A method for processing MALDI-TOF MS quality control data and a computing device

Abstract

The application discloses a MALDI-TOF MS quality control data processing method and a computing device. Quality control data is acquired; a sample type is identified and processed by a preset mapping rule, patient samples and quality control strain samples are respectively included in independent statistical indicators; a dynamic reference line is established based on a prior effective statistical index or an initial statistical window meeting a stable condition; trend analysis is performed based on a patient-side reference line, and quality control rule determination is performed based on a quality control strain-side reference line; when out-of-control is determined, an out-of-control record is generated, and a detection date corresponding to the corresponding out-of-control record is added to an exclusion set, so that data corresponding to the exclusion date does not participate in subsequent reference line updating; and sample types, original data sources and out-of-control records and other traceability information associated with statistical points are output. The application can reduce the influence of abnormal data on subsequent statistical benchmarks, and improve the traceability of quality control determination results.

Description

Technical Field

[0001] This application relates to the field of quality control data processing technology for mass spectrometry identification in microbial laboratories, and more specifically, to a method and computing device for processing MALDI-TOF MS quality control data. Background Technology

[0002] Matrix-assisted laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOF MS) is a commonly used technique in clinical microbiology laboratories for microbial identification. To ensure the accuracy and reliability of the identification results, laboratories need to conduct quality control on the daily operation of the mass spectrometer. Currently, the processing of MALDI-TOF MS quality control data typically relies on the identification results output by the instrument software, supplemented by manual recording or processing and observation using general office software. However, existing methods have the following shortcomings: Firstly, MALDI-TOF MS quality control data contains various types of samples, each with different functions and significance in quality control analysis. Current techniques, when using various types of samples for instrument condition assessment and quality observation, fail to establish a unified basis for judgment among the analytical results of different sample types. This necessitates viewing quality control data from multiple dimensions, increasing the complexity of quality control analysis and affecting the overall efficiency of quality control conclusions.

[0003] Secondly, quality control analysis requires comparing current test results with historical data to determine if there are any abnormal trends in instrument status. Existing technologies typically use fixed judgment criteria, while quality control data accumulates and changes continuously as daily testing work progresses. Fixed criteria are difficult to adapt to the dynamic changes in data, affecting the accuracy of trend judgment.

[0004] Third, when test results are abnormal, existing technologies usually only provide immediate alerts or manual recording. The impact of these abnormal information on subsequent quality control judgment criteria is unclear, and abnormal data may continue to interfere with subsequent quality control trend observation.

[0005] Fourth, there is insufficient correlation between the quality control data processing results and the original data sources. In existing technologies, original data, sample classification information, and anomaly judgment results are usually stored separately, and the correspondence between the various pieces of information is unclear. When there are doubts about the quality control judgment results, it is difficult to trace back to the specific original data source and processing information.

[0006] It is evident that the existing methods for processing MALDI-TOF MS quality control data have shortcomings in data classification, trend judgment, anomaly handling, and result traceability, making it difficult to guarantee the reliability and traceability of quality control judgment results. Summary of the Invention

[0007] The main purpose of this application is to provide a method and computing device for processing MALDI-TOF MS quality control data, so as to improve the traceability of quality control data, the reliability of rule judgment, and the integrity of the closed-loop anomaly handling.

[0008] To achieve the above objectives, a first aspect of this application proposes a method for processing MALDI-TOF MS quality control data. The method includes: generating and outputting the quality control judgment result and corresponding traceability information of the current batch of quality control data based on the patient-side quality control observation results and the results of the quality control rule judgment; the traceability information includes at least the sample type, the original data source, and the out-of-control record generated when there is an out-of-control judgment, and establishing the correlation between the traceability information and the corresponding statistical points.

[0009] To achieve the above objectives, a second aspect of this application provides a computing device including a processor and a memory storing a computer program, wherein when the processor runs the computer program, it implements the steps of the above-described method for processing MALDI-TOF MS quality control data.

[0010] To achieve the above objectives, a third aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of any of the above-described methods for processing MALDI-TOF MS quality control data.

[0011] The technical solutions provided by the embodiments of this application may include the following beneficial effects: In this application, business semantic parsing is performed on quality control data to identify the sample type of each sample (including at least patient samples and quality control strain samples). The data is then processed separately according to sample type, with patient samples and quality control strain samples included in independent statistical categories. This allows patient samples and quality control strain samples to be processed under independent statistical categories without interference, simultaneously addressing the quality control needs of both instrument operation status assessment and patient sample identification quality observation, thus improving the comprehensiveness and accuracy of quality control analysis. Furthermore, dynamic reference lines are established for patient samples and quality control strain samples based on their respective statistical indicators. Since the dynamic reference lines are based on accumulated statistical indicators, they can be dynamically updated as quality control data accumulates. Compared to fixed reference standards, dynamic reference lines can more accurately reflect the current data status, thereby improving the accuracy of quality control trend judgment. Finally, when the quality control rule determines that the data is out of control, an out-of-control record is generated, and the detection date corresponding to the out-of-control record is added to the exclusion set for dynamic reference line calculation. Therefore, data corresponding to excluded dates are no longer included in the calculation of subsequent dynamic reference lines, preventing abnormal data from continuously contaminating the benchmark for subsequent quality control judgments and improving the accuracy and stability of subsequent quality control trend observation. Finally, by generating quality control judgment results and outputting corresponding traceability information, which is at least associated with the sample type and out-of-control record of the quality control data, a correlation is established between the quality control judgment results and intermediate processing information such as sample type and out-of-control record. When there is doubt about the quality control judgment results, the traceability information can be used to trace back to the specific sample type and out-of-control record, improving the traceability of the quality control judgment results. In summary, this application embodiment, through techniques such as track-by-track processing, dynamic reference line establishment, out-of-control date exclusion, and traceability information output, forms a complete processing link from data acquisition to judgment result output. Compared with the prior art, it has significantly improved the comprehensiveness of quality control analysis, the accuracy of trend judgment, the reliability of anomaly handling, and the traceability of results. Attached Figure Description

[0012] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 A flowchart illustrating a method for processing MALDI-TOF MS quality control data provided in this application embodiment; Figure 2 This application provides a schematic diagram of the specific business process for a method of processing MALDI-TOF MS quality control data in an embodiment of the present application. Figure 3A schematic diagram of the homepage dashboard of the mass spectrometer quality control system provided in this application embodiment; Figure 4 A schematic diagram of the data import interface of the mass spectrometer quality control system provided in the embodiments of this application; Figure 5 A schematic diagram of the OCR verification interface of the mass spectrometer quality control system provided in the embodiments of this application; Figure 6 A schematic diagram of the OCR calibration interface of the mass spectrometer quality control system provided in the embodiments of this application; Figure 7 A schematic diagram of the interface for strain completion in a mass spectrometer quality control system provided in this application embodiment; Figure 8(a) is a schematic diagram of the daily trend of patient strains in the mass spectrometer quality control system provided in the embodiments of this application; Figure 8(b) is a schematic diagram of the interface of the mass spectrometer quality control system provided in the embodiment of this application for handling the runaway record; Figure 9 A schematic diagram of the alarm center interface of the mass spectrometer quality control system provided in this embodiment of the application; Figure 10 A schematic diagram of the maintenance interface of the mass spectrometer quality control system provided in the embodiments of this application; Figure 11 A schematic diagram of the interface between QC and standard bacteria management in the mass spectrometer quality control system provided in this application embodiment; Figure 12(a) is a schematic diagram of the interface of the conventional rules of the mass spectrometer quality control system provided in the embodiment of this application; Figure 12(b) is a schematic diagram of the interface for Westgard quality control / out-of-control rule management of the mass spectrometer quality control system provided in the embodiments of this application; Figure 13 A schematic diagram of the interface for managing the filtered bacteria in the mass spectrometer quality control system provided in this application embodiment; Figure 14 A schematic diagram of the interface for managing common bacterial strains in the mass spectrometer quality control system provided in this application embodiment; Figure 15 A schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0013] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0014] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0015] In this application, the terms "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are primarily for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation.

[0016] Furthermore, in addition to indicating location or positional relationship, some of the aforementioned terms may also have other meanings. For example, the term "above" may also be used in some cases to indicate a certain dependency or connection relationship. Those skilled in the art can understand the specific meaning of these terms in this application based on the specific circumstances.

[0017] Furthermore, the terms "installation," "setup," "equipped with," "connection," "linked," and "socketing" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0018] It should be noted that the embodiments of this application can be applied to the data processing field of daily quality control management of MALDI-TOF MS in clinical microbiology laboratories. For example, by deploying a mass spectrometer, data acquisition terminal, and computing device in the MALDI-TOF MS testing environment of a clinical microbiology laboratory, the mass spectrometry identification results data of the current batch are acquired, including the species name and identification confidence score of each sample; the computing device performs business semantic parsing on the mass spectrometry identification results data to identify the type of each sample (patient sample or quality control strain sample); based on the identification results, the data is processed by data segmentation, and patient samples and quality control strain samples are included in independent statistical calibers and corresponding statistical indicators are generated; the computing device establishes dynamic reference lines based on statistical indicators for patient samples and quality control strain samples respectively; and the statistical indicators of patient samples are calculated based on the dynamic reference lines on the patient side. Trend analysis is performed on the samples. Based on the dynamic reference line of the quality control strains, the identification confidence score of the quality control strain samples is judged according to the quality control rules. When the judgment result is out of control, an out-of-control record is generated, and the detection date corresponding to the out-of-control record is added to the exclusion set for subsequent dynamic reference line calculation. Finally, based on the patient-side quality control observation results, quality control rule judgment results and exclusion set, the quality control judgment result of the current batch and the corresponding traceability information are generated and output. This method aims to replace the technical limitations of traditional quality control data processing, such as not distinguishing sample types, using fixed reference lines, continuous contamination of subsequent benchmarks by outliers, and incomplete traceability links between the processed results and the original data.

[0019] Understandably, existing solutions for MALDI-TOF MS quality control data processing face the following main technical challenges. First, quality control data includes two different types of samples: patient samples and quality control bacterial strain samples. Existing technologies typically process both types of samples together or focus only on one type, making it difficult to simultaneously reflect information on both instrument operating status and patient sample identification quality. Second, existing technologies usually use fixed reference lines, while quality control data accumulates and changes continuously with daily testing, making it difficult for fixed reference lines to adapt to dynamic data changes. Furthermore, when the identification results of quality control bacterial strain samples are abnormal, existing technologies only provide immediate alerts or manual recording, and abnormal data may continue to affect the accuracy of subsequent quality control judgments. Finally, the traceability link between quality control judgment results and the original data source is incomplete in existing technologies, making it difficult to trace back to the specific original data and intermediate processing information when there are doubts about the quality control judgment results.

[0020] Based on this, the embodiments of this application aim to provide a method and computing device for processing MALDI-TOF MS quality control data. By performing business semantic parsing on the quality control data to identify sample types and performing track-based processing, independent dynamic reference lines are established for the patient side and the quality control strain side, respectively. The detection date corresponding to the out-of-control record is added to the exclusion set for subsequent reference line calculation, and traceability information of associated sample types and out-of-control records is output. This realizes systematic processing and full-link traceability of quality control data, improves the reliability of quality control judgment results, the accuracy of trend judgment, and the traceability of results, reduces reliance on human experience, and improves the intelligent level of quality control management in clinical microbiology laboratories.

[0021] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] Figure 1 A flowchart illustrating a method for processing MALDI-TOF MS quality control data provided in this application embodiment is shown below. Figure 1 As shown, applied to a computing device, the method includes: Step S1: Obtain the MALDI-TOF MS quality control data for the current batch; the quality control data shall include at least the sample identification results, which shall include at least the strain name and the identification confidence score.

