Traditional Chinese medicine uric acid reducing preparation quality evaluation method and system based on big data

Abstract

The invention discloses a traditional Chinese medicine uric acid reducing preparation quality evaluation method and system based on big data, and relates to the technical field of traditional Chinese medicine preparation analysis. The quality evaluation method comprises the following steps: S1, acquiring chromatographic characteristic data, active ingredient content data and physical and chemical detection data, performing preprocessing, and constructing a preparation quality database; s2, carrying out consistency analysis through batch detection data of characteristic peaks, and dynamically optimizing a peak identification and correction strategy; s3, fusing the batch co-occurrence frequency, the distinction degree index and the inter-batch fluctuation which are subjected to correction processing, and performing characteristic peak screening through weighted operation; s4, performing multi-dimensional fusion quantitative evaluation by integrating a high-priority characteristic peak screening result, the effective component content data and the physical and chemical detection data; and S5, based on the multi-dimensional fusion evaluation result and historical sample data iterative training and abnormal mapping, parameter optimization and dynamic updating are realized. The problems of inter-batch feature drift and insufficient feature recognition precision of the traditional Chinese medicine uric acid reducing preparation in batch detection are solved.

Description

Technical Field

[0001] This invention relates to the field of traditional Chinese medicine preparation analysis technology, specifically to a method and system for quality assessment of traditional Chinese medicine uric acid-lowering preparations based on big data. Background Technology

[0002] The quality assessment of existing traditional Chinese medicine uric acid-lowering preparations typically employs high-performance liquid chromatography (HPLC) to analyze the samples, collecting data such as chromatographic signal intensity, retention time, and peak area. This data, combined with internal standard correction and peak shape parameter analysis, identifies and compares characteristic peaks from different batches. The detection of active ingredients generally utilizes liquid chromatography-mass spectrometry (LC-MS), calibrating with standard solutions to quantitatively analyze the content of major components.

[0003] For example, the invention patent with publication number CN113917064A discloses a method for establishing an HPLC fingerprint spectrum of a heat-clearing and detoxifying oral liquid and its fingerprint spectrum, belonging to the field of traditional Chinese medicine preparation analysis. Specifically, it discloses a method for establishing an HPLC fingerprint spectrum of a heat-clearing and detoxifying oral liquid, which includes the preparation of reference and test solutions, the determination of HPLC chromatographic conditions, and the preparation of an HPLC standard fingerprint spectrum. This invention also discloses the HPLC standard fingerprint spectrum of the heat-clearing and detoxifying oral liquid obtained by this method, which has 30 common peaks. The quality detection method of this invention is simple to operate, highly stable, and reproducible. The obtained chromatogram has comprehensive characteristic peaks. By comparing the common peaks of the standard fingerprint spectrum, the quality of the heat-clearing and detoxifying oral liquid can be comprehensively evaluated and controlled, which is beneficial for accurately assessing the intrinsic quality of the preparation and ensuring the safety and effectiveness of clinical use.

[0004] For example, the invention patent with publication number CN116124937A discloses a method for constructing and applying the fingerprint spectrum of Zhangyanming tablets, which relates to the field of traditional Chinese medicine component detection and analysis technology, specifically involving a fingerprint spectrum of Zhangyanming tablets and its construction method and quality detection method. The construction method includes: extracting a Zhangyanming tablet sample to obtain a sample solution; dissolving a reference standard to obtain a reference standard solution; using the same high-performance liquid chromatography method to detect the sample solution and the reference standard solution respectively to obtain a sample spectrum and a reference spectrum; performing fingerprint spectrum similarity analysis on the sample spectrum to generate a standard fingerprint spectrum, and identifying the chromatographic peaks of the standard spectrum based on the reference spectrum to obtain the fingerprint spectrum of Zhangyanming tablets. This experiment establishes a fingerprint spectrum method for Zhangyanming tablets for the first time, screening a total of 18 characteristic common peaks for quality control and evaluation of Zhangyanming tablets, and can be used to examine production stability and batch-to-batch consistency.

[0005] However, with the large-scale development of the traditional Chinese medicine preparation industry, the number of samples and the volume of data continue to grow, making it difficult to meet the needs of large-scale, standardized quality evaluation by relying solely on manual interpretation or traditional algorithms. Existing detection methods suffer from problems such as fragmented processes, poor algorithm adaptability, and insufficient information integration in batch data collection, parameter standardization, feature fusion, and comprehensive analysis. In addition, factors such as process differences, raw material fluctuations, and differences in instrument platforms between batches also exacerbate the variability between active ingredients, physicochemical indicators, and fingerprint characteristics, affecting the scientific validity and consistency of batch quality determination.

[0006] Therefore, in order to address the above issues, there is an urgent need for a quality assessment method and system for traditional Chinese medicine uric acid-lowering preparations based on big data. Summary of the Invention

[0007] Technical problems to be solved

[0008] To address the shortcomings of existing technologies, this invention provides a method and system for quality assessment of traditional Chinese medicine uric acid-lowering preparations based on big data, which solves the problems of inter-batch feature drift and insufficient feature recognition accuracy in batch testing of traditional Chinese medicine uric acid-lowering preparations.

[0009] Technical solution

[0010] To achieve the above objectives, this invention provides the following technical solution: a method and system for quality assessment of traditional Chinese medicine uric acid-lowering preparations based on big data, comprising: S1, acquiring chromatographic characteristic data, effective ingredient content data, and physicochemical detection data and preprocessing them to construct a preparation quality database; S2, performing consistency analysis on batch detection data of characteristic peaks, dynamically optimizing peak identification and correction strategies, improving alignment accuracy, and archiving abnormal features; S3, fusing corrected batch co-occurrence frequency, discrimination index, and inter-batch fluctuations, performing characteristic peak screening through weighted calculations, and implementing characteristic peak grading screening and optimization processing based on the screening results; S4, performing multi-dimensional fusion quantitative evaluation by comprehensively considering high-priority characteristic peak screening results, effective ingredient content data, and physicochemical detection data, and achieving batch classification, re-inspection optimization, and quality archiving management based on the fusion evaluation; S5, achieving parameter optimization and dynamic updating based on iterative training and anomaly mapping using multi-dimensional fusion evaluation results and historical sample data.

