Disease name standardization method, device, storage medium and electronic device

Abstract

The present application discloses a method and device for standardizing disease names, a storage medium, and an electronic device. The method includes: dividing a target sentence into one or more target clauses according to a preset wildcard; dividing each target clause into one or more words, and determining the part of speech of each word respectively, and determining the part of speech set of the target clause according to the part of speech of the word; determining whether the target clause is complete according to the part of speech set; if incomplete, supplementing the target clause according to the part of speech that is missing in the part of speech set; and determining the target standard disease name corresponding to each target clause according to the standard disease names in the preset disease library. The method of the present application standardizes the disease names in the underwriting questionnaire of the insurance product in advance, so that uninsurable products can be quickly filtered out according to the standardized disease names, solving the problem of inefficiency caused by multiple underwritings in the existing insurance process.

Description

Technical Field

[0001] The present application relates to the field of natural language processing technology, and in particular to a disease name standardization method, device, storage medium and electronic device. Background Art

[0002] In recent years, public risk awareness has continued to grow, along with increasing acceptance of insurance products and a significant increase in the proportion of insurance purchases. For policyholders, the primary role of online insurance brokerages is to help them select suitable insurance products, assist with underwriting and insurance purchases, and facilitate subsequent claims processes.

[0003] When selecting an insurance product for a user, underwriting is required to determine whether the user meets the product's terms and conditions. Only after passing the underwriting process can the user purchase the insurance. The typical underwriting process includes reviewing the product's health disclosure. If the health disclosure requirements are met, the underwriting is passed and the user can purchase the insurance directly. If the health disclosure requirements are partially not met (such as for users with medical conditions), the process proceeds to intelligent underwriting, where the user's eligibility is determined based on the selected condition and the corresponding answers.

[0004] Health declarations are usually long paragraphs of text that list the diseases, family diseases, and occupational types that users cannot be insured for. The questionnaires used in smart underwriting are usually pre-set questionnaire trees. A certain disease or a certain type of disease may have a direct underwriting conclusion, or there may be multiple layers of questions. Choosing different answers may ultimately result in different underwriting conclusions. Usually, different insurance products correspond to different health declarations and underwriting questionnaires. The same disease may have different disease names in different health declarations or underwriting questionnaires (the disease name in product A is HCV antibody positive, the disease name in product B is hepatitis C antibody positive, and the disease name in product C is hepatitis C antibody positive).

[0005] In the actual insurance application process, the same user may receive different underwriting results from different insurance companies or different products. Even if rejected by a certain insurance product, there may be other products that are more tolerant of abnormal indicators and can be approved. However, because different insurance products may use different names for the same disease, users may need to go through underwriting multiple times to find an approved insurance product, which is cumbersome and inefficient. Summary of the Invention

[0006] In view of this, the present application provides a disease name standardization method, device, medium and equipment to standardize the disease names in the underwriting questionnaire of insurance products in advance, so that uninsurable products can be quickly filtered out based on the standardized disease names, solving the problem of inefficiency caused by multiple underwritings in the existing insurance process.

[0007] According to one aspect of the present application, a method for standardizing disease names is provided, comprising:

[0008] Split the target sentence into one or more target clauses based on the preset wildcards;

[0009] Segmenting each target clause into one or more words, determining the part of speech of each word, and determining the part of speech set of the target clause according to the part of speech of the word;

[0010] Determining whether the target clause is complete according to the part-of-speech set;

[0011] If it is incomplete, supplement the target clause according to the missing parts of speech in the part-of-speech set;

[0012] According to the standard disease names in the preset disease library, the target standard disease name corresponding to each target clause is determined respectively.

[0013] Optionally, dividing the target sentence into one or more target clauses according to a preset wildcard specifically includes:

[0014] Determining a position of the preset wildcard in the target sentence;

[0015] If the position of the preset wildcard is not within the brackets, the position of the preset wildcard is determined to be the first segmentation position;

[0016] Segmenting the target sentence into one or more middle clauses based on the first segmentation position;

[0017] If the middle clause does not contain brackets, the middle clause is treated as a target clause;

[0018] If the middle clause contains the bracket and the bracket does not contain the preset wildcard, determining the position of the bracket as the second segmentation position;

[0019] If the middle clause contains the bracket and the bracket contains the preset wildcard, determining the position of the bracket and the position of the preset wildcard in the bracket as the second segmentation position;

[0020] The middle clause is segmented into one or more target clauses based on the second segmentation position.

[0021] Optionally, the part-of-speech category of the part-of-speech includes a preposition category, a postposition category, and an independent category;

[0022] The determining whether the target clause is complete according to the part-of-speech set specifically includes:

[0023] Determine the part-of-speech category of each part of speech in the part-of-speech set to obtain a category set;

[0024] If the category set includes the independent category, determining that the target clause is complete;

[0025] If the category set includes the preceding category and the following category, determining that the target clause is complete;

[0026] Otherwise, it is determined that the target clause is incomplete.

[0027] Optionally, supplementing the target clause according to the missing parts of speech in the part-of-speech set specifically includes:

[0028] determining an arrangement order among the plurality of target clauses according to a position of each target clause in the target sentence;

[0029] Determining a target clause corresponding to the part-of-speech set as a clause to be completed;

[0030] If the part-of-speech set does not include the preceding category, the preceding category is regarded as a missing category, and the target clause arranged before the clause to be completed is regarded as a completion template clause;

[0031] If the part-of-speech set does not include the post-position category, the post-position category is used as the missing category, and the target clause arranged after the clause to be completed is used as the completion template clause;

[0032] Determine whether each of the completion template clauses is complete in descending order of distance from the clause to be completed;

[0033] If it is complete, then in the part-of-speech set corresponding to the completion template clause, determine that the part-of-speech category is the part-of-speech of the missing category and is the part-of-speech missing from the part-of-speech set, and use the words corresponding to the part-of-speech missing from the part-of-speech set to supplement the clause to be completed, and exit the step of judging whether each of the completion template clauses is complete in turn.

[0034] Optionally, determining the target standard disease name corresponding to each target clause based on the standard disease nouns in the preset disease library specifically includes:

[0035] Searching for a standard disease name corresponding to the target clause in the preset disease library, and if found, using the standard disease name as the target standard disease name;

[0036] If not found, searching for a disease synonym corresponding to the target clause in a preset disease synonym library, wherein the preset disease synonym library includes disease synonyms and standard disease names corresponding to the disease synonyms;

[0037] If a disease synonym corresponding to the target clause is found, the standard disease name corresponding to the disease synonym is determined to be the target disease name.

[0038] Optionally, the method further includes:

[0039] If no disease synonyms corresponding to the target clause are found in the preset disease synonym library, the standard disease name and the disease synonyms are used as similar disease names, and the similarity between the target clause and each of the similar disease names is calculated using the trained text matching model;

[0040] The target standard disease name is determined according to the similarity.

