Medical record text translation method and device based on artificial intelligence, equipment and medium

Abstract

The invention is suitable for the technical field of digital medical treatment, and particularly relates to a medical record text translation method and device based on artificial intelligence, equipment and a medium. According to the method, keywords of associated medical record texts are extracted and spliced to obtain keyword sequences, a target medical record text, M associated medical record texts and M keyword sequences are spliced to obtain a spliced medical record text, text features of the spliced medical record text are extracted, and a target translation direction is determined according to the text features. The text features are decoded to obtain translation features of the target medical record text, linear transformation is carried out on the translation features according to the preset linear transformation matrix corresponding to the target translation direction to obtain the target medical record translation, and the information representation capability of the target medical record text is enhanced by associating the medical record text with the keyword sequence; and decoding and linear transformation are performed on the text features according to the target translation direction in a targeted manner, so that the accuracy of the target medical record translation is improved, and the efficiency of services such as digital service and intelligent interaction in the medical process is improved.

Description

Technical Field

[0001] The present invention is applicable to the field of digital medical technology, and in particular relates to a medical record text translation method, device, equipment and medium based on artificial intelligence. Background Art

[0002] With the development of digital medical technology, in order to facilitate the management of patient information and improve the work efficiency of doctors, hospitals with a large number of patients generally develop internal medical systems for doctors to fill in patient medical records, view patient examination results, handle medical orders, discharge, etc. In order to improve the readability and wide application of medical records, medical records are translated into other languages ​​as needed to facilitate users of multiple languages ​​to browse, learn, communicate and study medical records, which is of great significance to improving the efficiency of digital services, intelligent interaction and other businesses in the medical process.

[0003] The medical record text translation task requires converting the patient's medical records from one text language to another text language. Based on the importance and rigor of patient medical records, as well as the close correlation between multiple medical records of the same patient at different times, and the close correlation between multiple medical records of the same patient for different patients with the same disease, the text translation of the target medical record is performed in combination with the associated medical records of the target medical record, which can effectively improve the accuracy of the translation results.

[0004] In order to draw on the association information between medical records, the current association-aware translation model usually simply splices the associated medical records and the target medical records, and translates the splicing result to obtain the medical record translation of the target medical record, which makes it difficult to extract effective association information from the associated medical records to assist the translation of the target medical record, resulting in low accuracy of the medical record translation of the target medical record, which cannot meet the needs of accurate translation of medical record texts.

[0005] Therefore, in the field of digital medical technology, how to improve the accuracy of medical record text translation, thereby improving the efficiency of digital services, intelligent interactions and other businesses in the medical process has become an urgent problem to be solved. Summary of the invention

[0006] In view of this, an embodiment of the present invention provides an artificial intelligence-based medical record text translation method, device, equipment and medium to solve the problem that the existing association-aware translation model cannot extract effective association information to assist the translation of the target medical record, resulting in low accuracy of the medical record translation of the target medical record.

[0007] In a first aspect, an embodiment of the present invention provides a medical record text translation method based on artificial intelligence, the medical record text translation method comprising: Obtain a target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extract a number of keywords of any associated medical record text, and splice the keywords into a keyword sequence corresponding to the associated medical record text according to the order of the keywords in the associated medical record text, where M is a positive integer; Traversing all the associated medical record texts, respectively obtaining the keyword sequence corresponding to each associated medical record text, obtaining M keyword sequences, and concatenating the target medical record text, the M associated medical record texts, and the M keyword sequences to obtain a concatenated medical record text; Extracting features of the concatenated medical record text to obtain text features of the concatenated medical record text, determining translation directions of the text features, and determining target translation directions of the target medical record text; Obtain a preset vocabulary and a preset linear transformation matrix corresponding to the target translation direction, decode the text features to obtain the translation features of the target medical record text, perform linear transformation on the translation features according to the preset linear transformation matrix, and obtain the target medical record translation of the target medical record text, wherein all texts in the target medical record translation are included in the preset vocabulary.

[0008] In a second aspect, an embodiment of the present invention provides a medical record text translation device based on artificial intelligence, the medical record text translation device comprising: A keyword splicing module is used to obtain a target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extract a number of keywords of any associated medical record text, and splice the keywords into a keyword sequence corresponding to the associated medical record text according to the order of the keywords in the associated medical record text, where M is a positive integer; A text splicing module is used to traverse all the associated medical record texts, obtain the keyword sequence corresponding to each associated medical record text, obtain M keyword sequences, and splice the target medical record text, the M associated medical record texts and the M keyword sequences to obtain a spliced ​​medical record text; A feature extraction module is used to extract features from the spliced ​​medical record text to obtain text features of the spliced ​​medical record text, determine the translation direction of the text features, and determine the target translation direction of the target medical record text; A text translation module is used to obtain a preset vocabulary and a preset linear transformation matrix corresponding to the target translation direction, decode the text features to obtain the translation features of the target medical record text, perform linear transformation on the translation features according to the preset linear transformation matrix, and obtain the target medical record translation of the target medical record text, wherein all texts in the target medical record translation are included in the preset vocabulary.