[0011] Further, the specific steps for acquiring chromatographic characteristic data, effective ingredient content data, and physicochemical detection data, and performing preprocessing to construct a formulation quality database are as follows: Acquiring data on traditional Chinese medicine (TCM) uric acid-lowering preparations, including chromatographic characteristic data, effective ingredient content data, and physicochemical detection data; Acquiring chromatographic characteristic data: After sample preparation, the samples are injected into a high-performance liquid chromatograph for detection, and the chromatographic signal intensity and retention time of each batch of samples are collected in real time; The retention time, peak area, and peak shape parameters of each batch's characteristic peaks are obtained through internal standard correction and time alignment algorithms; The peak position data required for the peak position standard deviation is obtained through the retention time series detected by the chromatograph; The peak area data required for the peak area standard deviation is calculated through an integral algorithm; The peak shape data required for the peak shape parameter standard deviation is obtained through extraction and processing by a chromatography workstation for statistical analysis, including the peak shape data such as half-peak width, symmetry factor, and tailing factor; The batch co-occurrence frequency is obtained by matching the batch numbers and statistically recording the occurrence frequency of the same characteristic peak in the chromatograms of all batches; The required peak shape parameters are calculated through information such as the signal intensity change and peak area distribution of the same characteristic peak in different batches. The input values ​​are processed by a chromatography data processing system to derive the discrimination index; effective component content data is obtained: target components are detected using liquid chromatography-mass spectrometry (LC-MS), and the actual concentration of each effective component is measured through a standard solution calibration curve. Then, the effective component value is obtained through linear normalization within a unified concentration range; physicochemical test data is obtained: the moisture, pH, ash, viscosity, and density of the samples are measured using physicochemical testing equipment. All physicochemical test results are corrected for temperature and humidity and normalized to obtain physicochemical performance values; the collected data of traditional Chinese medicine uric acid-lowering preparations are processed to unify the data format, timestamps, and batch codes; all numerical parameters are scaled to a unified range using linear normalization; state parameters are mapped to calculable range values ​​according to the operating conditions to ensure consistency in dimension and scale between parameters from different sources; after standardization and normalization of all parameters, the data of traditional Chinese medicine uric acid-lowering preparations are stored according to batch number and timestamp. A preparation quality database is constructed using a hybrid architecture of distributed relational structure and log storage, and high-priority batch tables, abnormal batch management tables, and verification sample tables are set up.

[0012] Furthermore, the specific steps for consistency analysis using batch detection data of characteristic peaks are as follows: obtain the standard deviation of peak position, standard deviation of peak area, and standard deviation of peak shape parameter for all target characteristic peaks in each batch; calculate the product of peak position standard deviation and peak position fluctuation weight factor to obtain the peak position fluctuation term; calculate the product of peak area standard deviation and peak area fluctuation weight factor to obtain the peak area fluctuation term; calculate the product of peak shape parameter standard deviation and peak shape fluctuation weight factor to obtain the peak shape fluctuation term; and add the three terms together to obtain the inter-batch fluctuation value of the i-th characteristic peak.

[0013] Furthermore, the specific steps for dynamically optimizing peak identification and correction strategies, improving alignment accuracy, and archiving abnormal features are as follows: Real-time comparison of inter-batch fluctuation values ​​with fluctuation thresholds; when the inter-batch fluctuation value is greater than the fluctuation threshold, the quantile of the signal-to-noise ratio greater than the overall batch distribution is selected as the characteristic peak retention standard, retaining only the characteristic peaks with stronger signals; further adjustments are made to improve peak alignment accuracy based on the top 10% range of the current error distribution; the range of peak width and peak shape parameters is narrowed, selecting the top 10% range of the current statistical distribution as the retention range, removing characteristic peaks whose width variation and symmetry deviation both exceed this range; after n rounds of optimization, if the fluctuation value is still higher than the threshold, all identification results of the characteristic peaks are exported to the abnormal batch management table; when the inter-batch fluctuation value is less than or equal to the fluctuation threshold, the characteristic peak is judged as a high-stability feature, and all characteristic parameters and judgment results are summarized in the high-priority batch table of the formulation quality database.

[0014] Furthermore, the specific steps for selecting characteristic peaks by integrating the corrected batch co-occurrence frequency, discrimination index, and inter-batch fluctuations through weighted calculation are as follows: obtain the occurrence frequency, signal strength, peak area distribution, and inter-batch fluctuation value of each characteristic peak in all batches; multiply the batch co-occurrence frequency by the discrimination index to obtain the characteristic peak fusion index; calculate the characteristic peak fusion index and divide it by the inter-batch fluctuation value plus one to obtain the selection retention value of the i-th characteristic peak.

[0015] Furthermore, the specific steps for implementing characteristic peak grading screening and optimization based on the screening results are as follows: Real-time comparison of the characteristic screening retention value and the screening threshold. When the characteristic screening retention value is greater than the screening threshold, the relevant characteristic peaks and their parameters are summarized and saved in the high-priority batch table, and the corresponding characteristic peaks are directly selected into the quality analysis process as the key analysis objects for report output. When the characteristic screening retention value is less than or equal to the screening threshold, characteristic peak optimization measures are initiated: The original spectral data of the characteristic peaks are reviewed to check for noise interference, baseline drift, and peak shape abnormalities; the peak identification window is adjusted, and peak identification and retention value calculation are performed again; for characteristic peaks that are still less than or equal to the screening threshold after optimization, a manual review method is attempted, with analysts determining whether they have potential retention value; all original data and screening results of characteristic peaks that failed screening are uniformly archived in the abnormal batch management table of the formulation quality database, and characteristic peaks confirmed by manual review to have potential retention value are written into the review sample table of the formulation quality database.

[0016] Furthermore, the specific steps for multi-dimensional fusion and quantitative evaluation based on the screening results of high-priority characteristic peaks, the content data of effective components, and the physicochemical test data are as follows: Obtain the screening retention value and characteristic peak characterization value of all high-priority characteristic peaks in each batch, the effective component value of each effective component, and the physicochemical performance of each physicochemical parameter. Multiply the screening retention value of the j-th batch by the characteristic peak characterization value, and sum all multiplications to obtain the comprehensive characteristic peak characterization value; sum all effective component values ​​in the j-th batch to obtain the comprehensive effective component value; sum all physicochemical performance values ​​in the j-th batch to obtain the comprehensive physicochemical parameter value; after adding the above three items, divide by the sum of the number of high-priority characteristic peaks, the number of effective components, and the number of physicochemical parameters to obtain the comprehensive quality value of the j-th batch.

[0017] Furthermore, the specific steps for batch classification, re-inspection optimization, and quality archiving management based on the integrated assessment are as follows: By comparing the overall batch quality value with the acceptable threshold (T1 and T2) in real time; when the overall batch quality value is lower than T1, the sample is deemed unqualified, and the warehousing, circulation, and subsequent production processes of the sample are suspended. Resampling and retesting are arranged to identify the main influencing factors causing the low overall value. All re-inspection and processing results are archived in the abnormal batch management table for subsequent traceability and quality improvement; when the overall batch quality value is between T1 and T2, it is determined to be a warning batch, and the characteristic peaks of the lower-scoring samples are summarized. 1. Conduct targeted retesting of effective components and physicochemical parameters; optimize extraction conditions, strengthen impurity removal, and improve detection methods; after adjustment and retesting, reassess the overall quality value. If the overall quality value of a batch is greater than or equal to T2, it can be converted into a qualified batch; otherwise, it is recorded as a warning batch in the abnormal batch management table; when the overall quality value of a batch is greater than or equal to T2, it is judged as a qualified batch, directly included in the high-priority batch table, and all test results, characteristic parameters, and quality judgment conclusions are archived; regularly analyze the distribution of main parameters and the trend of changes in overall quality value of qualified batches to form detailed data reports and distribution analysis results of qualified quality.

[0018] Furthermore, based on the multi-dimensional fusion evaluation results and iterative training and anomaly mapping of historical sample data, the specific steps for parameter optimization and dynamic updating are as follows: Data from the high-priority batch table, abnormal batch management table, and review sample table are retrieved from the formulation quality database; characteristic peak values, effective ingredient values, physicochemical properties, and comprehensive quality values ​​of each batch are extracted to establish a batch quality characteristic sample set; the quality characteristic sample set is periodically trained, and the inter-batch fluctuation weight factor, feature screening threshold, and qualification threshold parameters are updated to achieve dynamic parameter correction; based on the deviation characteristics of the new batch data, historical abnormal batch records are matched, and the feature screening strategy and fusion weight are adjusted; the optimized parameters and correction records are synchronously written to the formulation quality database log.

