Single-disease reporting method, system, terminal and medium

Abstract

The application provides a single disease reporting method and system, a terminal and a medium. The method comprises the following steps: obtaining disease reporting standard information, constructing a disease knowledge framework according to the disease reporting standard information; constructing an event set according to the disease knowledge framework, and obtaining event questions of each disease event in the event set respectively; determining relevant text in a sample case according to the event questions, constructing pre-annotation training data according to the relevant text and the event set; training an event implication model according to the pre-annotation training data, determining to-be-reasoned data according to a to-be-reported medical record; inputting the to-be-reasoned data into the trained event implication model for event prediction, and obtaining single disease reporting information. Based on the disease reporting standard information, the disease knowledge framework representing the key information of the corresponding single disease can be automatically constructed, so that the event implication model can effectively learn the key information in the pre-annotation training data, and the accuracy of single disease reporting is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to a single disease reporting method and system, a terminal and a medium. BACKGROUND

[0002] Single disease reporting is a series of problem reporting of patients admitted to hospital for a certain disease according to medical records and various diagnosis and treatment records according to the requirements of the state and the needs of doctors. Single disease reporting is of great significance to improve the level of medical service supervision and protect patient safety. In the single disease reporting process, key information of relevant data items needs to be extracted from medical record data, so as to select answers to the corresponding questions.

[0003] In the existing single disease reporting process, a reading comprehension model of a neural network is used for single disease reporting. In the existing reading comprehension model training process, medical record data is directly input into the reading comprehension model for model training. However, due to the overlong and complicated content of the medical record data, the reading comprehension model cannot learn the key information, which easily leads to inaccurate model prediction and reduces the accuracy of single disease reporting. SUMMARY

[0004] The purpose of the embodiment of the present application is to provide a single disease reporting method, system, terminal and medium, which aims to solve the problem of low accuracy of single disease reporting in the existing single disease reporting process.

[0005] The embodiment of the present application is implemented in the following manner. A single disease reporting method comprises the following steps:

[0006] Obtaining disease reporting standard information and constructing a disease knowledge framework according to the disease reporting standard information;

[0007] Constructing an event set according to the disease knowledge framework, and obtaining event questions of each disease event in the event set;

[0008] Determining relevant text in a sample case according to the event questions, and constructing pre-annotation training data according to the relevant text and the event set;

[0009] Training an event implication model according to the pre-annotation training data, and determining to-be-reasoned data according to a to-be-reported medical record;

[0010] Inputting the to-be-reasoned data into the trained event implication model to perform event prediction, and obtaining single disease reporting information.

[0011] Preferably, the disease knowledge framework is constructed according to the disease reporting standard information, and comprises the following steps:

[0012] respectively acquire data item information, event item information and question and answer item information in the disease reporting standard information;

[0013] combine the data item information, the event item information and the question and answer item information to obtain the disease knowledge framework.

[0014] Preferably, the respective acquisition of the data item information, the event item information and the question and answer item information in the disease reporting standard information comprises:

[0015] acquire disease reporting codes and disease reporting questions in each disease reporting standard information, and construct the data item information according to the disease reporting codes and the disease reporting questions;

[0016] acquire event occurrence positions and event contents in each disease reporting standard information, and construct the event item information according to the event occurrence positions and the event contents;

[0017] acquire question and answer answers and answer identifiers corresponding to each question and answer answer, and construct the question and answer item information according to the question and answer answers and the answer identifiers.

[0018] Preferably, the determination of the relevant text in the sample case according to the event question comprises:

[0019] determine an extraction position according to a question identifier of the event question, and perform text extraction on the sample case according to the extraction position to obtain the relevant text.

[0020] Preferably, the construction of the event set according to the disease knowledge framework comprises:

[0021] for each disease knowledge framework, the disease reporting codes, the event occurrence positions, the event contents and the question and answer answers are respectively stored to obtain the event set.

[0022] Preferably, the construction of the pre-annotation training data according to the relevant text and the event set comprises:

[0023] respectively acquire text semantics of the relevant text, and combine the text semantics with the event set to obtain combined data;

[0024] perform data annotation on the combined data to obtain the pre-annotation training data.

[0025] Preferably, after the determination of the relevant text in the sample case according to the event question, it further comprises:

[0026] perform word segmentation on the relevant text to obtain relevant segmented words, and respectively determine event correlation degrees between the relevant segmented words and the event question.

