Similar medical record retrieval method, device, equipment and storage medium

Abstract

The present application proposes a method, device, equipment and storage medium for retrieving similar medical records, which relates to the field of artificial intelligence technology, and in particular to the field of natural language processing and intelligent search technology. The specific implementation scheme is: receiving a search term; performing element recognition on the search term to obtain corresponding elements and element attributes; performing intent recognition on the search term to obtain the search intent; performing a search based on the search term, elements, and the search intent to obtain candidate search content; identifying a target stem and target options from the candidate search content based on the element attributes and the search intent; and performing a search based on the target stem and target options to obtain a target similar medical record. According to the present application, the efficiency of retrieving similar medical records and the accuracy of matching can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to the field of natural language processing and intelligent search technology. Specifically, the present application proposes a method, device, equipment and storage medium for retrieving similar medical records. Background Art

[0002] With the development of the Internet industry, intelligent search services and natural language processing are widely used in the medical field. Natural language processing includes natural language understanding and natural language generation, and is an important research direction in the fields of computer science and artificial intelligence. In the search scenario, users search for medical-related keywords to obtain relevant medical content, such as obtaining treatment or prevention methods through symptom search.

[0003] Currently, when users search for medical records based on medical-related keywords, the search results obtained often do not meet user needs due to the existence of massive amounts of medical data, making it difficult to accurately obtain the required search results and resulting in low medical record retrieval efficiency. Summary of the invention

[0004] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0005] To this end, the present application proposes a method, device, equipment and storage medium for retrieving similar medical records.

[0006] The first embodiment of the present application proposes a method for retrieving similar medical records, including:

[0007] receiving a search term;

[0008] Performing element recognition on the search term to obtain corresponding elements and element attributes;

[0009] Performing intent recognition on the search term to obtain the search intent;

[0010] Performing a search based on the search term, the element, and the search intent to obtain candidate search content;

[0011] According to the element attributes and the search intent, identifying a target topic and target options from the candidate search contents; and

[0012] According to the target question stem and the target options, a search is performed to obtain target similar medical records.

[0013] The second embodiment of the present application provides a similar medical record retrieval device, comprising:

[0014] A receiving module, used for receiving a search term;

[0015] An element recognition module, used to perform element recognition on the search term to obtain corresponding elements and element attributes;

[0016] An intention recognition module, used to perform intention recognition on the search term to obtain the search intention;

[0017] A first search module, used for searching according to the search term, the element, and the search intent to obtain candidate search content;

[0018] A determination module, configured to identify a target topic and target options from the candidate search contents according to the element attributes and the search intent; and

[0019] The second retrieval module is used to retrieve target similar medical records based on the target question stem and the target options.

[0020] The third aspect embodiment of the present application proposes an electronic device, comprising at least one processor, and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the similar medical record retrieval method as described in the first aspect embodiment.

[0021] The fourth aspect of the present application provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the similar medical record retrieval method as described in the first aspect of the present application.

[0022] One embodiment of the above application has the following advantages or beneficial effects: by receiving a search term, performing element recognition on the search term to obtain the corresponding element and element attribute, and performing intent recognition on the search term to obtain the search intent. Further, a search is performed based on the search term, element, and search intent to obtain candidate search content, and a target stem and target options are identified from the candidate search content based on the element attribute and the search intent, and a search is performed based on the target stem and target options to obtain a target similar medical record, which can improve the efficiency of similar medical record retrieval and the matching accuracy.

[0023] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present application.

[0025] Figure 1A flowchart of a method for retrieving similar medical records provided in an embodiment of the present application;

[0026] Figure 2 A flowchart of another similar medical record retrieval method provided in an embodiment of the present application;

[0027] Figure 3 A flowchart of another similar medical record retrieval method provided in an embodiment of the present application;

[0028] Figure 4 A flowchart of another similar medical record retrieval method provided in an embodiment of the present application;

[0029] Figure 5 A schematic diagram of the structure of a similar medical record retrieval device provided in an embodiment of the present application;

[0030] Figure 6 A schematic diagram of the structure of another similar medical record retrieval device provided in an embodiment of the present application;

[0031] Figure 7 A block diagram of an exemplary electronic device suitable for implementing the embodiments of the present application is shown. DETAILED DESCRIPTION

[0032] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0033] Figure 1 A flowchart of a method for retrieving similar medical records provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes:

[0034] Step 101, receiving a search term.