[0019] Furthermore, a second aspect of the present invention provides a quality assessment system for traditional Chinese medicine (TCM) uric acid-lowering preparations based on big data, applying the quality assessment method for TCM uric acid-lowering preparations based on big data as described in any one of claims 1-9, comprising: a TCM uric acid-lowering preparation data acquisition and preprocessing module, used to acquire chromatographic characteristic data, effective component content data, and physicochemical detection data and perform preprocessing to construct a preparation quality database; an inter-batch characteristic drift correction and feature recognition module, used to perform consistency analysis through characteristic peak batch detection data, dynamically optimize peak recognition and correction strategies, improve alignment accuracy, and achieve abnormal feature archiving; characteristic peak batch... The inter-batch matching and feature selection module is used to fuse corrected batch co-occurrence frequency, discrimination index, and inter-batch fluctuations, and to perform feature peak selection through weighted calculations. Based on the selection results, feature peaks are graded and optimized. The quality index fusion and comprehensive evaluation module is used to perform multi-dimensional fusion quantitative evaluation by integrating high-priority feature peak selection results, effective component content data, and physicochemical test data. Based on the fusion evaluation, batch classification, re-inspection optimization, and quality archiving management are achieved. The self-learning and dynamic optimization module is used to iteratively train and anomaly map based on multi-dimensional fusion evaluation results and historical sample data to achieve parameter optimization and dynamic updates.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) By adopting dynamic consistency analysis and peak position correction strategy, this invention can accurately capture the subtle differences in fingerprint spectrum between batches, optimize peak identification and alignment parameters in a timely manner, and thus achieve the effect of accurate identification of characteristic peaks across batches and automatic archiving of abnormal information. This effectively solves the problems of easy mismatch in peak identification between batches and difficulty in identifying and archiving abnormal peaks in the prior art.

[0023] (2) This invention introduces multiple indicators such as batch co-occurrence frequency, discrimination index and inter-batch fluctuation, and achieves priority sorting and screening of feature peaks through weighted fusion, thereby realizing the intelligent retention of high confidence feature peaks and the hierarchical screening of abnormal features, effectively solving the problems of single feature screening standard, insufficient peak signal-to-noise ratio and discrimination ability in the prior art.

[0024] (3) This invention integrates characteristic peak parameters, effective component content and physicochemical indicators in multiple dimensions to conduct comprehensive quantitative evaluation of each batch of samples, thereby achieving the integrated effect of quality grading, batch early warning and data archiving, effectively solving the problems of single quality evaluation, unscientific batch classification and re-inspection decision in the prior art.

[0025] (4) In the process of batch comprehensive evaluation, the present invention combines real-time threshold discrimination and hierarchical management mechanism to classify and manage batches of different quality levels and conduct targeted re-inspection, thereby achieving efficient archiving of high-quality batches and accurate traceability of abnormal batches, effectively solving the problems of unclear qualification judgment and delayed processing of abnormal batches in the prior art.

[0026] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0027] Figure 1 This is a flowchart of the method for quality evaluation of traditional Chinese medicine uric acid-lowering preparations based on big data according to the present invention.

[0028] Figure 2 This is a system structure diagram of the big data-based quality evaluation system for traditional Chinese medicine uric acid-lowering preparations of the present invention;

[0029] Figure 3 This is a chart showing the distribution and trend analysis of characteristic peak batch-to-batch fluctuations and retention priority of the quality evaluation method and system for traditional Chinese medicine uric acid-lowering preparations based on big data, as presented in this invention.

[0030] Figure 4 This is a flowchart of the multidimensional quality comprehensive evaluation and dynamic judgment process of the big data-based quality evaluation system for traditional Chinese medicine uric acid-lowering preparations of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figures 1-4This invention provides a technical solution: a method and system for quality assessment of traditional Chinese medicine uric acid-lowering preparations based on big data, comprising: S1, acquiring chromatographic feature data, effective component content data, and physicochemical detection data and preprocessing them to construct a preparation quality database; S2, performing consistency analysis on batch detection data of characteristic peaks, dynamically optimizing peak identification and correction strategies, improving alignment accuracy, and archiving abnormal features; S3, fusing corrected batch co-occurrence frequency, discrimination index, and inter-batch fluctuations, performing characteristic peak screening through weighted calculations, and implementing characteristic peak grading screening and optimization processing based on the screening results; S4, performing multi-dimensional fusion quantitative evaluation by comprehensively considering high-priority characteristic peak screening results, effective component content data, and physicochemical detection data, and realizing batch classification, re-inspection optimization, and quality archiving management based on the fusion evaluation; S5, achieving parameter optimization and dynamic updating based on iterative training and anomaly mapping using multi-dimensional fusion evaluation results and historical sample data.

[0033] Specifically, the steps for acquiring and preprocessing chromatographic characteristic data, effective ingredient content data, and physicochemical detection data to construct a formulation quality database are as follows: When acquiring data on traditional Chinese medicine uric acid-lowering preparations, it is necessary to systematically integrate three major data types: chromatographic characteristic data, effective ingredient content data, and physicochemical detection data. For chromatographic characteristic data acquisition, after a standardized sample preparation process, the samples are injected into a high-performance liquid chromatograph (HPLC) for detection, and the chromatographic signal intensity and retention time of each batch of samples are collected in real time and automatically. To ensure the accuracy and comparability of the data, internal standard correction and time alignment algorithms are further applied to preprocess the acquired raw chromatograms, thereby obtaining the retention time, peak area, and peak shape parameters of each batch of characteristic peaks. The peak position data required for calculating the peak position standard deviation is obtained from the retention time series automatically output by the chromatograph, while the basic data for the peak area standard deviation is accurately calculated using an integral algorithm; the peak shape parameter standard deviation is obtained through batch extraction and statistical analysis using a chromatography workstation, covering various key peak shape data such as half-peak width, symmetry factor, and tailing factor. The calculation of batch co-occurrence frequency is based on the number of times the same characteristic peak appears in the chromatograms of all batches. This is accomplished through automated matching and statistics of batch numbers, ensuring accurate quantification of the stability of characteristic peak occurrence. For the extraction of the discrimination index, it is necessary to comprehensively analyze input information such as the signal intensity changes and peak area distribution of the same characteristic peak in different batches, and then efficiently export it through the chromatographic data processing system. In the acquisition of active ingredient content data, qualitative and quantitative detection of target components is performed using liquid chromatography-mass spectrometry (LC-MS), and the actual concentration of each active ingredient is calculated using a standard solution calibration curve. Then, through linear normalization transformation within a unified concentration range, a standard-scale active ingredient value is obtained. In the physicochemical testing data acquisition section, professional physicochemical testing equipment is used to comprehensively test multiple physicochemical indicators of the samples, such as moisture, pH value, ash content, viscosity, and density. All test results are corrected according to the real-time environmental temperature and humidity, and then uniformly converted to standard physicochemical values. Before being stored, all collected data on traditional Chinese medicine uric acid-lowering preparations undergoes standardized processing in terms of data format, timestamps, and batch codes. All numerical data is normalized through range scaling, while status-based data is mapped to quantifiable range values ​​according to operational levels, ensuring consistency in dimensions and units across different data sources. After standardization and quantification of all parameters, the data is archived by batch number and timestamp, ultimately stored in a preparation quality database employing a hybrid architecture of distributed relational structure and log storage. High-priority batch tables, abnormal batch management tables, and verification sample tables are established to achieve efficient management and traceability of data throughout the entire process.

