Verification item reference interval construction system based on gender and age

Abstract

The invention discloses a sex and age-based test item reference interval construction system. The method comprises the following steps: acquiring a set of test data of physical examinees from a laboratory information system; carrying out duplicate removal on the set of the examination data of the health examinees and carrying out data collection and arrangement to obtain a set of alternative examination data of the health examinees; performing data grouping on the set of the test data of the alternative physical examinees based on age and gender to obtain a grouping result; performing outlier processing and data distribution description on a data subset of the test data of each candidate physical examinee in the grouping result; using a big data algorithm to determine a reference interval of a data subset of the test data of each candidate physical examinee to obtain a plurality of reference intervals; and constructing a continuous reference interval based on the plurality of reference intervals. In this way, the test item reference interval is constructed based on the test data of the real world, the feasibility is improved, and the cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of establishing and verifying reference intervals for test items, and specifically to a system for constructing reference intervals for test items based on gender and age. Background Art

[0002] Reference intervals are important criteria for clinical interpretation of laboratory test results and judgment of health or disease status. There are significant physiological differences between people of different age groups and genders, and these differences will lead to reference intervals that affect the results of some test items. However, it is difficult for most laboratories to establish reference intervals based on gender and age, resulting in inaccurate reference intervals, unnecessary diagnosis and treatment for patients, and waste of medical resources. The establishment of reference intervals is divided into direct and indirect methods. The direct method consumes a lot of time, resources and costs, and it is difficult to obtain the best results for items related to age and gender. In clinical practice, the use of indirect methods based on laboratory real-world big data to establish reference intervals suitable for the population served by the laboratory is of great significance for disease diagnosis, treatment and prevention.

[0003] However, there are also some problems with the algorithm models currently used. For example: 1. The establishment and verification of reference intervals is complicated, time-consuming, and has high technical bottlenecks, making it difficult to achieve functions such as data deduplication, out-of-threshold value exclusion, non-normal data conversion, intelligent grouping, and big data construction algorithms; 2. Due to the differences in data distribution characteristics, there are differences between reference intervals calculated by different algorithms, making it difficult to select a suitable algorithm; 3. The indirect method algorithm has restrictions on the proportion of pathological data mixed in the data. When the mixed pathological data exceeds the limit, the established reference interval may be biased.

[0004] Therefore, it is of great urgency to further understand the characteristics and principles of indirect methods and construct appropriate reference interval establishment schemes. Summary of the invention

[0005] One advantage of the present application is that it provides a system for constructing reference intervals for test items based on gender and age, wherein the system for constructing reference intervals for test items based on gender and age groups the test data according to gender and age, and uses a variety of big data algorithms to establish reference intervals corresponding to various age groups of different genders, which can improve the accuracy and suitability of the reference intervals to a certain extent.

[0006] Another advantage of the present application is that it provides a system for constructing reference intervals for test items based on gender and age, wherein the collection of test data of healthy physical examinees in the reference interval data processing system for test items based on gender and age comes from real-world patient medical records, and the source and use of the medical records are legal and compliant. In this way, on the one hand, the data source is reliable, and on the other hand, there is no need to specially recruit healthy physical examinees for testing in order to conduct reference interval experiments, which can improve feasibility and reduce costs to a certain extent.

[0007] According to one aspect of the present application, a system for constructing reference intervals for test items based on gender and age is provided, which includes: The test data collection module is used to obtain the collection of test data of healthy examinees from the laboratory information system; A deduplication module, used for deduplicating the set of the examination data of the healthy physical examinees based on age, gender and name or medical record number to obtain a set of the examination data of the healthy physical examinees after deduplication; A data screening module, for processing the set of healthy physical examinee test data after deduplication based on inclusion and exclusion criteria to obtain a set of candidate healthy physical examinee test data; A grouping module, used for grouping the set of the candidate healthy physical examinee test data based on age and gender to obtain a grouping result, wherein the grouping result includes a plurality of data subsets of the candidate healthy physical examinee test data; A data distribution description module, used for performing outlier processing and data distribution description on the data subsets of the test data of each candidate healthy physical examinee in the grouping result to obtain a plurality of data distribution description results; A reference interval determination module, for determining the reference interval of each of the candidate healthy physical examinee's test data subsets using a variety of big data algorithms based on the data distribution description results of each of the candidate healthy physical examinee's test data subsets; and The continuous reference interval construction module is used to construct a continuous reference interval based on gender, age and the multiple reference intervals.

[0008] In one embodiment of the system for constructing reference intervals for test items based on gender and age as described in the present application, the continuous reference interval construction module is further used to process the multiple reference intervals using a GAMLSS model to obtain the continuous reference interval.

[0009] In one embodiment of the system for constructing reference intervals for test items based on gender and age as described in the present application, it also includes: a consistency evaluation module for evaluating the consistency of results of different construction algorithms for reference intervals of data subsets of each of the candidate healthy examinee test data; and a pass rate verification module for verifying the pass rate of the reference intervals of data subsets of each of the candidate healthy examinee test data.

[0010] In one embodiment of the system for constructing reference intervals of test items based on gender and age as described in the present application, the deduplication module is further used to retain only the most recent test data of each healthy examinee in the set of healthy examinee test data.

[0011] In one embodiment of the system for constructing reference intervals for test items based on gender and age as described in the present application, the data screening module is further used to exclude abnormal healthy examinee test data that affects the reference intervals for test items from the set of healthy examinee test data after deduplication based on pre-made inclusion and exclusion criteria.

[0012] In one embodiment of the system for constructing reference intervals for test items based on gender and age as described in the present application, the grouping module is further used to group the set of test data of the candidate healthy examinees based on age and gender using the Harris & Boyd method or the standard deviation ratio method.

[0013] In one embodiment of the system for constructing reference intervals for test items based on gender and age as described in the present application, the data distribution description module is further used to perform distribution description on the test data of the candidate healthy examinees to obtain the multiple data distribution description results, wherein the manner of performing distribution description on the test data of the candidate healthy examinees is selected from at least one of the following manners: probability plot, quantile plot, kernel density plot, Shapiro-Wilk test, Kolmogorov-Smilov test; and eliminating outliers from data subsets of each candidate healthy examinee's test data in the grouping results to obtain the test data of the candidate healthy examinees after outlier processing.

