Rare disease CRF form generation method and device

Abstract

The invention provides a method and a device for generating a CRF form of a rare disease, in the application, when the CRF form of the rare disease is needed, a user can input keywords of the rare disease through a user terminal, and a medical report database is arranged in a server, so that a medical report can be sent to the user terminal; a medical report stored in a medical report database contains items which need to be filled in a CRF form corresponding to related diseases, such as basic information of a patient, past medical history, a personal physique detection result, a medical detection result, an image detection result, opinions and suggestions of a doctor and the like; therefore, various items capable of establishing the CRF form can be obtained from the medical reports corresponding to the rare disease, the medical reports with high relevancy with the keywords are determined by comparing the items with the medical words recorded in the medical dictionary, and then the CRF form related to the current rare disease is conveniently established. And the user can obtain the corresponding CRF form only by inputting the keyword, so that the manual workload is reduced.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and device for generating a CRF form for a rare disease. Background Art

[0002] Medical CRF (Case Report Form) form configuration usually customizes specific CRF form content according to the research needs of researchers such as clinical physicians, public health physicians, pharmacists, and health economists. For rare diseases, there is both homogeneity and heterogeneity between different diseases or different disease subcategories. Therefore, the data collection form content required for different diseases or different disease subcategories is also different. In this context, the customized CRF form lacks sufficient flexibility. When the research content or research disease changes, the customized CRF form needs to be manually modified, resulting in a large amount of manual workload. Summary of the invention

[0003] In view of this, an embodiment of the present application provides a method and device for generating a CRF form for a rare disease, so as to provide corresponding CRF forms according to different rare diseases, thereby reducing manual workload.

[0004] In a first aspect, an embodiment of the present application provides a method for generating a CRF form for a rare disease, the generating method being applied to a form generating system, the form generating system comprising a user terminal and a server, the server being provided with a medical report database and a medical dictionary, the generating method running on the server, the generating method comprising: In response to a keyword input by a user at the user terminal, frequency statistics are performed on each medical report in the medical report database to determine the candidate words with the highest frequency in each medical report; For each medical term and each to-be-compared term in the medical dictionary, construct a (m+1)*(n+1) matrix according to the first character string length of the medical term and the second character string length of the to-be-compared term, wherein m is the first character string length, n is the second character string length, and the to-be-compared term includes the keyword and the alternative term; Initialize the matrix elements of the first row of the matrix to integers from 0 to n in sequence, initialize the matrix elements of the first column of the matrix to integers from 0 to m in sequence, and determine the other matrix elements Tij in the matrix according to preset rules; wherein the preset rules include: if the i-th character of the medical term is the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the matrix element in the upper left matrix cell adjacent to the matrix cell where the matrix element Tij is located; if the i-th character of the medical term is not the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the sum of the smallest matrix elements in the three matrix cells adjacent to the upper left corner, the upper side and the left side of the matrix cell where the matrix element Tij is located and the value 1; After determining the matrix elements in the matrix cells in the matrix, the similarity between the medical term and the term to be compared is calculated using the following formula: S=1-B / A; Wherein, B is a matrix element in a matrix cell located at the lower right corner of the matrix, and A is the maximum value of m and n; After obtaining the first similarities between each candidate word and each medical word and the second similarities between the keyword and each medical word, determining a target medical word corresponding to the first similarity with the smallest difference from the target similarity among the first similarities between each candidate word and the target medical word, wherein the target medical word is the medical word corresponding to the largest first similarity of each candidate word, and the target similarity is the largest second similarity; A CRF form is constructed according to the words with matching numerical values ​​in the titles of each level in the target medical report corresponding to the target medical word and the text contents corresponding to each level of the titles.