[0027] According to the event correlation degree, the sentences composed of the related words are screened, and the related text is determined according to the screened sentences.

[0028] Another purpose of the embodiment of the application is to provide a single disease reporting system, which comprises:

[0029] A framework modeling module is configured to acquire disease reporting standard information and construct a disease knowledge framework according to the disease reporting standard information.

[0030] A question acquisition module is configured to construct an event set according to the disease knowledge framework and acquire event questions of each disease event in the event set.

[0031] A data construction module is configured to determine related text in a sample medical record according to the event questions and construct pre-annotation training data according to the related text and the event set.

[0032] A model training module is configured to train an event implication model according to the pre-annotation training data and determine to-be-reasoned data according to a to-be-reported medical record.

[0033] An event prediction module is configured to input the to-be-reasoned data into the trained event implication model to perform event prediction and obtain single disease reporting information.

[0034] Another purpose of the embodiment of the application is to provide a terminal device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above method when executing the computer program.

[0035] Another purpose of the embodiment of the application is to provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the above method.

[0036] In the embodiment of the application, the disease knowledge framework can be automatically constructed based on disease reporting standard information, the key information of the corresponding single disease can be effectively represented based on the disease knowledge framework, the event implication model can effectively learn the key information in the pre-annotation training data, the accuracy of single disease reporting is improved, the event set can be automatically constructed based on the disease knowledge framework, the data items of each disease event are independent of each other, the accuracy of the event implication model after model training is effectively improved, the event implication model is used instead of the reading comprehension model, the matching space of model learning is reduced, the task complexity is reduced, and the training efficiency of the event implication model is effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 is a flow chart of the single disease reporting method provided by the first embodiment of the present application;

[0038] Figure 2 is a structural schematic diagram of the single disease reporting system provided by the second embodiment of the present application;

[0039] Figure 3 is a structural schematic diagram of the single disease reporting system provided by the third embodiment of the present application;

[0040] Figure 4 is a flow chart of the specific implementation of the single disease reporting system provided by the third embodiment of the present application;

[0041] Figure 5 is a schematic diagram of the event implication model provided by the third embodiment of the present application;

[0042] Figure 6 is a structural schematic diagram of the terminal device provided by the fourth embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0044] In order to illustrate the technical solutions of the present application, specific embodiments will be described below.

[0045] Example One

[0046] Please refer to Figure 1 is a flow chart of the single disease reporting method provided by the first embodiment of the present application, which can be applied to any terminal device or system. The single disease reporting method comprises the following steps:

[0047] In step S10, the disease reporting standard information is obtained, and a disease knowledge framework is constructed according to the disease reporting standard information.

[0048] Among them, the disease reporting standard information is the national standard for single disease reporting, that is, the knowledge framework of the options is established according to the national standard for single disease reporting, and the disease knowledge framework is obtained.

[0049] Optionally, in this step, the disease knowledge framework is constructed according to the disease reporting standard information, which comprises:

[0050] respectively obtaining data item information, event item information and question and answer item information in the disease reporting standard information, wherein the data item information includes information data corresponding to a data item in the disease reporting standard information, the event item information includes information data corresponding to an event item in the disease reporting standard information, and the question and answer item information includes information data corresponding to a question and answer item in the disease reporting standard information;

[0051] combining the data item information, the event item information and the question and answer item information to obtain the disease knowledge framework, wherein the data item information, the event item information and the question and answer item information are combined to obtain key information of each disease.

[0052] Further, the respective obtaining of the data item information, the event item information and the question and answer item information in the disease reporting standard information comprises:

[0053] obtaining disease reporting codes and disease reporting questions in each disease reporting standard information, and constructing the data item information according to the disease reporting codes and the disease reporting questions;

[0054] obtaining event occurrence positions and event contents in each disease reporting standard information, and constructing the event item information according to the event occurrence positions and the event contents;

[0055] obtaining question and answer answers and answer identifiers corresponding to each question and answer answer, and constructing the question and answer item information according to the question and answer answers and the answer identifiers;

[0056] For example, ICH-1-3-5 whether the hematoma breaks the ventricle: y-yes, n-no, the following disease knowledge framework is established:

[0057] The data item information includes "itemId":"ICH-1-3-5"-data item code (disease reporting code) and "question":"whether the hematoma breaks into the ventricle"-data item content (disease reporting question);

