CN114464283A - Manual labeling processing method, device, processor and storage medium based on ICD-10 depression diagnosis and treatment standard interview text - Google Patents

Abstract

The invention relates to a method for implementing manual labeling processing based on interview text information of ICD-10 standard of diagnosis and treatment of depression, comprising the following steps: constructing a domain dictionary by using a statistical method, labeling a Chinese electronic medical record named entity identification and labeling corpus according to a defined labeling specification; Identify Chinese electronic medical record named entities. The invention also relates to a device, a processor and a computer-readable storage medium for realizing manual annotation processing of interview text information based on the ICD-10 standard of diagnosis and treatment of depression. The method, device, processor and computer-readable storage medium for realizing manual labeling processing based on ICD-10 depression diagnosis and treatment standard interview text information of the present invention are adopted to optimize user labeling experience, improve labeling efficiency, and improve the accuracy of labeling results. rate, adding review and editing functions, allowing deployment and operation on a variety of system environments.

Description

Manual annotation processing method, device, processor and storage medium for interview text based on ICD-10 standard for diagnosis and treatment of depression

technical field

The invention relates to the field of artificial intelligence, in particular to the field of natural language processing, and in particular to a method, a device, a processor and a computer-readable storage medium for realizing manual annotation processing based on ICD-10 standard of depression diagnosis and treatment interview text information.

Background technique

The current general scene text annotation tool does not support annotation functions in the field of mental health:

For example, interview medical records can include texts, doctor-patient dialogue scripts and other digital information. However, there is still a large amount of semi-structured or unstructured free text data in the interview records, and the useful information distributed in the free and messy text cannot be used quickly and effectively by the computer. In order to transform unstructured interview medical record data into a structured form that can be recognized by computer, it is essential to use natural language processing technology for text mining. Although the text information described in natural language in the interview medical records contains a wealth of depression symptoms and related medical knowledge, due to the professionalism and complexity of the field, there are often different language description methods from the general text, and the general field cannot be directly used. method to solve.

Therefore, named entity recognition (manual annotation), as the key technology of text data mining, has always been the research basis and hotspot of natural language processing in the field of mental health. The named entity in the general domain refers to the named entity composed of three major categories: entity class, time class and number class, as well as the named entity in a specific field (such as depression diagnosis). Defined under the entity type. Since this research is a brand new attempt, there is no historical experience to refer to, so the problem of manual annotation needs to be solved first.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art, and provide a method for realizing manual annotation processing based on the interview text information of the ICD-10 depression diagnosis and treatment standard, which meets the requirements of high accuracy, low cost consumption and relatively wide application range, Apparatus, processor, and computer-readable storage medium thereof.

In order to achieve the above object, the method, device, processor and computer-readable storage medium thereof for realizing manual labeling and processing based on ICD-10 depression diagnosis and treatment standard interview text information of the present invention are as follows:

The main feature of the method for implementing manual labeling and processing based on the interview text information of the ICD-10 standard for diagnosis and treatment of depression is that the method comprises the following steps:

(1) Construct a domain dictionary by accurately matching manual labels, and label the Chinese electronic medical record named entity recognition and labeling corpus according to the defined labeling specifications;

(2) Simultaneously perform manual labeling and automatic labeling to identify Chinese electronic medical record named entities.

Preferably, the step of constructing a domain dictionary in the step (1) is specifically:

Obtain keywords from Chinese electronic medical records or use external professional resources to obtain dictionary keywords to construct domain dictionaries.

Preferably, in the described step (2), carry out manual labeling, and specifically includes the following steps:

(2.1) Log in to the depression intelligent diagnosis and case generation system, and fill in the patient consultation information;

(2.2) Upload recording files;

(2.3) Perform recording recognition and convert the recording file into text content;

(2.4) Mark the text content;

(2.5) Generate a diagnostic report.

Preferably, the step of deep learning in the described step (2) is specifically:

Based on deep learning, domain corpus pre-training word embedding and domain preprocessing are used for related entity recognition models.

Preferably, the method further comprises the following steps:

(3) Build a manual labeling model, identify and interview medical entities of electronic medical records through the manual labeling model, and display and analyze the results of medical entity recognition in psychiatric medical records.

The device for realizing manual annotation processing of interview text information based on the ICD-10 standard for diagnosis and treatment of depression is mainly characterized in that the device includes:

a processor configured to execute computer-executable instructions;

The memory stores one or more computer-executable instructions, and when the computer-executable instructions are executed by the processor, each of the above-mentioned methods for implementing manual annotation processing based on the ICD-10 depression diagnosis and treatment standard interview text information is realized. step.