[0034] This implementation plan comprehensively collects and preprocesses chromatographic characteristic data, effective ingredient content data, and physicochemical detection data of traditional Chinese medicine (TCM) uric acid-lowering preparations. It fully integrates multiple types of data, such as batch co-occurrence frequency, discrimination index, effective ingredient value, and physicochemical performance, and relies on internal standard correction, time alignment, and normalization processing to achieve unified standardization of high-dimensional information of different batches of samples. This lays a solid data foundation for subsequent feature screening, grading evaluation, and quality judgment. At the same time, through the design of a distributed database architecture and batch grading management table, it ensures efficient storage, rapid retrieval, and anomaly verification and tracking of sample information, significantly improving the scientific nature, systematicness, and traceability of the entire process of quality evaluation of TCM uric acid-lowering preparations.

[0035] Specifically, the steps for consistency analysis using batch detection data of characteristic peaks are as follows: Obtain the standard deviation of peak position, peak area, and peak shape parameter for all target characteristic peaks in each batch; collect retention time data for characteristic peaks from all batches and statistically analyze the standard deviation of peak position for each characteristic peak between batches. Based on the actual impact of inter-batch peak position fluctuations on the overall peak identification accuracy, sensitivity analysis is used to calculate the actual proportion of discrimination failures or mismatches caused by peak position fluctuations, thus obtaining a peak position fluctuation weighting factor; statistically analyze the standard deviation of peak area for the same characteristic peak in each batch of samples, and combine the actual role of area fluctuations in inter-batch sample discrimination and identification sensitivity. Through correlation analysis with quality analysis errors, quantify the impact of area fluctuations on screening accuracy and consistency to obtain a peak area fluctuation weighting factor; collect the half-peak width, symmetry factor, and tailing factor of each characteristic peak between batches, statistically analyze their inter-batch standard deviations, and summarize them into an overall peak shape fluctuation index. By combining the contribution of peak shape parameters to the accuracy of peak identification and anomaly detection, the proportion of detection error caused by peak shape changes is analyzed to obtain the peak shape fluctuation weighting factor. The peak position fluctuation term is calculated by multiplying the peak position standard deviation by the peak position fluctuation weighting factor, the peak area fluctuation term by multiplying the peak area standard deviation by the peak area fluctuation weighting factor, and the peak shape fluctuation term by multiplying the peak shape parameter standard deviation by the peak shape fluctuation weighting factor. These three fluctuation terms are summed to obtain the batch-to-batch fluctuation value of the i-th characteristic peak. This provides a solid data foundation and evaluation standard for subsequent characteristic peak stability screening, spectral alignment accuracy improvement, and anomaly batch identification.

[0036] The specific formula for calculating inter-batch fluctuation is as follows:

[0037] ;

[0038] In the formula, This represents the inter-batch fluctuation value of the i-th characteristic peak, which is used to measure the consistency level of the characteristic peak in multiple batches of normalized data. This represents the standard deviation of the peak position of the i-th characteristic peak across all batches, measuring the degree of fluctuation in the peak position of the peak between different batches. This represents the standard deviation of the peak area of ​​the i-th characteristic peak across all batches, measuring the degree of fluctuation in the peak area of ​​the peak between different batches. represents the standard deviation of the peak shape parameter of the i-th characteristic peak across all batches, measuring the degree of variation in the peak shape among different batches; This represents the peak fluctuation weighting factor. This represents the peak area fluctuation weighting factor. Represents the peak-shaped fluctuation weighting factor, satisfying ;

[0039] In this implementation plan, by performing multidimensional statistics on all target characteristic peaks in each batch, the standard deviations of peak position, peak area, and peak shape parameters are combined with weighting factors to obtain the inter-batch fluctuation value, which can comprehensively reflect the fluctuation of characteristic peaks between different batches. This achieves a quantitative evaluation of the stability and consistency level of characteristic peaks, laying a precise analytical foundation for subsequent feature screening, inter-batch alignment, and abnormal data identification.

[0040] Specifically, the steps for dynamically optimizing peak identification and correction strategies to improve alignment accuracy and archive abnormal features are as follows: When comparing inter-batch fluctuation values ​​with fluctuation thresholds in real time, if an inter-batch fluctuation value is detected to be greater than the fluctuation threshold, feature peaks with a signal-to-noise ratio higher than the 90th percentile of the current batch's overall distribution are prioritized as the standard for retention and analysis. This eliminates background noise interference and focuses only on feature peaks with signal strength at a high level overall for the batch. Regarding improving peak alignment accuracy, the target is the top 10% range of the current batch error distribution. By refining the algorithm step size and optimizing correction parameters, more accurate peak registration is achieved. For peak width and peak shape parameters, the target is the top 10% range of the batch parameter statistical distribution. Feature peaks with the best peak width and symmetry are retained, while feature peaks with width variability and symmetry deviation exceeding this range are removed, thereby improving the consistency and comparability of feature peak shapes. Regarding the number of optimization iterations, the number of rounds (n) and the termination condition are jointly determined by the batch fluctuation value convergence rate, optimization magnitude, and maximum number of rounds. It can be set that further iterations terminate when the fluctuation value decreases by less than 5% for two consecutive rounds, or when n is set to less than or equal to 4 and 4 rounds of optimization have been performed. If the inter-batch fluctuation value is still greater than the fluctuation threshold after multiple rounds of optimization, all identification and processing results of this characteristic peak are archived in the abnormal batch management table. When the inter-batch fluctuation value is less than or equal to the fluctuation threshold, it is identified as a high-stability characteristic peak, and the relevant characteristic parameters and judgment results are included in the high-priority batch table of the formulation quality database, providing data support for subsequent batch quality analysis and graded control.

[0041] This implementation scheme achieves efficient differentiation between the volatility and stability of characteristic peaks through a hierarchical and dynamic screening and fine alignment mechanism. By comparing inter-batch volatility values ​​and volatility thresholds in real time, it focuses on characteristic peaks whose signal intensity is located in the high quantile of the overall batch distribution. At the same time, it continuously refines peak position registration and parameter optimization using the top 10% interval of error statistics, effectively improving the accuracy and consistency of peak identification. For characteristic peaks with large variations in morphological parameters, precise elimination is achieved through distribution interval constraints. Combining the dynamic adjustment of the number of optimization rounds and the convergence strategy of multi-round optimization, it can promptly archive abnormal characteristic peaks that are difficult to normalize, while ensuring that the screening criteria are scientific and reasonable. Ultimately, it ensures that high-stability characteristic peaks enter the high-priority database table, providing a highly reliable data foundation for subsequent batch quality evaluation and intelligent archiving.

