CN117409919A

CN117409919A - Data processing method, device, electronic equipment and storage medium

Info

Publication number: CN117409919A
Application number: CN202311382036.6A
Authority: CN
Inventors: 胡加学; 孔祥威; 王翔; 赵景鹤; 贺志阳; 鹿晓亮; 王士进; 魏思; 刘聪; 胡国平
Original assignee: iFlytek Co Ltd
Current assignee: iFlytek Co Ltd
Priority date: 2023-10-23
Filing date: 2023-10-23
Publication date: 2024-01-16

Abstract

The application provides a data processing method, a data processing device, electronic equipment and a storage medium, wherein the data processing method performs pre-training based on physical examination report information and fine-tuning training of physical examination report interpretation tasks on a pre-trained generated language model to obtain a report analysis model capable of performing physical examination report interpretation based on physical examination information. On the basis, physical examination information is obtained from the physical examination report, then a task prompt instruction which at least contains the physical examination information and the physical examination report interpretation task description information is generated, and the task prompt instruction is input into the pre-trained report analysis model, so that analysis results of the physical examination report can be obtained. The proposal realizes the automatic interpretation of the physical examination report, thereby improving the interpretation efficiency of the physical examination report.

Description

Data processing method, device, electronic equipment and storage medium

Technical Field

The present disclosure relates to the field of artificial intelligence technologies, and in particular, to a data processing method, an apparatus, an electronic device, and a storage medium.

Background

Physical examination can help people find and diagnose potential problems on the body early so as to take therapeutic measures or prevent diseases in early stages. Through physical examination report, people can know own physical condition in detail, such as weight, blood sugar, blood fat and the like, and can accurately grasp own physical condition. This is very beneficial for everyone to adjust his or her lifestyle and keep his or her body healthy.

The interpretation of current physical examination reports is mostly dependent on the report presented by the physical examination center, which is referred to by a physician to interpret the physical health of the patient. However, as the importance of physical health increases, the increase in the number and frequency of physical examination makes it difficult for physicians to fully address the needs of patient's medical report interpretation, and thus more efficient medical report interpretation methods are needed.

Disclosure of Invention

In view of the above state of the art, the present application provides a data processing method, apparatus, electronic device and storage interpretation.

According to a first aspect of an embodiment of the present application, there is provided a data processing method, including:

acquiring physical examination information from a physical examination report;

generating a first task prompt instruction at least comprising the physical examination information and first task description information, wherein the first task description information comprises description information of a physical examination report interpretation task, and the physical examination report interpretation task is used for interpreting the physical examination report based on the physical examination information;

inputting the first task prompting instruction into a pre-trained report analysis model to obtain an interpretation result of the physical examination report;

the report analysis model is at least obtained by pre-training the pre-trained generated language model based on physical examination report information and fine tuning training, wherein the fine tuning training comprises physical examination report interpretation training.

Optionally, the acquiring physical examination information from the physical examination report includes:

carrying out structuring treatment on the physical examination report according to structuring treatment rules corresponding to the format of the physical examination report to obtain structured physical examination information;

or,

and extracting a physical examination report conclusion and the test result of an abnormal test item related to the physical examination report conclusion from the physical examination report, and combining the physical examination report conclusion and the test result of the abnormal test item to obtain physical examination information.

Optionally, generating a first task prompting instruction at least including the physical examination information and the first task description information includes:

determining first task description information based on a physical examination report interpretation task, and determining filling descriptions of the physical examination report interpretation template according to a preset physical examination report interpretation template;

and combining the first task description information, the physical examination report interpretation template and the filling instruction of the physical examination report interpretation template to obtain a first task prompt instruction.