[0014] In one embodiment of the system for constructing reference intervals for test items based on gender and age as described in the present application, the following big data algorithms are used to determine the reference intervals of data subsets of the test data of each of the candidate healthy examinees to obtain multiple reference intervals: EP28 non-parametric method, EP28 parametric method, refineR algorithm, Kosmic algorithm, TMC algorithm.

[0015] In one embodiment of the system for constructing reference intervals for test items based on gender and age as described in the present application, the consistency assessment module is further used to calculate the bias ratio of reference intervals established by other algorithms based on the reference intervals established by the EP28 non-parametric method.

[0016] In one embodiment of a system for constructing reference intervals for test items based on gender and age as described in the present application, the pass rate verification module is further used to obtain physical examination data of test items in another time period from the laboratory information system as verification data to verify the reference interval; wherein, when the pass rate of the reference interval is ≥90%, the reference interval passes the verification; otherwise, the reference interval fails the verification.

[0017] Further objectives and advantages of the present application will be fully reflected through understanding of the following description and drawings.

[0018] These and other objects, features and advantages of the present application are fully reflected in the following detailed description, drawings and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components.

[0020] Figure 1 The figure shows a schematic diagram of the framework of a system for constructing reference intervals for test items based on gender and age according to an embodiment of the present application.

[0021] Figure 2 The figure illustrates a flowchart diagram of an example of constructing a reference interval for a test item based on gender and age according to an embodiment of the present application.

[0022] Figure 3 The diagram illustrates a scenario diagram in which a system for constructing reference intervals for test items based on gender and age according to an embodiment of the present application performs data distribution description on data subsets of test data of each candidate healthy examinee in the grouping results to obtain multiple data distribution description results.

[0023] Figure 4 The diagram illustrates a scenario diagram of a system for constructing reference intervals for test items based on gender and age according to an embodiment of the present application to construct continuous reference intervals.

[0024] Figure 5The diagram illustrates another scenario diagram of a system for constructing continuous reference intervals for test item reference intervals based on gender and age according to an embodiment of the present application.

[0025] Figure 6 Another schematic diagram of a framework diagram of a system for constructing reference intervals for test items based on gender and age according to an embodiment of the present application is illustrated. DETAILED DESCRIPTION

[0026] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0027] It is understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "one" cannot be understood as a limitation on the number. "Multiple" means greater than or equal to two.

[0028] Although ordinals such as "first," "second," and the like will be used to describe various components, those components are not limited herein. The term is used only to distinguish one component from another. For example, a first component may be referred to as a second component, and likewise, a second component may be referred to as a first component without departing from the teachings of the present application. The term "and / or" as used herein includes any and all combinations of one or more associated listed items.

[0029] The terms used herein are only used for the purpose of describing various embodiments and are not intended to be limiting. As used herein, the singular form is intended to include the plural form, unless the context clearly indicates an exception. It will also be understood that the terms "including" and / or "having" when used in this specification specify the presence of the described features, numbers, operations, components, elements, or combinations thereof, without excluding the presence or addition of one or more other features, numbers, operations, components, elements, or combinations thereof.

[0030] like Figures 1 to 6 As shown, a system 100 for constructing reference intervals for test items based on gender and age according to an embodiment of the present application is illustrated. The system for processing reference intervals for test items based on gender and age groups test data according to gender and age, and establishes reference intervals corresponding to each age group of different genders, which can improve the accuracy of the reference intervals to a certain extent.

[0031] Specifically, the reference interval data processing system for test items based on gender and age includes: a test data acquisition module 10, a deduplication module 20, a data screening module 30, a grouping module 40, a data distribution description module 50, a reference interval determination module 60 and a continuous reference interval construction module 70. The test data acquisition module 10 is used to obtain a set of healthy physical examiner test data from a laboratory information system (LIS); the deduplication module 20 is used to deduplicate the set of healthy physical examiner test data based on age, gender and name or medical record number to obtain a set of deduplicated healthy physical examiner test data; the data screening module 30 is used to process the set of deduplicated healthy physical examiner test data based on inclusion and exclusion criteria to obtain a set of candidate healthy physical examiner test data; the grouping module 40 is used to group the set of candidate healthy physical examiner test data based on age and gender to obtain a grouping result, and the grouping result includes a plurality of candidate healthy physical examiner test data. A data subset of the test data of healthy physical examinees is selected; the data distribution description module 50 is used to perform outlier processing and data distribution description on the data subsets of the test data of each candidate healthy physical examinee in the grouping result to obtain multiple data distribution description results; the reference interval determination module 60 is used to determine the reference interval of each data subset of the test data of the candidate healthy physical examinee based on the data distribution description results of the data subsets of the test data of each candidate healthy physical examinee, using multiple big data algorithms; the continuous reference interval construction module 70 is used to construct a continuous reference interval based on the multiple reference intervals.

[0032] Accordingly, in the process of constructing the reference intervals of the test items, the test item reference interval data processing system based on gender and age first obtains a set of healthy physical examinee test data from the laboratory information system through the test data acquisition module 10; then, the deduplication module 20 deduplicates the set of healthy physical examinee test data based on age, gender and name or medical record number to obtain a set of healthy physical examinee test data after deduplication; then, the data screening module 30 processes the set of healthy physical examinee test data after deduplication based on inclusion and exclusion criteria to obtain a set of candidate healthy physical examinee test data; then, the grouping module 40 sorts the candidate healthy physical examinee test data based on age and gender. A set of test data is grouped to obtain a grouping result, and the grouping result includes a plurality of data subsets of the test data of candidate healthy physical examiners; then, the data distribution description module 50 performs outlier processing and data distribution description on the data subsets of the test data of candidate healthy physical examiners in the grouping result to obtain a plurality of data distribution description results; thereafter, the reference interval determination module 60 determines the reference interval of each of the data subsets of the test data of the candidate healthy physical examiners based on the data distribution description results of the data subsets of the test data of the candidate healthy physical examiners and utilizes a plurality of big data algorithms; further, the continuous reference interval construction module 70 constructs a continuous reference interval based on the plurality of reference intervals.