[0005] Optionally, the step of constructing a CRF form according to the words with numerical values ​​matched in the titles of each level in the target medical report corresponding to the target medical word and the text contents corresponding to each level of the titles includes: After obtaining the titles of each level in the target medical report, segmenting the titles of each level to obtain candidate segmented words included in the titles of each level; For each candidate participle, after determining the part of speech of the candidate participle, the distribution ratio of the part of speech in various parts of speech in the medical field is determined as the weight of the candidate participle; The similarity between the candidate segmentation word and the preset medical field word is calculated using the following formula: S= ; Among them, K is the first word vector corresponding to the candidate word segmentation, and L is the second word vector corresponding to the preset medical field term. is the modulus of the first word vector, is the modulus of the second word vector, and d is the weight; For the text content corresponding to each level of title, text processing is performed on the text content to obtain text segmentation and dependency relationships between the text segmentation; According to the part of speech of each text segmentation, determine, from each text segmentation, a candidate text segmentation having a dependency relationship with a text segmentation whose part of speech is a numeral; Selecting a specified type of part of speech from the candidate text segmentations as a target text segmentation; The candidate segmentations with similarity higher than a preset threshold are taken as parent items, and for each parent item, the target text segmentations included in the text content under the title corresponding to the parent item are taken as child items to construct the CRF form.

[0006] Optionally, the generating method further includes: The CRF form is displayed on the user terminal.

[0007] Optionally, the user terminal includes a mobile terminal and a fixed terminal.

[0008] Optionally, the generating method further includes: In response to the user's adding operation, the item corresponding to the adding operation is added to the position specified by the adding operation.

[0009] In a second aspect, an embodiment of the present application provides a device for generating a CRF form for a rare disease, the generating device being applied to a form generating system, the form generating system comprising a user terminal and a server, the server being provided with a medical report database and a medical dictionary, the generating device being configured at the server, the generating device comprising: A statistical unit, configured to respond to a keyword input by a user at the user terminal, perform word frequency statistics on each medical report in the medical report database, and determine the candidate word with the highest word frequency in each medical report; A construction unit is used to construct, for each medical word and each to-be-compared word in the medical dictionary, a (m+1)*(n+1) matrix according to a first character string length of the medical word and a second character string length of the to-be-compared word, wherein m is the first character string length, n is the second character string length, and the to-be-compared word includes the keyword and the alternative word; A first determination unit is used to initialize the matrix elements of the first row of the matrix to integers from 0 to n, initialize the matrix elements of the first column of the matrix to integers from 0 to m, and determine other matrix elements Tij in the matrix according to preset rules; wherein the preset rules include: if the i-th character of the medical term is the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the matrix element in the upper left matrix cell adjacent to the matrix cell where the matrix element Tij is located; if the i-th character of the medical term is not the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the sum of the smallest matrix elements in the three matrix cells adjacent to the upper left corner, the upper side and the left side of the matrix cell where the matrix element Tij is located and the value 1; The calculation unit is used to calculate the similarity between the medical term and the term to be compared using the following formula after determining the matrix elements in the matrix cells in the matrix: S=1-B / A; Wherein, B is a matrix element in a matrix cell located at the lower right corner of the matrix, and A is the maximum value of m and n; A second determination unit is used to determine, after obtaining the first similarities between each candidate word and each medical word and the second similarities between the keyword and each medical word, a target medical word corresponding to the first similarity with the smallest difference from the target similarity among the first similarities between each candidate word and the target medical word, wherein the target medical word is the medical word corresponding to the largest first similarity of each candidate word, and the target similarity is the largest second similarity; The generating unit is used to construct a CRF form according to the words with numerical values ​​matched in the titles of each level in the target medical report corresponding to the target medical word and the text content corresponding to each level of the title.

[0010] Optionally, when the generating unit is used to construct a CRF form according to the titles at various levels in the target medical report corresponding to the target medical term and the words with numerical values ​​matched in the text content corresponding to the titles at various levels, it includes: After obtaining the titles of each level in the target medical report, segmenting the titles of each level to obtain candidate segmented words included in the titles of each level; For each candidate participle, after determining the part of speech of the candidate participle, the distribution ratio of the part of speech in various parts of speech in the medical field is determined as the weight of the candidate participle; The similarity between the candidate segmentation word and the preset medical field word is calculated using the following formula: S= ; Among them, K is the first word vector corresponding to the candidate word segmentation, and L is the second word vector corresponding to the preset medical field term. is the modulus of the first word vector, is the modulus of the second word vector, and d is the weight; For the text content corresponding to each level of title, text processing is performed on the text content to obtain text segmentation and dependency relationships between the text segmentation; According to the part of speech of each text segmentation, determine, from each text segmentation, a candidate text segmentation having a dependency relationship with a text segmentation whose part of speech is a numeral; Selecting a specified type of part of speech from the candidate text segmentations as a target text segmentation; The candidate segmentations with similarity higher than a preset threshold are taken as parent items, and for each parent item, the target text segmentations included in the text content under the title corresponding to the parent item are taken as child items to construct the CRF form.

