A smart storage and management method for clinical data of internal medicine nursing

By building clinical data space and quantifying testing projects, obtaining information value scales and selecting recommended indicators, and selecting classic cases, the problem of redundant and difficult conditions in internal medicine nursing data is solved, and high-quality data storage and scientific research are achieved value.

CN118737485BActive Publication Date: 2024-11-08自贡市第一人民医院
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411215069.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-08
Estimated Expiration
2044-09-02

AI Technical Summary

Technical Problem

In the clinical data storage of internal medicine nursing, due to the high similarity of patients in the same department, there is too much redundant data in the preferred data, and some difficult condition data account for a small proportion, resulting in difficult condition being unable to be effectively selected, thereby reducing scientific research the value of the data.

Method used

By building clinical data space, quantifying the numerical and text information of the test items, obtaining the information value scale and selecting recommendation indicators of each test item, selecting classic cases based on the optimal coefficient, and building a classic case database to realize the intelligent storage of data.

Benefits of technology

It improves the analyticity of the data, accurately reflects classic and rare cases in the department, reduces redundant cases, and improves the quality of classic case databases and the value of scientific research.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118737485B_ABST
    Figure CN118737485B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of clinical data storage, and specifically to a method for intelligent storage and management of clinical data of internal medicine nursing, including: constructing a clinical data space based on the patient's case; obtaining the information value scale of each test item after quantifying all test items; obtaining the selection recommendation index of each test item of each patient according to the polarized distribution of the results of the patient's test items and the universal distribution; obtaining the selection recommendation index of all test items of the same patient and the information value scale corresponding to the test items, and obtaining the optimization coefficient of each patient; selecting classic patient cases according to the optimization coefficient to construct a classic case database, and realizing the intelligent storage of clinical data. The present invention aims to solve the problem that when the classic case database is constructed by optimizing the case, the classic cases caused by redundant cases cannot be selected, so as to reduce the storage amount and improve the quality of the cases.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of clinical data storage, and in particular to an intelligent storage management method for internal medicine nursing clinical data. Background Art

[0002] The recording of clinical data of internal medicine nursing provides medical staff with detailed information and medical history of patients, which helps to make scientific and systematic clinical decisions and achieve long-term tracking and analysis of patients' condition changes and treatment effects. The long-term accumulation and analysis of clinical data of internal medicine nursing provides a valuable data source and medical education value for scientific research in medical care.

[0003] When storing clinical data, a database is usually built for the data of each department and then associated to achieve the storage of clinical data. However, when the data in the preferred database is used as the data basis for scientific research, the highly similar conditions of most patients in the same department result in too much identical redundant data in the preferred data, and the data of some difficult conditions account for a small proportion, which is affected by a large amount of highly similar clinical data. As a result, the base number is too large during the selection process, resulting in the inability to select difficult conditions, which in turn reduces the upper limit of the difficulty and universality of the data for scientific research, and fails to achieve actual educational value. Summary of the invention

[0004] The present invention provides an intelligent storage and management method for clinical data of internal medicine nursing, so as to solve the existing problem that the highly similar conditions of most patients in the same department result in too much identical redundant data in the optimized data, and the data of some difficult conditions account for a small proportion, so the selection proportion is affected by a large amount of highly similar clinical data, resulting in the base number being too large during optimization, resulting in the problem that difficult conditions cannot be optimized.

[0005] The present invention provides a method for intelligent storage and management of clinical data of internal medicine nursing using the following technical solutions:

[0006] An embodiment of the present invention provides a method for intelligent storage and management of clinical data of internal medicine nursing, the method comprising the following steps:

[0007] Construct a clinical data space based on the patient's case, wherein the horizontal axis of the clinical data space is a result list of the test items, and the vertical axis is the patient number;

[0008] After quantifying all test items in the clinical data space, the information value scale of each test item is obtained;

[0009] According to the polarized distribution and universality distribution of the test items of the patients, the selection recommendation index of each test item of each patient is obtained;

[0010] Based on the selection recommendation indicators of all test items for the same patient and the information value scale corresponding to the test items, the selection coefficient of each patient is obtained;

[0011] Classic patient cases are selected according to the optimization coefficient to build a classic case database to achieve intelligent storage of clinical data.

[0012] Preferably, the specific steps of acquiring the clinical data space include:

[0013] For the test items in the case that are numerical values ​​in the identity information and patient test values, the data of all patients in the same test item are normalized to obtain the standardized results of each patient in each test item in the format of numerical values;

[0014] For the test items in the case that are text information in the identity information and the description of the condition, the Jieba word segmentation algorithm is used to segment all the text test items of all patients to obtain several word segments;

[0015] The Word2vec algorithm model was used to construct vectors for all word segmentation results, and the vectors of each word segmentation for each test item in each patient's case were obtained;

[0016] For each patient's test item, all the word vectors are arranged in the order in which the word vectors appear in the description of the disease condition, and a full-text vector sequence of each test item of each patient is obtained, and the full-text vector sequence of each test item of each patient is used as a standardized result of each test item of each patient in a vector format;

[0017] A clinical data space is constructed, wherein the horizontal axis of the clinical data space is a result list of the test items, the vertical axis is a patient serial number, and each value in the clinical data space is a standardized result of each patient in each test item.