[0033] In the process of obtaining a set of healthy examinee test data from the laboratory information system through the test data acquisition module 10, the set of healthy examinee test data is a set of real test data of healthy examinees, which comes from the medical records of patients in the real world, that is, real patients are used as healthy examinees, and data from the medical records of multiple real patients are selected as the set of healthy examinee test data; in this way, on the one hand, the data source is reliable, and on the other hand, there is no need to specially recruit healthy examinees for testing in order to establish a reference interval experiment, which can improve feasibility and reduce costs to a certain extent.

[0034] In the process of obtaining a collection of healthy examinee's test data from the laboratory information system through the test data acquisition module 10, the database parameters are first configured in the laboratory information system. After the test database and the laboratory information system are communicatively connected, at least part of the medical records of the healthy examinees in the configured database are selected and exported to obtain preliminary exported data of the medical records of the healthy examinees in the database.

[0035] In the process of configuring the laboratory information system database, select an existing database or a new database to obtain basic database information and medical records of healthy examinees in the database; wherein the basic information in the database includes database name, database type (for example, Mysql, Oracle), database driver, database Internet Protocol address (i.e., database IP (Internet Protocol Address) address), database port number, database user name, and database password; the medical records of healthy examinees in the database include name, medical record number, sample number, gender, age, date of birth, test item name, test time, test instrument, test results, etc.

[0036] In the process of selecting and exporting at least part of the medical records of healthy examinees in the configured database, at least part of the medical records of healthy examinees in the configured database can be exported by writing Structured Query Language (SQL), selecting the name of the test item, year, or other selection conditions.

[0037] The preliminary exported data of the medical records of the healthy examinees in the database can be directly used as a collection of the inspection data of the healthy examinees, and the preliminary exported data of the medical records of the healthy examinees in the database can also be managed, for example, added, deleted, modified, etc., to obtain the managed exported data of the medical records of the healthy examinees in the database.

[0038] Correspondingly, the set of inspection data of healthy examinees includes inspection data of multiple healthy examinees, and the inspection data of each healthy examinee is the preliminary exported data of the medical records of the healthy examinee in the preliminary exported data of the medical records of the healthy examinees in the database, or the post-management exported data of the medical records of the healthy examinee in the post-management exported data of the medical records of the healthy examinees in the database.

[0039] In the process of deduplicating the set of healthy examinee's examination data based on age, gender and name or medical record number through the deduplication module 20 to obtain the deduplicated set of healthy examinee's examination data, only the most recent test data of each healthy examinee in the set of healthy examinee's examination data may be retained.

[0040] Specifically, the inspection data of each healthy physical examinee includes at least one healthy physical examinee inspection data of at least one inspection item. After the inspection time is arranged in descending order by a data block algorithm and a fingerprint generation and comparison algorithm, multiple healthy physical examinee inspection data of the same healthy physical examinee can be deleted based on age, gender, name or medical record number, and only the most recent inspection data of each healthy physical examinee is retained, so as to limit the inspection data of each healthy physical examinee in the set of the healthy physical examinee inspection data after deduplication to a single healthy physical examinee inspection data of each healthy physical examinee.

[0041] It should be understood that the deduplication module 20 can also deduplicate the set of the healthy physical examinee's examination data based on other data, for example, deduplicate the set of the healthy physical examinee's examination data based on age, gender and name.

[0042] In the process of processing the set of healthy examinee test data after deduplication based on the inclusion and exclusion criteria to obtain a set of alternative healthy examinee test data by the data screening module 30, abnormal healthy examinee test data that affects the reference interval of the test item in the set of healthy examinee test data after deduplication is excluded based on the pre-set inclusion and exclusion criteria.

[0043] Abnormal examination data of healthy examinees include: examination data of healthy examinees with diabetes, examination data of healthy examinees with endocrine disorders, examination data of healthy examinees with tumors, examination data of healthy examinees with inflammation, examination data of healthy examinees with abnormal blood test results, examination data of healthy examinees with abnormal cardiovascular test results, examination data of healthy examinees with abnormal kidney test results, examination data of healthy examinees with abnormal liver test results, examination data of healthy examinees who are pregnant, examination data of healthy examinees who have undergone surgery within a preset time interval, and examination data of healthy examinees with abnormal test results of other test items that have an impact on the reference interval.

[0044] In the process of grouping the set of candidate healthy examinee test data based on age and gender to obtain grouping results by the grouping module 40, the set of candidate healthy examinee test data can be grouped based on age and gender by the Harris & Boyd method or the standard deviation ratio method.

[0045] In one example of the present application, a group is established in the following manner.

[0046] First, the candidate health examinee test data of health examinees of different genders are sorted based on age to obtain sorted candidate health examinee test data samples, wherein the sorted candidate health examinee test data samples are recorded as a1. a 2. a 3... a n-1 , a n .

[0047] Then, the sorted candidate healthy physical examination data are divided into at least two groups of data subsets, wherein the two groups of data subsets are respectively recorded as the first data subsets X 1 and the second data subset X 2. Each data subset includes at least two of the sorted candidate healthy examinee test data samples. In one example of the present application, the initial first data subset X 1 is { a 1, a 2}, the initial second data subset X 2 is { a 3, a 4}.

[0048] Next, determine whether the data subset meets the preset grouping conditions; when the data subset meets the preset grouping conditions, group the data subset according to the preset grouping iteration method; when the data subset does not meet the preset grouping conditions, group configuration is performed based on the number of sorted candidate healthy physical examination data samples respectively contained in at least two groups of the data subsets.