[0058] The event item information includes "event_id":0-event id, "event_loc":"ventricle"-event occurrence position, "event_core":"hematoma breaks into"-event content and "event_time":"PRESENT"-event occurrence event;

[0059] The question and answer item information includes "opt_name":"y"-option name, "opt_content":"yes"-option content;

[0060] "opt_rules":"HAPPENED:0==HAPPENED","CONFLICT:0==CONFLICT","NOTREF:0==NOTREF" - the condition for this option to be true;

[0061] Where HAPPENED indicates that it has occurred, CONFLICT indicates that it has conflicted, NOTREF indicates that it has not been mentioned, and 0 represents the event_id in opt_events;

[0062] "opt_name":"n", "opt_content":"no", "opt_rules":[], "opt_events":[]

[0063] "item_rules":"y:y==HAPPENED","n:y!=HAPPENED" - When y==HAPPENED, select y - Yes; otherwise, select n - No;

[0064] The principles for defining "opt_events" include: events should be as general as possible and not be further subdivided; the semantics of events within the options should not overlap with each other; and event names should conform as closely as possible to doctors' spoken language habits.

[0065] Step S20: Construct an event set based on the disease knowledge framework, and obtain the event questions for each disease event in the event set;

[0066] Among them, the disease-based knowledge framework can effectively represent the key information of a single disease. For example, the event set corresponding to ICH-1-3-5 is "y ventricle_hematoma rupture_PRESENT". The disease-based knowledge framework can automatically construct the event set, making the event data items of each disease independent of each other, which effectively improves the accuracy of the event-entailed model after model training.

[0067] Optionally, in this step, constructing the event set based on the disease knowledge framework includes:

[0068] For each disease knowledge framework, the disease reporting code, the location of the event, the content of the event, and the question-and-answer answer are stored respectively to obtain the event set.

[0069] Step S30: Determine relevant texts in the sample cases based on the event question, and construct pre-labeled training data based on the relevant texts and the event set;

[0070] The relevant text is obtained by respectively querying the medical record positions of each event question in the medical record text and extracting the text at the medical record positions.

[0071] Optionally, in the step, the relevant text in the sample medical record is determined according to the event question, including:

[0072] The extraction position is determined according to the question identifier of the event question, and the relevant text is obtained by performing text extraction on the sample medical record according to the extraction position; the question identifier can be stored in the form of characters, numbers or codes.

[0073] Further, in the step, the pre-labeled training data is constructed according to the relevant text and the event set, including:

[0074] The text semantics of the relevant text are obtained respectively, the text semantics are combined with the event set to obtain combination data, the combination data are labeled to obtain the pre-labeled training data, for example, the pre-labeled training data includes: bilateral temporal lobe uncinate gyrus downward displacement ventricle_hematoma break into_PRESENT N, right cerebral hemisphere brain hemorrhage break into ventricle ventricle_hematoma break into_PRESENT H.

[0075] In step S40, the event implication model is trained according to the pre-labeled training data, and the to-be-reasoned data is determined according to the to-be-filled medical record;

[0076] By using the event implication model instead of the reading comprehension model, the matching space of model learning is reduced, the task complexity is reduced, and the training efficiency of the event implication model is effectively improved. In the step, the BERT learning method is used to control the event implication model to perform unsupervised learning on the pre-labeled training data, so that the event implication model automatically learns various information in the pre-labeled training data, the event implication model is controlled to perform linear prediction on the pre-labeled training data, the model loss of the event implication model is determined according to the linear prediction result, and the parameters of the event implication model are updated according to the model loss until the event implication model converges.

[0077] In step S50, the to-be-reasoned data is input into the trained event implication model to perform event prediction, and single-disease reporting information is obtained.

[0078] By inputting the to-be-reasoned data into the trained event implication model to perform event prediction, the option answers of each question in the to-be-filled medical record are output.

[0079] In this embodiment, the disease type reporting standard information can be used to automatically construct a disease type knowledge framework, and the disease type knowledge framework can effectively represent the key information of the corresponding single disease type, so that the event implication model can effectively learn the key information in the pre-labeled training data, thereby improving the accuracy of single disease type reporting. The event set can be automatically constructed based on the disease type knowledge framework, so that each disease event data item is independent of each other, thereby effectively improving the accuracy of the event implication model after model training. By using the event implication model instead of the reading comprehension model, the matching space of model learning is reduced, the task complexity is reduced, and the training efficiency of the event implication model is effectively improved.