[0023] Optionally, MALDI-TOF MS (Matrix-Assisted Laser Desorption / Ionization Time-of-Flight Mass Spectrometry) is a mass spectrometry technique used for microbial identification. Its basic principle is as follows: the sample to be tested is mixed with a matrix solution and spotted onto a target plate. After laser irradiation, the matrix absorbs the laser energy and transfers it to the sample molecules, causing them to desorb and ionize. The ionized sample ions are accelerated in an electric field and enter the time-of-flight mass analyzer. Ions with different mass-to-charge ratios (m / z) have different flight velocities. By detecting the time it takes for the ions to reach the detector, the protein fingerprint of the sample can be obtained. Comparing this fingerprint with a reference library of known microorganisms allows for the identification of the microorganism species. MALDI-TOF MS has advantages such as ease of operation, high detection speed, high throughput, and low cost, and has been widely used in routine pathogen identification in clinical microbiology laboratories.

[0024] Optionally, the current batch of MALDI-TOF MS quality control data refers to the raw mass spectrometry identification data acquired from the MALDI-TOF MS mass spectrometer and belonging to the same import. This data can originate from the result file exported by the mass spectrometer or from a screenshot of the mass spectrometer software interface. One import corresponds to one batch, and the system generates a unique batch identifier for each batch. The same testing date can contain multiple batches, each batch corresponding to one raw file. The system generates corresponding source numbers for multiple batches on the same day according to the import order, and samples from different batches are distinguished by source numbers in subsequent statistics. When calculating the daily average patient score, qualified patient samples from multiple confirmed batches within the same testing date will be aggregated for calculation. In one specific implementation, such as... Figure 4 As shown, the system supports users to import quality control data by uploading. Users can upload screenshots or structured result files (such as CSV format export files) generated by the mass spectrometer to the computing device through the import interface provided by the system. Each uploaded file creates a batch.

[0025] Optionally, the quality control data includes at least the sample identification results. Sample identification results refer to the identification conclusions and related information output by the MALDI-TOF MS mass spectrometer after identifying a single sample well. In one specific embodiment, the sample identification results include, but are not limited to, well number, sample ID, microbial species name, identification confidence score, anomaly text, and remarks. The well number is used to identify the physical location of the sample on the target plate. The sample ID is used to identify the sample in the current system and serves as the basis for sample type identification, batch location, and result traceability. The microbial species name is used to record the microbial species identified in the sample. The identification confidence score is used to quantify the reliability of the identification result. The anomaly text is used to record abnormal states that occur during the identification process, such as "no peaks found," "no reliable identification," or "false positive bottle."

[0026] Optionally, the strain name refers to the species name of the microorganism identified by the MALDI-TOF MS mass spectrometer. In a specific embodiment, the strain name may adopt a Latin scientific name, such as "Escherichia coli", "Staphylococcus aureus", "Pseudomonas aeruginosa", etc. The strain name may also be in an abbreviated form or a Chinese name, such as "E. coli" and "Escherichia coli". The strain name is used for service validity verification (checking whether the strain name hits the common strain dictionary or belongs to the filtered genus) and traceability information output in subsequent steps.

[0027] Optionally, the identification confidence score (Score) is a quantitative indicator output by the identification software matched with the MALDI-TOF MS mass spectrometer, which is used to reflect the reliability of the identification result. In a specific embodiment, the value of the identification confidence score ranges from 0 to 3, and a higher score indicates that the identification result is more credible. Taking a commercial mass spectrometry system as an example, it is generally considered that a score ≥ 2.0 indicates reliable identification at the species level, a score between 1.7 and 2.0 indicates reliable identification at the genus level, and a score < 1.7 indicates that the identification result is unreliable. The identification confidence score is used in multiple links in subsequent steps: it participates in service semantic parsing as a part of the parsing result in step S2; it is used to generate the patient daily average statistical index and QC strain points in step S3; it is used to establish a dynamic reference line in step S4; it is used for trend analysis of the statistical indicators of patient samples in step S5; it is used as the judgment object for quality control rule judgment in step S6. In the manual confirmation gating link, the identification confidence score is displayed as read-only in the original parsing result; whether the strain needs to be supplemented is determined together by the preset abnormal text and whether the strain name is missing.

[0028] In the above embodiment, step S1 is an acquisition step, and its specific implementation depends on the data output form of the MALDI-TOF MS mass spectrometer, the data management process of the laboratory, and the deployment mode of the system. Those skilled in the art can select an appropriate acquisition method according to the actual application scenario, all of which fall within the protection scope of the present application.

[0029] Step S2: performing service semantic parsing on the quality control data, and identifying the sample type of each sample, wherein the sample types at least include patient sample and quality control strain sample.

[0030] Optionally, business semantic parsing refers to the process of understanding and classifying the information in each field of the MALDI-TOF MS quality control data obtained in step S1, combined with the business rules and semantic knowledge of microbial mass spectrometry quality control. Business semantic parsing differs from ordinary text recognition or data import; its core lies not only in recognizing the character content in the data, but more importantly, in understanding the meaning of this content within the business scenario of microbial mass spectrometry quality control. For example, ordinary text recognition can read the text information of a sample as "QC" or "Escherichia coli," while business semantic parsing can further understand that "QC" represents that the sample belongs to the quality control sample, and "Escherichia coli" represents the strain identification result. Whether the sample belongs to QC, a standard strain, or a patient sample is still determined by the sample identifier and preset configuration. In one specific implementation, business semantic parsing utilizes preset parsing rules and mapping relationships (including but not limited to sample identifier prefix rules, matching relationships between sample identifiers and standard strain matching keywords, and mapping relationships between abnormal text and business status) to transform the original identification results obtained in step S1 into structured data with business meaning, for subsequent track-by-track processing, statistical analysis, and rule determination.

[0031] Optionally, sample type refers to the business category to which the sample belongs in MALDI-TOF MS quality control management. It should be noted that different sample types have different functions and uses in quality control analysis. In one specific implementation, the system classifies samples into types such as patient samples, quality control strain samples, standard strains, or blank wells based on the sample identification results obtained in step S1 and preset classification rules. Sample type identification is the basis for subsequent track-based processing, determining which statistical caliber each sample will be included in.

[0032] Optionally, patient samples refer to microbial samples to be identified from clinical patients. The identification results of patient samples form the basis for clinical diagnosis and treatment. In quality control management, the identification quality of patient samples (such as the distribution and trend of identification confidence scores) reflects the operating status of the mass spectrometer under real clinical testing conditions. When the mass spectrometer is in good condition, the identification confidence score of patient samples should remain stable at a high level; when the instrument experiences problems such as contamination, laser energy shift, or abnormal spectral library matching, the identification confidence score of patient samples may systematically decline. Therefore, statistical indicators of patient samples (such as daily average scores) are an important reference dimension for evaluating the overall performance of the instrument in routine clinical testing. In one specific embodiment, the system identifies patient samples as patient samples when the sample identifier does not conform to the quality control identifier format, fails to match the enabled standard strain identifier configuration, and does not conform to the semantics of blank wells. In another specific embodiment, the system also allows users to manually adjust the sample type on the OCR verification page to correct errors that may occur during automatic identification.

[0033] Optionally, the quality control strain sample refers to a quality control strain used to evaluate the stability of the MALDI-TOF MS instrument or identification process. The quality control strain is typically a reference strain of a known species, which is analyzed periodically or with patient samples during each batch of testing. Since the expected identification results of the quality control strain are known, the stability of the mass spectrometer or identification process can be evaluated by determining whether the identification results are consistent with expectations and comparing the relationship between its identification confidence score and the background dynamic reference line parameters. The identification confidence score of the quality control strain sample is used to form a QC-side statistical sequence, calculate the dynamic reference line parameters, and execute Westgard rule determinations; the dynamic reference line parameters can be calculated only in the background without requiring QC trend plotting at the front end. In one specific implementation, the system identifies the quality control strain sample based on the matching result between the sample identifier and the preset QC quality control strain configuration.

[0034] It should be added that, in one specific implementation, the system identifies sample types based on sample identifiers and preset configurations: when the QC strain identification master switch is enabled, sample identifiers starting with QC, such as QC_01, QC-01, and QC 01, are uniformly identified as quality control strain samples; for standard strains, identification is performed according to the configured identifier matching keywords and prefixes, inclusion, exact matching, or regular expression matching methods; for blank wells, judgment is made based on empty sample identifiers, blank semantics such as B, OB, and OB, or corresponding identification results. The filter genera configuration is used to determine whether identified patient samples participate in patient-side statistics and is not used as the basis for automatically classifying standard strains into QC-side statistics.

[0035] After step S2 is completed, each sample is assigned a sample type label, which provides a classification basis for subsequent track division processing.

[0036] Step S3: The quality control data is processed by separating the samples according to the sample type, and the patient samples and the quality control strain samples are included in independent statistical categories, and corresponding statistical indicators are generated based on the identification confidence score.

[0037] In some implementations, the quality control data is processed by categorizing it according to the sample type, including: Set the quality control data to a pending confirmation status; Perform business legality verification on the sample identification results of each sample in the quality control data to be confirmed; If the verification passes, the quality control data in the pending confirmation state will be marked as confirmed. The quality control data in the confirmed state are processed by categorizing them according to the sample type.

[0038] Optionally, the "pending confirmation" state refers to the intermediate business state where quality control data has been parsed but has not yet undergone manual review and confirmation. In one specific implementation, after the quality control data is imported through step S1, the system creates a batch record for it and initially sets the batch status to "pending confirmation." Although the identification results of each sample in the "pending confirmation" state have been converted into structured data through OCR recognition or document parsing, this data has not yet undergone manual review and business legality verification. Therefore, the system cannot use it for formal quality control statistics, trend analysis, and rule determination. The "pending confirmation" state is a core component of the "confirmation gating mechanism" in this invention. Its function is to establish an isolation buffer to logically isolate the original parsed data from the formal statistical data.

[0039] It should be noted that in actual laboratory operations, the raw identification results output by the MALDI-TOF MS mass spectrometer may present several situations requiring manual intervention. For example, text in instrument software screenshots may be garbled due to OCR errors; some abnormal identification results (such as "no reliable identification") in the mass spectrometer's exported results files may lack corresponding bacterial species names; the identified bacterial species of some samples may belong to genera that require filtering according to the system configuration; and the well positions of some samples may be misidentified. These situations all require manual review and intervention to ensure the accuracy and reliability of the data entering formal quality control statistics. The pending confirmation status provides a preliminary review step for this purpose.

[0040] In one specific implementation, quality control data in a pending confirmation state is not visible on the system homepage dashboard, trend analysis page, or alarm page. That is, pending confirmation data will not affect the homepage health score, will not appear in the trend chart, and will not trigger any alarm rules. (Reference) Figure 3 , Figure 3 This is a schematic diagram of the homepage dashboard of the mass spectrometer quality control system provided in this application embodiment. The various quality control indicators displayed on the homepage dashboard (such as instrument health score, patient daily average score, QC strain score, blank quality control observation, and daily alarm count) are all calculated and generated solely based on data from confirmed batches. Only when the user is on the OCR verification page (e.g., ...) Figure 5 (As shown) Only after the data review is completed and the batch is marked as confirmed will the data of that batch be included in the calculation scope of the homepage, trends and alarms.

[0041] In some implementations, the sample identification results of each sample in the quality control data to be confirmed are subject to business legality verification, including at least one of the following: When the sample identification result contains preset abnormal text that needs to be completed and the bacterial species name is missing, a prompt will be made and the bacterial species name will be required to be completed. Check whether the bacterial species names in the sample identification results match the preset common bacterial species dictionary; The test detects whether the bacterial genus to which the bacterial species name in the sample identification result belongs matches the preset filter bacterial genus configuration.

[0042] It should be added that, in one specific implementation, the quality control data in the pending confirmation state is obtained by the user on the OCR verification page (e.g., Figure 5 The system performs item-by-item or batch verification (as shown). Users can modify fields such as well position, sample number, sample type, bacterial strain name, abnormal text, and supplementary remarks; the identification confidence score is brought in from the original identification or file parsing results and is displayed read-only on the current verification interface, and cannot be modified by the user. After the user completes the modifications, clicking the confirmation button will trigger the system to perform the above business legality verification. If the verification passes, the system will mark the quality control data in the pending confirmation state as confirmed; if the verification fails, the system will prompt the reason for the failure (such as "There are abnormal results with incomplete bacterial strains, please complete them and confirm again") and prevent the user from completing the confirmation operation until all verification items pass.