[0011] Optionally, the generating device further includes: A display unit is used to display the CRF form on the user terminal.

[0012] Optionally, the user terminal includes a mobile terminal and a fixed terminal.

[0013] Optionally, the generating device further includes: The adding unit is used to respond to the adding operation of the user and add the item corresponding to the adding operation to the position specified by the adding operation.

[0014] The technical solution provided by the embodiments of the present application may have the following beneficial effects: In the present application, when a CRF form for a rare disease is needed, the user can input keywords of the rare disease through the user terminal. A medical report database is provided in the server. Since the medical reports stored in the medical report database contain items that need to be filled in the CRF form corresponding to the relevant disease, such as: basic information of the patient, past medical history, personal physical examination results, medical examination results, imaging examination results, doctor's opinions and suggestions, etc., various matters that can be used to establish a CRF form can be obtained from the medical report corresponding to the rare disease. By comparing with the medical terms included in the medical dictionary, a medical report with a high degree of relevance to the keyword can be determined, thereby facilitating the construction of a CRF form related to the current rare disease. Through the above method, the user only needs to enter the keyword to obtain the corresponding CRF form, which is conducive to reducing manual workload.

[0015] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A flowchart of a method for generating a CRF form for a rare disease provided in an embodiment of the present application; Figure 2 A flowchart of another method for generating a CRF form for a rare disease provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of a device for generating a CRF form for a rare disease provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0019] It should be noted in advance that the medical reports involved in this application are guiding medical documents that can serve as guidance for doctors or medical workers to carry out treatment and research on related rare diseases, for example: guiding medical documents on how to diagnose and treat related diseases, relevant precautions and relevant disease indicators. The specific medical reports stored in the medical report database can be set according to actual needs and are not specifically limited here.

[0020] Figure 1 A flow chart of a method for generating a CRF form for a rare disease provided in an embodiment of the present application is provided. The generation method is applied to a form generation system. The form generation system includes a user terminal and a server. The server is provided with a medical report database and a medical dictionary. The generation method runs on the server. Figure 1 As shown, the generation method comprises the following steps: Step 101 : In response to a keyword input by a user at the user terminal, frequency statistics are performed on each medical report in the medical report database to determine the candidate words with the highest frequency in each medical report.

[0021] Step 102: for each medical term and each to-be-compared term in the medical dictionary, construct a (m+1)*(n+1) matrix according to the first string length of the medical term and the second string length of the to-be-compared term, wherein m is the first string length, n is the second string length, and the to-be-compared term includes the keyword and the alternative term.

[0022] Step 103, initialize the matrix elements of the first row of the matrix to integers from 0 to n in sequence, initialize the matrix elements of the first column of the matrix to integers from 0 to m in sequence, and for other matrix elements Tij in the matrix, determine other matrix elements Tij according to preset rules; wherein the preset rules include: if the i-th character of the medical term is the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the matrix element in the upper left matrix cell adjacent to the matrix cell where the matrix element Tij is located; if the i-th character of the medical term is not the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the sum of the smallest matrix elements in the three matrix cells adjacent to the upper left corner, the upper side and the left side of the matrix cell where the matrix element Tij is located and the value 1.

[0023] Step 104: After determining the matrix elements in the matrix cells in the matrix, the similarity between the medical term and the term to be compared is calculated using the following formula 1: S=1-B / A; (Formula 1) Wherein, B is a matrix element in a matrix cell located at the lower right corner of the matrix, and A is the maximum value of m and n.

[0024] Step 105: After obtaining the first similarities between each candidate word and each medical word and the second similarities between the keyword and each medical word, determine the target medical word corresponding to the first similarity with the smallest difference with the target similarity among the first similarities between each candidate word and the target medical word, wherein the target medical word is the medical word corresponding to the maximum first similarity of each candidate word, and the target similarity is the maximum second similarity.