Optionally, the physical examination report interpretation template comprises a disease hidden danger description, a physical examination abnormal item description and a project description needing attention;

The abnormal physical examination item description comprises a to-be-diagnosed item description and a to-be-reviewed item description, wherein the to-be-diagnosed item description comprises a to-be-diagnosed item name, a to-be-diagnosed reason and a to-be-diagnosed department, and the to-be-reviewed item description comprises a to-be-reviewed item name, a review reason and a review department; the attention required item description comprises an attention required item name, clinical significance and guiding advice.

Optionally, the method further comprises:

acquiring consultation questions aiming at the physical examination report;

generating a second task prompt instruction at least comprising the consultation problem and second task description information, wherein the second task description information comprises description information of a problem solving task, and the problem solving task is used for solving the consultation problem;

and inputting the second task prompt instruction into the report analysis model to obtain a solution result of the consultation problem.

Optionally, the generating a second task prompt instruction at least including the consultation problem and second task description information includes:

screening out physical examination report information related to the consultation problems from the physical examination reports according to the consultation problems, and screening out physical examination report conclusions from the physical examination reports;

Determining second task description information according to the physical examination report information related to the consultation problem and the physical examination report conclusion, wherein the second task description information comprises description information of a problem solving task, and the problem solving task is used for solving the consultation problem based on the physical examination report information related to the consultation problem and the physical examination report conclusion;

and combining at least the consultation problem, the second task description information, the physical examination report information related to the consultation problem and the physical examination report conclusion to obtain a second task instruction.

Optionally, the screening the physical examination report information related to the consultation problem from the physical examination report according to the consultation problem includes:

inputting the consultation questions and the physical examination report into a pre-trained information screening model to obtain physical examination report information related to the consultation questions;

the information screening model is obtained through physical examination report information retrieval training.

Optionally, the second task instruction further includes a constraint condition description for solving the consultation problem;

wherein the constraint specification includes at least one of a reply request specification, an spam application specification, a fixed reply specification, and a forbidden reply specification.

According to a second aspect of embodiments of the present application, there is provided a data processing apparatus comprising:

an information processing unit for acquiring physical examination information from the physical examination report;

the instruction generation unit is used for generating a first task prompt instruction at least comprising the physical examination information and first task description information, wherein the first task description information comprises description information of a physical examination report interpretation task, and the physical examination report interpretation task is used for interpreting the physical examination report based on the physical examination information;

the report interpretation unit is used for inputting the first task prompting instruction into a pre-trained report analysis model to obtain an interpretation result of the physical examination report;

According to a third aspect of embodiments of the present application, there is provided an electronic device comprising a memory and a processor;

the memory is connected with the processor and used for storing programs;

the processor is used for realizing the data processing method by running the program in the memory.

According to a fourth aspect of embodiments of the present application, there is provided a storage medium having stored thereon a computer program which, when executed by a processor, implements the above-mentioned data processing method.

According to the data processing method, the pre-training based on the physical examination report information and the fine-tuning training of the physical examination report interpretation task are carried out on the pre-trained generated language model, so that the report analysis model capable of carrying out physical examination report interpretation based on the physical examination information is obtained. On the basis, physical examination information is obtained from the physical examination report, then a task prompt instruction which at least contains the physical examination information and the physical examination report interpretation task description information is generated, and the task prompt instruction is input into the pre-trained report analysis model, so that analysis results of the physical examination report can be obtained. The proposal realizes the automatic interpretation of the physical examination report, thereby improving the interpretation efficiency of the physical examination report.

In addition, the physical examination report interpretation method has strong natural language processing capability based on the pre-trained generated language model, and can ensure high physical examination report interpretation quality after the physical examination information-based pre-training and physical examination report interpretation task fine-tuning training.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present application, and that other drawings may be obtained according to the provided drawings without inventive effort to a person skilled in the art.

FIG. 1 is a flowchart of a data processing method according to an embodiment of the present application;

FIG. 2 is a block diagram of a data processing apparatus according to an embodiment of the present application;

fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.

In view of this, embodiments of the present application provide a data processing method, apparatus, electronic device, and storage medium.