[0042] Specifically, the steps for feature peak selection through weighted calculation, which integrates the corrected batch co-occurrence frequency, discrimination index, and inter-batch fluctuation, are as follows: First, obtain the occurrence frequency, signal strength, peak area distribution, and inter-batch fluctuation value of each feature peak across all batches. Second, multiply the batch co-occurrence frequency by the discrimination index to form a feature peak fusion index that combines occurrence probability and discriminative ability, significantly improving the comprehensive discrimination and representativeness during the selection process. Third, divide this fusion index by the inter-batch fluctuation value plus one to further reduce interference caused by inconsistencies across multiple batches, highlighting feature peaks with smaller fluctuations and stable performance. The resulting selection retention value for the i-th feature peak integrates occurrence frequency and discriminative features while dynamically compensating for the impact of inter-batch fluctuations, providing scientific and quantitative basic data support for subsequent high-priority feature peak selection and hierarchical processing.

[0043] The specific formula for calculating the retained values ​​is as follows:

[0044]

[0045] In the formula, Indicates the first The retention value of each characteristic peak is used to comprehensively evaluate the value of prioritizing the retention of that characteristic peak in all batches. The larger the value, the higher the retention priority. Indicates the first The batch co-occurrence frequency of a characteristic peak reflects the stability of the occurrence of the characteristic peak in all batches, and the value ranges from 0 to 1. This represents the discrimination index, which measures the information contribution of a characteristic peak across all batches. Indicates the first The inter-batch fluctuation value of a characteristic peak is used to measure the consistency level of the characteristic peak in multiple batches of normalized data.

[0046] Table 1 shows a list of multidimensional screening parameters and retention priorities for batch characteristic peaks of traditional Chinese medicine uric acid-lowering preparations provided in this application embodiment. In this embodiment, the batch co-occurrence frequency of sample characteristic peak 1 is 0.96, the discrimination index is 1.30, the inter-batch fluctuation value is 0.06, and the calculated screening retention value is 1.178; the batch co-occurrence frequency of sample characteristic peak 2 is 0.85, the discrimination index is 1.20, the inter-batch fluctuation value is 0.33, and the calculated screening retention value is 0.768; The batch co-occurrence frequency of characteristic peak 3 is 0.60, the discrimination index is 1.05, the inter-batch fluctuation value is 0.10, and the calculated screening retention value is 0.573; the batch co-occurrence frequency of sample characteristic peak 4 is 0.79, the discrimination index is 1.22, the inter-batch fluctuation value is 0.60, and the calculated screening retention value is 0.60; the batch co-occurrence frequency of sample characteristic peak 5 is 0.91, the discrimination index is 1.15, the inter-batch fluctuation value is 0.92, and the calculated screening retention value is 0.546.

[0047] Table 1. Overview of batch characteristic peak multidimensional screening parameters and retention priorities for traditional Chinese medicine uric acid-lowering preparations.

[0048]

[0049] like Figure 3 The figure shows the distribution and trend analysis of inter-batch fluctuation and retention priority of characteristic peaks provided in the embodiments of this application. According to Table 1 and the sample characteristic peaks in the figure, characteristic peaks with smaller inter-batch fluctuation values ​​generally have higher screening retention values, indicating that these characteristic peaks exhibit high stability and consistency between batches and have priority retention value. Conversely, characteristic peaks with larger inter-batch fluctuation values, even if they show some performance in batch co-occurrence frequency or discrimination index, still have significantly suppressed screening retention values ​​and lower priority. Further observation of the color distribution of the observation points shows that characteristic peaks with higher batch co-occurrence frequencies are mostly concentrated in the high screening retention value and low inter-batch fluctuation range, indicating that stable and highly discriminative characteristic peaks are more likely to pass multiple rounds of screening and be included in the high-priority batch list. For some samples, such as No. 1 and No. 5, even with high batch co-occurrence frequency and discrimination index and extremely low inter-batch fluctuation values, their screening retention values ​​are much higher than those of similar characteristic peaks, reflecting the significant influence of multi-factor synergy on characteristic retention decisions. Overall, the relationship diagram and table intuitively demonstrate the combined effects of batch-to-batch variation, co-occurrence frequency, and discriminative power on the characteristic peak screening and retention mechanism. This effectively supports automated high-priority characteristic peak screening and quality database archiving, providing a solid data foundation and visual criteria for consistency evaluation and characteristic optimization of multi-batch formulation data.

[0050] In this implementation plan, by comprehensively collecting and quantifying multi-source data such as the occurrence frequency, signal strength, peak area distribution, and inter-batch fluctuations of each characteristic peak in all batches, and by using a fusion algorithm of batch co-occurrence frequency and discrimination index, the representativeness and discrimination ability of the characteristic peaks are fully explored. At the same time, an inter-batch fluctuation compensation mechanism is introduced, which effectively improves the scientificity and accuracy of characteristic peak selection, realizes the selection of key characteristic peaks that are stable and have significant discrimination power, and provides a solid data foundation for subsequent quality assessment and hierarchical management.

[0051] Specifically, the steps for implementing characteristic peak grading screening and optimization based on the screening results are as follows: Real-time comparison of the characteristic peak retention value and the screening threshold. When the retention value is greater than the threshold, the relevant characteristic peaks and their parameters are summarized and saved in a high-priority batch table. The corresponding characteristic peaks are directly selected into the quality analysis process as key analysis objects in the report output, ensuring that critical characteristic peaks can participate in subsequent quality assessments and batch traceability analyses with priority. When the retention value is less than or equal to the threshold, characteristic peak optimization measures are initiated. This includes a detailed review of the original spectral data of the characteristic peak to identify influencing factors such as noise interference, baseline drift, and peak shape anomalies. The peak identification window is adjusted accordingly, and peak identification and retention value calculations are recalculated to increase the likelihood of retention. For characteristic peaks whose retention values ​​remain less than or equal to the threshold after optimization, a manual review is attempted. Analysts make a comprehensive judgment based on the data performance and practical application value. Characteristic peaks confirmed to have potential retention value are promptly written into the review sample table of the formulation quality database as an important basis for subsequent data supplementation and revision of screening standards. All characteristic peaks that fail the screening are archived together with their original data and screening results in the abnormal batch management table of the formulation quality database, thus improving the management loop and traceability of the entire screening process.

[0052] In this implementation plan, by comparing the feature selection retention value with the selection threshold in real time, key feature peaks in each batch can be accurately identified and summarized for retention, achieving efficient screening of high-priority feature peaks and their priority inclusion in the analysis process. Simultaneously, by reviewing the original spectra, optimizing parameters, and manually verifying feature peaks that do not meet the screening criteria, the sensitivity and accuracy of feature peak screening are significantly improved, ensuring that potentially valuable feature peaks are supplemented and archived as a reference for subsequent standard revisions. All feature peak results that fail screening are fully archived, providing a systematic data foundation for abnormal batch management and subsequent traceability analysis, further improving the closed-loop control system for feature screening in the quality evaluation process of traditional Chinese medicine uric acid-lowering preparations.

[0053] Specifically, the steps for multidimensional fusion and quantitative evaluation based on the screening results of high-priority characteristic peaks, the content data of effective components, and the physicochemical detection data are as follows: Obtain the screening retention value and characteristic peak characterization value of all high-priority characteristic peaks in each batch, the effective component value of each effective component, and the physicochemical performance of each physicochemical parameter. Multiply the screening retention value of each high-priority characteristic peak in the j-th batch by its corresponding characteristic peak characterization value, and sum all products to obtain the comprehensive characteristic peak characterization value. This comprehensively reflects the multidimensional performance and retention significance of each characteristic peak in the batch sample, thus providing more discriminative structural information for the batch sample. The comprehensive effective component value is obtained by summing all effective component values ​​in the j-th batch, which is used to measure the overall performance of the batch sample in terms of key components related to efficacy. The comprehensive physicochemical parameter value is obtained by summing all physicochemical performance values ​​in the j-th batch, reflecting the comprehensiveness of sample quality from multiple physicochemical dimensions. The comprehensive quality value of the j-th batch is obtained by adding the three comprehensive values ​​together and dividing by the sum of the number of high-priority characteristic peaks, the number of effective components, and the number of physicochemical parameters. This achieves multi-dimensional integration of key indicators, compatibility of structural weights, and accurate characterization of the overall level, providing a quantitative basis for subsequent quality judgment, risk classification, and batch traceability.