[0018] Preferably, the specific steps of obtaining the information value scale include:

[0019] Record any test item in the clinical data space as a target test item;

[0020] When the standardized result of the target detection item is a numerical value, the distribution degree and the proportion difference of the standardized results of all patients under the target detection item are analyzed to obtain the information difference degree of the target detection item; the distribution degree and the aggregation degree are positively correlated;

[0021] When the standardized result of the target detection project is a vector, the length of the longest full-text vector sequence in the full-text vector sequences of all standardized results under the target detection project is recorded as the standard length, and the full-text vector sequence of each patient under the target detection project is linearly interpolated to obtain the standard length sequence of each patient under the target detection project, and the consistency of the condition description of each patient is obtained based on the similarity between each patient and the standard length sequence of other patients, and the degree of information difference of the target detection project is obtained; the consistency of the condition description is positively correlated with the degree of information difference;

[0022] Obtain the degree of information difference of all test items, and record the variance of the consistency of the condition description of all patients under the target test item as the degree of information difference of the target test item;

[0023] The information difference degree of all test items is inversely normalized, and the inversely normalized result of each test item is recorded as the information value scale of each test item.

[0024] Preferably, the specific method for obtaining the distribution degree and the aggregation degree is:

[0025] The variance of the standardized results of all patients under the target detection item is recorded as the distribution degree of the target detection item;

[0026] The kurtosis of the standardized results of all patients under the target detection item is recorded as the degree of aggregation of the target detection item.

[0027] Preferably, the specific method for obtaining the consistency of the condition description is:

[0028] For the standard length sequence of the i-th patient under the target detection project, the average of the absolute values ​​of the Pearson correlation coefficients of the standard length sequence of the i-th patient and the standard length sequences of all patients in the target detection project except the i-th patient is calculated, and the reciprocal of the average of the absolute values ​​is recorded as the consistency of the condition description of the i-th patient.

[0029] Preferably, the specific steps of selecting and obtaining the recommendation indicator include:

[0030] When the standardized result of the target detection item is a numerical value, the maximum value and the minimum value of the standardized results of all patients under the target detection item are obtained, and the standardized results of all patients under the target detection item are arranged in descending order to obtain the median of the target detection item;

[0031] For the standardized result of the i-th patient under the target detection item, the result after inverse normalization of the difference between the maximum value of the target detection item and the standardized result of the i-th patient under the target detection item is recorded as the first universality coefficient of the i-th patient under the target detection item;

[0032] The result of inversely normalizing the difference between the standardized result of the i-th patient under the target detection item and the minimum value of the target detection item is recorded as the second universality coefficient of the i-th patient under the target detection item;

[0033] The absolute value of the difference between the standardized result of the i-th patient under the target detection item and the median of the target detection item is inversely normalized, and recorded as the third universality coefficient of the i-th patient under the target detection item;

[0034] The largest one among the first universality coefficient, the second universality coefficient and the third universality coefficient is recorded as the selection recommendation index for the i-th patient under the target detection item;

[0035] When the standardized result of the target detection project is a vector, an initial reference vector under the target detection project is obtained according to the vector similarity between the standardized result of each patient under the target detection project and the standardized results of other patients, and a recurrence dictionary of the target detection project is constructed according to the order consistency of the initial reference vectors of different patients under the target detection project;

[0036] Arrange the recurrence times of all vector sequences in the recurrence dictionary of the target detection project in descending order to obtain a recurrence descending sequence, and record the initial reference vectors in all vector sequences between the first vector sequence in the recurrence descending sequence and the first vector sequence in the recurrence descending sequence, which are the two vector sequences with the largest difference in the recurrence times of adjacent vector sequences in the recurrence descending sequence, as the final reference vectors of the target detection project;

[0037] According to the final benchmark vector, the universal preferred parameters of each patient under the target detection item are obtained, and the universal preferred parameters of each patient under the target detection item are used as the numerical values ​​of the standardized results of the target detection item. According to the method of calculating the selection recommendation index when the standardized results of the target detection item are numerical values, the selection recommendation index for each patient is obtained when the standardized results of the target detection item are vectors.

[0038] Preferably, the specific method of obtaining the initial reference vector is:

[0039] The number of occurrences of each vector value in all full-text vector sequences in the standardized results under the target detection project is obtained, and the vector value greater than the preset recurrence threshold in the ratio of the number of occurrences of each vector value under the target detection project to the number of patients is recorded as the initial reference vector under the target detection project.

