A traditional Chinese medicine new drug research and development conversion decision method, system and electronic equipment

Abstract

The application relates to the technical field of research and development decision, and discloses a traditional Chinese medicine new drug research and development conversion decision method, a system and an electronic device, the method comprising the following steps: analyzing input decision text to obtain a feature vector; dynamically scheduling relevant link execution processing based on feature matching with four predefined processing links; integrating link results and performing verification; performing research and development path verification and comprehensive quality evaluation on the integrated results; and outputting a decision report or triggering a feedback correction process according to the results. The application realizes accurate link calling through semantic understanding and intelligent scheduling, and ensures that decision basis is complete, consistent and reliable through multiple verification and closed-loop optimization, so that the efficiency of traditional Chinese medicine new drug research and development conversion decision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of research and development decision-making, in particular to a traditional Chinese medicine new drug research and development conversion decision-making method and system and electronic equipment. BACKGROUND

[0002] Traditional Chinese medicine, as a traditional medical method and an important part of traditional Chinese medicine, has thousands of years of rich clinical experience and unique theoretical system, and plays an important role in the prevention, treatment and rehabilitation of diseases. However, with the rapid development of modern medicine, traditional Chinese medicine faces many challenges in clinical research and drug research and development conversion. The traditional clinical research method of traditional Chinese medicine is not scientific and rigorous enough, resulting in uneven evidence quality, which to some extent limits the international recognition and promotion value of traditional Chinese medicine.

[0003] The traditional research and development conversion mode of traditional Chinese medicine mainly relies on expert experience and scattered clinical data, and lacks systematic and standardized research methods. This mode has inherent defects such as scattered data, low processing efficiency and strong subjectivity of decision-making, which is difficult to meet the requirements of modern evidence-based medicine for scientificity, rigor, repeatability and decision-making accuracy. The research and development process of traditional Chinese medicine itself is a complex process involving multiple disciplines and multiple links, covering medicinal material source research, pharmacological analysis, dosage form development, clinical conversion path planning and drug use scheme decision-making. Under the traditional mode, these links are usually completed independently by different teams or institutions, resulting in difficulties in data integration and poor information flow. At the same time, the decision-making process highly depends on manual experience and is highly subjective, which is prone to bias and error. This scattered and inefficient research and development mode not only restricts the modernization and internationalization process of traditional Chinese medicine, but also affects the reliability and promotion value of its clinical application. Therefore, how to use modern technical means to improve the efficiency and scientificity of traditional Chinese medicine clinical evidence-based drug research and development conversion decision-making has become a problem to be solved. SUMMARY

[0004] In view of this, the present application provides a traditional Chinese medicine new drug research and development conversion decision-making method, system and electronic equipment, which aims to solve the problems of low efficiency, insufficient flexibility, limited information mining depth and limited expandability of traditional Chinese medicine new drug research and development conversion decision-making in the prior art.

[0005] In one aspect, the present application provides a traditional Chinese medicine new drug research and development conversion decision-making method, comprising:

[0006] obtaining a decision-making text input by a user, performing normalization processing, keyword extraction and semantic analysis on the decision-making text to obtain a decision-making text feature vector;

[0007] The four processing links include a traditional Chinese medicine data management link, a traditional Chinese medicine drug research and development technology link, a traditional Chinese medicine drug conversion path link, and a traditional Chinese medicine decision application scene link. The association degree between the decision text feature vector and the link feature vector corresponding to the four processing links is calculated respectively, and the required processing link set is determined from the four processing links based on the association degree.

[0008] The processing flow of any processing link in the processing link set is executed to obtain the processing result of each link, and the processing results of all links are uniformly formatted and merged to obtain a merged result. The integrity and consistency of the merged result are checked;

[0009] The path verification result is obtained by verifying the converted path of the merged result. The quality evaluation result is obtained by synchronously evaluating the comprehensive quality of the merged result. The matching state and qualified state of the merged result are determined based on the path verification result and the quality evaluation result;

[0010] If the path verification result is matched and the quality evaluation result is qualified, a decision report is output. If the path verification result is not matched or the quality evaluation result is unqualified, the problem points in the merged result are located and feedback instructions are generated. Based on the feedback instructions, the process of the related processing link is re-executed and the processing result is updated, and the merging, path verification and quality evaluation are re-performed. If the path verification result and the quality evaluation result still do not meet the qualified condition after a continuous set of repeated execution times, a decision report containing unqualified reasons and missing item list is output.

[0011] Further, when the decision text is standardized, key words are extracted and semantic analysis is performed to obtain a decision text feature vector, including:

[0012] The stop words in the decision text are removed and the punctuation is standardized to obtain a standardized decision text;

[0013] Based on the standardized decision text, key words are extracted to obtain a key word set;

[0014] Based on the key word set, entity recognition and intent recognition are performed on the standardized decision text to obtain entity recognition results and intent recognition results. The entity recognition results include prescription entity, disease entity, drug component entity, task type entity and research stage entity. The intent recognition results include evidence collection and analysis intent, drug efficacy and safety evaluation intent, and conversion path planning intent;

[0015] The key word set, entity recognition result and intent recognition result are structured and vectorized to obtain a decision text feature vector.

[0016] Furthermore, when calculating the correlation between the feature vector of the decision text and the feature vectors of the four processing stages, and determining the set of processing stages, the following steps are included:

[0017] Predefine capability description texts for the TCM data management link, TCM drug research and development technology link, TCM drug transformation path link and TCM decision-making application scenario link respectively;

[0018] The capability description text is vectorized to obtain the corresponding link feature vector;

[0019] Based on the vector space similarity measurement method, the correlation degree between the feature vector of the decision text and the feature vector of each processing stage is calculated respectively.

[0020] Based on the degree of relevance of each processing step, the four processing steps are ranked according to their relative relevance. Then, based on the distribution of the degree of relevance among the multiple processing steps, the processing steps with a high degree of relevance to the decision text requirements are determined as the steps to be invoked.

[0021] When the correlation degree meets the basic matching conditions, the process to be called is filtered by referring to the preset matching threshold to form the set of processing processes.

