Hydrogel for injection and preparation and formula optimization method thereof

Abstract

The invention provides an injection hydrogel for vascular embolism treatment and a preparation and formula optimization method thereof. The hydrogel formula comprises two or more 2-arm or multi-arm polyethylene glycol polymers containing functional groups which can be cross-linked with each other, lithium magnesium silicate and a contrast agent. The formula optimization method comprises the following steps: designing an orthogonal experiment, systematically investigating the influence of raw material factors (such as polyethylene glycol molecular weight, functional groups, concentration and the like) on product performances of shear viscosity eta, compression strength sigma c and tensile strength sigma t, and obtaining basic experiment data; establishing a chemical reaction kinetic model and a product performance prediction model based on machine learning based on the data; determining a multi-objective optimization objective function according to product performance requirements; establishing a non-linear multi-objective equation set for formula optimization in combination with a material proportioning relationship; solving the optimization model by adopting an intelligent optimization algorithm to obtain a preliminary optimization formula; carrying out an optimization experiment under a small-scale test condition to obtain new experiment data, and updating model parameters; and the optimization iteration is repeated until convergence.

Description

[0001] This application is a divisional application of the invention patent application with the application number 202411331385.X, the application date of September 24, 2024, and the invention name of "A hydrogel for injection and its preparation and formulation optimization method". TECHNICAL FIELD

[0002] The present application belongs to the technical field of medical implants, and specifically relates to a hydrogel for vascular embolization therapy and a preparation and formulation optimization method thereof. BACKGROUND

[0003] In the process of clinical diagnosis and treatment, implantable hydrogels have wide application prospects. They can exist stably in the body for a long time as biomaterials and play special functional roles, such as hemostasis, embolization, stent reinforcement, and drug carriers. Due to high biocompatibility, high safety, and injectability, implantable hydrogels gradually replace traditional solid implants and become an important new type of medical material. In the prior art, hydrogel materials are mainly divided into two categories: natural-based materials and synthetic materials. Natural-based materials are mainly derived from animal and plant tissues or extracellular matrix proteins, such as gelatin, collagen, and hyaluronic acid. This type of material has good biocompatibility, but has problems such as purity, batch difference, and limited sources. Synthetic materials are mainly polymers, such as polyethylene glycol (PEG), polyvinyl alcohol (PVA), and polypropylene (PPE). Synthetic materials have the advantages of single composition and controllable structure, but some materials have poor biocompatibility and degradability. Common hydrogel crosslinking methods include physical crosslinking (such as electrostatic interaction, block copolymer self-assembly), chemical crosslinking (such as free radical initiation, ultraviolet light crosslinking, enzyme catalysis crosslinking), and composite crosslinking. Physical crosslinking hydrogels have injectability and self-repairing ability, but have low mechanical strength. Chemical crosslinking hydrogels have high strength, but have poor injectability and intelligent responsiveness. Composite crosslinking can combine the advantages of both, but the formulation design and reaction control are more complex. In addition, implant materials often also need to have imaging performance to facilitate accurate positioning during surgery and postoperative tracking observation. This requires the composite of contrast agents (such as iodide and barium salt) and the matrix. However, due to the particle state and super-affinity of the contrast agent itself, it often reduces the mechanical properties of the hydrogel and interferes with its rheological properties. Therefore, existing injectable hydrogels still face some challenges in actual application. First, in order to obtain sufficient viscosity and mechanical strength, it is often necessary to increase the polymer concentration, but this will greatly increase the initial viscosity of the hydrogel, making it difficult to be injected into the human body through a small injection needle. Second, the formulation optimization process of the hydrogel usually relies on a large number of trial and error experiments, which requires a lot of time and resources. Therefore, how to ensure the performance of the hydrogel while considering its injectability and establishing an efficient formulation optimization strategy has become a technical problem to be solved in this field.

[0004] That is, the existing injection hydrogel has the technical problem of difficulty in precisely controlling the strength and viscosity, and the formulation optimization process usually relies on a large number of trial and error experiments, which requires a large amount of time and resources. SUMMARY

[0005] Therefore, the present application provides a hydrogel for injection and a preparation and formulation optimization method thereof, which can solve the technical problem of the existing injection hydrogel that is difficult to precisely control the strength and viscosity, and the formulation optimization process usually relies on a large number of trial and error experiments, which requires a large amount of time and resources.

[0006] The present application is implemented as follows:

[0007] 1. A first aspect of the present application provides a hydrogel for injection, wherein a functional group-containing high molecular compound comprising polyethylene glycol amino or polyethylene glycol thiol or polyethylene glycol carboxyl or polyethylene glycol hydroxyl, and a polymer having other cross-linkable functional groups, and a lithium magnesium silicate component and a contrast agent, and a solvent.

[0008] The functional group is polyethylene glycol amide or polyethylene glycol acrylate or polyethylene glycol oxirane or polyethylene glycol isocyanate or polyethylene glycol aldehyde.

[0009] The lithium magnesium silicate component includes any one of sodium magnesium aluminum silicate, montmorillonite, zeolite, bentonite, or kaolin; the contrast agent uses any one of iodized oil, sodium iodide, barium sulfate; and the solvent is water or buffer.

[0010] The functional group is polyethylene glycol amide or polyethylene glycol acrylate or polyethylene glycol oxirane or polyethylene glycol isocyanate or polyethylene glycol aldehyde.

[0011] Optionally, the specific formulation is: 2-arm polyethylene glycol amino: 0.1-99.9% w / v, 2-arm polyethylene glycol amide: 0.1-99.9% w / v, sodium magnesium aluminum silicate: 0.1-99.9% w / v, iodized oil: 0.1-99.9% v / v. Optionally, the specific formulation is: 4-arm polyethylene glycol thiol: 0.1-99.9% w / v, 4-arm polyethylene glycol acrylate: 0.1-99.9% w / v, montmorillonite: 0.1-99.9% w / v, barium sulfate: 0.1-99.9% w / w. Optionally, the specific formulation is: 4-arm polyethylene glycol amino: 0.1-99.9% w / v, 4-arm polyethylene glycol isocyanate: 0.1-99.9% w / v, zeolite: 0.1-99.9% w / v, sodium iodide: 0.1-99.9 mol / L.

[0012] Optionally, the specific formulation is as follows: 4-arm polyethylene glycol thiol: 0.1–99.9% w / v, 4-arm polyethylene glycol ethylene oxide: 0.1–99.9% w / v, bentonite: 0.1–99.9% w / v, and iodized oil: 0.1–99.9% v / v. A second aspect of the present invention provides a method for preparing an injectable hydrogel, comprising the following steps:

[0013] Step 1: Prepare a polymer solution of polyethylene glycol amino, polyethylene glycol mercapto, polyethylene glycol carboxyl, or polyethylene glycol hydroxyl groups and functional groups. Specifically, dissolve polyethylene glycol amino or polyethylene glycol mercapto with an equivalent amount of polyethylene glycol amide, polyethylene glycol acrylate, polyethylene glycol ethylene oxide, polyethylene glycol isocyanate, or polyethylene glycol aldehyde in a solvent and stir to mix thoroughly.

[0014] Step 2: Prepare a magnesium lithium silicate suspension by mixing a certain amount of magnesium lithium silicate components with deionized water and stirring for a certain period of time to form a uniform nano magnesium lithium silicate suspension.

[0015] Step 3: Mix the magnesium lithium silicate suspension obtained in Step 2 with the two polymer solutions in Step 1 separately, and then mix with the polymer mixture containing magnesium lithium silicate. Alternatively, mix the two mixtures in Step 1 first and then mix with the magnesium lithium silicate suspension in Step 2. Add an appropriate amount of medical contrast agent and stir until uniform to obtain the desired high-viscosity and high-strength hydrogel.

[0016] Step 4: Perform physicochemical property tests on the hydrogel prepared in Step 3, including rheological property tests and mechanical strength tests, to ensure that it meets the requirements for use.

[0017] A third aspect of the present invention provides a method for optimizing the formulation of an injectable hydrogel, comprising the following steps:

[0018] S10. Design basic experiments to test the effects of different raw material components on product performance and obtain basic experimental data.

[0019] S20. Establish chemical reaction models and product performance prediction models based on basic experimental data;

[0020] S30. Determine the multi-objective optimization objective function based on product performance requirements;

[0021] S40. Establish a multi-objective optimization model for the formulation by combining material metering relationships;

[0022] S50. The multi-objective optimization model is solved using an intelligent optimization algorithm to obtain the preliminary optimized formulation component ratios;

[0023] S60. Use the preliminary optimized formula to conduct optimization experiments and obtain optimization experimental data;

[0024] S70. Update chemical reaction models and product performance prediction models using optimized experimental data;

[0025] S80. Reconstruct the multi-objective optimization model and solve it again using the intelligent optimization algorithm to obtain a new optimal formulation component ratio;

[0026] S90. Repeat steps S60 to S80 until the formula converges to the optimal solution.

[0027] Furthermore, the basic experiment specifically employs an orthogonal experimental design to investigate the effects of polyethylene glycol molecular weight, functional group type, polymer concentration, crosslinking agent dosage, magnesium lithium silicate type and dosage, and contrast agent type and dosage on the rheological properties during injection and the mechanical strength after curing.