[0035] The method of the embodiment of the present application can be applied to a search scenario of medical consultation. In this embodiment, in order to realize the search of similar medical records, a search term can be received to obtain similar medical records according to the search term.

[0036] The search term may be input by the user, and the user may input the search term in a manner including but not limited to voice input, on-screen input, etc. The user may conduct a medical consultation by inputting the search term, for example, the received search term may be "my stomach hurts a little".

[0037] Step 102: perform element recognition on the search term to obtain corresponding elements and element attributes.

[0038] In this embodiment, the element attribute may be the content contained in the search term, for example, the element attribute may be stomachache, lung cancer surgery, etc. The element is used to indicate the category of the element attribute, for example, the element may include surgery, disease, part, symptom, attribute degree, etc. The elements involved in the search term and the element attributes corresponding to the elements are obtained by performing element recognition on the search term. For example, for the search term "stomachache a little bit", the element-element attributes obtained are as follows: symptom-stomachache, attribute degree-a little bit.

[0039] There are many ways to implement element recognition for search terms.

[0040] As an example, a neural network model is pre-trained, the input of the neural network model is text, and the output is features and feature attributes corresponding to the text.

[0041] As another example, text is recognized by rule matching to obtain elements and element attributes corresponding to the text.

[0042] Step 103: perform intent recognition on the search term to obtain the search intent.

[0043] In this embodiment, the search intent is used to represent the user's search needs. After receiving the search term, the search intent is identified based on the search term to obtain the search intent corresponding to the search term. For example, the search intent may include asking about medicine, asking about prices, asking about similar medical records, etc.

[0044] As a possible implementation method, the intent of the search term is identified through an intent classifier to obtain the search intent. In this case, historical search terms are collected and the intent of the historical search terms is annotated. The historical search terms and the annotated intent are used as a training set. The intent classifier is trained based on the training set. The input of the intent classifier is the search term, and the output is the intent category. The intent classifier can be implemented through a neural network. Then, when the user enters a search term online, the intent of the search term is identified through the pre-trained intent classifier to obtain the search intent corresponding to the search term.

[0045] It should be noted that the above-mentioned implementation method of obtaining the retrieval intent is only an example. Intent identification can be performed according to needs through rule classification based on dictionary templates, log matching, classification model, etc., and no specific limitation is made here.

[0046] Step 104, search based on the search terms, elements, and search intent to obtain candidate search content.

[0047] In this embodiment, a search is performed in a large amount of data based on the search terms, elements, and search intent to obtain candidate search content. The candidate search content includes but is not limited to articles, conversation content, etc. For example, if the search term is "my stomach hurts a little", the candidate search content may include records of medical consultation conversations about the symptom of stomach pain.

[0048] As an example, based on the inverted text search method, candidate search contents are obtained according to search terms, elements, and search intent.

[0049] As another example, a retrieval method based on semantic generalization ANN (Approximate Nearest Neighbor) obtains candidate retrieval content according to retrieval terms, elements, and retrieval intent.

[0050] Thus, based on the above method, articles or conversation contents related to the user's search terms are obtained from massive data.

[0051] In one embodiment of the present application, historical medical record information may be collected, including but not limited to consultation and question-and-answer data, and the historical medical record information may be processed offline to obtain structured data, wherein the structured data may include, for example: the chief complaint of the initial diagnosis, the history of the present illness (including the attributes such as the yin and yang of the disease, symptoms, and signs, and duration), tests, past medical history, and family history, etc. In this embodiment, the structured data of the historical medical record information is obtained through offline processing, and in the online process, the massive data is searched through search terms, elements, and search intents, and then the structured data in the search results are matched, and the structured data in the search results are used as candidate search content.

[0052] Step 105, identifying target questions and target options from candidate search contents according to element attributes and search intent.

[0053] In this embodiment, a candidate intent set is pre-set, and the candidate intent set includes preset candidate intents, and the number of candidate intents can be one or more. If the search intent matches a candidate intent in the candidate intent set, it triggers the identification of the target stem and target options from the candidate search content according to the element attributes.