[0080] Example Two

[0081] Referring to Figure 2 is a flowchart of a secret embedding sharing method provided by the second embodiment of the present application, which is used to further refine the steps after step S50 in the first embodiment, including the following steps:

[0082] Step S60, performing word segmentation on the related text to obtain related words, and determining the event correlation degree between the related words and the event question respectively;

[0083] In this step, the association vocabulary corresponding to the event question is queried, and each related word is matched with the association vocabulary to obtain the event correlation degree. The association vocabulary stores the corresponding relationship between different words and the corresponding event correlation degree.

[0084] Step S70, screening the sentences composed of the related words according to the event correlation degree, and determining the related text according to the screened sentences;

[0085] In this step, the sum of the event correlation degrees between the related words in each sentence is calculated to obtain a total correlation value. If the total correlation value corresponding to any sentence is less than the correlation threshold, the sentence corresponding to the correlation threshold is deleted. The screened sentences are determined as the related text.

[0086] In this embodiment, by performing word segmentation on the related text, the determination of the event correlation degree between each related word and the event question is effectively guaranteed. By screening the sentences composed of the related words according to the event correlation degree, sentences with less correlation with the event question in the related text can be effectively deleted, thereby improving the accuracy of the related text.

[0087] In the embodiment, a certain knowledge point of whether the text contains an option is modeled, an event implication model is used instead of a reading comprehension model, a matching space learned by the model is reduced, task complexity is reduced, only the internal of each question data item in single disease reporting is modeled, and each disease data item is independent of each other, which is beneficial to improve the accuracy of the model. The embodiment is based on a national standard, has interpretability, reusability, can form effective knowledge accumulation, and is easy to extend to other tasks, and development efficiency is improved.

[0088] Example Three

[0089] Please refer to Figure 3 It is a structural schematic diagram of a single disease reporting system 100 provided by the third embodiment of the application, comprising: a framework modeling module 10, a question acquisition module 11, a data construction module 12, a model training module 13 and an event prediction module 14, wherein:

[0090] The framework modeling module 10 is used for acquiring disease reporting standard information and constructing a disease knowledge framework according to the disease reporting standard information.

[0091] Optionally, the framework modeling module 10 is further used for acquiring data item information, event item information and question and answer item information in the disease reporting standard information respectively.

[0092] The data item information, the event item information and the question and answer item information are combined to obtain the disease knowledge framework.

[0093] Further, the framework modeling module 10 is further used for acquiring disease reporting codes and disease reporting questions in each disease reporting standard information, and constructing the data item information according to the disease reporting codes and the disease reporting questions.

[0094] The event occurrence positions and the event contents in each disease reporting standard information are acquired, and the event item information is constructed according to the event occurrence positions and the event contents.

[0095] The question and answer answers and the answer identifiers corresponding to each question and answer answer in each disease reporting standard information are acquired, and the question and answer item information is constructed according to the question and answer answers and the answer identifiers.

[0096] The question acquisition module 11 is used for constructing an event set according to the disease knowledge framework, and acquiring event questions of each disease event in the event set respectively.

[0097] Optionally, the question acquisition module 11 is further used for storing the disease reporting codes, the event occurrence positions, the event contents and the question and answer answers corresponding to each disease knowledge framework respectively to obtain the event set.

[0098] The data construction module 12 is configured to determine relevant text in a sample case according to the event question, and construct pre-labeled training data according to the relevant text and the event set.

[0099] Optionally, the data construction module 12 is further configured to determine an extraction position according to a question identifier of the event question, and perform text extraction on the sample case according to the extraction position to obtain the relevant text.

[0100] Further, the data construction module 12 is further configured to obtain text semantics of the relevant text respectively, and combine the text semantics with the event set to obtain combined data.

[0101] The combined data is labeled to obtain the pre-labeled training data.

[0102] Further, the data construction module 12 is further configured to perform word segmentation on the relevant text to obtain relevant segmented words, and determine an event correlation degree between the relevant segmented words and the event question respectively.

[0103] The sentences composed of the relevant segmented words are filtered according to the event correlation degree, and the relevant text is determined according to the filtered sentences.

[0104] The model training module 13 is configured to train an event implication model according to the pre-labeled training data, and determine to-be-reasoned data according to a to-be-reported medical record.