The main feature of the processor for implementing manual annotation processing of interview text information based on the ICD-10 depression diagnosis and treatment standard is that the processor is configured to execute computer-executable instructions, and the computer-executable instructions are When the processor is executed, each step of the above-mentioned method for manual annotation processing based on the interview text information of the ICD-10 standard for diagnosis and treatment of depression is realized.

The computer-readable storage medium is mainly characterized in that a computer program is stored thereon, and the computer program can be executed by a processor to realize the above-mentioned method for manual annotation processing based on the interview text information of the ICD-10 standard for diagnosis and treatment of depression of the various steps.

Using the method, device, processor and computer-readable storage medium for manual annotation processing based on ICD-10 depression diagnosis and treatment standard interview text information of the present invention, each annotation result needs to be visually displayed in real time, and each annotation needs to be Supports selection of a large number of entity types. The implementation method of relationship annotation between two entities, and the implementation method of event extraction between multiple entities. Real-time visualization of nested entity annotations. In the iterative labeling of text support, the labeling content of the labelled text is recognized and imported into the database, which optimizes the user labeling experience and improves labeling efficiency. In order to improve the accuracy of labeling results, additional review and editing functions are added to allow deployment and operation in a variety of system environments. .

Description of drawings

FIG. 1 is a schematic flowchart of a method for implementing manual annotation processing based on ICD-10 standard of depression diagnosis and treatment interview text information according to the present invention.

FIG. 2 is a schematic diagram of manual annotation of a method for implementing manual annotation processing based on the interview text information of the ICD-10 standard for diagnosis and treatment of depression according to the present invention.

FIG. 3 is a schematic diagram of an inquiries according to an embodiment of the method for implementing manual annotation processing based on the interview text information of the ICD-10 standard for diagnosis and treatment of depression according to the present invention.

4 is a schematic diagram of the electronic medical record processing result of the manual labeling of the method for implementing manual labeling processing based on the interview text information of the ICD-10 depression diagnosis and treatment standard of the present invention.

Detailed ways

In order to describe the technical content of the present invention more clearly, further description will be given below with reference to specific embodiments.

The present invention is based on the ICD-10 depression diagnosis and treatment standard interview text information to realize the method of manual labeling and processing, which comprises the following steps:

(1) Construct a domain dictionary by accurately matching manual labels, and label the Chinese electronic medical record named entity recognition and labeling corpus according to the defined labeling specifications;

(2) Simultaneously perform manual labeling and automatic labeling to identify Chinese electronic medical record named entities.

As a preferred embodiment of the present invention, the step of constructing a domain dictionary in the step (1) is specifically:

Obtain keywords from Chinese electronic medical records or use external professional resources to obtain dictionary keywords to construct domain dictionaries.

As a preferred embodiment of the present invention, manual marking is performed in the step (2), which specifically includes the following steps:

(2.1) Log in to the depression intelligent diagnosis and case generation system, and fill in the patient consultation information;

(2.2) Upload recording files;

(2.3) Perform recording recognition and convert the recording file into text content;

(2.4) Mark the text content;

(2.5) Generate a diagnostic report.

As a preferred embodiment of the present invention, the step of deep learning in the step (2) is specifically:

Based on deep learning, domain corpus pre-training word embedding and domain preprocessing are used for related entity recognition models.

As a preferred embodiment of the present invention, the method further comprises the following steps:

(3) Build a manual labeling model, identify and interview medical entities of electronic medical records through the manual labeling model, and display and analyze the results of medical entity recognition in psychiatric medical records.

The device of the present invention for realizing manual annotation processing of interview text information based on the ICD-10 standard of diagnosis and treatment for depression, wherein the device includes:

a processor configured to execute computer-executable instructions;

The memory stores one or more computer-executable instructions, and when the computer-executable instructions are executed by the processor, each of the above-mentioned methods for implementing manual annotation processing based on the ICD-10 depression diagnosis and treatment standard interview text information is realized. step.

The processor of the present invention for implementing manual annotation processing of interview text information based on ICD-10 depression diagnosis and treatment standard, wherein the processor is configured to execute computer-executable instructions, and the computer-executable instructions are When the processor described above is executed, each step of the above-mentioned method for manual annotation processing based on the interview text information of the ICD-10 standard for diagnosis and treatment of depression is realized.

The computer-readable storage medium of the present invention has a computer program stored thereon, and the computer program can be executed by the processor to implement each of the above-mentioned methods for manual annotation processing based on the interview text information of the ICD-10 standard for diagnosis and treatment of depression. step.

