Method and system for screening and calling new excipients for preparations based on usage requirements

Abstract

The present invention discloses a method and system for screening and calling new auxiliary materials for preparations based on usage requirements, which belongs to the field of production control, wherein the method includes: establishing a preparation database for new auxiliary materials; establishing a capsule analysis model, performing performance requirement analysis, establishing a performance requirement set, performing feature similarity matching in the preparation database, sorting in order of matching values, and establishing a search space; randomly distributing N initial solutions, establishing an initial solution set, and performing optimal search of preparation process parameters in the search space; performing search constraints through a cost balance evaluation network, and establishing a process parameter set according to the iterative results of the optimal search as the production call parameter of the target new auxiliary material, and executing production control. Based on the capsule performance requirement analysis, a search space is established, and optimal search and search constraints of preparation process parameters are performed to improve the accuracy of screening and calling new auxiliary materials.

Description

Technical Field

[0001] The present invention relates to the field of production control, and in particular to a method and system for screening and calling new auxiliary materials for preparations based on usage requirements. Background Art

[0002] In the capsule production process, the selection and use of new excipients have an important impact on capsule production. In the prior art, the screening and use of new excipients mainly rely on the experience of pharmaceutical research and development personnel and a small amount of experimental data. This over-reliance on experience has obvious limitations. On the one hand, the scope of application of experience is limited, and it is difficult to meet new formulation requirements; on the other hand, manual experience is difficult to accurately grasp the complex relationship between excipient performance and preparation process parameters, resulting in improper matching of process parameters. These problems lead to low accuracy in the screening and use of new excipients, and the inability to give full play to the advantages of new excipients, which restricts the production of capsule preparations. Summary of the invention

[0003] This application provides a method and system for screening and calling new excipients for preparations based on usage needs, aiming to solve the technical problems that the existing screening and calling of new excipients is overly dependent on experience, has limitations in calling and improper matching of process parameters, resulting in low accuracy in the screening and calling of new excipients.

[0004] In view of the above problems, the present application provides a method and system for screening and calling new excipients for preparations based on usage requirements.

[0005] The first aspect disclosed in the present application provides a method for screening and calling new excipients for preparations based on usage requirements, the method comprising: establishing a preparation database for new excipients, wherein the preparation database stores preparation process parameters of N new excipients, and each new excipient corresponds to a plurality of preparation process parameters, and each preparation process parameter carries a performance requirement identifier, the performance requirement identifier comprises its own performance identifier and an inclusion requirement identifier; establishing a capsule analysis model, using the capsule analysis model to perform performance requirement analysis on the target new excipient, establishing a performance requirement set, using the performance requirement set to perform feature similarity matching in the preparation database, sorting in order of matching values, and establishing a search space; randomly distributing N initial solutions, establishing an initial solution set, and performing optimal search for preparation process parameters in the search space based on the initial solution set; performing search constraints through a cost balance evaluation network, and establishing a process parameter set according to the iterative results of the optimal search; using the process parameter set as a production call parameter for the target new excipient to perform production control.

[0006] Another aspect disclosed in the present application provides a screening and calling system for new excipients for preparations based on usage requirements, the system comprising: a database establishment module for establishing a preparation database for new excipients, wherein the preparation database stores preparation process parameters of N new excipients, and each new excipient corresponds to a plurality of preparation process parameters, and each preparation process parameter carries a performance requirement identifier, the performance requirement identifier comprises its own performance identifier and an inclusion requirement identifier; a search space establishment module for establishing a capsule analysis model, using the capsule analysis model to perform performance requirement analysis of the target new excipient, establishing a performance requirement set, using the performance requirement set to perform feature similarity matching in the preparation database, sorting in order of size of the matching values, and establishing a search space; an optimal search module for randomly distributing N initial solutions, establishing an initial solution set, and performing optimal search of the preparation process parameters in the search space based on the initial solution set; a parameter set establishment module for performing search constraints through a cost balance evaluation network, and establishing a process parameter set according to the iterative results of the optimal search; a production control module for using the process parameter set as a production call parameter for the target new excipient to perform production control.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] By establishing a structured and systematic preparation database for new excipients, a data basis is provided for subsequent performance analysis and process optimization, and performance requirement identification is introduced to achieve a direct correlation between process parameters and performance requirements, laying the foundation for accurate matching; a capsule analysis model is introduced to achieve quantitative characterization of the performance of new excipients, establish a performance requirement set, and establish a search space by matching the characteristics of the performance requirement set with the preparation database, providing effective constraints for optimal search; a random initialization strategy is adopted to randomly distribute N initial solutions, establish an initial solution set, and generate a diverse initial solution set, which is conducive to the search for the global optimal solution, and conduct optimal search in the search space to improve search efficiency and optimization quality; a cost leveling method is introduced The evaluation network is used to realize the dynamic constraint of cost factors on the optimization process, and the production cost is controlled while ensuring the performance. Through iterative optimization, a high-performance and low-cost process parameter set is obtained; the process parameter set is used as the production call parameter of the target new excipient, and production control is executed. It is applied to production process control to realize the technical solution of linking the new excipient screening results with production practice. It solves the technical problems that the existing new excipient screening and calling are overly dependent on experience, have calling limitations and improper matching of process parameters, resulting in low accuracy of new excipient screening and calling. It achieves the technical effect of establishing a search space based on capsule performance demand analysis, conducting optimal search and search constraints on preparation process parameters, and improving the accuracy of new excipient screening and calling.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A flow chart of a method for screening and calling new excipients for preparations based on usage requirements is provided for the embodiments of the present application;

[0011] Figure 2 A flowchart of search constraints in a method for screening and calling new excipients for preparations based on usage requirements is provided for an embodiment of the present application;

[0012] Figure 3 A structural schematic diagram of a system for screening and calling new excipients for preparations based on usage requirements is provided for an embodiment of the present application.