[0105] The event prediction module 14 is configured to input the to-be-reasoned data into the trained event implication model to perform event prediction, and obtain single-disease reporting information.

[0106] Specifically, please refer to Figure 4 is a schematic diagram of specific implementation steps of the single-disease reporting system 100 of the embodiment:

[0107] (1) Establish a knowledge framework of options according to the national standard of single-disease reporting;

[0108] For example, ICH-1-3-5 Hematoma breaks into ventricle: y-yes, n-no, the following disease knowledge framework is established:

[0109] The data item information includes "itemId":"ICH-1-3-5"-data item code (disease reporting code) and "question":"whether the hematoma breaks into the ventricle"-data item content (disease reporting question);

[0110] The event item information includes "event_id": 0-event id, "event_loc": "ventricle"-event occurrence location, "event_core": "hematoma break in"-event content, and "event_time": "PRESENT"-event occurrence event;

[0111] The question and answer item information includes "opt_name": "y"-option name, "opt_content": "yes"-option content;

[0112] "opt_rules": "HAPPENED: 0 == = HAPPENED", "CONFLICT: 0 == = CONFLICT", "NOTREF: 0 == = NOTREF"-the option is established condition;

[0113] Wherein, HAPPENED indicates occurrence, CONFLICT indicates conflict, NOTREF indicates not mentioned, and 0 is the event id in opt_events;

[0114] "opt_name": "n", "opt_content": "no", "opt_rules": [], "opt_events": []

[0115] "item_rules": "y: y == = HAPPENED", "n: y! = = HAPPENED"-when y == = HAPPENED, select y-yes, otherwise select n-no;

[0116] Wherein, the event "opt_events" definition principle includes: the event principle is as much as possible, cannot be subdivided, the event semantics in the option, do not cover each other, and the event name should be as much as possible to meet the doctor's oral habit.

[0117] (2) generating an event set according to the knowledge framework defined in (1)

[0118] For example, the event set corresponding to ICH-1-3-5 is "yventricle_hematoma break in_PRESENT";

[0119] (3) according to the location of the data item question in the medical record, extract the related text, and use the event set generated in (2) to generate the corresponding pre-labeled training data, such as ICH-1-3-5 corresponding to the content of "check report" in the medical record, the generated pre-labeled training data (N corresponds to NOTREF, H corresponds to HAPPEND, and C corresponds to CONFLICT):

[0120] Bilateral temporal lobe uncus downward displacement ventricle_hematoma break in_PRESENT N

[0121] Right cerebral hemisphere hemorrhage into the ventricle Ventricular hemorrhage into the present H

[0122] The pre-annotation training data is formed after manual review;

[0123] (4) Training using event implication model, model structure is as follows Figure 5 :

[0124] Wherein, by adopting BERT learning method, the event implication model is controlled to perform unsupervised learning on the pre-annotation training data, so that the event implication model automatically learns various information in the pre-annotation training data, the event implication model is controlled to perform linear prediction on the pre-annotation training data, the model loss of the event implication model is determined according to the linear prediction result, and the parameters of the event implication model are updated according to the model loss until the event implication model converges.

[0125] (5) Inference stage, according to (2) and (3), generate data to be inferred, and according to the knowledge framework "item_rules" defined in (1), give answer options; Specifically, input the data to be inferred into the trained event implication model to perform event prediction, and obtain single disease reporting information.

[0126] In this embodiment, whether the text contains a certain knowledge point of the option is modeled, the event implication model is used instead of the reading comprehension model, the matching space of model learning is reduced, the task complexity is reduced, only the internal modeling of each question data item in single disease reporting is performed, and each disease data item is independent of each other, which is beneficial to improve the accuracy of the model. This embodiment is based on the national standard, has interpretability, reusability, can form effective knowledge accumulation, and is easy to extend to other tasks, and improves the development efficiency.

[0127] Example Four

[0128] Figure 6 is a structural block diagram of a terminal device 2 provided by the fourth embodiment of the present application. As shown in Figure 6 , the terminal device 2 of this embodiment comprises a processor 20, a memory 21, and a computer program 22 stored in the memory 21 and executable on the processor 20, such as a single disease reporting method program. The processor 20 implements the steps in each embodiment of the above-mentioned various single disease reporting methods when executing the computer program 22.