[0041] Optionally, determining the target standard disease name according to the similarity specifically includes:

[0042] Determining a maximum similarity with the largest value among the plurality of similarities;

[0043] If the maximum similarity is greater than the preset threshold, and the similar disease name corresponding to the maximum similarity is the standard disease name, determining the similar disease name corresponding to the maximum similarity as the target standard disease name;

[0044] If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is a synonym of the disease, then determining the standard disease name corresponding to the similar disease name corresponding to the maximum similarity as the target standard disease name;

[0045] If the maximum similarity is less than the preset threshold, a prompt message indicating that the target standard disease name is not matched is output.

[0046] According to another aspect of the present application, a disease name standardization device is provided, comprising:

[0047] The first segmentation module is used to segment the target sentence into one or more target clauses according to preset wildcards;

[0048] a second segmentation module, configured to segment each target clause into one or more words, and determine the part of speech of each word, and determine the part of speech set of the target clause according to the part of speech of the word;

[0049] a splicing module, configured to determine whether the target clause is complete based on the part-of-speech set; if incomplete, supplement the target clause based on the missing parts of speech in the part-of-speech set;

[0050] The matching module is used to determine the target standard disease name corresponding to each target clause according to the standard disease names in the preset disease library.

[0051] Optionally, the first segmentation module is specifically configured to:

[0052] Determining a position of the preset wildcard in the target sentence;

[0053] If the position of the preset wildcard is not within the brackets, the position of the preset wildcard is determined to be the first segmentation position;

[0054] Segmenting the target sentence into one or more middle clauses based on the first segmentation position;

[0055] If the middle clause does not contain brackets, the middle clause is treated as a target clause;

[0056] If the middle clause contains the bracket and the bracket does not contain the preset wildcard, determining the position of the bracket as the second segmentation position;

[0057] If the middle clause contains the bracket and the bracket contains the preset wildcard, determining the position of the bracket and the position of the preset wildcard in the bracket as the second segmentation position;

[0058] The middle clause is segmented into one or more target clauses based on the second segmentation position.

[0059] Optionally, the part-of-speech category of the part-of-speech includes a preposition category, a postposition category, and an independent category;

[0060] The splicing module is specifically used for:

[0061] Determine the part-of-speech category of each part of speech in the part-of-speech set to obtain a category set;

[0062] If the category set includes the independent category, determining that the target clause is complete;

[0063] If the category set includes the preceding category and the following category, determining that the target clause is complete;

[0064] Otherwise, it is determined that the target clause is incomplete.

[0065] Optionally, the splicing module is further used to:

[0066] determining an arrangement order among the plurality of target clauses according to a position of each target clause in the target sentence;

[0067] Determining a target clause corresponding to the part-of-speech set as a clause to be completed;

[0068] If the part-of-speech set does not include the preceding category, the preceding category is regarded as a missing category, and the target clause arranged before the clause to be completed is regarded as a completion template clause;

[0069] If the part-of-speech set does not include the post-position category, the post-position category is used as the missing category, and the target clause arranged after the clause to be completed is used as the completion template clause;

[0070] Determine whether each of the completion template clauses is complete in descending order of distance from the clause to be completed;

[0071] If it is complete, then in the part-of-speech set corresponding to the completion template clause, determine that the part-of-speech category is the part-of-speech of the missing category and is the part-of-speech missing from the part-of-speech set, and use the words corresponding to the part-of-speech missing from the part-of-speech set to supplement the clause to be completed, and exit the step of judging whether each of the completion template clauses is complete in turn.

[0072] Optionally, the matching module is specifically configured to:

[0073] Searching for a standard disease name corresponding to the target clause in the preset disease library, and if found, using the standard disease name as the target standard disease name;

[0074] If not found, searching for a disease synonym corresponding to the target clause in a preset disease synonym library, wherein the preset disease synonym library includes disease synonyms and standard disease names corresponding to the disease synonyms;

[0075] If a disease synonym corresponding to the target clause is found, the standard disease name corresponding to the disease synonym is determined to be the target disease name.

[0076] Optionally, the matching module is further configured to:

[0077] If no disease synonyms corresponding to the target clause are found in the preset disease synonym library, the standard disease name and the disease synonyms are used as similar disease names, and the similarity between the target clause and each of the similar disease names is calculated using the trained text matching model;

[0078] The target standard disease name is determined according to the similarity.

[0079] Optionally, the matching module is further configured to:

[0080] Determining a maximum similarity with the largest value among the plurality of similarities;

[0081] If the maximum similarity is greater than the preset threshold, and the similar disease name corresponding to the maximum similarity is the standard disease name, determining the similar disease name corresponding to the maximum similarity as the target standard disease name;

[0082] If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is a synonym of the disease, then determining the standard disease name corresponding to the similar disease name corresponding to the maximum similarity as the target standard disease name;

[0083] If the maximum similarity is less than the preset threshold, a prompt message indicating that the target standard disease name is not matched is output.

[0084] According to another aspect of the present application, a storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned disease name standardization method is implemented.

[0085] According to another aspect of the present application, an electronic device is provided, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the above-mentioned disease name standardization method when executing the computer program.

[0086] By means of the above technical solution, the present application first performs some preprocessing operations to process the disease into multiple disease nouns, and then standardizes the disease nouns. First, the disease text is used as the target sentence, and a two-stage segmentation operation is performed. The first stage segments the target sentence into target clauses, and the second stage segments the target clauses into one or more words. Then, according to the part of speech of each word in the target clause, the incomplete target clause after segmentation is completed to obtain a complete target clause without missing components. After that, each target clause is analyzed one by one to obtain the final target standard disease name. In this way, uninsurable products can be quickly filtered according to the target standard disease name, which solves the problem of inefficiency caused by multiple underwritings in the existing insurance process.

[0087] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0089] Figure 1 A schematic diagram of a process for standardizing disease names provided in an embodiment of the present application is shown;

[0090] Figure 2 A schematic diagram showing a flow chart of determining target sentence preprocessing for another disease name standardization method provided in an embodiment of the present application is shown;

[0091] Figure 3 A schematic diagram showing a process of determining a target standard disease name in another disease name standardization method provided in an embodiment of the present application is shown;

[0092] Figure 4 A structural block diagram of a disease name standardization device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0093] In this embodiment, a method for standardizing disease names is provided. Figure 1 As shown, the method includes:

[0094] Step 101: Split the target sentence into one or more target clauses according to preset wildcards;

[0095] The disease name standardization method provided in the embodiment of the present application is applicable to scenarios such as health notification or underwriting questionnaires. In specific application scenarios, the disease may be a large text, that is, most diseases are not a simple noun, but a group of disease nouns, and some disease nouns have discontinuous problems. For example: atrial flutter, ventricular flutter, atrial fibrillation (atrial fibrillation, atrial fibrillation), ventricular fibrillation (ventricular fibrillation, ventricular fibrillation), ST segment or T wave abnormalities, sinus arrest, sinoatrial node block, sick sinus syndrome.