[0013] Explanation of the accompanying drawings: database establishment module 11, search space establishment module 12, optimization search module 13, parameter set establishment module 14, production control module 15. DETAILED DESCRIPTION

[0014] The overall idea of ​​the technical solution provided by this application is as follows:

[0015] The embodiments of the present application provide a method and system for screening and calling new excipients for preparations based on usage requirements. First, a preparation database for new excipients is established to store a variety of preparation process parameters for N new excipients, and a performance requirement identifier is introduced to achieve a direct association between process parameters and performance requirements. Secondly, the performance requirement analysis of the target new excipient is performed through the capsule analysis model, a performance requirement set is established, and feature similarity matching is performed with the preparation database to establish a search space. Then, N initial solutions are randomly distributed to form an initial solution set, and an optimal search for the preparation process parameters is performed within the constrained search space. At the same time, a cost balance evaluation network is introduced to dynamically constrain the optimal search process to control production costs while ensuring performance. Finally, the optimized process parameter set is used as the production call parameter of the target new excipient to guide the execution of production control.

[0016] Through the integrated application of key technologies such as preparation database construction, capsule analysis modeling, optimization search strategy, cost evaluation network, etc., a complete new excipient screening and calling solution is formed, which fully utilizes data resources to achieve precise, automated and intelligent screening and calling of new excipients, effectively improving the production efficiency and quality level of capsule preparations.

[0017] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0018] Embodiment 1

[0019] like Figure 1 As shown, the embodiment of the present application provides a method for screening and calling new excipients for preparations based on usage requirements, the method comprising:

[0020] S1: Establishing a preparation database of new excipients, wherein the preparation database stores preparation process parameters of N new excipients, and each new excipient corresponds to a plurality of preparation process parameters, and each preparation process parameter has a performance requirement identifier, and the performance requirement identifier includes its own performance identifier and an inclusion requirement identifier;

[0021] In the embodiment of the present application, all new auxiliary materials and their preparation process parameters are summarized and stored in a database in a structured manner to form a preparation database of new auxiliary materials. The preparation database of new auxiliary materials thus formed covers N kinds of new auxiliary materials and their preparation process parameters, where N is the number of types of new auxiliary materials.

[0022] At the same time, in the preparation database, the preparation process parameters of each new excipient are marked with a performance requirement mark. The performance requirement mark includes an inherent performance mark and an inclusion requirement mark. Among them, the inherent performance mark represents the performance indicators of the new excipient itself prepared by the preparation process parameters, such as purity, particle size distribution, bulk density, etc.; the inclusion requirement mark represents the applicability of the new excipient prepared by the preparation process parameters in the drug preparation, such as compatibility with the drug, disintegration time of the preparation, etc.

[0023] By systematically collecting and identifying the preparation process parameters of new excipients, a preparation database with wide coverage and comprehensive content is established, laying a solid data foundation for the subsequent screening and calling of new excipients.

[0024] S2: Establishing a capsule analysis model, using the capsule analysis model to analyze the performance requirements of the target new excipient, establishing a performance requirement set, using the performance requirement set to perform feature similarity matching in the preparation database, sorting in order of matching values, and establishing a search space;

[0025] In the embodiment of the present application, an analysis model capable of comprehensively evaluating capsule preparations, i.e., a capsule analysis model, is established to realize the inference from capsule preparations to the performance requirements of new excipients. First, the input items of the capsule analysis model, i.e., the various requirements of the capsule preparations, and the output items, i.e., the various performance requirements that the new excipients need to meet, are determined. Among them, the input items may include attributes such as drug release curves and disintegration time limits, and the output items may include performance parameters such as excipient particle size, specific surface area, and solubility. Data on the relationship between capsule prescriptions, process parameters, and preparation performance are widely collected, and the collected data are cleaned, formatted, and standardized to establish a structured data set. Based on the established data set, features that have a significant impact on capsule performance and excipient performance, such as the type, dosage, particle size, solubility, etc. of excipients, are extracted to form a comprehensive feature space that fully reflects the intrinsic relationship between capsule formulations, excipient properties, and preparation performance. On the basis of the feature space, machine learning algorithms, such as regression analysis, neural networks, etc., are used to establish a quantitative mapping relationship between capsule preparation requirements and excipient performance requirements, that is, through algorithm training, learn how to infer the various performance parameters that the excipients need to meet from a given set of capsule preparation requirements. The collected data is divided into training sets and test sets. The training set data is used to estimate parameters and optimize learning of the constructed capsule analysis model. The test set data is used to evaluate the performance of the trained model to examine its reasoning accuracy and generalization ability. Based on the model verification results, the model's feature selection, algorithm selection, and parameter setting are further optimized, and the cycle is repeated until the model performance meets the expected requirements and the capsule analysis model is obtained.

[0026] Then, clarify the current capsule preparation requirements, such as the range of drug release curve, the upper limit of disintegration time, the lower limit of content uniformity, etc. Transmit it as model input to the capsule analysis model, start the model's analytical reasoning program, and the model will infer the excipient performance required to achieve the current capsule preparation requirements based on the internally constructed capsule preparation-excipient performance association, and obtain the output results. Extract the performance requirements of the target new excipients from the output results of the capsule analysis model, such as particle size distribution range, specific surface area, solubility, etc. Summarize and format the performance requirements of all target new excipients inferred by the capsule analysis model to generate a structured performance requirement set, which includes the name of each requirement indicator, required value or range, importance weight and other attributes, and sort them according to priority.