[0028] Furthermore, the multi-objective optimization model for the formulation is a set of nonlinear equations with polyethylene glycol molecular weight, functional group type, polymer concentration, crosslinking agent dosage, magnesium lithium silicate type and dosage, and contrast agent type and dosage as independent variables, and minimum shear viscosity, maximum compressive strength, and maximum tensile strength at injection as dependent variables.

[0029] Specifically, the steps of the formulation optimization method include:

[0030] S10. Design basic experiments to test the effects of different raw material components on product performance and obtain basic experimental data. Specifically, the basic experiments are orthogonal experimental designs to investigate the effects of polyethylene glycol molecular weight, functional group type, polymer concentration, crosslinking agent dosage, magnesium lithium silicate type and dosage, and contrast agent type and dosage on the rheological properties during injection and the mechanical strength after curing.

[0031] S20. Establish a chemical reaction model and a product performance prediction model based on basic experimental data. Specifically, the chemical reaction model describes the polymer crosslinking reaction process based on kinetic equations, and the product performance prediction model is specifically based on nonlinear mapping modeling of experimental data using machine learning algorithms.

[0032] S30. Based on product performance requirements, determine a multi-objective optimization objective function, where multiple objectives include minimizing shear viscosity during injection, maximizing compressive strength, and maximizing tensile strength.

[0033] S40. Based on the material metering relationship, establish a multi-objective optimization model for the formulation. Specifically, it is a set of nonlinear equations with the molecular weight of polyethylene glycol, the type of functional group, the polymer concentration, the amount of crosslinking agent, the type and amount of lithium magnesium silicate, and the type and amount of contrast agent as independent variables, and the minimum shear viscosity, the maximum compressive strength, and the maximum tensile strength at the injection point as dependent variables.

[0034] S50. The multi-objective optimization model is solved using an intelligent optimization algorithm to obtain the preliminary optimized formulation component ratio.

[0035] S60. Conduct optimization experiments using the preliminary optimized formula to obtain optimization experimental data. Specifically, the optimization experiments involve preparing samples under small-scale production conditions and testing rheological properties, curing speed, and mechanical properties. S70. Update the chemical reaction model and product performance prediction model using the optimization experimental data.

[0036] S80. Reconstruct the multi-objective optimization model and solve it again using the intelligent optimization algorithm to obtain a new optimal formulation component ratio.

[0037] S90. Repeat steps S60 to S80 until the formula converges to the optimal solution.

[0038] Step S10 specifically includes:

[0039] Step S101: Determine the factors to be investigated and their different levels. The factors include polyethylene glycol molecular weight, functional group type, polymer concentration, crosslinking agent dosage, lithium magnesium silicate type and dosage, and contrast agent type and dosage.

[0040] Step S102: Select an appropriate orthogonal table based on the number of factors and the number of their respective levels. The orthogonal table is used to arrange the combination relationship of the levels of each factor.

[0041] Step S103: Arrange the formulation experimental groups according to the orthogonal table, and prepare at least three parallel samples for each experimental group;

[0042] Step S104: For all samples prepared in step S103, test their rheological properties and mechanical properties after curing.

[0043] Step S105: Process all experimental data using statistical analysis methods to identify the main factors that have the greatest impact on product performance;

[0044] Step S106: Obtain preliminary optimization directions, such as increasing polymer concentration or increasing functional group density.

[0045] Specifically, step S20 includes:

[0046] Step S201: Construct a chemical reaction kinetic model, wherein the chemical reaction kinetic model uses reaction kinetic equations to describe the crosslinking reaction process;

[0047] Step S202: Construct a product performance prediction model, wherein the product performance prediction model uses a machine learning algorithm to perform nonlinear mapping modeling on the experimental data;

[0048] Optionally, an objective function can be used to combine and optimize different desired performance indicators; and the material metering relationship can be introduced as a constraint into the multi-objective optimization model of the formulation.

[0049] In step S50, a particle swarm optimization algorithm is used to solve the multi-objective optimization model of the formula. In the particle swarm optimization algorithm, the position and velocity of each individual in the population are initialized, and the individual and global optimal positions are updated based on the fitness value of each individual.

[0050] In step S50, a genetic algorithm is used to solve the multi-objective optimization model of the formula; in the genetic algorithm, the population individuals are optimized through selection, crossover, and mutation operations.

[0051] Step S70 includes adding optimized experimental data to the original dataset to expand the amount of training data for the chemical reaction model and the product performance prediction model.

[0052] Step S80 includes adjusting the weight coefficients of each index in the multi-objective optimization objective function. Steps S60 to S80 are repeated until the formulation converges to the optimal solution. The criterion for convergence is that the optimal solution is considered to have converged to the globally optimal formulation when it shows no significant change over several generations.

[0053] Compared with existing technologies, the beneficial effects of the hydrogel for injection and its preparation and formulation optimization methods provided by this invention are:

[0054] 1. A mild chemical crosslinking reaction between polyethylene glycol amino groups, polyethylene glycol mercapto groups, polyethylene glycol carboxyl groups, or polyethylene glycol hydroxyl groups and polyethylene glycol amides, polyethylene glycol acrylates, polyethylene glycol ethylene oxide, polyethylene glycol isocyanates, or polyethylene glycol aldehydes can achieve in-situ solidification of hydrogels under physiological conditions without the need for high temperatures or initiators, thus improving biocompatibility. Furthermore, by adjusting the polymer molecular weight and functional group concentration, the strength and viscosity of the hydrogel can be precisely controlled.

[0055] 4. A chemical reaction kinetic model and a product performance prediction model based on machine learning were established, and a multi-objective optimization algorithm was used for intelligent optimization of the formulation. This method can significantly improve the efficiency and accuracy of formulation development, reduce a large number of redundant experiments, and lay the foundation for the industrial application of hydrogel materials.

[0056] In summary, the solution of this invention solves the technical problems of existing injectable hydrogels, such as the difficulty in precisely controlling strength and viscosity, and the fact that the formulation optimization process usually relies on a large number of trial-and-error experiments, which requires a lot of time and resources. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 A flowchart of the method provided by the present invention;

[0059] Figure 2 The shear viscosity curve of the hydrogel in experimental group 1;

[0060] Figure 3 The shear viscosity curve of the hydrogel in experimental group 2;

[0061] Figure 4 The shear viscosity curve of the hydrogel in experimental group 3;

[0062] Figure 5 The shear viscosity curve of the hydrogel in experimental group 4;

[0063] Figure 6 The shear viscosity curves of the hydrogel in experimental group 5 are shown. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. A first aspect of the present invention provides a hydrogel for injection, the formulation of which includes a polymer of polyethylene glycol amino, polyethylene glycol mercapto, polyethylene glycol carboxyl, or polyethylene glycol hydroxyl groups and functional groups, a magnesium lithium silicate component, a contrast agent, and a solvent. The magnesium lithium silicate component includes any one of sodium magnesium aluminum silicate, montmorillonite, zeolite, bentonite, or kaolin; the contrast agent is any one of iodized oil, sodium iodide, or barium sulfate; and the solvent is water or a buffer solution. The functional groups are polyethylene glycol amide, polyethylene glycol acrylate, polyethylene glycol ethylene oxide, polyethylene glycol isocyanate, or polyethylene glycol aldehyde.

[0065] Optionally, the functional groups should meet the following conditions:

[0066] (1) It can form covalent bonds or hydrogen bonds with polyethylene glycol to construct a three-dimensional network structure of hydrogel;

[0067] (2) It has good biocompatibility and will not cause toxicity or allergic reactions in the human body;

[0068] (3) It can provide specific functions, such as enhancing cell adhesion, promoting tissue regeneration, and controlling drug release; other optional functional groups include:

[0069] Hydroxyl group (-OH): such as polyethylene glycol hydroxyl group, polymethyl methacrylate hydroxyl ester;

[0070] Amino groups (-NH2): such as polyethylene glycol amino groups and polylysine;

[0071] Carboxyl group (-COOH): such as polyethylene glycol carboxylic acid, polyacrylic acid;

[0072] Sulfur groups (-SH): such as polyethylene glycol mercapto groups;

[0073] Alkenyl (-C=C-): such as polyethylene glycol acrylate;

[0074] Phosphates: such as phosphorylated polysaccharides;

[0075] Carbohydrates: such as chitosan and sodium hyaluronate;

[0076] Peptides: such as the Arg-Gly-Asp(RGD) peptide sequence;

[0077] Antibodies or receptor ligands: such as antibodies, growth factors, and cell adhesion proteins;

[0078] In summary, as long as the aforementioned conditions are met, all of the functional groups described above can serve as functional groups of polyethylene glycol, forming covalent or non-covalent bonds to construct hydrogel materials with specific biological functions. Among them, polar functional groups such as hydroxyl, amino, and carboxyl groups are more common, while sugars, peptides, and receptor ligands can endow the gel with biological activity.

[0079] Alternatively, lithium magnesium silicate can be made from: montmorillonite, kaolinite, bentonite, zeolite, Laponite, Saponite, Vermiculite, Synthetic fluorinated silicate, magnesium oxide (MgO), silicon dioxide (SiO2), hydrated magnesium silicate (Mg3Si4O10(OH)2), or hydrated aluminosilicate.