Exemplary method

The embodiment of the application firstly provides a data processing method, which is used for performing pre-training based on physical examination report information and fine-tuning training of physical examination report interpretation tasks on a pre-trained generated language model to obtain a report analysis model capable of performing physical examination report interpretation based on physical examination information. On the basis, physical examination information is obtained from the physical examination report, then a task prompt instruction which at least contains the physical examination information and the physical examination report interpretation task description information is generated, and the task prompt instruction is input into the pre-trained report analysis model, so that analysis results of the physical examination report can be obtained. The proposal realizes the automatic interpretation of the physical examination report, thereby improving the interpretation efficiency of the physical examination report.

In an alternative embodiment of the present application, the implementation subject of the data processing method may be a notebook computer, a tablet computer, a desktop computer, a mobile device (e.g., a mobile phone, a personal digital assistant, a dedicated messaging device), or any other type of user terminal or a combination of any two or more of these data processing devices, or may be a server.

As shown in fig. 1, the data processing method includes:

s101, acquiring physical examination information from the physical examination report.

The physical examination report refers to a written report made by a doctor or a medical institution according to the physical examination result after physical examination, and generally includes the health condition of an individual, the detection result of various indexes of the body, and the advice and diagnosis of the doctor.

In an alternative embodiment of the present application, the physical examination information may be all information in the physical examination report, but in consideration of the long length of the physical examination report, the existence of redundant information may cause imaging of subsequent interpretation of the physical examination report, so in another alternative embodiment of the present application, the obtaining physical examination information from the physical examination report includes:

for example, the structural content of the physical examination report may be shown in table 1, where table 1 is a structural physical examination information schematic table provided in the embodiment of the present application:

table 1:

the age, sex, height, weight, date and the physical examination report information result can be added according to the actual condition of the physical examination object, in addition, in order to protect the privacy of the physical examination object, the physical examination report is formatted, and meanwhile, the name of the physical examination object is desensitized, namely, a name column is added in the form of XX.

In another embodiment, the physical examination report is structured, so that the following structural physical examination information can be obtained:

name: age of XX: gender XX: XX height: weight of XX: XX (X)

Reporting date: { date }

Detailed results of physical examination report outliers:

the # radiology examination abnormal item: { examination results and nodules, including CT, X-ray, MRI }

# clinical laboratory examination of abnormal items: { abnormal item, value, normal value range, unit and nodule }

Ultrasound department examination of abnormal item: { examination results and nodules, including B-ultrasound and color ultrasound }

Surgery, gynecology, and other department abnormalities: { nodule }

Report conclusion of # physical examination:

{ summary }

And extracting the specific data content corresponding to each item of the structural physical examination information from the physical examination report, namely, realizing the structural information extraction of the physical examination report, namely, realizing the structural processing corresponding to the format of the physical examination report. The structure of the structured physical examination information can be used as a physical examination report structured template.

Because the formats of the physical examination reports presented by different medical institutions are not exactly identical, the physical examination report structuring template described above may not be suitable for structuring physical examination reports presented by all medical institutions. For this case, the formats of physical examination reports presented by different medical institutions in the market can be collected in advance, and then the structural processing rules of the shape shadows, such as the corresponding physical examination report structural templates, are respectively designed for the physical examination reports in different formats. When a physical examination report in a certain format is obtained, a structural processing rule corresponding to the format or a corresponding physical examination report structural template is adopted to carry out structural processing on the physical examination report, so that structural physical examination information similar to the structural information is obtained.

In another embodiment of the present application, some of the physical examination reports may be physical examination reports that cannot be structured, for example, a structural rule or a structural template of the physical examination report corresponding to the format of the physical examination report is not specified yet, and then the physical examination report cannot be structured directly according to the rule or the template. For those physical examination reports that cannot be structured, it becomes particularly critical to extract core information from the physical examination report to reduce redundancy of the physical examination report, and for this reason, the physical examination information may be specifically obtained from the physical examination report:

The physical examination report conclusion refers to physical examination conclusion information of each physical examination item, and in practical application, different physical examination items correspond to different departments of a hospital, so in the embodiment of the application, the physical examination items may include: radiology department, ultrasound department, surgery department, and the like.