[0040] Preferably, the specific steps of obtaining the recurring dictionary include:

[0041] Obtain the character separator in the description of the condition of each test item of each patient, use the character separator to segment the full-text vector sequence, and obtain several partial vector sequences in each test item of each patient;

[0042] For all partial vector sequences of all patients under the target detection project, the specific steps of constructing the recurrence dictionary of the target detection project are:

[0043] For the first partial vector sequence of the target detection project, an initial reference vector in the first partial vector sequence is obtained, a new vector sequence is formed according to the order of the initial reference vectors in the first partial vector sequence, and the new vector sequence is added to the initial recurrence dictionary as an element of the dictionary;

[0044] For the next partial vector sequence of the target detection project, the initial reference vector in the next partial vector sequence is obtained, and a new vector sequence is formed according to the order of the initial reference vectors in the next partial vector sequence. When the new vector sequence is in the initial recurrence dictionary, the number of recurrences of the elements in the dictionary corresponding to the new vector sequence is increased by one; when the new vector sequence is not in the recurrence dictionary, the new vector sequence is added to the initial recurrence dictionary as an element of the dictionary;

[0045] And so on, after traversing all partial vector sequences in the target detection project, the initial recurrence dictionary is recorded as the recurrence dictionary of the target detection project, and the recurrence dictionary of the target detection project contains each vector sequence and the number of recurrences of each vector sequence.

[0046] Preferably, the specific method of obtaining the universal optimal parameter is:

[0047] For the i-th patient under the target detection project, the ratio of the number of occurrences of the final benchmark vector in the full-text vector sequence of the i-th patient under the target detection project to the total number of vectors in the full-text vector sequence of the i-th patient under the target detection project is recorded as the universal optimal parameter of the i-th patient under the target detection project.

[0048] Preferably, the specific method of obtaining the optimization coefficient is:

[0049] Based on the selection recommendation index of each patient in each test item and the information value scale of the corresponding test item, the selection coefficient of each patient is obtained; the selection recommendation index and the information value scale are positively correlated.

[0050] The beneficial effects of the technical solution of the present invention are as follows: the present invention constructs a clinical data space for the patient's case, realizes the quantification of the test items in numerical format and the test items in text format, and improves the analyzability of the data; after all the test items are quantified, the information value scale of each test item is obtained, and the information value that each test item can express is obtained from the analysis of a single test item; according to the polarized distribution of the results of the patient's test items and the universal distribution, the selection recommendation index of each test item of each patient is obtained, and based on the information of each patient under the test items of different formats, the data showing universal characteristics and extreme characteristics are selected to realize more accurate reflection of the classic and uncommon cases in the department; based on the selection recommendation index of all test items of the same patient and the information value scale corresponding to the test items, the optimization coefficient of each patient is obtained, and the optimal patient is selected based on the information value of each test item and the universal characteristics that can be reflected; according to the optimization coefficient, the classic patient cases are selected to construct a classic case database, and the intelligent storage of clinical data is realized, so as to achieve the purpose of streamlining the redundant cases in the classic case database, and improve the case quality in the classic case database. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 The present invention is a flowchart of the steps of a method for intelligent storage and management of clinical data of internal medicine nursing. DETAILED DESCRIPTION

[0053] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of an intelligent storage and management method for clinical data of internal medicine nursing proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0054] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0055] The following is a detailed description of a specific solution of an intelligent storage and management method for clinical data of internal medicine nursing provided by the present invention in conjunction with the accompanying drawings.

[0056] See also Figure 1 , which shows a flowchart of a method for intelligent storage and management of clinical data of internal medicine nursing provided by an embodiment of the present invention, the method comprising the following steps:

[0057] Step S001: construct a clinical data space based on the patient's case.

[0058] The purpose of this embodiment is to select some classic cases from the cases of all patients, so as to build a classic case database as the data basis for scientific research. Therefore, it is necessary to obtain the cases of all patients first. Since the conditions described in the cases of various departments are highly similar, the optimization method for cases in different departments is the same. The subsequent analysis of this embodiment is described using the database of one department as an example. The analysis and optimization methods of other departments are the same as the methods of the database analyzed in this embodiment.

[0059] Preferably, this embodiment obtains the medical records of all patients from the department database, each of which includes several test items, and the test items mainly include patient identity information, patient test values, detailed description of the patient's condition, etc. In order to facilitate the subsequent analysis, each test item is taken as a target, and the information of all patients under the same test item is standardized;

[0060] For the test items in the case that are numerical values ​​in the identity information and patient test values, since they are numerical information, this embodiment normalizes the data of all patients of the same test item to obtain standardized results of each patient in each test item. This embodiment uses the maximum and minimum normalization model to normalize the numerical information. Other embodiments may select other normalization models, which are not specifically limited in this embodiment.