[0080] Contrast agents that can be used include: iodine-based contrast agents, Gadolinium-based contrast agents, platinum-based contrast agents, and other organic or grade 5 contrast agents. Iodine-based contrast agents include iohexol, iopamidol, iomeprol, and iopromide; Gadolinium-based contrast agents include gadoteratemeglumine, gadopentetate dimeglumine, and gadoteridol; platinum-based contrast agents include cyclic platinum complexes (Cisplatin) and platinum(IV) hydroxide (Oxaliplatin); nonpolar contrast agents can be iron sulfide nanoparticles (Ferumoxytol) or magnetic iron oxide nanoparticles (Superparamagnetic iron oxide). Organic contrast agents typically consist of small organic molecules or complexes containing heavy metals (such as iodine, gadolinium, and platinum) and are used in imaging techniques such as X-ray CT, MRI, and PET. Inorganic contrast agents, on the other hand, are mostly metal oxide / sulfide nanoparticles and are primarily used in magnetic resonance imaging (MRI) and ultrasound imaging. Both types of contrast agents have wide applications in the field of medical imaging.

[0081] The effect of this formulation is that clinical applications require high-viscosity and high-strength hydrogels, such as for vascular embolization. Therefore, it's necessary to increase the molecular weight and concentration of each component to achieve the desired viscosity and mechanical strength. However, high-viscosity and high-strength hydrogels are difficult, if not impossible, to inject through a syringe via a catheter. Introducing a viscosity thixotropic agent, such as lithium magnesium silicate, can reduce the concentration of each component, allowing the formed hydrogel to not only achieve the desired viscosity and mechanical strength but also to be easily injected manually without a power source. In this formulation, the crosslinking precursors are polyethylene glycol amino, polyethylene glycol mercapto, polyethylene glycol carboxyl, or polyethylene glycol hydroxyl, and polyethylene glycol amide, polyethylene glycol acrylate, polyethylene glycol ethylene oxide, polyethylene glycol isocyanate, or polyethylene glycol aldehyde. When mixed, a radical-catalyzed addition reaction occurs—for example, the mercapto group acts as a nucleophile, attacking the carbon-carbon double bond of the acrylate, thereby forming a covalent crosslink. This reaction mechanism differs from the traditional free radical-initiated acrylate crosslinking reaction and can proceed under mild conditions. Besides acrylates, ketone, aldehyde, and isocyanate groups can also undergo addition reactions with thiol groups to form stable thioether bonds. Therefore, by appropriately modifying the functional groups of the polymer, various types of crosslinking can be achieved, thereby precisely controlling the properties of the hydrogel.

[0082] Optionally, the specific formulation is as follows: 2-arm polyethylene glycol amino: 0.1–99.9% w / v, 2-arm polyethylene glycol amide: 0.1–99.9% w / v, sodium magnesium aluminum silicate: 0.1–99.9% w / v, iodized oil: 0.1–99.9% v / v. Optionally, the specific formulation is as follows: 4-arm polyethylene glycol mercapto: 0.1–99.9% w / v, 4-arm polyethylene glycol acrylate: 0.1–99.9% w / v, montmorillonite: 0.1–99.9% w / v, barium sulfate: 0.1–99.9% w / v. Optional, specific formulations include: 4-arm polyethylene glycol amino: 0.1–99.9% w / v, 4-arm polyethylene glycol isocyanate: 0.1–99.9% w / v, zeolite: 0.1–99.9% w / v, and sodium iodide: 0.1–99.9 mol / L.

[0083] Optionally, the specific formulation is as follows: 4-arm polyethylene glycol thiol: 0.1–99.9% w / v, 4-arm polyethylene glycol ethylene oxide: 0.1–99.9% w / v, bentonite: 0.1–99.9% w / v, and iodized oil: 0.1–99.9% v / v. A second aspect of the present invention provides a method for preparing an injectable hydrogel, comprising the following steps:

[0084] Step 1: Prepare a polymer solution of polyethylene glycol amino, polyethylene glycol mercapto, polyethylene glycol carboxyl, or polyethylene glycol hydroxyl groups with functional groups. Specifically, dissolve polyethylene glycol amino, polyethylene glycol mercapto, polyethylene glycol carboxyl, or polyethylene glycol hydroxyl groups with an equivalent amount of polyethylene glycol amide, polyethylene glycol acrylate, polyethylene glycol ethylene oxide, polyethylene glycol isocyanate, or polyethylene glycol aldehyde in a solvent and stir to mix thoroughly.

[0085] Step 2: Prepare a magnesium lithium silicate suspension by mixing a certain amount of magnesium lithium silicate components with deionized water and stirring for a certain period of time to form a uniform nano magnesium lithium silicate suspension.

[0086] Step 3: Mix the magnesium lithium silicate suspension obtained in Step 2 with the two polymer solutions in Step 1 separately, and then mix with the polymer mixture containing magnesium lithium silicate. Alternatively, mix the two mixtures in Step 1 first and then mix with the magnesium lithium silicate suspension in Step 2. Add an appropriate amount of medical contrast agent and stir until uniform to obtain the desired high-viscosity and high-strength hydrogel.

[0087] Step 4: Perform physicochemical property tests on the hydrogel prepared in Step 3, including rheological property tests and mechanical strength tests, to ensure that it meets the requirements for use.

[0088] Specifically, the implementation method of step 1 is as follows: First, select an appropriate solvent according to the type and molecular weight of the polymer. Common solvents include water, phosphate buffered saline (PBS), ethanol, and ethyl acetate. Add the polymer to be dissolved (such as polyethylene glycol amino, polyethylene glycol mercapto, polyethylene glycol carboxyl, or polyethylene glycol hydroxyl) and the functional group polymer (such as polyethylene glycol amide, polyethylene glycol acrylate, polyethylene glycol ethylene oxide, polyethylene glycol isocyanate, or polyethylene glycol aldehyde) to the solvent according to the theoretically calculated equivalent ratio, and stir to dissolve. Heating to a certain temperature during dissolution can accelerate the dissolution rate, but the temperature should not be too high to avoid polymer degradation. The solution should be completely transparent and homogeneous, without visible particles or precipitates. When the solution contains more than two polymers, the order of addition must be controlled. First, add the lower concentration component to the solvent to dissolve, and then gradually add the higher concentration component to avoid flocculation or precipitation. The stirring equipment should have an adjustable speed to achieve vigorous stirring or slow vortexing, and a constant temperature water bath should be provided to maintain a certain temperature. The amount of solvent used depends on the total mass concentration of the polymer, and is usually between a solid-liquid mass ratio of 1:5 and 1:50. The purpose of this process is to uniformly disperse the components and prepare them for the next reaction step.

[0089] The specific implementation method of step 2 is as follows: First, lithium magnesium silicate particles need to be dispersed in a polar solvent such as deionized water. The mass ratio of lithium magnesium silicate to water is usually controlled between 1:5 and 1:10. The dispersion process can be assisted by a mixer, stirrer, or ultrasonic crusher, so that the lithium magnesium silicate particles are mechanically sheared and broken into nanoscale particles, forming a high-concentration and stable suspension. The stirring speed, filler load, grinding time, and the material and diameter of the grinding balls all affect the grinding effect, and the optimal process parameters need to be determined through orthogonal experiments. During the stirring process, lithium magnesium silicate dissociates under mechanical action and slowly re-aggregates, forming a parallel layered nanosheet structure. The suspension can be further diluted with distilled water to the required concentration. The higher the concentration of the suspension, the greater the bulk viscosity and the stronger the rheological properties. The purpose of this step is to prepare a nano-lithium magnesium silicate suspension with ideal rheological properties.

[0090] The specific implementation method of step 3 is as follows: The polymer solution obtained in step 1 and the lithium magnesium silicate suspension obtained in step 2 are mixed according to the mass ratio or volume ratio specified in the formula, and the mixing order can be controlled. Then, the pre-prepared contrast agent solution or suspension is added, and the mixture is stirred thoroughly to ensure uniform mixing of all components, thus obtaining the hydrogel precursor solution. The stirring method can be mechanical stirring or magnetic stirring, and the stirring intensity and time need to be determined according to the actual situation. If obvious phase separation or gel lumps occur during the mixing process, the stirring intensity should be increased and the temperature should be appropriately raised. The precursor solution should be a homogeneous viscous liquid without obvious particles or precipitates. The purpose of this step is to fully mix all components, adjust the initial rheological properties of the hydrogel, and prepare for the next crosslinking reaction.

[0091] The specific implementation of step 4 is as follows: The precursor solution obtained in step 3 is transferred to the desired site using a syringe, catheter, or other container, keeping it away from the air. At this time, the polymer molecules are in a stretched state in the solvent, exhibiting good fluidity. Once the precursor solution is injected or removed, the solvent gradually evaporates, the molecular chains shorten, and a cross-linking reaction occurs. For example, a nucleophilic addition reaction occurs between polyethylene glycol amino or polyethylene glycol thiol molecules and polyethylene glycol amide or polyethylene glycol acrylate molecules, forming stable amide or thioether bonds, thus achieving cross-linking of the polymer network. The reaction proceeds spontaneously at room temperature without the need for additional heating or initiators. The reaction rate can be controlled by changing the concentrations of the two polymers and the concentration of functional groups. A high cross-linking density increases the strength of the hydrogel but decreases the viscosity. In addition, lithium magnesium silicate undergoes an intercalation reaction with water, altering the rheological behavior of the hydrogel. The purpose of this step is to achieve a transformation of the hydrogel from a low-viscosity, high-fluidity dynamic to a high-viscosity, low-fluidity dynamic through chemical reaction, meeting the needs of clinical use.