[0027] Next, similarity metrics are selected according to the characteristics of the performance requirements of new excipients, such as Euclidean distance, Manhattan distance, cosine similarity, etc. The excipient characteristics in the performance requirement set and the preparation database are standardized by using maximum and minimum value standardization, zero mean standardization, etc., and the indicators of different dimensions and scales are unified into the same metric space, eliminating the influence of the indicator dimension, making the matching results more reasonable and reliable. The similarity between each indicator in the performance requirement set and the corresponding indicator in the preparation database is calculated separately to obtain the single indicator similarity. On the basis of the single indicator similarity, the comprehensive similarity between each candidate excipient and the performance requirement set is calculated in combination with the importance weight of each indicator as the matching value to comprehensively evaluate the overall satisfaction of the excipients with various performances. According to the size of the matching value, the N new excipients in the preparation database are arranged to obtain a matching ranking table. The higher the ranking of the excipient, the higher the degree of matching between its characteristics and the performance requirement set, and the more likely it is to meet the current capsule preparation requirements for new excipients. From the matching ranking table, select the auxiliary materials with the highest ranking according to the preset number of selected materials, and construct a high-dimensional parameter space as the search space based on the various preparation process parameters corresponding to these auxiliary materials in the preparation database. Each dimension of the search space corresponds to a process parameter, such as reaction temperature, pH value, feed ratio, etc., and its value range is determined by the distribution of the corresponding parameters in the preparation database.

[0028] S3: randomly distributing N initial solutions, establishing an initial solution set, and performing an optimal search for preparation process parameters in the search space based on the initial solution set;

[0029] In an embodiment of the present application, in the established search space, N initial solutions are randomly selected as iterative starting points, where N is a positive integer, indicating the number of initial solutions, which is set according to the size of the search space and the computing resources. The N initial solutions constitute an initial solution set, and each initial solution contains a set of auxiliary material preparation process parameters. Then, a search strategy such as a simulated annealing algorithm, a genetic algorithm, and a particle swarm optimization algorithm is adopted to select N initial solutions as the starting point and carry out a local optimization search in its neighborhood. By iteratively optimizing the preparation process parameters of the N initial solutions, the corresponding auxiliary material performance is gradually approached to the performance requirement set. At the same time, in the search process, random perturbations and jump-out mechanisms are introduced to avoid the risk of falling into local optimality and improve the robustness and global convergence of the search. Moreover, in the search process, multi-point parallel search is adopted to increase the probability of finding the global optimal solution. After completing the preset number of iterations, the optimal solution obtained by searching at each iteration starting point is summarized as the iterative result of the optimization search.

[0030] S4: Search constraints through the cost balance evaluation network and establish a process parameter set based on the iterative results of the optimal search;

[0031] In the embodiment of the present application, the optimal solution obtained by searching for each iterative starting point is extracted, and the preparation process parameters of the N groups of auxiliary materials contained therein are extracted, that is, the iterative results of the optimization search. Secondly, each group of preparation process parameters is evaluated using a pre-constructed cost balance evaluation network. Among them, the cost balance evaluation network is based on a large amount of process practice data and empirical models, and a quantitative relationship between each cost element in the preparation process of auxiliary materials is established, which can comprehensively evaluate the cost rationality and economic feasibility of a group of preparation process parameters. The preparation process parameters are input into the cost balance evaluation network to obtain their cost balance scores. The higher the score, the better the cost-effectiveness of the process scheme. Again, according to the cost balance score, the preparation process parameters are screened, the preparation process parameters with lower cost balance scores are eliminated, and the preparation process parameters with higher scores are retained as the optimal solution, so as to establish a process parameter set.

[0032] By introducing the economic constraints of the cost balance evaluation network, the preparation process parameters obtained by the optimal search are evaluated and screened to ensure the cost rationality and feasibility of the process parameter set and avoid the risk of blindly pursuing performance while ignoring cost.

[0033] S5: Using the process parameter set as the production call parameter of the target new auxiliary material to perform production control.

[0034] In the examples of the present application, the obtained process parameter set is applied to production practice to guide and control the large-scale preparation process of the new auxiliary material to ensure the performance requirements and cost targets of the new auxiliary material product.

[0035] First, according to the requirements of the process parameter set, set the process conditions and operating parameters of the new excipient production line, such as reaction temperature, pressure, feed ratio, mixing time, etc., compile a complete production operation procedure and quality control standards, and clarify the control points and allowable range of the process steps. Secondly, set up online monitoring and automatic control devices in the key links of new excipient production to collect process parameter data such as reaction temperature, pH value, feed amount, etc. in real time, and compare them with the target values ​​of the process parameter set. Once it is found that the deviation exceeds the preset threshold, the automatic adjustment mechanism is immediately started to adjust the process parameters to restore the production process to the optimal state to ensure the consistency of the quality of the new excipient product. Thirdly, the new excipient products produced are systematically tested and evaluated for quality, including physical and chemical indicators, impurity spectrum, stability, etc., and compared with the expected performance requirements of the process parameter set. For batches of products whose test results meet the standards, they are released and their process parameters are collected as excellent process references; for batches of products that fail the test, analyze the reasons for their deviations, and make targeted adjustments and improvements to the process parameters, so as to achieve production control of new excipients, improve the accuracy of new excipient screening and calling, and meet the use requirements of capsule preparations.