Since the text annotation tools on the market cannot meet the needs of the actual depression field projects, this project independently developed a web-based text annotation tool to build a high-quality corpus. The tool needs to support basic labeling functions such as entity labeling, ICD-10 indicators, event extraction, text classification, etc. It requires that labeling specifications can be customized, and text can be iteratively labelled. It is suitable for labeling tasks of large-scale entity types, and can expand nested entity labeling. , standard name tagging and pre tagging based on dictionary matching and regular matching. On the premise of satisfying the annotation function, optimize the annotation experience as much as possible, reduce the user's workload and cost consumption, and ensure the accuracy of the annotation results. Annotation works well on mainstream operating systems Windows, Linux and Mac.

In a specific embodiment of the present invention, the present invention provides a proprietary manual labeling ICD-10 diagnosis and treatment model that supports the field of mental health, which is easy to support innovative research.

The invention adds a manual labeling ICD-10 label system to the depression intelligent diagnosis and case generation system, and conducts structured analysis on the interview results. The user can label the text record through the web visual interface, and edit the formed label.

The present invention solves the above-mentioned technical problems through the following technical means: considering the language characteristics of Chinese interview medical record texts, and the current situation that there is no unified labeling norm and public labeling corpus for research, this patent is devoted to the application of small-scale labeling corpus. It can improve the performance of Chinese medical entity recognition model under different circumstances. The work done by this patent can be summarized into the following three parts:

(1) Using statistical methods to obtain keywords from Chinese electronic medical records and using external professional resources to obtain dictionary keywords to build domain dictionaries. At the same time, according to the defined annotation specification, the Chinese electronic medical record named entity recognition and annotation corpus is annotated. Among them, the statistical methods are exact matching, manual labeling, and automatic storage of results. For example, using the results of a systematic doctor-patient interview, the doctor selects the keyword "can't sleep" by exact matching. Insomnia" symptom keyword: "can't sleep". At the same time, phrases that fully contain keywords (the word order cannot be reversed), such as: "I can't sleep", "I often can't sleep".

(2) Research on named entity recognition of Chinese electronic medical records, which makes doctors based on two categories of manual annotation and automatic annotation. The method based on manual annotation combines the conditional random field with the domain dictionary, and uses the pre-labeling-secondary labeling process of two different granularities to improve the effect of named entity recognition. The deep learning-based method uses two networks, BLSTM-CRF and TransformerCRF, and adopts two domain preprocessing methods, namely, domain corpus pre-training word embedding and Fine-turing for related entity recognition models, so that deep learning methods can be better applied to medical treatment. Medical standard entity recognition.

Manual and automatic labeling are performed simultaneously. Depression interview medical records and auxiliary diagnosis and treatment system, manual labeling and uploading steps are as follows:

a. Depression intelligent diagnosis and case generation system: login;

b. Depression intelligent diagnosis and case generation system: fill in the patient consultation information and click save;

c. Depression intelligent diagnosis and case generation system: click on the audio to upload the recording;

d. Depression intelligent diagnosis and case generation system: select the recording file and open it;

e. Depression intelligent diagnosis and case generation system: After uploading the file, click on the recording to identify;

f. Depression intelligent diagnosis and case generation system: After waiting for a few minutes, click to get the text and save it;

g. Depression intelligent diagnosis and case generation system: Click the mark in the inquiry list to mark it;

h. Depression intelligent diagnosis and case generation system: label operation, save, and generate diagnosis report.

(3) Other applications of manual annotation models. The universality of the manual annotation model for the recognition of medical entities in interview electronic medical records and the display and analysis of the results of medical entity recognition in real psychiatric medical records in hospitals are respectively studied.

After the tester submits the test report and the test development process, the developer confirms the test results and confirms that the ASR recognition accuracy rate is more than 90%, the manual labeling function can be realized, and the medical record report can be generated, and all bugs have been fixed, and the repair is completed. Submit and re-confirm. Once again, the testers conduct regression confirmation on the submitted repair results to confirm that all system bugs have been cleaned up.

For the specific implementation scheme of this embodiment, reference may be made to the relevant descriptions in the foregoing embodiments, which will not be repeated here.

It can be understood that, the same or similar parts in the above embodiments may refer to each other, and the content not described in detail in some embodiments may refer to the same or similar content in other embodiments.

It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for the purpose of description, and should not be construed as indicating or implying relative importance. Furthermore, in the description of the present invention, unless otherwise specified, the meaning of "plurality" means at least two.

Any description of a process or method in the flowcharts or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing a specified logical function or step of the process , and the scope of the preferred embodiments of the invention includes alternative implementations in which the functions may be performed out of the order shown or discussed, including performing the functions substantially concurrently or in the reverse order depending upon the functions involved, which should It is understood by those skilled in the art to which the embodiments of the present invention belong.