[0043] Optionally, the "confirmed" status refers to the business status of quality control data that has been officially adopted by the system after manual review and business legality verification. Data in the "confirmed" status participates in subsequent track segmentation, dynamic reference line calculation, trend analysis, and rule determination, and is presented on the homepage dashboard, trend chart, and alarm page. Users with appropriate permissions can still modify sample fields in the confirmed batch. After modification, the system will re-execute classification, filtering, and confirmation status refresh, and recalculate the statistical indicators, reference lines, and rule results for the relevant months from the batch's testing date.

[0044] Optionally, the quality control data with confirmed status participates in subsequent track-based processing. Track-based processing refers to diverting the confirmed quality control data to different statistical calibers and data processing paths according to sample type. In one specific implementation, data identified as patient samples are included in the patient-side statistical caliber for calculating the daily average patient score and generating patient trend data; samples that match the currently enabled QC quality control strain configuration are included in the QC-side statistical caliber for forming independent statistical points, calculating background dynamic reference line parameters, and executing quality control rule judgments; other standard strains are excluded from the patient-side statistical caliber but are not automatically included in the QC-side statistical caliber simply because they belong to standard strains; data identified as blank wells are independently judged for contamination status.

[0045] Specifically, refer to Figure 5 and Figure 6 Users Figure 5 After the data review and modification are completed on the OCR verification page shown, the system performs a pre-confirmation verification. After the verification is passed, the batch is marked as confirmed. The system then processes the samples according to their sample types, including patient samples and quality control strain samples in separate statistical categories. Figure 6The image demonstrates the system's verification and prompting for abnormal results before confirmation. In this example, the system detected the abnormal text "no reliable identification" and an empty bacterial strain name, prompting the user to complete the strain name before confirmation. This pre-confirmation verification is a crucial step in ensuring the quality of data processed in different tracks, ensuring that only data that has passed legality verification can be included in official statistical reporting.

[0046] In one specific implementation, after the quality control data with the confirmed status is processed in separate tracks, its sample type label is persistently stored and associated with information such as the original data source (e.g., original file name, source number), confirmation status (confirmer, confirmation time), and sample identification results (well location, strain name, score, abnormal text) for subsequent traceability information output.

[0047] It should be noted that, in one specific implementation, please refer to [link / reference needed]. Figure 7 For samples with missing bacterial species, the system provides a manual quick completion function. In one specific implementation, when the system detects during the verification process that a sample's identification result contains abnormal text (such as "no reliable identification") but the bacterial species name is empty, the bacterial species name field of that sample is marked as editable on the OCR verification page, and the user is prompted by a highlight or special indicator that the field needs to be completed. After the user clicks on the field, the system displays a drop-down selection list. The candidate list is derived from the enabled common bacterial species dictionary and filtered bacterial genus candidates. The user selects a standard name through the candidate search. Upon confirmation, the system verifies whether the non-empty bacterial name matches the enabled common bacterial full English name, or whether its genus matches the enabled filtered bacterial genus configuration. This design avoids multiple spellings of the same bacterial species due to different operator input habits (such as "E. coli", "Escherichiacoli", "Escherichia coli"), thereby ensuring the standardization of bacterial species names and the consistency of statistical standards. In another specific implementation, after the completed sample identification result is updated, the system automatically re-verifies the business validity of the sample, including reconfirming whether the completed bacterial species name matches the enabled full English names of common bacteria, or whether its genus matches the enabled filtered genus configuration. If the completed bacterial species name still fails the above verification, the system will continue to prompt the user to complete or confirm it again until all verification items pass. Figure 7 The manual quick completion function shown allows users to complete the missing bacterial species completion operation without returning to the previous page or switching to other modules, significantly improving the efficiency of handling abnormal results.

[0048] In this implementation, by setting multiple business legitimacy checks before the data to be confirmed is converted to confirmed data, including checking the completion of bacterial species for abnormal results, verifying that the bacterial species name matches the dictionary of common bacterial species, and detecting that the bacterial genus matches the configuration of the filtered bacterial genus, it is effectively prevented that incomplete abnormal results, non-standard bacterial species names, or bacterial genera that should be filtered are included in the formal statistics. This ensures the accuracy and compliance of the data participating in subsequent quality control calculations in terms of business semantics, and guarantees the reliability of the quality control conclusions from the data entry point.

[0049] In other embodiments, the quality control data are mass spectrometer screenshots, and before the quality control data is processed by track splitting according to the sample type, the method further includes: The mass spectrometer screenshots are subjected to OCR recognition to extract the well locations, sample identifiers, bacterial species names, identification confidence scores, abnormal texts, and remarks of each sample, generating structured analysis results. Contextual error correction is performed on the OCR recognition results by combining preset hole position rules.

[0050] The contextual correction of the OCR recognition results based on preset aperture position rules includes: When a character in the OCR recognition result that conforms to the semantics of a blank hole is misidentified as a non-blank hole identifier, it is corrected to a blank hole identifier.

[0051] For example, refer to Figure 5 and Figure 6 As shown. Figure 5 The OCR verification page is shown. The system converts mass spectrometer screenshots or result files into structured tabular data through OCR recognition or file parsing. The top of the page displays verification data such as well position, sample number, identified strain, score, abnormal text, and processing status, while the bottom of the page displays data source information and batch information. When the source is a CSV file, the data source area displays file information instead of the original screenshot. The system generates sample type labels based on sample identifiers, the QC strain identification master switch, standard strain matching configuration, and blank semantic rules; the strain name is not used to automatically identify the sample as a QC strain. Figure 6 After the system identifies a pre-defined abnormal text that requires completion and contains a missing bacterial species name, it marks the sample as needing manual review and prompts the user to complete it. The user can then click "edit" to view the results. Figure 7 The editing interface shown completes the bacterial species name. The system then re-executes the bacterial name validity and filtering genus verification, and refreshes the sample classification and filtering results based on the sample identifier and preset configuration, thereby ensuring that the subsequent track processing is consistent.

[0052] In another specific implementation, the system also supports the identification of standard strains and blank wells. Standard strains are reference strains identified based on preset identifier matching configurations and excluded from patient-side statistics; they are not automatically included in QC-side statistics unless designated as currently enabled QC control strains. Blank wells are used to monitor target plate contamination and matrix interference. For blank wells, if a valid bacterial species is identified or an identification confidence score exists, the blank well is considered abnormal, triggering a corresponding alarm. The identification method for standard strains and blank wells is similar to that for patient samples and quality control strain samples, based on a comprehensive judgment of sample identifiers, well location information, and preset configurations.

[0053] In some implementations, acquiring the current batch of MALDI-TOF MS quality control data includes: receiving the raw file of the quality control data, the raw file including mass spectrometer screenshots and / or structured result files; Parse the filename of the original file to extract the detection date and / or operator identifier of the quality control data; For multiple original files from the same testing date and the same instrument, a corresponding source number is generated, and the source number is associated with the quality control data of the current batch; and the traceability information is also associated with at least the original source of the quality control data; Based on the aforementioned authentication credibility score, corresponding statistical indicators are generated, including: For patient samples, the average of the identification confidence scores of all confirmed and unfiltered patient samples within the same testing date is taken as the statistical index of the daily average score of patients on that date. For quality control strain samples, select samples to be judged that match the configuration of the currently used QC quality control strains according to the confirmed batch and source number, and use the identification confidence score of the selected samples as an independent statistical indicator. For blank wells, determine whether the blank wells are normal based on the corresponding identification confidence score and / or bacterial species identification results.

[0054] In one specific implementation, the system provides a data import function, which users can use to import data. Figure 4 On the data import page shown, click "Read Data" and directly select the image or CSV file generated by the MALDI-TOF MS mass spectrometer to upload. Figure 4 The system displays the current data import interface and a list of quality control charts. Import records can be viewed by filename, date, operator, parsing status, and verification status, and operations such as parsing, verification, and deletion are provided. After upload, the system saves the original file and its storage path, associating the file with the current batch. Saving the original file allows for tracing back to the original source within the corresponding batch or import record, ensuring the integrity of the data chain.

[0055] Specifically, the system parses the filename of each received raw file according to preset naming rules to extract the testing date and operator identifier of the quality control data; the instrument identifier is determined by the import request or the current instrument configuration and is associated with the raw file and batch record. For CSV files, the filename may also contain serial number information used to determine the source number; for non-CSV files, multiple files from the same testing date can be assigned non-conflicting source numbers according to the import time and record order. If the filename does not conform to the preset naming rules, the system prompts that the filename parsing has failed and prevents the file from entering the formal parsing process.

[0056] In actual clinical microbiology laboratories, multiple raw files may be generated by the same MALDI-TOF MS instrument on the same testing date. For example, when the mass spectrometer target plate is large, the user may take multiple screenshots or export the results in batches; or different operators may export the results for their respective target plate areas on the same date. If the data from these files are simply merged by date, it is impossible to distinguish which raw file each data point comes from, making it impossible to pinpoint the specific source during subsequent tracing. Based on this, this invention proposes a source numbering mechanism.

[0057] In one specific implementation, the system assigns non-conflicting source numbers to multiple raw files from the same testing date and the same instrument. For CSV files, the source number is preferentially parsed from the filename that conforms to the preset naming rules; for non-CSV files or files whose filenames do not provide a source number, source numbers are assigned according to the import time and record order. The source number is stored as a batch attribute in the batch record, and each sample in the batch inherits this source number, which is used to distinguish different raw files from the same day.

[0058] In one specific implementation, the traceability information is at least associated with the original source of the quality control data. The original source includes the original file name, file storage path, testing date, instrument identifier, operator identifier, and source number. The anomaly display on the trend analysis page shows the testing date, statistical value, and rule details; users can first locate the batch of the day based on the testing date, and then combine this with the source number in the batch or imported records to view the corresponding original file, sample details, verification status, and confirmation information, thereby forming a traceability link from statistical points and out-of-control records to the original data source.

[0059] Specifically, for patient samples, the average identification confidence score of all confirmed and unfiltered patient samples within the same testing date is taken as the daily average score statistical index for that date. In one specific implementation, "confirmed" means that the batch containing the patient sample has passed the manual confirmation gate in step S3 and is marked as confirmed; "unfiltered" means that the bacterial species name of the patient sample does not belong to the preset filtering bacterial genus configuration. The system queries the database for the identification confidence scores of all confirmed and unfiltered patient samples within the current testing date and calculates the average value according to the following formula: Patient's average daily score = (Σ confidence score of each patient sample) / (total number of patient samples) Specifically, the patient's daily average score was used as a statistical point for subsequent establishment of a dynamic reference line and trend analysis on the patient side. (Reference) Figure 3 The homepage dashboard shown here displays a trend chart of "Patient Daily Average Score" which is based on the daily calculated average patient score. Referring to Figure 8(a), which is a schematic diagram of the interface of the mass spectrometer quality control system provided in this application embodiment, the trend analysis page shows the daily trend of patient sample identification quality with date on the horizontal axis and patient daily average score on the vertical axis.

[0060] It should be added that, in some embodiments, the method further includes: a trend testing step, the trend testing step including: In response to the trend test data input by the user, calculations are performed according to the criteria of track segmentation, dynamic reference line establishment, trend analysis, and quality control rule determination as described above. Output the inclusion status, reference source, and rule determination results for each data point in the test data; The calculation results of the trend testing step are not written into the formal business database.

[0061] Thus, this embodiment, through trend testing steps, ensures consistency in calculation methods while achieving logical isolation between test data and official data, providing users with a safe, efficient, and accurate tool for rule verification and trend review.