[0036] Furthermore, the embodiment of the present application also includes:

[0037] Establishing the importance ratio of each parameter in the performance requirement set, performing a comprehensive evaluation of the matching value based on the importance ratio, and establishing a basic search space in the order of the comprehensive evaluation results;

[0038] Perform matching and sorting of independent parameters, and establish an additional search space with single parameter identification according to the matching and sorting results;

[0039] A search space establishment is completed based on the basic search space and the additional search space.

[0040] In a preferred embodiment, another method for establishing a search space is provided. First, according to the role of the target new excipient in the capsule preparation, the weight of the influence of each parameter in the performance requirement set on the comprehensive performance of the excipient is determined as the importance ratio of each parameter in the performance requirement set. Among them, the parameters closely related to the quality attributes of the preparation, such as solubility, particle size distribution, etc., have a higher importance ratio; while the parameters that have a smaller impact on the performance of the preparation, such as residual solvent content, heavy metal content, etc., have a lower importance ratio. Secondly, the degree of matching between the parameters of each excipient in the preparation database and the performance requirement set is calculated using measurement methods such as weighted Euclidean distance and cosine similarity, taking into full account the differences in the importance of each parameter and highlighting the influence of key parameters. All excipients are sorted from high to low according to the matching value to form an orderly basic search space, which reflects the degree of matching between the comprehensive performance of the excipient and the performance requirement set, and provides a global perspective for subsequent screening.

[0041] In order to highlight the impact of key parameters and improve the targeted search, each parameter in the performance requirement set is matched and ranked separately. Specifically, for parameter A in the performance requirement set, the matching value of each excipient in the preparation database on parameter A is calculated to obtain a matching value ranking for parameter A. Repeat this process to obtain a matching value ranking for each parameter. The matching value ranking of each parameter is regarded as an additional search space and identified with the name of the parameter. For example, the matching ranking for the parameter "particle size distribution" is marked as "particle size distribution search space". These additional search spaces reflect the degree of fit between the excipient and the performance requirements on specific parameters, providing an entry point for focusing on key performance.

[0042] Afterwards, the obtained basic search space is used as the main body, and the obtained multiple additional search spaces are used as supplements to form a hierarchical search space structure. In the subsequent search process, firstly, the preparation process parameters with a high comprehensive matching degree are selected in the basic search space; then, for the selected preparation process parameters with a high comprehensive matching degree, their performance in each additional search space is evaluated, and the preparation process parameters with excellent performance in key parameters are further selected as the key optimization objects. By combining the basic search space and the additional search space, the matching degree of the comprehensive performance is taken into account, and the influence of the key parameters is strengthened, which improves the comprehensiveness and pertinence of the search, and provides an efficient and accurate starting point and direction for the optimization search.

[0043] Through the above-mentioned search space construction method, the information of the performance requirement set is fully utilized. Through parameter importance analysis and hierarchical matching sorting, the refined organization and optimization of the search space are achieved, providing strong technical support for the development of new auxiliary material calls.

[0044] Furthermore, the embodiment of the present application also includes:

[0045] Generate optimization constraints for parameters based on the importance ratios;

[0046] Performing iterative optimization of the preparation process parameters in the basic search space with the initial solution set, and performing constraint supervision of the iterative process through the optimization constraint;

[0047] When any iterative step cannot satisfy the optimization constraint, the additional search space is called according to the corresponding parameters to perform the next round of single parameter optimization to complete the optimization search.

[0048] In a preferred embodiment, another optimization search is proposed based on a method of constructing a search space using a basic search space and an additional search space.

[0049] First, the importance ratio of each parameter in the performance requirement set is used as the weight of the optimization constraint. For parameters with higher importance ratios, stricter constraints are imposed; for parameters with lower importance ratios, looser constraints are imposed. For example, for the particle size distribution parameter with an importance ratio of 0.3, its optimized value must fall within the range specified by the performance requirement set; and for the residual solvent amount parameter with an importance ratio of 0.1, its optimized value is constrained to be within ±10% of the value specified in the requirement set. Secondly, the optimization constraints of each parameter are mathematically expressed and quantified to form a set of constraint equations in the form of inequalities or equations, clarifying the boundary conditions and feasible domains for the optimization of each parameter, and providing clear optimization directions and constraints for optimal search.

[0050] Then, the randomly generated N initial solutions are used as the starting point for the optimization search, and each initial solution corresponds to a set of preparation process parameters. For each initial solution, local optimization is performed in the basic search space. Optimization algorithms such as pattern search and gradient descent are used to iteratively update the process parameters to continuously improve the matching degree between the comprehensive performance of the auxiliary materials corresponding to the preparation process parameters and the performance requirement set. After each round of iterative update, the updated preparation process parameters are substituted into the constructed constraint equation for inspection. If all optimization constraints are met, the round of iteration is considered to be valid and the next round of iteration is continued. If there is a violation of the optimization constraint in this round of iteration, that is, when the optimization constraint cannot be met, the basic search space is jumped out and the direction of improvement is explored in the additional search space. Specifically, when any iterative step cannot meet the optimization constraint, the parameter that causes the violation of the optimization constraint is identified, and the additional search space corresponding to the parameter is retrieved. In the additional search space, the parameter is focused on for targeted single-item optimization, and the climbing method, taboo search and other strategies are adopted to optimize the parameter while keeping other parameters unchanged. Once an improved solution that satisfies the optimization is found in the additional search space, it is brought back to the basic search space to replace the original solution that violates the optimization constraints, and this is used as a new starting point to continue comprehensive optimization iterations. This cycle repeats until the predetermined termination conditions are met and the optimization search is completed. By introducing optimization constraints, the optimization search process is effectively supervised and standardized, ensuring the priority and precision requirements of important parameter optimization, avoiding blind searches, and limiting the feasible domain of the preparation process parameters, accelerating search convergence, and improving optimization efficiency.

