Methods, systems, equipment and media for assessing the reliability of test results mutual recognition
By introducing mutual recognition units and hierarchical security controls, and using multi-dimensional indicators to assess the reliability of mutual recognition of test results among medical institutions, the problem of inaccurate assessment in existing technologies is solved, and the accurate reliability assessment and security control of test results are realized.
Patent Information
- Application Number
- CN202611301286.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-26
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies cannot effectively assess the mutual recognition and reliability of test results between medical institutions. They lack real-time, multi-source, dynamic comprehensive credibility evaluation of specific sending institutions, specific receiving institutions, and specific test items, and ignore the quality of evidence and mask uncertainty.
The concept of mutual recognition unit is introduced. Through access layer, dynamic supervision layer and hard threshold judgment, evidence perception mechanism and layered security control are adopted. The credibility of the test results is evaluated by combining multi-dimensional indicators, including key access indicators, supporting access indicators and dual consistency indicators, and different standardization functions and weight calculation methods are used.
It enables accurate and reliable assessment of test results between medical institutions, improves the accuracy of mutual recognition reliability assessment, ensures the integrity and security of assessment results, and avoids complete rejection due to minor issues.
Smart Images

Figure CN123120220A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of medical test result mutual recognition technology, and relates to the assessment of mutual recognition of medical test results between medical institutions (especially medical consortia), and particularly to a method, system, equipment and medium for assessing the reliability of mutual recognition of test results. Background Technology
[0002] To standardize medical examination and testing practices among medical consortia, ensure medical quality and safety, conserve medical resources, improve the efficiency of medical services, and enhance patients' medical experience, it is necessary to carry out and promote the mutual recognition of examination and testing results among various medical institutions (including medical consortia and medical alliances).
[0003] The so-called mutual recognition of medical test results means that if a patient has a test done at one hospital, the doctor at another hospital will also recognize the test results, eliminating the need for repeated testing. In other words, a medical institution can affix a corresponding mutual recognition label to mutually recognized test reports, allowing patients to bring these reports to other hospitals for treatment and thus achieving mutual recognition of the test results.
[0004] Internal quality control (IQC) and external quality assessment (EQA) are two complementary and irreplaceable core components of a clinical laboratory quality assurance system. The former focuses on routine monitoring within the laboratory, while the latter focuses on external comparisons between laboratories.
[0005] Internal quality control involves laboratory staff using a series of methods to continuously evaluate the reliability of the laboratory's work to determine whether a report can be issued. It is used to control the precision of routine laboratory work, improve the consistency of batch and day-to-day specimen testing, and eliminate or reduce random errors. External quality assessment involves an independent external organization organizing multiple laboratories to analyze the same specimen, using inter-laboratory comparisons to determine the calibration level and testing capabilities of each laboratory. It is used to evaluate the accuracy and comparability of results, identify inter-laboratory differences, and confirm whether a laboratory possesses the capability to perform specific tests.
[0006] Patent application No. 202410867387.4 discloses a method, apparatus, device, and storage medium for mutual recognition of clinical test results (Publication No. CN118866214A). The mutual recognition method includes: responding to a patient's request for test orders at any medical institution, determining the ordered test items and mutually recognized test reports at each medical institution, wherein each mutually recognized test report records at least one mutually recognized test item; and determining the mutually recognized test result corresponding to the ordered test item from the mutually recognized test report based on the mutual recognition time limit corresponding to the ordered test item, a preset mapping relationship between mutually recognized test items, and mutual recognition relationships between medical institutions. This enables effective mutual recognition of test results across different medical institutions, ensuring the accuracy and convenience of mutual recognition of test results across different medical institutions.
[0007] The invention patent application with application number 202510916087.5 also discloses a medical testing data management system and management method (publication number CN120878014A). For mutual recognition, the laboratory institution initiates a data mutual recognition application to the data center platform. The data center platform verifies and analyzes the application information submitted by the laboratory institution and the historical verification of its corresponding medical testing data, and grants the laboratory institution mutual recognition qualification according to preset conditions.
[0008] Similar to the aforementioned invention patents, existing technologies only record and disclose the historical verification and analysis of medical testing data to determine whether test results for a certain test item can be mutually recognized between medical institutions. However, existing technologies do not disclose how to determine or assess whether a test result can be mutually recognized between medical institutions, nor do they disclose how medical institutions conduct credibility assessments of the mutual recognition of test results. They also do not specify which indicator data needs to be obtained when assessing the credibility of mutual recognition of test results between medical institutions, or what algorithms and logical judgments are used to ultimately assess the credibility of mutual recognition. A multi-level credibility assessment and security control mechanism targeting specific providers, recipients, mutually recognized items, and assessment points has not yet been established. Therefore, it is necessary to provide a method for assessing the credibility of mutual recognition of test results, used to evaluate the credibility (i.e., priority acceptance, normal acceptance, cautious acceptance, and not recommended acceptance) of a test result issued by one medical institution in mutual recognition by another medical institution. Summary of the Invention
[0009] The purpose of this invention is to address the technical problem that existing technologies cannot effectively assess the credibility of a test result issued by one institution that is mutually recognized by another institution, or that the accuracy of mutual recognition credibility assessment is low. This invention provides a method, system, device, and medium for assessing the mutual recognition credibility of test results. This assessment method enables one institution to effectively and accurately assess the mutual recognition credibility of a test report issued by another institution, allowing the institution to draw conclusions such as priority acceptance, normal acceptance, cautious acceptance, or not recommending acceptance of the test report, thereby achieving the assessment of mutual recognition credibility and improving the accuracy of mutual recognition credibility assessment.
[0010] To achieve the above objectives, the present invention specifically adopts the following technical solution: A method for assessing the reliability of mutual recognition of test results includes the following steps: Step S1: Determine the mutual recognition unit; According to the results provider Intended to the recipient of the results Mutual recognition projects provided At the current assessment point To meet the mutual recognition credibility assessment needs, and to construct mutual recognition units. ; Step S2: Form the mutual recognition data field D(U); Around the mutual recognition unit From the perspective of the results provider Result recipient Data is collected from quality management systems, interlaboratory quality assessment systems, and / or regional mutual recognition platforms to obtain the mutual recognition data domain D(U); wherein, the mutual recognition data domain D(U) includes access layer data. Regular regulatory data from institutions Regular monitoring data of projects Real-time monitoring data of projects ; Admission layer data Including key access indicators Support for entry indicators And dual consistency index ; Step S3: Admission layer determination; First, the key access indicators Make a judgment; if key access indicators If all criteria are approved, then the support for the admission criteria will be established. Consistency index with duality Make a judgment; otherwise, key access indicators If the judgment fails, the output should not be accepted. Further support for access indicators Consistency index with duality Make a judgment, if the admission criteria are supported Support admission score, dual consistency index If the dual consistency index is greater than or equal to the corresponding qualified threshold, proceed to step S4; if the supporting admission score and the dual consistency index are both greater than or equal to the corresponding lower bound of the gray zone and less than the corresponding qualified threshold, supplementary evidence is required before making a judgment; if the supporting admission score and / or the dual consistency index are less than the corresponding lower bound of the gray zone, the output is "not recommended for acceptance". Step S4: Dynamic regulatory layer determination; The observation range is matched based on the raw value of indicator i in dimension d. Then, a standardization function is matched based on the observation range, and the standardized score of indicator i in dimension d is obtained based on the standardization function. Finally, the standardized score of indicator i in dimension d is compared with the corresponding weights, and the raw score is regularly monitored by the calculation agency. Project periodic monitoring raw score Project Real-time Monitoring Raw Score ; Step S5: Hard threshold determination; The indicators in dimension d are divided into indicators, and hard threshold conditions are determined. Based on the determination results, the periodic regulatory data for each institution is determined. Regular monitoring data of projects Real-time monitoring data of the project Corresponding hard threshold level ; Step S6: Final output of the linkage decision; Based on the raw scores of each dimension d , , Determine the corresponding original score level ; Based on the raw score levels of dimension d Hard threshold level Choose the lower level as the final level for this dimension d. According to the final level of each dimension The final output is obtained; the final output includes priority acceptance, normal acceptance, cautious acceptance, and not recommended acceptance.
[0011] Furthermore, in step S2, key access indicators Including instrument calibration status Difference between testing time and accreditation period These two indicators; Support for entry indicators Including the completeness rate of IQC records over the past 30 days Historical clinical acceptance rate These two indicators; Duality consistency index Including the consistency of testing instruments Methodological consistency Consistency of unit of result Consistency of reference interval mapping These four indicators; Regular regulatory data from institutions Including historical clinical mutual recognition rate This one indicator; Project periodic monitoring data Including the CV value of IQC over the past 30 days IQC pass rate in the past 30 days Recent first concentration EQA results Recent second concentration EQA results Recent EQA results for the third concentration Inter-laboratory comparison These 6 indicators; Real-time monitoring data of projects Including the indoor quality control status of the first concentration of the day The second concentration indoor quality control status on the same day The third concentration of the day in the indoor quality control status These three indicators.
[0012] Furthermore, the indicators in the mutually recognized data domain D(U) are classified into three categories: benefit-type indicators, cost-type indicators, and binary or ordinal indicators. For benefit-type indicators and cost-type indicators, different standardization functions are used to calculate the standardized scores; binary indicators are scored according to two states, and ordinal indicators are scored according to a level mapping table.
[0013] Furthermore, in step S3, the specific steps for determining the admission layer are as follows: Step S3-1: Determine key admission criteria ; The instrument calibration status was checked separately. Difference between testing time and accreditation period Perform a judgment; if both judgments pass, proceed to step S3-2; if the instrument is in calibration status... Difference between testing time and accreditation period If at least one of the results is deemed unacceptable, the output should not be accepted. Step S3-2: Calculate the support admission score and dual consistency index; Based on the completeness rate of IQC records over the past 30 days Standardized score Weight Calculate and obtain the contribution value of the integrity rate. Based on historical clinical acceptance rates Standardized score Weight Calculate and obtain the contribution value of the recognition rate. Based on the contribution value of the completeness rate , Contribution value of recognition rate Calculate and obtain the support admission score ; Based on the consistency of the testing instruments Matching degree Weight Calculate and obtain the contribution value of the detection instrument. According to methodological consistency Matching degree Weight Calculate and obtain the methodological contribution value. Based on the consistency of the units of the results Matching degree Weight Calculate and obtain the unit contribution value. Based on the consistency of the reference interval mapping Matching degree Weight Calculate and obtain the reference interval mapping contribution value. According to the contribution value of the testing instrument Methodological contribution value Result unit contribution value Reference interval mapping contribution value Calculate and obtain the duality index ; Step S3-3: Support admission and duality consistency determination; If admission points are supported Dual consistency index If all values are greater than or equal to the corresponding qualifying threshold, then the application is deemed approved and proceeds to step S4; if the application is supported... With duality index All are greater than or equal to the corresponding lower bound of the gray area, and support admission scoring. or / and duality index If the score is below the corresponding qualifying threshold, supplementary evidence will be required before a final determination is made; if the admission score is supported... or / and duality index If the value is less than the corresponding lower bound of the gray area, the output is not recommended for use.
[0014] Furthermore, in step S3-2, the standardized scores of the indicators for each category are calculated as follows: For benefit-type indicators, let... ,in , , All are level thresholds. Let be the lower bound of the acceptable numerical range; then, determine the standardization function based on the observed value of the i-th indicator under dimension d, and calculate the standardized score based on the standardization function; where the relationship between the range of observed values and the standardization function is:
[0015] For cost-type indicators, let... ,in , , All are level thresholds. Let be the upper limit of the acceptable numerical range; then, determine the standardization function based on the observed value of the i-th indicator under dimension d, and calculate the standardized score based on the standardization function; where the relationship between the observed value range and the standardization function is: .
[0016] Furthermore, in step S3-2, based on the sufficiency of samples at each stage, the weights of each indicator are calculated as follows: When the initial sample size is insufficient, calculate the subjective safety weights separately. Weight of objective information Then, calculate the combined weight based on both subjective and objective weights. Combined weights That is, the weight of indicator i; Once sufficient samples are available in the later stages, the empirical importance vector will be constructed. Combined weights By integrating and recalibrating based on historical experience, Recalibrating based on historical experience That is, the weight of indicator i; Among them, subjective safety weight The calculation formula is: ; , , , ; Objective information weight The calculation formula is: ; ; ; ; Combined weights The calculation formula is: ; Historical experience recalibration The calculation formula is: ; in, This indicates that the consistency index is achieved under the premise that the following conditions are met. Minimum, This represents the weight to be determined for the i-th indicator. This indicates the weight of the optimal indicator B. This represents the weight of the worst-case indicator W. This represents the importance scale value of the best indicator B given by the decision-maker relative to indicator i. This represents the importance scale value of indicator i given by the decision-maker relative to the worst-case indicator W. Indicates consistency index; Let represent the standard deviation of the i-th indicator in dimension d across all evaluated objects. The Pearson correlation coefficient represents the relationship between index i and index k. This indicates the number of indicators k. This indicates the amount of information contained in index i; This represents the original observation value of the j-th object in dimension d on the i-th index. Indicates that index i under dimension d is among all The arithmetic mean of a sample, This represents the total number of samples in dimension d; Indicates the degree of safety orientation. The weights represent the combined weights. This represents the empirical importance vector.
[0017] Furthermore, in step S6, the raw scores for each dimension d are... , , Determine the corresponding original score level At that time, the standard for classifying the original score level was: 90 ≤ original score ≤ 100. The answer is A; 80 ≤ raw score < 90. The answer is B; 70 ≤ raw score < 80. The answer is C; 0 ≤ original score < 70. The answer is D; Based on the raw score levels of the agency's regular monitoring. and the corresponding hard threshold level The final level of receiving regular supervision from the agency Based on the original score level of the project's regular monitoring. and the corresponding hard threshold level The final level of project regular monitoring Based on the original score level of the project's real-time monitoring and the corresponding hard threshold level The ultimate level of real-time project monitoring ; The final output standard is: the final rating of the institution under regular supervision. The final level of regular project supervision Ultimate level of real-time project monitoring If all are grade A, the final output will be the one that is accepted first; if the final grade is... , , If none of the values are C or D, and at least one is B, then the final output is considered normal; if the final level... , , If a grade C exists and no grade D exists, the final output is "cautious acceptance"; if the final grade... , , If a D-level is present, the final output will be "not recommended for adoption".