[0061] Furthermore, for the test items in the case that are text information and condition descriptions in the identity information, if the conditions in the same department are similar and therefore the diagnosis descriptions are similar, the condition descriptions can be digitized. The more frequent condition descriptions appear in the digitized results, the more common conditions of most patients in the department can be represented, which is conducive to the accurate selection of classic cases in the subsequent selection of medical records. The specific process of digitizing text information is as follows:

[0062] (1) Use the Jieba word segmentation algorithm to segment all text-based test items of all patients to obtain a number of word segments;

[0063] (2) Use the Word2vec algorithm model to construct vectors for all word segmentation results, and obtain the vector of each word segmentation for each test item in each patient's case;

[0064] (3) For each patient, all the word vectors in each test item are arranged in the order in which the word vectors appear in the description of the disease condition to obtain the full-text vector sequence of each test item for each patient, and the full-text vector sequence of each test item for each patient is used as the standardized result of each test item for each patient;

[0065] (4) Furthermore, character separators such as commas, periods, and semicolons are obtained from the description of the condition of each test item of each patient, and the full-text vector sequence is segmented using the character separators to obtain several partial vector sequences in each test item of each patient.

[0066] It should be noted that the Jieba word segmentation algorithm and Word2vec algorithm model described in this embodiment are existing well-known technologies and will not be described in detail in this embodiment.

[0067] At this point, the standardized results of numerical information and textual information in the test items of all patients have been obtained. All test items of all patients constitute a clinical data space. The horizontal axis of the clinical data space is the result list of the test items, and the vertical axis is the patient serial number. Each value in the clinical data space is the standardized result of each patient in each test item.

[0068] Step S002: After quantifying all test items in the clinical data space, obtain the information value scale of each test item.

[0069] It should be noted that when constructing a classic case data set through case optimization, in order to make the data set meet the universal characteristics as much as possible, that is, the preferred cases include the same manifestations of most cases in the department, and also need to include the special manifestations of difficult and complicated diseases in the department. Therefore, it is necessary to refer to the test results of different test items to show whether the condition is normal or abnormal, and then select cases with polarized characteristics to construct a classic case data set.

[0070] It should be further explained that since the case contains a large amount of information that is not related to the condition, such as the patient's identity information, and different test items have different reference values ​​for scientific research, it is usually necessary to manually select some test items as the basis for judging common conditions and difficult and complicated conditions. However, this method has subjective factors that make it less objective, and some test items also have potential reference value. Therefore, this embodiment obtains the information value scale of each test item through the distribution of standardized results of all patients under the same test item.

[0071] In some implementations of this embodiment, the information value scale of each test item is obtained by the distribution of standardized results of all patients under the same test item, including the following specific methods:

[0072] It should be noted that for test items with information value, since the descriptions are all descriptions of the disease conditions with the same characteristics, the standardized results are approximately distributed and continuous in distribution. However, for test items with very little information value, since all test results are standardized during the standardization process, the similarity between the standardized results of numerical or textual information is very small, which is reflected in the fact that the distribution range of the standardized results is very large and the number of the same standardized results is very small. Therefore, this embodiment quantifies the distribution range of the test items according to the difference in the standardized results to obtain the degree of information difference of each test item.

[0073] Preferably, according to the difference in the standardized results, the distribution range of the test items is quantified to obtain the degree of information difference of each test item. The specific process is:

[0074] Any test item in the clinical data space is recorded as the target test item. When the standardized result of the target test item is a numerical value, the distribution degree and the proportion difference of the standardized results of all patients under the target test item are analyzed to obtain the information difference degree of the target test item; the distribution degree and the aggregation degree are positively correlated. As an example, the information difference degree of the target test item can be obtained through the following example, and the specific acquisition method is:

[0075] The variance of the standardized results of all patients under the target detection item is recorded as the distribution degree of the target detection item;

[0076] The kurtosis of the standardized results of all patients under the target detection item is recorded as the degree of aggregation of the target detection item;

[0077] The product of the distribution degree and the aggregation degree of the target detection items is recorded as the information difference degree of the target detection items.

[0078] Optionally, in other embodiments, after obtaining the standardized results of all patients in all test items, the distribution degree and aggregation degree of the standardized results of all patients under the target detection item are not calculated, and the information entropy of the standardized results of all patients under the target detection item is recorded as the information difference degree of the target detection item.

[0079] It should be noted that the variance, kurtosis and information entropy described in this embodiment are existing public technologies and will not be described in detail in this embodiment.