[0025] Step 106: construct a CRF form according to the target medical terms and the words with matching numerical values ​​in the titles of each level in the target medical report corresponding to the target medical terms and the text contents corresponding to each level.

[0026] Specifically, when a user (including but not limited to: doctors, medical workers, disease researchers, etc.) needs to use a CRF form for a rare disease, a keyword related to the rare disease can be entered in the user terminal, for example: the name of the rare disease. After obtaining the keyword, the server will perform word frequency statistics on each medical report stored in the medical report database, and determine the candidate words with the highest word frequency in each medical report. By selecting the candidate words with the highest word frequency in the medical report, the specific disease type targeted by the corresponding medical report can be determined. For a disease, the description method may be different. In order to determine the standard terms corresponding to each candidate word and keyword, it can be determined through steps 102-104. For example, a medical word in the medical dictionary is "kitten" and the word to be compared is "sitting". The string length of "kitten" is 6 and the string length of "sitting" is 7. Then a 7*8 matrix is ​​constructed, and then the first row matrix elements of the 7*8 matrix are initialized to 0-7 in sequence, and the first column matrix elements are initialized to 0-6 in sequence. The resulting matrix is ​​as follows:

[0027] For other matrix elements Tij (the matrix composed of blank matrix cells in the above matrix), they are determined in sequence from left to right and from top to bottom, and the preset rule used for determination is: if the i-th character of the medical term is the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the matrix element in the upper left matrix cell adjacent to the matrix cell where the matrix element Tij is located; if the i-th character of the medical term is not the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the sum of the smallest matrix elements in the three matrix cells adjacent to the upper left, upper side and left side of the matrix cell where the matrix element Tij is located and the value 1.

[0028] Taking the matrix element in the second row and second column of the above array as an example, Tij is T11, that is, the first character of the medical term and the first character of the term to be compared, the first character of the medical term is k, and the first character of the term to be compared is s. The smallest matrix element in the three matrix cells adjacent to the cell T11 at the upper left corner, upper side and left side is 0, and the sum of the value 1 is 1, so the value 1 is used as the matrix element of the cell T11.

[0029] Taking the matrix element in the third row and third column of the above array as an example, Tij is T22, that is: the second character of the medical term and the second character of the term to be compared, the second character of the medical term is i, and the second character of the term to be compared is also i. The matrix cell in the upper left corner adjacent to the cell T22 is T11, so the matrix element in the cell T22 is 1.

[0030] The matrix elements in the cells of the above matrix can be obtained in the above method. At this time, the matrix elements in the matrix cell in the lower right corner of the above matrix can be obtained, and A is 7. Then, through Formula 1, the similarity when the medical term is "kitten" and the term to be compared is "sitting" can be obtained.

[0031] The first similarity can be used to determine the relevance of a candidate word to each medical word (i.e., standard medical terms), and the second similarity can be used to determine the relevance of the keyword input by the user to each medical word. When the relevance of a candidate word to the keyword is high, the difference in similarity between the two and the same medical word will generally be relatively small. Therefore, in order to determine which candidate word the keyword input by the user has a higher relevance to, after obtaining the first similarity between each candidate word and each medical word and the second similarity between the keyword and each medical word, the medical word with the highest similarity to each candidate word (i.e., target medical word) can be determined. At this time, each candidate word will determine a target medical word, which is the word with the highest correlation with the corresponding medical report, and the second similarity between the keyword and each medical word can be determined, and the target medical word can be determined from each second similarity. The maximum second similarity is obtained. After obtaining the target medical terms corresponding to each candidate word, the first similarity between each candidate word and the target medical term (hereinafter referred to as the similarity to be compared) can be obtained. Then, the difference between the maximum second similarity determined above and each similarity to be compared is calculated to determine the target medical term corresponding to the similarity to be compared with the smallest difference. The target medical term can be understood as the medical term with the highest relevance to the keyword. Since each candidate word corresponds to a target medical term, after obtaining the target medical term, the candidate word corresponding to the target medical term is determined, and then the medical report from which the candidate word comes can be determined, so that the medical report can be used as the target medical report. After obtaining the target medical report, a CRF form is constructed according to the words with matching values ​​in the titles at all levels in the target medical report and the text content corresponding to the titles at all levels.