[0062] It should be noted that for quality control strain samples, the system selects samples that match the currently enabled QC quality control strain configuration based on the confirmed batch and source number. The identification confidence score of the selected samples is used as an independent statistical indicator, and independent points are retained according to the testing date and source number, without daily merging. Different source files from the same testing date can form separate QC-side statistical points, each used for background dynamic reference line calculation and Westgard rule determination. The system can output the QC strain score for the selected date on the homepage dashboard and output the determination result through alarms or out-of-control records when a rule is hit; the current implementation does not require the display of a QC strain trend chart as a necessary condition.

[0063] Specifically, for blank wells, their normality is determined based on the corresponding identification confidence score and / or microbial species identification result. Blank wells refer to quality control positions on the target plate where no effective microbial species or identification confidence score should appear; they are used to monitor target plate contamination, matrix interference, and instrument status. In one specific implementation, the system retains the independent status of blank wells according to the detection date and source number. Blank wells with the same detection date but different source numbers are judged independently. When a blank well does not identify an effective microbial species, has no identification confidence score, and the abnormal text is "no peaks found," it is considered normal. When an effective microbial species is identified or an identification confidence score exists, it is considered abnormal and a blank well contamination alarm is triggered. (Reference) Figure 3 The homepage dashboard shown displays the status of blank holes for the day in the "Blank Hole Quality Control Observation" area; if any blank hole from any source is abnormal, an abnormal status will be displayed.

[0064] Thus, in this embodiment, by generating source numbers for multiple original files from the same testing date and the same instrument according to the import order and associating them with the current batch, the problem of losing source information due to simple merging of data from multiple sources on the same day is solved, realizing full-link traceability from quality control judgment results to the source of original data. At the same time, by averaging the identification confidence scores of confirmed and unfiltered patient samples within the same testing date to generate the daily average patient score statistical index, using the identification confidence scores of samples selected according to the currently enabled QC quality control strain configuration in each confirmed batch as independent statistical indicators, and independently judging whether blank wells are normal based on identification confidence scores and / or strain identification results, a complete statistical indicator system is formed from three dimensions: overall identification quality of patient samples, independent QC strain locations, and blank well contamination. This provides an accurate and reliable data foundation for the subsequent establishment of dynamic reference lines, trend analysis, and rule determination.

[0065] In this embodiment, by generating source numbers for multiple original files from the same testing date and the same instrument and associating them with the current batch, independent traceability of multi-source data on the same day is achieved, making the quality control judgment results traceable to specific original files and source numbers. At the same time, by averaging the identification confidence scores of confirmed and unfiltered patient samples to generate the daily average score for patients, the identification confidence scores of each quality control strain sample are used as independent statistical indicators, as well as to independently judge the contamination status of blank wells. A complete statistical indicator system is established from three dimensions: patient sample identification quality, QC strain location, and blank well contamination, providing a reliable data foundation for the subsequent establishment of dynamic reference lines and the determination of quality control rules.

[0066] Step S4: For patient samples and quality control strain samples respectively, establish corresponding dynamic reference lines based on historical data in the statistical indicators or based on an initial statistical window that meets preset stability conditions.

[0067] Historical data in statistical indicators refers to the accumulated valid statistical indicator values ​​prior to the current statistical point. It is clear that historical data excludes the current statistical point itself; that is, the current statistical point does not participate in the calculation of its corresponding dynamic reference line, thus avoiding the logical inconsistency problem of using the current point to evaluate the current point.

[0068] It is understandable that the correspondence includes: the patient-side historical data is the daily average score sequence of patients for each date prior to the current date; the QC-side historical data is the confidence score sequence of QC strain samples selected by source number for each confirmed batch prior to the current date.

[0069] In one specific implementation, let there be n valid statistical index values ​​in the reference window, denoted as Y={y1,y2,...,y...} n}, where y k This represents the k-th valid score, where k = 1, 2, ..., n. The system uses the detection date as the horizontal axis and the statistical indicators as the vertical axis, arranging the statistical indicators for each date in chronological order to form a statistical point sequence.

[0070] The system calculates the mean X using equation (1), the sample standard deviation SD using equation (2), and the coefficient of variation CV using equation (3). X = (1 / n) × Σ(k=1 to n) y k (1) SD = √{[Σ(k=1 to n)(y k X) 2 ] / (n 1)},n>1 (2) CV = (SD / |X|) × 100%, X ≠ 0 (3) The upper limit of the control line for the m-th standard deviation, U m and lower limit L m Determine according to formula (4): U m = X + m × SD, L m = X m × SD, m∈{1,2,3} (4) When n≤1, the sample standard deviation is not statistically significant, and the system does not establish a standard deviation control line for this reference window.

[0071] Preset stability conditions refer to preset rules used to filter valid historical data points and make the initial reference window reach a statistically stable state.

[0072] The initial statistical window refers to the first stable reference window formed after the dynamic reference line is established for the first time or the quality control cycle is reset, and after being filtered through the above-mentioned preset stability conditions.

[0073] The significance of the initial statistical window lies in the fact that, during the initial baseline construction, historical data has not yet been accumulated, and statistically significant means and standard deviations cannot be directly calculated. By pre-setting stability conditions to screen early candidate points, the minimum number of valid points is accumulated first, then outliers are removed, and new points are added, ultimately forming a stable benchmark window that is not contaminated by early outlier data.

[0074] In one specific implementation, after the initial baseline establishment or quality control cycle reset, the system establishes an initial reference baseline through a stable reference window mechanism. Let the set of candidate quality control points be C = {c1, c2, ..., c...}. q}, where c i =(d i ,r i ,y i ), d i Indicates the detection date, r i Indicates the source number, y i This indicates the score for that point. The system first presses (d) i ,r i The candidate points are arranged in ascending order to obtain an ordered sequence C′. The system scans point by point along C′ and forms a temporary candidate window W_t. When the number of valid points in the window reaches the preset minimum number of valid points N_min, the temporary mean X_t and the temporary sample standard deviation SD_t are calculated according to equations (1) and (2). For any candidate point c in the window i If its score y i Satisfy |y i If X_t|>2×SD_t, then the point is determined to deviate from the current stable window and added to the stable window removal set E. After removal, the system continues to add candidate points until the number of stable and valid points that have not been removed reaches N_min, forming a stable reference window W. Therefore, the stable window establishment process achieves a closed-loop screening of "sorting, accumulation, calculation, deviation removal, and continued point replenishment", avoiding early abnormal quality control points from directly entering the reference line calculation.

[0075] It should be noted that once the initial statistical window is established, it serves as the benchmark for subsequent dynamic reference lines. As new statistical indicators are continuously added, the reference lines are dynamically updated accordingly. However, outliers that have been removed will not re-enter the reference window.

[0076] In some implementations, a corresponding dynamic reference line is established based on statistical indicators, including: For the patient side, the average daily score of patients on the same testing date is used as a statistical point in the calculation of the dynamic reference line; For the quality control strains, the confidence scores of samples selected from different source numbers on the same testing date according to the currently enabled QC quality control strain configuration are used as independent statistical points in the calculation of the dynamic reference line.

[0077] In one specific implementation, for the target month, the system acquires the daily average score data of confirmed patients within the current quality control cycle that are prior to the target month but no earlier than the sixth month prior to the target month, and excludes the dates corresponding to valid out-of-control records; the daily average score of patients on the same testing date is used as a statistical point, and a patient-side statistical point sequence is formed according to the testing date order. The data for the target month is not included in the calculation of the reference line for that target month itself. If there is no benchmark for a previous month after the start of the quality control cycle, the system can collect candidate points in chronological order within the current quality control cycle, and establish an initial reference line after the number of valid points not excluded by the stable window reaches a preset minimum.

[0078] In one specific implementation, for the target month, the system acquires QC strain sample data selected according to confirmed batch and source number within the current quality control cycle, which are located before and no earlier than the sixth month prior to the target month, and excludes dates corresponding to valid out-of-control records; the scores of quality control strain samples with the same testing date but different source numbers are used as independent statistical points. The data for the target month are not included in the calculation of the reference line for that target month; when there is no prior month benchmark, an initial statistical window that meets preset stability conditions can be used to establish a reference line.

[0079] Dynamic reference lines are established independently for the patient side and the QC strain side. On the patient side, the average daily score of patients on the same date is used as the statistical point; on the QC side, QC strain samples from different sources on the same date are used as independent statistical points. The statistical results on the patient side can be used to generate a visual trend chart; the reference line parameters and statistical point sequences on the QC side are used in the background for Westgard rule determination, and the front-end output can be QC strain scores, rule determination results, alarms, and out-of-control records, without requiring the generation of a QC trend chart.

[0080] In this embodiment, by merging patient-side data into a single statistical point for dynamic reference line calculation on a daily basis, statistical fluctuations caused by multiple patient sample sources on the same date are avoided. This ensures that the patient-side reference line stably reflects the diurnal trend of the overall identification quality of patient samples. Simultaneously, by using QC strain data as independent statistical points based on different source numbers for dynamic reference line calculation, independent point information for QC strains from multiple sources on the same day is preserved. This allows the QC-side reference line to sensitively capture fluctuations in the instrument's operating status at different times within a day, avoiding the loss of time-series information due to daily merging and thus preventing the masking of potential equipment drift or contamination issues. The patient-side and QC strain-side reference lines are established independently, each reflecting different dimensions of instrument status. Together, they constitute a two-dimensional dynamic monitoring system for MALDI-TOF MS quality control, providing differentiated and reliable benchmarks for subsequent patient-side trend analysis and QC-side Westgard rule determination.

[0081] In some implementations, the quality control data is processed by separating the samples according to their types, including including patient samples and quality control strain samples in separate statistical categories: Obtain the sample identifier of each sample in the quality control data, and obtain the preset quality control identifier, standard strain identifier matching configuration and blank identifier configuration; Based on the matching results between the sample identifier and the quality control identifier, the standard strain identifier, and the blank identifier, each sample is classified into patient samples, quality control strain samples, standard strains, and blank wells. The patient samples were included in the patient-side statistical scope. The quality control strain samples were included in the statistical scope of the quality control strains, while the standard strains were excluded from the statistical scope of the patients. The contamination status of each blank hole is determined independently.

[0082] In one specific implementation, the system acquires the sample identifier and / or anomalous text of each sample from the quality control data imported in step S1. The sample identifier (Sample ID) is information used to uniquely identify the source of a sample, and in clinical laboratories, it typically includes a sample type prefix or a specific coding rule. The anomalous text is a description of the abnormal status output by the MALDI-TOF MS mass spectrometer during the identification process. Simultaneously, the system acquires several preset configuration information, including quality control identifiers, standard strain configurations, blank well identifiers, and filter genus configurations. These configuration information are pre-set by the administrator in the system backend to guide the system in automatically identifying the business type of each sample.

[0083] In one specific implementation, the quality control identifier is used to identify sample identifiers starting with QC. After the master switch is activated, samples such as QC_01, QC-01, and QC 01 are uniformly identified as quality control strain samples. The standard strain configuration includes strain code, strain name, sample identifier matching keywords, matching method, and activation status. Blank semantic rules are used to identify empty sample identifiers and blank identifiers such as B, OB, and OB. The filtered genera configuration is a list of genera that need to be excluded from the patient-side statistical caliber (such as fungi, Gram-positive bacteria, etc.).

[0084] refer to Figure 13 This is a schematic diagram of the interface for managing the filtered bacteria in the mass spectrometer quality control system provided in an embodiment of this application. Figure 13 The interface demonstrates the administrator's settings for filtering bacterial genera. On this page, the administrator can configure which bacterial genera are automatically filtered in the daily average score statistics of patient samples, excluding fungi, Gram-positive bacteria (G+b), and the bacterial genera filtered by the administrator from the calculation of the patient mean. It also supports setting specific bacterial genera to "not filter" so that they can be included in the statistics when they are identified as objects of laboratory concern. Figure 14 This is a schematic diagram of the interface for managing common bacterial strains in the mass spectrometer quality control system provided in this application embodiment, as shown below. Figure 14 The interface showcases the standardized bacterial name maintenance feature. When adding common bacterial species, administrators can enter their full English name, abbreviation, and full Chinese name. After addition, the system synchronizes the new species to the filtered genera and sets it to not be filtered by default. When editing existing common bacterial species, the current interface only allows modification of the full Chinese name. When completing abnormal results, users search for and select standardized entries from the enabled common bacterial species and filtered genera candidates to prevent non-standard names from entering the official statistics. Genera synchronized from common bacterial species are protected and can only be deleted within the common bacterial species management system, thus preventing accidental deletion of standard dictionary entries.