The detection results of the abnormal detection items related to the physical examination report conclusion specifically refer to specific detection results of each abnormal detection item extracted from the physical examination report conclusion. The detection result may be a detection result of information such as an image, a size, a shape, a density, and a color representing the abnormal portion.

For example, for radiology examination of abnormal items, including at least CT, X-ray, MRI detection results and nodules; checking abnormal items for a clinical laboratory, wherein the abnormal items at least comprise abnormal items, values, normal value ranges, units and nodules; checking abnormal items for ultrasonic department, wherein the abnormal items at least comprise detection results and nodules in B ultrasonic and color ultrasonic; for surgical, gynecological and other department abnormalities, at least the nodules are included.

S102, generating a first task prompt instruction at least comprising the physical examination information and first task description information, wherein the first task description information comprises description information of a physical examination report interpretation task, and the physical examination report interpretation task is used for interpreting the physical examination report based on the physical examination information.

The first task prompt instruction is used for indicating an object receiving the instruction to execute a physical examination report interpretation task.

In an embodiment of the present application, the first task prompting instruction is configured to instruct a report interpretation model to interpret the physical examination report based on the instruction.

The first task prompting instruction comprises physical examination information and first task description information, wherein the first task description information refers to description information of a physical examination report interpretation task. Description information of the physical examination report interpretation task, specifically, information describing the physical examination report interpretation task, including but not limited to: the purpose of the physical examination report interpretation task, the interpretation of the interpreted data object, the interpretation method, the interpretation rule, the interpretation notice, the output format requirement of the interpretation result, and the like. The specific content of each item of information of the description information of the physical examination report interpretation task, such as a specific interpretation method, an interpretation rule, an interpretation notice and the like, can be flexibly set according to the actual requirement of the interpretation task.

In some implementations, the generating the first task suggestion instruction that includes at least the physical examination information and the first task description information includes:

determining first task description information based on a physical examination report interpretation task, and determining filling instructions of the physical examination report interpretation template according to a preset physical examination report interpretation template; and combining the first task description information, the physical examination report interpretation template and the filling instruction of the physical examination report interpretation template to obtain a first task prompt instruction.

The first task description information may be description information written according to a specific physical examination report interpretation task by a related staff or a preset task description generating program or model, for example, the first task description information may be description sentences such as "the following is a report abnormal value and suggestion of physical examination, please give I a report interpretation result".

Similar to the first task description information, in order to facilitate the doctor and the patient to understand the interpretation result obtained by interpreting the physical examination report, the interpretation template of the physical examination report may be written by the relevant staff. In the physical examination report interpretation template, structured items to be interpreted are generally included, and physical examination report interpretation is realized by filling in specific contents of the items to be interpreted.

The description of filling the reading template of the physical examination report refers to the description of filling each item to be read in the physical examination report, and may be, for example, a filling requirement, a filling notice, etc. For the physical examination report interpretation template, a filling specification matched with the physical examination report interpretation template can be predetermined. When the physical examination report interpretation template corresponding to the physical examination report interpretation task is determined, the filling instruction corresponding to the physical examination report interpretation template can be determined according to the matching relation between the predetermined template and the filling instruction.

After the first task description information, the physical examination report interpretation template and the filling instruction of the physical examination report interpretation template are respectively obtained, the first task description information, the physical examination report interpretation template and the filling instruction of the physical examination report interpretation template are combined to obtain the first task prompt instruction.

In an alternative embodiment of the present application, the physical examination report interpretation template includes: description of hidden trouble of diseases, description of abnormal items of physical examination and description of items to be concerned;

In order to facilitate understanding of the filling explanation of the physical examination report interpretation template, the following details are described in connection with the filling explanation of the item to be interpreted in the specific physical examination report template.