[0080] When the standardized result of the target detection project is a vector, the length of the full-text vector sequence in the standardized results of all patients under the target detection project is standardized to obtain the standard length sequence of the standardized results of all patients under the target detection project, and the consistency of the description of each patient's condition is obtained based on the similarity between each patient and the standard length sequence of other patients, and the degree of information difference of the target detection project is obtained; the consistency of the condition description is positively correlated with the degree of information difference. As an example, the degree of information difference of the target detection project can be obtained through the following example, and the specific steps are as follows:

[0081] The length of the longest full-text vector sequence among all the standardized results of the target detection project is recorded as the standard length, and the full-text vector sequence of each patient in the target detection project is linearly interpolated to obtain the standard length sequence of each patient in the target detection project;

[0082] Further, for the standard length sequence of the i-th patient under the target detection project, the average of the absolute values ​​of the Pearson correlation coefficients of the standard length sequence of the i-th patient and the standard length sequences of all patients in the target detection project except the i-th patient is calculated, and the reciprocal of the average of the absolute values ​​is recorded as the consistency of the condition description of the i-th patient;

[0083] Furthermore, the consistency of the condition description of all patients under the target detection item is obtained in the same way, and the variance of the consistency of the condition description of all patients under the target detection item is recorded as the information difference degree of the target detection item.

[0084] Optionally, in other embodiments, after obtaining the full-text vector sequence of all standardized results under the target detection project, length normalization is not performed. Taking the full-text vector sequence of the i-th patient as an example, the mean of the cosine similarity between the full-text vector sequence of the i-th patient and the full-text vector sequences of all patients in the target detection project except the i-th patient is calculated, and the inverse is recorded as the consistency of the condition description of the i-th patient, and the sum of the consistency of the condition description of all patients is recorded as the degree of information difference of the target detection project.

[0085] Furthermore, similarly, the information difference degree of all detection items is obtained, and the information difference degree of all detection items is inversely normalized, and the inversely normalized result of each detection item is recorded as the information value scale of each detection item; this embodiment adopts Model to present inverse proportional relationship and normalization, As the input of the model, the implementer can set the inverse proportional function and normalization function according to the actual situation.

[0086] Step S003: obtaining a selection recommendation index for each test item of each patient according to the polarized distribution and universality distribution of the test item results of the patient.

[0087] It should be noted that after analyzing the information that can be expressed by all patients under each test item and calculating the information value scale of each test item, it is necessary to select appropriate patient cases as classic medical records based on the reference value that different patients can provide for disease analysis and scientific research. For the same test item, the selected cases should meet universal characteristics and differential characteristics, specifically including the same conditions of most patients, as well as conditions that are rarer than normal conditions, so as to provide more information during disease analysis and scientific research. Therefore, this embodiment obtains the selection recommendation indicators for each patient under each test item based on the difference between the patient's standardized results under the test item and other patients.

[0088] Preferably, based on the differences in standardized results, similarity and polarization analysis is performed on the standardized results of different patients under the same test item to obtain the selection recommendation index for each patient under each test item. The specific process is:

[0089] Preferably, when the standardized result of the target detection item is a numerical value, the selection recommendation index for each patient under the target detection item is obtained according to the distribution difference between the standardized result of each patient under the target detection item and the standardized results of other patients. As an example, the selection recommendation index when the standardized result of the target detection item is a numerical value can be obtained through the following example, and the specific operation steps are:

[0090] Obtain the maximum and minimum values ​​of the standardized results of all patients under the target detection item, and arrange the standardized results of all patients under the target detection item in descending order to obtain the median of the target detection item;

[0091] For the standardized result of the i-th patient under the target detection item, the difference between the maximum value of the target detection item and the standardized result of the i-th patient under the target detection item is inversely normalized, and recorded as the first universality coefficient of the i-th patient under the target detection item; it is used to represent the difference in the extreme value distribution of the standardized results of the i-th patient under the target detection item and other patients under the same detection item. The larger the value, the closer the i-th patient under the target detection item is to the special case under the detection item;

[0092] The result of inversely normalizing the difference between the standardized result of the i-th patient under the target detection item and the minimum value of the target detection item is recorded as the second universality coefficient of the i-th patient under the target detection item; it is used to represent the difference in extreme value distribution of the standardized results of the i-th patient under the target detection item and other patients under the same detection item. The larger the value, the closer the i-th patient under the target detection item is to the special case under the detection item;

[0093] The absolute value of the difference between the standardized result of the i-th patient under the target detection item and the median of the target detection item is inversely normalized and recorded as the third universality coefficient of the i-th patient under the target detection item; it is used to represent the normal distribution difference between the standardized results of the i-th patient under the target detection item and other patients under the same detection item. The larger the value, the closer the i-th patient under the target detection item is to the most normal case under the detection item;

[0094] It should be noted that this embodiment adopts Model to present inverse proportional relationship and normalization, As the input of the model, the implementer can set the inverse proportional function and normalization function according to the actual situation.

[0095] The largest one among the first universality coefficient, the second universality coefficient and the third universality coefficient is recorded as the selection recommendation index for the i-th patient under the target detection item.