[0096] In this step, the disease text is used as the target sentence, and the target sentence is segmented using the preset wildcard to obtain the target clause, wherein the segmented target clause does not include the wildcard.

[0097] For example, taking the above-mentioned disease text as the target sentence, we can segment it into the following target clauses: ['atrial flutter', 'ventricular flutter', 'atrial fibrillation', 'atrial fibrillation', 'atrial fibrillation', 'ventricular fibrillation', 'ventricular fibrillation', 'ventricular fibrillation', 'ST segment', 'T wave abnormality', 'sinus arrest', 'sinoatrial block', 'sick sinus syndrome'].

[0098] Step 102: Segment each target clause into one or more words, determine the part of speech of each word, and determine the part of speech set of the target clause based on the part of speech of the word;

[0099] In this step, a dictionary of common words and a dictionary of disease components can be pre-built for word segmentation and used as the basis for word segmentation. The segmentation results of the previous step are traversed and a dictionary-based word segmentation operation is performed using the forward maximum matching method.

[0100] For example, by segmenting the aforementioned multiple target clauses, the following multiple words can be obtained: [['atrium', 'flutter'], ['ventricle', 'flutter'], ['atrium', 'fibrillation'], ['atrial fibrillation'], ['atrium', 'fibrillation'], ['ventricle', 'fibrillation'], ['ventricular fibrillation'], ['ventricular', 'fibrillation'], ['ST segment'], ['T wave', 'abnormal'], ['sinus', 'arrest'], ['sinoatrial node', 'conduction block'], ['pathological', 'sinoatrial node', 'syndrome']]. Among them, the target clause 'atrial flutter' is split into two words, 'atrium' and 'flutter', the target clause 'ventricular flutter' is split into two words, 'ventricle' and 'flutter', and so on.

[0101] Next, based on the disease component dictionary, each word is classified into its disease component, i.e., its part of speech. The disease component dictionary includes words and their corresponding disease components. For example, the disease component dictionary might include the word "atrium," and its corresponding disease component is "part," meaning that "atrium" refers to the affected body part.

[0102] For example, by dividing the aforementioned words into disease components, the following parts of speech are obtained: [['part', 'suffix'], ['part', 'suffix'], ['part', 'suffix'], ['independent disease'], ['part', 'suffix'], ['part', 'suffix'], ['independent disease'], ['part', 'suffix'], ['test'], ['test', 'suffix'], ['property', 'disease'], ['part', 'suffix'], ['prefix', 'part', 'suffix']]. Among them, the part of speech of the word 'atrium' is 'part', and the part of speech of the word 'flutter' is 'suffix'. The part of speech set corresponding to the target clause 'atrial flutter' is ['part', 'suffix'], and so on.

[0103] Step 103, determining whether the target clause is complete based on the part-of-speech set;

[0104] Step 104: If the target clause is incomplete, the target clause is supplemented according to the missing parts of speech in the part-of-speech set;

[0105] In this step, if the target clause contains non-consecutive disease names, such as "numbness in the limbs and head," then when it is split, it may obtain two target clauses: "limbs" and "numbness in the head." It is understandable that the target clause "limbs" does not clearly describe the disease condition and is therefore not a complete target clause.

[0106] To address this situation, the present invention determines whether each target clause is complete based on its part of speech, and then completes the incomplete target clause. Specifically, the target clause can be determined to be complete based on a part of speech set, and if it is incomplete, the missing part of speech is analyzed based on the part of speech set.

[0107] For example, the target clause "limbs" is segmented into the word "limbs," whose part of speech is 'body part'. Therefore, the part-of-speech set corresponding to the target clause "limbs" is ['body part']. According to the preset verification rules, if the part-of-speech set only contains 'body part', the part-of-speech set is incomplete, and therefore the target clause "limbs" is also incomplete. Then, based on the verification rules, the missing part of speech in the part-of-speech set is determined and "numbness" is added to the target clause, resulting in the complete target clause "limbs numbness."

[0108] Step 105 : Determine the target standard disease name corresponding to each target clause based on the standard disease nouns in the preset disease database.

[0109] In this embodiment, a standardization operation is performed on each target clause respectively, so that a mapping relationship is formed between the target clause and a standard disease name in a preset disease library, and a corresponding target standard disease name is obtained.

[0110] For example, the preset disease library includes multiple preset standard disease nouns. If the target clause is 'atrial flutter', and the noun 'atrial flutter' is a standard disease name in the preset disease library, a mapping relationship is established between the target clause and the standard disease name, and the target standard disease noun corresponding to the target clause is determined to be 'atrial flutter'.

[0111] By applying the technical solution of this embodiment, some preprocessing operations are first performed to process the disease into multiple disease nouns, and then the disease nouns are standardized. First, the disease text is used as the target sentence, and a two-stage segmentation operation is performed. The first stage segments the target sentence into target clauses, and the second stage segments the target clauses into one or more words. Then, according to the part of speech of each word in the target clause, the incomplete target clause after segmentation is completed to obtain a complete target clause without missing components. After that, each target clause is analyzed one by one to obtain the final target standard disease name. In this way, uninsurable products can be quickly filtered according to the target standard disease name, solving the problem of inefficiency caused by multiple underwritings in the existing insurance process.

[0112] Further, as a refinement and extension of the specific implementation manner of the above embodiment, in order to completely illustrate the specific implementation process of this embodiment, another method for standardizing disease names is provided. The target statement is segmented into one or more target clauses according to a preset wildcard, which specifically includes:

[0113] Step 201, determine the position of the preset wildcard in the target statement;

[0114] Step 202, if the position of the preset wildcard is not within parentheses, determine the position of the preset wildcard as the first segmentation position;

[0115] Step 203, segment the target statement into one or more intermediate clauses based on the first segmentation position;

[0116] In steps 201 - 203, the stage of segmenting the target statement into target clauses can be executed in two steps. The first segmentation step first segments the target statement into intermediate clauses, and the second segmentation step then segments the intermediate clauses into target clauses. In the first step, the first segmentation position is determined according to the preset wildcard, and then the target statement is segmented at the first segmentation position to obtain intermediate clauses.

[0117] Specifically, first determine the position of the preset wildcard in the target statement. Among them, "and", " / ", ",", ",", ";", ":", "or", "and" etc. can be used as the preset wildcard, or other preset wildcards can be set according to actual business requirements. If these preset wildcards are inside parentheses, the position where the preset wildcard is located is not used as the first segmentation position; otherwise, it is used as the first segmentation position.

[0118] For example, the target statement is as follows: atrial flutter, ventricular flutter, atrial fibrillation (AF, atrial fibrillation), ventricular fibrillation (VF, ventricular fibrillation), ST segment or T wave abnormality, sinus arrest, sinoatrial block, sick sinus syndrome.