[0009] In a third aspect, an embodiment of the present invention provides a computer device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the medical record text translation method as described in the first aspect when executing the computer program.

[0010] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the medical record text translation method as described in the first aspect is implemented.

[0011] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: by obtaining a target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extracting a number of keywords of the associated medical record texts, splicing the keywords into a keyword sequence corresponding to the associated medical record text according to the order of the keywords in the associated medical record texts, splicing the target medical record text, the M associated medical record texts and the M keyword sequence to obtain a spliced ​​medical record text, performing feature extraction on the spliced ​​medical record text to obtain text features of the spliced ​​medical record text, judging the translation direction of the text features, and determining the target translation direction of the target medical record text. The preset vocabulary and preset linear transformation matrix corresponding to the target translation direction are obtained, the text features are decoded to obtain the translation features of the target medical record text, and the translation features are linearly transformed according to the preset linear transformation matrix to obtain the target medical record translation of the target medical record text. The information representation ability of the target medical record text is enhanced by associating the medical record text and the keyword sequence, and the corresponding preset vocabulary and preset linear transformation matrix are obtained by determining the target translation direction, and the text features of the target medical record text are decoded and linearly transformed in a targeted manner, thereby improving the accuracy of the translation results and improving the efficiency of digital services, intelligent interaction and other businesses in the medical process. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0013] Figure 1 This is a schematic diagram of an application environment of a medical record text translation method based on artificial intelligence provided in Embodiment 1 of the present invention; Figure 2 It is a flowchart of a medical record text translation method based on artificial intelligence provided in Embodiment 1 of the present invention; Figure 3 It is a structural schematic diagram of a medical record text translation device based on artificial intelligence provided in Embodiment 2 of the present invention; Figure 4 It is a structural diagram of a computer device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0014] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0015] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0016] It should also be understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0017] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]", depending on the context.

[0018] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0019] References to "one embodiment" or "some embodiments" etc. described in the present specification mean that one or more embodiments of the present invention include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0020] The embodiments of the present invention can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0021] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0022] It should be understood that the order of execution of the steps in the following embodiments does not imply a precedence of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0023] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0024] The first embodiment of the present invention provides a medical record text translation method based on artificial intelligence, which can be applied in the following aspects: Figure 1 In an application environment, a client communicates with a server. The client includes but is not limited to a handheld computer, a desktop computer, a laptop computer, a super mobile personal computer, a netbook, a cloud computer device, a personal digital assistant and other computer devices. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0025] See also Figure 2 , is a flowchart of a medical record text translation method based on artificial intelligence provided by the first embodiment of the present invention. The medical record text translation method can be applied to Figure 1In the client, the medical record text translation method may include the following steps: Step S201, obtain the target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extract several keywords of the associated medical record text for any associated medical record text, and splice the several keywords into a keyword sequence corresponding to the associated medical record text according to the order of the several keywords in the associated medical record text.

[0026] Among them, medical records are systematic records of the occurrence, development, diagnosis, and treatment of patients' diseases. They are disease file materials obtained by clinical physicians through interviews, physical examinations, inspections, and detailed observations of the condition. They are summarized, analyzed, organized, and written. They can truly reflect the patient's condition and directly reflect the hospital's medical quality, academic level, and management level. They not only provide valuable basic data for medical treatment, scientific research, and teaching, but also provide indispensable medical information for hospital management.

[0027] This embodiment converts the target medical record text from one text language into another required text language, which can improve the readability and wide application of the target medical record, and facilitate users of multiple languages ​​to browse, study, communicate and research the target medical record, thereby improving the efficiency of digital services, intelligent interaction and other businesses in the medical process.

[0028] Among them, the target medical record text and the M associated medical record texts corresponding to the target medical record text can be multiple medical record texts for the same patient at different times, or multiple medical record texts for different patients with the same disease, etc., which are medical record texts with certain correlation. Therefore, this embodiment combines M associated medical record texts closely related to the target medical record text to improve the information representation ability of the target medical record text, thereby improving the translation accuracy, wherein M is a positive integer.

[0029] Specifically, in order to improve translation efficiency, this embodiment extracts several keywords of the associated medical record text, and splices the several keywords into a keyword sequence corresponding to the associated medical record text according to the order of the several keywords in the associated medical record text, so as to improve the information representation capability of the target medical record text.

[0030] Optionally, the keywords include predicate, predicate issuer, and predicate recipient; Extracting a number of keywords from the associated medical record text, and splicing the keywords into a keyword sequence corresponding to the associated medical record text according to the order of the keywords in the associated medical record text includes: Extract all predicates, predicate issuers and predicate recipients in the associated medical record text, and according to the order of all predicates, predicate issuers and predicate recipients in the associated medical record text, splice all predicates, predicate issuers and predicate recipients into a keyword sequence of the associated medical record text.

