CN113033155A - Automatic coding method for medical concepts by combining sequence generation and hierarchical word lists - Google Patents

Abstract

The invention discloses and provides a medical concept automatic coding method and device combining sequence generation and hierarchical vocabulary. The medical concept coding events in clinical medical texts are transformed into sequence generation problems, and the concept of hierarchical vocabulary is introduced to enhance the According to the relationship between medical terms, the standard medical terms corresponding to the clinical medical text are accurately determined and automatically encoded in the process of sequence generation according to the hierarchical vocabulary. It solves the problems of high cost, limited efficiency and low accuracy by using manual coding method in the prior art to manually map medical concepts in clinical medical texts to standard medical term codes.

Description

An automatic coding method for medical concepts combining sequence generation and hierarchical vocabulary

technical field

The invention relates to the field of medical concept coding, in particular to a medical concept automatic coding method combining sequence generation and hierarchical vocabulary.

Background technique

The automatic coding of medical concepts is an important research direction in the field of medical information processing. In the medical information system, the same standard medical term may have many different expressions of medical concepts. The inconsistency and inaccuracy of such expressions seriously hinder the integration, sharing and utilization of medical big data. , teaching and research have brought a lot of inconvenience. Medical coding is a number and letter labeling system that provides a unique and uniform coded representation for each diagnosis, symptom, or combination of symptoms. At present, medical institutions need to manually map medical concepts in clinical medical texts to standard medical term codes by means of manual coding, and manual coding requires a large number of professionals with medical knowledge to operate, which is costly, limited in efficiency and inaccurate.

Therefore, the existing technology still needs to be improved and developed.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to provide a method and device for automatic coding of medical concepts combining sequence generation and hierarchical vocabulary, aiming at solving the problem of using manual coding in the prior art. Manual mapping of medical concepts in clinical medical texts to standard medical term codes is not only costly, limited in efficiency, but also inaccurate.

The technical scheme adopted by the present invention to solve the problem is as follows:

In a first aspect, an embodiment of the present invention provides an automatic coding method for medical concepts combining sequence generation and hierarchical vocabulary, wherein the method includes:

Obtain clinical medical text, input the clinical medical text into a preset encoder, and obtain initial vector data of the clinical medical text;

Obtain pre-built hierarchical vocabulary data, input the hierarchical vocabulary data into a preset learning algorithm, and obtain standard medical term vector data of the hierarchical vocabulary;

Input the initial vector data of the clinical medical text and the standard medical term vector data that have been generated into the preset decoder, sequentially generate encoded data corresponding to several standard medical terms, and form the encoded data according to the encoded data. The encoded data corresponding to the clinical medical text.

In one embodiment, the obtaining of the clinical medical text, inputting the clinical medical text into a preset encoder, and obtaining the initial vector data of the clinical medical text includes:

Inputting the clinical medical text into the word embedding layer, and mapping the clinical medical text through the word embedding layer to obtain mapping data;

The mapping data is input to an encoder, and initial vector data encoded and generated by the encoder based on the mapping data is obtained.

In one embodiment, the obtaining pre-built hierarchical vocabulary data, inputting the hierarchical dictionary data into a preset learning algorithm, and obtaining the standard medical term vector data of the hierarchical vocabulary includes:

Obtain the coding information of the standard medical term data in the term dictionary data, and divide the standard medical term data into a parent node and a child node according to the coding information;

Obtain the parent node, the child node, and the parent-child relationship information between the parent node and the child node, according to the parent node, the child node, and the parent node and the child node. The parent-child relationship information builds hierarchical vocabulary data;

Inputting the hierarchical vocabulary data into a preset learning algorithm to obtain vector data representing the parent node, the child node and the parent-child relationship information;

The vector data representing the parent node, the child node and the parent-child relationship information is used as the standard medical term vector data of the hierarchical vocabulary.

In one embodiment, the encoded information includes alphabetic segment information and numeric segment information.