[0119] For this target statement, the positions of "," and "or" outside the parentheses can be used as the first segmentation position, and then the target statement is segmented at the first segmentation position.

[0120] The multiple intermediate clauses obtained by the first - step segmentation are as follows: ['atrial flutter','ventricular flutter', 'atrial fibrillation (AF, atrial fibrillation)', 'ventricular fibrillation (VF, ventricular fibrillation)', 'ST segment', 'T wave abnormality','sinus arrest','sinoatrial block','sick sinus syndrome'].

[0121] Step 204, if the intermediate clause does not contain parentheses, use the intermediate clause as a target clause;

[0122] Step 205: If the middle clause contains brackets and the brackets do not contain a preset wildcard, the position of the brackets is determined to be the second segmentation position;

[0123] Step 206: If the middle clause contains brackets and the brackets contain a preset wildcard, determine the position of the brackets and the position of the preset wildcard in the brackets as the second segmentation position;

[0124] Step 207: Split the middle clause into one or more target clauses based on the second segmentation position.

[0125] In steps 204-207, the second segmentation step is performed to segment the middle clause into target clauses. Specifically, if a middle clause does not contain parentheses, the second segmentation step is not performed on it, and the middle clause is directly used as a target clause. If a middle clause contains parentheses, the second segmentation step is performed on the middle clause.

[0126] During the second segmentation step, the middle clause's parentheses and the position of the preset wildcard within the parentheses are used as the second segmentation position. If the middle clause's parentheses do not contain a preset wildcard, only the position of the parentheses is used as the second segmentation position. The middle clause is segmented at the second segmentation position to obtain the target clause.

[0127] For example, the multiple middle clauses obtained from the first step of segmentation are: ['atrial flutter', 'ventricular flutter', 'atrial fibrillation (atrial fibrillation, atrial fibrillation)', 'ventricular fibrillation (ventricular fibrillation, ventricular fibrillation)', 'ST segment', 'T wave abnormality', 'sinus arrest', 'sinoatrial block', 'sick sinus syndrome'].

[0128] The second segmentation step is performed on the middle clauses 'atrial fibrillation (atrial fibrillation, atrial fibrillation)' and 'ventricular fibrillation (ventricular fibrillation, ventricular fibrillation)'. 'atrial fibrillation (atrial fibrillation, atrial fibrillation)' is segmented into 'atrial fibrillation', 'atrial fibrillation', and 'atrial fibrillation'; 'ventricular fibrillation (ventricular fibrillation, ventricular fibrillation)' is segmented into 'ventricular fibrillation', 'ventricular fibrillation', and 'ventricular fibrillation'. The other middle clauses do not contain brackets and are therefore treated directly as target clauses.

[0129] Finally, the following target clauses are obtained: ['atrial flutter', 'ventricular flutter', 'atrial fibrillation', 'atrial fibrillation', 'atrial fibrillation', 'ventricular fibrillation', 'ventricular fibrillation', 'ventricular fibrillation', 'ST segment', 'T wave abnormality', 'sinus arrest', 'sinoatrial block', 'sick sinus syndrome'].

[0130] In this embodiment, two segmentation steps are used to segment the target sentence into one or more target clauses based on preset wildcards, that is, a string of disease nouns is split, and then each disease noun can be segmented, completed, and other operations can be performed in subsequent steps to complete the preprocessing of the target sentence.

[0131] Furthermore, in another disease name standardization method, the part-of-speech categories include preposition category, postposition category, and independent category. Based on the part-of-speech set, whether the target clause is complete is determined, specifically including:

[0132] Step 301, determining the part-of-speech category of each part-of-speech in the part-of-speech set to obtain a category set;

[0133] In this embodiment, each part of speech has a part-of-speech category, wherein the part-of-speech category can be a pre-category, a post-category, or an independent category. Therefore, a preset matching rule can be used to determine the part-of-speech category of each part of speech in the part-of-speech set, and a category set can be obtained based on all the part-of-speech categories.

[0134] For example, for the following part-of-speech sets: [['part', 'suffix'], ['part', 'suffix'], ['part', 'suffix'], ['part', 'suffix'], ['independent disease'], ['part', 'suffix'], ['part', 'suffix'], ['independent disease'], ['part', 'suffix'], ['test'], ['test', 'suffix'], ['property', 'disease'], ['part', 'suffix'], ['prefix', 'part', 'suffix']].

[0135] Using the preset matching rules, we can determine that the part-of-speech category corresponding to the part-of-speech 'part' is a pre-positioned category; the part-of-speech category corresponding to the part-of-speech 'suffix' is a post-positioned category; the part-of-speech category corresponding to the part-of-speech 'test' is a pre-positioned category; the part-of-speech category corresponding to the part-of-speech 'independent disease' is an independent category; and the part-of-speech category corresponding to the part-of-speech 'suffix' is a post-positioned category. This allows us to obtain the category set corresponding to each target clause. For example, the part-of-speech set corresponding to the target clause 'atrial flutter' is ['part', 'suffix'], and its corresponding category set is ['pre-positioned category', 'post-positioned category'].

[0136] For the above-mentioned part-of-speech set, it can be determined that the category set corresponding to the target clause is [['pre-category', 'post-category'], ['pre-category', 'post-category'], ['pre-category', 'post-category'], ['pre-category', 'post-category'], ['independent category'], ['pre-category', 'post-category'], ['pre-category', 'post-category'], ['independent category'], ['pre-category', 'post-category'], ['pre-category'], ['pre-category', 'post-category'], ['nature', 'disease'], ['pre-category', 'post-category'], ['pre-category', 'pre-category', 'post-category']].

[0137] Step 302: If the category set includes an independent category, then the target clause is determined to be complete;

[0138] Step 303: If the category set includes the preceding category and the following category, then the target clause is determined to be complete;

[0139] Step 304: Otherwise, determine that the target clause is incomplete.

[0140] In this embodiment, the completeness of the target clause is determined based on the part-of-speech categories in the category set. If the category set includes an independent category, the target clause is considered to clearly represent the disease and is therefore complete. For example, if the target clause "atrial fibrillation" corresponds to the part-of-speech set ['independent disease'] and the category set ['independent category'], the target clause is considered complete.

[0141] If the category set includes both the pre- and post-categories, the target clause is considered to clearly represent the disease, and thus the target clause is determined to be complete. For example, the target clause 'atrial flutter' corresponds to the part-of-speech set of ['part', 'suffix'], and the category set of ['pre-categories', 'post-categories']. In this case, the target clause is determined to be complete.

[0142] Otherwise, that is, the category set does not include an independent category and does not include one of the preceding or following categories, then it is considered that the target clause cannot clearly express the disease, and the target clause is determined to be incomplete. For example: the part-of-speech set corresponding to the target clause 'st segment' is ['inspection'], and the category set is ['preceding category'], which does not include an independent category and does not include a following category. In this case, the target clause is determined to be incomplete.