[0022] Furthermore, when executing the corresponding processing flow for any processing step in the set of processing steps to obtain the set of processing results, it includes:

[0023] Based on the feature vector of the processing step, determine the processing rule set corresponding to the processing step;

[0024] Based on the processing rule set, data content matching the processing steps is filtered from data resources related to the decision text to obtain the step input dataset;

[0025] The input dataset for the aforementioned process is processed by data organization and semantic alignment.

[0026] The processed input dataset is analyzed and reasoned to obtain the processing results.

[0027] The processing results of each step are summarized to form a processing result set.

[0028] Furthermore, the process of unifying the format of the processing result set and obtaining a merged result, and performing integrity and consistency checks on the merged result, includes:

[0029] Obtain the processing results corresponding to each processing step;

[0030] Based on a pre-defined unified data model, the processing results are subjected to field mapping and structural normalization.

[0031] The standardized processing results of each step are correlated and summarized to obtain a merged result.

[0032] The merging result is subjected to an integrity check to verify whether it covers the key information elements required for each stage corresponding to the decision text.

[0033] The merging results are subjected to consistency verification to check whether the results of different processing stages are consistent in semantic expression, conclusion orientation and logical relationship.

[0034] Further, when performing path verification on the verified merged result to obtain a path verification result, and determining whether the merged result matches the preset R&D transformation path, the process includes:

[0035] Extract a set of path elements from the merged results. The set of path elements includes research stage elements, evidence type elements, technology maturity elements, and transformation condition elements.

[0036] The stage affiliation and sequential relationship of each element in the path element set are identified to obtain the actual R&D transformation path expression;

[0037] The actual R&D transformation path is compared and analyzed with the preset standard stage evolution relationship to determine whether there are any missing, jumps or contradictions in the path elements in terms of stage order, element completeness and logical connection.

[0038] If the actual R&D transformation path is expressed in a reasonable order of stages, complete key path elements, and consistent logical connection, then the path verification result is determined to be a match.

[0039] If the actual R&D transformation path is expressed in a way that is not logically ordered, lacks key path elements, or has inconsistent logical connections, then the path verification result is determined to be a mismatch.

[0040] Furthermore, when performing a quality assessment on the merged results to obtain a quality assessment judgment, it includes:

[0041] Extract the evidence source elements, conclusion statement elements, and transformation support elements from the merged results respectively;

[0042] The source consistency and citation correlation analysis of the evidence source elements are performed. By identifying the citation relationship, supporting relationship and stage correspondence relationship between the evidence elements, it is determined whether the evidence forms a mutually supportive chain of evidence.

[0043] A conclusion convergence analysis is performed on the elements of the conclusion statement. By performing cross-stage mapping and constraint consistency verification on the conclusion elements in the output results of different processing stages, it is determined whether the conclusion is consistent in terms of directionality and constraint conditions.

[0044] A feasibility analysis is performed on the aforementioned transformation support elements. By verifying the completeness of the association and closed-loop correspondence between transformation condition elements, application scenario elements and implementation requirement elements, it is determined whether a closed-loop support relationship is formed.

[0045] If the evidence source elements meet the evidence-based correlation requirement, the conclusion expression elements meet the consistency requirement, and the transformation support elements meet the closed-loop requirement, then the quality assessment result is determined to be passed.

[0046] If any assessment step fails to meet the corresponding requirements, the quality assessment result will be deemed as unsuccessful.

[0047] Furthermore, when the quality assessment result is "pass", it includes:

[0048] A structured decision report is generated based on the merging results and output to the user through a human-computer interaction interface;

[0049] If the quality assessment result is "fail", perform the following steps:

[0050] Based on the specific circumstances where any one of the aforementioned evidence-based correlation requirements, consistency requirements, and closed-loop requirements is not met, locate the relevant problem-handling links in the merged results;

[0051] Generate optimization feedback instructions for the problem-solving process, including supplementary evidence requirements, conclusion adjustment directions, or transformation condition constraints;

[0052] Based on the optimization feedback instruction, the processing flow corresponding to the problem handling step is re-executed to update the processing results and subsequent merging results;

[0053] The integrity check, consistency check, path verification, and quality assessment are re-executed on the updated merge results.

[0054] After each output of the structured decision report or generation of a termination report for a failed result, the following is included:

[0055] The entire process data of this processing is added to the historical case library. The entire process data includes at least the decision text feature vector, the set of processing steps, the processing results of each step, the path verification results, and the quality assessment results.

[0056] Based on the historical case library, the matching model, matching criteria, or evaluation requirements are periodically optimized or the rules are updated.

[0057] Furthermore, an electronic device is characterized in that the electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the decision-making method for the research and development of new traditional Chinese medicine drugs as described in any one of claims 1-8.

[0058] Compared with existing technologies, the beneficial effects of this invention are as follows: By semantically parsing the decision text input by the user and associating and matching it with multiple processing stages in the entire process of R&D and transformation of new TCM drugs, the collaborative linkage between R&D and transformation decision-making needs and data management, R&D technology, transformation paths and application scenarios is realized; by unifying and merging the results of each processing stage and performing integrity and consistency checks, decision-making biases caused by fragmented information from different sources are avoided; and by introducing an R&D and transformation path verification and comprehensive quality assessment mechanism, a feedback re-execution process is triggered when conditions are not met, so that the decision results can be gradually improved through multiple rounds of correction, thereby improving the systematicness, reliability and traceability of R&D and transformation decisions for new TCM drugs, reducing the risk of blind decision-making, and improving the efficiency and quality of R&D and transformation.

[0059] On the other hand, this application also provides a decision-making system for the research and development of new traditional Chinese medicine drugs, used to implement the above-mentioned decision-making method for the research and development of new traditional Chinese medicine drugs, including:

[0060] The input preprocessing module is configured to acquire the decision text input by the user, perform normalization processing, keyword extraction and semantic parsing on the decision text, and obtain the decision text feature vector;

[0061] The intelligent scheduling module is configured to have four preset processing stages, which include a TCM data management stage, a TCM drug research and development technology stage, a TCM drug transformation path stage, and a TCM decision-making application scenario stage. The module calculates the correlation between the feature vector of the decision text and the feature vector of the corresponding stage of the four processing stages, and determines the set of processing stages to be called from the four processing stages based on the correlation.