In one embodiment, the acquiring coding information of the standard medical term data in the term dictionary data, and dividing the standard medical term data into parent nodes and child nodes according to the coding information includes:

Treat each standard medical term data as a node;

Using the same type of all letter segment information, and several nodes with the same number before the preset sequence bit of the digital segment information as the same type of node;

Among the nodes of the same type, the node with the shortest digital field information is used as the parent node, and the nodes other than the parent node are used as child nodes.

In one embodiment, the decoder includes a classifier, the classifier includes labels of a plurality of standard medical terms, the initial vector data of the clinical medical text and the generated standard The medical term vector data is input into the preset decoder, the encoded data corresponding to several standard medical terms are sequentially generated, and the standard medical term sequence data corresponding to the clinical medical text is formed according to the encoded data, including:

Obtain sequence data composed of all historical standard medical term vector data output by the decoder; the sequence data is the standard medical term vector data corresponding to the encoding output by the decoder before the current time step;

Determine the encoded data output by the decoder at the current time step corresponding to the clinical medical text based on the initial vector data and the sequence data by the classifier; repeat this process until there is no encoded data available until generated;

Standard medical term sequence data corresponding to the clinical medical text is formed according to the encoded data.

In an embodiment, the classifier includes a probability function, and the classifier determines, based on the initial vector data and the sequence data, that at the current time step corresponding to the clinical medical text, the the encoded data output by the decoder;

Fusing the initial vector data with the vector data corresponding to the sequence data to obtain fusion vector data;

Inputting the fusion vector data into the probability function, and obtaining the probability values of several possible encoded data generated by the probability function based on the fusion vector data;

Sort the probability values according to the numerical value, and use the encoded data with the largest probability value as the encoded data output by the decoder at the current moment.

In a second aspect, an embodiment of the present invention further provides an apparatus for automatic coding of medical concepts combining sequence generation and hierarchical vocabulary, wherein the apparatus includes:

an acquisition module, configured to acquire clinical medical text, input the clinical medical text into a preset encoder, and obtain initial vector data of the clinical medical text;

A learning module for acquiring pre-built hierarchical vocabulary data, inputting the hierarchical vocabulary data into a preset learning algorithm, and obtaining standard medical term vector data of the hierarchical vocabulary;

The encoding module is used to input the initial vector data of the clinical medical text and the standard medical term vector data that have been generated into the preset decoder, and sequentially generate encoded data corresponding to several standard medical terms, and according to the The encoded data forms standard medical term sequence data corresponding to the clinical medical text.

Beneficial effects of the present invention: In the embodiment of the present invention, the medical concept coding events in clinical medical texts are transformed into sequence generation problems, and the concept of hierarchical vocabulary is introduced to enhance the relationship between medical terms. During the sequence generation process, the standard medical terms corresponding to the clinical medical text are accurately determined and automatically encoded. The method solves the problems of high cost, limited efficiency and low accuracy of manually mapping medical concepts in clinical medical texts to standard medical term codes by using manual coding methods in the prior art.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments described in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort.

FIG. 1 is a schematic flowchart of a method for automatic coding of medical concepts that combines sequence generation and hierarchical vocabulary according to an embodiment of the present invention.

FIG. 2 is a schematic diagram of a framework of a Seq2Seq model provided by an embodiment of the present invention.

FIG. 3 is a module connection diagram of an apparatus for automatic encoding of medical concepts that combines sequence generation and hierarchical vocabulary according to an embodiment of the present invention.

FIG. 4 is a principle block diagram of a terminal provided by an embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer and clearer, the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

It should be noted that if there are directional indications (such as up, down, left, right, front, back, etc.) involved in the embodiments of the present invention, the directional indications are only used to explain a certain posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication also changes accordingly.

With the rapid development and popularization of information technologies such as the Internet, big data, cloud computing and artificial intelligence, the production and life of human beings have been affected unprecedentedly. In recent years, information technology has been gradually applied to all aspects of social life. In all walks of life, information technology has changed the way humans manage, analyze and use data, and many fields such as economics and culture are now inseparable from the assistance of information technology. Among these application fields of information technology, medical treatment is one of the important fields with unlimited potential. In the medical field, a large amount of information processing is involved. This medical information has the following characteristics:

1) The amount of data is large and the growth rate is fast;

2) High demand for sharing.