[0018] A reliability assessment system for mutual recognition of test results, comprising: The mutual recognition unit determination module is used to determine the result provider. Intended to the recipient of the results Mutual recognition projects provided At the current assessment point To meet the mutual recognition credibility assessment needs, and to construct mutual recognition units. ; The mutual recognition data field D(U) forming module is used to form a mutual recognition unit. From the perspective of the results provider Result recipient Data is collected from quality management systems, interlaboratory quality assessment systems, and / or regional mutual recognition platforms to obtain the mutual recognition data domain D(U); wherein, the mutual recognition data domain D(U) includes access layer data. Regular regulatory data from institutions Regular monitoring data of projects Real-time monitoring data of projects ; Admission layer data Including key access indicators Support for entry indicators And dual consistency index ; The admission tier determination module is used to first determine key admission indicators. Make a judgment; if key access indicators If all criteria are approved, then the support for the admission criteria will be established. Consistency index with duality Make a judgment; otherwise, key access indicators If the judgment fails, the output should not be accepted. Further support for access indicators Consistency index with duality Make a judgment, if the admission criteria are supported Support admission score, dual consistency index If both the dual consistency index and the dual consistency index are greater than or equal to the corresponding qualified threshold, the system will proceed to the dynamic regulatory layer judgment module. If both the support admission score and the dual consistency index are greater than or equal to the corresponding lower bound of the gray zone and less than the corresponding qualified threshold, supplementary evidence will be provided before judgment is made. If the support admission score and / or the dual consistency index are less than the corresponding lower bound of the gray zone, the system will output "not recommended for acceptance". The dynamic regulatory layer determination module is used to match the observation range with the raw value of indicator i in dimension d, then match a standardization function with the observation range and obtain the standardized score of indicator i in dimension d based on the standardization function. Finally, it calculates the original score of periodic supervision by the agency based on the standardized score of indicator i in dimension d and the corresponding weight. Project periodic monitoring raw score Project Real-time Monitoring Raw Score ; The hard threshold determination module is used to classify the indicators in dimension d, determine the hard threshold conditions, and determine the periodic regulatory data of the institution based on the determination results. Regular monitoring data of projects Real-time monitoring data of the project Corresponding hard threshold level ; The final output module of the linked decision-making process is used to calculate the raw scores of each dimension d. , , Determine the corresponding original score level ; Based on the raw score levels of dimension d Hard threshold level Choose the lower level as the final level for this dimension d. According to the final level of each dimension The final output is obtained; the final output includes priority acceptance, normal acceptance, cautious acceptance, and not recommended acceptance.
[0019] A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method described above.
[0020] A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the above-described method.
[0021] The beneficial effects of this invention are as follows: 1. This invention innovatively introduces three core innovations when assessing the credibility of mutual recognition of test results: ① Relational assessment: Shifting the assessment object from "institution" to "mutual recognition unit (U)," solving the problem of existing technologies lacking real-time, multi-source, and dynamic comprehensive credibility evaluation for specific sending institutions, receiving institutions, and test items; ② Evidence perception mechanism: Introducing "evidence correction coefficient" and "evidence level upper limit," solving the problem of existing technologies ignoring evidence quality and masking uncertainty; ③ Layered security control: Constructing a three-layer security network of "hard threshold - evidence upper limit - circuit breaker," solving the problem of existing technologies using total score to mask key risks. Based on the above three core innovations, accurate assessment of the credibility of mutual recognition of the same test report between medical institutions can be achieved, improving the accuracy of medical institutions' assessment of the credibility of mutual recognition of test reports.
[0022] 2. Mutual recognition reliability is not a single attribute. This invention designs a comprehensive status composed of multiple indicators from four dimensions: access level, periodic institutional supervision, periodic project supervision, and real-time project supervision. It covers a multi-dimensional analysis and judgment of whether the mutual recognition basis is established (i.e., whether the result provider and the result recipient have the sameness, compatibility, and comparability in a specific project), whether the operational status is stable (i.e., whether the result provider continuously maintains an acceptable range at the institutional and project levels), and whether it is safe at the current time (i.e., whether there is an immediate risk of anomalies in the project at the current time). The judgment of mutual recognition reliability assessment is entirely based on complete, direct, and timely data. For a certain test report, it can effectively assess the credibility of the test report in mutual recognition among medical institutions, realize the accurate assessment of the mutual recognition reliability of the same test report among medical institutions, and improve the accuracy of medical institutions' assessment of the mutual recognition reliability of test reports.
[0023] 3. In this invention, during the credibility assessment, the following steps are taken: access layer judgment, dynamic supervision layer judgment, and hard threshold judgment. The access layer judgment is used to determine whether the mutual recognition relationship can be established, which is a prerequisite for mutual recognition. If the access foundation is not established, the dynamic supervision and hard threshold judgment will not be carried out. The dynamic supervision layer no longer discusses whether the mutual recognition relationship is established, but focuses more on whether the subject H(A) is continuously in a mutually recognized state for project X. In the hard threshold judgment, key indicators have veto power, which can prevent "high scores from covering up major defects" and non-key indicators from having controllable shortcomings, avoiding complete rejection due to minor issues. It is particularly suitable for high-risk scenarios such as mutual recognition of medical results, which not only ensures the bottom line of safety, but also provides necessary management flexibility. Through the above three judgments, the accuracy of medical institutions' credibility assessment of mutual recognition of test reports can be greatly improved.
[0024] 4. In this invention, for benefit-type indicators, cost-type indicators, and target-type indicators, when calculating the standardized score of the i-th indicator under dimension d, different standardized functions are used to calculate the standardized score for different indicator types. The standardized function adopts piecewise linear standardization, which transforms indicators with different dimensions and directions into a unified comparable score. Therefore, the calculation of the standardized score of the i-th indicator under dimension d is more accurate and efficient, improving the accuracy of medical institutions' assessment of the mutual recognition reliability of test reports.
[0025] 5. In this invention, when determining the weight of the i-th indicator under dimension d, the weights within the dimension and between dimensions are not "completely copied from the same algorithm," but rather follow the same theoretical mechanism and adopt different intensities at different levels. Within a dimension, a complete three-stage combination of weights is suitable, while between dimensions, expert semantic prior guidance and gentle correction based on historical data are more appropriate. The objective information weights innovatively adopt a data-driven, unbiased weighting method, avoiding biases caused by subjective judgments. This ensures that the calculation of objective information weights can automatically "emphasize" those indicators that truly distinguish between good and bad and provide unique information, making the final mutual recognition reliability assessment more objective and robust, and improving the accuracy of medical institutions' mutual recognition reliability assessment of test reports. Attached Figure Description
[0026] Figure 1 This is a flowchart illustrating the present invention; Figure 2 This is a schematic diagram of the admission layer determination process in this invention; Figure 3 This is a schematic diagram illustrating the generation of the weight of index i in dimension d in this invention. Detailed Implementation
[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.
[0028] Therefore, all other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0029] In addition, it should be noted that the main steps, measures and means in the technical solution described in the following embodiments and test examples are not innovative points. The relevant parameter settings, threshold settings and technical details that have not been described / clearly described are not innovative points. Those skilled in the art can set them themselves according to actual conditions and needs and implement them with reference to existing technologies without creative effort.
[0030] Example 1 This embodiment provides a method for assessing the reliability of mutual recognition of test results. This method is applicable to test items included in the mutual recognition program of county-level medical consortia and actually retrieved by the platform. Typical applicable scenarios include: 1) the result provider H(A) intends to provide mutually recognized results to the result recipient H(B); 2) the platform conducts a prior review of whether a certain item has the basis for mutual recognition; 3) the platform continuously monitors items already included in the mutual recognition system; 4) the platform makes a final acceptance judgment on a specific retrieved result; and 5) the platform conducts rectification review and restoration management for abnormal items. Each step of this reliability assessment method is implemented based on computer software programs, and specifically includes the following steps: Step S1: Determine the mutual recognition unit; According to the results provider Intended to the recipient of the results Mutual recognition projects provided At the current assessment point To meet the mutual recognition credibility assessment needs, and to construct mutual recognition units. .
[0031] For the results provider In mutual recognition projects A certain test result, the recipient of the result At the current assessment point The need for mutual recognition reliability of the test results will be addressed by constructing mutual recognition units through computer programs. This mutual recognition unit Including mutual recognition parties and mutual recognition projects and the current assessment time point Different mutual recognition parties, different mutual recognition projects. and different current assessment points Different mutual recognition units can be built. Construct this mutual recognition unit. The purpose is to facilitate the subsequent acquisition of relevant data.
[0032] This method defines the basic object of mutual recognition assessment as the mutual recognition unit and represents it using a unified notation: ; in, Indicates the provider of the results. Indicates the recipient of the result. Indicates mutual recognition projects, This indicates the current assessment point in time.
[0033] Therefore, this method evaluates not the institutions themselves, but the mutual recognition relationships between institutions at a specific project and point in time. Furthermore, this definition of the object has the following significance: first, it transforms "mutual recognition" from a general quality discussion into a relational judgment; second, it avoids simply equating the overall level of an institution with the credibility of project-level results; and third, it ensures that all subsequent stages, including access, supervision, release, and rectification, can revolve around the same object.
[0034] Step S2: Form the mutual recognition data field D(U); Around the mutual recognition unit From the perspective of the results provider Result recipient Data is collected from quality management systems, interlaboratory quality assessment systems, and / or regional mutual recognition platforms to obtain the mutual recognition data domain D(U); wherein, the mutual recognition data domain D(U) includes access layer data. Regular regulatory data from institutions Regular monitoring data of projects Real-time monitoring data of projects ; Admission layer data Including key access indicators Support for entry indicators And dual consistency index It should be noted that the data and indicators in each dimension here refer to the values of each indicator in the corresponding dimension of data, and the values of related indicators in each indicator.
[0035] The mutually recognized data domain D(U) consists of the specific values of the corresponding indicators in four dimensions, i.e., the access layer data. Values of various indicators and regular regulatory data from institutions Values of various indicators and regular monitoring data of projects The values of each indicator and real-time monitoring data of the project. Values of various indicators, access layer data The values of each indicator include key access indicators. The values of each indicator and the supporting access indicators The values of each indicator, and the dual consistency index The values of each indicator (and the other three dimensions are similar) can be calculated based on the collected data and the meaning of each indicator. Based on the collected data, the specific values of each indicator can be calculated by those skilled in the art according to the meaning of the corresponding indicator, without requiring creative effort; that is, the algorithm for the corresponding indicator value is pre-embedded in the program. After obtaining basic data from relevant objects, systems, and platforms, the program can directly calculate and obtain the value of each specific indicator (which can be a specific benefit / cost value, such as an acceptance rate of 95%, a pass rate of 90%, and a completeness rate of 100%; or it can be binary or ordinal, such as the instrument calibration status indicating that the most recent calibration was within the allowable time). The specific values of each indicator will be used for subsequent judgment.
[0036] Around the mutual recognition unit The data to be collected is collectively referred to as the mutual recognition data domain D(U), which can be represented as: ; in, This represents admission layer data, used to determine whether mutual recognition relationships meet basic conditions such as identity, compatibility, and comparability; This indicates that the agency regularly monitors data to reflect the long-term foundation of the results provider in overall quality governance, periodic stability, and management support. This represents the periodic monitoring data of the project, used to reflect the long-term performance, cyclical instability, and project-level governance status of the mutual recognition project X; This represents real-time monitoring data for the project, reflecting the immediate quality status, trend changes, and current risks of the mutually recognized project X in the current period.
[0037] When collecting data, the data collection should meet five principles: completeness, consistency of terminology, temporal consistency, traceability, and multi-source corroboration. Completeness means that the data can support a closed-loop judgment in all stages of access, supervision, release, and rectification. Consistency of terminology means that the definition, unit, direction, and statistical scope of the same indicator remain consistent across different institutions and windows. Temporal consistency means that data from different sources have comparable time benchmarks. Traceability means that all data can be traced back to original business records, quality records, or platform records. Multi-source corroboration means that key conclusions should be based on direct evidence as much as possible, with acceptable alternative evidence introduced only when necessary.
[0038] In the time window settings, a basic window can be set. With extended window ,in The basic window is used for routine assessment, while the expanded window is used for supplementary judgment when the sample is insufficient. However, the conclusions drawn from the expanded window should reflect their uncertainty in subsequent evidence revision.
[0039] For data from periodic monitoring (including institutional and project-based periodic monitoring), not all projects must be included in the annual ongoing evaluation in exactly the same way. To ensure representativeness, continuity, and comparability, an annual project sample set should be compiled, denoted as: ; in, This represents the set of items to be extracted in year y. This indicates the actual number drawn that year. Indicates the minimum number of samples to be drawn. This represents the xth actual item extracted.
[0040] Minimum number of draws It is not a single constant, but can be understood as a function of the combined effects of the number of organizational layers, project scale, and implementation capabilities. ;in, Indicates organizational hierarchy. Indicates the project scale. It indicates the ability to implement.
[0041] The significance of this type of project sampling lies in the fact that the sampling quantity should not be arbitrarily determined based on experience, but should be jointly constrained by "assessment coverage needs" and "implementation capacity." As long as n ≥ a, it means that the annual sampling meets the minimum evaluable requirements in terms of quantity; if n is much lower than a, the annual evaluation cannot guarantee representativeness.
[0042] In addition to meeting the minimum quantity requirement, project sampling should also meet coverage requirements. If projects are divided into H main categories based on professional categories, the annual sampling should ensure sufficient coverage of the main categories, with an available coverage rate of [missing information]. It indicates its sufficiency requirement and satisfies it. Here Determined by business management objectives, the theoretical implication is that the extracted set should not be concentrated on a few easily evaluated items in the long term, but should be able to represent the main professional aspects of the mutual recognition work among various institutions.
[0043] In the specific extraction and ranking, a project extraction priority Q(X) can be introduced to characterize the relative priority of a project entering the annual extraction set. Theoretically, Q(X) can be composed of factors such as "actual implementation level, mutual recognition relevance, usage frequency, sample sufficiency, and professional representativeness." The purpose of this approach is not to create a complex algorithm, but to clarify that project extraction itself should be a theoretically grounded selection process, rather than relying entirely on arbitrary manual designation.