[0062] The result integration and verification module is configured to execute the corresponding processing flow for any processing step in the set of processing steps, obtain the processing results of each step, unify and merge the processing results of all steps to obtain a merged result, and perform integrity verification and consistency verification on the merged result.

[0063] The R&D path verification module is configured to verify the R&D transformation path of the merged results after verification, and obtain the path verification result; simultaneously perform a comprehensive quality assessment on the merged results to obtain a quality assessment judgment result; and determine the matching status and qualified status of the merged results based on the path verification result and the quality assessment judgment result.

[0064] The report output module is configured to output a decision report if the path verification result is a match and the quality assessment result is qualified; if the path verification result is a mismatch or the quality assessment result is unqualified, it locates the problem points in the merged result and generates feedback instructions, re-executes the relevant processing steps based on the feedback instructions and updates the processing results, and re-merges, verifies the path, and assesses the quality; if the path verification result and the quality assessment result still do not meet the qualification conditions after a set number of consecutive re-executions, it outputs a decision report containing the reasons for the non-compliance and a list of missing items.

[0065] It is understandable that the aforementioned decision-making method, system, and electronic equipment for the research and development of new traditional Chinese medicine drugs have the same beneficial effects, and will not be elaborated further here. Attached Figure Description

[0066] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0067] Figure 1 A flowchart of a decision-making method for the research and development of new traditional Chinese medicine drugs is provided in an embodiment of the present invention;

[0068] Figure 2 This is a schematic diagram of an electronic device provided in an embodiment of the present invention.

[0069] Figure 3 A functional block diagram of a decision-making system for the research and development of new traditional Chinese medicine drugs provided in this embodiment of the invention. Detailed Implementation

[0070] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0071] See Figure 1 As shown, this application proposes a decision-making method for the research and development of new traditional Chinese medicine drugs, including:

[0072] S1: Obtain the decision text input by the user, perform normalization processing, keyword extraction and semantic parsing on the decision text to obtain the decision text feature vector;

[0073] S2: Four processing stages are preset, including TCM data management, TCM drug research and development technology, TCM drug transformation path, and TCM decision-making application scenario. The correlation between the feature vector of the decision text and the feature vector of the corresponding stage of the four processing stages is calculated, and the set of processing stages to be called is determined from the four processing stages based on the correlation.

[0074] S3: Execute the corresponding processing flow for any processing step in the set of processing steps, obtain the processing results of each step, unify and merge the processing results of all steps to obtain the merged result, and perform integrity and consistency checks on the merged result.

[0075] S4: Verify the R&D transformation path of the merged results after verification to obtain the path verification results; simultaneously conduct a comprehensive quality assessment of the merged results to obtain the quality assessment judgment results; determine the matching status and qualified status of the merged results based on the path verification results and the quality assessment judgment results.

[0076] S5: If the path verification result is a match and the quality assessment result is qualified, a decision report is output. If the path verification result is a mismatch or the quality assessment result is unqualified, the problem points in the merged result are located and feedback instructions are generated. Based on the feedback instructions, the relevant processing steps are re-executed and the processing results are updated. The merging, path verification, and quality assessment are then performed again. If the path verification result and the quality assessment result still do not meet the qualification conditions after the set number of repeated executions, a decision report containing the reasons for non-compliance and a list of missing items is output.

[0077] Specifically, the decision text refers to natural language text input by users that describes the needs for the research and development or transformation of new TCM drugs. Its content includes one of the following: research objectives, disease or indication focus, proposed formula or drug components, current research stage, and direction of transformational application. The standardization process for the decision text includes cleaning up invalid symbols, stop words, and non-standard expressions, and standardizing terminology. Keyword extraction is used to extract core terms representing research intentions and technological focuses from the standardized text. Semantic parsing is used to identify semantic relationships between terms and encode the text content into a decision text feature vector. These four processing stages correspond to different capabilities in the TCM new drug research and transformation process: TCM data management corresponds to the ability to organize data sources and evidence types; TCM drug research and development technology corresponds to the ability to organize research technology routes and research stage information; TCM drug transformation path corresponds to the ability to organize stage connections, technology maturity, and transformation conditions information; and TCM decision application scenario corresponds to the ability to organize application objectives, decision constraints, and output formats. By calculating the correlation between the decision text feature vector and the feature vectors of each processing stage, processing stages with a high correlation to the current decision needs are selected for subsequent processing. After executing the corresponding processing flow for each selected processing step, the processing results of each step are subjected to field mapping, structural normalization, and semantic alignment. Based on this, a combined result is obtained through correlation and aggregation. Integrity and consistency checks are performed to confirm the coverage of key information elements in the combined result and the logical consistency between the conclusions of each step. R&D transformation path verification may include extracting research stage elements and transformation condition elements from the combined result to form an R&D transformation path expression, and performing alignment analysis with standard R&D transformation path constraints to obtain path verification results. Comprehensive quality assessment may include performing consistency and closure analysis on evidence source elements, conclusion expression elements, and transformation support elements to obtain quality assessment judgment results. A decision report is output when the path verification result is a match and the quality assessment judgment result is qualified; otherwise, the problem points in the combined result are located and feedback instructions are generated, causing the relevant processing steps to be re-executed and the processing results updated. Subsequently, the merging, path verification, and quality assessment are repeated. The termination condition for repeated execution can be configured by the operating rules. For example, if both the path verification result and the quality assessment judgment result remain in a failing state within a continuous execution cycle and no state change occurs, the repeated execution is terminated, and a decision report containing the reasons for non-compliance and a list of missing items is output.

[0078] In some embodiments of this application, when performing normalization processing, keyword extraction, and semantic parsing on the decision text to obtain the decision text feature vector, the following are included:

[0079] Stop words were removed and punctuation was standardized from the decision text to obtain a standardized decision text.

[0080] Keyword set is obtained by extracting keywords from standardized decision text;

[0081] Based on the keyword set, entity recognition and intent recognition are performed on the standardized decision text to obtain entity recognition results and intent recognition results. Entity recognition results include prescription entities, disease entities, drug component entities, task type entities, and research stage entities; intent recognition results include evidence collection and analysis intent, efficacy and safety assessment intent, and translational pathway planning intent.

[0082] The keyword set, entity recognition results, and intent recognition results are structured, encoded, and quantized to obtain the decision text feature vector.