[0092] The specific implementation of step 5 is as follows: Based on the known composition and ratio of the hydrogel, it is necessary to further test whether its physicochemical properties meet the usage requirements. First, the shear viscosity of the hydrogel at different shear rates can be tested using a rotational rheometer or a cone-plate rheometer to evaluate its rheological properties. A typical shear viscosity curve should exhibit shear thinning characteristics, that is, the viscosity gradually decreases as the shear rate increases. The rheological parameters can be obtained by fitting the data using a power-law model or a Herschel-Bulkley model. Second, the compressive strength of the hydrogel is tested using a pressure testing instrument, and its tensile strength and elastic modulus are tested using a universal testing machine. These parameters must meet the standards for clinical use. In addition, other performance evaluations such as biocompatibility, degradation behavior, hydration rate, and porosity can be performed as needed. Only when all indicators meet the requirements can the formulation be determined to be suitable for subsequent process research.

[0093] like Figure 1 As shown, a third aspect of the present invention provides a method for optimizing the formulation of an injectable hydrogel, comprising the following steps:

[0094] S10. Design basic experiments to test the effects of different raw material components on product performance and obtain basic experimental data. Specifically, the basic experiments are orthogonal experimental designs to investigate the effects of raw material factors such as polyethylene glycol molecular weight, functional group type, polymer concentration, crosslinking agent dosage, type and dosage of lithium magnesium silicate, and type and dosage of contrast agent on the rheological properties during injection (e.g., shear viscosity) and the mechanical strength after curing (e.g., compressive strength, tensile strength).

[0095] S20. Establish a chemical reaction model and a product performance prediction model based on basic experimental data. The chemical reaction model is specifically based on kinetic equations to describe the polymer crosslinking reaction process, and the product performance prediction model is specifically based on machine learning algorithms (such as artificial neural networks, support vector machines, etc.) to perform nonlinear mapping modeling on the experimental data.

[0096] S30. Based on product performance requirements, determine a multi-objective optimization objective function, where multiple objectives include minimizing shear viscosity during injection, maximizing compressive strength, and maximizing tensile strength.

[0097] S40. Based on the material metering relationship, establish a multi-objective optimization model for the formulation. Specifically, it is a set of nonlinear equations with the molecular weight of polyethylene glycol, the type of functional group, the polymer concentration, the amount of crosslinking agent, the type and amount of lithium magnesium silicate, and the type and amount of contrast agent as independent variables, and the minimum shear viscosity, the maximum compressive strength, and the maximum tensile strength at the injection point as dependent variables.

[0098] S50. Intelligent optimization algorithms (such as particle swarm optimization, simulated annealing, etc.) are used to solve the multi-objective optimization model to obtain the preliminary optimized formulation component ratio.

[0099] S60. Conduct optimization experiments using the preliminary optimized formula to obtain optimization experimental data. Specifically, the optimization experiments involve preparing samples under small-scale production conditions and testing rheological properties, curing speed, and mechanical properties. S70. Update the chemical reaction model and product performance prediction model using the optimization experimental data.

[0100] S80. Reconstruct the multi-objective optimization model and solve it again using the intelligent optimization algorithm to obtain a new optimal formulation component ratio.

[0101] S90. Repeat steps S60 to S80 until the formula converges to the optimal solution.

[0102] The specific implementation methods of the above steps are described in detail below:

[0103] Step S10: In basic experiments testing the impact of different factors on product performance, orthogonal experimental design is an effective method. It can screen out the main influencing factors while reducing the number of experiments. For example, polymer molecular weight can be selected at three levels: 3000, 6000, and 12000; functional group types can be selected at three levels: amino, thiol, and amide; and polymer concentration can be selected at three levels: 5%, 10%, and 20%. Then, according to the orthogonal array, the combination relationship of each factor level is reasonably arranged to construct experimental groups with different formulations. Each formulation should be repeated at least three times in parallel to eliminate random errors. The preparation steps are similar to steps 4 and 5 above. After all experimental group samples are prepared, various performance indicators are tested. The specific operation is as follows:

[0104] 1) Determine the factors and levels to be considered. The main factors to be considered are the molecular weight of polyethylene glycol, the type of functional groups, the polymer concentration, the amount of crosslinking agent, the type and amount of lithium magnesium silicate, and the type and amount of contrast agent. Each factor should be set with 3-5 different levels.

[0105] 2) Select an appropriate orthogonal array based on the number of factors and levels, such as L9(3^4), L16(4^5), etc.

[0106] 3) Arrange the experiment according to the orthogonal array, and prepare 3 parallel samples for each experimental point. The preparation method is the same as in steps 1-5.

[0107] 4) Test the key indicators such as rheological properties (shear viscosity, etc.) and mechanical strength (compressive strength, tensile strength, etc.) of each sample after curing.

[0108] 5) Use mathematical statistics methods (such as analysis of variance) to analyze the primary and secondary effects of factors and screen out the main influencing factors.

[0109] 6) Obtain preliminary optimization directions, such as increasing polymer concentration or increasing functional group density.

[0110] Step S20: Based on basic experimental data, construct chemical reaction models and product performance prediction models.

[0111] Chemical reaction models can employ reaction kinetic equations, for example:

[0112]

[0113] In the formula, c is the crosslinking density, t is time, [A] and [B] are the reactant concentrations, and k, m, and n are the reaction order and rate constant, respectively.

[0114] The cross-linking reaction process can be simulated by using numerical integration or analytical solutions, combined with reaction kinetic equations and material equilibrium equations.

[0115] Product performance prediction models can employ machine learning algorithms such as Artificial Neural Networks (ANNs) and Support Vector Machines (SVMs). Taking shear viscosity η and compressive strength σ as examples:

[0116] 1) Use the raw material ratio (polymer concentration, crosslinking agent dosage, etc.) as the input variable x.

[0117] 2) The experimentally measured η and σ are used as output variables y.

[0118] 3) Construct the training dataset (x, y).

[0119] 4) Select a suitable network structure (such as a 3-layer BP network) and optimization algorithm (such as gradient descent), and set the training parameters.

[0120] 5) Train an ANN or SVM model on the training dataset to obtain the prediction functions of η(x) and σ(x).

[0121] 6) Validate the model accuracy on the retained test dataset, and retrain if necessary.

[0122] Step S30: Determine the multi-objective optimization objective function, which can take the following form:

[0123]

[0124] Where η * σ * ,∈ * These represent the target values ​​for performance indicators such as shear viscosity, compressive strength, and tensile strain.

[0125] Equation (2) achieves multi-objective optimization, and the weights of each indicator can be adjusted according to actual needs.

[0126] Step S40: Establish a multi-objective optimization model for the formulation by combining the material metering relationship.

[0127] Let the independent variable x be (polyethylene glycol molecular weight M, functional group type f, polymer concentration c, crosslinking agent amount r, lithium magnesium silicate type s and amount h, contrast agent type d and amount v).

[0128] The dependent variable y consists of performance indices such as η, σ, and ε, which are given by the prediction model in step S20.

[0129] y=f(x) (3)

[0130] Substituting (2) into (3) yields the nonlinear optimization model:

[0131]

[0132] stΣx i =1 (Material measurement relationship)

[0133] x lb ≤x≤x ub (Variable Boundaries)

[0134] In the formula x lb x ub These are the lower and upper limits for each variable.

[0135] Step S50: Use an intelligent optimization algorithm to solve the multi-objective optimization model established in S40 to obtain a preliminary optimized formula.

[0136] Commonly used intelligent optimization algorithms include:

[0137] 1) Particle Swarm Optimization (PSO) guides particle motion using information on the optimal position of the swarm and the optimal position of the individual particles;

[0138] 2) Simulated Annealing (SA) algorithm, which gradually approaches the global optimum through a controlled "cooling" process;

[0139] 3) Genetic Algorithm (GA): It simulates the biological evolution process and optimizes the population through operations such as selection, crossover, and mutation.

[0140] Taking the PSO algorithm as an example, the specific process is as follows:

[0141] 1) Initialize the position and velocity of the population individuals (feasible solutions).

[0142] 2) Assess the fitness of each individual and calculate the objective function value F(x) according to equation (2).

[0143] 3) Update the historical best position for each individual, as well as the global best position.

[0144] 4) Update the individual's speed and position according to the following formula:

[0145] v = wv + c1r1(px) + c2r2(gx)

[0146] x = x + v

[0147] Where w is the inertia weight, c1 and c2 are acceleration constants, r1 and r2 are 0-1 uniform random numbers, p is the individual's historical best position, and g is the global best position.

[0148] 5) Repeat steps 2-4 until the stopping criteria are met (e.g., the maximum number of iterations or the objective function value tends to remain constant). Based on the optimization results of algorithms such as PSO, the preliminary optimal formulation component ratios can be obtained.

[0149] Step S60: Use the preliminary optimized formula to conduct optimization experiments and obtain optimization experimental data.

[0150] 1) Prepare small-scale samples according to the methods in steps 1-5, and test their rheological and mechanical properties.

[0151] 2) Compare the optimized experimental data with the model predictions and calculate the prediction bias.

[0152] 3) Analyze the shortcomings of the model by combining experimental phenomena and test data.

[0153] Step S70: Update the chemical reaction model and product performance prediction model using optimized experimental data.

[0154] 1) Add the optimized experimental data to the original dataset to increase the amount of training data for the model.

[0155] 2) Based on the optimized experimental phenomena, modify the model form (such as reaction order, network structure, etc.).