[0049] The preset grouping conditions are: Z Represents the average value of all the sorted candidate healthy examinee test data samples of the first data subset; Represents the average value of all the sorted candidate healthy examinee test data samples of the second data subset; S 1 represents the standard deviation of all the sorted candidate healthy examinee test data samples of the first data subset; S 2 represents the standard deviation of all the sorted candidate healthy physical examinee test data samples of the second data subset; The variance of all the sorted candidate healthy examinee test data samples of the first data subset is represented; The variance of all the sorted candidate healthy examinee test data samples of the second data subset is represented; represents the absolute value of the difference between the standard deviation of all the sorted candidate healthy physical examinee test data samples of the first data subset and the standard deviation of all the sorted candidate healthy physical examinee test data samples of the second data subset; The maximum value of the standard deviation of all the sorted candidate healthy physical examinee test data samples of the first data subset and the standard deviation of all the sorted candidate healthy physical examinee test data samples of the second data subset; n 1 represents the number of all the sorted candidate healthy examinee test data samples in the first data subset; Represents the number of all the sorted candidate healthy physical examinee test data samples of the second data subset; k is the default value, usually 3 to 5.

[0050] The preset grouping iteration method is: select the second data subset X 2 replaces the first data subset with at least two of the sorted candidate healthy examinee test data samples after The sorted candidate healthy examinee test data samples in the above, wherein the first data subset is used to replace X At least two of the sorted candidate health examinee inspection data samples in 1 are defined as first substitute health examinee inspection data samples; at least two of the sorted candidate health examinee inspection data samples after the first substitute health examinee inspection data samples are selected to fill in the second data subset X The sorted candidate healthy physical examination data samples in 2.

[0051] In the process of grouping and configuring based on the number of the sorted candidate healthy physical examinee test data samples respectively contained in at least two groups of the data subsets, when the first data subset X The number of the selected healthy physical examination test data samples after sorting in 1 is equal to the second data subset X When the number of selected healthy examinee test data samples after sorting in 2 is X At least one of the sorted candidate healthy physical examination data samples in 2 is filled into the first data subset X 1, will be ranked in the second data subset X At least one of the sorted candidate healthy examinee test data samples after the healthy examinee test data sample in 2 is added to the second data subset X 2, to obtain the updated first data subset X 1 and the updated second data subset X 2. For example, the second data subsetX 2. A sample of the test data of the selected healthy examinees after sorting Fill to the first data subset X 1, will be ranked in the second data subset X 2. At least one of the healthy examinee test data samples after a 5 Supplement to the second data subset X 2, to obtain the updated first data subset X 1 and the updated second data subset X 2; wherein the updated first data subset X 1 is { a 1, a 2, a 3}; Updated second data subset X 2 is { a 4, a 5}. When the first data subset after update X 1 and the updated second data subset X 2. When the preset grouping condition is met, the updated first data subset X 1 and the updated second data subset X 2 Group according to the preset grouping iteration method. When the first data subset after update X 1 and the updated second data subset X 2 does not meet the preset grouping condition, and when the updated second data subset X The number of the selected healthy physical examinee test data samples after the sorting in 1 is less than the first data subset after the update X When the number of candidate healthy physical examination data samples after sorting described in 1 is equal to, it will be ranked in the updated second data subset X At least one of the sorted candidate healthy examinee test data samples after the healthy examinee test data sample in 2 is added to the second data subset X 2, to obtain the second updated data subset X 2, for example, the second data subset after the update X At least one of the selected healthy examinee test data samples after the healthy examinee test data sample in 2 a 6 Fill in the second data subset X 2, to obtain the second updated data subset X 2 is { a 4, a 5, a 6}. When the first data subset after update X 1 and the second data subset after updating again X2. When the preset grouping condition is met, the updated first data subset X 1 and the second data subset after updating again X 2 Group according to the preset grouping iteration method. When the first data subset after update X 1 and the second data subset after updating again X 2. If the preset grouping condition is not met, the updated first data subset is checked. X 1 and the second data subset after updating again X The number of candidate healthy examinee test data samples after sorting as described in 2.

[0052] The data distribution description module 50 performs outlier processing and data distribution description on the data subsets of the test data of each candidate healthy physical examinee in the grouping result (such as Figure 3 In the process of obtaining multiple data distribution description results by using the Tukey algorithm to eliminate the abnormal values ​​of the data subsets of the candidate healthy physical examinee test data in the grouping results, so as to obtain the candidate healthy physical examinee test data after outlier processing; the distribution description method of the candidate healthy physical examinee test data after outlier processing is selected from at least one of the following methods: probability map, quantile map, kernel density map, Shapiro-Wilk test, Kolmogorov-Smirnov test. The subset data that does not meet the normality is converted into normal distribution or approximate normal distribution data using Box-Cox.

[0053] In the process of eliminating the outliers of the data subsets of the test data of each candidate healthy physical examiner in the grouping result by using the Tukey algorithm, the quartiles of the data subsets of the test data of each candidate healthy physical examiner in the grouping result are calculated, and the formula P 25 -1.5*(P 75 -P 25 ) calculates the upper limit for identifying outliers, and uses the formula P 75 +1.5*(P 75 -P 25 ) calculates the lower limit for identifying outliers and removes data outside the range. 25 is the 25% percentile among the percentiles. 75 The 75% percentile among the percentiles.

[0054] It should be understood that the outliers of the data subsets of the test data of each candidate healthy physical examiner in the grouping result can also be eliminated by other means to obtain the test data of the candidate healthy physical examiner after outlier processing, for example, the outliers of the data subsets of the test data of each candidate healthy physical examiner in the grouping result are eliminated by the ±3SD method. In the process of eliminating the outliers of the data subsets of the test data of each candidate healthy physical examiner in the grouping result by the ±3SD method, the mean value and standard deviation of the data subsets of the test data of each candidate healthy physical examiner are calculated, and the upper and lower limits for identifying outliers are determined, so that the data outside the range is eliminated.

[0055] In another technical solution of the present application, outliers of the data subsets of the candidate healthy physical examinee test data in the grouping result are eliminated to obtain the candidate healthy physical examinee test data after outlier processing, including the following steps: S151: vectorize and encode each candidate healthy physical examiner's test data in the data subset of the candidate healthy physical examiner's test data to obtain a set of candidate healthy physical examiner's test data embedded coding vectors. The purpose of vectorizing and encoding each candidate healthy physical examiner's test data in the data subset of the candidate healthy physical examiner's test data is to convert non-numeric data into a numerical vector form that can be processed by a computer, so that mathematical operations can be used to analyze these data.