[0085] In one specific implementation, the system identifies the sample type based on the sample identifier, the QC strain identification master switch, the standard strain identifier matching configuration, and the blank semantic rules; after the sample is identified as a patient sample, it determines whether the patient sample should participate in the patient-side statistics based on the matching result of the genus to which the identified bacterial species belongs and the filter genus configuration.

[0086] Specifically, when the QC strain identification master switch is enabled, sample identifiers beginning with QC are preferentially identified as quality control strain samples; sample identifiers not identified as QC strains are identified as standard strains if they match an enabled standard strain configuration; samples that meet the blank semantics are identified as blank wells; and the remaining samples are identified as patient samples. Only samples that match the currently enabled QC quality control strain configuration are included in the QC-side statistics; other standard strains are excluded from the patient-side statistics but are not automatically included in the QC-side statistics.

[0087] In one specific implementation, after a sample is initially identified as a patient sample, the system further determines whether the sample should be excluded from the patient-side statistical scope based on the filtering genus configuration. Specifically, the system extracts the genus to which the bacterial species name of the patient sample belongs and compares it with the preset filtering genus configuration. If the genus matches the filtering genus configuration, the patient sample is marked as "filtered" and is not included in the calculation of the patient's daily average score, but its identification results are still saved and can be used for other purposes. The filtering genus configuration also supports setting specific genera to "not filter" so that they can be included in the statistics when identified as objects of laboratory concern. This mechanism ensures that genera that should not be included in the patient mean do not interfere with the quality control trend. (Reference) Figure 13 Administrators can view the list of filtered and unfiltered genera, search by genera name and filtering status, and batch set selected maintainable genera to be filtered or unfiltered; protected items from common species synchronization need to be maintained on the common species management page.

[0088] After the above classification, patient samples are included in the patient-side statistical caliber for calculating the daily average score; samples matching the currently enabled QC control strain configuration are included in the QC-side statistical caliber for background dynamic reference line calculation and Westgard rule determination; other standard strains are excluded from patient statistics but are not automatically included in the QC-side statistical caliber; blank wells are independently assessed for contamination status. Different types of samples enter the data processing path corresponding to their configurations to avoid mutual interference.

[0089] In this implementation, the system combines sample identification and preset configurations to classify samples into patient samples, quality control strain samples, standard strains, or blank wells. Patient samples are used to generate daily average scores and patient trend data; samples matching the currently enabled QC quality control strain configuration are used to generate QC-side statistical sequences, calculate background dynamic reference line parameters, and execute Westgard rule judgments; other standard strains are excluded from patient statistics; blank wells are independently judged for contamination status. This classification method avoids both non-patient samples contaminating daily average scores and unconditionally incorporating all standard strains into the QC-side statistical scope.

[0090] Step S5: Based on the dynamic reference line corresponding to the patient sample, perform trend analysis on the statistical indicators of the patient sample to generate patient-side quality control observation results.

[0091] In one specific implementation, the system obtains the daily average score statistics for all confirmed and unexcluded patients within the current quality control cycle. The daily average score refers to the average confidence score of all confirmed and unfiltered patient samples within the same testing date; its calculation method has been described in detail in the relevant implementation of step S3. The system arranges the daily average scores of patients for each date in chronological order, with the testing date as the horizontal axis and the daily average score as the vertical axis, forming a sequence of patient-side statistical indicators. Simultaneously, the system obtains the patient-side dynamic reference lines established in step S4, including the mean X, sample standard deviation SD, and control lines X±1SD, X±2SD, and X±3SD.

[0092] In one specific implementation, the system compares each statistical point in the patient-side statistical index sequence with the patient-side dynamic reference line to determine whether each statistical point exceeds a preset control range. Specifically, the system performs the following analysis: For single-point deviation and cumulative over-limit rules, when the deviation of the patient's daily average score from the mean is less than 2SD, R010, R011 or R012 will not be triggered due to the deviation; whether the statistical point is out of control needs to be determined in combination with other rules such as continuous ipsilateral, cross-lateral jump values ​​and unidirectional trends. When a patient's daily average score meets the condition 2SD≤|yX|<3SD and is the first statistical point to meet this condition within the same calendar month, an R010 warning is triggered; starting from the second statistical point to meet this condition in the same month, an R012 out-of-control condition is triggered. When a patient's daily score deviates from the mean by 3 SD, R011 out of control is triggered.

[0093] In another specific implementation, the system also executes rules such as continuous ipsilateral, cross-lateral jump values, and unidirectional trends. The specific window length, ipsilateral condition, and deviation threshold of the continuous rule are determined by preset rule codes, and the judgment results are incorporated into the patient-side quality control observation results.

[0094] In one specific implementation, the system applies preset patient-side quality control rules to the daily average patient score sequence. The patient-side and QC-side use independent statistical sequences and dynamic reference lines; the same rule code or corresponding rule template can be applied to both the daily average patient score statistical point and the independent QC strain statistical point, but the judgment objects, statistical calibers, and traceability targets on both sides are independent. Patient-side rules may include rules such as single-point deviation from 2SD or 3SD thresholds, continuous same-side deviations, cross-side jumps, and unidirectional trends; among these, preset patient scoring rules participate in health score deductions, and preset direct maintenance rules are used to determine recommended maintenance status.

[0095] In one specific implementation, the patient-side quality control observation results include, but are not limited to, the following: the average daily score of patients on each testing date, the in-control / warning / out-of-control status of each statistical point, the list of hit patient-side rules, a description of trend changes (such as "seven consecutive statistical points rising"), a list of abnormal dates and corresponding abnormal descriptions. These observation results are used in subsequent step S7 to generate comprehensive quality control judgment results and are displayed on the homepage dashboard and trend analysis page.

[0096] Referring to the daily trend interface of patient strains shown in Figure 8(a), the system displays the diurnal variation of patient sample identification quality with date on the horizontal axis and daily patient score on the vertical axis, and plots dynamic reference lines for X, X±1SD, X±2SD, and X±3SD; warning points and out-of-control points are displayed with different markers. Referring to the out-of-control handling record interface shown in Figure 8(b), users can fill in the handling results, handling time, handling measures, handling personnel, and remarks for the out-of-control date and rule, and view the corresponding historical records.

[0097] In one specific implementation, the patient-side quality control observation results retain the detection date, statistical value, hit rule, and dynamic reference line parameters. The system can also query batch data for the same day based on the detection date, and further distinguish the specific original source by combining the source number in the batch or imported record. When the user clicks on the out-of-control statistical point on the interface shown in Figure 8(a), the system opens the corresponding rule details; after the user fills in the processing information on the interface shown in Figure 8(b), the system writes the processing information into the corresponding out-of-control record, forming a correlation link from trend anomaly, rule determination to processing result.

[0098] Thus, by comparing the daily average scores of patients on each date with the X±2SD and X±3SD dynamic control lines and executing patient-side rule judgments, the system can automatically identify abnormal fluctuations, systematic shifts, and trend drifts in patient sample identification quality, generating structured patient-side quality control observation results. These observation results are independent of and complementary to the QC-side rule judgments based on independent statistical sequences. The patient-side results reflect the overall performance of the instrument under real clinical testing conditions, while the QC-side results reflect the identification stability and rule compliance of the instrument under standard quality control conditions. Together, they constitute a two-dimensional evaluation system for MALDI-TOF MS quality control, providing a basis for comprehensive quality control judgments and instrument status assessments.

[0099] Step S6: Based on the dynamic reference line corresponding to the quality control strain sample, perform quality control rule judgment on the identification confidence score of the quality control strain sample; when the judgment result is out of control, generate an out-of-control record corresponding to the quality control strain sample, add the detection date corresponding to the out-of-control record to the exclusion set, and prevent the data corresponding to the detection date in the exclusion set from participating in the subsequent update of the dynamic reference line; It should be added that the specific calculation method of the dynamic reference line (including the calculation of the mean X, sample standard deviation SD, the upper and lower limits of the control line, and the process of establishing a stable reference window) has been described in detail in step S4, and will not be repeated here.

[0100] In one specific implementation, the system performs Westgard multi-rule quality control judgment on the QC strain sequence. Westgard rules are a commonly used quality control judgment rule system in clinical laboratories, including 1-2s, 1-3s, cumulative exceedance rules R012, R-4s, 4-1s, 10x, 7T, and 7x rules. The judgment logic for each rule is as follows: 1-2s rule (early warning): Within the same calendar month, when the identification confidence score of a quality control strain sample first satisfies 2SD≤|yX|<3SD, a 1-2s early warning is triggered. This rule indicates an early signal of potential random or systematic errors, but does not generate an out-of-control record; the second and subsequent statistical points in the same month that meet this interval condition are handled according to the cumulative exceedance rule R012.

[0101] 1-3s Rule (Out of Control): When the confidence score of a quality control strain sample deviates from the mean by 3 SD, a 1-3s out-of-control situation is triggered. This rule indicates an out-of-control situation, typically indicating a large random error or severe systematic bias.

[0102] Cumulative Exceedance Rule R012 (Out of Control): Within the same calendar month, only quality control points satisfying 2SD≤|yX|<3SD are counted. The first point meeting the condition is only used as a warning. Starting from the second point meeting the condition, the current point and subsequent points meeting the condition in that month are judged as out of control. Points meeting the condition can be located on the same side or opposite side of the mean. Points that have not exceeded the limit are allowed between two points meeting the condition. Accumulation will restart in the next calendar month.

[0103] R-4s rule (out of control): When the identification confidence scores of two adjacent quality control strain samples are located on opposite sides of the mean, and the difference between them (i.e., the difference between the larger and smaller values) reaches or exceeds 4 SD, R-4s out of control is triggered. This rule indicates the presence of a large random error.

[0104] The 4-1s rule (out of control): When the confidence scores of four consecutive quality control strain samples are on the same side of the mean, and the deviation of each point from the mean is at least 1 SD, a 4-1s out of control is triggered. This rule indicates the presence of a systematic bias.

[0105] 10x Rule (Out of Control): When the confidence scores of ten consecutive quality control strain samples are all on the same side of the mean, 10x out of control is triggered. This rule indicates the existence of a persistent systematic bias.

[0106] 7T Rule (Out of Control): When the confidence scores of seven consecutive quality control strain samples show a unidirectional continuous upward or downward trend, 7T out of control is triggered. This rule indicates the existence of a trend drift.

[0107] 7x Rule (Out of Control): 7x out of control is triggered when the confidence scores of seven consecutive quality control strain samples are all on the same side of the mean. This rule indicates the presence of a persistent systematic bias.

[0108] In one specific implementation, the 1-2s rule serves only as an early warning and does not generate an out-of-control record; the 1-3s, cumulative over-limit rules R012, R-4s, 4-1s, 10x, 7T, and 7x generate corresponding out-of-control records when the judgment result is out of control. Administrators can adjust the enabling status and alarm display level of the rules, but the out-of-control judgment semantics of the preset out-of-control rule code remain fixed.

[0109] In one specific implementation, when any of the aforementioned out-of-control rules (1-3s, cumulative over-limit rules R012, R-4s, 4-1s, 10x, 7T, 7x) is triggered, the system generates a corresponding out-of-control record. The out-of-control record stores fields such as trigger date, source type, source number, rule code, rule result identifier, processing status, reason, processing measures, handler, processing time, processing result, and remarks. The reason field on the current interface is generated based on the rule code selected by the user, and the user enters the processing measures, handler, processing time, processing result, and remarks. The associated rule result stores the rule level, out-of-control description, target object, and suggested action. The system associates the out-of-control record with the rule result through the rule result identifier and can trace the corresponding batch and original source by combining the detection date and source number.