[0054] For example, the candidate intent set includes the intent to retrieve similar medical records. When the identified retrieval intent is to retrieve similar medical records, it triggers the operation of identifying the target stem and target options from the candidate retrieval content based on the element attributes. Otherwise, the above operation of identifying the target stem and target options from the candidate retrieval content based on the element attributes is not triggered.

[0055] The target stem can be used to describe the causes and scenarios of some symptoms, and may include the types of occurrences corresponding to the elements and element attributes, and the target options describe the case segmentation types associated with the element attributes. There can be one or more target stems, and each target stem may correspond to one or more target options.

[0056] For example, the element-element attribute is symptom-headache. According to the predefined set, the corresponding question stems include cold headache, diabetic headache, cold headache, etc. For the question stem of diabetic headache, the options are primary diabetes and secondary diabetes.

[0057] Step 106, searching for target similar medical records based on the target question stem and target options.

[0058] In this embodiment, after determining the target stem and target options, the target stem and target options are provided to the user for the user to filter. When the user selects the stem and option, the candidate search content is secondary searched according to the stem and option selected by the user to obtain the target similar medical records corresponding to the search term.

[0059] The method for retrieving similar medical records in the embodiment of the present application receives a search term, identifies the elements of the search term to obtain the corresponding elements and element attributes, and identifies the intent of the search term to obtain the search intent. Then, a search is performed based on the search term, elements, and search intent to obtain candidate search content, and a target stem and target options are identified from the candidate search content based on the element attributes and the search intent, and a search is performed based on the target stem and target options to obtain target similar medical records. According to the present application, the efficiency of similar medical record retrieval can be improved, guidance can be provided to users based on their medical record retrieval needs, and similar medical records can be accurately matched in massive data.

[0060] It should be noted that the above description of similar medical record retrieval is exemplary, and the method of the embodiment of the present application can also be used for retrieval of other archival files.

[0061] Based on the above embodiments, the method of the embodiments of the present application can realize the identification of continuous element types and non-continuous element types when performing element identification on the search term. Figure 2 This is a flow chart of another similar medical record retrieval method provided in an embodiment of the present application. In this method, the above step 102 includes:

[0062] Step 201 , semantic recognition is performed on the search term, and the element classification corresponding to the search term is determined according to the semantic recognition result.

[0063] The element classification includes continuous element type and non-continuous element type. The search terms of the continuous element type are, for example, "lung cancer surgery", and the search terms of the non-continuous element type are, for example, "stomach ache, and swelling".

[0064] In this embodiment, the search term can be semantically recognized by using relevant natural language processing technology, wherein the input can be each character of the search term and the position feature corresponding to each character, and the output is the semantic recognition result. Further, according to the semantic recognition result, it is determined whether the search term corresponds to a continuous element type or a non-continuous element type, and after determining the element classification, the recognition method corresponding to the element classification is used to obtain the element and the element attribute, as shown in the following steps 202 and 203.

[0065] Step 202, when the element classification corresponding to the search term is a continuous element type, the semantic recognition result is input into the pre-trained first artificial intelligence model to obtain the elements and element attributes output by the first artificial intelligence model.

[0066] In this embodiment, first training sample data is obtained, and the first training sample data includes, for example, sample text of a continuous element type, corresponding annotated elements and element attributes, and a first artificial intelligence model is pre-trained based on the first training sample data.

[0067] Among them, the first pre-trained artificial intelligence model can be an entity recognition model based on a neural network, and the implementation methods include but are not limited to: a sequentially connected bidirectional LSTM (Long Short-Term Memory) and a flexible maximum transfer function softmax logistic regression model, a biGRU (Gate Recurrent Unit) and a flexible maximum transfer function softmax model, a pre-trained model and a flexible maximum transfer function softmax model, a pre-trained model and a CRF (Conditional Random Fields) model, and a CRF model. Thus, the problem of word embedding can be solved, for example, for lung cancer surgery, the analysis is surgery-lung cancer surgery, disease-lung cancer, and part-lung.

[0068] Step 203: When the element classification corresponding to the search term is a non-continuous element type, the elements and element attributes are identified by rule matching.