For example, the following description is written about the item to be interpreted, namely the hidden trouble of the disease: the whole physical examination report is combined to give the disease risk/hidden danger which is mainly the disease hidden danger in the doctor-seeing and the rechecking, and the disease names are needed to be separated.

For another example, the following description is provided for the to-be-read item to be treated: the abnormal physical examination item which needs to be immediately treated is judged according to clinical experience and medical knowledge, the illness state is probably delayed, and the standard of the part of the standard is strict (whether symptoms exist or not, the patient needs to be immediately treated in a hospital, or serious consequences or illness state are probably caused).

For another example, the filling of the item to be interpreted about the reason of the visit is as follows: the language is required to be refined, popular and easy to understand, words are not required to be too many, and the description which can be understood by common patients is used, and comprises the following four aspects: (1) clinical significance of abnormal outcome, (2) what disease or symptoms the outcome may lead to, what harm may be if not taken to the hospital, etc. (3) for the purpose of taking hospital visits, xx auxiliary examination/determination of treatment regimen/s may be needed to exclude xx disease.

In another optional embodiment of the present application, the first task prompting instruction may further include a format requirement of an interpretation result of the physical examination report, for example, the interpretation result is in json format.

As an example, the following is a first task hint instruction generated according to the above embodiments of the present application:

the following is a report of abnormal values and advice for physical examination, please give I a report of physical examination interpretation.

The output format is according to json format, the output template is as follows:

the "hidden trouble of disease" is described as follows:

the whole physical examination report is combined to give the disease risk/hidden danger which is mainly the disease hidden danger in the treatment and the to-be-rechecked, and the disease risk/hidden danger is the disease name, and a plurality of disease names are separated by using.

The "physical examination anomaly" is described as follows:

if there are multiple abnormal items, the abnormal items need to be expressed separately, and two points of a third layer are arranged behind each abnormal item which is analyzed separately;

if multiple abnormal items appear, they can be interpreted jointly, then no separate description is needed. Note that: only clinically there is an association that requires joint interpretation, such as: the clear association between the leucorrhea cleanliness higher and the leucorrhea leucocyte++ can be read in a combined way; "mammary gland color Doppler ultrasound: left breast nodules may "sum" abdominal ultrasound: fatty liver' does not have association relationship, and can be read separately.

Report with "# physical examination: "in { } the content inside is in control, in the case of joint interpretation, multiple abnormal terms are connected by" | ", such as: "leucorrhea cleanliness is higher than that of |leucocyte++";

criteria for blood and urine routine and other examination items whether joint interpretation is required: the condition of diagnosis of the disease in the textbook or guideline is based on: if a certain examination item and abnormal results of blood routine/urine routine are simultaneously mentioned in the diagnosis confirming condition, the diagnosis can be combined; if no definitive conditions are present in this regard, the description is separate;

if a result abnormality is mentioned in the examination result abnormality/imaging result abnormality/ultrasound department result abnormality/pathology department result abnormality, but not written in the conclusion, the abnormal item is written in our interpretation, and the writing format of the abnormal item is as follows: "detailed abnormal item + clinical meaning of outcome".

The description of "need to visit" is as follows:

judging abnormal physical examination items which need to be immediately treated according to clinical experience and medical knowledge, wherein the abnormal physical examination items can not cause delay of illness state, and the standard of the part can be strict (whether symptoms exist or not, the patient needs to be immediately treated in a hospital, or serious consequences or delay of illness state can be caused);

The explanation about the "reason for visit" is as follows:

the language is required to be refined, popular and easy to understand, words are not required to be too many, and the description which can be understood by common patients is used, and comprises the following four aspects: (1) clinical significance of abnormal outcome, (2) what disease or symptoms the outcome may lead to, what harm may be if not taken to the hospital, etc. (3) for the purpose of taking hospital visits, xx auxiliary examination/determination of treatment regimen/s may be needed to exclude xx disease.