[0056] S152: extracting the i-th candidate healthy physical examinee test data embedding coding vector from the set of candidate healthy physical examinee test data embedding coding vectors as a query vector. The i-th selected here refers to any individual data vector, which is used for comparison and analysis in subsequent steps.

[0057] S153: inputting the query vector and the set of candidate healthy physical examination subject test data embedding coding vectors into a feature dynamic query analysis network to obtain a query response coding vector within the candidate test data set; S154: Input the query response coding vector in the alternative test data set into the outlier identifier based on the support vector machine model to obtain a recognition result, and the recognition result is used to indicate whether the embedded coding vector of the test data of the i-th alternative healthy physical examinee is an outlier. Finally, the response coding vector obtained in the previous step is input into an outlier identifier based on a support vector machine (SVM) model. Support vector machine is a supervised learning model, which is used here to distinguish between normal values ​​and outliers. Through the trained SVM model, it can be determined whether the embedded coding vector of the test data of the i-th alternative healthy physical examinee is an outlier. If the model determines that it is abnormal, the data point will be regarded as an outlier and will be eliminated.

[0058] In the technical solution of the present application, the step S153 includes: firstly, inputting the query vector and each candidate healthy physical examinee test data embedding coding vector in the set of candidate healthy physical examinee test data embedding coding vectors into an adaptive decision parser to obtain a set of query-candidate test data decision point state implicit coding vectors. This process can be expressed as: in, represents the set of embedded coding vectors of candidate healthy examinee test data, , , , Respectively represent the first, second, and and The test data of candidate healthy examinees are embedded in the coding vector, represents the total number of embedded coding vectors of candidate healthy examinee test data, represents the query vector, Indicates The first weight matrix, Indicates The first bias vector, represents matrix multiplication, represents the S-type activation function, represents the set of implicit encoding vectors of the query-alternative test data decision point states, , , , Respectively represent the first, second, and and The query-alternative test data decision point state implicitly encoded vector.

[0059] It should be understood that the nonlinear decision response unit is intended to capture the decision response result of the candidate healthy examinee's test data embedded encoding vector given the query vector. At the technical level, the nonlinear decision response unit uses nonlinear mapping and feature interaction to construct the decision response state latent space of the two. In this latent space, the features that were originally difficult to distinguish may be effectively separated, and the complex decision boundary is also simplified. The core of the nonlinear decision response unit is that it uses nonlinear mapping and feature interaction technology. Nonlinear mapping allows the model to capture complex relationships in the data, which are difficult to identify or express in linear models. For example, in health examination data, some potential health risk factors may not be significantly manifested in a single indicator alone, but require comprehensive consideration of multiple related indicators and their interactions to be accurately identified. Interaction between features further enhances this ability. It not only focuses on the performance of a single feature, but also analyzes the mutual influence between different features. In this way, the nonlinear decision response unit can create a new representation in the latent space, in which each point (i.e., the latent encoding vector) reflects the state and characteristics of a candidate healthy examinee relative to the query vector. In this latent space, the features that were originally difficult to distinguish may be effectively separated. This means that individuals that appear similar in the original high-dimensional space but are actually subtly different can be more clearly distinguished in the latent space.

[0060] Next, based on the set of implicitly encoded vectors of the query-candidate test data decision point states, the query-candidate test data decision point state class neighborhood matrix is ​​calculated. This process can be expressed as: in, Indicates query-alternative test data decision point state implicit encoding vector, represents the square of the one-norm of a vector, represents the inverse hyperbolic cosine function, Represents the query-alternative test data decision point state class neighborhood matrix The value of the position, represents the query-alternative test data decision point state class neighborhood matrix, , , , Respectively represent the query-alternative test data decision point state class neighborhood matrix , , , The value of the position.

[0061] It should be understood that the process of calculating the class neighborhood matrix is ​​to explicitly model the local structural information between the implicit encoding vectors of the query-alternative test data decision point states. Specifically, in the context of health examination data, the class neighborhood matrix reflects the relative position and connection relationship between different healthy examinees in the feature space. For example, if the implicit encoding vectors of the query-alternative test data decision point states of two healthy examinees are very close, then the corresponding elements of the two healthy examinees in the class neighborhood matrix will show a high similarity value, indicating that there is a strong local connection relationship between them. This connection relationship not only depends on numerical proximity, but also considers the complex patterns brought by nonlinear feature interactions. The core significance of constructing the class neighborhood matrix is ​​that it can capture the manifold structure inside the data. The manifold structure refers to the intrinsic geometric shape that may exist in the data in the high-dimensional space, and this shape is often not described by a simple linear method. By calculating the class neighborhood matrix, this complex geometric structure can be converted into a graph form, where each node represents a healthy examinee and the weight of the edge reflects the similarity or distance between nodes. This method is particularly suitable for processing high-dimensional, complex data sets that may have nonlinear relationships. In practical applications, the neighborhood matrix provides a key basis for subsequent spectral analysis.

[0062] Then, based on the set of implicit coding vectors of the query-alternative test data decision point states, the query-alternative test data decision point state class matrix is ​​calculated. This process can be expressed as: in, represents the first on the diagonal of the query-alternative test data decision point state class matrix The eigenvalues ​​at the positions, represents the query-alternative test data decision point state class matrix, , Indicates the first and second diagonal values ​​in the query-alternative test data decision point state class matrix The feature value at each position.

[0063] It should be understood that the process of calculating the degree matrix is ​​to quantify the connection relationship between the implicit encoding vectors of the query-alternative test data decision point states, and evaluate the "importance" and "centrality" of each node (i.e., individual) in the entire network. The degree matrix is ​​a diagonal matrix that records the number of connections to each node, that is, the strength of the connection between each individual and other individuals. In the application scenario of health examination data, this means that the relative position and influence of each individual in its group can be quantified. For example, if an individual's health data is highly similar to that of most other individuals, then its value in the degree matrix will be higher, indicating that it is in a more "central" position in the network structure. On the contrary, if an individual's data is significantly different from that of other individuals, then its degree is lower, which may mean that it is a potential outlier or outlier. By constructing the degree matrix, the structural information inside the data can be seen more clearly. For example, in a healthy group, the physiological indicators of most individuals may show certain similarities, forming a tightly connected core network. Individuals who deviate from the normal range will show lower connectivity, which helps to identify individuals who may have health problems. This approach not only helps spot outliers, but can also reveal underlying subpopulation structures, i.e. small groups with similar health characteristics.