[0069] In this embodiment, rule matching is used to obtain elements and element attributes corresponding to search terms of non-continuous element types. For example, for the search term "stomach ache and swelling", the element-element attributes obtained are: symptoms-stomach ache, symptoms-swollen stomach.

[0070] In this embodiment, it is possible to identify the element classification of the search terms, select a neural network model or rule matching method based on the element classification, and realize element-element attribute recognition of search terms of continuous element types and non-continuous element types, thereby ensuring the accuracy of element and element attribute recognition and further improving the accuracy of similar medical record retrieval.

[0071] In practical applications, different users have different expression habits. The search terms input by users are colloquial and have large differences in expression, so they need to be normalized.

[0072] In one embodiment of the present application, after performing element identification on the search terms to obtain the corresponding elements and element attributes, it also includes: performing entity normalization processing on the corresponding elements and element attributes to obtain standard elements and standard element attributes; retrieving candidate search content based on the search terms, standard elements, and search intent; and identifying target questions and target options from the candidate search content based on the standard element attributes.

[0073] Among them, there are many ways to perform entity normalization. Optionally, entity normalization is achieved based on dictionary matching, for example, the stomach is normalized to the abdomen. Optionally, entity normalization is achieved based on rules. For example, the element-element attribute corresponding to "my stomach hurts a little" is "symptoms-stomach pain, attribute degree-a little", and "abdominal pain" is obtained by "removing the attribute degree words" and performing "dictionary matching" according to the rule. Optionally, synonyms are normalized by semantic generalization. By establishing a standard database, parsing the word slot components and the concept library to calculate the correlation, the standard elements and standard element attributes that are most similar to the elements and element attributes are obtained. Optionally, regular expressions are used to convert time, numerical values, etc. into standardized formats. Therefore, the accuracy of subsequent retrieval and identification of target questions and target options is further improved through the normalization of elements and element attributes.

[0074] The following is an explanation of the target question stem and target options.

[0075] Figure 3 A flowchart of another similar medical record retrieval method provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, in this method, the above step 105 may include:

[0076] Step 301: Identify attribute content that matches the feature attribute from candidate search content.

[0077] In this embodiment, the candidate search content may include attribute content and non-attribute content. The attribute content may include, for example, accompanying symptoms, duration, age, population, suspected diseases, etc.

[0078] Optionally, an attribute set is pre-established, the attribute set includes attribute content, and the attribute content in the candidate search content is determined according to the pre-established attribute set, for example, content related to accompanying symptoms, age, population, etc. is determined from the candidate search content, so as to identify attribute content matching the element attribute. As an example, the candidate search content includes multiple structured data fragments, and the attribute content can be, for example, several structured data fragments matching the element attribute.

[0079] Step 302: perform feature recognition on the attribute content to obtain corresponding attribute features.

[0080] In this embodiment, the implementation methods of feature recognition of attribute content include but are not limited to methods based on extraction of candidate content attributes, extraction based on search log attributes, and extraction based on artificial prior experience. The corresponding method can be selected to extract attribute features according to actual application needs.

[0081] Step 303: input the attribute content and attribute features into a pre-trained second artificial intelligence model to obtain at least one target question stem output by the second artificial intelligence model.

[0082] In this embodiment, the second artificial intelligence model is pre-trained, and the second artificial intelligence model input includes attribute content and attribute features, and the output is at least one target stem. Among them, the at least one target stem output can be arranged in sequence. When the number of target stems output is greater than the preset number, the first N target stems in the sorting are selected. Optionally, the second artificial intelligence model can be trained based on the second training sample data, and the second training sample data includes sample content, sample features, and at least one labeled stem. Among them, the second artificial intelligence model includes but is not limited to a neural network model, a machine learning model, and rules, which are not specifically limited here.

[0083] As an example, the element attribute of the search term is "headache", and attribute content matching the element attribute is identified from the candidate search content, such as accompanying symptoms of headache, suspected diseases, etc., and then input into the second artificial intelligence model based on the attribute content and attribute characteristics. The output target questions include headache caused by cold, headache caused by diabetes, headache caused by cold, etc.

[0084] Step 304, semantically parse the target question stem to obtain target case information corresponding to the target question stem.

[0085] In this embodiment, each target question stem corresponds to target case information, for example, wind-cold headache corresponds to case information of wind-cold headache, and the target case information corresponding to the target question stem is obtained through semantic analysis.