[0110] In one specific implementation, when the determination result is out of control, the system automatically adds the detection date corresponding to the out-of-control record to the exclusion set for subsequent dynamic reference line calculation. The exclusion set is used to mark detection dates that should not participate in subsequent reference line calculations; when executing step S4, the system first queries the exclusion set and removes the data corresponding to the exclusion date from the reference window so that it does not participate in the subsequent calculation of mean and standard deviation. When there are multiple source numbers for the same detection date, the current implementation performs exclusion according to the detection date.

[0111] In one specific implementation, for rules such as 4-1s, 10x, 7T, and 7x that rely on continuous window triggering, the system adds the detection dates corresponding to the accompanying statistical points involved in the rule hit to the exclusion set. For the cumulative over-limit rule R012, the first point in the month that meets the condition 2SD≤|yX|<3SD is only used as a warning, and its detection date is not added to the exclusion set; starting from the second point that meets the condition, the detection dates corresponding to the current statistical point that actually triggered the out-of-control situation and the subsequent statistical points that meet the condition in the month are added to the exclusion set.

[0112] In one specific implementation, the system calculates dynamic reference line parameters in the background based on the statistical sequence of QC strains and executes Westgard rule judgments accordingly. When a QC point triggers an out-of-control rule, the system generates corresponding rule results, alarms, and out-of-control records. The alarm page displays the rule code, level, date, target object, description, suggested action, and confirmation status, and is used to view and confirm alarms; users enter handling measures, handlers, handling time, handling results, and remarks in the out-of-control handling record area of ​​the trend analysis page. The current implementation does not require displaying the Levey-Jennings trend chart of QC strains on the front end or accessing details by clicking on points in the QC chart.

[0113] In one specific implementation, the data corresponding to the detection date included in the exclusion set is not included in the subsequent calculation of the dynamic reference line, rather than just being excluded from the current calculation. This means that even if more valid data points are accumulated later, the data corresponding to the exclusion date is still excluded from the reference window and will not affect the reference line again due to being "diluted" by subsequent data. When the out-of-control record status is invalid or ignored, the backend will no longer include the date corresponding to that record in the current trend exclusion; the processing result field in the trend analysis page is used to record the handling results such as resolved, under observation, requiring retesting, or unresolved, and is saved separately from the above-mentioned valid exclusion status.

[0114] In some implementations, the quality control rule determination includes at least one of the following rules: When, within the same preset statistical period, the deviation of a statistical indicator from the mean reaches or exceeds a preset first deviation threshold but is less than a preset second deviation threshold, and the statistical indicator is the first statistical indicator to meet the deviation condition within the same statistical period, a single-point warning is triggered. When the deviation of a statistical indicator from the mean reaches or exceeds a preset second deviation threshold, a single point of loss of control is triggered, where the second deviation threshold is greater than the first deviation threshold. When the cumulative number of times a statistical indicator deviates from the mean within the same preset statistical period reaches or exceeds the first deviation threshold but is less than the second deviation threshold reaches a predetermined number, cumulative over-limit loss of control is triggered for the current statistical indicator that reaches the predetermined number and subsequent statistical indicators that meet the condition. When a predetermined number of statistical indicators are all located on the same side of the mean, and the deviation of each statistical indicator from the mean reaches the preset third deviation threshold, continuous same-side loss of control is triggered. When a predetermined number of statistical indicators are all located on the same side of the mean, an offset loss of control is triggered. When two adjacent statistical indicators are located on opposite sides of the mean and the difference between them reaches the preset fourth deviation threshold, cross-side jump value loss is triggered. When a predetermined number of statistical indicators show a unidirectional continuous upward or downward trend, it triggers a unidirectional trend out of control. The construction of the exclusion set also includes: Add the detection date corresponding to the out-of-control record to the exclusion set; For rules that rely on continuous window triggering, the detection dates corresponding to the accompanying statistical points that participate in the rule's hit are added to the exclusion set; for the cumulative out-of-control situation, only the detection dates corresponding to the current statistical indicator that actually triggers the cumulative out-of-control situation and subsequent qualified statistical indicators are added to the exclusion set.

[0115] Specifically, within the same calendar month, when a statistical indicator first satisfies 2SD≤|yX|<3SD, an R010 warning is triggered; When the deviation of a statistical indicator from the mean reaches or exceeds 3SD, 3SD runaway is triggered. Within the same calendar month, when the cumulative number of statistical indicators satisfying 2SD≤|yX|<3SD reaches the second one, the cumulative over-limit rule R012 is triggered from the current statistical indicator; the statistical indicators that meet the conditions can be located on the same side or opposite side of the mean, and there are allowed to be points that have not exceeded the limit in between; When four consecutive statistical indicators are on the same side of the mean and the deviation of each statistical indicator from the mean reaches 1SD, a 4-1s loss of control is triggered. When ten consecutive statistical indicators are on the same side of the mean, a 10x loss of control is triggered. When two adjacent statistical indicators are located on opposite sides of the mean and the difference between them reaches or exceeds 4SD, R-4s runaway is triggered. When seven consecutive statistical indicators show a unidirectional continuous upward or downward trend, 7T out-of-control is triggered. When seven consecutive statistical indicators are all on the same side of the mean, 7x out of control is triggered.

[0116] In the above implementation, by applying Westgard multi-rules to the identification confidence scores of MALDI-TOF MS quality control strain samples in the background, the system can identify different types of anomalies from rule dimensions such as single-point exceedance, cross-side jump values, continuous same-side, and unidirectional changes. By converting the judgment results of out-of-control rules such as 1-3s, cumulative exceedance rules R012, R-4s, 4-1s, 10x, 7T, and 7x into structured out-of-control records, the persistence and traceability of rule hit results are realized. By adding the detection dates corresponding to the out-of-control points and the accompanying points of applicable continuous rules to the exclusion set of subsequent dynamic reference line calculations, the impact of abnormal data on subsequent judgment benchmarks is reduced. By saving processing information in the out-of-control records and allowing independent maintenance of records to selectively associate alarms, a traceable link from anomaly discovery, rule judgment to processing records is formed, without relying on the front-end display of QC trend charts as a necessary component of this closed loop.

[0117] Step S7: Based on the patient-side quality control observation results and the results of the quality control rule judgment, generate and output the quality control judgment results and corresponding traceability information of the current batch of quality control data; the traceability information includes at least the sample type, the original data source, and the out-of-control record generated when there is an out-of-control judgment, and establish the correlation between the traceability information and the corresponding statistical points.

[0118] In summary, the embodiments of this application form a complete processing link from data acquisition to judgment result output by means of technical means such as track splitting, dynamic reference line establishment, out-of-control date exclusion and traceability information output. Compared with the prior art, it has significantly improved the comprehensiveness of quality control analysis, the accuracy of trend judgment, the reliability of anomaly handling and the traceability of results.

[0119] In some implementations, the method further includes: Receive processing information input by the user for the out-of-control record, the processing information including processing measures, processing personnel, processing time, processing result and remarks; The processing information is written into the corresponding out-of-control record, or the maintenance record containing the processing information is associated with and stored in conjunction with the corresponding alarm record; wherein, when the traceability information is output, the processing information is displayed in conjunction with it.

[0120] In one specific implementation, when the system generates a loss of control record (as described in step S6) or triggers an alarm, the user can view the alarm on the alarm page (e.g., Figure 9 View and confirm alarms (as shown in Figure 8(b)); you can fill in the out-of-control date, rule, processing result, processing time, processing measures, handler and remarks in the out-of-control processing record area on the trend analysis page (as shown in Figure 8(b)); you can also view and confirm alarms on the maintenance page (as shown in Figure 8(b)). Figure 10(As shown) Create maintenance records separately. Maintenance records can be selectively associated with corresponding alarms via alarm identifiers.

[0121] The anomaly handling information includes the reason field for rule code generation, as well as the handling measures, handler, handling time, handling result, and remarks entered by the user. The out-of-control handling records on the trend analysis page are used to save the handling process of rule out-of-control situations, with handling result options including "Resolved," "Under Observation," "Requires Retesting," and "Unresolved." The independent maintenance records on the maintenance page are used to save instrument repair, calibration, or maintenance information, with maintenance result options including "Resolved," "Pending Repair," and "Reported." These two types of records are saved separately and can be linked through rule result identifiers or alarm identifiers.

[0122] In one specific implementation, the out-of-control record itself stores a cause field generated by rule codes, as well as processing fields such as processing measures, handler, processing time, processing result, and remarks. Users can directly update the corresponding out-of-control record to form a closed-loop handling process. The system can also set up an independent maintenance record table and associate corresponding alarms through alarm record identifiers for centralized management of instrument repair and maintenance information.

[0123] In one specific implementation, the out-of-control record is associated with the corresponding alarm through a rule result identifier, and the processing information is stored in the out-of-control record; the independently maintained record can be associated with the corresponding alarm through an alarm identifier. When the system outputs traceability information, it can display the corresponding out-of-control judgment and handling content based on the rule result identifier and alarm identifier.

[0124] In one specific implementation, when outputting traceability information, the system displays associated exception handling information. This includes the following display scenarios: Scenario 1: QC and standard strain configuration. For example... Figure 11 As shown, the system manages QC identifiers uniformly through a master switch for QC strain identification and maintains standard strain codes, strain names, matching keywords, matching methods, and activation status. The system can also designate a QC quality control strain for homepage score display and QC-side rule determination. Other standard strains are excluded from patient-side statistics, but are not automatically included in QC-side statistics simply because they are standard strains.

[0125] Scenario 2: Alarm and Out-of-Control Record Association Display. When a QC strain sample triggers an out-of-control rule, the alarm page displays the rule code, level, detection date, target object, out-of-control description, suggested action, and confirmation status. The out-of-control handling record area saves the handling measures, handler, handling time, handling results, and remarks according to the detection date and rule. Users can view and confirm the corresponding alarms on the alarm page and view or fill in the handling information in the out-of-control handling record area on the trend analysis page.

[0126] Scenario 3: Centralized management of maintenance pages. For example... Figure 10 The maintenance page shown provides a centralized management view of maintenance records. Users can view information such as maintenance type, repair date, problem description, repair measures, creation time, and processing status in a list. When setting alarm tags, maintenance records can be associated with corresponding alarm information, making it easy for users to centrally view and trace the instrument's repair, calibration, and maintenance status.

[0127] Scenario 4: Traceability Information Association Display. In the traceability information generated in step S7, the system associates the original source of the quality control data, sample type, rule results, out-of-control records, and saved processing information. Users can locate the corresponding batch or abnormal record based on the detection date and view the corresponding original data by combining the source number in the batch or imported record; independently maintained records are only associated with the corresponding alarm when an alarm flag is set.

[0128] In one specific implementation, the out-of-control record itself stores a cause field generated by the rule code, as well as processing fields such as processing measures, handler, processing time, processing result, and remarks. For example, auditors can locate the out-of-control record based on the detection date and rule code to view the corresponding processing measures, handler, processing time, and processing result. Independent maintenance records are used to centrally store instrument repair and maintenance information, and can selectively associate corresponding alarms through alarm identifiers; a mandatory one-to-one correspondence is not required between the two.

[0129] In this implementation, the system associates rule-based judgments, out-of-control records, and handling results by storing processing information within the out-of-control log. Different business records are stored according to their respective purposes by allowing users to view and confirm alarms on the alarm page, fill in out-of-control handling records on the trend analysis page, and independently manage maintenance information on the maintenance page. Independent maintenance records can be optionally associated with corresponding alarms when setting alarm identifiers, but it is not mandatory for each maintenance record to be bound to an out-of-control record. This reduces manual verification work and improves the traceability of anomaly handling.