[0051] By utilizing the global perspective of the basic search space and the local focus of the additional search space for optimal search, we can achieve hierarchical progression and dynamic guidance of the optimization process, and use optimization constraints for embedded supervision to ensure the priority and accuracy of parameter optimization. By elastically switching the additional space, we can break through the limitations of local optimality and improve the global optimization capability.

[0052] Furthermore, the embodiment of the present application also includes:

[0053] During the search iteration process, each round of search iteration generates a new solution;

[0054] Set solution set preservation constraints that shrink as the number of iterations increases;

[0055] When it is determined that the iteration is in the middle of the iteration, the newly added solution is evaluated for retention in the initial solution set by using the solution set retention constraint, and the initial solution set is updated with the retention evaluation result;

[0056] The optimization search is completed based on the updated results of the initial solution set.

[0057] In a preferred embodiment, in the iterative process of optimal search, for each iteration starting point, each round of iteration will generate a new candidate solution, called a newly added solution, which reflects the local optimization result of the current search position. The matching degree of its performance index with the performance requirement set determines whether it is included in the initial solution set to update the search starting point. Then, a solution set retention constraint that shrinks dynamically with the increase of the number of iterations is designed to determine the performance index threshold that the newly added solution needs to meet to enter the initial solution set. For example, in the early stage of the search iteration, the solution set retention constraint is relatively loose, and the performance index of the newly added solution only needs to reach 80% of the performance requirement set to be included in the initial solution set, or the newly added solution is directly added to the initial solution set, which helps to expand the search range and improve the diversity and globality of the search. As the search continues to deepen, the solution set retention constraint is gradually tightened. In the middle of the search iteration, the performance index of the newly added solution needs to reach 90% or even 95% of the performance requirement set to be included in the initial solution set, so as to gradually eliminate inferior solutions while expanding the search range, focus on high-quality solution areas, and balance the breadth and depth of the search. In the later stages of the search, the solution set retention constraints are further tightened. Only new solutions with performance indicators better than the worst solution in the initial solution set are eligible to enter the initial solution set, ensuring that the search continues to deepen into the area of ​​better solutions and accelerates the location of the global optimal solution. The judgment of whether the search enters the early, mid, and late stages of the iteration is determined based on a combination of factors such as the number of iterations, the magnitude of performance indicator improvement, and the frequency of solution set updates. For example, when the number of iterations reaches 1 / 3 to 1 / 2 of the total preset number, and the frequency of solution set updates drops significantly, it can be considered to have entered the mid-stage of the search iteration.

[0058] When it is determined that the iteration is in the middle of the iteration, the newly added solutions generated in each round of iteration are substituted into the solution set retention constraints for testing to obtain the retention evaluation results. If the newly added solution meets the current solution set retention constraints, it is added to the initial solution set and the initial solution set is updated; if the newly added solution does not meet the solution set retention constraints, it is discarded. Through the dynamic screening of the solution set retention constraints, the initial solution set can be continuously updated and optimized, low-quality solutions are gradually eliminated, high-quality solutions are retained and strengthened, and the search focus gradually shifts to the optimal solution area.

[0059] After dynamic screening and updating of solution set retention constraints, the initial solution set has evolved from a randomly dispersed state to a cluster distribution that gathers high-quality solution areas, and the updated results of the initial solution set are obtained. Starting from the updated initial solution set, the optimization search is restarted, and refined local exploration is carried out in the high-quality solution area to further improve the quality of the solution and the convergence speed. At the same time, the solution set retention constraints continue to be applied to screen and update the initial solution set with more stringent standards to ensure that the search continues to approach the global optimal solution. When the number of iterations reaches the preset upper limit, or the initial solution set fails to be substantially updated after several consecutive iterations, the search process terminates, and the solution in the current initial solution set is output as the global optimal solution to complete the optimization search.

[0060] By introducing the adaptive search guidance mechanism of solution set preservation constraint, a dynamic balance between search breadth and search depth is achieved at different stages of the search, and the quality of the initial solution set is continuously optimized and improved. Through the dynamic update of the solution set, the search focus can be gradually guided from the random state to the optimal solution area, thereby improving the accuracy and efficiency of the search.

[0061] Furthermore, the embodiment of the present application also includes:

[0062] When the iteration is determined to be in the late stage of iteration, an elimination constraint is generated;

[0063] Based on the elimination constraints, the initial solution set and the newly added solutions are eliminated to complete the optimal search.

[0064] In a feasible implementation, when the optimization search enters the late stage of iteration, an elimination constraint mechanism is introduced to further guide and optimize the search process. When the number of iterations reaches more than 2 / 3 of the total preset number of times, and the performance improvement of consecutive iterations is lower than the preset threshold, it can be regarded as entering the late stage of iteration and the elimination constraint mechanism is activated.

[0065] Specifically, we design elimination constraints, define the strict requirements that the solution performance indicators must meet, and the elimination mechanism when the requirements are not met. For example, we set the matching degree between the solution and the performance requirement set to be better than 95% of the current optimal solution, otherwise it will enter the elimination queue; or we require that the performance improvement of the solution in multiple iterations must exceed a certain threshold, otherwise it will be considered as search stagnation and eliminated.