[0096] It should be further explained that for all the descriptive language of the condition under the same department, the basic text of the description is roughly the same, and other language descriptions, such as the details of special cases, are all added to the basic text. Therefore, for the test items whose standardized results are vectors, the same part of the vector exists in different full-text vector sequences, and the order is roughly fixed. Therefore, for the full-text vector sequences of different patients under the same test item, if the proportion of the same part of the full-text vector sequence of any patient and the full-text vector sequence of other patients is smaller, it means that some special descriptions have been added to the description of the patient's condition, and the patient's condition is more likely to be a special case, and the possibility of being selected when selecting classic cases is greater;

[0097] On the contrary, if most of the vector values ​​contained in the full-text vector sequence of any patient are used as components of the full-text vector sequences of other patients, then it means that the case described by the full-text vector sequence of this patient is a classic case in this department, and other cases are further derived on the basis of this case. Therefore, this case needs to be selected as a classic case.

[0098] Preferably, when the standardized result of the target detection project is a vector, the initial reference vector under the target detection project is obtained according to the vector similarity between the standardized result of each patient under the target detection project and the standardized results of other patients, a recurrence dictionary of the target detection project is constructed according to the sequential consistency of the initial reference vectors of different patients under the target detection project, and a final reference vector is obtained according to the recurrence dictionary of the target detection project; the universal preferred parameters of each patient under the target detection project are obtained according to the final reference vector, and the selection recommendation index of each patient under the target detection project is obtained according to the universal preferred parameters. As an example, the selection recommendation index when the standardized result of the target detection project is a vector can be obtained through the following example, and the specific operation steps are as follows:

[0099] Specifically, the number of occurrences of each vector value in all full-text vector sequences in the standardized results under the target detection project is obtained, and the vector value that is greater than the preset recurrence threshold in the ratio of the number of occurrences of each vector value under the target detection project to the number of patients is recorded as the initial reference vector under the target detection project; it should be noted that the recurrence threshold described in this embodiment is described using 0.6 as an example, and other embodiments may set other values, which are not specifically limited in this embodiment.

[0100] The case information described for the standard cases in the department is approximately the same, and the order of the initial reference vectors is the same. Therefore, the final reference vector under the target detection project can be obtained by judging whether the initial reference vectors have the same specific order in all patients of the target detection project.

[0101] Preferably, for all partial vector sequences of all patients under the target detection project, a recurrence dictionary of the target detection project is constructed. The specific method for obtaining the recurrence dictionary of the target detection project is:

[0102] For the first partial vector sequence of the target detection project, an initial reference vector in the first partial vector sequence is obtained, a new vector sequence is formed according to the order of the initial reference vectors in the first partial vector sequence, and the new vector sequence is added to the initial recurrence dictionary as an element of the dictionary;

[0103] For the next partial vector sequence of the target detection project, the initial reference vector in the next partial vector sequence is obtained, and a new vector sequence is formed according to the order of the initial reference vectors in the next partial vector sequence. When the new vector sequence is in the initial recurrence dictionary, the number of recurrences of the elements in the dictionary corresponding to the new vector sequence is increased by one; when the new vector sequence is not in the recurrence dictionary, the new vector sequence is added to the initial recurrence dictionary as an element of the dictionary;

[0104] And so on, after traversing all partial vector sequences in the target detection project, the initial recurrence dictionary is recorded as the recurrence dictionary of the target detection project, and the recurrence dictionary of the target detection project contains each vector sequence and the number of recurrences of each vector sequence.

[0105] For uncommon descriptions, they are very different from standard cases in the department, so the final benchmark vector of the target detection project can be obtained through the polarized distribution of the number of recurrences.

[0106] Furthermore, the number of recurrences of all vector sequences in the recurrence dictionary of the target detection project is arranged in descending order to obtain a recurrence descending sequence, and the initial reference vector in all vector sequences between the first vector sequence in the recurrence descending sequence and the first vector sequence in the recurrence descending sequence, which are the two vector sequences with the largest difference in the number of recurrences of adjacent vector sequences in the recurrence descending sequence, is recorded as the final reference vector of the target detection project.

[0107] Furthermore, for the i-th patient under the target detection project, the ratio of the number of occurrences of the final reference vector in the full-text vector sequence of the i-th patient under the target detection project to the total number of vectors in the full-text vector sequence of the i-th patient under the target detection project is recorded as the universal optimal parameter of the i-th patient under the target detection project;

[0108] The universal optimal parameter of each patient under the target detection item is used as the numerical value of the standardized result of the target detection item. According to the method of calculating the selection recommendation index when the standardized result of the target detection item is a numerical value, the selection recommendation index of each patient is obtained when the standardized result of the target detection item is a vector.

[0109] Similarly, the selection recommendation indicators for each patient under all test items are obtained.

[0110] Step S004: based on the selection recommendation indexes of all the test items of the same patient and the information value scales corresponding to the test items, obtain the optimization coefficient of each patient.