The biggest advantage of information technology lies in its efficiency of data processing. Therefore, information technology is widely used in the medical field, and thus the computer application direction of medical information processing has emerged. Medical information processing refers to the organic combination of computer-related technologies and the needs of the medical and health industry to meet the needs of medical institutions and related departments for the collection, sorting, storage and analysis of medical and health information, improve the efficiency of the health industry, and meet customer functional needs. . Medical information processing technology not only improves the efficiency of medical information processing, but also improves the accuracy of medical information processing, bringing the development of medical information to a new level. For a long time, how to use medical information processing technology to effectively improve medical level and promote medical development has been one of the hot issues that scholars in related fields have been studying.

The automatic coding of medical concepts is an important research direction in the field of medical information processing. Clinical medical texts usually refer to the text data formed by medical staff in medical activities to describe the clinical manifestations of patients, which may contain several medical-related concepts. In a medical information system, the same standard medical term may have many different representations of medical concepts. First, because medical staff may have different recording styles, sometimes in pursuit of work efficiency, the medical texts they record may contain many synonyms, acronyms, loanwords or colloquial expressions. Therefore, in clinical medical texts, it is obvious that one term corresponds to multiple expressions. For example, in Chinese clinical medical texts, the term "congenital scoliosis" can be expressed as "congenital scoliosis" or "congenital scoliosis deformity"; in English clinical medical texts, "heart attack" ", "MI" and "myocardial infarction" can all mean "myocardial infarction". Second, in some cases, medical concepts related to multiple diagnoses or symptoms are closely related and easily confused. The same medical concept in clinical medical texts may correspond to different standard medical terms due to different contexts, such as In Chinese clinical medical texts, the diagnosis-related medical concept "nasopharyngeal fistula" may correspond to the standard medical term "nasosinus fistula" or the standard medical term "pharyngeal fistula" according to contextual information. The inconsistency and inaccuracy of such expressions have seriously hindered the integration, sharing and utilization of medical big data, and brought a lot of inconvenience to clinical, teaching and scientific research in the medical field. Medical coding is a number and letter labeling system that provides a unique coded representation for each diagnosis, symptom, or combination of symptoms. Therefore, it is particularly urgent to standardize medical concepts in clinical medical texts into codes corresponding to standard medical terms in the medical coding system in accordance with a unified approach to promoting medical informatization. At present, some medical institutions use manual coding to manually map medical concepts in clinical medical texts to standard medical term codes. In this process, coders need to look up medical concepts or other related information in clinical medical texts, and then manually assign appropriate standard medical term codes to these medical concepts according to the coding guidelines. As medical institutions generate massive amounts of text information every day, manual coding requires a large number of professionals with medical knowledge to operate, which is costly, limited in efficiency and inaccurate.

Aiming at the above-mentioned defects of the prior art, the present invention provides a medical concept automatic coding method combining sequence generation and hierarchical vocabulary, by transforming medical concept coding events in clinical medical texts into sequence generation problems, and introducing hierarchical vocabulary To enhance the relationship between medical terms, the standard medical terms corresponding to the clinical medical text are accurately determined and automatically encoded in the process of sequence generation according to the hierarchical vocabulary. The method solves the problems of high cost, limited efficiency and low accuracy of manually mapping medical concepts in clinical medical texts to standard medical term codes by using manual coding methods in the prior art.

As shown in FIG. 1 , the present embodiment provides a medical concept automatic coding method combining sequence generation and hierarchical vocabulary, and the method includes the following steps:

Step S100: Obtain clinical medical text, input the clinical medical text into a preset encoder, and obtain initial vector data of the clinical medical text.