[0064] Then, based on the query-alternative test data decision point state class neighborhood matrix and the query-alternative test data decision point state class degree matrix, the query-alternative test data decision point state Laplace matrix is ​​calculated. This process can be expressed as: in, Represents the query-alternative test data decision point state Laplacian matrix.

[0065] The query-candidate test data decision point state Laplace matrix is ​​spectrally decomposed to obtain a set of query-candidate test data decision point core component encoding vectors. This process can be expressed as: in, represents the spectral decomposition, represents the set of core component encoding vectors of the query-alternative test data decision points, , , Indicates the first, second and query-alternative test data decision point core component encoding vector, represents the diagonal matrix after spectral decomposition, represents a diagonal matrix, , , The first, second, and third diagonal columns of the spectral diagonal matrix are feature values.

[0066] It should be understood that based on the query-alternative test data decision point state class neighborhood matrix and the query-alternative test data decision point state class degree matrix. The Laplace matrix is ​​a core tool in spectral graph theory, which converts the structural information of the graph into an algebraic representation. The eigenvalues ​​and eigenvectors of the Laplace matrix contain important information about the data distribution, such as frequency components and basis functions. Through further spectral decomposition, the original data can be re-examined from a low-dimensional perspective, the most representative features can be extracted, and the computational complexity can be reduced. In practical applications, the calculation of the Laplace matrix and the subsequent spectral decomposition process are of great significance. For example, when identifying outliers or outliers, the Laplace matrix can help determine which individuals are significantly different from most other individuals. In a healthy population, some individuals may be far away from other individuals in the feature space due to potential health problems or undiscovered risk factors. Through the Laplace matrix and its spectral decomposition, these individuals can be more easily identified because their connection relationships show obvious deviations. In addition, the Laplace matrix provides a basis for constructing low-dimensional embeddings. Through spectral decomposition, the eigenvectors corresponding to the smallest eigenvalues ​​can be extracted. These eigenvectors are regarded as capturing the low-dimensional embedding representation of the data manifold and representing the main structural direction of the data. This low-dimensional embedding not only retains the key structural information of the data, but also significantly reduces the computational complexity, alleviates the dimensionality curse problem, and improves the efficiency and robustness of query analysis.

[0067] Finally, the set of query-candidate test data decision point core component encoding vectors is adaptively fused to obtain the query response encoding vector in the candidate test data set. This process can be expressed as: in, Indicates query-alternative test data decision point core component encoding vector, Indicates modulation vector, Indicates A weight matrix, Indicates A bias vector, represents the normalized exponential function, Indicates Normalized query-alternative test data decision point core component encoding factors, represents a predetermined threshold, represents the mask function, Indicates Normalized decision fusion factor, represents adaptive fusion, Represents the query response encoding vector in the candidate test data set.

[0068] It should be understood that the query-alternative test data decision point core component encoding vectors are extracted from the spectral decomposition of the Laplacian matrix. These vectors represent the main structural directions of the data manifold and capture the key features of the data. However, a single core component encoding vector may not fully reflect all important information, so they need to be fused to better utilize the unique contribution of each vector. The core of adaptive fusion lies in integrating the complementary information of these core component encoding vectors. For example, in health examination data, the physiological indicators of some individuals may be normal in one dimension, but show potential risks in another dimension. Through adaptive fusion, the information of these different dimensions can be comprehensively considered to more comprehensively evaluate the health status of each individual. Specifically, the adaptive fusion method can be implemented through an attention mechanism or a gating mechanism. The attention mechanism allows the model to learn the contribution or importance of each core component encoding vector to a specific query task and dynamically assign weights. This means that for each query vector, the system can focus on the most relevant feature components and suppress noise or irrelevant information according to its specific needs. This flexibility enables the system to generate the optimal query response encoding vector for different query inputs and state changes. Through adaptive fusion, the embedding encoding vectors of these individuals can be compared with the query vector, and the weights can be dynamically adjusted according to the performance of each individual in different feature dimensions. In this way, individuals that show abnormalities or potential risks in certain key indicators will be given higher weights, making the final query response encoding vector more discriminative and robust. In addition, adaptive fusion is not just a simple weighted summation, but a learning context-dependent feature combination strategy. This approach enables the dynamic query response encoding vector to best adapt to different query inputs and state changes.

[0069] In the process of determining the reference interval of each data subset of the test data of the candidate healthy examinees based on the data distribution description results of the data subset of the test data of each candidate healthy examinee by using the big data algorithm through the reference interval determination module 60, the reference interval of each data subset of the test data of the candidate healthy examinees is determined by using the following big data algorithms to obtain multiple reference intervals: EP28 non-parametric method, EP28 parametric method, refineR algorithm, TMC algorithm, Kosmic algorithm.

[0070] In the EP28 nonparametric method, it does not depend on the specific distribution form of the data. n The test data of the candidate healthy physical examination subjects are arranged in ascending order and recorded as:x 1. x 2. x 3… x n ; x 1≤ x 2≤ x 3≤…≤ x n ,in, x 1 for n The minimum value among the test data of the candidate healthy examinees, x n for n The maximum value among the test data of the candidate healthy examinees. n The test data of the candidate healthy examinees are divided into 100 equal parts, and the number corresponding to the r% rank is the r The reference lower limit and the reference upper limit of the reference interval can be ranked by P and P, respectively. 2.5 and P 97.5 express. r =0.025×( n +1), r =0.975×( n +1), if 0.025×( n +1) and 0.975×( n +1) is not an integer, so 0.025×( n +1) and 0.975×( n +1) Round to the nearest integer.

[0071] In the EP28 parameter method, if the data subsets of the test data of each candidate healthy examiner obey the normal distribution, or the data subsets of the test data of each candidate healthy examiner obey the normal distribution after transformation, the data subset mean ±1.96s of the test data of each candidate healthy examiner represents a distribution range of 95%, or the data subset mean ±2.58s of the test data of each candidate healthy examiner represents a distribution range of 99%.