[0044] For projects entering the evaluation, attention should be paid to the sufficiency of their sample size within the basic evaluation window. If... Indicates project In the basic window The effective sample size within, in If represents the minimum sample requirement, then when When, a judgment can be made directly within the basic window; when At that time, it can be expanded to Recalculate; if the condition is still not met after expansion. In such cases, numerical values should not be used to fill in the gaps and real evidence should be treated as "insufficient evidence".
[0045] This rule indicates that insufficient sample size should be considered an evidentiary constraint rather than a statistical blank. The purpose of extending the window is to obtain as many observational bases as possible to support stable judgments; if the bases are still insufficient, uncertainty should be retained rather than forcing definitive conclusions.
[0046] The annual list of sampled items should, in principle, remain stable throughout the year to ensure the interpretability of horizontal and vertical comparisons. If adjustments are necessary, the reasons for the adjustments, the affected items, and the effective date should be recorded, along with an explanation of whether the adjustments affect the continuity of previous evaluations. The list of sampled items, window setting rules, minimum sample requirements, and exception handling should be filed and managed. The theoretical significance of filing is that the evaluation framework itself should also be subject to constraints; otherwise, the same item may lose comparability at different times due to changes in the sampling rules.
[0047] After obtaining the relevant data, the data for each indicator can be obtained directly from the acquired data, or it may require certain statistical or computational steps. Once the relevant data is obtained, those skilled in the art can calculate and obtain the corresponding data for each indicator based on its meaning, without requiring any creative effort.
[0048] Admission layer data Including key access indicators Support for entry indicators and duality consistency index These indicators are used to determine whether a mutual recognition relationship meets basic conditions such as identity, compatibility, and comparability. A comparison of the three indicators at the admission layer is shown in Table 1: Table 1 Comparison of the three types of indicators for the admission tier
[0049] The core of the admission layer is not simply "checking whether a few conditions are met," but rather determining whether the result provider H(A) and the result receiver H(B) have established a mutually recognized basic relationship on project X. Therefore, the admission layer data... Classified as key access indicators Support for entry indicators and duality consistency index : , , , ; in, and The assessment focuses on the "qualifications and conditions of an individual institution".
[0050] Key entry indicators These key entry indicators are used to assess whether the result provider H(A) or the result recipient H(B) meets the most basic and mandatory requirements for conducting Project X testing. (Although the system directly obtains the relevant indicator data during mutual recognition reliability assessment, and calculation and evaluation can be performed as long as the relevant data is obtained, some special requirements are still required. For example, compliance with relevant laws and regulations is necessary, such as whether the laboratory is certified, whether the staff has the corresponding qualifications, and whether the equipment is calibrated on time. If the laboratory is not certified or the staff does not have the corresponding qualifications, even if the data is standardized and reasonable, it should not be included in the calculation and evaluation.) These key entry indicators are the "entry tickets," and failure to meet them results in a veto. Key Entry Indicators Including instrument calibration status Difference between testing time and accreditation period These two indicators; among them: Instrument calibration status This refers to whether the time interval between the most recent calibration and the current detection is within the allowable time. Difference between testing time and accreditation period This refers to the time difference between the testing time and the accreditation of an inspection item.
[0051] Support for entry indicators This is used to evaluate whether the result provider H(A) or the result recipient H(B) has better or more comprehensive conditions in Project X (such as participation in higher-level interlaboratory quality assessment, shorter reporting cycle, etc.). These conditions are bonus points and are used to distinguish between excellent and good. Supporting admission indicators Including the completeness rate of IQC records over the past 30 days Historical clinical acceptance rate These two indicators; among them: IQC record completeness rate over the past 30 days This refers to the completeness of the laboratory's IQC records over the past 30 days; Historical clinical acceptance rate This refers to the historical acceptance of the testing unit by the recipient of the results for this project.
[0052] Duality consistency index The duality consistency index is used to assess the comparability of the result provider H(A) and the result recipient H(B) on project X. It does not evaluate the quality of individual institutions, but specifically assesses whether the systems, methods, and standards of the result provider H(A) and the result recipient H(B) are compatible, comparable, and convertible on the specific project X. This index focuses on paired, relational attributes, such as whether the units of the test results need to be converted, whether the reference intervals are consistent, whether the project codes can be mapped, and whether the specimen types are the same. Even if both institutions are excellent ( and High scores), but if their results cannot be directly compared ( (Low score), mutual recognition also cannot be established. Duality consistency index Including the consistency of testing instruments Methodological consistency Consistency of unit of result Consistency of reference interval mapping These four indicators; among them: Consistency of testing instruments This refers to determining whether the testing instruments used by the result provider and the result recipient are from the same manufacturer and of the same model, from the same manufacturer but different models, or from different manufacturers. Methodological consistency This refers to whether the method of the actual testing instrument used by the result provider and the method used by the result recipient are consistent. Consistent results This refers to whether the actual testing instruments used by the result provider and the result receiving unit are consistent. Reference interval mapping consistency This refers to whether the actual testing instruments used by the result provider and the reference range used by the result recipient are consistent.
[0053] Regular regulatory data from institutions Including historical clinical mutual recognition rate This single indicator reflects the foundation of long-term mutual recognition between the two parties; among which: Historical clinical mutual recognition rate This refers to the historical clinical acceptance rate over the past 3 months or 6 months.
[0054] Project periodic monitoring data Including the CV value of IQC over the past 30 days IQC pass rate in the past 30 days Recent first concentration EQA results Recent second concentration EQA results Recent EQA results for the third concentration Inter-laboratory comparison These six indicators are used to reflect the results provider Mutual recognition projects The long-term performance; among which: CV value of IQC in the past 30 days This refers to the CV value calculated when the IQC quality control products from the past 30 days are from the same batch. IQC pass rate in the past 30 days This refers to the proportion of the results provided by the result provider that are in "control / qualified" within 30 days before and after the testing time of the test item; Recent first-concentration EQA results This refers to the evaluation status of the first concentration quality assessment sample in the most recent EQA batch of the project provided by the results provider before the evaluation time point; Recent second concentration EQA results This refers to the evaluation status of the second concentration quality assessment sample in the most recent EQA batch of the project provided by the results provider before the evaluation time point; Recent third concentration EQA results This refers to the evaluation status of the third concentration quality assessment sample in the most recent EQA batch of the project provided by the results provider before the evaluation time point; Inter-laboratory comparison This refers to whether the project result provider and the result recipient have conducted a result comparison and whether the comparison passed.
[0055] Real-time monitoring data of projects Including the indoor quality control status of the first concentration of the day The second concentration indoor quality control status on the same day The third concentration of the day in the indoor quality control status These three indicators are used to reflect the results provider Mutual recognition projects The immediate quality status during the inspection period; where: The first concentration of the day was under indoor quality control status. This refers to whether the tested item passed the first concentration quality control on the actual testing day; The second concentration of the day was under indoor quality control status. This refers to whether the tested item passed the second concentration quality control on the actual testing day; The third concentration of the day was under indoor quality control status. This refers to whether the tested item passed the third concentration quality control on the actual testing day.
[0056] The values of the specific indicators for the above four dimensions are calculated using the available data and the definition of the indicators, and are used for subsequent determinations of each dimension.
[0057] Step S3: Admission layer determination; First, the key access indicators Make a judgment; if key access indicators If all criteria are approved, then the support for the admission criteria will be established. Consistency index with duality Make a judgment; otherwise, key access indicators The judgment fails, and the output is not recommended for acceptance. Support for access indicators Consistency index with duality Make a judgment, if the admission criteria are supported Support admission score, dual consistency index If both the dual consistency index and the dual consistency index are greater than or equal to the corresponding qualified threshold, proceed to step S4; if both the support admission score and the dual consistency index are greater than or equal to the corresponding lower bound of the gray zone, and the support admission score or / and the dual consistency index are less than the corresponding qualified threshold, supplementary evidence is required before making a judgment; if the support admission score or / and the dual consistency index are less than the corresponding lower bound of the gray zone, the output "not recommended for acceptance" is provided.
[0058] This step determines whether the mutual recognition relationship between the two institutions can be established based on the values of relevant indicators at the access layer, which is a prerequisite for mutual recognition. If the values of the relevant indicators at the access layer are deemed to be qualified, the subsequent dimensions of the judgment will proceed. If the values of the relevant indicators at the access layer are deemed to be unqualified, the subsequent dimensions of the judgment will not proceed, and "not recommended for adoption" will be output directly. The entire access layer judgment is also implemented based on program code.
[0059] The specific determination method is as follows: Step S3-1: Determine key admission criteria ; The instrument calibration status was checked separately. Difference between testing time and accreditation period Perform a judgment; if both judgments pass, proceed to step S3-2; if the instrument is in calibration status... Difference between testing time and accreditation period If at least one of the results is deemed unacceptable, the output is not recommended for use.
[0060] It should be noted that the instrument is in calibration status. Difference between testing time and accreditation period The criteria for passing or failing can be determined and preset by a person skilled in the art based on the actual situation; if the preset criteria are met, it is considered a pass; if the preset criteria are not met, it is considered a fail. (Based on instrument calibration status) For example, the preset standard allows a time interval of 10 days between the most recent calibration and the current test. However, during this test, the system retrieved a time interval of 3 days between the most recent calibration and the current test (i.e., the instrument calibration status). (3 days), 3 days is less than 10 days, therefore the instrument calibration status is... This indicator is used to determine whether something passes; similarly, the difference between the testing time and the approval period is used. For example, the preset standard is a 5-day time difference threshold between the testing time and the accreditation time for this project, and the instrument calibration status is obtained from the system. The timeframe is 2 days, which is less than 5 days; therefore, the difference between the testing time and the acceptance period is significant. This indicator was also deemed acceptable.
[0061] Step S3-2: Calculate the support admission score and dual consistency index; Based on the completeness rate of IQC records over the past 30 days Standardized score Weight Calculate and obtain the contribution value of the integrity rate. Based on historical clinical acceptance rates Standardized score Weight Calculate and obtain the contribution value of the recognition rate. Based on the contribution value of the completeness rate , Contribution value of recognition rate Calculate and obtain the support admission score Among them, support for access points The calculation formula is: ; .
[0062] Similarly, based on the consistency of the testing instruments Matching degree Weight Calculate and obtain the contribution value of the detection instrument. According to methodological consistency Matching degree Weight Calculate and obtain the methodological contribution value. Based on the consistency of the units of the results Matching degree Weight Calculate and obtain the unit contribution value. Based on the consistency of the reference interval mapping Contribution Weight Calculate and obtain the reference interval mapping contribution value. According to the contribution value of the testing instrument Methodological contribution value Result unit contribution value Reference interval mapping contribution value Calculate and obtain the duality index Among them, the duality index The calculation formula is: ; .
[0063] The calculation of the above contribution values involves standardized scores, which are obtained after standardization based on the category of the acquired indicators and the original indicator values. Specifically: I. Indicator Classification; The indicators in the mutual recognition data domain D(U) are categorized. These categories include benefit-based indicators, cost-based indicators, target-based indicators, and binary or ordinal indicators. For example, benefit-based indicators include historical clinical acceptance rate, historical clinical mutual acceptance rate, IQC pass rate in the past 30 days, and IQC record completeness rate in the past 30 days; cost-based indicators include the CV value of IQC in the past 30 days; binary or ordinal indicators include instrument calibration status, difference between testing time and acceptance period, instrument consistency, methodology consistency, result unit consistency, reference interval mapping consistency, recent first-concentration EQA result, recent second-concentration EQA result, recent third-concentration EQA result, and interlaboratory comparison. This embodiment only involves the above three types of indicators. In addition, target-based indicators can also be set, including EQA deviation, target value deviation, and interlaboratory comparison deviation, but these are not included in this embodiment.
[0064] II. Standardization of original indicator values; For benefit-type, cost-type, and target-type indicators, the collected raw indicator values are standardized, and different standardization functions are used to calculate and obtain the standardized scores corresponding to each indicator. Binary indicators are scored according to two states, and ordinal indicators are scored according to a level mapping table.
[0065] For benefit-type, cost-type, and target-type indicators, piecewise linear standardization is used. Specifically, the original values (i.e., observed values) of the indicator are matched to an observation range; then, a standardization function is matched to the observation range; finally, a standardized score is calculated based on the standardization function. In other words, different standardization methods are used for different types of indicators. (1) Benefit-oriented indicators (the higher the value, the better); For benefit-type indicators, let... ,in , , These are the grade thresholds for C, B, and A, respectively. This represents the lower limit of the acceptable numerical range. Table 2 shows the standardized function values and observed value ranges for the benefit-type indicators.
[0066] Table 2. Comparison of Standardized Function Values and Observation Ranges for Benefit-Type Indicators
[0067] Table 2 shows that scores above the A-level threshold are directly assigned as 100 points, the B-level range is mapped to 80-89 points, the C-level range to 70-79 points, and the D-level range to 0-69 points. This represents the observed value of the i-th indicator under dimension d.
[0068] Taking the indicator "historical clinical acceptance rate (%)" as an example, the original value (i.e., the observed value) of "historical clinical acceptance rate" obtained from the system is "95%", that is... =95, its threshold is set to =60、 =80、 =90、 =97; then the range of matched observations is " ≤ < The formula for calculating the standardized score is: Therefore, using this standardized calculation formula, the standardized score of this indicator is finally calculated to be 96.4.
[0069] (2) Cost-related indicators (the smaller the value, the better); set up ,in , , These are the grade thresholds for C, B, and A, respectively. This represents the upper limit of the acceptable numerical range. Table 3 shows the comparison of the standardized function values and observation ranges for cost-related indicators.
[0070] Table 3. Comparison of Standardized Function Values and Observation Ranges for Cost Indicators
[0071] As shown in Table 3, cost-based indicators and benefit-based indicators are perfectly paired and are suitable for indicators such as non-compliance rate, instability rate, alarm rate, and number of times exceeding limits, which are "the smaller the better".
[0072] Taking the indicator "CV value (%) of IQC in the past 30 days" as an example, the raw value (i.e., the observed value) of "CV value of IQC in the past 30 days" obtained from the system is "1.5%", that is... =1.5, its threshold is set to =3.0、 =3.6、 =4.5, =5.0; then the range of matched observations is " ≤ The formula for calculating the standardized score is: Therefore, the standardized score for this indicator is 100.