[0143] This embodiment obtains a category set based on the part-of-speech category of each word in the target clause, and determines whether the target clause is complete based on whether the category set lacks necessary components, thereby supplementing the incomplete target clause to avoid unclear expression of the target clause.

[0144] Furthermore, in another disease name standardization method, the target clause is supplemented according to the part of speech that is missing from the part of speech set, specifically including:

[0145] Step 401, determining the arrangement order of multiple target clauses according to the position of each target clause in the target sentence;

[0146] In this embodiment, if the target clause is incomplete, the target clause can be supplemented according to the missing parts of speech in the part-of-speech set, so that the target clause can clearly indicate the disease.

[0147] Specifically, the target clauses are first sorted according to their positions in the target sentence. For example, for the target sentences: atrial flutter, ventricular flutter, atrial fibrillation (atrial fibrillation, atrial fibrillation), ventricular fibrillation (ventricular fibrillation, ventricular fibrillation), ST segment or T wave abnormality, sinus arrest, sinus node block, sick sinus syndrome, multiple target clauses can be sorted as: ['atrial flutter', 'ventricular flutter', 'atrial fibrillation', 'atrial fibrillation', 'atrial fibrillation', 'ventricular fibrillation', 'ventricular fibrillation', 'ventricular fibrillation', 'ST segment', 'T wave abnormality', 'sinus arrest', 'sinus node block', 'sick sinus syndrome'].

[0148] Step 402: determining the target clause corresponding to the part-of-speech set as the clause to be completed;

[0149] In this embodiment, the target clause corresponding to the part-of-speech set of the missing part of speech is used as the clause to be completed, that is, the incomplete target clause is used as the clause to be completed.

[0150] For example, with respect to the multiple target clauses in the embodiment of step 401, the target clause 'st segment' is used as the clause to be completed.

[0151] Step 403: If the part-of-speech set does not include the preceding category, the preceding category is treated as a missing category, and the target clause arranged before the clause to be completed is used as the completion template clause;

[0152] Step 404: If the part-of-speech set does not include a postposition category, the postposition category is treated as a missing category, and the target clause arranged after the clause to be completed is used as a completion template clause;

[0153] In this embodiment, the part-of-speech set corresponding to the clause to be completed is searched, and it is determined whether the part-of-speech category missing from the part-of-speech set is a preposition category or a postposition category. If the preposition category is missing, the component used for completion is searched in the target clause that precedes the clause to be completed. Therefore, the target clause that precedes the clause to be completed is used as the completion template clause. If the postposition category is missing, the component used for completion is searched in the target clause that follows the clause to be completed. Therefore, the target clause that follows the clause to be completed is used as the completion template clause.

[0154] For example, for the aforementioned target clauses sorted as ['atrial flutter', 'ventricular flutter', 'atrial fibrillation', 'atrial fibrillation', 'atrial fibrillation', 'ventricular fibrillation', 'ventricular fibrillation', 'ST segment', 'T wave abnormality', 'sinus arrest', 'sinoatrial block', 'sick sinus syndrome'], the clause to be completed is 'ST segment'. At this time, because the part-of-speech set is ['test'], which only contains the preceding category and lacks the following category, the component used to complete 'ST segment' is searched in the target clause sorted after 'ST segment'. In other words, ['T wave abnormality', 'sinus arrest', 'sinoatrial block', 'sick sinus syndrome'] is used as the completion template clause.

[0155] Step 405: determine whether each completion template clause is complete in descending order of distance from the clause to be completed.

[0156] Step 406: If it is complete, then in the part-of-speech set corresponding to the completion template clause, determine the part-of-speech category as the missing category as the part-of-speech set missing, and use the words corresponding to the part-of-speech set missing The clause to be completed is supplemented, and exit the step of judging whether each completion template clause is complete in turn.

[0157] In this embodiment, whether each completion template clause is complete is judged in turn in descending order of distance from the clause to be completed, the complete completion template clause closest to the clause to be completed is found, and the components used to complete the clause to be completed are obtained from it, so that the clause to be completed is completed using the found components.

[0158] For example, for the aforementioned completion template clauses ['T wave abnormality', 'sinus arrest', 'sinoatrial block', 'sick sinus syndrome'], it can be determined that the complete completion template clause closest to the clause to be completed 'st segment' is 'T wave abnormality', so the components used to complete 'st segment' can be obtained from 'T wave abnormality'.

[0159] Specifically, since the missing category in the category set corresponding to 'st segment' is a postpositioned category, and the part-of-speech set corresponding to 'T wave abnormality' is ['test', 'suffix'], the part-of-speech category of 'test' is a prepositioned category, and the part-of-speech category of 'suffix' is a postpositioned category. Therefore, in 'T wave abnormality', the part-of-speech category of the postpositioned category is determined, that is, the part-of-speech category missing from the part-of-speech set with the part-of-speech 'suffix' being 'st segment'. Since the word corresponding to the part-of-speech 'suffix' in 'T wave abnormality' is 'abnormal', this word 'abnormal' can be used to complete 'st segment', resulting in the completed target clause 'st segment abnormality'.

[0160] This embodiment realizes the completion of an incomplete target clause, so that the target clause can clearly express the disease and avoid unclear expressions affecting subsequent standardization operations.

[0161] Furthermore, Figure 2 A schematic diagram illustrates the preprocessing process for determining target sentences in another disease name standardization method. As shown, the target sentence is first identified as a text segment describing the disease in the underwriting document. Wildcard characters such as commas and parentheses are then used to segment the target sentence into target clauses. Each target clause is then segmented using a pre-set dictionary of common words and a dictionary of disease components, with each word's part of speech labeled. Finally, a component analysis is performed on each target clause using a set of categories to determine its completeness. If a target clause lacks a component to form a complete disease name, or if a prepended category can be added to a target clause that does not contain an independent category, the target clause is considered incomplete and completion is performed. Specifically, some components from target clauses preceding or following the target clause are added to the target clause, completing the incomplete target clauses. Using the preprocessing steps shown in the figure, a text segment consisting of a string of disease names can be transformed into multiple clearly defined disease names, facilitating standardization in subsequent steps.

[0162] Furthermore, in another disease name standardization method, the target standard disease name corresponding to each target clause is determined based on the standard disease nouns in the preset disease library, specifically including:

[0163] Step 501: Search for a standard disease name corresponding to the target clause in a preset disease database. If found, use the standard disease name as the target standard disease name.

[0164] Step 502: If not found, search for a disease synonym corresponding to the target clause in a preset disease synonym library, wherein the preset disease synonym library includes disease synonyms and standard disease names corresponding to the disease synonyms;

[0165] Step 503: If a disease synonym corresponding to the target clause is found, the standard disease name corresponding to the disease synonym is determined as the target disease name.