[0032] Since the target medical report has guiding content for the disease corresponding to the keyword input by the user, guiding items can be obtained through the various levels of titles in the target medical report and the corresponding text content under each level of titles, so that the items in the constructed CRF form are highly relevant to the diseases corresponding to the keywords, and the items are relatively complete for users to use. Through the above method, users only need to enter keywords to get the corresponding CRF form, which is conducive to reducing manual workload.

[0033] In one possible embodiment, Figure 2A flow chart of another method for generating a CRF form for a rare disease provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, when executing step 106, it can be achieved by the following steps: Step 201: After obtaining the titles of each level in the target medical report, segment the titles of each level to obtain candidate segmented words included in the titles of each level.

[0034] Step 202: for each candidate participle, after determining the part of speech of the candidate participle, the distribution ratio of the part of speech in various parts of speech in the medical field is determined as the weight of the candidate participle.

[0035] Step 203: Calculate the similarity between the candidate word segment and the preset medical field word using the following formula 2: S= ; (Formula 2) Among them, K is the first word vector corresponding to the candidate word segmentation, and L is the second word vector corresponding to the preset medical field term. is the modulus of the first word vector, is the modulus of the second word vector, and d is the weight.

[0036] Step 204: For the text content corresponding to each level of title, text processing is performed on the text content to obtain text segmentation and dependency relationships between text segmentation.

[0037] Step 205: According to the part of speech of each text segmentation, determine from among the text segmentations a candidate text segmentation that has a dependency relationship with a text segmentation whose part of speech is a numeral.

[0038] Step 206: Select a specified type of part of speech from the candidate text segmentations as a target text segmentation.

[0039] Step 207: Take the candidate segmented words with similarity higher than the preset threshold as parent items, and for each parent item, take the target text segmented words included in the text content under the title corresponding to the parent item as child items to construct the CRF form.

[0040] Specifically, after obtaining the target medical report, a parent item can be constructed according to the titles at each level, and then the child items under the parent item can be constructed according to the content under the titles at each level, so as to construct a CRF form. For the titles at each level, the titles at each level are first segmented to obtain the candidate words included in the titles at each level. In the medical field, the usage frequencies of different parts of speech are different. The higher the usage frequency, the higher the importance of the part of speech in the medical field. Therefore, weights can be assigned to the corresponding candidate analysis according to the distribution ratio of the parts of speech in various parts of speech in the medical field. For example: the distribution ratio of nouns in various parts of speech in the medical field is 30%, and the assigned weight of the noun part of speech is 0.3. After obtaining the corresponding weights of each candidate participle, the similarity between each candidate participle and the preset medical field word is calculated using Formula 2. The higher the similarity, the higher the relevance of the candidate word to the preset medical field.

[0041] After obtaining titles at all levels, perform text segmentation on the text content under each level of title, and determine the dependency relationship between each text segmentation. The dependency relationship can characterize the syntactic interdependence between segmentations. Since CRF forms are generally filled in with content related to numerical values, after obtaining the text segmentation, determine the numeral segmentations related to numerals in the text segmentation, and then use the text segmentations that have a dependency relationship with the numeral segmentations as candidate text segmentations (candidate text segmentations can be used as sub-items).

[0042] Since sub-items are generally words of one or several categories of this part of speech, such as nouns, after obtaining the candidate text segmentation words, the specified type of part of speech is selected from the candidate text segmentation words as the target text segmentation words, and then the candidate segmentation words with a similarity higher than a preset threshold are used as parent items, and for each parent item, the target text segmentation words included in the text content under the title corresponding to the parent item are used as child items to construct the CRF form, so that the node relationship expressed by the constructed CRF is the same as the directory relationship of constructing the CRF form by taking the target text segmentation words included in the text content under the title corresponding to the parent item as child items for each parent item.

[0043] It should be noted that different search terms and preset medical field terms corresponding to each search term can be pre-set. After obtaining the candidate participles, the cosine similarity is used to determine the similarity between each candidate participle and each search term, and the search term with the highest similarity is used as the target search term for the candidate participle. Then, the preset medical field term corresponding to the target search term is used as the preset medical field term for the candidate participle. Alternatively, multiple preset medical field terms can be pre-set based on experience. How to obtain the preset medical field terms can be set according to actual needs and is not specifically limited here.