[0083] Specifically, the purpose of standardizing, extracting keywords, and semantically parsing the decision text is to transform the TCM R&D transformation needs input by users in natural language into a structured representation that can be uniformly understood and calculated by subsequent processing stages. Stop word removal and punctuation standardization involve removing functional words that contribute little to semantic understanding and unifying punctuation differences across different input texts to reduce interference from language variations in semantic analysis results, thus obtaining a more semantically stable standardized decision text. Keyword extraction, based on this, identifies core terms from the standardized decision text that centrally reflect R&D goals, technological objects, and stage characteristics, representing the main focus of the decision text. Entity recognition and intent recognition based on the keyword set further refine the text content at the semantic level. Entity recognition distinguishes and labels different types of professional objects in the text, such as prescriptions, diseases, drug components, task types, and research stages. Intent recognition determines whether the decision text focuses on evidence collection and analysis, efficacy and safety assessment, or the planning of R&D transformation paths, thereby clarifying the processing direction of the user's decision needs. Finally, by structurally encoding and quantizing the keyword set, entity recognition results, and intent recognition results, the originally discrete and heterogeneous text semantic information can be transformed into decision text feature vectors with unified dimensions and representation forms. This enables the feature vectors of subsequent processing stages to be correlated and matched, providing basic data support for the selection of processing stages and the scheduling of decision processes.

[0084] In some embodiments of this application, when calculating the correlation between the decision text feature vector and the feature vectors of the four processing stages and determining the set of processing stages, the following steps are included:

[0085] Predefine capability description texts for the TCM data management, TCM drug research and development technology, TCM drug transformation pathway, and TCM decision-making application scenarios.

[0086] The capability description text is vectorized to obtain the corresponding link feature vector;

[0087] Based on the vector space similarity measurement method, the correlation degree between the feature vector of the decision text and the feature vector of each corresponding processing stage is calculated.

[0088] Based on the degree of relevance of each processing step, the four processing steps are ranked according to their relative relevance. Then, based on the distribution of relevance among multiple processing steps, the processing steps with a high degree of relevance to the decision text requirements are identified as the steps to be invoked.

[0089] When the relevance meets the basic matching conditions, the process to be called is filtered by referring to the preset matching threshold to form a set of processing steps.

[0090] Specifically, the processing steps, originally existing as functional modules, are transformed into computable objects with stable semantic boundaries. Therefore, for the TCM data management step, TCM drug development technology step, TCM drug transformation path step, and TCM decision-making application scenario step, capability description texts corresponding to their respective capability scopes are pre-constructed. These capability description texts abstractly express the types of processing objects covered, key focuses, applicable R&D stages, and supported decision-making tasks covered by the corresponding processing step in the TCM new drug R&D transformation process. Their construction is based on the functional positioning of the processing step within the methodology, the types of data or knowledge that can be processed, existing R&D transformation process specifications, and the actual processing content that the step participated in in historical decision-making cases, thus forming a stable and reusable capability semantic representation for each processing step. Furthermore, the capability description texts are processed using the same structured encoding and vectorization method as the decision text to obtain the corresponding step feature vectors, placing them in the same vector space as the decision text feature vectors. Based on this, the correlation between the feature vector of the decision text and the feature vectors of each stage is calculated using vector space similarity measurement methods, such as similarity calculations based on vector angle, vector distance, or vector dot product. This reflects the degree of semantic matching between the current decision requirement and the capability focus of each processing stage. After calculating the correlation, the correlation results are first relatively ranked among the four processing stages, and a comprehensive analysis is performed based on the differences in the overall distribution of each correlation to identify one or more processing stages that are relatively prominently related to the decision text requirement in the current decision scenario. When multiple processing stages are all in a high range in the correlation distribution, multiple processing stages are allowed to enter the scope of invocation simultaneously to ensure the integrity of the decision analysis. Furthermore, based on the ranking and distribution analysis, the stages to be invoked are filtered by referring to a preset matching threshold. The matching threshold is used to exclude processing stages with significantly low semantic correlation to the current decision text and limited contribution to the decision. Its value can be statistically determined based on the correlation distribution corresponding to the decision scenarios that have been verified in the historical case library. Through the combined process of capability description text construction, vectorization representation, correlation calculation, relative ranking, distribution analysis, and threshold filtering, a set of processing steps adapted to the current decision text requirements is finally formed, thereby providing an interpretable and consistent scheduling basis for the specific execution of each subsequent processing step.

[0091] In some embodiments of this application, when executing the corresponding processing flow for any processing step in the processing step set to obtain a processing result set, the process includes:

[0092] Based on the feature vectors of the processing steps, determine the set of processing rules corresponding to each processing step;

[0093] Based on the processing rule set, data content matching the processing steps is selected from data resources related to the decision text to obtain the step input dataset;

[0094] Perform data organization and semantic alignment on the input dataset of each stage;

[0095] The processed input dataset is analyzed and reasoned to obtain the processing results.

[0096] The processing results of each step are summarized to form a processing result set.

[0097] Specifically, based on the feature vector of each processing stage, a set of processing rules corresponding to that stage is determined. This set of rules constrains the focus and processing boundaries of that stage during data screening, analysis, and reasoning. Its formation is based on the stage's capability description text, the types of objects it can process, and its specific responsibilities in the R&D transformation process, ensuring that different stages have a clear focus when executing their processes. Furthermore, based on the processing rule set, data content semantically matching that of the processing stage is selected from data resources related to the decision text. These resources can include structured databases, literature repositories, experimental records, historical decision cases, or transformation path information. Rule constraints limit the data scope, resulting in an input dataset that matches the needs of the current processing stage. Data organization and semantic alignment of the input dataset eliminate differences in expression, terminology, and structural levels between different data sources, ensuring semantic consistency with the processing rules of that stage, thus providing a unified data foundation for subsequent analysis. After sorting and alignment, analysis and reasoning are performed on the processed input dataset to generate output reflecting the judgment results of each processing stage. These outputs can demonstrate the analytical conclusions, supporting evidence, or interim judgments made by that stage regarding the issues addressed in the decision text. Finally, the processing results from each stage are aggregated to form a result set, providing a multi-stage collaborative input foundation for subsequent result merging, path verification, and quality assessment.