[0166] In this embodiment, the target clause is first retrieved in the preset disease library. If the target clause can be directly retrieved in the preset disease library, the retrieved standard disease noun can be directly used as the target standard disease noun corresponding to the target clause.

[0167] If the target clause is not found in the preset disease database, a search is performed in the preset disease synonym database. If the target clause is found in the preset disease synonym database, the standard disease noun corresponding to the retrieved disease synonym can be used as the target disease noun.

[0168] For example, first search for the target clause "hepatitis C antibody positive" in the preset disease library. If not found, search for the target clause "hepatitis C antibody positive" in the preset disease synonym library. If the disease synonym "hepatitis C antibody positive" is retrieved, the standard disease name corresponding to the disease synonym "hepatitis C antibody positive" is used as the target standard disease name.

[0169] The preset disease synonym library includes disease synonyms and standard disease names corresponding to the disease synonyms. That is, the standard disease names corresponding to the disease synonyms can be searched in the preset disease synonym library.

[0170] Furthermore, another method for standardizing disease names further includes:

[0171] Step 601: If no disease synonyms corresponding to the target clause are found in the preset disease synonym library, the standard disease name and the disease synonyms are used as similar disease names, and the similarity between the target clause and each similar disease name is calculated using the trained text matching model;

[0172] Step 602: Determine the target standard disease name based on the similarity.

[0173] In this embodiment, if the target clause is not retrieved in the preset disease library and the preset disease synonym library, the trained text matching model can be used to standardize the disease name of the target clause by similarity calculation.

[0174] The text matching model is a pre-trained model trained with sentence-BERT (sentence-Bidirectional Encoder Representation from Transformers). It's understandable that sentence-BERT is a twin network based on the pre-trained BERT, capable of obtaining semantically meaningful paragraph vectors. Sentence-BERT can be used to address the significant time overhead of BERT semantic similarity retrieval and the unsuitability of its sentence representations for unsupervised tasks such as clustering and sentence similarity calculation. Sentence-BERT uses a twin network structure to obtain vector representations of sentence pairs, and then pre-trains a similarity model, which is sentence-BERT.

[0175] Specifically, each standard disease name and disease synonym is used as a similar disease name. Using the pre-trained model trained with sentence-BERT, the similarity between the target clause and each similar disease name can be calculated, and then the target standard disease name corresponding to the target clause can be determined.

[0176] Furthermore, in another disease name standardization method, a target standard disease name is determined based on similarity, specifically including:

[0177] Step 701, determining a maximum similarity with the largest value among multiple similarities;

[0178] Step 702: If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is a standard disease name, then the similar disease name corresponding to the maximum similarity is determined as the target standard disease name;

[0179] Step 703: If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is a disease synonym, then the standard disease name corresponding to the similar disease name corresponding to the maximum similarity is determined as the target standard disease name;

[0180] Step 704: If the maximum similarity is less than a preset threshold, a prompt message indicating that the target standard disease name is not matched is output.

[0181] In this embodiment, after obtaining the similarity between the target clause and each similar disease name, the target standard disease name is determined according to the similarity.

[0182] It can be understood that the higher the similarity between the target clause and a similar disease name, the closer the semantics represented by the target clause and the similar disease name are. Therefore, the similarity with the largest value, that is, the maximum similarity, is first determined. The similar disease name corresponding to the maximum similarity is more likely to be the disease name to be expressed by the target clause.

[0183] On this basis, the maximum similarity is compared with a preset threshold. If the maximum similarity is greater than the preset threshold, it can be considered that the corresponding similar disease name is likely to be the disease name expressed in the target clause. At this point, the standard disease name corresponding to the similar disease name can be determined as the target standard disease name.

[0184] Since each standard disease name and disease synonym is used as a similar disease name in the previous step, in this step, it is determined whether the similar disease name corresponding to the maximum similarity is a standard disease name or a disease synonym.

[0185] If the similar disease name corresponding to the maximum similarity is a standard disease name, then the similar disease name is directly used as the target standard disease name; if the similar disease name corresponding to the maximum similarity is a disease synonym, that is, the similar disease name is not a standard disease name, then the standard disease name corresponding to the disease synonym is used as the target standard disease name.

[0186] Furthermore, if the maximum similarity is less than a preset threshold, the probability that the corresponding similar disease name is the disease name intended by the target clause is low, and the disease name intended by the target clause may be a new disease name. Therefore, a prompt message indicating that the target standard disease name was not matched is output to prompt the staff to further confirm.

[0187] This embodiment uses a similarity calculation method to standardize diseases for target clauses that have not been retrieved in the preset disease library and the preset disease synonym library. Taking into account that the target clause may represent a new disease that has not appeared in the preset disease library and the preset disease synonym library, a preset threshold is set, which is applicable to the situation where a new underwriting disease name is added.

[0188] Further, Figure 3 The figure shows a flow chart of determining the target standard disease name of another disease name standardization method. As shown in the figure, the combined and spliced ​​clauses are the complete target clauses after completion. The complete target clause is used as the disease name. The disease name is first retrieved in the disease database. If it is retrieved, the retrieved name is directly used as the target standard disease name. If it is not retrieved in the disease database, the disease name is retrieved in the synonym database. If it is retrieved in the synonym database, the standard disease name corresponding to the retrieved synonym disease is used as the target standard disease name. If it is not retrieved in both the disease database and the synonym database, the disease standardization is achieved using the pre-trained model trained by sentence-BERT. Specifically, the model is used to obtain the vector representations of the target clause, the standard disease name in the disease database, and the synonym disease in the synonym database, and then the vector representation is used to calculate the cosine similarity between the target clause and each standard disease name and synonym to obtain the maximum similarity. If the maximum similarity is not greater than the threshold, it is considered that the disease name corresponding to the target clause cannot be mapped to the database and may be a new disease name. If the maximum similarity is greater than the threshold and corresponds to a standard disease name in the disease database, the standard disease name is used as the target standard disease name. If the maximum similarity is greater than the threshold and corresponds to a synonym disease in the synonym database, the standard disease name corresponding to the synonym disease is used as the target standard disease name.

[0189] Furthermore, in another disease name standardization method, assume that the target sentence is "breast nodule, mass, cyst or tumor; breast duct dilatation".

[0190] First, according to punctuation marks such as semicolons, commas, and brackets, the target sentence is divided into multiple target clauses as follows: ['breast nodule', 'lump', 'cyst', 'tumor', 'breast duct dilatation'].

[0191] For each target clause of the above segmentation, word segmentation is performed based on the pre-built vocabulary to obtain a word list: [['breast', 'nodule'], ['lump'], ['cyst'], ['tumor'], ['mammary duct', 'dilatation']]; then based on the pre-built part-of-speech dictionary, the part of speech of each word in the previous clause is obtained: [['part', 'disease'], ['disease'], ['disease'], ['suffix'], ['part', 'suffix']].