[0054] The specific formula for calculating the overall quality value is as follows:

[0055] ;

[0056] In the formula, Indicates the first The overall quality value of each batch is used to comprehensively reflect the overall quality level of the batch of samples. Indicates the first The retention value of each characteristic peak is used to measure the stability and discriminative power of that characteristic peak; Indicates the first The first batch Characteristic peak values; Indicates the first The first batch One effective ingredient value; Indicates the first The first batch Individual physical and chemical expression quantities; This indicates the number of high-priority feature peaks. Indicates the quantity of active ingredients. Indicates the number of physicochemical parameters.

[0057] In this implementation plan, by integrating the screening and retention values ​​and characteristic values ​​of high-priority characteristic peaks, the effective component values ​​of each active ingredient, and the physicochemical properties of each physicochemical parameter, a comprehensive evaluation system reflecting the multidimensional quality characteristics of each batch of samples is constructed using a weighted accumulation and normalization allocation method. This system not only effectively integrates the three core information categories of fingerprint characteristics, pharmacodynamic substances, and physicochemical properties, but also fully reflects the relative contributions and complementary effects of various elements within a batch, thereby providing a scientific and quantitative basis for judging the overall quality level of the samples and improving the objectivity and discriminative power of quality evaluation.

[0058] Specifically, the steps for batch classification, re-inspection optimization, and quality archiving management based on integrated assessment are as follows: real-time comparison of the overall batch quality value and the pass threshold, where the pass threshold is further subdivided into T1 and T2, to achieve multi-level discrimination of the batch sample quality status. For example... Figure 4 The diagram shows the multi-dimensional quality comprehensive assessment and dynamic judgment flowchart of the big data-based quality assessment system for traditional Chinese medicine uric acid-lowering preparations of this invention. When the comprehensive quality value of a batch is less than T1, the batch sample is immediately judged as unqualified, its warehousing, circulation and subsequent production are suspended, and resampling and retesting are arranged in a timely manner. Combining multi-dimensional data such as the characteristic peak values, effective ingredient values ​​and physicochemical properties within the batch, the main influencing factors causing the low comprehensive value are accurately located. All re-inspection and processing results are simultaneously archived in the abnormal batch management table, providing a data foundation for subsequent traceability and continuous quality improvement. When the comprehensive quality value of a batch is greater than T1 and less than T2, it is judged as a warning batch. By summarizing and analyzing the characteristic peaks, effective ingredients and physicochemical parameters of the samples with lower scores, targeted re-inspection and process optimization are carried out, and the removal of impurities and improvement of detection methods are strengthened. After the re-inspection and adjustment are completed, the comprehensive value is re-evaluated. If the result is greater than or equal to T2, the batch can be converted to qualified; otherwise, it is filed as a warning batch in the abnormal batch management table, forming a closed-loop management. When the overall quality value of a batch is greater than or equal to T2, it is directly judged as a qualified batch, included in the high-priority batch list, and all test results, characteristic parameters, and final quality judgment conclusions are archived. Simultaneously, the distribution of key parameters and the trend of overall quality value changes of all qualified batches are regularly statistically analyzed to generate detailed qualified quality data reports and distribution characteristic analyses, providing quantitative support for enterprise process quality control and risk early warning.

[0059] This implementation plan achieves tiered quality management of batch samples throughout the entire process, from non-compliance and warning to compliance, by setting multi-level qualification thresholds and dynamically comparing and classifying the comprehensive quality values ​​of each batch. Combined with in-depth analysis of multi-dimensional indicators such as characteristic peak values, active ingredient values, and physicochemical properties of each batch, it not only enables timely screening and identification of batches with quality abnormalities and the development of targeted re-inspection and adjustment measures, but also ensures the orderly archiving and traceability of all processing results. Simultaneously, by periodically analyzing the parameter distribution and comprehensive value trends of qualified batches, it promotes the refinement and dynamism of quality assessment results, providing solid data support and decision-making basis for risk control and continuous improvement in the formulation production process.

[0060] Specifically, the steps for parameter optimization and dynamic updating based on multi-dimensional fusion evaluation results and iterative training and anomaly mapping using historical sample data are as follows: Data from the high-priority batch table, abnormal batch management table, and review sample table are retrieved from the formulation quality database. Characteristic peak values, effective ingredient values, physicochemical properties, and comprehensive quality values ​​of each batch are comprehensively extracted to systematically construct a batch quality characteristic sample set. For this sample set, combined with batch numbers and time series, periodic training is conducted using training paradigms such as sliding window statistics, progressive updates, and rolling aggregation. As the sample coverage expands, the robustness and generalization ability of the weight parameters are continuously optimized. During training, the deviation characteristics between new batch data and existing distributions are captured in real time. Through multiple rounds of dynamic parameter correction, the inter-batch fluctuation weight factor, feature screening threshold, and qualification threshold parameters are iteratively optimized to improve the sensitivity and adaptability of the parameters. Simultaneously, based on the anomaly mapping mechanism, historical abnormal batch records are automatically compared and matched, and the feature screening strategy and fusion weights are adjusted in a timely manner to achieve adaptive compensation for abnormal characteristic peaks. All optimized parameters and correction records are synchronously written to the formulation quality database log, forming a closed-loop parameter evolution trajectory throughout the entire process, further supporting intelligent quality assessment and traceability improvement for subsequent batches.

[0061] In this implementation plan, by periodically aggregating and training batch quality feature sample sets, the core parameters such as batch-to-batch fluctuation weight factor, feature screening threshold, and qualification threshold are dynamically corrected. Combined with the mapping and comparison of historical abnormal batches, the feature screening strategy and weight allocation are intelligently adjusted, which comprehensively improves the evaluation method's responsiveness to batch-to-batch differences and abnormal situations. This provides a data-driven and adaptive support foundation for the quality evaluation and continuous optimization of traditional Chinese medicine uric acid-lowering preparations.