[0066] When the iteration is in the late stage, in each round of iteration, first, each solution in the initial solution set is evaluated by applying the elimination constraint. For solutions that do not meet the elimination constraint, they are removed from the initial solution set and added to the elimination queue; for solutions that meet the elimination constraint, they are retained in the initial solution set and used as the starting point for the next round of search. Secondly, for the new solutions generated by the current iteration, the elimination constraint is also applied for evaluation. If the new solution meets the elimination constraint, it is compared with the solution in the initial solution set and replaced with the initial solution with poor performance; if the new solution does not meet the elimination constraint, it is directly discarded and not included in the search process.

[0067] By introducing the elimination constraint mechanism, both the initial solution set and the newly added solutions are subject to strict quality screening and the pressure of survival of the fittest. Low-quality solutions are constantly eliminated, high-quality solutions are retained and updated, and the search process continues to converge to the area of ​​better solutions. As the search deepens, the area of ​​high-quality solutions continues to shrink, and the search focus is gradually locked near the optimal solution until the termination condition is met, the global optimal solution is output, and the optimization search is completed.

[0068] Further, such as Figure 2As shown, the embodiment of the present application also includes:

[0069] When any new solution is generated, the new solution is input into the cost balance evaluation network;

[0070] Call the duration cost evaluation subnetwork to perform the first constraint evaluation of the newly added solution;

[0071] Call the process cost evaluation subnetwork to perform the second constraint evaluation of the newly added solution;

[0072] The first constraint evaluation and the second constraint evaluation are integrated and output to complete the search constraint.

[0073] In a preferred embodiment, the cost balance evaluation network is a composite neural network composed of sub-networks for different cost factors, including a time cost evaluation sub-network and a process cost evaluation sub-network, which evaluates the feasibility and superiority of new solutions from the perspectives of time cost and process cost, respectively, and achieves balanced optimization among cost factors through information interaction and feedback between networks.

[0074] Whenever any new solution is added, it is input into the cost balance evaluation network, and the cost balance evaluation network calls the time cost evaluation subnetwork to input the new solution into the time cost evaluation subnetwork. The time cost evaluation subnetwork adopts a long short-term memory neural network structure, receives the process parameter sequence of the new solution, analyzes the duration and connection efficiency of each process parameter implementation step, and predicts the total time cost of the production implementation corresponding to the solution. Then, combined with historical production data and empirical models, the total time cost of the new solution is judged and a normalized time score is given. The time score of the new solution is judged against the preset time score threshold as the first constraint evaluation of the new solution. Among them, the preset time score threshold is set in advance by the expert group. For example, if the time constraint score of the new solution is greater than or equal to the preset time threshold, it is considered that the new solution does not meet the time cost constraint; if the time constraint score of the new solution is less than the preset time threshold, it is considered that the new solution meets the time cost constraint.

[0075] At the same time, the cost balance evaluation network calls the process cost evaluation subnetwork and inputs the newly added solution into the process cost evaluation subnetwork. The process cost evaluation subnetwork adopts a convolutional neural network structure, takes the preparation process parameters of the newly added solution as input, and extracts the key cost characteristics of the process route, such as raw material consumption, energy consumption level, equipment loss, etc. through multi-layer convolution and pooling operations. On the basis of feature extraction, the fully connected layer is used to predict and judge the total process cost and give a normalized process score. The process score of the newly added solution is judged against the preset process score threshold as the second constraint evaluation of the newly added solution. Among them, the preset process score threshold is set in advance by the expert group. For example, if the process score of the newly added solution is greater than the preset process score threshold, it means that the production process cost corresponding to the solution is too high, and the newly added solution does not meet the process cost constraint; on the contrary, if the process score of the newly added solution is less than or equal to the preset process score threshold, it means that the solution has an advantage in process cost and the newly added solution meets the process cost constraint.

[0076] Subsequently, the weight coefficients of the first constraint evaluation and the second constraint evaluation are set, and the deviation between the duration score and the preset duration score threshold and the deviation between the process score and the preset process score threshold are weighted and summed to obtain the comprehensive cost constraint score, which quantifies the comprehensive performance of the new solution in terms of time and process cost. The larger the comprehensive cost constraint score, the higher the cost of the new solution. Afterwards, the new solution is selected based on the comprehensive cost constraint score to complete the search constraint.

[0077] Through the iterative constraints of the cost balance evaluation network, the optimization search process is guided by multi-angle and all-round cost optimization, which takes into account both the performance of auxiliary materials and the process costs to ensure the economic feasibility of new auxiliary materials.

[0078] Furthermore, the embodiment of the present application also includes:

[0079] Performing production proofing control with the process parameter set;

[0080] Steady-state evaluation of production based on proofing results;

[0081] Taking the steady-state evaluation results as additional influences, a comprehensive evaluation and ranking of the process parameter sets is performed;

[0082] The production parameters are obtained by screening according to the comprehensive evaluation sorting results.

[0083] In a feasible implementation, multiple alternative process parameter sets are selected, and various operating parameters and control indicators of the production line, such as temperature, pressure, time, feed ratio, etc., are set according to the requirements of the process parameter sets. Start the production proofing process, strictly monitor and dynamically adjust the process nodes in real time according to the process parameter set, ensure that the production process runs smoothly within the feasible domain defined by the process parameter set, and obtain the proofing results, including multiple trial production new auxiliary materials corresponding to each alternative process parameter set. Then, set the steady-state evaluation content, including but not limited to the consistency of product quality, the qualified rate of key quality attributes, the balance of production rhythm, the level of defective product rate, etc. Subsequently, according to the steady-state evaluation content, analyze the data of the trial production of new auxiliary materials, judge whether the production process is in a controlled state, and whether the performance changes of the trial production of new auxiliary materials are within the allowable range, and obtain the steady-state evaluation results.