[0192] Because "lump", "cyst", and "tumor" are not independent diseases, they can be preceded by a part of the body to form a disease. In other words, "lump", "cyst", and "tumor" are incomplete target clauses. Therefore, "breast" can be combined with "lump", "cyst", and "tumor" respectively to complete the incomplete target clauses. The completion results are as follows: [['breast', 'nodule'], ['breast', 'lump'], ['breast', 'cyst'], ['breast', 'tumor'], ['mammary duct', 'dilatation']], and the final target clause list is: ['breast nodule', 'breast lump', 'breast cyst', 'breast tumor', 'mammary duct dilatation']

[0193] Since "breast nodules", "breast cysts" and "breast duct dilatation" are all diseases in the disease library, standardized results can be obtained directly; "breast mass" is neither in the disease library nor in the synonym library. Therefore, after obtaining the vector representation of "breast mass" through the pre-trained model obtained by sentence-BERT training, the cosine similarity is calculated with the vectors corresponding to all disease names in the disease library and synonym library. The highest similarity score with "breast mass" is 0.95, and its corresponding disease name is "breast mass". Since "breast mass" is a disease in the disease library, the standardized disease name corresponding to "breast mass" is "breast mass", that is, the target standard disease name of "breast mass" is "breast mass".

[0194] Furthermore, as a specific implementation of the above-mentioned disease name standardization method, the embodiment of the present application provides a disease name standardization device, such as Figure 4 As shown, the disease name standardization device includes: a first segmentation module, a second segmentation module, a splicing module and a matching module.

[0195] The first segmentation module is used to segment the target sentence into one or more target clauses according to preset wildcards;

[0196] A second segmentation module is used to segment each target clause into one or more words, and determine the part of speech of each word respectively, and determine the part of speech set of the target clause according to the part of speech of the word;

[0197] The splicing module is used to determine whether the target clause is complete based on the part-of-speech set; if it is incomplete, the target clause is supplemented based on the missing part-of-speech set;

[0198] The matching module is used to determine the target standard disease name corresponding to each target clause based on the standard disease names in the preset disease library.

[0199] In a specific application scenario, optionally, the first segmentation module is specifically used to:

[0200] Determine the position of the preset wildcard in the target sentence;

[0201] If the position of the preset wildcard is not within the brackets, the position of the preset wildcard is determined to be the first segmentation position;

[0202] Segmenting the target sentence into one or more middle clauses based on the first segmentation position;

[0203] If the middle clause does not contain parentheses, the middle clause is treated as a target clause;

[0204] If the middle clause contains brackets and the brackets do not contain a preset wildcard, the position of the brackets is determined to be the second segmentation position;

[0205] If the middle clause contains brackets and the brackets contain a preset wildcard, the position of the brackets and the position of the preset wildcard in the brackets are determined as the second segmentation position;

[0206] The middle clause is segmented into one or more target clauses based on the second segmentation position.

[0207] In a specific application scenario, optionally, the part-of-speech category includes a prepositive category, a postpositive category, and an independent category;

[0208] Splicing module, specifically used for:

[0209] Determine the part-of-speech category of each part-of-speech in the part-of-speech set to obtain a category set;

[0210] If the category set includes independent categories, the target clause is determined to be complete;

[0211] If the category set includes the preceding category and the following category, the target clause is determined to be complete;

[0212] Otherwise, the target clause is determined to be incomplete.

[0213] In a specific application scenario, optionally, the splicing module is also used to:

[0214] Determine the arrangement order of multiple target clauses according to the position of each target clause in the target sentence;

[0215] Determine the target clause corresponding to the part-of-speech set as the clause to be completed;

[0216] If the part-of-speech set does not include the preceding category, the preceding category is treated as the missing category, and the target clause that precedes the clause to be completed is used as the completion template clause;

[0217] If the part-of-speech set does not include a post-positioned category, the post-positioned category is treated as a missing category, and the target clause arranged after the clause to be completed is used as the completion template clause;

[0218] Determine whether each completion template clause is complete in order from closest to farthest from the clause to be completed;

[0219] If it is complete, then in the part-of-speech set corresponding to the completion template clause, determine the part-of-speech category as the missing category as the part-of-speech set missing part-of-speech, and use the words corresponding to the part-of-speech set missing part-of-speech to supplement the clause to be completed, and exit the step of judging whether each completion template clause is complete in turn.

[0220] In a specific application scenario, the matching module is optionally used to:

[0221] Search the preset disease database for the standard disease name corresponding to the target clause. If found, use the standard disease name as the target standard disease name.

[0222] If not found, search for a disease synonym corresponding to the target clause in a preset disease synonym library, wherein the preset disease synonym library includes disease synonyms and standard disease names corresponding to the disease synonyms;

[0223] If a disease synonym corresponding to the target clause is found, the standard disease name corresponding to the disease synonym is determined as the target disease name.

[0224] In a specific application scenario, optionally, the matching module is further used to:

[0225] If no disease synonyms corresponding to the target clause are found in the preset disease synonym library, the standard disease name and disease synonyms are used as similar disease names, and the trained text matching model is used to calculate the similarity between the target clause and each similar disease name;

[0226] The target standard disease name is determined based on the similarity.

[0227] In a specific application scenario, optionally, the matching module is further used to:

[0228] Among multiple similarities, determining a maximum similarity with the largest value;

[0229] If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is a standard disease name, then the similar disease name corresponding to the maximum similarity is determined to be the target standard disease name;

[0230] If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is a disease synonym, then the standard disease name corresponding to the similar disease name corresponding to the maximum similarity is determined as the target standard disease name;

[0231] If the maximum similarity is less than the preset threshold, a prompt message indicating that the target standard disease name is not matched is output.

[0232] According to another aspect of the present application, a storage medium is provided, on which a computer program is stored. When the program or instructions are executed by a processor, the above-mentioned disease name standardization method is implemented.

[0233] According to another aspect of the present application, an electronic device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the above-mentioned disease name standardization method when executing the program.

[0234] It should be noted that for other corresponding descriptions of the functional modules involved in the disease name standardization device provided in the embodiment of the present application, please refer to Figures 1 to 3 The corresponding description in will not be repeated here.

[0235] Based on the above Figures 1 to 3 The method shown in FIG. 1 is a method for performing the above-mentioned operation. Accordingly, the embodiment of the present application further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned operation is performed. Figures 1 to 3 The disease name standardization method shown.

[0236] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling an electronic device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0237] Based on the above Figures 1 to 3 The method shown, and Figure 4 In order to achieve the above-mentioned purpose, the embodiment of the disease name standardization device shown in the embodiment of the present application also provides an electronic device, which can be a personal computer, a server, a network device, etc. The electronic device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figures 1 to 3 The disease name standardization method shown.