In order to realize automatic encoding of medical concepts in clinical medical texts, in this embodiment, the encoded clinical medical texts need to be obtained first. Since the main goal of this embodiment is to automatically map medical concepts to encoded data corresponding to standard medical terms using computer-related technologies, it is necessary to perform certain processing on the clinical medical text so that the computer can perform calculations based on the processed clinical medical text. Specifically, as shown in FIG. 2 , in this embodiment, the Seq2Seq model is mainly used as the main generation framework, and the so-called Seq2Seq model refers to the model used when the output length is uncertain. For example, when the task of translating a sentence of Chinese into English appears in the task of machine translation, the length of the translated English may be shorter than that of Chinese, or it may be longer than that of Chinese, so the length of the output may be uncertain. , and the Seq2Seq model is just right for this situation. In this embodiment, the Seq2Seq model includes an encoder and a decoder, wherein the encoder can use a bidirectional LSTM, and the bidirectional LSTM is responsible for converting the input text data into data in the form of a vector, and this vector can be regarded as the semantic vector of the input text. Specifically, after obtaining the clinical medical text, input the clinical medical text into the encoder in the preset Seq2Seq model, and after the encoder obtains the clinical medical text, it will input the clinical medical text through learning input. Encoding into a vector form, and then obtaining the initial vector data of the clinical medical text.

In an implementation manner, the step S100 specifically includes the following steps:

Step S110, input the clinical medical text into a word embedding layer, and obtain mapping data after mapping the clinical medical text through the word embedding layer;

Step S120: Input the mapping data into an encoder, and obtain initial vector data encoded by the encoder to generate the clinical medical text based on the mapping data.

中的单词通过词嵌入层映射为向量表示的

，然后对

使用双向LSTM进行编码，得到隐藏层表示：In order to obtain the initial vector data, in this embodiment, the clinical medical text needs to be input into the word embedding layer first. After the word embedding layer obtains the clinical medical text, it maps the features of the words in the text to a lower dimension, and outputs the mapping data after the mapping is completed, so that the model has fewer parameters and faster training. Then, the mapping data is input into the encoder, and after the encoder obtains the mapping data, it encodes the mapping data and generates initial vector data. For example, in this embodiment, a bidirectional LSTM is used as the encoder. First, on the encoder side, the clinical medical text is first

The words in the word embedding layer are mapped to vector representation

, then right

Use a bidirectional LSTM for encoding to get the hidden layer representation:

Since the encoder used in this embodiment is a bidirectional LSTM, it is necessary to splicing the two hidden layer representations output by the bidirectional LSTM to obtain the final hidden layer representation:

即为所述临床医疗文本的初始向量数据。in,

That is, the initial vector data of the clinical medical text.

As shown in Figure 1, the method further includes the following steps:

Step S200: Obtain pre-built hierarchical vocabulary data, input the hierarchical vocabulary data into a preset learning algorithm, and obtain standard medical term vector data of the hierarchical vocabulary.

In order to realize the automatic coding process of medical concepts, this embodiment will need to acquire pre-built hierarchical vocabulary data, which actually refers to vocabulary data containing hierarchical relationships between various standard medical terms. In the embodiment, a deep learning model needs to be used, so it is also necessary to convert the standard medical terms contained in the hierarchical vocabulary into a vector format that can be executed by the deep learning model, so it is necessary to input the hierarchical vocabulary data into the preset learning. algorithm, and obtain the standard medical term vector data of the hierarchical vocabulary.

In an implementation manner, the step S200 specifically includes the following steps:

Step S210, obtaining the coding information of the standard medical term data in the term dictionary data, and dividing the standard medical term data into a parent node and a child node according to the coding information;

Step S220: Obtain the parent node, the child node, and the parent-child relationship information between the parent node and the child node, according to the parent node, the child node, and the parent node and the child node. The parent-child relationship information between them constructs hierarchical vocabulary data;

Step S230, inputting the hierarchical vocabulary data into a preset learning algorithm to obtain vector data representing the parent node, the child node and the parent-child relationship information;

Step S240: Use the vector data representing the parent node, the child node and the parent-child relationship information as the standard medical term vector data of the hierarchical vocabulary.