[0098] In some embodiments of this application, the processing result set of the processing stage is formatted and merged to obtain a merged result, and integrity and consistency checks are performed on the merged result, including:

[0099] Obtain the processing results corresponding to each processing step;

[0100] Based on a pre-defined unified data model, the processing results are subjected to field mapping and structural normalization.

[0101] The standardized processing results of each step are correlated and summarized to obtain a merged result.

[0102] Perform a completeness check on the merged results to verify whether they cover the key information elements required for each stage of the decision text.

[0103] Perform a consistency check on the merged results to verify whether the results from different stages are consistent in semantic expression, conclusion orientation, and logical relationship.

[0104] Specifically, due to differences in processing objectives, data sources, and analytical focuses across different processing stages, the outputs from each stage typically exhibit inconsistencies in structure, field naming, and semantic expression. Therefore, it is necessary to unify and integrate the processing results before proceeding to subsequent path verification and quality assessment. First, obtaining the processing results corresponding to each stage involves compiling the analytical conclusions, supporting information, or interim judgments generated from the TCM data management stage, TCM drug development technology stage, TCM drug transformation path stage, and TCM decision-making application scenario stage as independent result items. Then, based on a pre-defined unified data model, the above processing results undergo field mapping and structural standardization. This unified data model defines the standard expression methods for common information elements across different processing results. For example, it uses a unified field structure and semantic identifier for information such as research stage, evidence type, key conclusions, and transformation conditions, thereby eliminating differences in expression across different processing stages. After completing field mapping and structural standardization, the processing results from each stage are correlated and summarized. Information that is semantically related or points to the same decision-making objective is integrated to form a merged result that comprehensively reflects the current decision-making analysis status. Based on this, an integrity check is performed on the merged results to determine whether the merged results cover the key processing steps involved in the decision text and the corresponding core information elements, so as to avoid insufficient decision-making basis due to the absence of results from a certain processing step. At the same time, a consistency check is performed on the merged results to identify whether there are contradictions, conflicting directions or logical breaks by comparing the degree of coordination between the output results of different processing steps in terms of semantic expression, conclusion orientation and logical relationship, thereby ensuring that the merged results have a consistent and coherent decision support basis as a whole.

[0105] In some embodiments of this application, when performing path verification on the verified merged result to obtain a path verification result, and determining whether the merged result matches a preset R&D transformation path, the process includes:

[0106] Extract the path element set from the merged results. The path element set includes research stage elements, evidence type elements, technology maturity elements, and transformation condition elements.

[0107] The stage affiliation and sequential relationship of each element in the path element set are identified to obtain the actual R&D transformation path expression;

[0108] The actual R&D transformation path is compared and analyzed with the pre-set standard stage evolution relationship to determine whether there are any missing, jumps or contradictions in the path elements in terms of stage order, element completeness and logical connection.

[0109] If the actual R&D transformation path is expressed in a reasonable order of stages, complete key path elements, and consistent logical connection, then the path verification result is determined to be a match.

[0110] If the actual R&D transformation path is expressed in a way that is not logically ordered, lacks key path elements, or has inconsistent logical connections, then the path verification result is determined to be mismatched.

[0111] Specifically, the purpose of verifying the R&D transformation path of the merged results is to determine whether the overall analytical results formed by multiple processing stages conform to the reasonable evolutionary logic of new TCM drugs from research to transformation, rather than simply judging the correctness of a single conclusion. To this end, firstly, a set of path elements reflecting the state of the R&D transformation process is extracted from the merged results. Among these, the research stage element represents the R&D stage position corresponding to the current analysis content; the evidence type element represents the data or evidence form supporting the stage judgment; the technology maturity element reflects the maturity level of the relevant technology or solution in the R&D process; and the transformation condition element describes the preconditions required to achieve a stage leap or enter the application stage. Then, the stage affiliation and sequence relationship of each element in the path element set are identified. That is, based on the R&D stage attributes indicated by the elements themselves and their order of appearance in the merged results, an expression reflecting the actual R&D transformation path implicit in the current decision analysis is constructed. After obtaining the actual R&D transformation path, it is compared and analyzed with the preset standard stage evolution relationship. The standard stage evolution relationship is used to describe the reasonable connection sequence and necessary conditions between the stages in the R&D transformation of new traditional Chinese medicine drugs. Through this comparison, it can be identified whether there are situations such as the stage sequence being advanced or delayed, the key path elements being missing, or the logical connection between different stages in the actual path. When the actual R&D transformation path maintains a reasonable continuity in the stage sequence, the key path elements are complete, and the logical relationship between each element is consistent, the merged result is judged to match the preset R&D transformation path; conversely, if there are stage jumps, missing elements, or logical contradictions in the actual path, the path verification result is judged to be mismatched, thus providing a basis for subsequent feedback correction or decision termination.

[0112] In some embodiments of this application, when performing a quality assessment on the merged results to obtain a quality assessment result, the following is included:

[0113] Extract the evidence source elements, conclusion statement elements, and transformation support elements from the merged results respectively;

[0114] The source consistency and citation relevance analysis of the evidence source elements are conducted. By identifying the citation relationship, supporting relationship and stage correspondence among the evidence elements, it is determined whether the evidence forms a mutually supportive chain of evidence.

[0115] A conclusion convergence analysis is performed on the elements of the conclusion statement. By mapping the conclusion elements in the output results of different processing stages across stages and verifying the consistency of constraints, it is determined whether the conclusion is consistent in terms of directionality and constraints.

[0116] A feasibility analysis is conducted on the transformation support elements. By verifying the completeness of the correlation and closed-loop correspondence between the transformation condition elements, application scenario elements and implementation requirement elements, it is determined whether a closed-loop support relationship has been formed.

[0117] If the evidence source elements meet the evidence-based relevance requirement, the conclusion expression elements meet the consistency requirement, and the transformation support elements meet the closed-loop requirement, then the quality assessment result is deemed to be passed.

[0118] If any assessment step fails to meet the corresponding requirements, the quality assessment result will be deemed as unsuccessful.