[0044] In a feasible implementation manner, after the CRF form is obtained, the CRF form is displayed on the user terminal for the user to view.

[0045] In a feasible implementation manner, the user terminal includes a mobile terminal and a fixed terminal.

[0046] In a feasible implementation scheme, after obtaining the CRF form, it is also possible to respond to the user's adding operation and add the item corresponding to the adding operation to the position specified by the adding operation to meet the user's specific needs.

[0047] Figure 3 A schematic diagram of the structure of a device for generating a CRF form for a rare disease provided in an embodiment of the present application, wherein the device is applied to a form generation system, wherein the form generation system includes a user terminal and a server, wherein the server is provided with a medical report database and a medical dictionary, and wherein the device is configured at the server. Figure 3 As shown, the generating device comprises: A statistical unit 31, configured to respond to a keyword input by a user at the user terminal, perform word frequency statistics on each medical report in the medical report database, and determine the candidate word with the highest word frequency in each medical report; A construction unit 32 is used to construct, for each medical term and each to-be-compared term in the medical dictionary, a matrix of (m+1)*(n+1) according to the first string length of the medical term and the second string length of the to-be-compared term, wherein m is the first string length, n is the second string length, and the to-be-compared term includes the keyword and the alternative term; The first determination unit 33 is used to initialize the matrix elements of the first row of the matrix to integers from 0 to n, initialize the matrix elements of the first column of the matrix to integers from 0 to m, and determine the other matrix elements Tij in the matrix according to a preset rule; wherein the preset rule includes: if the i-th character of the medical term is the same as the j-th character of the term to be compared, the matrix element Tij is equal to the matrix element in the upper left matrix cell adjacent to the matrix cell where the matrix element Tij is located; if the i-th character of the medical term is not the same as the j-th character of the term to be compared, the matrix element Tij is equal to the sum of the smallest matrix elements in the three matrix cells adjacent to the upper left corner, the upper side and the left side of the matrix cell where the matrix element Tij is located and the value 1; The calculation unit 34 is used to calculate the similarity between the medical term and the term to be compared using the following formula after determining the matrix elements in the matrix cells in the matrix: S=1-B / A; Wherein, B is a matrix element in a matrix cell located at the lower right corner of the matrix, and A is the maximum value of m and n; A second determination unit 35 is used to determine, after obtaining the first similarities between each candidate word and each medical word and the second similarities between the keyword and each medical word, a target medical word corresponding to the first similarity with the smallest difference from the target similarity among the first similarities between each candidate word and the target medical word, wherein the target medical word is the medical word corresponding to the largest first similarity of each candidate word, and the target similarity is the largest second similarity; The generating unit 36 ​​is used to construct a CRF form according to the words with matching numerical values ​​in the titles of each level in the target medical report corresponding to the target medical word and the text contents corresponding to each level of the titles.

[0048] In a feasible embodiment, the generating unit is used to construct a CRF form according to the words with numerical values ​​matched in the titles of each level in the target medical report corresponding to the target medical word and the text content corresponding to each level of the title, including: After obtaining the titles of each level in the target medical report, segmenting the titles of each level to obtain candidate segmented words included in the titles of each level; For each candidate participle, after determining the part of speech of the candidate participle, the distribution ratio of the part of speech in various parts of speech in the medical field is determined as the weight of the candidate participle; The similarity between the candidate segmentation word and the preset medical field word is calculated using the following formula: S= ; Among them, K is the first word vector corresponding to the candidate word segmentation, and L is the second word vector corresponding to the preset medical field term. is the modulus of the first word vector, is the modulus of the second word vector, and d is the weight; For the text content corresponding to each level of title, text processing is performed on the text content to obtain text segmentation and dependency relationships between the text segmentation; According to the part of speech of each text segmentation, determine, from each text segmentation, a candidate text segmentation having a dependency relationship with a text segmentation whose part of speech is a numeral; Selecting a specified type of part of speech from the candidate text segmentations as a target text segmentation; The candidate segmentations with similarity higher than a preset threshold are taken as parent items, and for each parent item, the target text segmentations included in the text content under the title corresponding to the parent item are taken as child items to construct the CRF form.