[0073] (3) Target-type indicators (the closer to the target value, the better); Let the target value be Deviation amount is ; set again , here This represents the upper limit of acceptable deviation. Table 4 shows the standardized function values and observation ranges for the target indicators.
[0074] Table 4. Comparison of Standardized Function Values and Observation Ranges for Target Indicators
[0075] As shown in Table 4, target-type indicators are suitable for expressing states of "deviation from the target, deviation from the center, or deviation from the target value," and are particularly suitable for consistency, drift, and trend deviation indicators. This embodiment does not include target-type indicators; their standardized score calculation method can refer to the calculation methods for the aforementioned benefit-type and cost-type indicators.
[0076] (4) Binary or ordinal indicators; Binary indicators use a two-level scoring system of 0 / 100, or a four-level scoring system of 0 / 70 / 85 / 100; ordinal indicators use a pre-established level mapping table.
[0077] Taking the "interlaboratory comparison" indicator as an example, the original value of the "interlaboratory comparison" indicator obtained from the system is "pass", and a two-level scoring method is used (pass, 100; fail, 0). Therefore, the standardized score of this indicator is 100.
[0078] The calculation of the aforementioned contribution values also involves the matching degree. In this embodiment, the matching degree is calculated using a "normalized distance / difference measurement model based on a preset threshold". The core logic of this model is to calculate the "difference value" between the organization to be evaluated and the standard benchmark, and then map this difference value to a range of [0,1]. The smaller the difference, the closer the matching degree is to 1; the larger the difference, the closer the matching degree is to 0. The specific calculation method for the matching degree of these four consistency indicators is as follows: I. Unified computing framework; This embodiment defines a general matching degree function. , (Difference value (U, baseline)); Here, f is a normalization function that converts the difference value into a 0-1 fraction (usually a piecewise function or a simple subtraction, which can be directly applied to existing algorithms, and this algorithm is not an innovation).
[0079] II. Specific adaptation calculation methods for the matching degree of the four consistency indicators; 1. Consistency of testing instruments ( — Matching degree ; (1) Optimal approach: Total Medical Allowable Error (TEa); (2) Calculation logic: In clinical testing, there may be a certain small deviation between the results of different instruments. Calculate the relative deviation between the results of the instrument to be evaluated and the standard instrument. (3) Matching degree The calculation formula is: ; in, This indicates the test results obtained by the instrument being evaluated on the same sample. This indicates the standard reference result obtained by a standard reference instrument (reference instrument, calibration reference device) on an identical sample. This represents the total allowable error. (Note: If the deviation is within the allowable range, the score is 1; if it exceeds the allowable range, the score decreases linearly.) 2. Methodological consistency — Matching degree ; (1) Optimal solution: The method combining constant systematic error and proportional systematic error; (2) Calculation logic: Different detection methods (such as electrochemiluminescence vs chemiluminescence) may have intercept bias or slope bias. Calculate the average relative bias between the two. (3) Matching degree The calculation formula is: ; in, Indicates bias. This represents the absolute value of the systematic deviation between two testing methods (to be evaluated / standard). The coefficient of variation represents the bias value itself, characterizing the degree of random fluctuation in the bias. This represents the adjustment parameter. (Note: It combines bias and coefficient of variation, and limits the range to 0-1.) 3. Consistency of unit of measurement in results ( — Matching degree ; (1) Optimal solution: Conversion factor comparison method; (2) Calculation logic: This dimension deals with the consistency of units such as "mmol / L" and "mg / dL". In essence, it examines whether the institution being evaluated has used the correct unit conversion factor.
[0080] (3) Matching degree The calculation formula is: ; (Note: If the units are consistent, directly assign the value 1).
[0081] 4. Reference interval mapping consistency ( — Matching degree ; (1) Optimal solution: Interval overlap method; (2) Calculation logic: Evaluate the degree of overlap between two intervals, which is the most rigorous interval consistency index in statistics; (3) Formula: ; (Note: The value range is naturally between [0,1], which perfectly matches the definition of matching degree) III. Supplementary Explanation; (1) In this scheme, the above matching degree to The specific calculation model can be adaptively adjusted according to the actual application scenario and data availability. As a preferred embodiment of this application, when sufficient sample data is available for quantitative comparison, the above-mentioned refined calculation model based on the medically permissible total error or interval overlap can be used. As another optional embodiment of this application, if limited by data acquisition conditions, those skilled in the art can also use a rule-based logical judgment method to simplify the calculation: for example, regarding instrument consistency, if the difference between the detection results of the two instruments is within the clinically acceptable range, it is assigned a value of 1; otherwise, it is assigned a value of 0. Regarding unit consistency, if the units are consistent, it is assigned a value of 1; if they are inconsistent but can be converted and the conversion logic is correct, it is assigned a value of 1; otherwise, it is assigned a value of 0.
[0082] (2) In this scheme, the above matching degree The calculations are based on standard evaluation methods in the field of clinical laboratory testing. Specifically: for instrument consistency, the permissible total error (TEa) method is used, with TEa referencing the permissible total error values for the corresponding items specified in WS / T403—2024, CLIA'88, or ISO15189; for methodological consistency, Bland-Altman analysis is used to calculate bias and CV, with the adjustment coefficient k determined according to error allocation principles, preferably k=2; for unit consistency, the relative deviation between the conversion coefficient entered on the platform and the standard conversion coefficient (based on the Ministry of Health Clinical Laboratory Center or the authoritative standard for the corresponding item) is compared; for reference interval consistency, the interval overlap method is used. The weights of the above indicators are calculated using a multi-stage combination of "subjective safety weight + objective information weight + historical experience recalibration (sufficient sample)"; specifically, when the initial sample is insufficient, a two-stage combination of "subjective safety weight + objective information weight" is used, first calculating the subjective safety weight separately. Weight of objective information Then, calculate the combined weight based on both subjective and objective weights. Combined weights This is the weight of indicator i; after sufficient samples are available in the later stages, a three-stage combination of "subjective safety weight + objective information weight + historical experience recalibration" is adopted. First, the subjective safety weight is calculated separately. Weight of objective information Then, calculate the combined weight based on both subjective and objective weights. Then construct the empirical importance vector. Combined weights By integrating and recalibrating based on historical experience, Recalibrating based on historical experience This refers to the weight of indicator i. Specifically: (1) Calculate the subjective safety weight ; The Best-Worst Method (BWM) is employed. The core idea of BWM is to first select the most important indicator B and the weakest indicator W within the same dimension, and then obtain a set of weights that meet consistency requirements through comparison vectors of "best versus others" and "others versus worst". Its basic form can be written as: , , , ; in, This indicates that the consistency index is achieved under the premise that the following conditions are met. Minimum, This represents the weight to be determined for the i-th indicator (i.e., the subjective safety weight). This indicates the weight of the optimal indicator B. This represents the weight of the worst-case indicator W. This represents the importance scale value of the best indicator B relative to indicator i, as given by the decision-maker (this scale value is the judgment given in advance by the expert, i.e., how many times more important the best indicator B is than indicator i). This represents the importance scale value of indicator i relative to the worst indicator W given by the decision-maker (this scale value is the judgment given in advance by the experts, that is, how many times more important indicator i is than the worst indicator W). The consistency index (obtained after solving) The smaller the value, the higher the consistency of the decision-makers' judgments, and the more reliable the weighting results.
[0083] (2) Calculate the weight of objective information ; For the i-th indicator within the same dimension, the CRITIC method is used to calculate the objective information weight. The weight of this objective information This can be expressed as: ; ; in, This represents the dispersion within the index, which is the standard deviation of the i-th index across all evaluated objects under dimension d. This represents the correlation coefficient between indicators, specifically the Pearson correlation coefficient between indicator i and indicator k. This indicates the number of indicators k. This represents the amount of information contained in index i.
[0084] Assuming there are n evaluated entities (e.g., hospitals) and m evaluation indicators under an evaluation dimension d, the goal of this step is to calculate the objective weights for these m indicators. The specific steps are as follows: Step 1: Calculate the dispersion within the index ; It is the in-range dispersion of the indicator, that is, the standard deviation of the i-th indicator in dimension d among all the evaluated objects. Its calculation formula is: ; in, This represents the original observation value of the j-th object in dimension d on the i-th index. Indicates that index i under dimension d is among all The arithmetic mean of a sample, This represents the total number of samples in dimension d.
[0085] The larger the value, the more significant the differences in performance among hospitals on this indicator, the higher the comparative strength of the indicator, and the richer the objective information it contains.
[0086] Step 2: Calculate the correlation coefficients between the indicators. ; Let represent the Pearson correlation coefficient between indicator i and indicator k, with a value range of [−1, 1]. Its calculation formula is: ; in, This represents the total number of samples in dimension d. This represents the original observation value of the j-th object in dimension d on the i-th index. Indicates that index i under dimension d is among all The arithmetic mean of a sample, This represents the original observation value of the j-th object in dimension d on the k-th index. Indicates that index k under dimension d is in all The arithmetic mean of the samples.
[0087] The larger the value (closer to 1), the more similar the information provided by the two indicators and the higher the redundancy. If an indicator is highly correlated with all other indicators, then it can provide very little independent information.
[0088] Step 3: Calculate the information content of index i ; Information content of indicators The calculation formula is: ; in, Indicates the dispersion within the index. This represents the correlation coefficient between the indicators. This indicates the number of indicators k.
[0089] In the formula for calculating information content, the first part... This represents the information content of the indicator itself, the second part. This represents the independent information content of the indicator. When A value close to 1 indicates high redundancy. Smaller; when When the value is close to 0 or negative, it indicates that the information provided by indicator i and indicator k differs more significantly. The difference is relatively large; multiplying the two means that the total information content of an indicator is equal to its internal contrast strength multiplied by its external independence. Only indicators that simultaneously possess "large difference" and "uniqueness" will achieve a high score. value.
[0090] Step 4: Calculate the objective information weights of the indicators. ; Objective information weight The calculation formula is: ; in, This represents the amount of information contained in index i. This indicates the number of indicators k.
[0091] Information content of each indicator Normalization is performed. That is, the information content of a single indicator is divided by the sum of the information contents of all indicators, resulting in... It is a value between 0 and 1, and the sum of the objective weights of all indicators is 1. The more information a factor contains, the higher its objective weight.
[0092] (3) Calculate the combined weights ; To reduce the absolute dominance of a single path on the outcome, multiplicative normalization fusion is used to calculate and obtain the combined weight. Combined weights The calculation formula is: ; in, Indicates subjective safety weight. Indicates the weight of objective information; This indicates the degree of safety orientation. For high-safety scenarios like medical mutual recognition, a value slightly higher than 0.5 is advisable, allowing experts to maintain a dominant role in judging key risks without completely suppressing information provided by objective data. The combined weights ensure the assessment system is both robust (less prone to bias due to the limitations of a single method) and balanced (achieving the optimal trade-off between safety and objectivity, experience and data), ultimately providing a solid and reliable quantitative foundation for assessing the credibility of mutual recognition of results among medical institutions.
[0093] When the initial sample size is insufficient, the combined weights of index i under dimension d are calculated here. This refers to the weights used to calculate the original scores for the corresponding dimension d (institutional periodic supervision G, project periodic supervision P, and project real-time supervision R). For later stages with sufficient samples, in addition to calculating the combined weights using the aforementioned method... In addition, historical experience recalibration is also required. Specifically: For a sufficient number of historical labeled samples that have been accumulated, such as results from manual review, expert decisions, or posterior adverse events, an empirical importance vector can be constructed. and combined weights By integrating and recalibrating based on historical experience, Recalibrating based on historical experience This refers to the weight of indicator i under dimension d, which can be used to calculate the original score. Historical experience recalibration is also included. The calculation formula is: ; in, The weights represent the combined weights; The empirical importance vector can be obtained by normalizing the absolute value of logistic regression coefficients, tree model importance, or parameters of ordered classification models.
[0094] The aforementioned "sufficient accumulated historical label samples" is determined by a rigorous dual mechanism of "statistical significance" and "model stability." Specifically, it can be defined from the following two dimensions: 1. Statistical baseline: the ratio of sample size to feature size; In calculating the empirical importance vector When performing such tests, logistic regression or tree models are typically used, but these models have basic requirements regarding the amount of data.
[0095] Mathematical standard: In order to ensure that the variance of the weight calculation is controllable and to avoid "random weights", the total sample size N must be significantly greater than the total number of features (indicators) P by 10 to 20 times, that is, N≥(10P~20P).
[0096] For example, if there are 15 specific indicators (P=15) under the dimensions to be evaluated, then at least 150 to 300 complete data records need to be accumulated in the historical sample library to barely qualify for the basic qualification to calculate weights.
[0097] 2. Business coverage: Prior probability and probability occurrence rate; In medical mutual recognition data, cases of "direct acceptance" (normal / good) are usually far more numerous than cases of "expert recommendation for rectification" (abnormal / poor).
[0098] Mathematical standard: In order to ensure that the algorithm can learn the "error correction experience" of experts, it is necessary to ensure that the number of samples in the minority class (abnormal / adverse events) can also meet the above requirement of N≥10P.
[0099] Judgment logic: If there are only a few hundred "normal" records in the historical data, but only a few "rejected by experts", the model will tend to predict all new samples as "normal". This is called "sample distribution skew". Only when the positive and negative samples (or samples at each level) reach a certain threshold (e.g., at least 50-100 cases per class) are the samples considered "sufficient" in terms of business.
[0100] Therefore, the aforementioned "sufficient sample accumulation" is not a fixed Arabic numeral, but a vector inequality (which can be defined by those skilled in the art based on the above two dimensions without creative effort); only when the above-mentioned dual standards of "statistical bottom line" and "business coverage" are triggered will "historical experience recalibration" be initiated.
[0101] Furthermore, regardless of the weighting method used, a minimum weight constraint should be set for key indicators to prevent them from being "squeezed" too low by objective methods. ; in, This represents the minimum historical experience recalibration of index i under dimension d.