First, this embodiment will build a hierarchical vocabulary in advance. Specifically, in order to construct the hierarchical vocabulary, this embodiment needs to acquire the coding information of the standard medical term data in the term dictionary data, and divide the standard medical term data into parent nodes and child nodes according to the coding information to determine Inclusions between standard medical terms. In an implementation method, the encoding information may include letter segment information and number segment information. For example, the encoding corresponding to a standard medical term may be J001. Then, each standard medical term data is used as a node, and all the letter segment information is of the same type, and several nodes with the same number before the preset sequence bit of the digital segment information are used as the same type of node, and then in the same type of node. Among the class nodes, the node with the shortest digital field information is regarded as the parent node, and the nodes other than the parent node are regarded as child nodes.

In an implementation manner, this embodiment can design an algorithm to determine the parent node and the child node, and the structure composed of the parent node and the child node is a tree-shaped hierarchical structure. The specific algorithm as follows:

A. Define the data structure of the nodes in the tree, each tree node contains two parts: the encoded string and the list of child nodes, go to step B;

B. Initialize the root node of the tree, the corresponding code of the root node is an empty string, and the list of child nodes is empty, go to step C;

C. If the standard medical term dictionary is empty, the algorithm ends. Otherwise, take a (code, term) pair from the standard medical term dictionary and go to step D;

D. Set the current node as the root node, if the code of the current node is the prefix of the code (letter segment information) and the two are different, then set the jump out of the loop flag to false, go to step E, otherwise go to step H;

E. If the child node list of the current node is empty, go to step G. Otherwise, take a child node and go to step F;

F. If the code of the child node is the prefix of the code taken and the two are different, set the current node as the child node, set the jump-out loop flag to be true, and jump to step F, otherwise go to step E;

G. If the out of loop flag is true, go to step C. Otherwise, go to step D;

H. Initialize a new node, the corresponding code of the new node is the selected code, and the child node is empty. Add this new node to the list of child nodes of the current node and go to step C.

After determining the parent node and the child node, it is also necessary to obtain the parent node, the child node and the parent-child relationship information between the parent node and the child node, and according to the parent node, the child node and the Describe the parent-child relationship information to construct hierarchical vocabulary data, and then input the hierarchical vocabulary data into a preset learning algorithm, such as the TransE algorithm, so as to obtain the information representing the parent node, the child node and the parent-child relationship information. vector data. Finally, the vector data representing the parent node, the child node and the parent-child relationship information is used as the standard medical term vector data of the hierarchical vocabulary. In short, this embodiment hopes to represent standard medical terms by low-dimensional and dense vectors, and to determine the inclusion relationship between each standard medical term vector data, so that standard medical terms with similar semantics are distributed in approximate and adjacent ones. vector space. In addition, this embodiment determines the subordination or hierarchical relationship of each standard medical term by constructing a hierarchical vocabulary, thereby making the hierarchical relationship between the standard medical terms clearer and enabling better screening of clinical medical texts The standard medical terms corresponding to the concepts of traditional Chinese medicine, and finally achieve the purpose of automatic coding of medical concepts.

As shown in Figure 1, the method further includes the following steps:

Step S300, inputting the initial vector data of the clinical medical text and the generated standard medical term vector data into a preset decoder, sequentially generating encoded data corresponding to several standard medical terms, and according to the encoded data Standard medical term sequence data corresponding to the clinical medical text is formed.

The Seq2Seq model used in this embodiment further includes a decoder, and in an implementation manner, a unidirectional LSTM may be used as the decoder. Specifically, after obtaining the initial vector data of the clinical medical text and the vector data of standard medical terms, these two kinds of data need to be input into the decoder. After obtaining the two kinds of data, the decoder will sequentially generate multiple standard medical terms. The encoded data corresponding to the terms, according to the encoded data, the standard medical term sequence data corresponding to the clinical medical text can be formed. In short, when coding the medical concepts in the clinical medical text, this embodiment not only refers to the hierarchical relationship between standard medical terms, but also refers to the sequence generated by multiple standard medical terms, and finally determines the clinical medical text. The correct coding for the medical concepts in .