[0031] Among them, the keywords in this embodiment include predicate, predicate issuer and predicate recipient, which can effectively capture semantic information in the form of "who did what to whom", and are used to analyze text information such as referential relationships and omitted content in the corresponding associated medical record text to improve the representation ability of the associated medical record text.

[0032] Then, according to the order of the extracted keywords in the associated medical record text, the keywords are spliced ​​into a keyword sequence corresponding to the associated medical record text as a basis for translating the target medical record text, thereby improving the efficiency of information extraction and text translation.

[0033] This embodiment extracts the predicate, predicate issuer and predicate recipient in the associated medical record text to analyze the corresponding text information such as the referential relationship and omitted content in the associated medical record text as the basis for translating the target medical record text, thereby improving the representation ability of the associated medical record text, as well as the efficiency of information extraction and text translation.

[0034] Optionally, the associated medical record text includes the original medical record text and a medical record translation corresponding to the original medical record text; Extract all predicates, predicate issuers and predicate recipients in the associated medical record text, and according to the order of all predicates, predicate issuers and predicate recipients in the associated medical record text, splice all predicates, predicate issuers and predicate recipients into a keyword sequence of the associated medical record text, including: Extract all first predicates, first predicate issuers, and first predicate recipients in the original medical record text, and according to the order of all first predicates, first predicate issuers, and first predicate recipients in the original medical record text, splice all first predicates, first predicate issuers, and first predicate recipients into a first keyword sequence of the original medical record text; Extract all the second predicates, second predicate issuers and second predicate recipients in the medical record translation, and according to the order of all the second predicates, second predicate issuers and second predicate recipients in the medical record translation, splice all the second predicates, second predicate issuers and second predicate recipients into a second keyword sequence of the medical record translation; The first keyword sequence and the second keyword sequence are determined as keyword sequences associated with the medical record text.

[0035] Among them, the associated medical record text in this embodiment includes the original medical record text and the medical record translation corresponding to the original medical record text. The original medical record text and the corresponding medical record translation can assist in representing the text semantic information, text structure information and other related information of the target medical record text. The medical record translation can also assist in representing the translation information of the target medical record text, thereby improving the information representation capability of the target medical record text and further improving the accuracy of the translation results of the target medical record text.

[0036] Correspondingly, when extracting all predicates, predicate issuers and predicate recipients in the associated medical record text, all predicates, predicate issuers and predicate recipients in the original medical record text and the medical record translation are extracted respectively, and all first predicates, first predicate issuers and first predicate recipients in the corresponding original medical record text, and all second predicates, second predicate issuers and second predicate recipients in the corresponding medical record translation are obtained, and then all first predicates, first predicate issuers and first predicate recipients are spliced ​​into the first keyword sequence of the original medical record text in order, and all second predicates, second predicate issuers and second predicate recipients are spliced ​​into the second keyword sequence of the medical record translation. Then, the keyword sequence of the associated medical record text can be determined, including the corresponding first keyword sequence and second keyword sequence, as the basis for assisting the representation of the text semantic information, text structure information and translation information of the target medical record text.

[0037] The associated medical record text in this embodiment includes the original medical record text and the medical record translation corresponding to the original medical record text. The first keyword sequence corresponding to the original medical record text and the second keyword sequence corresponding to the medical record translation are extracted respectively, and the corresponding associated medical record text is used to assist in representing the text semantic information, text structure information and translation information of the target medical record text, thereby improving the information representation capability of the target medical record text and further improving the accuracy of the translation results of the target medical record text.

[0038] The above steps of obtaining the target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extracting several keywords of the associated medical record text for any associated medical record text, and splicing the several keywords into a keyword sequence corresponding to the associated medical record text according to the order of the several keywords in the associated medical record text, combining the M associated medical record texts closely related to the target medical record text to improve the information representation ability of the target medical record text, and using the keyword sequence to replace the corresponding associated medical record text, improve the efficiency of information extraction and text translation, as well as the accuracy of text translation.

[0039] Step S202, traverse all the associated medical record texts, obtain the keyword sequence corresponding to each associated medical record text, obtain M keyword sequences, splice the target medical record text, the M associated medical record texts and the M keyword sequences, and obtain a spliced ​​medical record text.

[0040] Among them, all the related medical record texts are traversed to obtain M keyword sequences corresponding to the M related medical record texts, and then the target medical record text, the M related medical record texts and the M keyword sequences are spliced ​​to obtain the spliced ​​medical record text as the data basis for translating the target medical record text.