[0102] Step S3-3: Support admission and duality consistency determination; If admission points are supported Dual consistency index If all values are greater than or equal to the corresponding qualifying threshold, then the application is deemed approved and proceeds to step S4; if the application is supported... With duality index All are greater than or equal to the corresponding lower bound of the gray area, and support admission scoring. or / and duality index If the score is below the corresponding qualifying threshold, supplementary evidence will be required before a final determination is made; if the admission score is supported... or / and duality index If the value is less than the corresponding lower bound of the gray area, the output is not recommended for use.
[0103] For key access indicators Total score compensation will not be used; for support access indicators This can form auxiliary admission scores; for indicators specifically reflecting the comparability of ABX, a dual consistency index is formed. .
[0104] Key entry indicators This is suitable for describing conditions where their failure would undermine the basis of mutual recognition, such as semantic equivalence of projects, consistency of specimen types, convertibility of units, comparability of methodological or traceability bases, and the establishment of regulatory prerequisites. For these indicators, an average total score is not appropriate; their status should be uniformly represented using... To express, ; in, This indicates that the key access indicator i has been passed. This indicates a lack of key evidence. This indicates that the application was rejected. Specifically: If the time interval between the most recent calibration and the current test is within the allowable time, then the instrument's calibration status is... of ,otherwise ; If the time difference between the testing time and the accreditation period is within the allowable time, then the difference between the testing time and the accreditation period is considered acceptable. of ,otherwise ; If the instrument is in calibration status Difference between testing time and accreditation period of If both values are 1, the judgment is passed, and the process proceeds to step S3-2; otherwise, the judgment is failed.
[0105] Support for entry indicators It does not directly determine whether the mutual recognition basis exists, but it does determine whether the mutual recognition basis is sufficiently stable, clear, and traceable. Typical aspects may include reference range mapping, document consistency, historical consistency support, and the completeness of quality records. Let... The auxiliary admission threshold for formal admission. To assist in defining the lower boundary of the gray area for entry; let... The duality consistency threshold for formal admission. As the lower bound of the duality-consistency gray region, the auxiliary admission score can be expressed as: , .
[0106] The essence of mutual recognition is not unilateral compliance of H(A), but rather the formation of a mutually understandable, mutually convertible, and mutually interpretable relationship between H(A) and H(B) on item X. Therefore, the duality consistency index... It is extracted separately from the admission layer and not simply mixed with the supporting admission components.
[0107] Dual Consistency Index Indicators that directly describe the correspondence between the two should be prioritized, such as consistency in project code mapping, effectiveness of reporting unit conversion, consistency in specimen type, sufficiency of reference interval mapping, and sufficiency of conversion model validation. The dual consistency index is defined as follows: ; in, This describes the degree of matching between the two parties in areas such as project coding mapping, reporting unit conversion, specimen type consistency, reference range mapping, and conversion model validation. When Even if one party's structure appears to be of good quality, it should not be readily assumed that the two parties have formed a basis for mutual recognition when the threshold is too low.
[0108] Combining key access criteria, supporting access criteria, and duality consistency criteria, the criteria for determining whether access is granted are as follows: This means that the application has been approved. ,and make ,or ,or This means that access is suspended; In other cases, admission is denied.
[0109] The comparison of the three access statuses is shown in Table 5: Table 5 Comparison of the Three Types of Access Status
[0110] In the reference settings, the corresponding qualified threshold and lower boundary of the gray area can be set by those skilled in the art according to actual needs, without requiring creative effort. Preferably, it can be made... =80、 =70、 =0.85、 =0.75 was used as the initial calibration value, and then optimized based on trial operation and historical data.
[0111] This definition clearly expresses three layers of logic: first, key admission indicators cannot be compensated with high scores; second, the comparability between the two parties needs to be characterized separately; and third, a transitional state of "suspension and re-evaluation after supplementary evidence" is reserved for cases with insufficient evidence or marginal status.
[0112] Step S4: Dynamic regulatory layer determination; The observation range is matched based on the raw values of indicator i in dimension d. A standardization function is then matched based on the observation range, and the standardized score of indicator i in dimension d is obtained using the standardization function. Finally, the periodic regulatory data of the institution are calculated based on the standardized score of indicator i in dimension d and its corresponding weight. The agency regularly monitors the raw scores Regular monitoring data of projects Project regular monitoring of raw scores Real-time monitoring data of projects Real-time monitoring of project raw scores .
[0113] The computing agency regularly monitors the raw scores. At that time, based on historical clinical mutual recognition rates Standardized score Weight (Since there is only one indicator in the institutional periodic supervision dimension in this embodiment, the weight of this indicator is...) The value is 1; if other indicators are added to this dimension, the weights of the corresponding indicators are calculated and determined based on the subjective safety weight and the objective information weight, and the institutional periodic regulatory contribution value is calculated and obtained. Regular regulatory contribution value of institutions This refers to the raw scores from the institution's regular supervision. .
[0114] The original score is monitored regularly during the calculation of the project. At that time, based on the CV value of IQC over the past 30 days Standardized score With weight Calculate and obtain the CV contribution value. Based on the IQC quality control pass rate over the past 30 days Standardized score With weight Calculate and obtain the contribution value of the quality control pass rate. According to the recent first concentration EQA results Standardized score With weight The EQA contribution value of a concentration was calculated and obtained. According to the recent second concentration EQA results Standardized score With weight The contribution value of the two-concentration EQA was calculated and obtained. According to recent third-concentration EQA results Standardized score With weight The contribution values of the three concentrations of EQA were calculated and obtained. Based on inter-laboratory comparison Standardized score With weight Calculate and obtain the interlaboratory comparison contribution value. Finally, based on the contribution value of the CV value... Contribution value of quality control pass rate One-concentration EQA contribution value EQA contribution value of 2 concentration Three-concentration EQA contribution value Interlaboratory comparison contribution value The original score for periodic project supervision was calculated. Among them, the original score for regular project monitoring. The calculation formula is: ; .
[0115] Calculate the raw score for real-time monitoring of the project. At that time, based on the indoor quality control status of the first concentration of the day Standardized score With weight The IQC contribution value was calculated and obtained. According to the indoor quality control status of the second concentration on that day Standardized score With weight The contribution value of the two-concentration IQC was calculated and obtained. According to the indoor quality control status of the third concentration on that day Standardized score With weight The contribution values of the three concentrations of IQC were calculated and obtained. According to the IQC contribution value Second concentration IQC contribution value Three-concentration IQC contribution value The raw score for real-time monitoring of the project was calculated. Among them, the raw score for real-time project monitoring. The calculation formula is: ; .
[0116] Regarding the regular regulatory data of the aforementioned institutions Regular monitoring data of projects Real-time monitoring data of the project The standardized score corresponding to indicator i in the three dimensions can be calculated by referring to the standardized score calculation method in step S3-2 above.
[0117] Regarding the regular regulatory data of the aforementioned institutions Regular monitoring data of projects Real-time monitoring data of the project The weights corresponding to index i in the three dimensions ( , , It also adopts a multi-stage combination of "subjective safety weight + objective information weight + historical experience recalibration (sufficient samples)"; for details, please refer to the weight calculation method in step S3-2 above.
[0118] Step S5: Hard threshold determination; The indicators in dimension d are divided into indicators, and hard threshold conditions are determined. Based on the determination results, the periodic regulatory data for each institution is determined. Regular monitoring data of projects Real-time monitoring data of the project Corresponding hard threshold level .
[0119] For institutional periodic regulatory data Regular monitoring data of projects Real-time monitoring data of the project For any one of the three dimensions d, classify the indicator types of dimension d, determine the hard threshold level of dimension d based on the hard threshold judgment criteria, and obtain the hard threshold level of dimension d. The specific method is as follows: Step S5-1: Classification of indicator types; Divide the indicators in dimension d into a set of key indicators. Non-key indicator set .
[0120] When classifying indicators, classification can be based on pre-set criteria or on the analysis and judgment of clinical experts, who can then classify them according to the actual situation. In this embodiment, clinical experts classify the aforementioned specific indicators into indicator types based on the actual situation.
[0121] After classifying the indicators by type, the key indicator set Non-key indicator set Conditions met: ; ; in, This indicates the number of specific indicators in dimension d. This represents the empty set.
[0122] Step S5-2: Determine the hard threshold condition; For any dimension d, the upper limit of the hard threshold level for dimension d is defined as the highest level that satisfies the hard threshold determination condition, thus obtaining the hard threshold level of dimension d. It can be represented as: ; If none of the three levels (A, B, C) are satisfied, then it is defined as That is, it is unqualified.
[0123] The hard threshold determination criteria are as follows: Condition 1: All key indicators meet the requirements; That is, all key indicators in dimension d must meet level q or higher; if any key indicator fails to meet the requirement, the condition is not met. Therefore, all key indicators meeting the requirement can be expressed as: ; in, Indicates the first One indicator, Represents the set of key indicators. Represents the th element in dimension d The indicator level of each key indicator is q; Condition 2: The gap in non-critical indicators is controllable; The sum of the weight gaps of all non-critical indicators that fail to meet the standards does not exceed the tolerance limit of level q. , represented as: ; in, Indicates the first One indicator, This represents the set of non-critical indicators; Represents the th element in dimension d If the indicator level of a non-critical indicator is q, then the indicator... If the threshold requirement for level q is met, then =1, if the indicator If there is a prohibited state that conflicts with level q (such as a serious violation), then =0; Represents the th element in dimension d The weights of non-critical indicators, and satisfying the following conditions: ; This represents the upper limit of tolerance for weight gaps in non-critical indicators at level q, and satisfies the following conditions: , , , These respectively indicate that the sum of the weight gaps of all non-critical indicators that have not met the standards does not exceed the tolerance limit for levels A, B, and C; This indicates that the non-critical indicators were not met. The resulting weight gap; Step S5-3: Determine the hard threshold level; If the hard threshold criteria are not met, then the hard threshold level is... The value is D; if the hard threshold determination condition is met, then the hard threshold level is... The determination method is as follows: Based on the data of key indicators collected in dimension d, determine the hard threshold level of the key indicators; The sum of the weight gaps of the corresponding non-critical indicators in the non-critical indicators of the calculation dimension d is compared with the tolerance upper limit of levels A, B, and C to obtain the hard threshold level of the non-critical indicators. Based on the hard threshold levels of the key indicators and non-key indicators of dimension d, select the lower level as the hard threshold level for dimension d. For example, if the hard threshold level of the key metric for dimension d is A, and the hard threshold level of the non-key metric is B, then the hard threshold level of dimension d is... That is, B; the subsequent "choose a lower level" is explained in the same way as here.
[0124] When determining the hard threshold level of key indicators, the observations in Table 2-4 of step S2 are used. The scope of the measurement needs to be determined. For example, in Table 2, if the observed value of a key indicator in dimension d... Greater than or equal to If the hard threshold level of this key indicator is A, then the hard threshold level is A; similarly, if the observed value... Greater than or equal to and less than If the observed value is B, then the hard threshold level of this key indicator is B; Greater than or equal to and less than If the hard threshold level of a key indicator is A, then the hard threshold level of that key indicator is C; the rest are D. Based on the hard threshold levels of all key indicators in dimension d, the lower level is selected as the hard threshold level for the key indicator in dimension d. For example, if there are 5 key indicators in dimension d, and 3 of them have a hard threshold level of A and 2 have a hard threshold level of B, then the hard threshold level of the key indicator in that dimension will be limited to level B by the lower-level key indicator, meaning all key indicators in condition 1 are met.
[0125] When determining the hard threshold level for non-critical indicators, if the sum of the weight gaps of the non-critical indicators that fail to meet the standard is less than or equal to the tolerance limit for level A... If the hard threshold level for non-critical indicators is A, then the hard threshold level for non-critical indicators is A. And less than or equal to the tolerance limit of level B. If the hard threshold level for non-critical indicators is B, then the hard threshold level for non-critical indicators is B. And less than or equal to the tolerance limit of level C. If the hard threshold level for non-critical indicators is C, then the hard threshold level for non-critical indicators is C. If so, the hard threshold level for non-critical indicators is D.
[0126] It should be noted that the sum of the gap thresholds for non-critical indicators in dimension d refers to the sum of the weights of non-critical indicators in that dimension that have not reached level q. Therefore, the upper limit of tolerance can be directly calculated from the sum of the gaps in the weights of non-critical indicators. , , By comparison, determine the upper limit of the threshold level for non-critical indicators in this dimension due to limitations imposed by non-critical indicators. For example, in dimension d, the upper limit for critical indicators is level B, while the sum of the weights of non-critical indicators below level B is 10%, and the tolerance upper limit for level B is... The tolerance limit is 9%, which is the upper limit for Category C. If the weight gap for non-critical indicators is 15%, then the non-critical indicator weight gap is greater than the tolerance limit for level B but less than the tolerance limit for level C. Therefore, the hard threshold level for non-critical indicators is level C. Since the hard threshold level for critical indicators is level B and the hard threshold level for non-critical indicators is level C, the final hard threshold level for dimension d is chosen as the lower of the two, which is level C.
[0127] Step S6: Final output of the linkage decision; Based on the raw scores of each dimension d , , Determine the corresponding original score level ; Based on the raw score levels of dimension d Hard threshold level Choose the lower level as the final level for this dimension d. According to the final level of each dimension The final output is obtained; the final output includes priority acceptance, normal acceptance, cautious acceptance, and not recommended acceptance.
[0128] For any dimension d, the original score ( , , The original score levels are divided according to the original score level classification criteria to obtain the original score levels for the corresponding dimension d. The original score grading criteria are as follows: 90 ≤ original score ≤ 100, grade A; 80 ≤ original score < 90, grade B; 70 ≤ original score < 80, grade C; 0 ≤ original score < 70, grade D.
[0129] For any dimension d, the original score level Hard threshold level Choose the lower level as the final level for this dimension d. Therefore, the final level of the institution's periodic supervision can be obtained. The final level of regular project supervision Ultimate level of real-time project monitoring ; based on the final rank of all dimensions d This yields the final output. The criteria for the final output are: If the final level of the agency's regular supervision The final level of regular project supervision Ultimate level of real-time project monitoring If all are grade A, the final output will be the one that is accepted first; if the final grade is... , , If none of the values are C or D, and at least one is B, then the final output is considered normal; if the final level... , , If a grade C exists and no grade D exists, the final output is "cautious acceptance"; if the final grade... , , If a D-level is present, the final output will be "not recommended for adoption".