[0111] Preferably, based on the selection recommendation indexes of all test items of the same patient and the information value scale corresponding to the test items, the selection coefficient of each patient is obtained. The specific process is:

[0112] It should be noted that after obtaining the recommended selection indicators for each patient in each test item and the information value scale of the corresponding test item, the greater the maximum necessity of selection among all test items for the same patient, the more standard or abnormal the data of the patient under the test item with the greatest necessity of selection is, so it is more necessary to consider the patient's case when selecting classic cases, analyze the universal characteristics of the same patient under all test items, and combine the test items with the greatest necessity of selection to obtain the optimization coefficient of each patient.

[0113] Preferably, the selection coefficient of each patient is obtained based on the selection recommendation index of each patient in each test item and the information value scale of the corresponding test item; the selection recommendation index and the information value scale are positively correlated. As an example, the selection coefficient of each patient can be obtained through the following example, and the specific operation steps are:

[0114] For the i-th patient, calculate the cumulative sum of the product of the selection recommendation index and the information value scale of the i-th patient under the same test item, and record the cumulative sum and the product of the maximum selection recommendation index among all the test items of the i-th patient as the selection coefficient of the i-th patient.

[0115] Similarly, obtain the merit coefficients of all patients.

[0116] Step S005: Select classic patient cases according to the optimization coefficient to build a classic case database to achieve intelligent storage of clinical data.

[0117] The maximum and minimum values ​​of the optimization coefficients of all patients are normalized, and the result of normalization is the normalized value of the optimization coefficient of each patient. The optimization threshold is preset to 0.8, and the cases of all patients whose normalized optimization coefficient value is greater than the optimization threshold are used as classic medical records. All classic medical records construct the classic case database of the department. By selecting some cases with reference value from all cases to form a classic case database, the redundant data in valuable medical records can be reduced, and the intelligent storage management of internal medicine nursing clinical data can be realized.

[0118] This is the end of the present invention.