[0238] Optionally, the electronic device may further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a Wi-Fi module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a Wi-Fi interface), etc.

[0239] Those skilled in the art will understand that the electronic device structure provided in this embodiment does not limit the electronic device, and may include more or fewer components, or combine certain components, or arrange the components differently.

[0240] The storage medium may also include an operating device and a network communication module. The operating device is a program that manages and stores the hardware and software resources of the electronic device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various controls within the storage medium and with other hardware and software in the physical device.

[0241] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform, or by hardware.

[0242] Those skilled in the art will understand that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the units or processes in the accompanying drawings are not necessarily required to implement the present application. Those skilled in the art will understand that the units in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The units of the above-mentioned implementation scenario can be combined into one unit, or can be further split into multiple sub-units.

[0243] The serial numbers of the above application are for descriptive purposes only and do not represent the advantages or disadvantages of the implementation scenarios. The above disclosure only discloses several specific implementation scenarios of the present application, but the present application is not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present application.

Claims

1. A method for standardizing disease names, characterized in that: The method comprises: Split the target sentence into one or more target clauses based on the preset wildcards; Segmenting each target clause into one or more words, determining the part of speech of each word, and determining the part of speech set of the target clause according to the part of speech of the word; Determine the part-of-speech category of each part of speech in the part-of-speech set to obtain a category set; wherein the part-of-speech category of the part-of-speech includes a pre-category, a post-category, and an independent category; If the category set includes the independent category, then the target clause is determined to be complete; if the category set includes the preceding category and the following category, then the target clause is determined to be complete; otherwise, the target clause is determined to be incomplete; If the target clause is incomplete, supplementing the target clause according to the missing parts of speech in the part-of-speech set; Determine the target standard disease name corresponding to each target clause according to the standard disease name in the preset disease library; The step of supplementing the target clause according to the missing parts of speech in the part-of-speech set specifically includes: determining an arrangement order among the plurality of target clauses according to a position of each target clause in the target sentence; Determining a target clause corresponding to the part-of-speech set as a clause to be completed; If the part-of-speech set does not include the preceding category, the preceding category is regarded as a missing category, and the target clause arranged before the clause to be completed is regarded as a completion template clause; If the part-of-speech set does not include the post-position category, the post-position category is used as the missing category, and the target clause arranged after the clause to be completed is used as the completion template clause; Determine whether each of the completion template clauses is complete in descending order of distance from the clause to be completed; If it is complete, then in the part-of-speech set corresponding to the completion template clause, determine that the part-of-speech category is the part-of-speech of the missing category and is the part-of-speech missing from the part-of-speech set, and use the words corresponding to the part-of-speech missing from the part-of-speech set to supplement the clause to be completed, and exit the step of judging whether each of the completion template clauses is complete in turn.

2. The method according to claim 1, characterized in that The step of dividing the target sentence into one or more target clauses according to the preset wildcards specifically includes: Determining a position of the preset wildcard in the target sentence; If the position of the preset wildcard is not within the brackets, the position of the preset wildcard is determined to be the first segmentation position; Segmenting the target sentence into one or more middle clauses based on the first segmentation position; If the middle clause does not contain brackets, the middle clause is treated as a target clause; If the middle clause contains the bracket and the bracket does not contain the preset wildcard, determining the position of the bracket as the second segmentation position; If the middle clause contains the bracket and the bracket contains the preset wildcard, determining the position of the bracket and the position of the preset wildcard in the bracket as the second segmentation position; The middle clause is segmented into one or more target clauses based on the second segmentation position.

3. The method according to claim 1, characterized in that Determining the target standard disease name corresponding to each target clause based on the standard disease nouns in the preset disease library specifically includes: Searching for a standard disease name corresponding to the target clause in the preset disease library, and if found, using the standard disease name as the target standard disease name; If not found, searching for a disease synonym corresponding to the target clause in a preset disease synonym library, wherein the preset disease synonym library includes disease synonyms and standard disease names corresponding to the disease synonyms; If a disease synonym corresponding to the target clause is found, the standard disease name corresponding to the disease synonym is determined to be the target standard disease name.

4. The method according to claim 3, further comprising: If no disease synonyms corresponding to the target clause are found in the preset disease synonym library, the standard disease name and the disease synonyms are used as similar disease names, and the similarity between the target clause and each of the similar disease names is calculated using the trained text matching model; The target standard disease name is determined according to the similarity.

5. The method according to claim 4, characterized in that Determining the target standard disease name according to the similarity specifically includes: Determining a maximum similarity with the largest value among the plurality of similarities; If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is the standard disease name, then determining the similar disease name corresponding to the maximum similarity as the target standard disease name; If the maximum similarity is greater than a preset threshold, and the similar disease name corresponding to the maximum similarity is a synonym of the disease, then determining the standard disease name corresponding to the similar disease name corresponding to the maximum similarity as the target standard disease name; If the maximum similarity is less than the preset threshold, a prompt message indicating that the target standard disease name is not matched is output.

6. A disease name standardization device, characterized in that: The device comprises: The first segmentation module is used to segment the target sentence into one or more target clauses according to preset wildcards; a second segmentation module, configured to segment each target clause into one or more words, and determine the part of speech of each word, and determine the part of speech set of the target clause according to the part of speech of the word; a splicing module, configured to determine the part-of-speech category of each part of speech in the part-of-speech set to obtain a category set; wherein the part-of-speech category of the part-of-speech includes a pre-category, a post-category, and an independent category; if the category set includes the independent category, then the target clause is determined to be complete; if the category set includes the pre-category and the post-category, then the target clause is determined to be complete; otherwise, the target clause is determined to be incomplete; if the target clause is incomplete, then the target clause is supplemented according to the missing part of speech in the part-of-speech set; A matching module, configured to determine the target standard disease name corresponding to each target clause according to the standard disease names in a preset disease library; Wherein, the splicing module is also used for: determining an arrangement order among the plurality of target clauses according to a position of each target clause in the target sentence; Determining a target clause corresponding to the part-of-speech set as a clause to be completed; If the part-of-speech set does not include the preceding category, the preceding category is regarded as a missing category, and the target clause arranged before the clause to be completed is regarded as a completion template clause; If the part-of-speech set does not include the post-position category, the post-position category is used as the missing category, and the target clause arranged after the clause to be completed is used as the completion template clause; Determine whether each of the completion template clauses is complete in descending order of distance from the clause to be completed; If it is complete, then in the part-of-speech set corresponding to the completion template clause, determine that the part-of-speech category is the part-of-speech of the missing category and is the part-of-speech missing from the part-of-speech set, and use the words corresponding to the part-of-speech missing from the part-of-speech set to supplement the clause to be completed, and exit the step of judging whether each of the completion template clauses is complete in turn.

7. A storage medium having a program or instruction stored thereon, characterized in that: When the program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

8. An electronic device, characterized in that: The method comprises a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 5 when executing the program.

Images