[0130] Example: As shown in Table 6, the original score levels corresponding to the three dimensions G, P, and R. These are the hard threshold levels corresponding to the three dimensions: B, A, and A. The scores are B, B, and A, respectively. Therefore, for dimension G, its original score levels are... Hard threshold level The lowest level is B, therefore the final level of dimension G is... That is, B; for dimension P, its original score level Level A, Hard Threshold For B, therefore the final rank of dimension R is... That is, B; for dimension R, its original score level Hard threshold level All are A, therefore the final rank of dimension R is... That is, A.
[0131] Table 6. GPR Level Comparison Table (Three Dimensions)
[0132] Based on the aforementioned original score levels of dimension d Hard threshold level This yields the final output (preferred, normal, cautious, and not recommended for acceptance). Regarding the reliability of this final output, this embodiment uses a comprehensive approach to separately differentiate the reliability levels within the same acceptance result, effectively providing supplementary explanations of the reliability of the acceptance result.
[0133] For example, a test result might have an initial score of AAA across its three dimensions, resulting in a composite score of 95. However, due to a hard threshold, the final score for each dimension might be ABA, leading to an output of "normal acceptance." Theoretically, a score of 95 for normal acceptance is more reliable (higher credibility) than a score of 88. Therefore, the composite score can be used to assess the credibility of a given result.
[0134] Among them, the overall score The calculation method is as follows: ; in, This indicates that the institution regularly monitors the raw scores. This indicates the original score for regular monitoring of the project. This indicates the raw score for real-time monitoring of the project. , , These represent the weights of the three dimensions: periodic institutional supervision, periodic project supervision, and real-time project supervision. This comprehensive score is only intended for identifying differences within the same level, monitoring projects nearing the de-escalation threshold, and as a reference for regulatory resource allocation; it cannot replace hard thresholds, evidence caps, or circuit breaker controls.
[0135] It should be noted that this overall score It is only used for ranking within the same grade, edge monitoring, and allocation of regulatory resources, and does not replace dimension levels, hard thresholds, evidence caps, or circuit breaker decisions; the weights within a dimension are used for indicator synthesis, while the weights between dimensions are only used for comprehensive scoring, and the two have different uses.
[0136] The weights of the above three dimensions , , Scientific determination is also necessary, but since there are only three dimensions, and each dimension has a distinct business semantic, it is not advisable to simply copy the weighting methods within the dimensions (such as the weight calculation method for index i under dimension d described in step S4 above). In this embodiment, the weights of each dimension are determined using a method of "expert prior guidance + gentle calibration with historical data". Specifically: First, experts provide prior values across the three dimensions using the best-worst method (BWM). (For details on the best-worst method (BWM), please refer to the section on the calculation of subjective safety weights mentioned above.) If there are insufficient historical label samples, expert prior values can be used directly. As the weight of the corresponding dimension ( , , ); If the historical label sample is sufficient, then estimate the empirical contribution of the three dimensions to the final ruling. And perform linear calibration; the specific formula is: ; in, This represents the proportion of subjective weights, used to adjust the contribution ratio of expert subjective weighting results and historical label weighting results in the final dimension weights. The value range is [0,1], and it is advisable to take a higher value to ensure that the meaning of the dimension itself is not distorted by pure empirical fluctuations.
[0137] For example, suppose there are expert prior values for a certain dimension G. The estimated empirical contribution is 0.50. It is 0.30; if we set If the weight is 0.6, then the weight of dimension G is... The weight is 0.6*0.5+(1-0.6)*0.3=0.42. This means that the weight of dimension G is 0.42, of which subjective expert opinions account for 60% and objective data accounts for 40%.
[0138] Example 2 In step S4, when calculating the original scores for each dimension, it is considered that the evaluation conclusion depends not only on the object's state but also on the quality of evidence; some dimensions may use complete direct evidence, while some indicators in some dimensions may have imputed evidence due to insufficient sample size. Therefore, based on Example 1, evidence correction is needed for dimension d where imputed evidence has been enabled, and the upper limit of the evidence level is determined according to the imputed evidence corresponding to dimension d. In the final output of the linkage decision in step S6, an upper limit for the level of evidence is introduced. Specifically: In high-security scenarios, simply reducing the original score to indicate insufficient evidence is often inadequate when revising evidence. This embodiment introduces an upper limit for the level of evidence. If the evidence is complete and direct, then the upper limit of the level of evidence is... This does not constitute an additional limitation; however, if the evidence comes from an alternative source, an extended window, or has a critical gap, its highest possible rating should be limited according to the degree of uncertainty. The upper limit of the evidence rating should be determined based on the completeness and directness of the evidence. They are divided into five levels, with the specific criteria as follows: If the evidence comes directly from official original records or first-hand measured data, then the upper limit of the evidence level is as follows. The highest level, Level A, is set, which does not constitute an additional restriction; if the evidence comes from indirect sources, delayed reporting data, or uses alternative testing methods, then the upper limit of the evidence level is [not specified]. The next highest level is B; if the evidence is of poor quality and not completely broken, the upper limit of the evidence level can be raised depending on the circumstances. Set to Level C or lower; if key elements of the evidence are missing or the logical chain is broken, the maximum evidence level is set. It is classified as Level D; if the evidence is insufficient, it does not belong to any of the levels A, B, C, or D, but is an independent "pending review" status, directly outputting "not recommended for acceptance" or "pending review", and does not participate in the rating assessment.
[0139] In determining the upper limit of the level of evidence Afterwards, evidence correction is required for any dimension d that has enabled supplementary evidence. The corrected original score for dimension d is then calculated. for: ; To ensure a unified expression for evidence revision, the quality of evidence in dimension d can be summarized as an evidence quality score. Evidence quality score It can be comprehensively formed by sub-items such as completeness, directness, timeliness, sufficiency, and consistency of evidence; correspondingly, the evidence correction coefficient. Can be considered as evidence quality score The monotonic mapping of evidence, i.e., the higher the quality of the evidence, the better. The closer the value is to 1, the lower the quality of the evidence. The smaller. , It can be represented as: ; ; in, This represents the raw score of dimension d. If dimension d represents periodic supervision by the institution, then... for If dimension d represents periodic project monitoring, then for If dimension d represents real-time project monitoring, then for ; The evidence correction coefficient for dimension d is determined by the quality of evidence in dimension d. This indicates the evaluation result of the h-th evidence sub-item (which can be determined by a person skilled in the art based on the actual situation). This indicates the relative importance of the h-th piece of evidence. This represents the mapping function from the evidence quality score to the correction coefficient.
[0140] Because evidence correction was performed, the final output of the linked decision in step S6, for dimension d, is based on the corrected original score. The level of evidence, the upper limit of the level of evidence Choose the lower grade as the score-corrected grade for this dimension d. The grade is then adjusted based on the score of dimension d. Hard threshold level Choose a lower level as the final level And according to the final level The final output is obtained. This includes the score-corrected rank for dimension d. , and the original score level of dimension d The method of determination is the same; the grade is adjusted according to the score. With hard threshold level The method for determining the final grade differs from the method for determining the original score grade based on dimension d. With hard threshold level The method for determining the final grade is the same. See Example 1 for details.
[0141] This can also be understood as: the grade after score correction for any dimension d. Hard threshold level Upper limit of the level of evidence The final level of dimension d can be uniformly represented as: ; Of the three levels mentioned above, the lowest level is taken as the final level, which also reflects the theoretical structure of this scheme: "total score follows the threshold, and the threshold follows the security of evidence".
[0142] This "dual correction" mechanism—that is, the coexistence of score reduction and grade capping—can more realistically express the impact of evidence quality on the evaluation conclusion. The former reflects "decreased credibility," while the latter reflects "it is not appropriate to directly declare a high grade when the evidence is insufficient."
[0143] Regarding the evidence mentioned above, this embodiment further classifies the evidence according to its status. The specific impact of each evidence status on subsequent judgments can be found in Table 7. Table 7 Comparison of Evidence Status and Corresponding Impact on Subsequent Judgments
[0144] "Insufficient evidence" is not the same as "qualified" or "Grade D". It means that there is currently insufficient evidence to prove that the mutual recognition unit or dimension has reached a stable and credible state. Therefore, insufficient evidence is an uncertain state with clear risk implications.
[0145] In front-end business presentations, insufficient evidence can be linked to "cautious acceptance" or "pending review"; however, at the underlying methodology level, it should retain its independent status to avoid misinterpreting "not yet able to prove security" as "security has been proven".
[0146] Example 3 During the data collection process from the information system in step S2, if any data is missing, corresponding missing value processing is required. Specifically, based on Example 1 or Example 2, the process is as follows: After obtaining the mutually recognized data domain D(U), if missing data exists, missing value handling is required. Missing value handling is not considered a simple statistical imputation problem, but rather an evidence management problem. In other words, when faced with missing values, the core issue is not "how to fill in the missing values," but rather "whether the current evidence is sufficient to support a stable judgment, and how this deficiency should affect the conclusion." If the mutual recognition reliability assessment is simply understood as a data calculation task, it is easy to mask the true uncertainty using mean substitution, historical imputation, or conventional interpolation methods; however, in high-security scenarios, the masked uncertainty itself may pose a risk. Therefore, different handling methods are adopted for different types of missing values, as detailed in Table 8: Table 8 Comparison of Treatment Principles for Different Types of Missing Data
[0147] As shown in Table 8, the corresponding processing rules for different missing types are as follows: For structural deficiencies, it should first be determined whether they are objectively non-existent in business operations; if they are objectively inapplicable, they can be marked as "inapplicable"; if a complete judgment is required, alternative evidence with equivalent explanatory significance should be introduced; if the alternative evidence itself is insufficient, it should not be considered "resolved" but should enter the path of insufficient evidence.
[0148] For insufficient samples, the principle of "expand the window first, then judge" should be followed. This involves increasing the sample size by extending the time window. The purpose of expanding the window is not to make the project appear compliant, but to verify whether the current project truly has a sufficient observational basis. If the sample size is still insufficient after expanding the window, it indicates that stable evaluation conditions are not yet available. For example, calculating the "CV value of IQC in the last 30 days" requires statistical analysis of the internal quality control (IQC) results for the most recent 30 days. However, if there are few IQC results within those 30 days, the calculated CV value will be inaccurate. Therefore, the time window needs to be extended to 60 days or longer to ensure a sufficient number of IQC results, making the calculated CV value statistically significant and more accurate. Thus, evidence-based adjustments to the score for this dimension are necessary after subsequent hard threshold determination.
[0149] For non-critical missing information, outright rejection should be avoided; such missing information often does not change the main direction of the conclusion, but it reduces the certainty and traceability of the conclusion, making it more suitable as a basis for evidence revision or grade limitation. Therefore, evidence revision for this dimension should be carried out after the subsequent hard threshold determination.
[0150] When key evidence is missing, we should avoid substantively ...
[0151] For the above-mentioned missing types, the specific details regarding whether to correct the score, the upper limit or limitation of the evidence level, and whether manual review is required are shown in Table 9.
[0152] Table 9. Classification and Processing Rules for Missing Data
[0153] It should be noted that "alternative evidence" in Table 9 belongs to "Level 2 evidence", with its highest level limited to Level B; "extended window evidence" also belongs to "Level 2 evidence", with its highest level also limited to Level B.
[0154] According to Table 9, key evidence is missing, which has affected the level determination. Therefore, the case is judged as insufficient evidence, and admission is suspended. Hence, there is no maximum level limit. Example 4 The aforementioned scoring mechanism is suitable for identifying general differences and gradual changes, but in practical applications, some unacceptable risks may also exist. Therefore, a circuit breaker mechanism can also be set up, independent of the aforementioned scoring and threshold system, specifically for identifying unacceptable risks. The scoring mechanism and the circuit breaker mechanism operate in parallel; when a red line condition is encountered, the process should not wait for the comprehensive score (details of which are discussed later). The circuit breaker should not be processed again until the value drops to a certain threshold (according to the calculation formula), but should instead directly enter a paused acceptance state. Therefore, this embodiment adds a parallel circuit breaker step.
[0155] In this embodiment, there are two circuit breaker triggering methods: the first is absolute triggering, which means that as long as a certain type of major anomaly, a low rating, or a major risk that has not yet been addressed is detected, the circuit breaker will be triggered directly; the second is cumulative triggering, which means that a single risk event is not sufficient to trigger the circuit breaker alone, but if multiple moderate risks occur simultaneously and accumulate to a certain extent, the circuit breaker should still enter a state of suspended acceptance. A comparison between absolute triggering and cumulative triggering can be found in Table 10. Table 10 Comparison of Absolute Trigger and Cumulative Trigger
[0156] In this regard, based on the “Typical Situations” in Table 10, those skilled in the art can set the specific conditions for absolute triggering according to the actual situation.
[0157] If we take The cumulative risk intensity of the mutual recognition unit can be expressed as: ; in, Indicates relative risk weights. Indicates the first The degree to which a moderate risk state occurs. When When the preset threshold is reached, the circuit breaker can be triggered.
[0158] Therefore, a circuit breaker can not only be a veto, but also a safety protection action after risks accumulate to a critical point.
[0159] The relevant judgment rules for various circuit breaker triggering scenarios are shown in Table 11.
[0160] Table 11 Circuit Breaker Triggering Circumstances and Judgment Rules
[0161] The following three conditions must be met for a "serious abnormality" to be considered a serious abnormality: (1) the abnormality has been confirmed, rather than just a suspected indication; (2) the abnormality can directly affect the authenticity of the report, the accuracy of the results, or the patient's safety; (3) the impact of the abnormality has not been eliminated, or the treatment has not been closed. The judgment of "persistent deviation or abnormal trend" can be made according to the type of indicator by adopting the internal quality control rules or the existing grading standards of the indicator. The same number of consecutive times and the same range of change are not uniformly set for all indicators. If there are no grading standards or the number of consecutive times and the range of change, those skilled in the art can also discuss and supplement them according to the actual situation. In addition, the judgment basis for various circuit breaker triggering situations in Table 7 does not list all situations in detail, but only some examples. In practical applications, those skilled in the art can discuss and supplement them according to the situation.