[0041] Optionally, the target medical record text, the M associated medical record texts and the M keyword sequences are concatenated to obtain a concatenated medical record text including: Obtaining a preset start delimiter and a sequence end character, and for any associated medical record text and a keyword sequence corresponding to the associated medical record text, concatenating the start delimiter, the keyword sequence, and the associated medical record text in a first preset order to obtain a first concatenated medical record text corresponding to the associated medical record text; Traverse the M associated medical record texts and the M keyword sequences corresponding to the M associated medical record texts to obtain M first concatenated medical record texts; splicing the start delimiter, the target medical record text, and the sequence end character according to a second preset order to obtain a corresponding second spliced ​​medical record text; The M first spliced ​​medical record texts and the second spliced ​​medical record texts are spliced ​​according to a third preset order to obtain a spliced ​​medical record text.

[0042] Among them, in order to improve the readability of the spliced ​​medical record text and facilitate the accurate extraction of feature information corresponding to the target medical record text, M related medical record texts and M keyword sequences, this embodiment obtains a preset starting separator and sequence end character, and splices the starting separator, keyword sequence and related medical record text in a first preset order to obtain a first spliced ​​medical record text, completing the splicing combination of each group of keyword sequences and corresponding related medical record texts; splices the starting separator, target medical record text and sequence end character in a second preset order to obtain a second spliced ​​medical record text, completing the splicing combination of the target medical record text; finally, splices the M first spliced ​​medical record texts and the second spliced ​​medical record text in a third preset order to obtain a spliced ​​medical record text, completing the splicing combination of the target medical record text, M related medical record texts and M keyword sequences.

[0043] Optionally, the first preset sequence includes a fourth preset sequence and a fifth preset sequence, and the keyword sequence includes a first keyword sequence and a second keyword sequence; The start delimiter, the keyword sequence and the original medical record text are concatenated in a first preset order to obtain a first concatenated medical record text corresponding to the associated medical record text, including: splicing the start delimiter, the first keyword sequence, and the associated medical record text according to a fourth preset order to obtain a corresponding first spliced ​​medical record subtext; splicing the start delimiter, the second keyword sequence and the medical record translation according to a fifth preset order to obtain a corresponding second spliced ​​medical record subtext; The first spliced ​​medical record subtext and the second spliced ​​medical record subtext are spliced ​​to obtain a first spliced ​​medical record text corresponding to the associated medical record text.

[0044] Among them, corresponding to the associated medical record text including the original medical record text and the medical record translation corresponding to the original medical record text, the first preset order in this embodiment includes the fourth preset order and the fifth preset order, and the keyword sequence includes the first keyword sequence and the second keyword sequence.

[0045] Correspondingly, when splicing each group of keyword sequences and the corresponding associated medical record text, the starting separator, the first keyword sequence and the associated medical record text are spliced ​​respectively to obtain the first spliced ​​medical record sub-text, the starting separator, the second keyword sequence and the medical record translation are spliced ​​to obtain the second spliced ​​medical record sub-text, and then the first spliced ​​medical record sub-text and the second spliced ​​medical record sub-text are spliced ​​into the corresponding first spliced ​​medical record text, thereby improving the representation ability of the spliced ​​medical record text for the associated medical record text.

[0046] The above steps of traversing all the associated medical record texts, respectively obtaining the keyword sequence corresponding to each associated medical record text, obtaining M keyword sequences, and splicing the target medical record text, M associated medical record texts and M keyword sequences to obtain the spliced ​​medical record text, and fusing the information of the target medical record text, M associated medical record texts and M keyword sequences into the spliced ​​medical record text, can provide additional information such as text position and text type for feature extraction, thereby improving the translation accuracy of the target medical record text.

[0047] Step S203, extracting features from the concatenated medical record text to obtain text features of the concatenated medical record text, determining the translation direction of the text features, and determining the target translation direction of the target medical record text.

[0048] Among them, feature extraction is performed on the spliced ​​medical record text to obtain the text features of the spliced ​​medical record text. The text features simultaneously represent the semantic information of the target medical record text, M related medical record texts and M keyword sequences, the correlation information between them, the translation information between the original text and the corresponding translation, and other effective information, which can serve as the basis for translating the target medical record text.

[0049] Then, the translation direction is determined based on the text features to determine the target translation direction of the target medical record text. For example, when the target medical record text is in Chinese, the target translation direction can be determined to be "Chinese to English". By clarifying the target translation direction and translating the target medical record text in a targeted manner, the rationality of the translation can be ensured and the accuracy of the translation results can be improved.

[0050] In one embodiment, a trained encoder is used to perform feature extraction on the concatenated medical record text to obtain text features of the concatenated medical record text, so as to improve the accuracy of text feature extraction.

[0051] Optionally, determining the translation direction of the text features to determine the target translation direction of the target medical record text includes: Use the trained multi-layer perceptron model to determine the translation direction of text features and obtain the probability value corresponding to the text features; A preset mapping table between translation directions and probability values ​​is obtained, and a target translation direction of the target medical record text is determined according to the probability values ​​and the mapping table.

