results showed that the composite hydrogel material had good cell viability and proliferation activity.

[0094] Figure 17This study showcases the changes in hepatocyte function indicators at 0, 5, 10, 15, and 20 days of culture using five different culture models (Gel group, ULA ultra-low adsorption plate group, HD hanging drop culture group, Scaf scaffold group, and 2D two-dimensional culture group). The results demonstrate the advantages of the hydrogel microsphere system (Gel group) of this invention in maintaining hepatocyte function, particularly compared to traditional culture models such as ULA, HD, Scaf, and 2D, providing experimental data support for the scientific validity and reliability of the 3D hepatotoxicity testing model.

[0095] Example 3 A computer device 800, such as Figure 18 As shown, the system includes a memory 810, a processor 820, and a computer program 830 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for preparing a monodisperse microgel for 3D hepatotoxicity testing. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here.

[0096] Example 4 A computer-readable storage medium, such as Figure 19 As shown, a computer program is stored thereon, which, when executed by a processor, implements the steps of a method for preparing a monodisperse microgel for 3D hepatotoxicity testing. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, which will not be repeated here.

[0097] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0098] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

[0099] The apparatus, computer device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, computer device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, computer device, and non-volatile computer storage medium will not be repeated here.

[0100] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software units implementing the method and structures within a hardware component.

[0101] The systems, apparatuses, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above apparatuses are described separately as various units based on their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0102] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0104] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0106] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0107] This specification may be described in the general context of computer-executable instructions, such as program units, that are executed by a computer. Generally, program units include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program units may reside in local and remote computer storage media, including storage devices.

[0108] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0109] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A process for the preparation of 3D hepatotoxicity test monodisperse microgels, characterized in that, Includes the following steps: S1. Collect at least one set of control parameters and their corresponding microgel ball quality indicators from the historical preparation process, and train an adaptive neural network using the control parameters and their corresponding quality indicators to establish a mapping relationship between the control parameters and the microgel ball quality indicators; wherein, the control parameters include at least electric field parameters, working distance and propulsion speed, and the quality indicators include at least the particle size and coefficient of variation of the microgel balls; S2. Using a trained adaptive neural network as a prediction model, and through a model prediction controller, under the condition of satisfying the preset physical constraints, solve for the optimal control input sequence that minimizes the error between the predicted microgel ball quality index and the target quality index in the future finite time domain. S3. Apply the first control input in the optimal control input sequence to the electrohydrodynamic microsphere generator to prepare microgel spheres from the gel suspension containing cells. S4. Acquire images of the microgel spheres prepared in the current batch, obtain their actual particle size and coefficient of variation through image analysis, and feed this actual quality index back to the control system. S5. Using the control parameters and actual quality indicators of the current batch as new training data, incrementally learn the adaptive neural network and update the network parameters in real time. S6. Repeat steps S2 to S5 to achieve closed-loop iterative control of the microgel sphere preparation process.

2. A process for the preparation of monodisperse microgels for 3D hepatotoxicity testing according to claim 1, characterized in that, The architecture of the adaptive neural network includes: The input layer contains neuron nodes corresponding to the control parameters; Multiple hidden layers are used, employing a fully connected structure and a non-linear activation function to learn the non-linear mapping relationship between control parameters and quality indicators; The output layer contains two neuron nodes, corresponding to the predicted microgel ball size and coefficient of variation, respectively.

3. A process for the preparation of monodisperse microgels for 3D hepatotoxicity testing according to claim 2, characterized in that, The training loss function of the adaptive neural network is: where, and are the actual measured microgel sphere diameters and coefficient of variation, and are the neural network predicted diameters and coefficient of variation, is the weight hyperparameter balancing the diameter error and coefficient of variation error.

4. A process for the preparation of monodisperse microgels for 3D liver toxicity testing according to claim 1, characterized in that, The objective function of the model predictive controller is: in, For at any time The predicted output, For the desired output, To control the input, and Let N be the weighted matrix, and N be the prediction range.

5. The method for preparing a monodisperse microgel for 3D hepatotoxicity testing as described in claim 4, characterized in that, The physical constraints include: Control input constraints: ; System output constraints: ; in, and The physical adjustable range of electric field parameters, working distance, and propulsion speed. and The allowable range for the particle size and coefficient of variation of the target microgel spheres.