[0162] Based on the final rank of all dimensions d The final output is obtained by determining the corresponding circuit breaker state. The standard for the final output is as follows: If the final level of the agency's regular supervision The final level of regular project supervision The ultimate level of real-time project monitoring With hard threshold level If all final levels are A and no circuit breaker is triggered, then the final output will be the one that is prioritized for acceptance; if the final level is A... , , , If none of the values are C or D, and at least one is B, and the circuit breaker has not been triggered, then the final output is considered normal; if the final level... , , , If a Class C level exists but no Class D level exists, and the circuit breaker is not triggered, then the final output is cautiously accepted; if the final level... , , , If a Class D event exists, or the circuit breaker is triggered, the final output will be "Not Recommended," as shown in Table 8. That is, the final output will be: ; in, This indicates the final level of the agency's regular regulatory dimensions. This indicates the final level of the project's periodic monitoring dimension. This indicates the final level of the project's real-time monitoring dimension. Indicates the hard threshold level; Indicates the circuit breaker status. =0 indicates that the circuit breaker has not been triggered. =1 indicates that the circuit breaker has been triggered; This represents the final output of the mutual recognition unit U; This represents a linked decision mapping, which essentially compresses the "long-term quality status," "current risk status," "evidence status," and "security control status" into a final business conclusion.
[0163] Table 12 Comparison of Linkage Conditions and Final Output
[0164] Example: As shown in Table 12, if the final level and hard threshold level of all three dimensions are A, and the circuit breaker is not triggered, the final output is "Preferred Adoption"; if there are no C or D in the final level and hard threshold level of the three dimensions, and at least one B exists, and the circuit breaker is not triggered, the final output is "Normal Adoption"; if there is C in the final level and hard threshold level of the three dimensions, and no D exists, and the circuit breaker is not triggered, the final output is "Cautious Adoption"; if there is D in the final level and hard threshold level of the three dimensions, or if the circuit breaker is triggered, the final output is "Not Recommended Adoption".
[0165] Example 5 Based on the above embodiments, this embodiment adds a "rectification and re-evaluation process", which is as follows: When a mutual recognition unit U falls under any of the following circumstances, it should enter the rectification and re-evaluation process: 1. Access is not approved; 2. Access is suspended or key evidence is insufficient; 3. The dynamic supervision level is rated C and requires continuous rectification; 4. Any dimension is downgraded to D; 5. Circuit breaker is triggered; 6. Although it is currently temporarily credible, the comprehensive score continues to approach the downgrade boundary.
[0166] The correspondence between each scenario and the key points of the re-evaluation is shown in Table 13.
[0167] Table 13 Comparison of Scenarios and Key Points of Re-evaluation
[0168] The rectification and re-evaluation should adhere to the principles of "same object, same criteria, and same rules." "Same object" means the re-evaluation should still focus on the original mutually recognized unit or its direct counterpart; "same criteria" means the definition of indicators, timeframes, and evidentiary standards should remain consistent before and after the re-evaluation; and "same rules" means that original problems should not be concealed by changing evaluation standards afterward.
[0169] The rectification and re-evaluation should also follow the principle of "eliminating red-line risks first, and then restoring regular evaluation." For projects that have triggered circuit breakers, normal acceptance should not be directly restored before major risks have been closed; instead, it should be demonstrated first that the risks have been eliminated.
[0170] The significance of the rectification and re-evaluation mechanism lies not in providing a formal remedial opportunity for "failure," but in transforming mutual recognition evaluation from a single judgment into a continuous governance process. Only by establishing a closed loop can the mutual recognition system avoid the one-sided tendencies of "evaluating without improving" or "concluding without explaining."
[0171] Example 6 This embodiment provides a system for assessing the mutual recognition reliability of test results, which includes: A reliability assessment system for mutual recognition of test results, comprising: The mutual recognition unit determination module is used to determine the result provider. Intended to the recipient of the results Mutual recognition projects provided At the current assessment point To meet the mutual recognition credibility assessment needs, and to construct mutual recognition units. ; The mutual recognition data field D(U) forming module is used to form a mutual recognition unit. From the perspective of the results provider Result recipient Data is collected from quality management systems, interlaboratory quality assessment systems, and / or regional mutual recognition platforms to obtain the mutual recognition data domain D(U); wherein, the mutual recognition data domain D(U) includes access layer data. Regular regulatory data from institutions Regular monitoring data of projects Real-time monitoring data of projects ; Admission layer data Including key access indicators Support for entry indicators And dual consistency index ; The admission tier determination module is used to first determine key admission indicators. Make a judgment; if key access indicators If all criteria are approved, then the support for the admission criteria will be established. Consistency index with duality Make a judgment; otherwise, key access indicators If the judgment fails, the output should not be accepted. Further support for access indicators Consistency index with duality Make a judgment, if the admission criteria are supported Support admission score, dual consistency index If both the dual consistency index and the dual consistency index are greater than or equal to the corresponding qualified threshold, the system will proceed to the dynamic regulatory layer judgment module. If both the support admission score and the dual consistency index are greater than or equal to the corresponding lower bound of the gray zone and less than the corresponding qualified threshold, supplementary evidence will be provided before judgment is made. If the support admission score and / or the dual consistency index are less than the corresponding lower bound of the gray zone, the system will output "not recommended for acceptance". The dynamic regulatory layer determination module is used to match the observation range with the raw value of indicator i in dimension d, then match a standardization function with the observation range and obtain the standardized score of indicator i in dimension d based on the standardization function. Finally, it calculates the original score of periodic supervision by the agency based on the standardized score of indicator i in dimension d and the corresponding weight. Project periodic monitoring raw score Project Real-time Monitoring Raw Score ; The hard threshold determination module is used to classify the indicators in dimension d, determine the hard threshold conditions, and determine the periodic regulatory data of the institution based on the determination results. Regular monitoring data of projects Real-time monitoring data of the project Corresponding hard threshold level ; The final output module of the linked decision-making process is used to calculate the raw scores of each dimension d. , , Determine the corresponding original score level ; Based on the raw score levels of dimension d Hard threshold level Choose the lower level as the final level for this dimension d. According to the final level of each dimension The final output is obtained; the final output includes priority acceptance, normal acceptance, cautious acceptance, and not recommended acceptance.
[0172] Example 7 A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform steps of a method for assessing the mutual recognition reliability of test results.
[0173] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.
[0174] The memory includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or 3D interface display memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory may be an internal storage unit of the computer device, such as the hard disk or memory of the computer device. In other embodiments, the memory may also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device. Of course, the memory may include both internal storage units and external storage devices of the computer device. In this embodiment, the memory is often used to store the operating system and various application software installed on the computer device, such as the program code of the mutual recognition reliability assessment method for the test results. In addition, the memory can also be used to temporarily store various types of data that have been output or will be output.
[0175] In some embodiments, the processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor is typically used to control the overall operation of the computer device. In this embodiment, the processor is used to run program code stored in the memory or process data, for example, to run program code for the method of assessing the mutual recognition reliability of test results.
[0176] Example 8 A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of a method for assessing the mutual recognition reliability of test results.
[0177] The computer-readable storage medium stores an interface display program that can be executed by at least one processor to cause the at least one processor to perform the steps of the mutual recognition reliability assessment method for test results as described above.
[0178] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the mutual recognition reliability assessment method of the test results described in the embodiments of this application.
[0179] Test case A patient underwent a serum creatinine test at the First People's Hospital of Jintang County on May 9, 2026. After receiving the test results, the patient went to West China Hospital of Sichuan University for treatment on May 12, 2026. When the doctors at West China Hospital of Sichuan University receive these test results, they need to determine the reliability of the serum creatinine test performed at the First People's Hospital of Jintang County. This requires using the inter-recognition reliability assessment method for test results described in this application to calculate, analyze, and evaluate the inter-recognition reliability of the serum creatinine test. The specific method is as follows: Step S1: Determine the mutual recognition unit; Results Provider The receiving party is the Laboratory Department of the First People's Hospital of Jintang County. Mutual recognition project for the Department of Nephrology, West China Hospital of Sichuan University Serum creatinine (Scr), current assessment time point The time is 09:30 on May 12, 2026. Therefore, the mutual recognition unit can be represented as: U = (A, B, serum creatinine, t).
[0180] Step S2: Form the mutual recognition data field D(U); First, through the interface from the result provider Result recipient The data required for credibility assessment are collected by the quality management system, interlaboratory quality assessment system and / or regional mutual recognition platform, as shown in Table 14.
[0181] Table 14 shows the data obtained from the result provider H(A) via the port.
[0182] The above indicators are then classified, and the classification results are shown in Table 15.
[0183] Table 15 Classification Results of Each Indicator
[0184] Then, each indicator is standardized to obtain a standardized score. In this experimental example, only the calculation process of three representative indicators is shown; the remaining indicators are processed in the same way. The calculation process of these three indicators and the obtaining of their standardized scores are shown in Table 16.
[0185] Table 16 shows the process of obtaining standardized scores for the three representative indicators.
[0186] As shown in Table 16, the final standardized score for historical clinical acceptance rate was 86.4, the final standardized score for the CV value of IQC in the past 30 days was 100, and the final standardized score for interlaboratory comparison was 100.
[0187] Step S3: Admission layer determination; First, the key access indicators Instrument calibration status Difference between testing time and accreditation period The two indicators were used for judgment, and the judgment results are shown in Table 17.
[0188] Table 17 Key Access Indicators and Access Judgment Results
[0189] Since both of the above key entry criteria have been met, that is Next, we will move on to supporting access indicators. Consistency index with duality Make a judgment; Subsequently, the support access indicators were calculated. Support for access points It is 87.45:
[0190] Calculated duality index Duality index It is 0.940:
[0191] In this experimental example, the preset initial threshold is: the auxiliary admission threshold for formal admission. The lower bound of the auxiliary access gray zone is 80. The duality consistency threshold for formal admission is 70. The lower bound of the duality-consistent gray region is 0.85. The value is 0.75. Therefore, the support admission score is 87.45 ≥ 80, the duality consistency index is 0.940 ≥ 0.85, and all key admissions are approved.
[0192] Step S4: Dynamic regulatory layer determination; Before calculating the raw scores for each indicator, it is necessary to calculate the weight of each indicator. In this experimental example, we will only use the periodic monitoring dimension of the project as an example to demonstrate how to determine the weight of each indicator. The other indicators will be processed in the same way: 1. Subjective safety weight ; Assuming expert judgment: The most important indicator is: P1 interlaboratory comparison (as mentioned above). The least important indicator (Worst): P6's IQC pass rate over the past 30 days (the aforementioned is...). The remaining values are: P2's CV values for IQC over the past 30 days. P3 represents the most recent first-concentration EQA result, P4 represents the most recent second-concentration EQA result, and P5 represents the most recent third-concentration EQA result. After experts assigned relative importance, the subjective weights of each indicator were obtained through BWM optimization:
[0193] Note: In the table, P1 represents interlaboratory comparison, and P2 represents the CV value of IQC over the past 30 days. P3 represents the recent first concentration EQA result, P4 represents the recent second concentration EQA result, P5 represents the recent third concentration EQA result, and P6 represents the IQC quality control pass rate over the past 30 days.
[0194] 2. Weighting of objective information ; The information content of each indicator is calculated based on historical data. for:
[0195] Therefore, the objective information weight of each indicator is calculated. for:
[0196] 3. Combined weights ; In this experimental example, the subjective safety weight ratio is α=0.6, then the combined weight... The calculation formula is: ; Therefore, it can be calculated that:
[0197] Ultimately, the weights of these six indicators were: P1=32.1%, P2=20.4%, P3=18.4%, P4=12.8%, P5=10.8%, and P6=5.5%.
[0198] In this experimental example, the calculated raw score for periodic institutional supervision was obtained. for:
[0199] The original score for periodic supervision of the project was then calculated. for:
[0200] The final calculated raw score for real-time project monitoring for: .
[0201] Step S5: Hard threshold determination; First, divide the indicators in the dynamic management system into key indicators and non-key indicators:
[0202] Then, a hard threshold determination is performed, and the hard threshold result is:
[0203] This shows that although the project regularly monitors the original score It reached 95.989, but since the key trend indicator can only reach B, its hard threshold level can only be B.
[0204] Step S6: Evidence correction and determination of the maximum level; In this experimental case, the G dimension used complete direct evidence, the P dimension's "30-day IQC pass rate" used imputational evidence due to a small sample size, and the R dimension still used complete evidence from the current period. Therefore:
[0205] Of particular note here is the P dimension, whose original score was 95.898, placing it at A. However, due to some evidence coming from the extended window, the revised score dropped to 93.11, and the upper limit of the grade was limited to B. In other words, even if the original score reaches A, it is not advisable to directly declare it as A until the extended window evidence is replaced by complete direct evidence.
[0206] Step S7: Final output of the linkage decision; First, determine the circuit breaker status.
[0207]
[0208] This test case examines whether a circuit breaker is triggered. Assume there are currently no red-line events, such as no IQC out of control or any dimension reaching level D. Simultaneously, the platform records three risk events: a continuous increase in the IQC CV value over the past 30 days, a near-passing EQA single concentration result, and a continuous decline in historical acceptance rate. Let their risk levels be 0.40, 0.20, and 0.10, respectively, with corresponding weights of 0.50, 0.30, and 0.20. Then: Cumulative Risk Intensity for: Ω(U)=0.50×0.40+0.30×0.20+0.20×0.10=0.28.
[0209] The preset circuit breaker threshold in this experimental example Since 0.28 < 0.60, the cumulative circuit breaker will not be triggered this time. Also, there is no red line event. Therefore, the circuit breaker status is: F(U) = 0.
[0210] Based on the levels obtained above, the final levels for each dimension are as follows:
[0211] Therefore, the final grades for the three dimensions are: B for the G dimension, B for the P dimension, and A for the R dimension.
[0212] In this experimental example, the combined weight of the three dimensions , , The scores are 0.16, 0.29, and 0.55 respectively, so the overall score is: =0.16×100+0.29×95.989+0.55×100=98.84; The overall score of 98.84 indicates that it only represents a high ranking within the same final output level; the final output is still determined by the final level of the dimension and the circuit breaker status. Since this experimental example has two B-level dimensions, no D-level dimensions, and no circuit breaker, the linked output is: Π(U) = Normal acceptance.
[0213]
[0214] This experimental case effectively demonstrates the advantages of the proposed method. While the overall score of 98.84 might easily give the impression of "preferred acceptance," the platform ultimately assigns "normal acceptance" rather than "preferred acceptance" because the key trend indicator in the P dimension only reaches B. This indicates that the overall score does not mask the current risks.