[0062] Specifically, such as Figure 2The diagram shown is a system structure diagram of the quality assessment method and system for traditional Chinese medicine uric acid-lowering preparations based on big data provided in this application embodiment. This embodiment provides a quality assessment system for traditional Chinese medicine uric acid-lowering preparations based on big data. The above-mentioned quality assessment method for traditional Chinese medicine uric acid-lowering preparations based on big data includes a data acquisition and preprocessing module for traditional Chinese medicine uric acid-lowering preparations, a batch-to-batch feature drift correction and feature recognition module, a batch-to-batch feature peak matching and feature screening module, a quality index fusion and comprehensive evaluation module, and a self-learning and dynamic optimization module. Each module works together. The data acquisition and preprocessing module for traditional Chinese medicine uric acid-lowering preparations is specifically designed for acquiring chromatographic characteristic data, effective ingredient content data, and physicochemical detection data throughout the entire process. It performs standardization and consistency preprocessing on various data types, including format conversion, timestamp and batch code normalization, and interval mapping between numerical and status data. This constructs a well-structured and hierarchically managed preparation quality database for subsequent high-quality data analysis. The inter-batch characteristic drift correction and feature recognition module, based on characteristic peak information obtained from batch detection, uses multivariate statistical analysis of peak position standard deviation, peak area standard deviation, and peak shape parameter standard deviation to quantitatively assess the inter-batch consistency of target characteristic peaks. It dynamically optimizes peak recognition and correction strategies, improves spectral peak alignment accuracy based on standards such as signal-to-noise ratio and error range, and archives characteristic peaks that remain abnormal after multiple rounds of screening to an abnormal batch management table, ensuring reliable hierarchical data storage. The characteristic peak inter-batch matching and feature screening module integrates multiple indicators such as batch co-occurrence frequency, discrimination index, and inter-batch fluctuation after inter-batch drift correction processing. The data is processed through a weighted aggregation and hierarchical screening mechanism to determine the retention priority of each characteristic peak. A review and optimization process is designed for characteristic peaks at the edge of the screening threshold to achieve hierarchical archiving, review, and optimization of characteristic peaks. The quality index fusion and comprehensive evaluation module integrates the screening results of high-priority characteristic peaks, the content data of each effective component, and the physicochemical test data for each batch. It uses multiplicative accumulation and weighted fusion to construct a comprehensive quality evaluation model covering multidimensional indicators of the entire sample. Through hierarchical threshold comparison, it automatically completes batch classification, re-inspection optimization, and historical archiving to achieve dynamic quality monitoring and data traceability. The self-learning and dynamic optimization module, based on the multidimensional fusion evaluation results, combines historical data of high-priority batches, abnormal batches, and review samples to periodically aggregate and train the characteristic sample set. Through adaptive updates of weight factors and discrimination thresholds, it automatically corrects parameter configurations to achieve continuous evolution and dynamic optimal adjustment of the evaluation method, significantly improving the accuracy, robustness, and real-time response capability of batch quality analysis of traditional Chinese medicine uric acid-lowering preparations.

[0063] This implementation plan achieves a high degree of integration and closed-loop management of traditional Chinese medicine (TCM) uric acid-lowering preparations, encompassing the entire process from raw data collection, standardized processing, batch-to-batch consistency correction, multi-dimensional screening and grading of characteristic peaks, comprehensive evaluation of quality indicators, to self-learning dynamic optimization. This significantly improves the accuracy and efficiency of characteristic peak identification and batch alignment, while also strengthening data management capabilities for abnormal batches and verification samples. This makes the fusion judgment of high-priority characteristics with effective components and physicochemical properties more scientific and dynamically adaptable. Through periodic parameter updates and historical anomaly mapping, the system continuously optimizes the screening and evaluation process, ultimately forming a well-structured, clearly defined, and self-evolving quality control platform. This provides solid data support and an intelligent decision-making foundation for large-scale, heterogeneous preparation quality assessment and continuous traceability.

[0064] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0065] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for evaluating the quality of traditional Chinese medicine (TCM) preparations for reducing uric acid based on big data, comprising the following steps: S1. Obtaining chromatographic feature data, effective component content data, and physicochemical detection data and performing preprocessing to construct a preparation quality database; S2. Performing consistency analysis on feature peak batch detection data, dynamically optimizing peak identification and correction strategies, improving alignment accuracy, and archiving abnormal features; S3. Fusing the batch co-occurrence frequency, discriminability index, and batch-to-batch fluctuation of the corrected data, performing feature peak screening through weighted operation, and implementing feature peak hierarchical screening and optimization based on the screening results; S4. Integrating high-priority feature peak screening results, effective component content data, and physicochemical detection data for multi-dimensional fusion and quantitative evaluation, and performing batch classification, recheck optimization, and quality archiving management based on the fusion evaluation; S5. Iterative training and abnormal mapping based on multi-dimensional fusion evaluation results and historical sample data to achieve parameter optimization and dynamic updating.

2. The big data-based traditional Chinese medicine preparation for reducing uric acid quality evaluation method according to claim 1, characterized in that: The specific steps for obtaining chromatographic feature data, effective component content data, and physicochemical detection data and performing preprocessing to construct a preparation quality database are as follows: Obtaining TCM preparations for reducing uric acid data, which includes obtaining chromatographic feature data, effective component content data, and physicochemical detection data; Obtaining chromatographic feature data: After sample preparation, the samples are injected into a high-performance liquid chromatograph for detection, and the chromatographic signal intensity and retention time of each batch of samples are collected in real time. The retention time, peak area, and peak shape parameters of each batch of feature peaks are obtained through internal standard correction and time alignment algorithms. The peak position data required for peak position standard deviation are obtained from the retention time sequence detected by the chromatograph. The peak area data required for peak area standard deviation are calculated by integral algorithm. The peak shape data required for peak shape parameter standard deviation are obtained by extracting and processing the chromatograph workstation for statistical analysis, including half-height peak width, symmetry factor, and tailing factor. The batch co-occurrence frequency is obtained by matching the batch number to record the number of occurrences of the same feature peak in all batch chromatograms. The input values required are calculated from the signal intensity changes and peak area distribution of the same feature peak in different batches, and the discriminability index is exported by the chromatographic data processing system; Obtaining effective component content data: The target components are detected by liquid chromatography-mass spectrometry, and the actual concentrations of each effective component are measured by standard solution calibration curve. After linear normalization of the unified concentration interval, the effective component values are obtained; Obtaining physicochemical detection data: The moisture, pH value, ash content, viscosity, and density physicochemical indicators of the sample are measured using physicochemical detection equipment. After temperature and humidity correction and normalization processing, the physicochemical performance quantities are obtained. The collected traditional Chinese medicine hypouricemic preparation data is uniformly processed in data format, timestamp and batch coding; linear normalization method is used for all numerical parameters to scale the data to a unified interval range; the state type parameters are mapped to interval values according to the grade working condition to ensure the consistency of parameters from different sources in dimension and unit; after the standardization and normalization of all parameters, the traditional Chinese medicine hypouricemic preparation data is stored according to batch number and timestamp, and a mixed architecture of distributed relational structure and log storage is used to design and build the preparation quality database, and high priority batch table, abnormal batch management table and review sample table are set.

3. The big data-based traditional Chinese medicine preparation for reducing uric acid quality evaluation method according to claim 1, characterized in that: The specific steps of consistency analysis by characteristic peak batch detection data are: Obtain the peak position standard deviation, peak area standard deviation and peak shape parameter standard deviation of all target characteristic peaks of each batch of characteristic peaks; Calculate the product of the peak position standard deviation and the peak position fluctuation weight factor to obtain the peak position fluctuation term; Calculate the product of the peak area standard deviation and the peak area fluctuation weight factor to obtain the peak area fluctuation term; Calculate the product of the peak shape parameter standard deviation and the peak shape fluctuation weight factor to obtain the peak shape fluctuation term, and add the three terms to obtain the batch-to-batch fluctuation value of the i th characteristic peak.