[0072] The refineR algorithm can be divided into three steps: data preprocessing, model optimization, and reference interval derivation. In the process of preprocessing the data subset of the candidate healthy examinee's test data, the refineR algorithm can identify the main peak of the mixed distribution, obtain the search area of ​​the Box-Cox transformation and the normal distribution parameters (λ, μ, σ), and calculate the histogram within the given concentration range around the peak. After preprocessing the data subset of the candidate healthy examinee's test data, the optimal values ​​λ, μ, σ are determined by multi-layer grid search to find the optimal model of the data subset of the candidate healthy examinee's test data. Finally, the non-pathological distribution is determined by the optimized model, and the reference interval is derived based on the non-pathological distribution.

[0073] The TMC algorithm can be divided into six steps: (1) select an age / sex stratum for analysis; (2) draw a stratified histogram; (3) obtain an initial estimate of the power normal distribution (PND) parameter λ and derive μ and σ from a series of quantile-quantile (QQ) plots; (4) obtain improved estimates of the PND parameters through the TMC procedure for different candidate cutoff values; (5) find the best PND parameter from the candidate parameters considered in step (4); and (6) calculate the reference interval based on the best PND parameter.

[0074] The Kosmic algorithm makes no assumptions about the distribution of pathological samples and minimizes the Kolmogorov-Smirnov distance between the estimated normal distribution and the truncated portion of the observed distribution of the test results after the Box-Cox transformation. The Kolmogorov-Smirnov test is a well-established normality test that numerically optimizes the parameters of the normal distribution (µ, σ), the Box-Cox transformation parameter (λ), and the truncation interval T.

[0075] like Figure 4 and Figure 5 As shown, in the process of constructing a continuous reference interval based on the multiple reference intervals through the continuous reference interval construction module 70, the multiple reference intervals are processed using the GAMLSS model based on the multiple reference intervals to obtain the continuous reference interval, thereby achieving a natural, continuous and smooth transition between age groups.

[0076] Specifically, the GAMLSS model is able to flexibly model the location (mean) and scale (variance) parameters of the data distribution at the same time. More specifically, GAMLSS allows the use of smooth functions to describe the changes of these parameters with covariates (e.g., age), rather than simply assuming that they are constant or linear.

[0077] In the traditional reference interval construction, a separate reference interval is usually established for each predefined age group (e.g., 0-1 years old, 1-5 years old, 5-10 years old, etc.). The problem with this approach is that the boundaries between adjacent age groups may appear abrupt and lack continuity, and may not accurately reflect the actual situation of the continuous change of physiological parameters with age. In addition, this segmented reference interval may also lead to diagnostic uncertainty at critical age points.

[0078] The GAMLSS model can capture the nonlinear changes in test results as age increases by introducing smooth functions, such as spline functions, which means that for any given age, an estimated reference interval can be obtained. This interval is based on the information of all available data points and reflects the complex relationship between age and test results. In this way, when age increases, the upper and lower limits of the reference interval will gradually change to form a smooth curve instead of a sudden jump.

[0079] Below, a specific example is used to illustrate the mechanism of the GAMLSS model in constructing continuous reference intervals.

[0080] The reference intervals of platelet concentrations in women of different age groups were processed using the GAMLSS model to obtain continuous reference intervals of platelet concentrations in women. A statistical distribution family suitable for describing the distribution of platelet concentrations was selected. For platelet counts, the normal distribution may be appropriate, but if the data presents skewness or other non-normal characteristics, other distributions, such as the Gamma distribution or the Log-Normal distribution, may also be considered. Then, the gamlss package in the R language was used for modeling. Smoothing functions were used to capture the trend of platelet concentration changing with age. Other covariates may also be added to the model, such as whether or not the individual is in a specific physiological stage (represented by a binary variable), to further refine the model.

[0081] Through the GAMLSS model, the platelet concentration distribution parameters at each age point (for example, mean μ and standard deviation σ) can be obtained; based on these parameters, a 95% reference interval and a 90% confidence interval can be calculated. This reference interval is not fixed, but changes continuously with age, forming a natural, continuous and smooth transition; this means that for any given age, an accurate reference interval can be provided that reflects the true distribution of platelet concentration in women of that age group.

[0082] In one embodiment of the present application, Figure 6As shown, the system 100 for constructing reference intervals of test items based on gender and age also includes a consistency evaluation module 80 and a pass rate verification module 90. The consistency evaluation module 80 is used to evaluate the consistency of the results of different construction algorithms of the reference intervals of each of the data subsets of the candidate healthy examinee test data; the pass rate verification module 90 is used to verify the pass rate of the reference intervals of each of the data subsets of the candidate healthy examinee test data.

[0083] Accordingly, the gender- and age-based test item reference interval construction system 100 can evaluate the consistency of results of different construction algorithms of the reference intervals of each data subset of the alternative healthy examinee test data through the consistency evaluation module 80, and then verify the pass rate of the reference intervals of each data subset of the alternative healthy examinee test data through the pass rate verification module 90.

[0084] In the process of evaluating the consistency of the reference intervals of the data subsets of the test data of each of the alternative healthy examinees through the consistency evaluation module 80, the bias degree of the reference interval established by other methods is calculated based on the reference interval established by the EP28 non-parametric method; when the bias degree of the reference interval established by other methods is within a preset range, the consistency of the reference interval is determined to be good; when the bias degree of the reference interval established by other methods is not within the preset range, the consistency of the reference interval is determined to be poor.

[0085] In a specific example of the present application, the bias of the reference interval established by the nonparametric method is calculated based on the reference interval established by other methods. Specifically, the reference value is calculated by the following formula: in, SD RI Indicates the base value; UL 0 represents the upper limit of the reference interval established based on the EP28 nonparametric method; LL 0 represents the lower limit of the reference interval established by the EP28 nonparametric method. The degree of bias of the lower limit of the reference interval established by other methods is calculated by the following formula: in, BR LL Indicates the degree of bias based on the lower limit of the reference interval established by other methods; LL Indicates the lower limit of a reference interval established based on other methods.