[0084] Afterwards, the steady-state evaluation result is used as an additional influencing factor to rank the alternative process parameter sets and obtain the comprehensive evaluation ranking result. If a process parameter set performs well in the steady-state evaluation, its comprehensive score is increased; if a process parameter set performs poorly in the steady-state evaluation, its comprehensive score is reduced. By introducing the steady-state evaluation result as an additional influence, the evaluation result of the process parameter set is more comprehensive and balanced, taking into account both cost-effectiveness and production stability. Subsequently, according to the comprehensive evaluation ranking results, all alternative process parameter sets are ranked from high to low, and the process parameter set with the highest ranking is selected as the production parameter as the process plan for large-scale production of new auxiliary materials.

[0085] By introducing a production practice perspective through steady-state evaluation, the feasibility and reliability of the process parameter set are comprehensively evaluated, economic benefits and quality risks are balanced, and the accuracy of new excipient screening and calling is further improved.

[0086] In summary, the method for screening and calling new excipients for preparations based on usage requirements provided in the embodiments of the present application has the following technical effects:

[0087] A preparation database for new excipients is established, wherein the preparation database stores the preparation process parameters of N new excipients, and each new excipient corresponds to multiple preparation process parameters, and each preparation process parameter is accompanied by a performance requirement identifier. The performance requirement identifier includes its own performance identifier and an included requirement identifier, thereby realizing a direct association between the process parameters and the performance requirements, and laying a data foundation for subsequent precise matching.

[0088] Establish a capsule analysis model, use the capsule analysis model to analyze the performance requirements of the target new excipients, establish a performance requirement set, use the performance requirement set to perform feature similarity matching in the preparation database, sort in order of matching values, establish a search space, and provide constraints for optimal search. Randomly distribute N initial solutions, establish an initial solution set, and conduct optimal search for preparation process parameters in the search space based on the initial solution set. Conduct optimal search in the search space to improve search efficiency while ensuring optimization quality. Perform search constraints through the cost balance evaluation network, establish a process parameter set based on the iterative results of the optimal search, and control production costs while ensuring the performance of new excipients. Use the process parameter set as the production call parameter of the target new excipient, perform production control, realize the connection between the screening results of new excipients and production practice, and improve the accuracy of new excipient screening and calling.

[0089] Embodiment 2

[0090] Based on the same inventive concept as the method for screening and calling new excipients for preparations based on usage requirements in the aforementioned embodiment, Figure 3 As shown, the embodiment of the present application provides a system for screening and calling new excipients for preparations based on usage requirements, the system comprising:

[0091] A database establishment module 11 is used to establish a preparation database of new excipients, wherein the preparation database stores preparation process parameters of N kinds of new excipients, and each new excipient corresponds to a plurality of preparation process parameters, and each preparation process parameter has a performance requirement identifier, and the performance requirement identifier includes its own performance identifier and an inclusion requirement identifier;

[0092] A search space establishment module 12 is used to establish a capsule analysis model, use the capsule analysis model to analyze the performance requirements of the target new excipient, establish a performance requirement set, use the performance requirement set to perform feature similarity matching in the preparation database, sort in order of matching values, and establish a search space;

[0093] An optimization search module 13 is used to randomly distribute N initial solutions, establish an initial solution set, and perform an optimization search for preparation process parameters in the search space based on the initial solution set;

[0094] A parameter set establishment module 14 is used to perform search constraints through a cost balance evaluation network and establish a process parameter set according to the iterative results of the optimization search;

[0095] The production control module 15 is used to use the process parameter set as the production call parameter of the target new auxiliary material to perform production control.

[0096] Furthermore, the search space establishing module 12 includes the following execution steps:

[0097] Establishing the importance ratio of each parameter in the performance requirement set, performing a comprehensive evaluation of the matching value based on the importance ratio, and establishing a basic search space in the order of the comprehensive evaluation results;

[0098] Perform matching and sorting of independent parameters, and establish an additional search space with single parameter identification according to the matching and sorting results;

[0099] A search space establishment is completed based on the basic search space and the additional search space.

[0100] Furthermore, the optimization search module 13 includes the following execution steps:

[0101] Generate optimization constraints for parameters based on the importance ratios;

[0102] Performing iterative optimization of the preparation process parameters in the basic search space with the initial solution set, and performing constraint supervision of the iterative process through the optimization constraint;

[0103] When any iterative step cannot satisfy the optimization constraint, the additional search space is called according to the corresponding parameters to perform the next round of single parameter optimization to complete the optimization search.

[0104] Furthermore, the optimization search module 13 also includes the following execution steps:

[0105] During the search iteration process, each round of search iteration generates a new solution;

[0106] Set solution set preservation constraints that shrink as the number of iterations increases;

[0107] When it is determined that the iteration is in the middle of the iteration, the newly added solution is evaluated for retention in the initial solution set by using the solution set retention constraint, and the initial solution set is updated with the retention evaluation result;

[0108] The optimization search is completed based on the updated results of the initial solution set.

[0109] Furthermore, the optimization search module 13 also includes the following execution steps:

[0110] When the iteration is determined to be in the late stage of iteration, an elimination constraint is generated;

[0111] Based on the elimination constraints, the initial solution set and the newly added solutions are eliminated to complete the optimal search.

[0112] Furthermore, the parameter set establishing module 14 includes the following execution steps:

[0113] When any new solution is generated, the new solution is input into the cost balance evaluation network;

[0114] Call the duration cost evaluation subnetwork to perform the first constraint evaluation of the newly added solution;

[0115] Call the process cost evaluation subnetwork to perform the second constraint evaluation of the newly added solution;

[0116] The first constraint evaluation and the second constraint evaluation are integrated and output to complete the search constraint.