[0086] Step 305, determine the option corresponding to the target case information, and use the corresponding option as the target option.

[0087] The target option describes the case segmentation type associated with the element attribute. Optionally, the relevant options can be recalled according to the target case information. Optionally, a correspondence between the target case information and the options is pre-established, and the option corresponding to the target case information is determined according to the correspondence and used as the target option.

[0088] In one embodiment of the present application, determining the options corresponding to the target case information includes: determining multiple candidate options corresponding to the target case information; determining the co-occurrence probability between each candidate option and the standard option based on the attribute content, the co-occurrence probability being the probability that the candidate option and the standard option exist in the attribute content at the same time; and selecting the corresponding option from the multiple candidate options based on the co-occurrence probability. In this embodiment, the correspondence between the target case information and the multiple candidate options is pre-set, the standard option can be determined based on the entity, and the multiple candidate options are sorted from high to low based on the co-occurrence probability, which can provide more relevant target options and further improve the efficiency of similar case retrieval.

[0089] Optionally, the following features can be extracted and sorted options can be obtained through rules or machine learning. The features include: the frequency of occurrence of candidate content options, the frequency of occurrence of options in similar search terms, co-occurrence information prompted in the input information, the relationship between entities in the knowledge graph, and the number of co-occurrences of entities and options in the candidate content.

[0090] Optionally, in the case where the obtained options may be repeated or contain relationships, the options are normalized and upwardly normalized through the entity relationship dictionary to achieve deduplication and dispersion of the options.

[0091] In this embodiment, attribute content that matches the element attributes is identified from the candidate search content, at least one corresponding target stem is determined according to the attribute content and attribute characteristics, and a target option corresponding to the target case information is determined. The case subdivision type associated with the element attributes is described through the target option. Thus, by performing a secondary search based on the target stem and the target option, similar medical records can be accurately retrieved, thereby improving the retrieval efficiency and matching accuracy of similar medical records.

[0092] The following is an explanation combining offline and online processing.

[0093] Reference Figure 4 , where the offline process includes:

[0094] 1) Obtain historical medical record information, which includes question and answer and consultation data, and extract the elements and element attributes corresponding to the historical medical record information based on the trained parser.

[0095] 2) Perform entity normalization on features and feature attributes.

[0096] 3) Generate structured data based on the content after entity normalization.

[0097] The online process includes:

[0098] 1) Receive the search terms input by the user, and extract the elements and element attributes of the corresponding main complaint based on the search terms and the above-mentioned parser.

[0099] 2) Perform entity normalization on features and feature attributes.

[0100] 3) Perform intent recognition on the search terms to obtain the search intent.

[0101] 4) Search massive data based on search terms, elements, and search intent to obtain candidate search content.

[0102] 5) Generate question stem; 6) Generate options; 7) Perform secondary search based on the question stem and options to obtain similar medical record information.

[0103] According to the method of the embodiment of the present application, the efficiency of retrieving similar medical records can be improved, as well as the matching accuracy, thereby ensuring the effect of medical record retrieval.

[0104] In order to implement the above embodiment, the present application also proposes a similar medical record retrieval device.

[0105] Figure 5 A schematic diagram of a similar medical record retrieval device provided in an embodiment of the present application is shown in FIG. Figure 5 As shown, the device includes: a receiving module 51, an element recognition module 52, an intention recognition module 53, a first retrieval module 54, a determination module 55, and a second retrieval module 56.

[0106] The receiving module 51 is used to receive search terms.

[0107] The element recognition module 52 is used to perform element recognition on the search term to obtain corresponding elements and element attributes.

[0108] The intention recognition module 53 is used to perform intention recognition on the search term to obtain the search intention.

[0109] The first search module 54 is used to perform a search based on the search terms, the elements, and the search intent to obtain candidate search content.

[0110] The determination module 55 is used to identify the target topic and target options from the candidate search content according to the element attributes and the search intent.

[0111] The second retrieval module 56 is used to retrieve target similar medical records according to the target question stem and the target options.

[0112] In one embodiment of the present application, the element identification module 52 includes: a determination unit, used to perform semantic identification on the search term, and determine the element classification corresponding to the search term based on the semantic identification result; an acquisition unit, used to adopt an identification method corresponding to the element classification to acquire the element and element attributes.