[0086] The degree of bias based on the upper limit of the reference interval constructed by other methods was calculated by the following formula: in, BR UL Indicates the degree of bias based on the upper limit of the reference interval established by other methods; UL Indicates the upper limit of a reference interval established by other methods. BR UL or BR LL When it is greater than 0.375, it means that the reference interval established by other methods has a large degree of bias and the consistency between different algorithms is poor; BR UL and BR LL If it is less than or equal to 0.375, it means that the bias of the reference interval established by other methods is small and the consistency between different algorithms is good.

[0087] In one embodiment of the present application, the reference interval data processing system for the test items based on gender and age further includes step: S190, verifying the reference interval. When the reference interval meets the preset conditions, the reference interval passes the verification; when the reference interval does not meet the preset conditions, the reference interval fails the verification.

[0088] In the process of verifying the pass rate of the reference interval of the data subset of each of the candidate healthy examinees' test data through the pass rate verification module 90, the test data acquisition module is used to obtain the physical examination data of the corresponding test items in another time period as verification data, and group verification is performed according to the grouping of the reference intervals. If the pass rate of the verification data (i.e., the probability of matching the continuous reference interval) is greater than or equal to 90%, it means that the verification is passed, and the verification results are displayed through visual pictures.

[0089] In summary, the reference interval data processing system for test items based on gender and age according to the embodiment of the present application is explained. The reference interval data processing system for test items based on gender and age groups the test data according to gender and age, and establishes reference intervals corresponding to each age group of different genders, which can improve the accuracy and suitability of the reference intervals to a certain extent. The collection of test data of healthy examinees in the reference interval data processing system for test items based on gender and age comes from the medical records of patients in the real world. The source and use of the medical records are legal and compliant. In this way, on the one hand, the data source is reliable, and on the other hand, there is no need to specially recruit healthy examinees for testing in order to construct a reference interval experiment, which can improve feasibility and reduce costs to a certain extent.

[0090] The present application and its implementation methods are described above, and such description is not restrictive. The drawings show only one implementation method of the present application, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it and design structural methods and embodiments similar to the technical solution without creative design without departing from the inventive purpose of the present application, they should all fall within the protection scope of the present application.

Claims

1. A system for constructing reference intervals for test items based on gender and age, characterized in that: include: The test data collection module is used to obtain the collection of test data of healthy examinees from the laboratory information system; A deduplication module, used for deduplicating the set of the examination data of the healthy physical examinees based on age, gender and name or medical record number to obtain a set of the examination data of the healthy physical examinees after deduplication; A data screening module, for processing the set of healthy physical examinee test data after deduplication based on inclusion and exclusion criteria to obtain a set of candidate healthy physical examinee test data; A grouping module, used for grouping the set of the candidate healthy physical examinee test data based on age and gender to obtain a grouping result, wherein the grouping result includes a plurality of data subsets of the candidate healthy physical examinee test data; A data distribution description module, used for performing outlier processing and data distribution description on the data subsets of the test data of each candidate healthy physical examinee in the grouping result to obtain a plurality of data distribution description results; A reference interval determination module, for determining the reference interval of each of the candidate healthy physical examinee's test data subsets using a variety of big data algorithms based on the data distribution description results of each of the candidate healthy physical examinee's test data subsets; as well as The continuous reference interval construction module is used to construct a continuous reference interval based on gender, age and the multiple reference intervals.

2. The system for constructing reference intervals for test items based on gender and age according to claim 1, characterized in that: The continuous reference interval construction module is further used to process the multiple reference intervals using the GAMLSS model to obtain the continuous reference interval.

3. The system for constructing reference intervals of test items based on gender and age according to claim 2, characterized in that: Also includes: A consistency evaluation module, used for evaluating the consistency of the results of different construction algorithms of the reference intervals of the data subsets of the test data of the candidate healthy examinees; as well as The pass rate verification module is used to verify the pass rate of the reference interval of each data subset of the test data of the candidate healthy physical examinees.

4. The system for constructing reference intervals for test items based on gender and age according to claim 1, characterized in that: The deduplication module is further used to retain only the most recent examination data of each healthy examinee in the set of examination data of healthy examinees.

5. The system for constructing reference intervals of test items based on gender and age according to claim 1, characterized in that: The data screening module is further used to exclude abnormal healthy examinee test data that affects the reference interval of the test items from the set of healthy examinee test data after deduplication based on pre-set inclusion and exclusion criteria.

6. The system for constructing reference intervals of test items based on gender and age according to claim 1, characterized in that: The grouping module is further used to group the set of candidate healthy examinee test data based on age and gender using the Harris & Boyd method or the standard deviation ratio method.

7. The system for constructing reference intervals of test items based on gender and age according to claim 1, characterized in that: The data distribution description module is further used to perform distribution description on the test data of the candidate healthy examinees to obtain the multiple data distribution description results, wherein the manner of performing distribution description on the test data of the candidate healthy examinees is selected from at least one of the following manners: probability plot, quantile plot, kernel density plot, Shapiro-Wilk test, Kolmogorov-Smilov test; and eliminating outliers of data subsets of the test data of each candidate healthy examinee in the grouping results to obtain the test data of the candidate healthy examinee after outlier processing.

8. The system for constructing reference intervals for test items based on gender and age according to claim 1, characterized in that: The reference intervals of the data subsets of the test data of each candidate healthy examinee are determined by using the following big data algorithms to obtain multiple reference intervals: EP28 non-parametric method, EP28 parametric method, refineR algorithm, Kosmic algorithm, TMC algorithm.

9. The system for constructing reference intervals for test items based on gender and age according to claim 3, characterized in that: The consistency assessment module is further used to calculate the bias ratio of the reference intervals established by the other algorithms based on the reference intervals established by the EP28 non-parametric method.

10. The system for constructing reference intervals of test items based on gender and age according to claim 3, characterized in that: The pass rate verification module is further used to obtain physical examination data of test items in another time period from the laboratory information system as verification data to verify the reference interval; wherein, when the pass rate of the reference interval is ≥90%, the reference interval passes the verification; otherwise, the reference interval fails the verification.

Images