[0117] Furthermore, the embodiment of the present application also includes a production data acquisition module, which includes the following execution steps:

[0118] Performing production proofing control with the process parameter set;

[0119] Steady-state evaluation of production based on proofing results;

[0120] Taking the steady-state evaluation results as additional influences, a comprehensive evaluation and ranking of the process parameter sets is performed;

[0121] The production parameters are obtained by screening according to the comprehensive evaluation sorting results.

[0122] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0123] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for screening and calling new excipients for preparations based on usage requirements, characterized in that: The method comprises: Establishing a preparation database of new excipients, wherein the preparation database stores preparation process parameters of N kinds of new excipients, and each new excipient corresponds to a plurality of preparation process parameters, and each preparation process parameter has a performance requirement identifier, and the performance requirement identifier includes a self-performance identifier and an inclusion requirement identifier; Establishing a capsule analysis model, using the capsule analysis model to analyze the performance requirements of the target new excipient, establishing a performance requirement set, using the performance requirement set to perform feature similarity matching in a preparation database, sorting in order of matching values, and establishing a search space; Randomly distributing N initial solutions, establishing an initial solution set, and performing an optimal search for preparation process parameters in the search space based on the initial solution set; The search constraints are performed through the cost balance evaluation network, and the process parameter set is established according to the iterative results of the optimal search; Using the process parameter set as a production call parameter for a target new auxiliary material to perform production control; The performance requirement set is used to perform feature similarity matching in the preparation database, and the matching values ​​are sorted in order of magnitude to establish a search space, including: Establishing the importance ratio of each parameter in the performance requirement set, performing a comprehensive evaluation of the matching value based on the importance ratio, and establishing a basic search space in the order of the comprehensive evaluation results; Perform matching and sorting of independent parameters, and establish an additional search space with single parameter identification according to the matching and sorting results; A search space establishment is completed based on the basic search space and the additional search space.

2. The method according to claim 1, characterized in that The randomly distributed N initial solutions are used to establish an initial solution set, and an optimal search for the preparation process parameters is performed in the search space based on the initial solution set, including: Generate optimization constraints for parameters based on the importance ratios; Performing iterative optimization of the preparation process parameters in the basic search space with the initial solution set, and performing constraint supervision of the iterative process through the optimization constraint; When any iterative step cannot satisfy the optimization constraint, the additional search space is called according to the corresponding parameters to perform the next round of single parameter optimization to complete the optimization search.

3. The method according to claim 2, characterized in that The method further comprises: During the search iteration process, each round of search iteration generates a new solution; Set solution set preservation constraints that shrink as the number of iterations increases; When it is determined that the iteration is in the middle of the iteration, the newly added solution is evaluated for retention in the initial solution set by using the solution set retention constraint, and the initial solution set is updated with the retention evaluation result; The optimization search is completed based on the updated results of the initial solution set.

4. The method according to claim 3, characterized in that The method further comprises: When the iteration is determined to be in the late stage of iteration, an elimination constraint is generated; Based on the elimination constraints, the initial solution set and the newly added solutions are eliminated to complete the optimal search.

5. The method according to claim 1, characterized in that The search constraint is performed through the cost balance evaluation network, including: When any new solution is generated, the new solution is input into the cost balance evaluation network; Call the duration cost evaluation subnetwork to perform the first constraint evaluation of the newly added solution; Call the process cost evaluation subnetwork to perform the second constraint evaluation of the newly added solution; The first constraint evaluation and the second constraint evaluation are integrated and output to complete the search constraint.

6. The method according to claim 1, characterized in that The method further comprises: Performing production proofing control with the process parameter set; Steady-state evaluation of production based on proofing results; Taking the steady-state evaluation results as additional influences, a comprehensive evaluation and ranking of the process parameter sets is performed; The production parameters are obtained by screening according to the comprehensive evaluation sorting results.

7. A system for screening and calling new excipients for preparations based on usage requirements, characterized in that: A system for implementing the method for screening and calling new excipients for preparations based on usage requirements according to any one of claims 1 to 6, the system comprising: A database establishment module, the database establishment module is used to establish a preparation database of new excipients, wherein the preparation database stores preparation process parameters of N kinds of new excipients, and each new excipient corresponds to a plurality of preparation process parameters, and each preparation process parameter has a performance requirement identifier, and the performance requirement identifier includes a self-performance identifier and an inclusion requirement identifier; A search space establishment module, which is used to establish a capsule analysis model, use the capsule analysis model to analyze the performance requirements of the target new excipient, establish a performance requirement set, use the performance requirement set to perform feature similarity matching in the preparation database, sort in order of matching values, and establish a search space; An optimization search module, the optimization search module is used to randomly distribute N initial solutions, establish an initial solution set, and perform an optimization search for preparation process parameters in the search space based on the initial solution set; A parameter set establishment module, the parameter set establishment module is used to search constraints through a cost balance evaluation network and establish a process parameter set according to iterative results of the optimization search; A production control module, the production control module is used to use the process parameter set as a production call parameter of the target new auxiliary material to perform production control; The search space establishment module includes the following execution steps: Establishing the importance ratio of each parameter in the performance requirement set, performing a comprehensive evaluation of the matching value based on the importance ratio, and establishing a basic search space in the order of the comprehensive evaluation results; Perform matching and sorting of independent parameters, and establish an additional search space with single parameter identification according to the matching and sorting results; A search space establishment is completed based on the basic search space and the additional search space.

Images