[0156] 3) Retrain the model to obtain more accurate dynamic equations and performance prediction functions.

[0157] Step S80: Reconstruct the multi-objective optimization model and solve it again using the intelligent optimization algorithm.

[0158] 1) Substitute the updated performance prediction model into equation (3).

[0159] 2) The weight coefficients of each item in the objective function (2) can be adjusted appropriately.

[0160] 3) Use the algorithm described in S50 (such as PSO) to optimize and solve the new model.

[0161] 4) Obtain the optimal proportions of the new formulation components.

[0162] Step S90: Repeat steps S60 to S80 until the formula converges to the optimal solution.

[0163] 1) Conduct small-scale trials with the new optimal formula to obtain experimental data.

[0164] 2) Update the model and rebuild the optimization model to find the next generation of optimal solutions.

[0165] 3) When the optimal solution does not change significantly for several generations, it is considered to have converged, and the global optimal formula is obtained.

[0166] The key to this formulation optimization method is establishing accurate reaction kinetics and product performance prediction models, and employing a reasonable multi-objective optimization strategy. This involves iteratively alternating between experimental and simulation calculations, allowing the model and experimental data to mutually correct and continuously improve, ultimately converging to the optimal formulation that meets all performance requirements. This method can significantly improve formulation optimization efficiency, reduce the number of experiments, and shorten the development cycle.

[0167] Specifically, the principle of this invention is that the key lies in the rational design of the crosslinking system and the use of intelligent optimization algorithms for multi-objective formulation optimization.

[0168] First, chemical crosslinking is achieved by selecting polyethylene glycol derivatives (such as PEG-NH2 and PEG-SH) and crosslinking agents containing appropriate functional groups (such as amide groups and acrylates), enabling in-situ curing of the hydrogel under mild conditions. Compared with traditional free radical-initiated acrylate crosslinking, the nucleophilic addition mechanism and reaction conditions employed in this invention are milder. For example, PEG-SH molecules, acting as nucleophiles, can attack the carbon-carbon double bonds of PEG-acrylate molecules to form stable thioether bonds, thereby achieving crosslinking of the polymer network. Furthermore, ketone and aldehyde functional groups can also undergo similar addition reactions with thiol groups. Therefore, by designing the functional groups of the polymer, various different types of crosslinking mechanisms can be achieved, thereby precisely controlling the properties of the hydrogel.

[0169] Secondly, introducing inorganic fillers with good rheological properties, such as diatomaceous earth (e.g., montmorillonite, kaolinite), into the hydrogel formulation can effectively reduce the polymer concentration while improving the shear-thinning properties of the hydrogel. Diatomaceous earth forms a stable, dispersed nanosheet structure in water and undergoes intercalation reactions with water, giving the hydrogel excellent shear-thinning characteristics. During injection, the high shear stress causes a significant decrease in viscosity, which facilitates injection through a fine syringe. Once the hydrogel is expelled from the syringe and enters the target site, the shear stress rapidly decreases, and the viscosity returns to a higher level, ensuring successful solidification and the provision of the required mechanical support.

[0170] Furthermore, to impart excellent visualization imaging effects to the hydrogel, this invention incorporates medical contrast agents, such as iodized oil and barium sulfate, into the formulation. These inorganic contrast agents can be uniformly dispersed within the hydrogel matrix, producing significant contrast enhancement during X-ray, CT, and other imaging examinations, providing crucial information for positioning guidance in minimally invasive clinical surgery.

[0171] Finally, for several key performance indicators of hydrogels (such as shear viscosity, compressive strength, and tensile strength), this invention establishes a mathematical model based on reaction kinetics and machine learning, and employs intelligent optimization algorithms (such as particle swarm optimization and simulated annealing algorithms) to perform multi-objective optimization of the formulation. This method not only significantly improves development efficiency and reduces redundant experiments, but also obtains a globally optimal formulation that meets the requirements of various performance indicators. Specifically:

[0172] 1) The reaction kinetic model describes the chemical reaction kinetics during polymer crosslinking, such as reaction rate constant and reaction order, laying the foundation for accurate prediction of crosslinking density.

[0173] 2) The product performance prediction model based on machine learning establishes a nonlinear mapping relationship between polymer ratios (such as molecular weight and concentration) and final performance indicators (such as viscosity and strength) by fitting experimental data, providing a reliable prediction method for multi-objective optimization.

[0174] 3) The multi-objective optimization model uses minimizing shear viscosity and maximizing compressive / tensile strength as objective functions, while simultaneously satisfying various material balance constraints. An intelligent optimization algorithm is used to solve for the optimal formulation. This method can significantly shorten the formulation development cycle and improve the predictability of product performance.

[0175] To better implement the formulation optimization method provided by this invention, a specific embodiment of this formulation optimization method is provided below with reference to a specific formula:

[0176] The specific implementation method of step S10 is as follows:

[0177] Orthogonal experimental design is typically used to consider the main influencing factors and their different levels. Assume that m influencing factors X1, X2, ..., Xn need to be considered. m Each factor is assigned l1, l2, ..., l m There are [number] levels. Therefore, the total number of levels in an orthogonal experiment is [number].

[0178] Orthogonal arrays are an experimental design method that allows the study of the main effects of factors on results using a relatively small number of experimental cases. Each row of the design matrix D represents an experimental case, and each column represents different levels of an influencing factor. Matrix elements d ij The value is taken as the level number of the factor. For example, an L9(3) 4 Orthogonal array:

[0179]

[0180] This table corresponds to 9 test points for 4 factors with 3 levels each. n parallel samples are prepared for each test point, requiring a total of L×n tests.

[0181] For any given test point, the actual values ​​of the influencing factors are:

[0182]

[0183] in This represents the actual value of the k-th level of the i-th factor.

[0184] Test all samples for the required performance indicators Y1, Y2, ..., Y p Examples include shear viscosity and compressive strength. For each index, calculate the mean value for each level:

[0185]

[0186] in This represents the test value of the l-th parallel sample corresponding to the j-th performance index at the k-th level of the i-th factor.

[0187] Then calculate the range of each factor for each indicator:

[0188] The larger the value, the greater the influence of the i-th factor on the j-th indicator. This can be used to screen out the main influencing factors.

[0189] Based on the analysis results of different indicators, the optimal levels for each influencing factor were initially determined, such as increasing the polymer concentration. This lays the foundation for further optimization.

[0190] The specific implementation method of step S20 is as follows:

[0191] Chemical reaction kinetic models can use reaction kinetic equations to describe cross-linking reaction processes, such as:

[0192]

[0193] Where c is the crosslinking density, t is time, [A] and [B] are reactant concentrations, and k, m, and n are the reaction order and rate constant, respectively.

[0194] By combining the material conservation equation and the reaction kinetic equation, the crosslinking reaction can be numerically simulated.

[0195] Product performance prediction models can employ machine learning algorithms, such as Artificial Neural Networks (ANN) and Support Vector Machines (SVM). Taking shear viscosity η and compressive strength σ as examples:

[0196] η = f η (x;w η ,b η )

[0197] σ=f σ (x;w σ ,b σ )

[0198] Where x is the input variable (such as polymer molecular weight, crosslinking agent dosage, etc.), and w and b are the weight and bias parameters, respectively.

[0199] For a single hidden layer network, its form is:

[0200]

[0201] Where σ(·) is the activation function, such as ReLU or Sigmoid; U and v are network weight parameters; b is the bias parameter; and c is a constant offset term.

[0202] By optimizing the network parameters using training data, a performance prediction model can be obtained. Step S10 uses statistical methods for calculation. Used to assess the average value of each level.

[0203] Step S20 utilizes machine learning to directly learn the prediction functions of η and σ from the training data.

[0204] To prevent overfitting, a regularization term, such as the L1 norm or L2 norm, can be added to X.

[0205] Step S30: Determine the multi-objective optimization objective function

[0206]

[0207] Where, η * ,σ * ,∈ * These represent the target values ​​for the desired shear viscosity, compressive strength, and tensile strain, respectively; w1, w2, and w3 are the corresponding weighting coefficients, reflecting the importance of each indicator; η(x), σ(x), and ∈(X) are given by the prediction model established in step S20.

[0208] Step S40: Establish a multi-objective optimization model for the formulation

[0209]

[0210] Where, x lb ≤x≤x ub

[0211] Where, x = (x1, x2, ..., x...) n ) is a vector of independent variables, including polymer molecular weight, concentration, crosslinking agent dosage, etc.; constraint condition ∑x i =1 represents the total quantity measurement relationship of materials; x lb ,x ub These are the lower and upper bounds for each variable.

[0212] Step S50: Solve using the Particle Swarm Optimization (PSO) algorithm.

[0213] Algorithm flow:

[0214] 1) Initialize the population X = (x1, ..., x2) N ) and velocity V = (v1, ..., v N )

[0215] 2) Calculate the fitness f of each particle. i =F(x) i Find the current optimal x. g and its fitness f g

[0216] 3) Update the individual's optimal position p i and the group optimal x g

[0217] 4) Update speed and location:

[0218] v i ←ωv i +c1r1(p i -x i )+c2r2(x g -x i )

[0219] x i ←x i +v i

[0220] 5) Repeat steps 2)-4) until the stopping condition is met.

[0221] Where ω is the inertial weight; c1, c2 are the acceleration constants; and r1, r2 ~ U(0,1) are random numbers.