[0215] Interpretation of Results: In this trial example, "normal acceptance" does not mean that the project has reached the "priority acceptance" level. Rather, it indicates that the overall quality of the project is relatively stable, the main credible evidence is basically complete, and no high-risk anomalies that significantly affect the security of mutual recognition have been found. Therefore, it can be used as a routine mutual recognition result in the clinical application process. First, Jintang County First People's Hospital can directly include this serum creatinine result in the mutual recognition chain and use it as an important reference for clinical diagnosis and treatment. Second, the recipient of the result should be simultaneously informed that the overall quality of the project is within a stable and credible range, but it is still recommended to make a comprehensive judgment based on the patient's clinical manifestations and dynamic trends. Third, routine dynamic monitoring of the project should continue, focusing on recent trend drift, quality control fluctuations, and changes in outliers to prevent the risk of subsequent persistent deviations.
[0216] If subsequent reassessment shows that the stability of interlaboratory comparisons, trend control capabilities, and supporting evidence of the project have further improved, and the key dimensions have continued to remain at the A level, then the project may be upgraded from "normal acceptance" to "priority acceptance." Conversely, if subsequent indicators show increased drift, continuous quality control anomalies, decreased evidence integrity, or some key dimensions drop to below the C level, then the project should be adjusted to "cautious acceptance" according to the degree of risk, and manual review and key supervision should be initiated if necessary.
[0217] In summary, taking the example of Jintang County First People's Hospital providing serum creatinine test results to West China Hospital of Sichuan University, this study first determines whether a basis for mutual recognition has been established between the two parties through key access indicators, supporting access scores, and dual consistency indices. Then, dynamic monitoring results are calculated from three dimensions: periodic institutional monitoring, periodic project monitoring, and real-time project monitoring, with hard thresholds and evidence corrections added. Finally, the final coordinated decision is formed by combining the circuit breaker status. The case study shows that even if the overall score is in the "priority acceptance" range, if the key indicator drops to level B, "normal acceptance" should be output instead of "priority acceptance"; if a red-line risk is triggered, it should directly switch to "not recommended for acceptance." This result indicates that this method can simultaneously identify the basis for mutual recognition, long-term quality status, and immediate risk status, possessing strong theoretical completeness and practical explanatory power.
Claims
1. A method for assessing the reliability of the mutual recognition of test results, characterized in that, Includes the following steps: Step S1: Determine the mutual recognition unit; According to the result provider The result receiver Provided mutual recognition project At the current evaluation point The mutual recognition credibility evaluation requirement, construct mutual recognition unit ; Step S2: Form the mutual recognition data field D(U); Around the mutual recognition unit From the results provider Result recipient Data is collected from quality management systems, interlaboratory quality assessment systems, and / or regional mutual recognition platforms to obtain the mutual recognition data domain D(U); wherein, the mutual recognition data domain D(U) includes access layer data. Regular regulatory data from institutions Regular monitoring data of projects Real-time monitoring data of projects ; Admission layer data Including key access indicators Support for entry indicators And dual consistency index ; Step S3: Admission layer determination; First, the key access indicators Make a judgment; if key access indicators If all criteria are approved, then the support for the admission criteria will be established. Consistency index with duality Make a judgment; otherwise, key access indicators If the judgment fails, the output should not be accepted. Further support for access indicators Consistency index with duality Make a judgment, if the admission criteria are supported Support admission score, dual consistency index If the dual consistency index is greater than or equal to the corresponding qualified threshold, proceed to step S4; if the supporting admission score and the dual consistency index are both greater than or equal to the corresponding lower bound of the gray zone and less than the corresponding qualified threshold, supplementary evidence is required before making a judgment; if the supporting admission score and / or the dual consistency index are less than the corresponding lower bound of the gray zone, the output is "not recommended for acceptance". Step S4: Dynamic regulatory layer determination; The observation range is matched based on the raw value of indicator i in dimension d. Then, a standardization function is matched based on the observation range, and the standardized score of indicator i in dimension d is obtained based on the standardization function. Finally, the standardized score of indicator i in dimension d is compared with the corresponding weights, and the raw score is regularly monitored by the calculation agency. Project periodic monitoring raw score Project Real-time Monitoring Raw Score ; Step S5: Hard threshold determination; The indicators in dimension d are divided into indicators, and hard threshold conditions are determined. Based on the determination results, the periodic regulatory data for each institution is determined. Regular monitoring data of projects Real-time monitoring data of the project Corresponding hard threshold level ; Step S6: Final output of the linkage decision; determine the corresponding raw score level , , determine the corresponding raw score level ; Based on the raw score levels of dimension d Hard threshold level Choose the lower level as the final level for this dimension d. According to the final level of each dimension The final output is obtained; the final output includes priority acceptance, normal acceptance, cautious acceptance, and not recommended acceptance.
2. The method for assessing the mutual recognition reliability of test results as described in claim 1, characterized in that, In step S2, key access indicators Including instrument calibration status Difference between testing time and accreditation period These two indicators; Support for entry indicators Including the completeness rate of IQC records over the past 30 days Historical clinical acceptance rate These two indicators; Duality consistency index Including the consistency of testing instruments Methodological consistency Consistency of unit of result Consistency of reference interval mapping These four indicators; Regular regulatory data from institutions Including historical clinical mutual recognition rate This one indicator; Project periodic monitoring data Including the CV value of IQC over the past 30 days IQC pass rate in the past 30 days Recent first concentration EQA results Recent second concentration EQA results Recent EQA results for the third concentration Inter-laboratory comparison These 6 indicators; Real-time monitoring data of projects Including the indoor quality control status of the first concentration of the day The second concentration indoor quality control status on the same day The third concentration of the day in the indoor quality control status These three indicators.
3. The method for assessing the mutual recognition reliability of test results as described in claim 2, characterized in that, The indicators in the mutually recognized data domain D(U) are classified into three categories: benefit indicators, cost indicators, and binary or ordinal indicators. For benefit-type indicators and cost-type indicators, different standardization functions are used to calculate the standardized scores; binary indicators are scored according to two states, and ordinal indicators are scored according to a level mapping table.
4. The method for assessing the mutual recognition reliability of test results as described in claim 1, characterized in that, In step S3, the specific steps for determining the admission layer are as follows: Step S3-1: Determine key admission criteria ; The instrument calibration status was checked separately. Difference between testing time and accreditation period Perform a judgment; if both judgments pass, proceed to step S3-2; If the instrument is in calibration status Difference between testing time and accreditation period If at least one of the results is deemed unacceptable, the output should not be accepted. Step S3-2: Calculate the support admission score and dual consistency index; Based on the completeness rate of IQC records over the past 30 days Standardized score Weight Calculate and obtain the contribution value of the integrity rate. Based on historical clinical acceptance rates Standardized score Weight Calculate and obtain the contribution value of the recognition rate. Based on the contribution value of the completeness rate , Contribution value of recognition rate Calculate and obtain the support admission score ; Based on the consistency of the testing instruments Matching degree Weight Calculate and obtain the contribution value of the detection instrument. According to methodological consistency Matching degree Weight Calculate and obtain the methodological contribution value. Based on the consistency of the units of the results Matching degree Weight Calculate and obtain the unit contribution value. ; Based on the consistency of the reference interval mapping Matching degree Weight Calculate and obtain the reference interval mapping contribution value. ; Based on the contribution value of the testing instrument Methodological contribution value Result unit contribution value Reference interval mapping contribution value Calculate and obtain the duality index ; Step S3-3: Support admission and duality consistency determination; If admission points are supported Dual consistency index If all values are greater than or equal to the corresponding qualifying threshold, then the application is deemed approved and proceeds to step S4; if the application is supported... With duality index All are greater than or equal to the corresponding lower bound of the gray area, and support admission scoring. or / and duality index If the score is below the corresponding qualifying threshold, supplementary evidence will be required before a final determination is made; if the admission score is supported... or / and duality index If the value is less than the corresponding lower bound of the gray area, the output is not recommended for use.
5. The method for assessing the mutual recognition reliability of test results as described in claim 4, characterized in that, In step S3-2, the standardized scores of each category of indicators are calculated as follows: For benefit-type indicators, let... ,in , , All are level thresholds. Let be the lower bound of the acceptable numerical range; then, determine the standardization function based on the observed value of the i-th indicator under dimension d, and calculate the standardized score based on the standardization function; where the relationship between the range of observed values and the standardization function is: For cost-type indicators, let... ,in , , All are level thresholds. Let be the upper limit of the acceptable numerical range; then, determine the standardization function based on the observed value of the i-th indicator under dimension d, and calculate the standardized score based on the standardization function; where the relationship between the observed value range and the standardization function is: 。 6. The method for assessing the mutual recognition reliability of test results as described in claim 4, characterized in that, In step S3-2, based on the sufficiency of samples at each stage, the weights of each indicator are calculated as follows: When the initial sample size is insufficient, calculate the subjective safety weights separately. Weight of objective information Then, calculate the combined weight based on both subjective and objective weights. Combined weights That is, the weight of index i; Once sufficient samples are available in the later stages, the empirical importance vector will be constructed. Combined weights By integrating and recalibrating based on historical experience, Historical experience recalibration That is, the weight of index i; Among them, subjective safety weight The calculation formula is: ; , , , ; Objective information weight The calculation formula is: ; ; ; ; Combined weights The calculation formula is: ; Historical experience recalibration The calculation formula is: ; in, This indicates that the consistency index is achieved under the premise that the following conditions are met. Minimum, This represents the weight to be determined for the i-th indicator. This indicates the weight of the optimal indicator B. This represents the weight of the worst-case indicator W. This represents the importance scale value of the best indicator B given by the decision-maker relative to indicator i. This represents the importance scale value of indicator i given by the decision-maker relative to the worst-case indicator W. Indicates consistency index; Let represent the standard deviation of the i-th indicator in dimension d across all evaluated objects. The Pearson correlation coefficient represents the relationship between index i and index k. This indicates the number of indicators k. This indicates the amount of information contained in index i; This represents the original observation value of the j-th object in dimension d on the i-th index. Indicates that index i under dimension d is among all The arithmetic mean of a sample, This represents the total number of samples in dimension d; Indicates the degree of safety orientation. The weights represent the combined weights. This represents the empirical importance vector.
7. The method for assessing the mutual recognition reliability of test results as described in claim 1, characterized in that, In step S6, the raw scores for each dimension d are obtained. , , Determine the corresponding original score level At that time, the standard for classifying the original score level was: 90 ≤ original score ≤ 100. The answer is A; 80 ≤ raw score < 90. The answer is B; 70 ≤ raw score < 80. The answer is C; 0 ≤ original score < 70. The answer is D; Based on the raw score levels of the agency's regular monitoring. and the corresponding hard threshold level The final level of receiving regular supervision from the agency Based on the original score level of the project's regular monitoring. and the corresponding hard threshold level The final level of project regular monitoring Based on the original score level of the project's real-time monitoring and the corresponding hard threshold level The ultimate level of obtaining real-time project monitoring ; The final output standard is: the final rating of the institution under regular supervision. The final level of regular project supervision Ultimate level of real-time project monitoring If all are grade A, the final output will be the one that is accepted first; if the final grade is... , , If none of the values are C or D, and at least one is B, then the final output is considered normal; if the final level... , , If a grade C exists and no grade D exists, the final output is "cautious acceptance"; if the final grade... , , If a D-level is present, the final output will be "not recommended for adoption".
8. A system for assessing the reliability of mutual recognition of test results, characterized in that, include: The mutual recognition unit determination module is used to determine the result provider. Intended to the recipient of the results Mutual recognition projects provided At the current assessment point To meet the mutual recognition credibility assessment needs, and to construct mutual recognition units. ; The mutual recognition data field D(U) forming module is used to form a mutual recognition unit. From the results provider Result recipient Data is collected from quality management systems, interlaboratory quality assessment systems, and / or regional mutual recognition platforms to obtain the mutual recognition data domain D(U); wherein, the mutual recognition data domain D(U) includes access layer data. Regular regulatory data from institutions Regular monitoring data of projects Real-time monitoring data of projects ; Admission layer data Including key access indicators Support for entry indicators And dual consistency index ; The admission tier determination module is used to first determine key admission indicators. Make a judgment; if key access indicators If all criteria are approved, then the support for the admission indicators will be supported. Consistency index with duality Make a judgment; otherwise, key access indicators If the judgment fails, the output should not be accepted. Further support for access indicators Consistency index with duality Make a judgment, if the admission criteria are supported Support admission score, dual consistency index If both the dual consistency index and the dual consistency index are greater than or equal to the corresponding qualified threshold, the system will proceed to the dynamic regulatory layer judgment module. If both the support admission score and the dual consistency index are greater than or equal to the corresponding lower bound of the gray zone and less than the corresponding qualified threshold, supplementary evidence will be provided before judgment is made. If the support admission score and / or the dual consistency index are less than the corresponding lower bound of the gray zone, the system will output "not recommended for acceptance". The dynamic regulatory layer determination module is used to match the observation range based on the raw value of indicator i in dimension d, then match a standardization function based on the observation range, and obtain the standardized score of indicator i in dimension d based on the standardization function. Finally, it calculates the original score of periodic supervision by the agency based on the standardized score of indicator i in dimension d and the corresponding weight. Project periodic monitoring raw score Project Real-time Monitoring Raw Score ; The hard threshold determination module is used to classify the indicators in dimension d, determine the hard threshold conditions, and determine the periodic regulatory data of the institution based on the determination results. Regular monitoring data of projects Real-time monitoring data of the project Corresponding hard threshold level ; The final output module of the linked decision-making process is used to calculate the raw scores of each dimension d. , , Determine the corresponding original score level ; Based on the raw score levels of dimension d Hard threshold level Choose the lower level as the final level for this dimension d. According to the final level of each dimension The final output is obtained; the final output includes priority acceptance, normal acceptance, cautious acceptance, and not recommended acceptance.
9. A computer device, characterized in that: It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The device stores a computer program that, when executed by a processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 7.
Citation Information
Patent Citations
Clinical examination result mutual recognition method and device, equipment and storage medium
CN118866214A
Medical detection data management system and management method
CN120878014A