In an implementation manner, the decoder includes a classifier, and the classifier includes a plurality of labels of standard medical terms, and the step S300 specifically includes the following steps:

Step S310, obtaining sequence data composed of all historical standard medical term vector data output by the decoder; the sequence data is the standard medical term vector data corresponding to the encoding output by the decoder before the current time step;

Step S320: Determine the encoded data output by the decoder at the current time step corresponding to the clinical medical text based on the initial vector data and the sequence data by the classifier; repeat this process until there is no until the encoded data can be generated;

Step S330 , forming standard medical term sequence data corresponding to the clinical medical text according to the encoded data.

In order to realize automatic encoding of medical concepts in clinical medical texts, this embodiment first acquires sequence data composed of all historical standard medical term vector data output by the decoder. It can be understood that the sequence data is the current time The code corresponding to the standard medical term vector data output by the decoder before the step. Then, in order to determine the vector data of a standard medical term with the highest similarity with the clinical medical text at the current time step, in this embodiment, the vector data and the sequence data are also input into the classifier, and the The classification space can actually be seen as a set of labels, each label corresponding to a standard medical term. In this embodiment, it is hoped that the classifier can determine the encoded data output by the decoder at the current time step corresponding to the clinical medical text according to the initial vector data and the sequence data. Specifically, in this embodiment, a donkey chasing function is also set in the classifier. In order to determine the encoded data output by the decoder at the current time step, this embodiment needs to firstly compare the initial vector data with the sequence data corresponding to the sequence data. The vector data is fused to obtain fused vector data. Then, the fusion vector data is input into the probability function, and the probability values of several possible encoded data generated by the probability function based on the fusion vector data are obtained. It can be understood that the probability values are to a certain extent The correlation or degree of association between each spliced sequence data and the clinical medical text can be indicated, that is, an attention mechanism is introduced in this embodiment to determine the standard medical term corresponding to the clinical medical text.

解码器的隐藏层表示

计算如下，其中

为

时刻输出分布中概率最大的编码

对应的标准医学术语的向量表示，即用TransE算法学习得到的向量表示；For example, on the decoder side, the moment

The hidden layer representation of the decoder

Calculated as follows, where

for

The code with the highest probability in the output distribution at the moment

The vector representation of the corresponding standard medical terms, that is, the vector representation learned by the TransE algorithm;

与临床医疗文本之间的相关性系数，具体地，注意力机制计算过程如下，

表示

时刻利用注意力机制计算得到的向量：Among them, the attention mechanism needs to be used when calculating the probability, so as to obtain

The correlation coefficient with clinical medical text, specifically, the calculation process of the attention mechanism is as follows,

express

The vector calculated by the attention mechanism at all times:

表示

与临床医疗文本中第i个字之间的相关度，

表示

与临床医疗文本中第i个字之间的相关度系数，

、

、

、均为学习参数。in,

express

the correlation with the i -th word in the clinical medical text,

express

is the correlation coefficient with the i -th word in the clinical medical text,

,

,

, are learning parameters.

Then output the probability distribution through the classifier:

表示

时刻

与

经过非线性激活函数f（如tanh、ReLu等）进行融合后的向量，

表示t时刻解码器端输出的分布，

、

、

、均为学习参数。in,

express

time

and

The vector after fusion by nonlinear activation function f (such as tanh, ReLu, etc.),

represents the distribution of the output of the decoder at time t ,

,

,

, are learning parameters.

Then, the obtained probability values are sorted by numerical value, and the encoded data with the largest probability value is used as the encoded data output by the decoder at the current moment. This process is repeated until no encoded data can be generated. Finally, standard medical term sequence data corresponding to the clinical medical text is formed according to the encoded data, thereby realizing automatic encoding of the clinical medical text.