[0222] Step S60: Conduct optimization experiments, obtain optimization data, prepare samples for small-scale testing according to the optimization results of S50, and test actual indicators.

[0223] Step S70: Update the model using the new data

[0224] 1) Add the new data to the original training set

[0225] 2) Modify the model structure based on new phenomena, such as adjusting the number of network layers or introducing new features.

[0226] 3) Retrain the reaction kinetics model and performance prediction model

[0227] Step S80: Reconstruct the optimization model and solve it.

[0228] 1) Substitute the updated performance prediction model into the objective function.

[0229] 2) Adjust the weights w of the objective function i

[0230] 3) Solve the new optimization model to obtain the new round of optimal formulation. Step S90: Repeat S60-S80 until convergence.

[0231] Experiments were conducted using the new optimal formula to obtain new data, the model was updated, the optimization model was rebuilt, and the solution was obtained. When the optimal solution did not change significantly for several consecutive generations, it was considered to have converged to the globally optimal formula.

[0232] The key to this formulation optimization strategy is that the predictive model is continuously improved through interactive iteration between modeling and experimental data, thereby obtaining a more accurate objective function that guides the optimization algorithm to converge toward the global optimum. The efficiency of formulation optimization will continuously improve with the accumulation of data and models.

[0233] The following are specific application examples of the injectable hydrogel of the present invention: A medical device company has developed an injectable hydrogel product to address clinical needs such as vascular embolism. This hydrogel uses polyethylene glycol (PEG) derivatives as its main component and achieves in-situ curing through a chemical cross-linking reaction of mercapto-acrylate. Simultaneously, inorganic nanofillers and medical contrast agents are introduced to impart good rheological properties and imaging effects to the material. To optimize the formulation of this hydrogel, the company designed the following five comparative experiments: Experimental Group 1 (Control Group): PEG-SH (Mw 6000): 10% w / v, PEG-acrylate (Mw 8000): 8% w / v, montmorillonite: 3% w / v, iodized oil: 20% v / v;

[0234] Experimental group 2: PEG-SH (Mw 4000): 12% w / v, PEG-acrylate (Mw 6000): 10% w / v, kaolin: 4% w / v, barium sulfate: 25% w / w;

[0235] Experimental group 3: PEG-SH (Mw 8000): 15% w / v, PEG-ethylene oxide (Mw 6000): 12% w / v, bentonite: 5% w / v, sodium iodide: 0.4 mol / L;

[0236] Experimental group 4: PEG-SH (Mw 5000): 14% w / v, PEG-isocyanate: 10% w / v, zeolite: 3% w / v, iodized oil: 22% v / v;

[0237] Experimental group 5: PEG-SH (Mw 7000): 13% w / v, PEG-amide (Mw 8000): 11% w / v, montmorillonite: 3.5% w / v, barium sulfate: 28% w / w.

[0238] The preparation process and performance test results of these five formulations are described below:

[0239] 1. Experimental Group 1 (Control Group)

[0240] (1) Preparation steps

[0241] First, PEG-SH (Mw 6000) and PEG-acrylate (Mw 8000) were dissolved in PBS buffer at an equivalent ratio of 1:1 and stirred until completely dissolved. Simultaneously, montmorillonite powder was mixed with deionized water at a mass ratio of 3:10 and stirred for 1 hour to obtain a homogeneous nano-montmorillonite suspension. Then, the two solutions were mixed at a volume ratio of 1:1, and 20% v / v iodized oil contrast agent was added, followed by stirring for 5 minutes. Finally, the resulting precursor solution was loaded into a syringe and allowed to stand for 4 minutes to complete the in-situ solidification of the hydrogel.

[0242] (2) Performance Testing

[0243] The prepared hydrogel samples were first tested for shear viscosity at different shear rates using a rotational rheometer. The results are as follows: Figure 2 As shown, the hydrogel exhibits good shear-thinning properties, with a high viscosity (approximately 3.2 Pa·s) at low shear rates, but the viscosity gradually decreases to around 0.6 Pa·s as the shear rate increases, meeting the flowability requirements of the injection process.

[0244] Next, the compressive strength and tensile strength of the hydrogel were tested using a pressure testing instrument and a universal testing machine, respectively. The results showed that the compressive strength of the hydrogel was 205 kPa and the tensile strength was 78 kPa, both of which meet the mechanical performance standards for clinical applications.

[0245] Furthermore, during X-ray imaging experiments, it was found that the hydrogel had a good imaging effect due to the introduction of iodized oil, and its distribution in blood vessels could be clearly shown.

[0246] Based on the above performance indicators, it can be concluded that the hydrogel prepared in experimental group 1 meets all the requirements for clinical application and provides a control sample for subsequent studies.

[0247] 2. Experimental Group 2

[0248] (1) Preparation steps

[0249] The preparation steps were basically the same as those in Experimental Group 1. The differences were: the molecular weight of PEG-SH was reduced to 4000, and the molecular weight of PEG-acrylate was reduced to 6000; kaolin was used instead of montmorillonite as filler, and the amount was increased to 4% w / v; the contrast agent was changed from iodized oil to barium sulfate, and the amount was 25% w / w.

[0250] (2) Performance Testing

[0251] The shear viscosity curve of the hydrogel sample is as follows: Figure 3As shown in the figure, compared to experimental group 1, the viscosity of this hydrogel is slightly lower at low shear rates (approximately 2.8 Pa·s), but the viscosity decreases more significantly with increasing shear rate, reaching only 0.4 Pa·s at high shear rates. This is mainly due to the reduced molecular weight of the PEG chains and the increased amount of kaolin, which endows the hydrogel with stronger shear-thinning properties. In mechanical property tests, the compressive strength of this hydrogel is 195 kPa and the tensile strength is 72 kPa, slightly lower than that of experimental group 1, which may be related to the reduction in PEG chain length.

[0252] X-ray imaging tests showed that barium sulfate contrast agent can also provide good visualization of hydrogels, with imaging resolution comparable to that of iodized oil.

[0253] In summary, although the hydrogel in Experimental Group 2 showed a slight decrease in mechanical properties, its excellent rheological properties could meet the requirements of injection applications, and its relatively low cost made it an alternative formulation option.

[0254] 3. Experimental Group 3

[0255] (1) Preparation steps

[0256] In this group of experiments, the molecular weight of PEG-SH was increased to 8000, and the molecular weight of PEG-ethylene oxide was 6000. The dosages were 15% w / v and 12% w / v, respectively. Bentonite was used as the filler at a dosage of 5% w / v. Sodium iodide was used as the contrast agent at a concentration of 0.4 mol / L. The preparation steps were similar to the previous two groups.

[0257] (2) Performance Testing

[0258] like Figure 4 As shown, the viscosity of this hydrogel is as high as 4.1 Pa·s at low shear rates, but it decreases rapidly to 0.7 Pa·s as the shear rate increases, still meeting the injection requirements. Compared with the previous two groups, the high molecular weight and high concentration of this formulation endow the hydrogel with superior rheological properties.

[0259] In mechanical property testing, the hydrogel exhibited a compressive strength of 225 kPa and a tensile strength of 85 kPa, significantly superior to the previous two groups. This is mainly attributed to the higher crosslinking density and the reinforcing effect of bentonite. X-ray imaging experiments showed that sodium iodide, as a contrast agent, could also effectively visualize the distribution of the hydrogel within blood vessels.

[0260] In summary, the hydrogel prepared in experimental group 3 showed significant improvements in both rheological and mechanical properties, making it a relatively ideal formulation choice. However, it should also be noted that excessively high polymer concentrations may affect its injection performance.

[0261] 4. Experimental Group 4

[0262] (1) Preparation steps

[0263] In this group of experiments, the molecular weight of PEG-SH was 5000, and the dosage was 14% w / v; the dosage of PEG-isocyanate was 10% w / v. Zeolite was used as the filler, and the dosage was 3% w / v. Iodized oil was used as the contrast agent, and the dosage was 22% v / v. The preparation steps were similar to those described above.

[0264] (2) Performance Testing

[0265] like Figure 5 As shown, the viscosity of this hydrogel at low shear rates is approximately 3.5 Pa·s, slightly lower than that of experimental group 3, but it still maintains a low viscosity (0.6 Pa·s) at high shear rates, meeting the injection requirements.

[0266] In mechanical property testing, the hydrogel exhibited a compressive strength of 215 kPa and a tensile strength of 82 kPa, which were superior to experimental groups 1 and 2, but slightly lower than experimental group 3. This is likely due to the lower molecular weight and relatively lower crosslinking density of PEG-SH.

[0267] X-ray imaging results showed that iodized oil, as a contrast agent, also provided good visualization of the hydrogel. Overall, the hydrogel in experimental group 4 exhibited excellent comprehensive performance, making it a relatively ideal formulation choice.

[0268] 5. Experimental group 5

[0269] (1) Preparation steps

[0270] This group used PEG-SH (Mw 7000) and PEG-amide (Mw 8000) as the main components, at amounts of 13% w / v and 11% w / v, respectively. Montmorillonite was used as the filler at 3.5% w / v. Barium sulfate was used as the contrast agent at a concentration of 28% w / w. The preparation steps were the same as before.

[0271] (2) Performance Testing

[0272] like Figure 5 As shown, the viscosity of this hydrogel at low shear rates is approximately 3.8 Pa·s, slightly higher than that of experimental group 4, but it still exhibits good shear-thinning properties. At high shear rates, the viscosity drops to 0.8 Pa·s.