4. The big data-based traditional Chinese medicine preparation for reducing uric acid quality evaluation method according to claim 1, characterized in that: The specific steps of dynamic optimization of peak identification and correction strategy, improvement of alignment accuracy and realization of abnormal feature archiving are: Real-time comparison of batch-to-batch fluctuation value and fluctuation threshold, when the batch-to-batch fluctuation value is greater than the fluctuation threshold, select the signal-to-noise ratio greater than the quantile of the overall distribution of the batch as the characteristic peak retention standard, and only retain the characteristic peak with stronger signal; further adjust and improve the peak position alignment accuracy according to the top 10% interval of the current error distribution; narrow the peak width and peak shape parameter range, select the top 10% range of the current statistical distribution as the retention interval, and remove the characteristic peaks whose width variation and symmetry deviation exceed the interval; after n rounds of optimization, if the fluctuation value is still higher than the threshold, export all the identification results of the characteristic peak to the abnormal batch management table; When the batch-to-batch fluctuation value is less than or equal to the fluctuation threshold, the characteristic peak is determined as a high-stability characteristic, and all characteristic parameters and determination results are summarized into the high-priority batch table of the preparation quality database.

5. The big data-based traditional Chinese medicine preparation for reducing uric acid quality evaluation method according to claim 1, characterized in that: The specific steps of fusing the batch co-occurrence frequency, discriminability index and batch-to-batch fluctuation after correction processing, and screening the characteristic peaks by weighted operation are: Obtain the number of occurrences, signal intensity, peak area distribution and batch-to-batch fluctuation value of each characteristic peak in all batches; Multiply the batch co-occurrence frequency and the discriminability index to obtain the characteristic peak fusion index, calculate the characteristic peak fusion index divided by the batch-to-batch fluctuation value plus one to obtain the screening retention value of the i th characteristic peak.

6. The big data-based traditional Chinese medicine preparation quality evaluation method for reducing uric acid according to claim 1, characterized in that: The specific steps of implementing characteristic peak grading screening and optimization processing according to the screening results are: Real-time comparison of characteristic screening retention value and screening threshold, when the characteristic screening retention value is greater than the screening threshold, the related characteristic peak and its parameters are summarized and saved to the high-priority batch table, and the corresponding characteristic peak is directly selected into the quality analysis process as the key analysis object of report output; When the characteristic screening retention value is less than or equal to the screening threshold, the characteristic peak optimization processing measures are started: review the original spectral data of the characteristic peak to check whether there is noise interference, baseline drift and peak shape abnormality; Adjust the peak identification window, re-identify peaks and calculate retention values; for the characteristic peaks still less than or equal to the screening threshold after optimization, try to use artificial review method to determine whether they have potential retention value by the analyst; Archive all the original data and screening results of the characteristic peaks that do not pass the screening to the abnormal batch management table of the preparation quality database, and write the characteristic peaks confirmed by artificial review to have potential retention value into the review sample table of the preparation quality database.

7. The big data-based traditional Chinese medicine preparation quality evaluation method for reducing uric acid according to claim 1, characterized in that: The specific steps of the multi-dimensional fusion quantitative evaluation of the comprehensive high-priority characteristic peak screening results, the effective component content data and the physicochemical detection data are: Obtain the screening retention values and characteristic peak representation values of all high-priority characteristic peaks in each batch, the values of each effective component and the physicochemical performance quantities; Multiply the screening retention values and characteristic peak representation values of the jth batch, and add all the products to obtain the characteristic peak comprehensive representation quantity; add all the effective component values of the jth batch to obtain the effective component comprehensive value, and add all the physicochemical performance quantities of the jth batch to obtain the physicochemical parameter comprehensive quantity; add the above three items, and divide by the sum of the number of high-priority characteristic peaks, the number of effective components and the number of physicochemical parameters to obtain the quality comprehensive value of the jth batch.

8. The big data-based traditional Chinese medicine preparation quality evaluation method for reducing uric acid according to claim 1, characterized in that: The specific steps of batch classification, re-inspection optimization and quality archiving management according to the fusion evaluation are: Compare the batch quality comprehensive value with the qualified threshold in real time, the qualified threshold includes T1 and T2; when the batch quality comprehensive value is lower than T1, the sample is determined to be unqualified, the sample is suspended from warehousing, circulation and subsequent production links, re-sampling and re-measurement are arranged, the main influencing factors leading to the low comprehensive value are located, and all the review and processing results are archived to the abnormal batch management table for subsequent tracing and quality improvement; When the batch quality comprehensive value is between T1 and T2, it is determined to be a warning batch, the characteristic peaks, effective components and physicochemical parameter data with low scores are summarized, and targeted re-inspection is arranged; the extraction conditions are optimized, impurity removal is strengthened, and the detection method is improved; after adjustment and re-inspection, the quality comprehensive value is re-evaluated, and if the batch quality comprehensive value is greater than or equal to T2, it can be converted into a qualified batch, otherwise it is left as a warning batch in the abnormal batch management table; When the batch quality comprehensive value is greater than or equal to T2, it is determined to be a qualified batch, and all the detection results, characteristic parameters and quality determination conclusions are directly archived to the high-priority batch table and the qualified batch management table; the main parameter distribution and quality comprehensive value change trend of the qualified batch are statistically analyzed to form detailed qualified quality data report and distribution analysis results. 9.The method for evaluating the quality of a traditional Chinese medicine preparation for reducing uric acid based on big data according to claim 1, characterized in that: The specific steps of parameter optimization and dynamic update based on the iterative training and abnormal mapping of the multi-dimensional fusion evaluation results and historical sample data are: The data of high-priority batch table, abnormal batch management table and review sample table are called from the preparation quality database, the characteristic peak characteristic value, active ingredient value, physical and chemical performance value and quality comprehensive value of each batch are extracted, and a batch quality characteristic sample set is established; the quality characteristic sample set is periodically trained, the inter-batch fluctuation weight factor, characteristic screening threshold and qualified threshold parameters are updated, and dynamic correction of the parameters is realized; according to the deviation characteristics of the data of the new batch, the historical abnormal batch records are matched, the characteristic screening strategy and the fusion weight are adjusted, and the optimized parameters and the correction records are written into the preparation quality database log synchronously.

10. A quality evaluation system for traditional Chinese medicine hypouricemic preparation based on big data, applying the quality evaluation method for traditional Chinese medicine hypouricemic preparation based on big data according to any one of claims 1-9, characterized in that , comprising: A traditional Chinese medicine preparation data acquisition and preprocessing module is used for acquiring chromatographic characteristic data, active ingredient content data and physical and chemical detection data and preprocessing, and constructing a preparation quality database; An inter-batch characteristic drift correction and characteristic identification module is used for consistency analysis through batch detection data of characteristic peaks, dynamic optimization of peak identification and correction strategy, improvement of alignment accuracy and realization of abnormal characteristic archiving; A characteristic peak inter-batch matching and characteristic screening module is used for screening characteristic peaks through weighted operation by fusing batch co-occurrence frequency, discriminability index and inter-batch fluctuation after correction processing, and implementing hierarchical screening and optimization processing of characteristic peaks according to the screening results; A quality index fusion and comprehensive evaluation module is used for multi-dimensional fusion quantitative evaluation of high-priority characteristic peak screening results, active ingredient content data and physical and chemical detection data, batch classification, review optimization and quality archiving management are realized according to the fusion evaluation; A self-learning and dynamic optimization module is used for iterative training and abnormal mapping based on multi-dimensional fusion evaluation results and historical sample data, and parameter optimization and dynamic updating are realized.

Images