[0049] In a feasible embodiment, the generating device further comprises: A display unit is used to display the CRF form on the user terminal.

[0050] In a feasible implementation manner, the user terminal includes a mobile terminal and a fixed terminal.

[0051] In a feasible embodiment, the generating device further comprises: The adding unit is used to respond to the adding operation of the user and add the item corresponding to the adding operation to the position specified by the adding operation.

[0052] about Figure 3 For a description of the relevant principles, please refer to Figure 1-Figure 2 The description of the relevant contents will not be repeated in detail here.

[0053] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0054] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0055] In addition, each functional unit in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0056] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0057] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0058] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application. They should all be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for generating a CRF form for a rare disease, characterized in that: The generation method is applied to a form generation system, the form generation system includes a user terminal and a server, the server is provided with a medical report database and a medical dictionary, the generation method runs on the server, and the generation method includes: In response to a keyword input by a user at the user terminal, frequency statistics are performed on each medical report in the medical report database to determine the candidate words with the highest frequency in each medical report; For each medical term and each to-be-compared term in the medical dictionary, construct a (m+1)*(n+1) matrix according to the first character string length of the medical term and the second character string length of the to-be-compared term, wherein m is the first character string length, n is the second character string length, and the to-be-compared term includes the keyword and the alternative term; Initialize the matrix elements of the first row of the matrix to integers from 0 to n in sequence, initialize the matrix elements of the first column of the matrix to integers from 0 to m in sequence, and determine the other matrix elements Tij in the matrix according to preset rules; wherein the preset rules include: if the i-th character of the medical term is the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the matrix element in the upper left matrix cell adjacent to the matrix cell where the matrix element Tij is located; if the i-th character of the medical term is not the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the sum of the smallest matrix elements in the three matrix cells adjacent to the upper left corner, the upper side and the left side of the matrix cell where the matrix element Tij is located and the value 1; After determining the matrix elements in the matrix cells in the matrix, the similarity between the medical term and the term to be compared is calculated using the following formula: S=1-B / A; Wherein, B is a matrix element in a matrix cell located at the lower right corner of the matrix, and A is the maximum value of m and n; After obtaining the first similarities between each candidate word and each medical word and the second similarities between the keyword and each medical word, determining a target medical word corresponding to the first similarity with the smallest difference from the target similarity among the first similarities between each candidate word and the target medical word, wherein the target medical word is the medical word corresponding to the largest first similarity of each candidate word, and the target similarity is the largest second similarity; A CRF form is constructed according to the words with matching numerical values ​​in the titles of each level in the target medical report corresponding to the target medical word and the text contents corresponding to each level of the titles.

2. The generation method according to claim 1, characterized in that: The step of constructing a CRF form according to the words with numerical values ​​matched in the titles of each level in the target medical report corresponding to the target medical word and the text contents corresponding to each level of the titles includes: After obtaining the titles of each level in the target medical report, segmenting the titles of each level to obtain candidate segmented words included in the titles of each level; For each candidate participle, after determining the part of speech of the candidate participle, the distribution ratio of the part of speech in various parts of speech in the medical field is determined as the weight of the candidate participle; The similarity between the candidate segmentation word and the preset medical field word is calculated using the following formula: S= ； Among them, K is the first word vector corresponding to the candidate word segmentation, and L is the second word vector corresponding to the preset medical field term. is the modulus of the first word vector, is the modulus of the second word vector, and d is the weight; For the text content corresponding to each level of title, text processing is performed on the text content to obtain text segmentation and dependency relationships between the text segmentation; According to the part of speech of each text segmentation, determine, from each text segmentation, a candidate text segmentation having a dependency relationship with a text segmentation whose part of speech is a numeral; Selecting a specified type of part of speech from the candidate text segmentations as a target text segmentation; The candidate segmentations with similarity higher than a preset threshold are taken as parent items, and for each parent item, the target text segmentations included in the text content under the title corresponding to the parent item are taken as child items to construct the CRF form.

3. The generation method according to claim 1, characterized in that: The generating method further comprises: The CRF form is displayed on the user terminal.