The explanation about "department of medical science" is as follows:

the department that needs to visit the doctor to express the abnormal item is called the second-level department as much as possible, and a plurality of departments are connected by the joint.

The description of "to be reviewed" is as follows:

the medical examination report is not written with detailed information such as the size, quantity, nature and boundary of the three-dimensional model, and the guidance/suggestion in the medical examination report is not serious, and is defaulted to be mild/common/not serious.

The explanation about the "review reason" is as follows:

the items to be reviewed in combination with the physical examination report are written, and the following five points are included: (1) the items to be reviewed have a small point (the items to be reviewed are written as much as possible, for example, abnormal items are "abdominal ultrasound: liver cyst", the items to be reviewed you can write "liver ultrasound"), (2) the meaning of these items to be reviewed, (3) the review time, (4) what risk may occur if no review is done, and (5) what symptoms need to be taken to a hospital in time. Note that: if there are multiple items to review, a single separate write is also required. Note that: the review time is expressed in the physical examination report, and the review time is required to be corresponding to the report, if not, the review time is expressed according to medical knowledge.

The explanation about "review department" is as follows:

the departments needing to visit the doctor when the abnormal item is reexamined are expressed, and the departments are connected with each other by using the second-level departments as much as possible.

The description of "attention required" is as follows:

according to medical knowledge judgment, the patient can relieve or restore to normal by changing diet, strengthening exercise, improving life habit and the like (the examination index item is slightly deviated from the normal value and is concerned, the life is not influenced temporarily, and the problem/index result can be improved by optimizing life habit, diet and the like).

The description of "clinical meaning" is as follows:

the abnormal result represents clinical significance.

The explanation about "living instruction" is as follows:

advice on diet, exercise and lifestyle and attention in daily life care need to be separately and simply explained.

Physical examination report outliers and advice:

in the first task prompting instruction, the following is a report of abnormal values and advice for physical examination, please report the interpretation result for me. "is the description information of the reading task of the physical examination report, namely the description information of the proposal task.

The output format is the interpretation result output format requirement according to json format.

"input templates are as follows: … … the "life guidance" is "xxx" } } "part of the content is the physical examination report interpretation template.

The "related to" disease risk "specification is described below … … in a separate and simplified manner. And the partial content is the filling description of the physical examination report interpretation template.

In "report outliers and advice for physical examination: "after that, the first task presentation instruction composed of the first task description information, the physical examination report interpretation template, and the filling description of the physical examination report interpretation template is obtained by adding the physical examination information acquired from the physical examination report according to step S101.

S103, inputting the first task prompt instruction into a pre-trained report analysis model to obtain an interpretation result of the physical examination report.

The generative language model (Generative Language Model) is a machine learning based natural language processing model for generating text conforming to grammatical and semantic rules. The generative language model may learn a probability distribution of text data and then use this distribution to generate new text. The core idea of the generative language model is that by learning a large amount of text data, the model is able to capture the associations between words and context information. Based on these learned associations, the generative language model may predict the likelihood of the next word or piece of text, thereby generating coherent, meaningful text. Common generative language models include statistical-based models, such as n-gram models and Hidden Markov Models (HMMs), and neural network-based models, such as Recurrent Neural Networks (RNNs) and variants, such as long-short term memory networks (LSTMs) and Transformer models.

Further, the pre-training of the pre-trained generated language model based on the physical examination report refers to text pre-training and/or relation deduction pre-training of the pre-trained generated language model based on the physical examination report.

The pre-training is a common large model training task, wherein the text prediction training refers to masking certain contents in a text, so that the generated language model predicts the masked contents; the relation derivation pre-training refers to that the generated language model predicts the relation between two entities or two sentences, or directly predicts the relation between two sentences in the same text, wherein the relation between sentences can be implication (one sentence is contained between the other sentences), contradiction (no front-back relation exists between the two sentences, such as the two sentences are from different articles), neutral (the two sentences are from the same article but have no containing relation).