[0119] Specifically, the merged results extract evidence source elements, conclusion expression elements, and transformation support elements. Evidence source elements characterize the data source type and background upon which the analytical conclusions of each processing stage are based. Conclusion expression elements characterize the judgment conclusions output by each processing stage and their constraints. Transformation support elements characterize the conditions, scenarios, and implementation requirements necessary to transform R&D results into practical applications. Furthermore, the evidence source elements undergo source consistency and citation correlation analysis. By identifying whether there are clear citation relationships, mutual support relationships, and matching relationships with the corresponding R&D stages among different evidence elements, it is determined whether various types of evidence can form a coherent and mutually corroborating chain of evidence, thereby avoiding conclusions based on isolated or insufficient evidence. Based on this, the conclusion expression elements undergo conclusion convergence analysis. By mapping the conclusion elements in the output results of different processing stages across stages and verifying the consistency of the applicable conditions and limiting factors attached to the conclusions, it is determined whether the conclusions formed in each processing stage are consistent in their target and constraints, avoiding situations where conclusions are scattered or conflicting. Simultaneously, a feasibility analysis is conducted on the supporting elements for transformation. By verifying whether the transformation conditions, application scenarios, and implementation requirements correspond in content and form a closed-loop relationship logically, it is determined whether the merged result has a supporting foundation for actually promoting the transformation of R&D results. When the evidence source elements can form a complete chain of evidence-based connections, the conclusion expression elements are consistent in direction and constraints, and the transformation supporting elements form a closed-loop support relationship between conditions, scenarios, and implementation requirements, the quality assessment result is deemed passed. Conversely, if any assessment link fails to meet the corresponding requirements, the quality assessment result is deemed failed and serves as the basis for subsequent feedback correction or decision termination.

[0120] In some embodiments of this application, when the quality assessment result is passed, the following is included:

[0121] A structured decision report is generated based on the merge results and output to the user through a human-computer interaction interface;

[0122] If the quality assessment result is "fail", perform the following steps:

[0123] Based on the specific circumstances where any one of the requirements for evidence-based correlation, consistency of direction, and closed-loop is not met, locate the relevant problem-handling links in the merged results;

[0124] Generate optimized feedback instructions for the problem-solving process. These instructions include requirements for supplementary evidence, adjustments to the conclusion, or constraints on transformation conditions.

[0125] Based on the optimized feedback instructions, the processing flow corresponding to the problem handling stage is re-executed to update the processing results and subsequent merging results;

[0126] Re-perform integrity checks, consistency checks, path verifications, and quality assessments on the updated merge results;

[0127] After each output of a structured decision report or generation of a termination report for a failed result, the following is included:

[0128] Add the entire process data of this processing to the historical case library. The entire process data should include at least the decision text feature vector, the set of processing links, the processing results of each link, the path verification results, and the quality assessment results.

[0129] Based on a historical case library, parameters are regularly optimized or rules are updated for the matching model, matching criteria, or evaluation requirements.

[0130] Specifically, when the quality assessment result is "pass," it indicates that the merged results meet the basic requirements for R&D transformation decisions in terms of evidence completeness, conclusion consistency, and feasibility of transformation. At this point, a structured decision report is generated based on the merged results. This structured decision report is not simply a text summary; rather, it follows a unified data model and presentation standards, hierarchically organizing and presenting information from the R&D phase, supporting evidence, key conclusions, and transformation recommendations. This information is then output to the user through a human-computer interaction interface, enabling the user to clearly understand the decision basis, the source of the conclusions, and their applicable boundaries. Conversely, when the quality assessment result is "fail," it indicates that the merged results have at least one deficiency in the evidence chain, the direction of the conclusions, or the support for transformation. In this case, the decision-making process is not directly terminated; instead, a feedback and correction mechanism is initiated. Specifically, this includes: performing reverse analysis on the merged results based on unmet requirements for evidence-based correlation, consistency of direction, and closed-loop processing to pinpoint the specific processing steps or multiple related processing steps that caused the problem, thereby clarifying the source of the problem; generating optimization feedback instructions for the problem-handling steps, which clearly indicate the types of evidence that need to be supplemented, the direction of conclusion revision, or the content that needs further limitation or improvement of transformation conditions to guide subsequent correction operations; and, based on the optimization feedback instructions, re-triggering and executing the processing flow corresponding to the problem-handling step, updating the relevant processing results, and synchronously updating the subsequent merged results; finally, further re-analyzing the updated merged results... The process involves performing integrity checks, consistency checks, path verification, and quality assessments to form a closed-loop decision-making mechanism. This improves the reliability and executability of R&D transformation decisions. Integrity checks include at least checking the existence of fields related to research stage, evidence type, key conclusions, and transformation conditions. This is used to determine whether the merged results cover the basic information elements that each processing step should provide under the current semantic meaning of the decision text. Consistency checks include at least checking the consistency of conclusions, research stage attribution, and the relationship between evidence and conclusions in the output results of different processing steps. This is used to identify whether there are conflicts in conclusions, conflicts in stage positioning, or breaks in the supporting relationship.

[0131] Regardless of whether the final output is a structured decision report or a termination report generated for cases where the quality assessment fails, all data generated during the decision-making process will be uniformly included in the historical case library for retention and management. This comprehensive data includes not only the decision text feature vectors that trigger the decision process, but also the set of actual processing steps invoked, the processing results generated by each step within its corresponding processing flow, and the resulting path verification and quality assessment results. This fully reflects the entire trajectory of a research and development transformation decision from input, processing, to output. By continuously accumulating this data in the historical case library, a foundation of case samples covering different decision-making needs, research and development stages, and transformation scenarios can be formed. Based on this, and using the case data accumulated in the historical case library, the relevant mechanisms for process matching and quality assessment are updated according to a preset maintenance cycle. This includes adjusting the parameter configurations involved in the matching model, revising the rules used to determine relevance or priority in the matching criteria, or supplementing or refining the assessment requirements used in the quality assessment process. This allows the agent to continuously revise its decision-making basis and judgment logic based on the actual performance of existing decision results, gradually improving the stability, consistency, and adaptability of subsequent research and development transformation decisions.

[0132] In another preferred embodiment based on the above embodiments, see [reference] Figure 2 As shown, this embodiment provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to realize the steps of the intelligent, integrated production method for the entire process of traditional Chinese medicine clinical practice.