[0130] In some implementations, the method further includes: Based on the preset patient scoring rules triggered by each confirmed batch of patient samples within the current testing date, the level of each rule, and the number of samples on the current testing date, the corresponding score is deducted from the preset initial score to obtain the current instrument's health score; When a preset direct maintenance rule is met, the current instrument status is marked as requiring maintenance; the quality control judgment result includes the health score and the current instrument status.

[0131] In one specific implementation, the health score is a score calculated by the system based on preset patient scoring rules, rule levels, and the number of samples taken that day, triggered by confirmed batches of patient samples within the current testing date. This score is used to quantitatively evaluate the current operational status of the MALDI-TOF MS instrument. The health score uses a percentage scale from 0 to 100. A higher health score indicates a more stable and reliable instrument operation. The health score provides laboratory personnel with an intuitive and quantitative basis for judging the instrument's status, allowing users to quickly understand the overall operational status of the instrument without having to check each quality control indicator individually.

[0132] In one specific implementation, the health score starts from a preset initial value, and deductions are made only according to preset patient scoring rules based on fixed deduction values ​​corresponding to the rule level, with the deduction values ​​scaled according to the number of samples taken that day. Direct maintenance rules such as QC strain out-of-control issues and blank well contamination do not further deduct health scores; instead, the current instrument status is directly marked as requiring maintenance. Administrators can maintain the rule's enabled status, display level, and allowed rule parameters, but the mapping between health score deduction values ​​and rule levels is preset by the system.

[0133] In one specific implementation, the health score starts from a preset initial score. For preset patient scoring rules, the system obtains a fixed deduction value based on the rule level, and then scales and deducts the deduction value based on the number of samples on the current testing date; for preset direct maintenance rules, the system directly marks the instrument status as requiring maintenance, rather than using whether the health score is below a threshold as the sole condition.

[0134] Referring to Figure 12(a), the system displays the rule code, name, description, level, activation status, and parameters of a regular rule; referring to Figure 12(b), the system displays the Westgard quality control / out-of-control rules and their fixed judgment criteria, where R010 serves as a warning rule, and other activated out-of-control rules generate out-of-control records upon being hit and participate in subsequent benchmark elimination. Administrators can adjust the activation status and display level of rules, but this does not change the judgment semantics corresponding to each preset rule code.

[0135] In one specific implementation, the deduction value of the patient scoring rule is determined by its rule level and scaled according to the number of samples on that day; out-of-control rules such as 1-3s on the QC strain side, cumulative over-limit rules R012, 4-1s, 10x, R-4s, 7T, and 7x are direct maintenance rules, and when triggered, the instrument status is directly marked as recommended for maintenance. 1-2s is only used as a warning rule and is not judged as out of control on its own.

[0136] The deduction levels for different rules on the patient side are determined by the rule template; the specific deduction values ​​corresponding to each level and the sample size scaling method are preset by the system. Preset rules such as uncontrolled quality control strains or contamination of blank wells can directly trigger a maintenance suggestion.

[0137] In one implementation, the health score is displayed on the homepage dashboard. (See reference) Figure 3 The homepage dashboard shown is illustrated. It displays information such as instrument health score, daily average patient score, QC strain score, blank quality control observation, daily alarm count, and recent trend overview. The "Instrument Health Score" area displays the current instrument's health rating in numerical form, allowing users to intuitively understand the instrument's current status. When the health score falls below a preset threshold, the homepage alerts the user via a color change (e.g., from green to yellow or red) or a text prompt, reminding them to pay attention to the instrument's status.

[0138] For unconfirmed alarms, rule level changes will synchronously update their display level; confirmed alarms will retain their original level. When rule parameters or activation status change, the system will recalculate the rule results for the relevant months with existing quality control data.

[0139] In one specific implementation, the health score is output as part of the quality control judgment result, along with the patient-side quality control observation results, QC-side rule judgment results, and traceability information. Users can view the current health score and the deduction details of the patient scoring rules on the homepage dashboard, view the daily health score and status on the trend analysis page, and view the daily health score, status, and abnormal details in the exported monthly report; this implementation does not require the display of the health score through a QC trend chart.

[0140] In this implementation, a health score is calculated based on the patient scoring rules triggered by confirmed patient samples on the current testing date, the rule level, and the number of samples on that day. The recommended maintenance status is then determined by direct maintenance rules, generating a quantitative health score reflecting the instrument's current operating status, which is displayed on the homepage dashboard. Figure 3 The system provides an intuitive display, allowing users to quickly and easily understand the overall operating status of the instrument without having to check each quality control indicator individually. This significantly lowers the barrier to interpreting quality control data and improves the efficiency of daily quality control management. By maintaining the rule activation status, display level, and allowed rule parameters in the rule template management page, the system can adapt to the quality control management requirements of different laboratories while maintaining the stability of the judgment semantics of the rule code and the mapping of health score deductions. By incorporating health scores into the quality control judgment results and associating them with traceability information, users can view the deduction details and basis at any time, ensuring the auditability and traceability of the scoring process. Through a protection mechanism that prevents confirmed historical alarms from being overwritten in batches, the system supports dynamic adjustment of rule configurations while ensuring the stability and traceability of historical data, avoiding inconsistencies between historical quality control conclusions and original records due to rule changes.

[0141] Based on the same inventive concept as the foregoing embodiments, this application also provides a schematic diagram of a specific business process for processing MALDI-TOF MS quality control data, as shown below. Figure 2 As shown, the method includes: Step 1: User login.

[0142] Specifically, users enter their username and password through the input interface of the computing device, and the system identifies user permissions based on the user account's role configuration. The system supports two types of account roles: administrators and ordinary users, each with different operational permissions. For example, administrators have management permissions such as rule template maintenance, strain configuration, filter rule configuration, and account configuration, while ordinary users have daily operational permissions such as data import, OCR verification, homepage viewing, trend viewing, alarm confirmation, and maintenance operations. Through role-based access control, it ensures that critical configuration and confirmation operations in the quality control process are performed by authorized personnel, reducing the risk of misoperation or unauthorized modifications affecting quality control conclusions.

[0143] Step 2: Data import and batch creation.

[0144] Specifically, after logging in, users access the data import page and upload the raw files generated by the MALDI-TOF MS mass spectrometer. The raw files include mass spectrometer screenshots and / or structured result files. The system saves the raw files and their storage paths, extracting the detection date and operator identifier from the filenames according to preset naming rules. The instrument identifier is determined by the import request or the current instrument configuration. For multiple raw files from the same detection date, the system assigns a source number based on the source sequence number in the CSV filename or the import order of non-CSV files, and associates the source number with the current batch. The system creates a record for this batch that needs parsing or confirmation.

[0145] Step 3: OCR parsing.

[0146] Specifically, for raw files in mass spectrometer screenshot format, the system performs OCR recognition on the screenshots, extracting well locations, sample identifiers, bacterial species names, identification confidence scores, anomalous text, and remarks for each sample, generating structured analysis results. The system also performs contextual error correction on the OCR results based on preset well location rules, such as correcting misidentifications of characters that conform to the semantics of blank wells (e.g., misidentifying "Q11" as "B11"), but without performing global replacements to avoid incorrectly identifying bacterial names or sample numbers. For structured result files (such as CSV format), the system directly parses the file content to generate structured data.

[0147] Step 4: Business semantic analysis and sample type identification.

[0148] Specifically, the system performs business semantic analysis on the structured data obtained after OCR parsing or document parsing, combining the business rules and semantic knowledge of microbial mass spectrometry quality control. Business semantic analysis utilizes preset parsing rules and mapping relationships (including sample identifier prefix rules, matching relationships between sample identifiers and standard strain matching keywords, and mapping relationships between abnormal text and business status) to transform the original identification results into data with business meaning. The system acquires the sample identifier and / or abnormal text of each sample, and obtains preset quality control identifiers, standard strain configurations, blank well identifiers, and filter genus configurations, then integrates the above information to identify the sample type of each sample. Specifically, when the QC strain identification master switch is enabled, sample identifiers starting with QC are identified as quality control strain samples; sample identifiers not identified as QC strains are identified as standard strains if they match the enabled standard strain identifier matching configuration; samples conforming to blank semantics are identified as blank wells; and the remaining samples are identified as patient samples. Sample types include at least patient samples, quality control strain samples, standard strains, and blank wells.

[0149] Step 5: Manual confirmation of gate control.

[0150] Specifically, after step four is completed, the quality control data for the current batch is set to a pending confirmation status. While the identification results of each sample in the pending confirmation status have been converted into structured data through OCR recognition or document parsing, this data has not yet undergone manual review and business legitimacy verification. Therefore, the system cannot use it for formal quality control statistics, trend analysis, or rule determination. Quality control data in the pending confirmation status is not visible on the system's homepage dashboard, trend analysis page, or alarm page.

[0151] Specifically, users can view the comparison between the original screenshot or file content and the structured parsing results on the OCR verification page. They can modify fields such as well location, sample identifier, sample type, bacterial species name, abnormal text, and supplementary remarks. The identification confidence score is brought in from the original recognition or file parsing results and is displayed in read-only mode on the current verification interface. The pre-confirmation verification includes at least the following: when the sample identification result contains preset abnormal text that needs to be completed and the bacterial species name is missing, check and require the completion of the bacterial species name; check whether the non-empty bacterial species name matches the enabled common bacterial species full English name, or whether its genus matches the enabled filtered genus configuration.

[0152] Please see Figure 6 , Figure 6 This is a schematic diagram of the OCR calibration interface of the mass spectrometer quality control system provided in this application embodiment. Figure 6 As shown, when the system detects a preset abnormal text that needs to be completed and the strain name is empty, it prompts the user to complete the strain name in the pending list; the user clicks "edit" to proceed. Figure 7The sample editing interface shown allows users to select a standard bacterial name through a candidate search. Candidate sources include the enabled common bacterial species dictionary and filter genera candidates. After completion, the system re-verifies whether the non-empty bacterial name matches the full English name of an enabled common bacterium, or whether its genera matches the enabled filter genera configuration; if the verification fails, confirmation remains blocked.

[0153] If the verification fails, the system will display the reason and prevent confirmation; if the verification passes, the system will mark the batch as confirmed and include it in the formal quality control calculation. Users with the appropriate permissions can still modify the sample fields in the confirmed batch. After the modification, the system will re-execute the classification, filtering, and confirmation status refresh, and recalculate the statistical indicators, reference lines, and rule results for the relevant months from the date of the test.

[0154] Step Six: Quality Control Statistical Calculation and Track Separation Processing.

[0155] Specifically, the system processes the confirmed quality control data in separate tracks: patient samples are included in the patient-side statistical caliber; samples that match the currently enabled QC quality control strain configuration are included in the QC-side statistical caliber; other standard strains are excluded from the patient-side statistical caliber but are not automatically included in the QC-side statistical caliber; and blank wells are independently judged for contamination status.

[0156] For patient samples, the system averages the identification confidence scores of all confirmed and unfiltered patient samples within the same testing date, using this average as the daily average score for that date. "Confirmed" means the sample batch has passed the manual confirmation gate in step five, and "unfiltered" means the bacterial genus of the sample does not belong to the preset filtering genus configuration.

[0157] For quality control strain samples, the system uses the identification confidence score of the samples selected from each confirmed batch according to the currently enabled QC quality control strain configuration as an independent statistical indicator, retaining independent sites according to the detection date and source number, without performing daily merging. For blank wells, the system determines whether they are normal based on the identification confidence score and / or the strain identification result. When a valid strain is identified or an identification confidence score exists, the blank well is judged as abnormal.

[0158] Step 7: Establish bilateral dynamic reference lines.

[0159] The system establishes dynamic reference lines for both the patient and quality control strain sides. For the patient side, the average daily score of patients on the same testing date is used as a statistical point in the dynamic reference line calculation. For the quality control strain side, the confidence scores of samples selected from different source numbers on the same testing date according to the currently enabled QC quality control strain configuration are used as independent statistical points in the dynamic reference line calculation.

[0160] Step 8: Patient-side trend analysis.