6. The method for preparing a monodisperse microgel for 3D hepatotoxicity testing as described in claim 1, characterized in that, The electrohydrodynamic microsphere generator uses a high-voltage power supply; the method further includes: By adjusting the voltage amplitude of the high-voltage power supply, the number of microgel balls in each spraying cycle can be precisely controlled, thereby precisely controlling the number of microgel balls entering each well of the multi-well plate or culture dish to ensure the consistency of the number of 3D cultured cells in each well.

7. The method for preparing a monodisperse microgel for 3D hepatotoxicity testing as described in claim 1, characterized in that, The steps for obtaining the actual particle size and coefficient of variation through image analysis include: Images of microgel spheres were acquired using a microscopic imaging system; Image recognition algorithms are used to identify individual microgel spheres in an image and the diameter of each microgel sphere is calculated. The particle size distribution of all microgel spheres was statistically analyzed, and the average diameter and coefficient of variation were calculated as the actual quality indicators.

8. The method for preparing a monodisperse microgel for 3D hepatotoxicity testing as described in claim 1, characterized in that, It also includes S7 and fully automated process steps: After the preparation in step S3 is completed, the culture container containing microgel spheres is automatically transferred to the medium exchange area through an automated material transfer and liquid handling device. The gel forming solution is removed, the washing solution is added and removed, and the cell culture medium is added and replaced in sequence. Then the culture container is transferred to the culture area for three-dimensional culture.

9. A 3D hepatotoxicity testing monodisperse microgel preparation system, characterized in that, include: Automation platform; The monodisperse microgel preparation module, set on the automated platform, is based on the principle of electrohydrodynamics and uses adjustable electric field parameters, working distance and propulsion speed to prepare cell-containing gel suspensions into microgel spheres. The gel microsphere statistical analysis module, set on the automated platform, is used to image the prepared microgel spheres and statistically analyze their particle size distribution and coefficient of variation. The material transfer and liquid handling module is set on the automated platform and is used to automatically transfer culture containers within the automated platform and perform the addition and replacement of gel forming solution, washing solution and cell culture medium; The control system is communicatively connected to the monodisperse microgel preparation module, the gel microsphere statistical analysis module, and the material transfer and liquid handling module, and is configured to perform the method steps as described in any one of claims 1 to 8.

10. The 3D hepatotoxicity testing monodisperse microgel preparation system as described in claim 9, characterized in that, The monodisperse microgel preparation module includes: Syringe clamp, used to hold a syringe containing a gel suspension; A syringe plunger drive mechanism, connected to the syringe plunger, is used to push the syringe plunger at an adjustable advance speed to extrude the gel suspension; A high-voltage power supply is used to generate an adjustable high-voltage electrical signal, the voltage amplitude of which is adjustable. The injection needle is connected to the outlet of the syringe; A ring electrode, encapsulated and disposed near the injection needle and electrically connected to the high-voltage power supply, is used to create an electric field between the injection needle and the culture container containing the gel-forming solution.

11. The 3D hepatotoxicity testing monodisperse microgel preparation system as described in claim 9, characterized in that, The gel microsphere statistical analysis module includes: The imaging unit, including a surface light source, a microscope objective, and an image sensor, is used to acquire high-resolution images of microgel spheres inside the culture container; The image processing unit is configured to identify, segment, and measure the acquired images, calculate the particle size distribution and coefficient of variation of the microgel spheres, and send the statistical results to the control system.

12. The 3D hepatotoxicity testing monodisperse microgel preparation system as described in claim 9, characterized in that, The material transfer and liquid handling module includes: At least one pipetting module with liquid level detection function is used for quantitative transfer, addition and aspiration of gel forming solution, washing solution and cell culture medium; At least one rotating gripper module is used to grip and transfer culture containers between different workstations on the automated platform.

13. The 3D hepatotoxicity testing monodisperse microgel preparation system as described in claim 9, characterized in that, The automated platform includes a HEPA filtration system and / or ultraviolet sterilization device to provide a sterile environment for cell manipulation and three-dimensional culture.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 8.

Images