[0133] Specifically, the processor can be at least one of a general-purpose processor, a central processing unit, a digital signal processor, or a multi-core processor. The processor is used to perform calculations and control the execution of instructions in a computer program. The memory can include at least one of non-volatile memory and volatile memory. Non-volatile memory includes read-only memory, flash memory, or solid-state memory. Volatile memory includes random access memory. The processor is used to store computer programs, running data, and intermediate processing results.

[0134] In another preferred embodiment based on the above embodiments, see [reference] Figure 3 As shown, this embodiment provides a decision-making system for the research and development of new traditional Chinese medicine drugs, used to implement the above-mentioned decision-making method for the research and development of new traditional Chinese medicine drugs, including:

[0135] The input preprocessing module is configured to acquire the decision text input by the user, perform normalization processing, keyword extraction and semantic parsing on the decision text, and obtain the decision text feature vector;

[0136] The intelligent scheduling module is configured to have four preset processing stages, including TCM data management, TCM drug research and development technology, TCM drug transformation path, and TCM decision-making application scenario. It calculates the correlation between the feature vector of the decision text and the feature vector of the corresponding stage of the four processing stages, and determines the set of processing stages to be called from the four processing stages based on the correlation.

[0137] The result integration and verification module is configured to execute the corresponding processing flow for any processing step in the set of processing steps, obtain the processing results of each step, unify and merge the processing results of all steps to obtain the merged result, and perform integrity verification and consistency verification on the merged result.

[0138] The R&D path verification module is configured to verify the R&D transformation path of the merged results after verification, and obtain the path verification result; simultaneously, it performs a comprehensive quality assessment on the merged results and obtains the quality assessment judgment result; and determines the matching status and qualified status of the merged results based on the path verification result and the quality assessment judgment result.

[0139] The report output module is configured to output a decision report if the path verification result is a match and the quality assessment result is qualified; if the path verification result is a mismatch or the quality assessment result is unqualified, the problem points in the merged result are located and feedback instructions are generated. Based on the feedback instructions, the relevant processing steps are re-executed and the processing results are updated, and the merging, path verification, and quality assessment are performed again. If the path verification result and the quality assessment result still do not meet the qualification conditions after a set number of consecutive re-executions, a decision report containing the reasons for non-compliance and a list of missing items is output.

[0140] It is understood that in this implementation, the decision-making process for the research and development of new TCM drugs is organized into an executable workflow in the order of input processing, selection of processing links, integration and verification of results, path verification and quality assessment, report output and feedback iteration. The input preprocessing module standardizes, extracts keywords, and performs semantic parsing on the decision text to form a decision text feature vector for subsequent correlation calculations. The intelligent scheduling module determines the set of processing steps to be called based on the correlation between the decision text feature vector and the feature vector of each processing step, thus mapping the decision requirements to the corresponding processing steps. The result integration and verification module performs field mapping, structure standardization, and merging on the processing results output by each processing step, and performs integrity and consistency verification on the merged results. The R&D path verification module performs R&D transformation path verification on the verified merged results and outputs the path verification results. At the same time, it performs comprehensive quality assessment and outputs the quality assessment judgment results. Based on the two, it determines the matching status and qualified status of the merged results. The report output module outputs a decision report when the matching status and qualified status meet the conditions. When the conditions are not met, it generates feedback instructions and triggers the re-execution of relevant processing steps, thereby realizing the iterative update of the decision results. When the termination conditions are met, it outputs a decision report containing the reasons for non-compliance and a list of missing items.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A decision-making method for the research and development and transformation of new traditional Chinese medicine drugs, characterized in that, include: The decision text input by the user is obtained, and the decision text is normalized, keywords are extracted, and semantic parsing is performed to obtain the decision text feature vector; Four processing stages are preset, namely, TCM data management stage, TCM drug research and development technology stage, TCM drug transformation path stage, and TCM decision-making application scenario stage; the correlation degree between the feature vector of the decision text and the feature vector of the corresponding stage of the four processing stages is calculated, and the set of processing stages to be called is determined from the four processing stages based on the correlation degree; For any processing step in the set of processing steps, execute its corresponding processing flow to obtain the processing results of each step, and unify and merge the processing results of all steps to obtain a merged result. Perform integrity verification and consistency verification on the merged result. The merged results after verification are used to verify the R&D transformation path and obtain the path verification results; at the same time, the merged results are used to conduct a comprehensive quality assessment and obtain the quality assessment judgment results. The matching status and qualification status of the merged result are determined based on the path verification results and the quality assessment results. If the path verification result is a match and the quality assessment result is qualified, a decision report is output. If the path verification result is not a match or the quality assessment result is unqualified, the problem points in the merged result are located and feedback instructions are generated. Based on the feedback instructions, the relevant processing steps are re-executed and the processing results are updated. Then, the merging, path verification and quality assessment are performed again. If the path verification results and quality assessment results still do not meet the qualification criteria after the set number of repeated executions, a decision report containing the reasons for non-compliance and a list of missing items will be output. When performing normalization, keyword extraction, and semantic parsing on decision text to obtain its feature vector, the process includes: The decision text is processed by stop word removal and punctuation standardization to obtain a standardized decision text; Keyword extraction is performed based on the standardized decision text to obtain a keyword set; Based on the keyword set, entity recognition and intent recognition are performed on the standardized decision text to obtain entity recognition results and intent recognition results. The entity recognition results include prescription entities, disease entities, drug component entities, task type entities, and research stage entities. The intent recognition results include evidence collection and analysis intent, efficacy and safety assessment intent, and translational path planning intent. The keyword set, entity recognition results, and intent recognition results are structured, encoded, and quantized to obtain a decision text feature vector; When calculating the correlation between the feature vector of the decision text and the feature vectors of the four processing stages, and determining the set of processing stages, the following steps are included: Predefine capability description texts for the TCM data management link, TCM drug research and development technology link, TCM drug transformation path link and TCM decision-making application scenario link respectively; The capability description text is vectorized to obtain the corresponding link feature vector; Based on the vector space similarity measurement method, the correlation degree between the feature vector of the decision text and the feature vector of each processing stage is calculated respectively. Based on the degree of relevance of each processing step, the four processing steps are ranked according to their relative relevance. Then, based on the distribution of the degree of relevance among the multiple processing steps, the processing steps with a high degree of relevance to the decision text requirements are determined as the steps to be invoked. When the correlation degree meets the basic matching conditions, the process to be called is filtered by referring to the preset matching threshold to form the set of processing processes; The processing results from each stage are formatted and merged. The merged results are then subjected to integrity and consistency checks, including: Obtain the processing results corresponding to each processing step; Based on a pre-defined unified data model, the processing results are subjected to field mapping and structural normalization. The standardized processing results of each step are correlated and summarized to obtain a merged result. The merging result is subjected to an integrity check to verify whether it covers the key information elements required for each stage corresponding to the decision text. The merging results are subjected to consistency verification to check whether the results of different processing stages are consistent in semantic expression, conclusion orientation and logical relationship. When performing path verification on the merged result after verification to obtain the path verification result, and determining whether the merged result matches the preset R&D transformation path, the following steps are taken: Extract a set of path elements from the merged results. The set of path elements includes research stage elements, evidence type elements, technology maturity elements, and transformation condition elements. The stage affiliation and sequential relationship of each element in the path element set are identified to obtain the actual R&D transformation path expression; The actual R&D transformation path is compared and analyzed with the preset standard stage evolution relationship to determine whether there are any missing, jumps or contradictions in the path elements in terms of stage order, element completeness and logical connection. If the actual R&D transformation path is expressed in a reasonable order of stages, complete key path elements, and consistent logical connection, then the path verification result is determined to be a match. If the actual R&D transformation path is expressed in a way that is not logically ordered, lacks key path elements, or has inconsistent logical connections, then the path verification result is determined to be mismatched. When performing a quality assessment on the merged results to obtain the quality assessment judgment, the following should be included: Extract the evidence source elements, conclusion statement elements, and transformation support elements from the merged results respectively; The source consistency and citation correlation analysis of the evidence source elements are performed. By identifying the citation relationship, supporting relationship and stage correspondence relationship between the evidence elements, it is determined whether the evidence forms a mutually supportive chain of evidence. A conclusion convergence analysis is performed on the elements of the conclusion statement. By performing cross-stage mapping and constraint consistency verification on the conclusion elements in the output results of different processing stages, it is determined whether the conclusion is consistent in terms of directionality and constraint conditions. A feasibility analysis is performed on the aforementioned transformation support elements. By verifying the completeness of the association and closed-loop correspondence between transformation condition elements, application scenario elements and implementation requirement elements, it is determined whether a closed-loop support relationship is formed. If the evidence source elements meet the evidence-based correlation requirement, the conclusion expression elements meet the consistency requirement, and the transformation support elements meet the closed-loop requirement, then the quality assessment result is deemed qualified. If any assessment step fails to meet the corresponding requirements, the quality assessment result will be deemed unqualified.