[0129] For example, the computer program 22 can be divided into one or more modules, which are stored in the memory 21 and executed by the processor 20 to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program 22 in the terminal device 2. The terminal device can include, but is not limited to, the processor 20, the memory 21.

[0130] The processor 20 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0131] The memory 21 can be an internal storage unit of the terminal device 2, such as a hard disk or a memory of the terminal device 2. The memory 21 can also be an external storage device of the terminal device 2, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 21 can include both the internal storage unit and the external storage device of the terminal device 2. The memory 21 is used to store the computer program and other programs and data required by the terminal device. The memory 21 can also be used to temporarily store data that has been output or will be output.

[0132] In addition, each functional module in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically independently, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0133] The integrated module, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Among them, the computer readable storage medium can be non-volatile or volatile. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can be executed by a processor to implement the steps of each method embodiment described above. Among them, the computer program includes computer program code, which can be in the form of source code, object code, executable file or some intermediate form, etc. The computer readable storage medium can include any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable storage medium does not include electric carrier signal and telecommunication signal.

[0134] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A single disease reporting method, characterized by, The method comprises: acquiring disease reporting standard information, and constructing a disease knowledge framework according to the disease reporting standard information; constructing an event set according to the disease knowledge framework, and acquiring event questions of each disease event in the event set respectively; determining relevant text in a sample case according to the event questions, and constructing pre-labeled training data according to the relevant text and the event set; training an event implication model according to the pre-labeled training data, and determining to-be-reasoned data according to a to-be-reported medical record; inputting the to-be-reasoned data into the trained event implication model for event prediction to obtain single-disease reporting information.

2. The single disease reporting method of claim 1, wherein, The disease knowledge framework is constructed according to the disease reporting standard information, comprising: acquiring data item information, event item information and question and answer item information in the disease reporting standard information respectively; combining the data item information, the event item information and the question and answer item information to obtain the disease knowledge framework.

3. The single disease reporting method of claim 2, wherein, The data item information, the event item information and the question and answer item information in the disease reporting standard information are acquired respectively, comprising: acquiring disease reporting codes and disease reporting questions in each disease reporting standard information, and constructing the data item information according to the disease reporting codes and the disease reporting questions; acquiring event occurrence positions and event contents in each disease reporting standard information, and constructing the event item information according to the event occurrence positions and the event contents; acquiring question and answer answers and answer identifiers corresponding to each question and answer answer, and constructing the question and answer item information according to the question and answer answers and the answer identifiers.

4. The single disease reporting method of claim 1, wherein, The relevant text in the sample case is determined according to the event questions, comprising: determining an extraction position according to a question identifier of the event question, and performing text extraction on the sample case according to the extraction position to obtain the relevant text.

5. The single disease reporting method of claim 3, wherein, The event set is constructed according to the disease knowledge framework, comprising: for each disease knowledge framework, the disease reporting codes, the event occurrence positions, the event contents and the question and answer answers are stored correspondingly to obtain the event set.

6. The single disease reporting method of claim 1, wherein, The pre-labeled training data is constructed according to the relevant text and the event set, comprising: acquiring text semantics of the relevant text respectively, and combining the text semantics with the event set to obtain combined data; performing data labeling on the combined data to obtain the pre-labeled training data.

7. The single disease category reporting method according to any one of claims 1 to 6, wherein, After the relevant text in the sample case is determined according to the event questions, the method further comprises: performing word segmentation on the relevant text to obtain relevant segmented words, and determining event correlation degrees between the relevant segmented words and the event questions respectively; performing screening on a sentence composed of the relevant segmented words according to the event correlation degrees, and determining the relevant text according to the screened sentence.

8. A single disease reporting system characterized by, The system comprises: a framework construction module, configured to acquire disease reporting standard information, and construct a disease knowledge framework according to the disease reporting standard information; a question acquisition module, configured to construct an event set according to the disease knowledge framework, and acquire event questions of each disease event in the event set respectively; a data construction module configured to determine relevant text in a sample case according to the event question, and construct pre-labeled training data according to the relevant text and the event set; a model training module configured to train an event implication model according to the pre-labeled training data, and determine to-be-reasoned data according to a to-be-reported medical record; an event prediction module configured to input the to-be-reasoned data into the trained event implication model to perform event prediction, and obtain single-disease reporting information.

9. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1-9. The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.

Images