4. The generation method according to claim 1, characterized in that: The user terminals include mobile terminals and fixed terminals.

5. The generation method according to claim 1, characterized in that: The generating method further comprises: In response to the user's adding operation, the item corresponding to the adding operation is added to the position specified by the adding operation.

6. A device for generating a CRF form for a rare disease, characterized in that: The generating device is applied to a form generating system, the form generating system comprises a user terminal and a server, the server is provided with a medical report database and a medical dictionary, the generating device is configured at the server, and the generating device comprises: A statistical unit, configured to respond to a keyword input by a user at the user terminal, perform word frequency statistics on each medical report in the medical report database, and determine the candidate word with the highest word frequency in each medical report; A construction unit is used to construct, for each medical word and each to-be-compared word in the medical dictionary, a (m+1)*(n+1) matrix according to a first character string length of the medical word and a second character string length of the to-be-compared word, wherein m is the first character string length, n is the second character string length, and the to-be-compared word includes the keyword and the alternative word; A first determination unit is used to initialize the matrix elements of the first row of the matrix to integers from 0 to n, initialize the matrix elements of the first column of the matrix to integers from 0 to m, and determine other matrix elements Tij in the matrix according to preset rules; wherein the preset rules include: if the i-th character of the medical term is the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the matrix element in the upper left matrix cell adjacent to the matrix cell where the matrix element Tij is located; if the i-th character of the medical term is not the same as the j-th character of the term to be compared, then the matrix element Tij is equal to the sum of the smallest matrix elements in the three matrix cells adjacent to the upper left corner, the upper side and the left side of the matrix cell where the matrix element Tij is located and the value 1; The calculation unit is used to calculate the similarity between the medical term and the term to be compared using the following formula after determining the matrix elements in the matrix cells in the matrix: S=1-B / A; Wherein, B is a matrix element in a matrix cell located at the lower right corner of the matrix, and A is the maximum value of m and n; A second determination unit is used to determine, after obtaining the first similarities between each candidate word and each medical word and the second similarities between the keyword and each medical word, a target medical word corresponding to the first similarity with the smallest difference from the target similarity among the first similarities between each candidate word and the target medical word, wherein the target medical word is the medical word corresponding to the largest first similarity of each candidate word, and the target similarity is the largest second similarity; The generating unit is used to construct a CRF form according to the words with numerical values ​​matched in the titles of each level in the target medical report corresponding to the target medical word and the text content corresponding to each level of the title.

7. The generating device according to claim 6, characterized in that The generating unit is used to construct a CRF form according to the words with numerical values ​​matched in the titles of each level in the target medical report corresponding to the target medical word and the text content corresponding to each level of the titles, including: After obtaining the titles of each level in the target medical report, segmenting the titles of each level to obtain candidate segmented words included in the titles of each level; For each candidate participle, after determining the part of speech of the candidate participle, the distribution ratio of the part of speech in various parts of speech in the medical field is determined as the weight of the candidate participle; The similarity between the candidate segmentation word and the preset medical field word is calculated using the following formula: S= ； Among them, K is the first word vector corresponding to the candidate word segmentation, and L is the second word vector corresponding to the preset medical field term. is the modulus of the first word vector, is the modulus of the second word vector, and d is the weight; For the text content corresponding to each level of title, text processing is performed on the text content to obtain text segmentation and dependency relationships between the text segmentation; According to the part of speech of each text segmentation, determine, from each text segmentation, a candidate text segmentation having a dependency relationship with a text segmentation whose part of speech is a numeral; Selecting a specified type of part of speech from the candidate text segmentations as a target text segmentation; The candidate segmentations with similarity higher than a preset threshold are taken as parent items, and for each parent item, the target text segmentations included in the text content under the title corresponding to the parent item are taken as child items to construct the CRF form.

8. The generating device according to claim 6, characterized in that The generating device further comprises: A display unit is used to display the CRF form on the user terminal.

9. The generating device according to claim 6, characterized in that: The user terminals include mobile terminals and fixed terminals.

10. The generating device according to claim 6, characterized in that The generating device further comprises: The adding unit is used to respond to the adding operation of the user and add the item corresponding to the adding operation to the position specified by the adding operation.

Images