2. The decision-making method for the research and development of new traditional Chinese medicine drugs according to claim 1, characterized in that, When executing the corresponding processing flow for any processing step in the set of processing steps to obtain the set of processing results, it includes: Based on the feature vector of the processing step, determine the processing rule set corresponding to the processing step; Based on the processing rule set, data content matching the processing steps is filtered from data resources related to the decision text to obtain the step input dataset; The input dataset for the aforementioned process is processed by data organization and semantic alignment. The processed input dataset is analyzed and reasoned to obtain the processing results. The processing results of each step are summarized to form a processing result set.

3. The decision-making method for the research and development and transformation of new traditional Chinese medicine drugs according to claim 1, characterized in that, If the path verification result is a match and the quality assessment result is qualified, it includes: A structured decision report is generated based on the merging results and output to the user through a human-computer interaction interface; If the path verification result does not match or the quality assessment result is unqualified, perform the following steps: Based on the specific circumstances where any one of the aforementioned evidence-based correlation requirements, consistency requirements, and closed-loop requirements is not met, locate the relevant problem-handling links in the merged results; Generate optimization feedback instructions for the problem-solving process, including supplementary evidence requirements, conclusion adjustment directions, or transformation condition constraints; Based on the optimization feedback instruction, the processing flow corresponding to the problem handling step is re-executed to update the processing results and subsequent merging results; The integrity check, consistency check, path verification, and quality assessment are re-executed on the updated merge results. After each output of the structured decision report or generation of a termination report for unsatisfactory results, the following is included: The entire process data of this processing is added to the historical case library. The entire process data includes at least the decision text feature vector, the set of processing steps, the processing results of each step, the path verification results, and the quality assessment results. Based on the historical case library, the matching model, matching criteria, or evaluation requirements are periodically optimized or the rules are updated.

4. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the decision-making method for the research and development of new traditional Chinese medicine drugs as described in any one of claims 1-3.

5. A decision-making system for the research and development of new traditional Chinese medicine drugs, used to implement the decision-making method for the research and development of new traditional Chinese medicine drugs as described in any one of claims 1-3, characterized in that, include: The input preprocessing module is configured to acquire the decision text input by the user, perform normalization processing, keyword extraction and semantic parsing on the decision text, and obtain the decision text feature vector; The intelligent scheduling module is configured to have four preset processing stages, which include a TCM data management stage, a TCM drug research and development technology stage, a TCM drug transformation path stage, and a TCM decision-making application scenario stage. The module calculates the correlation between the feature vector of the decision text and the feature vector of the corresponding stage of the four processing stages, and determines the set of processing stages to be called from the four processing stages based on the correlation. The result integration and verification module is configured to execute the corresponding processing flow for any processing step in the set of processing steps, obtain the processing results of each step, unify and merge the processing results of all steps to obtain a merged result, and perform integrity verification and consistency verification on the merged result. The R&D path verification module is configured to verify the R&D transformation path of the merged results after verification, and obtain the path verification result; simultaneously, it performs a comprehensive quality assessment on the merged results to obtain the quality assessment judgment result. The matching status and qualification status of the merged result are determined based on the path verification results and the quality assessment results. The report output module is configured to output a decision report if the path verification result is a match and the quality assessment result is qualified; if the path verification result is not a match or the quality assessment result is unqualified, the module will locate the problem points in the merged result and generate feedback instructions. Based on the feedback instructions, the relevant processing steps will be re-executed and the processing results will be updated. The merge, path verification and quality assessment will be performed again. If the path verification results and quality assessment results still do not meet the qualification criteria after the set number of consecutive executions, a decision report containing the reasons for non-compliance and a list of missing items will be output.

Images