[0113] In one embodiment of the present application, the element classification includes: continuous element type, wherein the acquisition unit is specifically used to: when the element classification corresponding to the search term is the continuous element type, the semantic recognition result is input into the pre-trained first artificial intelligence model to obtain the elements and element attributes output by the first artificial intelligence model.

[0114] In one embodiment of the present application, the element classification includes: non-continuous element type, wherein the acquisition unit is specifically used to: when the element classification corresponding to the search term is the non-continuous element type, the element and element attribute are identified by rule matching.

[0115] Figure 6 This is a schematic diagram of the structure of another similar medical record retrieval device provided in an embodiment of the present application. Figure 6 The similar medical record retrieval device shown includes: a receiving module 61, an element recognition module 62, an intention recognition module 63, a first retrieval module 64, a determination module 65, a second retrieval module 66, and a normalization module 67.

[0116] The normalization module 67 is used to perform entity normalization processing on the corresponding elements and element attributes to obtain standard elements and standard element attributes.

[0117] The first search module 64 is specifically used to retrieve candidate search content according to the search term, the standard elements, and the search intent.

[0118] The determination module 65 is specifically used to identify the target topic and target options from the candidate search content according to the standard element attributes.

[0119] In one embodiment of the present application, the determination module 65 is specifically used to: if the retrieval intent matches a candidate intent in the candidate intent set, trigger the identification of a target topic and target options from the candidate retrieval content according to the element attributes.

[0120] In one embodiment of the present application, the determination module 65 includes: a stem determination unit, used to identify attribute content that matches the element attribute from the candidate retrieval content; perform feature recognition on the attribute content to obtain corresponding attribute features; and input the attribute content and the attribute features into a pre-trained second artificial intelligence model to obtain at least one of the target stems output by the second artificial intelligence model.

[0121] In one embodiment of the present application, the determination module 65 also includes: an option determination unit, which is used to perform semantic analysis on the target stem to obtain target case information corresponding to the target stem; determine the option corresponding to the target case information, and use the corresponding option as the target option.

[0122] In one embodiment of the present application, the option determination unit is specifically used to: determine a plurality of candidate options corresponding to the target case information; determine the co-occurrence probability between each of the candidate options and the standard option based on the attribute content, the co-occurrence probability being the probability that the candidate option and the standard option simultaneously exist in the attribute content; and select the corresponding option from the plurality of candidate options based on the co-occurrence probability.

[0123] In one embodiment of the present application, the target option describes the case segmentation type associated with the element attribute.

[0124] In one embodiment of the present application, the pre-trained first artificial intelligence model is a sequentially connected bidirectional LSTM and flexible maximum transfer function softmax logistic regression model.

[0125] It can be understood that the receiving module 61, the element recognition module 62, the intention recognition module 63, the first retrieval module 64, the determination module 65, and the second retrieval module 66 in this embodiment can have the same function and structure as the receiving module 51, the element recognition module 52, the intention recognition module 53, the first retrieval module 54, the determination module 55, and the second retrieval module 56 in the above-mentioned embodiment.

[0126] The explanation of the method for retrieving similar medical records in the aforementioned embodiment is also applicable to the device for retrieving similar medical records in this embodiment, and will not be repeated here.

[0127] The similar medical record retrieval device of the embodiment of the present application receives a search term, identifies the elements of the search term to obtain the corresponding elements and element attributes, and identifies the intent of the search term to obtain the search intent. Then, a search is performed based on the search term, elements, and search intent to obtain candidate search content, and a target stem and target options are identified from the candidate search content based on the element attributes and the search intent, and a search is performed based on the target stem and target options to obtain a target similar medical record. According to the present application, the efficiency of similar medical record retrieval and the matching accuracy can be improved.

[0128] In order to implement the above embodiments, the present application also proposes a computer program product, which implements the similar medical record retrieval method as described in any of the above embodiments when the instructions in the computer program product are executed by a processor.

[0129] According to an embodiment of the present application, the present application also provides an electronic device and a readable storage medium.