[0161] Specifically, based on the patient-side dynamic reference line, the system performs trend analysis and rule determination on the daily average statistical points of patients: for single-point deviation and cumulative exceedance rules, R010, R011, or R012 is not triggered when the deviation is less than 2SD; the first statistical point in the same calendar month that meets 2SD≤|yX|<3SD triggers an R010 warning, and R012 out-of-control is triggered starting from the second statistical point that meets this condition; R011 out-of-control is triggered when the deviation reaches or exceeds 3SD. The system also executes rules such as continuous ipsilateral, cross-lateral jump values, and unidirectional trends, and generates patient-side quality control observation results by integrating the results of each rule.

[0162] Step 9: QC side Westgard rule determination and out-of-control removal.

[0163] Specifically, the system uses a dynamic reference line on the quality control strain side to apply quality control rules to determine the identification confidence score of the quality control strain samples. The system performs Westgard multi-rule quality control judgments on the QC strain sequences, including 1-2s, 1-3s, cumulative exceedance rules R012, R-4s, 4-1s, 10x, 7T, and 7x. Among these, 1-2s serves as an early warning and does not generate out-of-control records; 1-3s, cumulative exceedance rules R012, R-4s, 4-1s, 10x, 7T, and 7x generate corresponding out-of-control records when the judgment result is out of control.

[0164] Specifically, when any out-of-control rule is triggered, the system generates an out-of-control record, saving the trigger date, source type, source number, rule code, rule result identifier, and processing status. The system also associates the rule result identifier with the corresponding rule level, out-of-control description, target object, and suggested action. Simultaneously, the system adds the detection date corresponding to this out-of-control record to the exclusion set for subsequent dynamic reference line calculations. When consecutive rule triggers occur, the system not only marks the current trigger point as an out-of-control point but also adds the detection dates corresponding to the accompanying statistical points participating in the continuous window to the exclusion set.

[0165] Step 10: Comprehensive quality control judgment and traceability information output.

[0166] Specifically, the system generates quality control judgment results for the current batch of quality control data based on patient-side quality control observation results, quality control rule judgment results, and exclusion sets. The quality control judgment results include patient-side trend observation information, QC-side out-of-control judgment information, and the instrument's health score.

[0167] Specifically, the health score starts from a preset initial value and deducts points only according to the preset patient scoring rules, based on the fixed deduction value corresponding to the rule level. The deduction value is scaled according to the number of samples on that day. When a direct maintenance rule such as QC out of control or blank well contamination is triggered, the health score is not further deducted; instead, the instrument status is marked as requiring maintenance. The health score is displayed numerically on the homepage dashboard.

[0168] Specifically, the traceability information output by the system is linked to at least the sample type, out-of-control record, and original source. The original source includes the original file name, testing date, instrument identifier, operator identifier, and source number. Users can locate the corresponding batch based on the testing date, and then view the original file, sample details, verification status, and confirmation information by combining the source number in the batch or imported record.

[0169] Step 11: Alarm and maintenance closed loop.

[0170] Specifically, alarms generated by the rule engine are displayed on the alarm page, where users can view and confirm them. On the trend analysis page, users enter the out-of-control date, rule, handling measures, handler, handling time, handling result, and remarks in the out-of-control handling record area; this information is saved in the corresponding out-of-control record. Users can also create separate instrument repair or maintenance records on the maintenance page and selectively associate them with corresponding alarms using alarm identifiers. This forms a traceable chain: "anomaly detection → rule determination → out-of-control record → handling record → subsequent elimination → trend observation."

[0171] In this way, a complete quality control data processing flow from data import to maintenance closed loop is completed.

[0172] Based on the same inventive concept as the foregoing embodiments, this application also provides a computing device, such as... Figure 15 As shown. The computing device includes a processor 310, a memory 311 storing a computer program, and a network interface 312, all connected via a bus system 313. When the processor 310 runs the computer program, it implements the aforementioned MALDI-TOF MS quality control data processing method. The number of processors and memories can be one or more.

[0173] The memory 311 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferroelectric random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memory 311 described in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0174] In this embodiment, the memory 311 stores the computer program and business data required to implement the MALDI-TOF MS quality control data processing method described above. The business data includes original quality control documents, sample identification results, sample types, statistical indicators, dynamic reference line parameters, rule determination results, out-of-control records, and traceability information. When the processor 310 executes the computer program, it completes data acquisition, business semantic parsing, track separation processing, reference line calculation, rule determination, and traceability information output.

[0175] Based on the same inventive concept as the foregoing embodiments, this embodiment also provides a computer-readable storage medium storing a computer program; when the computer program is executed by a processor, it implements the above-described MALDI-TOF MS quality control data processing method. The computer-readable storage medium can be at least one of a read-only memory, random access memory, programmable memory, flash memory, magnetic disk, or optical disk.

[0176] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0177] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0178] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for processing MALDI-TOF MS quality control data, characterized in that, Includes the following steps: Obtain the current batch of MALDI-TOF MS quality control data; the quality control data includes at least the sample identification results, and the sample identification results include at least the strain name and the identification confidence score; The quality control data is subjected to business semantic parsing to identify the sample type of each sample, and the sample type includes at least patient samples and quality control strain samples; The quality control data is processed by separating the samples according to the sample type, and the patient samples and the quality control strain samples are included in independent statistical categories. The corresponding statistical indicators are generated based on the identification confidence score. For patient samples and quality control strain samples respectively, based on historical data in the statistical indicators or based on an initial statistical window that meets preset stability conditions, corresponding dynamic reference lines are established. Based on the dynamic reference line corresponding to the patient sample, trend analysis is performed on the statistical indicators of the patient sample to generate patient-side quality control observation results. Based on the dynamic reference line corresponding to the quality control strain sample, the identification confidence score of the quality control strain sample is determined by quality control rules; when the determination result is out of control, an out-of-control record corresponding to the quality control strain sample is generated, the detection date corresponding to the out-of-control record is added to the exclusion set, and the data corresponding to the detection date in the exclusion set is not included in the subsequent update of the dynamic reference line; Based on the patient-side quality control observation results and the results of the quality control rule determination, the quality control determination results and corresponding traceability information of the current batch of quality control data are generated and output; the traceability information includes at least the sample type, the original data source, and the out-of-control record generated when there is an out-of-control determination, and the correlation between the traceability information and the corresponding statistical points is established.

2. The method according to claim 1, characterized in that, The step of segmenting the quality control data according to the sample type includes: Set the quality control data to a pending confirmation status; Perform business legality verification on the sample identification results of each sample in the quality control data to be confirmed; If the verification passes, the quality control data in the pending confirmation state will be marked as confirmed. The quality control data with confirmed status are processed by tiering according to the sample type.

3. The method according to claim 2, characterized in that, The business legality verification of the sample identification results of each sample in the quality control data to be confirmed includes at least one of the following: When the sample identification result contains preset abnormal text that needs to be completed and the bacterial species name is missing, a prompt will be made and the bacterial species name will be required to be completed. Check whether the bacterial species names in the sample identification results match the preset common bacterial species dictionary; The test checks whether the bacterial genus to which the bacterial species name in the sample identification result belongs matches the preset filter genus configuration.

4. The method according to claim 2, characterized in that, The process of obtaining the current batch of MALDI-TOF MS quality control data includes: Receive the raw file of the quality control data, the raw file including mass spectrometer screenshots and / or structured result files; Parse the filename of the original file to extract the detection date and / or operator identifier of the quality control data; For multiple original files from the same testing date and the same instrument, a corresponding source number is generated, and the source number is associated with the quality control data of the current batch; and the traceability information is also associated with at least the original source of the quality control data; The generation of corresponding statistical indicators based on the authentication credibility score includes: For patient samples, the average of the identification confidence scores of all confirmed and unfiltered patient samples within the same testing date is taken as the statistical index of the daily average score of patients on that date. For quality control strain samples, select samples to be judged that match the configuration of the currently used QC quality control strains according to the confirmed batch and source number, and use the identification confidence score of the selected samples as an independent statistical indicator. For blank wells, determine whether the blank wells are normal based on the corresponding identification confidence score and / or bacterial species identification results.

5. The method according to claim 4, characterized in that, The establishment of a corresponding dynamic reference line based on the statistical indicators includes: For the patient side, the average daily score of patients on the same testing date is used as a statistical point in the calculation of the dynamic reference line; For the quality control strains, the confidence scores of samples selected from different source numbers on the same testing date according to the currently enabled QC quality control strain configuration are used as independent statistical points in the calculation of the dynamic reference line.

6. The method according to claim 1, characterized in that, The step of separating the quality control data according to the sample type, and including patient samples and quality control strain samples in independent statistical categories, includes: Obtain the sample identifier of each sample in the quality control data, and obtain the preset quality control identifier, standard strain identifier matching configuration and blank identifier configuration; Based on the matching results between the sample identifier and the quality control identifier, the standard strain identifier, and the blank identifier, each sample is classified into patient samples, quality control strain samples, standard strains, and blank wells. The patient samples were included in the patient-side statistical scope. The quality control strain samples were included in the statistical scope of the quality control strains, while the standard strains were excluded from the statistical scope of the patients. The contamination status of each blank hole is determined independently.

7. The method according to claim 1, characterized in that, The method further includes: Receive processing information input by the user for the out-of-control record, the processing information including processing measures, processing personnel, processing time, processing result and remarks; The processing information is written into the corresponding out-of-control record, or the maintenance record containing the processing information is associated with and stored in conjunction with the corresponding alarm record; wherein, when the traceability information is output, the processing information is displayed in conjunction with it.

8. The method according to claim 1, characterized in that, The method further includes: Based on the preset patient scoring rules triggered by each confirmed batch of patient samples within the current testing date, the level of each rule, and the number of samples on the current testing date, the corresponding score is deducted from the preset initial score to obtain the current instrument's health score; When a preset direct maintenance rule is met, the current instrument status is marked as requiring maintenance; the quality control judgment result includes the health score and the current instrument status.

9. The method according to claim 1, characterized in that, The quality control rule determination includes at least one of the following rules: When, within the same preset statistical period, the deviation of a statistical indicator from the mean reaches or exceeds a preset first deviation threshold but is less than a preset second deviation threshold, and the statistical indicator is the first statistical indicator to meet the deviation condition within the same statistical period, a single-point warning is triggered. When the deviation of a statistical indicator from the mean reaches or exceeds a preset second deviation threshold, a single point of loss of control is triggered, where the second deviation threshold is greater than the first deviation threshold. When the cumulative number of times a statistical indicator deviates from the mean within the same preset statistical period reaches or exceeds the first deviation threshold but is less than the second deviation threshold reaches a predetermined number, cumulative over-limit loss of control is triggered for the current statistical indicator that reaches the predetermined number and subsequent statistical indicators that meet the condition. When a predetermined number of statistical indicators are all located on the same side of the mean, and the deviation of each statistical indicator from the mean reaches the preset third deviation threshold, continuous same-side loss of control is triggered. When a predetermined number of statistical indicators are all located on the same side of the mean, an offset loss of control is triggered. When two adjacent statistical indicators are located on opposite sides of the mean and the difference between them reaches the preset fourth deviation threshold, cross-side jump value loss is triggered. When a predetermined number of statistical indicators show a unidirectional continuous upward or downward trend, it triggers a unidirectional trend out of control. The construction of the exclusion set also includes: Add the detection date corresponding to the out-of-control record to the exclusion set; For rules that rely on continuous window triggering, the detection dates corresponding to the accompanying statistical points that participate in the rule's hit are added to the exclusion set; for the cumulative out-of-control situation, only the detection dates corresponding to the current statistical indicator that actually triggers the cumulative out-of-control situation and subsequent qualified statistical indicators are added to the exclusion set.

10. A computing device, characterized in that, It includes a processor and a memory storing a computer program; when the processor runs the computer program, it implements the steps of the method for processing MALDI-TOF MS quality control data according to any one of claims 1 to 9.

11. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the steps of the method for processing MALDI-TOF MS quality control data as described in any one of claims 1 to 9.

Images