[0119] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for intelligent storage and management of clinical data of internal medicine nursing, characterized in that: The method comprises the following steps: Construct a clinical data space based on the patient's case, wherein the horizontal axis of the clinical data space is a list of test item results, and the vertical axis is the patient number; After quantifying all test items in the clinical data space, the information value scale of each test item is obtained; According to the polarized distribution and universality distribution of the test items of the patients, the selection recommendation index of each test item of each patient is obtained; Based on the selection recommendation indicators of all test items for the same patient and the information value scale corresponding to the test items, the selection coefficient of each patient is obtained; Select classic patient cases according to the merit coefficient to build a classic case database to achieve intelligent storage of clinical data; The specific steps of obtaining the clinical data space include: For the test items in the case that are numerical values ​​in the identity information and patient test values, the data of all patients in the same test item are normalized to obtain the standardized results of each patient in each test item in the format of numerical values; For the test items in the case that are text information in the identity information and the description of the condition, the Jieba word segmentation algorithm is used to segment all the text test items of all patients to obtain several word segments; The Word2vec algorithm model was used to construct vectors for all word segmentation results, and the vectors of each word segmentation for each test item in each patient's case were obtained; For each patient's test item, all the word vectors are arranged in the order in which the word vectors appear in the description of the disease condition, and a full-text vector sequence of each test item of each patient is obtained, and the full-text vector sequence of each test item of each patient is used as a standardized result of each test item of each patient in a vector format; Constructing a clinical data space, wherein the horizontal axis of the clinical data space is a result list of the test items, the vertical axis is a patient serial number, and each value in the clinical data space is a standardized result of each patient in each test item; The specific steps of obtaining the information value scale include: Record any test item in the clinical data space as a target test item; When the standardized result of the target detection item is a numerical value, the distribution degree and the proportion difference of the standardized results of all patients under the target detection item are analyzed to obtain the information difference degree of the target detection item; the distribution degree and the aggregation degree are positively correlated; When the standardized result of the target detection project is a vector, the length of the longest full-text vector sequence in the full-text vector sequences of all standardized results under the target detection project is recorded as the standard length, and the full-text vector sequence of each patient under the target detection project is linearly interpolated to obtain the standard length sequence of each patient under the target detection project, and the consistency of the condition description of each patient is obtained based on the similarity between each patient and the standard length sequence of other patients, and the degree of information difference of the target detection project is obtained; the consistency of the condition description is positively correlated with the degree of information difference; Obtain the degree of information difference of all test items, and record the variance of the consistency of the condition description of all patients under the target test item as the degree of information difference of the target test item; The information difference degree of all test items is inversely normalized, and the inversely normalized result of each test item is recorded as the information value scale of each test item; The specific method for obtaining the distribution degree and the aggregation degree is as follows: The variance of the standardized results of all patients under the target detection item is recorded as the distribution degree of the target detection item; The kurtosis of the standardized results of all patients under the target detection item is recorded as the degree of aggregation of the target detection item; The specific method for obtaining the consistency of the condition description is as follows: For the standard length sequence of the i-th patient under the target detection project, calculate the average of the absolute values ​​of the Pearson correlation coefficients of the standard length sequence of the i-th patient and the standard length sequences of all patients in the target detection project except the i-th patient, and take the reciprocal of the average of the absolute values ​​and record it as the consistency of the condition description of the i-th patient; The specific steps of selecting and obtaining the recommended indicators include: When the standardized result of the target detection item is a numerical value, the maximum value and the minimum value of the standardized results of all patients under the target detection item are obtained, and the standardized results of all patients under the target detection item are arranged in descending order to obtain the median of the target detection item; For the standardized result of the i-th patient under the target detection item, the result after inverse normalization of the difference between the maximum value of the target detection item and the standardized result of the i-th patient under the target detection item is recorded as the first universality coefficient of the i-th patient under the target detection item; The result of inversely normalizing the difference between the standardized result of the i-th patient under the target detection item and the minimum value of the target detection item is recorded as the second universality coefficient of the i-th patient under the target detection item; The absolute value of the difference between the standardized result of the i-th patient under the target detection item and the median of the target detection item is inversely normalized, and recorded as the third universality coefficient of the i-th patient under the target detection item; The largest one among the first universality coefficient, the second universality coefficient and the third universality coefficient is recorded as the selection recommendation index for the i-th patient under the target detection item; When the standardized result of the target detection project is a vector, an initial reference vector under the target detection project is obtained according to the vector similarity between the standardized result of each patient under the target detection project and the standardized results of other patients, and a recurrence dictionary of the target detection project is constructed according to the order consistency of the initial reference vectors of different patients under the target detection project; Arrange the recurrence times of all vector sequences in the recurrence dictionary of the target detection project in descending order to obtain a recurrence descending sequence, and record the initial reference vectors in all vector sequences between the first vector sequence in the recurrence descending sequence and the first vector sequence in the recurrence descending sequence, which are the two vector sequences with the largest difference in the recurrence times of adjacent vector sequences in the recurrence descending sequence, as the final reference vectors of the target detection project; Obtaining a universal optimal parameter for each patient under the target detection item according to the final reference vector, using the universal optimal parameter for each patient under the target detection item as the numerical value of the standardized result of the target detection item, and obtaining a selection recommended indicator for each patient when the standardized result of the target detection item is a vector according to a method for calculating the selection recommended indicator when the standardized result of the target detection item is a numerical value; The specific method of obtaining the initial reference vector is: Obtain the number of occurrences of each vector value in all full-text vector sequences in the standardized results under the target detection project, and record the vector value greater than the preset recurrence threshold in the ratio of the number of occurrences of each vector value under the target detection project to the number of patients as the initial reference vector under the target detection project; The specific steps of obtaining the recurring dictionary include: Obtain the character separator in the description of the condition of each test item of each patient, use the character separator to segment the full-text vector sequence, and obtain several partial vector sequences in each test item of each patient; For all partial vector sequences of all patients under the target detection project, the specific steps of constructing the recurrence dictionary of the target detection project are: For the first partial vector sequence of the target detection project, an initial reference vector in the first partial vector sequence is obtained, a new vector sequence is formed according to the order of the initial reference vectors in the first partial vector sequence, and the new vector sequence is added to the initial recurrence dictionary as an element of the dictionary; For the next partial vector sequence of the target detection project, the initial reference vector in the next partial vector sequence is obtained, and a new vector sequence is formed according to the order of the initial reference vectors in the next partial vector sequence. When the new vector sequence is in the initial recurrence dictionary, the number of recurrences of the elements in the dictionary corresponding to the new vector sequence is increased by one; when the new vector sequence is not in the recurrence dictionary, the new vector sequence is added to the initial recurrence dictionary as an element of the dictionary; And so on, after traversing all partial vector sequences in the target detection project, the initial recurrence dictionary is recorded as the recurrence dictionary of the target detection project, and the recurrence dictionary of the target detection project contains each vector sequence and the number of recurrences of each vector sequence; The specific method of obtaining the universal optimal parameters is: For the i-th patient under the target detection project, the ratio of the number of occurrences of the final benchmark vector in the full-text vector sequence of the i-th patient under the target detection project to the total number of vectors in the full-text vector sequence of the i-th patient under the target detection project is recorded as the universal optimal parameter of the i-th patient under the target detection project.

2. According to claim 1, a method for intelligent storage and management of clinical data of internal medicine nursing, characterized in that: The specific method for obtaining the optimization coefficient is as follows: Based on the selection recommendation index of each patient in each test item and the information value scale of the corresponding test item, the selection coefficient of each patient is obtained; the selection recommendation index and the information value scale are positively correlated.