[0130] like Figure 7 , is a block diagram of an electronic device according to a method for retrieving similar medical records according to an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0131] like Figure 7 As shown, the electronic device includes: one or more processors 701, a memory 702, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are interconnected using different buses and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In other embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 7 A processor 701 is taken as an example.

[0132] The memory 702 is a non-transient computer-readable storage medium provided in the present application. The memory stores instructions executable by at least one processor to enable the at least one processor to perform the method for retrieving similar medical records provided in the present application. The non-transient computer-readable storage medium of the present application stores computer instructions, which are used to enable a computer to perform the method for retrieving similar medical records provided in the present application.

[0133] The memory 702 is a non-transient computer-readable storage medium that can be used to store non-transient software programs, non-transient computer executable programs and modules, such as program instructions / modules corresponding to the similar medical record retrieval method in the embodiment of the present application (for example, the attached Figure 5 The processor 701 executes various functional applications and data processing of the server by running the non-transient software programs, instructions and modules stored in the memory 702, that is, the retrieval method of similar medical records in the above method embodiment is implemented.

[0134] The memory 702 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 702 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 702 may optionally include a memory remotely arranged relative to the processor 701, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0135] The electronic device of the similar medical record retrieval method may further include: an input device 703 and an output device 704. The processor 701, the memory 702, the input device 703 and the output device 704 may be connected via a bus or other means. Figure 7 The example of connecting through bus is taken in the following.

[0136] The input device 703 can receive input digital or character information, and generate key signal input related to user settings and function control of the electronic device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick and other input devices. The output device 704 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display and a plasma display. In some embodiments, the display device may be a touch screen.

[0137] Various implementations of the systems and techniques described herein can be realized in digital electronic circuit systems, integrated circuit systems, dedicated ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0138] These computer programs (also referred to as programs, software, software applications, or code) include machine instructions for programmable processors and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or means (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0139] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0140] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0141] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS (Virtual Private Server) services.

[0142] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this application can be executed in parallel, sequentially or in different orders, as long as the expected results of the technical solution disclosed in this application can be achieved, and this document is not limited here.

[0143] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for retrieving similar medical records, comprising: receiving a search term input by a user; Performing element recognition on the search term to obtain corresponding elements and element attributes; Performing intent recognition on the search term to obtain the search intent; Performing a search based on the search term, the element, and the search intent to obtain candidate search content; According to the element attributes and the search intent, identifying a target topic and target options from the candidate search contents; as well as According to the target question stem and the target options, a secondary search is performed on the candidate search content to obtain target similar medical records.

2. The method according to claim 1, wherein: The performing element recognition on the search term to obtain corresponding elements and element attributes includes: Performing semantic recognition on the search term, and determining the element classification corresponding to the search term according to the semantic recognition result; The elements and element attributes are obtained by using an identification method corresponding to the element classification.

3. The method according to claim 2, wherein the element classification comprises: A continuous feature type, wherein the adopting of an identification method corresponding to the feature classification to obtain the feature and feature attributes comprises: When the element classification corresponding to the search term is the continuous element type, the semantic recognition result is input into the pre-trained first artificial intelligence model to obtain the elements and element attributes output by the first artificial intelligence model.

4. The method according to claim 2, wherein the element classification comprises: A non-continuous feature type, wherein the use of an identification method corresponding to the feature classification to obtain the feature and feature attributes includes: When the element classification corresponding to the search term is the non-continuous element type, the element and element attribute are identified by rule matching.

5. The method according to claim 1, after performing element recognition on the search term to obtain corresponding elements and element attributes, further comprises: Performing entity normalization processing on the corresponding elements and element attributes to obtain standard elements and standard element attributes; The searching according to the search term, the element, and the search intent to obtain candidate search content includes: Retrieving candidate search content according to the search term, the standard elements, and the search intent; Then, identifying a target topic and target options from the candidate search content according to the element attributes and the search intent includes: According to the standard element attributes, target questions and target options are identified from the candidate search contents.

6. The method according to claim 1, wherein: The step of identifying a target topic and target options from the candidate search contents according to the element attributes and the search intent includes: In response to the search intent matching a candidate intent in the candidate intent set, a trigger is provided to identify a target topic and target options from the candidate search content according to the element attributes.