Generally speaking, although there are technologies for medical concept coding using machine learning methods in the prior art, most of them use a greedy search strategy to generate codes. The greedy search strategy converts the probability of The largest vector of standard medical terms is filtered out, so the search space is limited. However, this embodiment actually adopts beam search, that is, at each time step, the first few sequences with the highest probability are considered as candidate sequences, and the sequence with the highest probability is selected as the final target sequence among the candidate sequences. The relationship between sequences and clinical medical texts will be considered, and the first few sequences with the largest probability values are used as candidate sequences. Therefore, compared with the greedy search strategy, the search space of the beam search strategy adopted in this embodiment is larger. , the encoding is more accurate.

Based on the above embodiment, the present invention also provides a medical concept automatic coding device combining sequence generation and hierarchical vocabulary. As shown in FIG. 3 , the device includes:

An acquisition module 01, configured to acquire clinical medical text, input the clinical medical text into a preset encoder, and acquire initial vector data generated by the encoder based on the clinical medical text;

The learning module 02 is used to obtain pre-built hierarchical vocabulary data, input the hierarchical vocabulary data into a preset learning algorithm, and obtain the standard medical term vector data of the hierarchical vocabulary;

The encoding module 03 is used to input the initial vector data of the clinical medical text and the standard medical term vector data that have been generated into the preset decoder, and sequentially generate encoded data corresponding to several standard medical terms, and according to the The encoded data forms standard medical term sequence data corresponding to the clinical medical text.

Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which may be as shown in FIG. 4 . The terminal includes a processor, a memory, a network interface, and a display screen connected through a system bus. Among them, the processor of the terminal is used to provide computing and control capabilities. The memory of the terminal includes a non-volatile storage medium and an internal memory. The nonvolatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The network interface of the terminal is used to communicate with external terminals through a network connection. The computer program, when executed by a processor, implements a method for automatic coding of medical concepts that combines sequence generation and hierarchical vocabulary. The display screen of the terminal may be a liquid crystal display screen or an electronic ink display screen.

Those skilled in the art can understand that the principle block diagram shown in FIG. 4 is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal to which the solution of the present invention is applied. The specific terminal may include There are more or fewer components than shown in the figures, or some components are combined, or have a different arrangement of components.

In one implementation, one or more programs are stored in a memory of the terminal and are configured to be executed by one or more processors, including for combining sequence generation and hierarchy. Instructions for automatic coding methods for medical concepts of the vocabulary.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage In the medium, when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other medium used in the various embodiments provided by the present invention may include non-volatile and/or volatile memory. Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM) and so on.

To sum up, the present invention discloses a medical concept automatic coding method and device combining sequence generation and hierarchical vocabulary, by transforming medical concept coding events in clinical medical texts into sequence generation problems, and introducing hierarchical vocabulary Concepts are used to enhance the relationship between medical terms, and standard medical terms corresponding to the clinical medical text are accurately determined and automatically encoded in the process of sequence generation according to the hierarchical vocabulary. The method solves the problems of high cost, limited efficiency and low accuracy of manually mapping medical concepts in clinical medical texts to standard medical term codes by using manual coding methods in the prior art.

It should be understood that the application of the present invention is not limited to the above examples. For those of ordinary skill in the art, improvements or transformations can be made according to the above descriptions, and all these improvements and transformations should belong to the protection scope of the appended claims of the present invention.

Claims (9)

1. A method for automatic coding of medical concepts in conjunction with sequence generation and hierarchical vocabularies, the method comprising:

acquiring a clinical medical text, and inputting the clinical medical text into a preset encoder to obtain initial vector data of the clinical medical text;

acquiring pre-constructed hierarchical word list data, inputting the hierarchical word list data into a preset learning algorithm, and acquiring standard medical term vector data of the hierarchical word list;

inputting the initial vector data of the clinical medical text and the generated standard medical term vector data into a preset decoder, sequentially generating encoded data corresponding to a plurality of standard medical terms, and forming standard medical term sequence data corresponding to the clinical medical text according to the encoded data.