[0052] Among them, the multi-layer perceptron model is a neural network composed of fully connected layers with at least one hidden layer, which can be used to judge the category of input content. The category of translation direction and the preset mapping table between translation direction and probability value can be set according to actual conditions. For example, the translation direction corresponding to the probability value in [0, 0.5] can be set as "Chinese to English", and the translation direction corresponding to the probability value in (0.5, 1] ​​can be set as "English to Chinese".

[0053] In this embodiment, a trained multi-layer perceptron model is used to judge the translation direction of text features to obtain the probability value corresponding to the text feature, and further based on the probability value and the mapping table, the translation direction corresponding to the probability value in the mapping table is determined as the target translation direction of the target medical record text.

[0054] This embodiment determines the probability value corresponding to the text feature based on the trained multi-layer perceptron model, and combines the preset mapping table between the translation direction and the probability value to determine that the translation direction corresponding to the probability value in the mapping table is the target translation direction of the target medical record text, thereby improving the accuracy of the target translation direction.

[0055] The above steps of extracting features from the spliced ​​medical record text, obtaining text features of the spliced ​​medical record text, judging the translation direction of the text features, and determining the target translation direction of the target medical record text, extracting the text features of the spliced ​​medical record text and clarifying the target translation direction, and translating the target medical record text in a targeted manner can ensure the rationality of the translation and improve the accuracy of the translation results.

[0056] Step S204, obtain the preset vocabulary and preset linear transformation matrix corresponding to the target translation direction, decode the text features, obtain the translation features of the target medical record text, perform linear transformation on the translation features according to the preset linear transformation matrix, and obtain the target medical record translation of the target medical record text. All texts in the target medical record translation are included in the preset vocabulary.

[0057] Among them, the preset vocabulary and the preset linear transformation matrix are set according to the corresponding translation direction. For example, when the translation direction is "Chinese to English", the corresponding preset vocabulary is composed of English text, and the corresponding preset linear transformation matrix is ​​used to perform linear transformation on the input features to obtain the corresponding English translation, and all texts in the English translation are included in the corresponding preset vocabulary; when the translation direction is "English to Chinese", the corresponding preset vocabulary is composed of Chinese text, and the corresponding preset linear transformation matrix is ​​used to perform linear transformation on the input features to obtain the corresponding Chinese translation, and all texts in the Chinese translation are included in the corresponding preset vocabulary.

[0058] Therefore, this embodiment obtains the preset vocabulary and preset linear transformation matrix corresponding to the target translation direction to translate the target medical record text in a targeted manner, which can ensure the rationality of the translation and improve the accuracy of the translation result. Specifically, the text features are first decoded to obtain the translation features of the target medical record text. Then, the translation features are linearly transformed according to the preset linear transformation matrix to obtain the target medical record translation of the target medical record text. Correspondingly, all texts in the target medical record translation are included in the preset vocabulary.

[0059] Optionally, the target medical record text includes N text characters, where N is a positive integer; Decode the text features to obtain the translation features of the target medical record text, including: Use the trained decoder to decode the text features and the preset features to obtain the first translation sub-feature corresponding to the first text character in the target medical record text, where the preset feature is a zero feature sequence; Use the trained decoder to decode the text features and the first n-1 translation sub-features to obtain the nth translation sub-feature corresponding to the nth text character in the target medical record text, n=2, 3, ..., N; According to the N translation sub-features corresponding to the N text characters in the target medical record text, the translation features of the target medical record text are obtained.

[0060] Among them, the target medical record text includes N text characters. The trained decoder in this embodiment is used to decode the N text characters respectively, and decode the next text character according to the decoding features of the previous text character to improve the accuracy of the translation features obtained by decoding.

[0061] Since the first text character does not have a corresponding previous text character, in order to ensure the decoding accuracy of the target medical record text by the decoder, this embodiment obtains a zero feature sequence as a preset feature to replace the decoding feature of the previous text character of the first text character, and finally obtains N translation sub-features corresponding to the N text characters in the target medical record text, and concatenates the N translation sub-features to obtain the translation features of the target medical record text.

[0062] In this embodiment, a trained decoder is used to decode N text characters respectively, and the next text character is decoded according to the decoding feature of the previous text character. Considering that there is no corresponding previous text character for the first text character, a zero feature sequence is obtained as a preset feature to replace the decoding feature of the previous text character of the first text character, thereby improving the accuracy of the translation feature obtained by decoding.

[0063] The above-mentioned steps of obtaining the preset vocabulary and preset linear transformation matrix corresponding to the target translation direction, decoding the text features, obtaining the translation features of the target medical record text, linearly transforming the translation features according to the preset linear transformation matrix, obtaining the target medical record translation of the target medical record text, and all the texts in the target medical record translation are included in the preset vocabulary, obtaining the preset vocabulary and preset linear transformation matrix corresponding to the target translation direction, and specifically translating the target medical record text ensures the rationality of the translation and improves the accuracy of the translation results.