7. The method according to claim 6, wherein: Identifying the target topic from the candidate search content according to the element attributes includes: Identifying attribute content that matches the element attribute from the candidate search content; Performing feature recognition on the attribute content to obtain corresponding attribute features; and The attribute content and the attribute features are input into a pre-trained second artificial intelligence model to obtain at least one of the target questions output by the second artificial intelligence model.

8. The method according to claim 7, wherein: Identifying the target option from the candidate search content according to the element attribute includes: Performing semantic analysis on the target question stem to obtain target case information corresponding to the target question stem; Determine the option corresponding to the target case information, and use the corresponding option as the target option.

9. The method according to claim 8, wherein: The determining of the options corresponding to the target case information includes: Determining multiple candidate options corresponding to the target case information; Determine, according to the attribute content, a co-occurrence probability between each of the candidate options and the standard option, wherein the co-occurrence probability is a probability that the candidate option and the standard option exist in the attribute content at the same time; and According to the co-occurrence probability, the corresponding option is selected from the multiple candidate options.

10. The method of claim 8, wherein the target option describes a case segmentation type associated with the feature attribute.

11. A device for searching similar medical records, comprising: A receiving module, used for receiving a search term input by a user; An element recognition module, used to perform element recognition on the search term to obtain corresponding elements and element attributes; An intention recognition module, used to perform intention recognition on the search term to obtain the search intention; A first search module, used for searching according to the search term, the element, and the search intent to obtain candidate search content; A determination module, used to identify a target topic and target options from the candidate search content according to the element attributes and the search intent; as well as The second retrieval module is used to perform a secondary search on the candidate search content according to the target question stem and the target options to obtain target similar medical records.

12. The device of claim 11, wherein: The element recognition module comprises: A determination unit, configured to perform semantic recognition on the search term and determine a factor classification corresponding to the search term according to the semantic recognition result; The acquisition unit is used to acquire the elements and element attributes by adopting an identification method corresponding to the element classification.

13. The apparatus according to claim 12, wherein the element classification comprises: Continuous element type, wherein the acquisition unit is specifically used for: When the element classification corresponding to the search term is the continuous element type, the semantic recognition result is input into the pre-trained first artificial intelligence model to obtain the elements and element attributes output by the first artificial intelligence model.

14. The apparatus according to claim 12, wherein the element classification comprises: Non-continuous element type, wherein the acquisition unit is specifically used for: When the element classification corresponding to the search term is the non-continuous element type, the element and element attribute are identified by rule matching.

15. The apparatus of claim 11, further comprising: A normalization module, used for performing entity normalization processing on the corresponding elements and element attributes to obtain standard elements and standard element attributes; The first search module is specifically used to retrieve candidate search content according to the search term, the standard element, and the search intention; The determination module is specifically used to identify the target topic and target options from the candidate search content according to the standard element attributes.

16. The device of claim 11, wherein: The determination module is specifically used for: In response to the search intent matching a candidate intent in the candidate intent set, a trigger is provided to identify a target topic and target options from the candidate search content according to the element attributes.

17. The device of claim 16, wherein: The determination module comprises: A question stem determination unit, used to identify attribute content matching the element attribute from the candidate search content; Performing feature recognition on the attribute content to obtain corresponding attribute features; and The attribute content and the attribute features are input into a pre-trained second artificial intelligence model to obtain at least one of the target questions output by the second artificial intelligence model.

18. The device of claim 17, wherein: The determination module also includes: An option determination unit, used for performing semantic analysis on the target question stem to obtain target case information corresponding to the target question stem; Determine the option corresponding to the target case information, and use the corresponding option as the target option.

19. The device of claim 18, wherein: The option determination unit is specifically used for: Determining multiple candidate options corresponding to the target case information; Determine, according to the attribute content, a co-occurrence probability between each of the candidate options and the standard option, wherein the co-occurrence probability is a probability that the candidate option and the standard option exist in the attribute content at the same time; and According to the co-occurrence probability, the corresponding option is selected from the multiple candidate options.

20. The apparatus of claim 18, wherein the target option describes a case segmentation type associated with the element attribute.

21. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the similar medical record retrieval method described in any one of claims 1-10.

22. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the similar medical record retrieval method according to any one of claims 1 to 10.

23. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method for retrieving similar medical records according to any one of claims 1 to 10.

Images