2. The method of claim 1, wherein the obtaining clinical medical texts, inputting the clinical medical texts into a preset encoder, and obtaining initial vector data of the clinical medical texts comprises:

inputting a clinical medical text into a word embedding layer, and mapping the clinical medical text through the word embedding layer to obtain mapping data;

inputting the mapping data into an encoder, and acquiring initial vector data of the clinical medical text generated by the encoder based on the mapping data.

3. The method as claimed in claim 1, wherein the step of obtaining pre-constructed hierarchical vocabulary data, inputting the hierarchical vocabulary data into a predetermined learning algorithm, and obtaining standard medical term vector data of the hierarchical vocabulary comprises:

acquiring coding information of standard medical term data in term dictionary data, and dividing the standard medical term data into a parent node and a child node according to the coding information;

acquiring the father node, the child node and the father-child relationship information between the father node and the child node, and constructing hierarchical vocabulary data according to the father node, the child node and the father-child relationship information between the father node and the child node;

inputting the hierarchical vocabulary data into a preset learning algorithm to obtain vector data representing the father node, the child nodes and the father-child relationship information;

and taking vector data representing the parent node, the child nodes and the parent-child relationship information as standard medical term vector data of the hierarchical vocabulary.

4. The method as claimed in claim 3, wherein the coding information comprises letter field information and number field information.

5. The method as claimed in claim 4, wherein the obtaining of encoding information of standard medical term data in term dictionary data, and the dividing of the standard medical term data into parent nodes and child nodes according to the encoding information comprises:

taking each standard medical term data as a node;

taking a plurality of nodes with the same number before the preset sequence bit of the digital field information as the same type of nodes, wherein the types of all the letter field information are the same;

and in the same type of nodes, taking the node with the shortest digital field information as a father node and taking the nodes except the father node as child nodes.

6. The method as claimed in claim 1, wherein the decoder comprises a classifier containing labels of a plurality of standard medical terms, the inputting the initial vector data of the clinical medical text and the generated vector data of the standard medical terms into a preset decoder sequentially generates encoded data corresponding to a plurality of standard medical terms, and the forming of the sequence data of the standard medical terms corresponding to the clinical medical text from the encoded data comprises:

acquiring sequence data consisting of all historical standard medical term vector data output by the decoder; the sequence data is standard medical term vector data corresponding to codes output by the decoder before the current time step;

determining, by the classifier, encoded data output by the decoder at a current time step corresponding to the clinical medical text based on the initial vector data and the sequence data; this process is repeated until no encoded data can be generated;

and forming standard medical term sequence data corresponding to the clinical medical text according to the coded data.

7. The method of claim 6, wherein the classifier comprises a probability function, and the classifier determines the encoded data output by the decoder at the current time step corresponding to the clinical medical text based on the initial vector data and the sequence data;

fusing the initial vector data with vector data corresponding to the sequence data to obtain fused vector data;

inputting the fusion vector data into the probability function, and acquiring probability values of a plurality of possible coded data generated by the probability function based on the fusion vector data;

and sequencing the probability values according to the numerical values, and taking the coded data with the maximum probability value as the coded data output by the decoder at the current moment.

8. An apparatus for automatic coding of medical concepts in conjunction with sequence generation and hierarchical vocabularies, the apparatus comprising:

the acquisition module is used for acquiring a clinical medical text, inputting the clinical medical text into a preset encoder and acquiring initial vector data generated by the encoder based on the clinical medical text;

the learning module is used for acquiring pre-constructed hierarchical word list data, inputting the hierarchical word list data into a preset learning algorithm and acquiring standard medical term vector data of the hierarchical word list;

and the encoding module is used for inputting the initial vector data of the clinical medical text and the generated standard medical term vector data into a preset decoder, sequentially generating encoded data corresponding to a plurality of standard medical terms, and forming standard medical term sequence data corresponding to the clinical medical text according to the encoded data.

9. A computer readable storage medium having stored thereon a plurality of instructions adapted to be loaded and executed by a processor to perform the steps of a method for automatic encoding of medical concepts in conjunction with sequence generation and hierarchical vocabularies of any of the preceding claims 1-7.

Images