[0064] The embodiment of the present invention obtains a target medical record text and M associated medical record texts corresponding to the target medical record text, extracts a plurality of keywords from the associated medical record texts, splices the plurality of keywords into a keyword sequence corresponding to the associated medical record text according to the order of the plurality of keywords in the associated medical record texts, splices the target medical record text, the M associated medical record texts and the M keyword sequence to obtain a spliced ​​medical record text, extracts features from the spliced ​​medical record text to obtain text features of the spliced ​​medical record text, determines a translation direction for the text features, and determines a target translation direction for the target medical record text, and obtains a translation direction corresponding to the target translation direction. The preset vocabulary and preset linear transformation matrix are used to decode the text features and obtain the translation features of the target medical record text. The translation features are linearly transformed according to the preset linear transformation matrix to obtain the target medical record translation of the target medical record text. The information representation ability of the target medical record text is enhanced by associating the medical record text and the keyword sequence. The corresponding preset vocabulary and preset linear transformation matrix are obtained by determining the target translation direction, and the text features of the target medical record text are decoded and linearly transformed in a targeted manner, which improves the accuracy of the translation results and improves the efficiency of digital services, intelligent interaction and other businesses in the medical process.

[0065] Corresponding to the medical record text translation method in the above embodiment, Figure 3 A structural block diagram of an artificial intelligence-based medical record text translation device provided in Embodiment 2 of the present invention is given. For ease of explanation, only the parts related to the embodiment of the present invention are shown.

[0066] See also Figure 3 , the medical record text translation device comprises: The keyword splicing module 31 is used to obtain the target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extract a number of keywords of the associated medical record text for any associated medical record text, and splice the keywords into a keyword sequence corresponding to the associated medical record text according to the order of the keywords in the associated medical record text, where M is a positive integer; The text concatenation module 32 is used to traverse all the associated medical record texts, obtain the keyword sequence corresponding to each associated medical record text, obtain M keyword sequences, and concatenate the target medical record text, the M associated medical record texts and the M keyword sequences to obtain a concatenated medical record text; The feature extraction module 33 is used to extract features from the concatenated medical record text, obtain text features of the concatenated medical record text, judge the translation direction of the text features, and determine the target translation direction of the target medical record text; The text translation module 34 is used to obtain the preset vocabulary and preset linear transformation matrix corresponding to the target translation direction, decode the text features, obtain the translation features of the target medical record text, perform linear transformation on the translation features according to the preset linear transformation matrix, and obtain the target medical record translation of the target medical record text. All texts in the target medical record translation are included in the preset vocabulary.

[0067] Optionally, the keyword includes a predicate, a predicate issuer and a predicate recipient, and the keyword concatenation module 31 includes: The keyword splicing submodule is used to extract all predicates, predicate issuers and predicate recipients in the associated medical record text, and splice all predicates, predicate issuers and predicate recipients into a keyword sequence of the associated medical record text according to the order of all predicates, predicate issuers and predicate recipients in the associated medical record text.

[0068] Optionally, the associated medical record text includes the original medical record text and the medical record translation corresponding to the original medical record text, and the above-mentioned keyword splicing submodule includes: A first keyword concatenation unit is used to extract all first predicates, first predicate issuers and first predicate recipients in the original medical record text, and concatenate all first predicates, first predicate issuers and first predicate recipients into a first keyword sequence of the original medical record text according to the order of all first predicates, first predicate issuers and first predicate recipients in the original medical record text; A second keyword concatenation unit is used to extract all second predicates, second predicate issuers and second predicate recipients in the medical record translation, and concatenate all second predicates, second predicate issuers and second predicate recipients into a second keyword sequence of the medical record translation according to the order of all second predicates, second predicate issuers and second predicate recipients in the medical record translation; The third keyword concatenation unit is used to determine the first keyword sequence and the second keyword sequence as a keyword sequence associated with the medical record text.

[0069] Optionally, the text splicing module 32 includes: A first text concatenation submodule is used to obtain a preset start delimiter and a sequence end character, and for any associated medical record text and a keyword sequence corresponding to the associated medical record text, concatenate the start delimiter, the keyword sequence and the associated medical record text according to a first preset order to obtain a first concatenated medical record text corresponding to the associated medical record text; The first traversal submodule is used to traverse the M associated medical record texts and the M keyword sequences corresponding to the M associated medical record texts to obtain M first concatenated medical record texts; A second text concatenation submodule is used to concatenate the start delimiter, the target medical record text and the sequence end symbol according to a second preset order to obtain a corresponding second concatenated medical record text; The third text splicing submodule is used to splice the M first spliced ​​medical record texts and the second spliced ​​medical record texts according to a third preset order to obtain a spliced ​​medical record text.