[0273] In the mechanical property test, the hydrogel had a compressive strength of 220 kPa and a tensile strength of 84 kPa, which is close to the level of experimental group 3, reflecting a high crosslinking density.

[0274] X-ray imaging results show that barium sulfate can also provide good imaging effects for hydrogels.

[0275] 6. Summary

[0276] Through the above five sets of experiments, the company systematically evaluated the effects of the molecular weight, functional group type, concentration of PEG derivatives, as well as the type and amount of inorganic fillers and contrast agents on the hydrogel properties. The experimental results showed that: (1) Increasing the molecular weight and concentration of PEG is beneficial to improving the rheological properties and mechanical strength of the hydrogel, but it will also increase the initial viscosity, which is not conducive to injection. Therefore, a balance needs to be struck between performance and injectability.

[0277] (2) Using thiol-acrylate or thiol-isocyanate chemical crosslinking methods can achieve higher crosslinking density and mechanical strength. While using amide crosslinking results in a slight decrease in performance, the cost is relatively low.

[0278] (3) Appropriate amounts of inorganic fillers (such as 3-5% w / v) can effectively improve the shear thinning properties of hydrogels, but excessive addition will reduce mechanical strength.

[0279] (4) Different types of contrast agents, such as iodides and sulfates, can provide good contrast effects for hydrogels and can be selected according to actual application needs.

[0280] Based on the above results, the company initially determined that polyethylene glycol thioglycolate (Mw 8000), polyethylene glycol amide (Mw 8000), montmorillonite, and iodized oil were the optimal formulation combination, with the dosages of each component being 13% w / v, 11% w / v, 3.5% w / v, and 25% v / v, respectively. This formulation not only possesses ideal rheological properties and mechanical strength but also exhibits good visualization effects in X-ray imaging, meeting various requirements for clinical applications. The following is an example of a specific application scenario of the formulation optimization method provided by this invention: Assume a medical device company has developed an injectable hydrogel for vascular embolization treatment. The company hopes to prepare a hydrogel with high viscosity, high strength, and good injectability through reasonable formulation optimization to meet the needs of clinical applications. Based on the formulation optimization method proposed in this invention, the company carried out the following series of work: 1. Basic experiments

[0281] First, the company designed a series of orthogonal experiments to systematically evaluate the effects of six factors—polymer molecular weight, functional group type, polymer concentration, crosslinking agent dosage, diatomaceous earth type and dosage, and contrast agent type and dosage—on the rheological properties and mechanical strength of the hydrogel. The specific experimental design is as follows:

[0282] The factor levels are set as shown in Table 1.

[0283] Table 1. Factor Level Table for Orthogonal Experiment

[0284] Factor Level 1 Level 2 Level 3 Polyethylene glycol molecular weight (g / mol) 3000 6000 12000 Functional group type Amino Mercapto Amido Polymer concentration (% w / v) 5 10 20 Crosslinker amount (% w / v) 4 8 12 Diatomaceous earth type Montmorillonite clay Kaolin clay Bentonite clay Contrast agent type Iodinated oil Barium sulfate Sodium iodide Contrast agent amount 20% v / v 30% w / w 0.5 mol / L

[0285] According to L16(4) 5 An orthogonal experimental design was used, comprising 16 groups of experiments, with 3 parallel samples in each group. All experimental samples were prepared according to the following method:

[0286] (1) Weigh out the raw materials such as polyethylene glycol, crosslinking agent, and diatomaceous earth according to the formula, dissolve them in an appropriate amount of buffer solution (such as PBS), and stir thoroughly until uniform.

[0287] (2) Mix the obtained polymer solution and diatomaceous earth suspension in a certain proportion, then add the corresponding contrast agent and continue stirring until completely homogeneous.

[0288] (3) The precursor fluid is loaded into a syringe and left to stand for a certain period of time to undergo a cross-linking reaction to form the desired hydrogel.

[0289] (4) Rheological properties (shear viscosity) and mechanical strength (compressive strength, tensile strength) of the prepared hydrogel samples were tested.

[0290] Through experiments, the company obtained the basic experimental data shown in Table 2.

[0291] Table 2. Results of the Orthogonal Experiment

[0292]

[0293] Through statistical analysis of these experimental data, the company reached the following conclusions:

[0294] (1) The molecular weight, functional group type and concentration of polymer are the main factors affecting the performance of hydrogel, followed by the amount of crosslinking agent and the type of diatomaceous earth. The type and amount of contrast agent have a relatively small impact on the performance.

[0295] (2) Increasing the molecular weight and concentration of the polymer can improve the shear viscosity and mechanical strength of the hydrogel, but it will also significantly increase the initial viscosity, which is not conducive to injection. Using functional groups such as thiol or amide groups can achieve higher crosslinking density and strength.

[0296] (3) Appropriate amounts of diatomaceous earth (such as 2-5% w / v) can effectively improve the rheological properties of hydrogels, but excessive addition will reduce mechanical strength.

[0297] (4) Different types of contrast agents have little effect on the performance of hydrogels, but iodides have better contrast effects than sulfates.

[0298] Based on the above analysis results, the company has initially determined that increasing the molecular weight and concentration of polymers, using thiol or amide functional groups, adding an appropriate amount of diatomaceous earth, and selecting iodide contrast agents are the main directions for subsequent optimization.

[0299] 2. Model Building and Optimization

[0300] Next, based on basic experimental data, the company's R&D personnel established a chemical reaction kinetic model and a product performance prediction model for hydrogels.

[0301] The chemical reaction kinetic model employs the classic Michaelis-Menten kinetic equations:

[0302]

[0303] In the formula, c is the crosslinking density, t is the reaction time, [E] is the enzyme (crosslinking agent) concentration, and [S] is the substrate (polymer) concentration k. cat K is the enzyme activity constant. m is the Michaelis constant. By fitting experimental data, the specific values ​​of each parameter were determined: k cat =0.18s -1 , Using this kinetic model, the company can predict how the crosslinking density of hydrogels changes over time under different formulation conditions, providing a basis for the regulation of curing behavior.

[0304] Meanwhile, the company has also established a hydrogel performance prediction model based on artificial neural networks. This model uses shear viscosity η and compressive strength σ as parameters... c and tensile strength σ t Using polymer molecular weight M, functional group type f, polymer concentration c, crosslinking agent dosage r, diatomaceous earth type s and dosage h, and contrast agent type d and dosage v as input variables, a three-layer backpropagation (BP) neural network model was constructed. The training data for the model came from the aforementioned orthogonal experimental results, and the gradient descent algorithm was used during training. The average prediction errors of the model on the retained test dataset were: Δη = 8.7%, Δσ c =7.2%, Δσ t =6.5%, indicating high prediction accuracy. With the above two models as support, the company's R&D personnel then conducted multi-objective optimization of the hydrogel formulation. They set the following optimization objective function:

[0305]

[0306] Where, η * = 3 Pa·s, Let be the desired performance target value. Combining the above objective function with a neural network-based performance prediction model, we obtain the following multi-objective optimization problem:

[0307]

[0308] stΣx i =1

[0309] x lb ≤x≤x ub

[0310] In the formula, x = (M, f, c, r, s, h, d, v) is the optimization variable vector, and f1(x), f2(x), and f3(x) are the η, σ, and σ predicted by the neural network, respectively. c σ t value.

[0311] To solve this multi-objective optimization problem, the company adopted an improved particle swarm optimization algorithm (MOPSO). The specific steps are as follows:

[0312] (1) Initialize the position and velocity of the population individuals (feasible solutions).

[0313] (2) Assess the fitness of each individual and calculate the objective function value F(x) according to equation (2).

[0314] (3) Update the historical best position of each individual and the global non-dominated best solution set.

[0315] (4) Update the individual's velocity and position according to the following formula:

[0316] v = wv + c1r1(px) + c2r2(gx)

[0317] x = x + v

[0318] Where w is the inertia weight, c1 and c2 are acceleration constants, r1 and r2 are 0-1 uniform random numbers, p is the individual's historical best position, and g is the global non-dominated optimal solution.

[0319] (5) Repeat steps (2)-(4) until the algorithm convergence condition is met.

[0320] After 20 iterations, the MOPSO algorithm finally converged to the Pareto optimal solution set shown in Table 3.

[0321] Table 3 Optimization results of the MOPSO algorithm

[0322]

[0323]

[0324] 3. Optimize experiments and model updates

[0325] Based on the Pareto optimal solution set obtained by the MOPSO algorithm, the company selected three representative formulations for small-scale experimental verification, as follows:

[0326] (1) Formula 1:

[0327] - Polyethylene glycol mercapto (Mw 8000): 12% w / v

[0328] - Polyethylene glycol amide (Mw 8000): 10% w / v

[0329] - Montmorillonite: 3% w / v

[0330] -Iodized oil: 22% v / v

[0331] (2) Formula 2:

[0332] - Polyethylene glycol amide (Mw 6000): 15% w / v

[0333] - Polyethylene glycol acrylate (Mw 6000): 8% w / v

[0334] - Kaolin: 4% w / v

[0335] -Barium sulfate: 28% w / w

[0336] (3) Formula 3:

[0337] - Polyethylene glycol amino (Mw 10000): 10% w / v

[0338] - Polyethylene glycol isocyanate: 7% w / v

[0339] - Zeolite: 2.5% w / v

[0340] - Sodium iodide: 0.45 mol / L

[0341] Following the aforementioned hydrogel preparation steps, the company prepared small-scale samples of the three formulations and conducted detailed performance tests on them. The experimental results are shown in Table 4.