It should be understood that various parts of the present invention may be implemented in hardware, software, firmware or a combination thereof. In the above-described embodiments, various steps or methods may be implemented in software or firmware stored in memory and executed by suitable instruction execution means. For example, if implemented in hardware, as in another embodiment, it can be implemented by any one or a combination of the following techniques known in the art: Discrete logic circuits, application specific integrated circuits with suitable combinational logic gates, Programmable Gate Arrays (PGA), Field Programmable Gate Arrays (FPGA), etc.

Those skilled in the art can understand that all or part of the steps carried by the methods of the above embodiments can be completed by instructing the relevant hardware through a program, and the corresponding program can be stored in a computer-readable storage medium, and the program can be executed when the program is executed. , including one or a combination of the steps of the method embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing module, or each unit may exist physically alone, or two or more units may be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware, and can also be implemented in the form of software function modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, and the like.

In the description of this specification, description with reference to the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples", etc., mean specific features described in connection with the embodiment or example , structure, material or feature is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Using the method, device, processor and computer-readable storage medium for manual annotation processing based on ICD-10 depression diagnosis and treatment standard interview text information of the present invention, each annotation result needs to be visually displayed in real time, and each annotation needs to be Supports selection of a large number of entity types. The implementation method of relationship annotation between two entities, and the implementation method of event extraction between multiple entities. Real-time visualization of nested entity annotations. In the iterative labeling of text support, the labeling content of the labelled text is recognized and imported into the database, which optimizes the user labeling experience and improves labeling efficiency. In order to improve the accuracy of labeling results, additional review and editing functions are added to allow deployment and operation in a variety of system environments. .

In this specification, the invention has been described with reference to specific embodiments thereof. However, it will be evident that various modifications and changes can still be made without departing from the spirit and scope of the invention. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense.

Claims (8)

1. A method for realizing manual labeling processing based on ICD-10 depression diagnosis and treatment standard interview text information is characterized by comprising the following steps:

(1) constructing a domain dictionary in a statistical mode of accurately matching handlabels, and labeling a named entity recognition labeling corpus of the Chinese electronic medical record according to a defined labeling specification;

(2) and performing manual labeling and automatic labeling at the same time, and identifying the named entity of the Chinese electronic medical record.

2. The method for realizing manual labeling processing based on ICD-10 depression diagnosis and treatment standard interview text information according to claim 1, wherein the step of constructing the domain dictionary in the step (1) is specifically as follows:

and obtaining keywords from the Chinese electronic medical record or obtaining dictionary keywords by using external professional resources to construct a domain dictionary.

3. The method for realizing manual labeling processing based on the ICD-10 depression diagnosis and treatment standard interview text information according to claim 1, wherein the manual labeling in the step (2) specifically comprises the following steps:

(2.1) logging in an intelligent depression diagnosis and case generation system and filling out inquiry information of a patient;

(2.2) uploading the recording file;

(2.3) carrying out sound recording identification, and converting the sound recording file into text content;

(2.4) labeling the text content;

and (2.5) generating a diagnosis report.

4. The method for realizing manual labeling processing based on the ICD-10 depression diagnosis and treatment standard interview text information according to claim 1, wherein the step (2) of deep learning comprises the following specific steps:

and embedding the domain linguistic data pre-training words and performing a domain preprocessing mode on the relevant entity recognition model based on deep learning.

5. The method for realizing manual labeling processing based on ICD-10 depression diagnosis and treatment standard interview text information according to claim 1, wherein the method further comprises the following steps:

(3) and constructing a manual labeling model, identifying the interview electronic medical record medical entity through the manual labeling model, and displaying and analyzing the identification result of the psychiatric medical record medical entity.

6. An apparatus for implementing manual annotation processing based on ICD-10 depression diagnosis and treatment standard interview text information, characterized in that the apparatus comprises:

a processor configured to execute computer-executable instructions;

a memory storing one or more computer-executable instructions that, when executed by the processor, perform the steps of the method of any one of claims 1 to 5 for performing a manual tagging process based on ICD-10 depression medical standard interview text information.

7. A processor for implementing a manual tagging process based on ICD-10 depression diagnosis and treatment standard interview text information, characterized in that the processor is configured to execute computer executable instructions which, when executed by the processor, implement the steps of the method for implementing a manual tagging process based on ICD-10 depression diagnosis and treatment standard interview text information as claimed in any one of claims 1 to 5.

8. A computer-readable storage medium, having stored thereon a computer program executable by a processor to perform the steps of the method of any one of claims 1 to 5 for performing manual tagging based on ICD-10 depressive illness treatment standard interview text information.

Images