[0070] Optionally, the first preset order includes a fourth preset order and a fifth preset order, the keyword sequence includes a first keyword sequence and a second keyword sequence, and the first text concatenation submodule includes: A first text concatenation unit, configured to concatenate the start delimiter, the first keyword sequence, and the associated medical record text according to a fourth preset order to obtain a corresponding first concatenated medical record subtext; A second text concatenation unit is used to concatenate the start delimiter, the second keyword sequence and the medical record translation according to a fifth preset order to obtain a corresponding second concatenated medical record subtext; The third text splicing unit is used to splice the first spliced ​​medical record sub-text and the second spliced ​​medical record sub-text to obtain a first spliced ​​medical record text corresponding to the associated medical record text.

[0071] Optionally, the feature extraction module 33 includes: The probability value determination submodule is used to use the trained multi-layer perceptron model to determine the translation direction of the text features and obtain the probability value corresponding to the text features; The target translation direction determination submodule is used to obtain a preset mapping table between the translation direction and the probability value, and determine the target translation direction of the target medical record text according to the probability value and the mapping table.

[0072] Optionally, the target medical record text includes N text characters, where N is a positive integer, and the text translation module 34 includes: A first feature decoding submodule is used to decode the text feature and the preset feature using a trained decoder to obtain a first translation subfeature corresponding to the first text character in the target medical record text, where the preset feature is a zero feature sequence; The second feature decoding submodule is used to decode the text feature and the first n-1 translation sub-features using the trained decoder to obtain the nth translation sub-feature corresponding to the nth text character in the target medical record text, where n=2, 3, ..., N; The translation feature determination submodule is used to obtain the translation features of the target medical record text according to N translation sub-features corresponding to N text characters in the target medical record text.

[0073] It should be noted that the information interaction, execution process and other contents between the above modules are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0074] Figure 4 This is a schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention. Figure 4 As shown, the computer device of this embodiment includes: at least one processor ( Figure 4 Only one is shown in the figure), a memory, and a computer program stored in the memory and executable on at least one processor, wherein when the processor executes the computer program, the steps in any of the above-mentioned medical record text translation method embodiments are implemented.

[0075] The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that Figure 4 These are merely examples of computer devices and do not constitute limitations on the computer devices. The computer devices may include more or fewer components than those shown in the figure, or a combination of certain components, or different components. For example, they may also include a network interface, a display screen, and an input device.

[0076] The processor may be a CPU, or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0077] The memory includes a readable storage medium, an internal memory, etc., wherein the internal memory may be the memory of a computer device, and the internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. The readable storage medium may be a hard disk of a computer device, and in other embodiments, it may also be an external storage device of the computer device, for example, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device. Further, the memory may also include both an internal storage unit of the computer device and an external storage device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of a computer program, etc. The memory may also be used to temporarily store data that has been output or is to be output.

[0078] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned device can refer to the corresponding process in the above-mentioned method embodiment, which will not be repeated here. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned method embodiment can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device capable of carrying computer program code, recording medium, computer memory, read-only memory (ROM), random access memory, electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0079] The present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by a computer program product. When the computer program product runs on a computer device, the computer device can implement the steps in the above-mentioned method embodiment when executing.

[0080] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0081] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0082] In the embodiments provided by the present invention, it should be understood that the disclosed devices / computer equipment and methods can be implemented in other ways. For example, the device / computer equipment embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0083] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0084] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A medical record text translation method based on artificial intelligence, characterized in that: The medical record text translation method comprises: Obtain a target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extract a number of keywords of any associated medical record text, and splice the keywords into a keyword sequence corresponding to the associated medical record text according to the order of the keywords in the associated medical record text, where M is a positive integer; Traversing all the associated medical record texts, respectively obtaining the keyword sequence corresponding to each associated medical record text, obtaining M keyword sequences, and concatenating the target medical record text, the M associated medical record texts, and the M keyword sequences to obtain a concatenated medical record text; Extracting features of the concatenated medical record text to obtain text features of the concatenated medical record text, determining translation directions of the text features, and determining target translation directions of the target medical record text; Obtain a preset vocabulary and a preset linear transformation matrix corresponding to the target translation direction, decode the text features to obtain the translation features of the target medical record text, perform linear transformation on the translation features according to the preset linear transformation matrix, and obtain the target medical record translation of the target medical record text, wherein all texts in the target medical record translation are included in the preset vocabulary.

2. The medical record text translation method according to claim 1, characterized in that: The keywords include predicate, predicate sender and predicate receiver; The step of extracting a plurality of keywords from the associated medical record text and splicing the plurality of keywords into a keyword sequence corresponding to the associated medical record text according to the order of the plurality of keywords in the associated medical record text comprises: Extract all predicates, predicate issuers and predicate recipients in the associated medical record text, and according to the order of all predicates, predicate issuers and predicate recipients in the associated medical record text, splice all predicates, predicate issuers and predicate recipients into a keyword sequence of the associated medical record text.