[0342] Table 4 Optimization Experiment Results

[0343] Formulation η (Pa s) c (kPa)]]> ​ t (kPa)]]> ​ Curing time (min) Injectability 1 3.05±0.35 198±22 75±11 4 Good 2 3.79±0.41 182±20 68±10 6 Good 3 2.86±0.33 205±23 79±12 3 Good Good

[0344] The results show that all three optimized formulations met the expected performance targets: shear viscosity around 3 Pa·s, compressive strength and tensile strength around 200 kPa and 80 kPa respectively, curing time between 3 and 6 minutes, and good injectability. Compared with the prediction results of the MOPSO algorithm, there are some deviations between the experimental data and the model predictions, with formulation 2 showing a slightly larger deviation, mainly due to the limitations of modeling the crosslinking reaction kinetics.

[0345] To further improve model accuracy, the company supplemented the original training set with optimized experimental data and retrained the chemical kinetics model and neural network performance prediction model. The new chemical reaction kinetic equations are as follows:

[0346] The average prediction error of the neural network model was also reduced to: Δη = 6.3%, Δσ c =5.1%, Δσ t =4.8%.

[0347] 4. Final Optimization and Validation

[0348] Based on the updated model, the company conducted another multi-objective optimization of the hydrogel formulation. The optimization results are shown in Table 5.

[0349] Table 5 Final Optimization Results

[0350] The formula was verified through multiple experiments under small-scale conditions, and the results are as follows:

[0351] - Shear viscosity: 3.12 ± 0.30 Pa·s

[0352] -Compressive strength: 218±24kPa

[0353] - Tensile strength: 84±13kPa

[0354] - Curing time: 4±1 min

[0355] - Injectability: Good; can be smoothly injected through a 2.5mm injection needle.

[0356] All of the above performance indicators met the expected targets and were in good agreement with the model prediction results.

[0357] To further verify the performance of this hydrogel in practical applications, the company invited clinical experts to conduct animal experiments. A vascular embolization experiment was performed on a rat abdominal aorta model. The results showed that the hydrogel could rapidly solidify within the target blood vessel, providing good mechanical support and successfully blocking blood flow, achieving the expected embolization effect. The animal experiments lasted for 24 weeks, and no significant degradation of the hydrogel material was observed, indicating good tissue compatibility and no obvious inflammatory or toxic reactions.

[0358] Through the above series of efforts, the company finally determined the optimal formulation and preparation process of the injectable hydrogel, laying a solid foundation for its further clinical trials and industrial applications.

[0359] In summary, the hydrogel formulation optimization method proposed in this invention fully utilizes chemical reaction kinetics models and machine learning techniques, effectively solving the injectability problem while ensuring hydrogel performance. The above description is merely a specific embodiment of this invention, but the scope of protection of this invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this invention should be included within the scope of protection of this invention.

Claims

1. An injectable hydrogel for the treatment of vascular embolization, the composition of which is selected from: a) 12% w / v polyethylene glycol mercapto Mw 8000, 10% w / v polyethylene glycol amide Mw 8000, 3% w / v montmorillonite, 22% v / v iodized oil and solvent; b) 15% w / v polyethylene glycol amide Mw 6000, 8% w / v polyethylene glycol acrylate Mw 6000, 4% w / v kaolin, 28% w / w barium sulfate and solvent; c) 13% w / v polyethylene glycol amino Mw 10000, 7% w / v polyethylene glycol isocyanate, 2.5% w / v zeolite, 0.45 mol / L sodium iodide, and solvent; or d) 13% w / v polyethylene glycol mercapto Mw 8500, 9% w / v polyethylene glycol amide Mw 8500, 3.5% w / v montmorillonite, 25% v / v iodized oil and solvent.

2. The injectable hydrogel for vascular embolization treatment according to claim 1, wherein the solvent is water or a buffer solution.

3. A method for optimizing the formulation of an injectable hydrogel for the treatment of vascular embolism, characterized in that, The formulation optimization method described above is used to optimize the formulation of hydrogels for injection, and includes the following steps: S10. Design basic experiments to test the effects of different raw material components on product performance and obtain basic experimental data; the basic experiments are orthogonal experimental designs to investigate the effects of polyethylene glycol molecular weight, functional group type, polymer concentration, crosslinking agent dosage, magnesium lithium silicate type and dosage, and contrast agent type and dosage on rheological properties during injection and mechanical strength after curing. S20. Establish chemical reaction models and product performance prediction models based on basic experimental data; (a) The chemical reaction model is specifically the Michaelis-Menten kinetic equation (1). In the formula, c is the crosslinking density, t is the reaction time, [E] is the enzyme (crosslinking agent) concentration, [S] is the substrate (polymer) concentration, and k cat K is the enzyme activity constant. m K is the Michaelis constant. cat =0.18s -1 , It describes the polymer crosslinking reaction process based on kinetic equations; (b) The product performance prediction model is specifically based on nonlinear mapping modeling of experimental data using machine learning algorithms; it uses shear viscosity η and compressive strength σ as parameters. c and tensile strength σ t As the output variable, a three-layer BP neural network model was constructed with polymer molecular weight M, functional group type f, polymer concentration c, crosslinking agent amount r, diatomaceous earth type s and amount h, and contrast agent type d and amount v as input variables. The training data of this model came from the aforementioned orthogonal experimental results, and the training process adopted the gradient descent algorithm. S30. According to the product performance requirements, the multi-objective optimization objective function as shown in equation (2) was determined, where multiple objectives include minimum shear viscosity during injection, maximum compressive strength, and maximum tensile strength. Where, η * = 3 Pa·s, This represents the desired performance target value. S40. Establish a multi-objective optimization model for the formulation by combining material metering relationships; S50. The multi-objective optimization model is solved using an intelligent optimization algorithm to obtain the preliminary optimized formulation component ratios; the intelligent optimization algorithm is an improved particle swarm optimization algorithm (MOPSO). S60. Use the preliminary optimized formula to conduct optimization experiments and obtain optimization experimental data; S70. Update chemical reaction models and product performance prediction models using optimized experimental data; The updated chemical reaction model is as described in equation (3): S80. Reconstruct the multi-objective optimization model and solve it again using the intelligent optimization algorithm to obtain a new optimal formulation component ratio; S90. Repeat steps S60 to S80 until the formula converges to the optimal solution. The formulation of the injectable hydrogel includes polymers containing functional groups such as polyethylene glycol amino, polyethylene glycol thiol, polyethylene glycol carboxyl, or polyethylene glycol hydroxyl, polymers with other functional groups capable of cross-linking, lithium magnesium silicate components, contrast agents, and solvents; the functional groups are polyethylene glycol amide, polyethylene glycol acrylate, polyethylene glycol ethylene oxide, polyethylene glycol isocyanate, or polyethylene glycol aldehyde.

4. The method for optimizing the formulation of the injectable hydrogel for vascular embolization treatment according to claim 3, characterized in that, The levels of each factor in the orthogonal experimental design are: The molecular weight of polyethylene glycol is selected from 3000, 6000, and 12000 g / mol; The functional groups are selected from amino, thiol, and amide groups; The polymer concentration was selected from 5, 10, and 20% w / v; The amount of crosslinking agent used is selected from 4, 8, or 12% w / v; The magnesium lithium silicate is selected from montmorillonite, kaolinite, and bentonite; The contrast agents are selected from iodized oil, barium sulfate, and sodium iodide.

5. The method for optimizing the formulation of the injectable hydrogel for vascular embolization therapy according to claim 3 or 44, characterized in that, The multi-objective optimization model for the formulation in step S40 is specifically a set of nonlinear equations with polyethylene glycol molecular weight, functional group type, polymer concentration, crosslinking agent dosage, magnesium lithium silicate type and dosage, and contrast agent type and dosage as independent variables, and minimum shear viscosity, maximum compressive strength, and maximum tensile strength at injection as dependent variables.

6. The method for optimizing the formulation of an injectable hydrogel for vascular embolization therapy according to claim 5, characterized in that, In step S40, the objective function is combined with a neural network-based performance prediction model to obtain the following multi-objective optimization model for the formulation: s.t.∑x i =1 x lb ≤x≤x ub In the formula, x = (M, f, c, r, s, h, d, v) is the optimization variable vector, and f1(x), f2(x), and f3(x) are the η, σ, and σ predicted by the neural network, respectively. c σ t value.

7. A method for optimizing the formulation of an injectable hydrogel for vascular embolization therapy according to claim 3 or 4, characterized in that, The intelligent optimization algorithm is an improved particle swarm optimization (MOPSO) algorithm. The specific steps are as follows: (1) Initialize the position and velocity of each individual (feasible solution) in the population; (2) Assess the fitness of each individual; calculate the objective function value F(x) according to equation (2); (3) Update the historical best position of each individual; and the global non-dominated best solution set; (4) Update the individual's velocity and position according to the following formula: v = wv + c1r1(px) + c2r2(gx) x = x + v; Where w is the inertia weight, c1 and c2 are acceleration constants, r1 and r2 are 0-1 uniform random numbers, p is the individual's historical best position, and g is the global non-dominated optimal solution; (5) Repeat steps (2)-(4) until the algorithm convergence condition is met.

Images