3. The medical record text translation method according to claim 2, characterized in that: The associated medical record text includes the original medical record text and the medical record translation corresponding to the original medical record text; The extracting of all predicates, predicate issuers and predicate recipients in the associated medical record text, and splicing all predicates, predicate issuers and predicate recipients into a keyword sequence of the associated medical record text according to the order of all predicates, predicate issuers and predicate recipients in the associated medical record text includes: Extracting all first predicates, first predicate issuers, and first predicate recipients in the original medical record text, and splicing all first predicates, first predicate issuers, and first predicate recipients into a first keyword sequence of the original medical record text according to the order of all first predicates, first predicate issuers, and first predicate recipients in the original medical record text; Extracting all the second predicates, second predicate issuers and second predicate recipients in the medical record translation, and splicing all the second predicates, second predicate issuers and second predicate recipients into a second keyword sequence of the medical record translation according to the order of all the second predicates, second predicate issuers and second predicate recipients in the medical record translation; The first keyword sequence and the second keyword sequence are determined as the keyword sequences of the associated medical record text.

4. The medical record text translation method according to claim 3, characterized in that: The step of splicing the target medical record text, the M associated medical record texts and the M keyword sequences to obtain a spliced ​​medical record text includes: Obtaining a preset start delimiter and a sequence end character, and for any associated medical record text and a keyword sequence corresponding to the associated medical record text, concatenating the start delimiter, the keyword sequence and the associated medical record text in a first preset order to obtain a first concatenated medical record text corresponding to the associated medical record text; Traverse the M associated medical record texts and the M keyword sequences corresponding to the M associated medical record texts to obtain M first concatenated medical record texts; splicing the start delimiter, the target medical record text and the sequence end symbol according to a second preset order to obtain a corresponding second spliced ​​medical record text; The M first spliced ​​medical record texts and the second spliced ​​medical record text are spliced ​​in a third preset order to obtain a spliced ​​medical record text.

5. The medical record text translation method according to claim 4, characterized in that: The first preset sequence includes a fourth preset sequence and a fifth preset sequence, and the keyword sequence includes the first keyword sequence and the second keyword sequence; The step of splicing the start delimiter, the keyword sequence and the original medical record text in a first preset order to obtain a first spliced ​​medical record text corresponding to the associated medical record text includes: splicing the start delimiter, the first keyword sequence and the associated medical record text according to a fourth preset order to obtain a corresponding first spliced ​​medical record subtext; splicing the start delimiter, the second keyword sequence and the medical record translation according to a fifth preset order to obtain a corresponding second spliced ​​medical record subtext; The first spliced ​​medical record subtext and the second spliced ​​medical record subtext are spliced ​​to obtain a first spliced ​​medical record text corresponding to the associated medical record text.

6. The medical record text translation method according to claim 1, characterized in that: The determining of the translation direction of the text feature to determine the target translation direction of the target medical record text includes: Using the trained multi-layer perceptron model to determine the translation direction of the text feature, and obtaining a probability value corresponding to the text feature; A preset mapping table between translation directions and probability values ​​is obtained, and a target translation direction of the target medical record text is determined according to the probability values ​​and the mapping table.

7. The medical record text translation method according to claim 1, characterized in that: The target medical record text includes N text characters, where N is a positive integer; The decoding of the text features to obtain the translation features of the target medical record text includes: Decoding the text feature and the preset feature using a trained decoder to obtain a first translation sub-feature corresponding to the first text character in the target medical record text, wherein the preset feature is a zero feature sequence; Decoding the text feature and the first n-1 translation sub-features using the trained decoder to obtain the nth translation sub-feature corresponding to the nth text character in the target medical record text, where n=2, 3, ..., N; The translation features of the target medical record text are obtained according to the N translation sub-features corresponding to the N text characters in the target medical record text.

8. A medical record text translation device based on artificial intelligence, characterized in that: The medical record text translation device comprises: A keyword splicing module is used to obtain a target medical record text to be translated and M associated medical record texts corresponding to the target medical record text, extract a number of keywords of any associated medical record text, and splice the keywords into a keyword sequence corresponding to the associated medical record text according to the order of the keywords in the associated medical record text, where M is a positive integer; A text splicing module is used to traverse all the associated medical record texts, obtain the keyword sequence corresponding to each associated medical record text, obtain M keyword sequences, and splice the target medical record text, the M associated medical record texts and the M keyword sequences to obtain a spliced ​​medical record text; A feature extraction module is used to extract features from the spliced ​​medical record text to obtain text features of the spliced ​​medical record text, determine the translation direction of the text features, and determine the target translation direction of the target medical record text; A text translation module is used to obtain a preset vocabulary and a preset linear transformation matrix corresponding to the target translation direction, decode the text features to obtain the translation features of the target medical record text, perform linear transformation on the translation features according to the preset linear transformation matrix, and obtain the target medical record translation of the target medical record text, wherein all texts in the target medical record translation are included in the preset vocabulary.

9. A computer device, characterized in that: The computer device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the medical record text translation method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the medical record text translation method according to any one of claims 1 to 7 is implemented.

Images