The embodiment of the application firstly performs the pre-training on the generated language model by utilizing a large number of physical examination reports, and then performs the fine-tuning training of the physical examination report interpretation task on the generated language model after the pre-training on the generated language model is completed. Specifically, through a physical examination report sample, a physical examination report interpretation prompt and a physical examination report interpretation label, the supervised physical examination report interpretation training is performed on the generated language model, so that the generated language model derives the physical examination report interpretation capability based on the powerful self-language processing capability of the generated language model, namely the report interpretation model is obtained through training.

In some embodiments, the generative language model described above may employ a large language model.

It will be appreciated that the report parsing model is built based on a pre-trained generative language model, and that the inputs to the generative language model include a prompt for prompting the model to perform a particular task. In this embodiment of the present application, the first task prompting instruction may be used as a prompt of the report parsing model.

The report analysis model is used for reading the physical examination information contained in the first task prompt instruction according to the first task prompt instruction, and further obtaining the reading result.

Specifically, as can be seen from the foregoing description of the first task prompting instruction, the first task instruction includes the physical examination report interpretation template and the filling explanation of the physical examination report interpretation template, so that in the practical application process, the report analysis model interprets the physical examination report to obtain the interpretation result, and the interpretation result is actually filled in the filling explanation of the physical examination report according to the filling explanation of the physical examination report.

And inputting the generated first task prompting instruction into the pre-trained report analysis model to obtain an interpretation result of the physical examination report output by the model.

As can be seen from the above description, according to the data processing method provided in the embodiment of the present application, the pre-training based on the physical examination report information and the fine-tuning training of the physical examination report interpretation task are performed on the pre-trained generated language model, so as to obtain a report analysis model capable of performing physical examination report interpretation based on the physical examination information. On the basis, physical examination information is obtained from the physical examination report, then a task prompt instruction which at least contains the physical examination information and the physical examination report interpretation task description information is generated, and the task prompt instruction is input into the pre-trained report analysis model, so that analysis results of the physical examination report can be obtained. The proposal realizes the automatic interpretation of the physical examination report, thereby improving the interpretation efficiency of the physical examination report.

In another optional embodiment of the present application, the report parsing model is further capable of answering a related problem of a physical examination object, so as to confuse the physical examination object with respect to the physical examination report, and specifically, the data processing method further includes the following steps S104 to S106:

S104, acquiring the consultation problem aiming at the physical examination report.

The consultation problem for the physical examination report refers to a problem which is presented by a physical examination object (patient) according to the content and the personal condition of the physical examination report.

For example, the consultation question may be "based on my physical examination results, i are at risk of suffering from certain diseases? What measures can reduce the risk? "can also be" do medications or other therapeutic measures need to be taken based on my physical examination results? "and the like.

S105, generating a second task prompt instruction at least comprising the consultation problem and second task description information, wherein the second task description information comprises description information of a problem solving task, and the problem solving task is used for solving the consultation problem.

And the second task prompt instruction is used for indicating an object receiving the second task prompt instruction to execute the problem solving task.

The task of solving the problem is specifically a task of solving the consultation problem. And executing the problem solving task, namely solving the consultation problem.

In the generated second task prompting instruction, the consultation problem is carried and the description information of the problem solving task is carried, so that the object receiving the second task prompting instruction can solve the consultation problem in the instruction based on executing the problem solving task in the instruction.

Specifically, the generating a second task prompting instruction at least including the consultation problem and the second task description information includes the following steps B1 to B3:

b1, screening out physical examination report information related to the consultation problems from the physical examination reports according to the consultation problems, and screening out physical examination report conclusions from the physical examination reports;

specifically, the purpose of the step B1 is to determine, from the physical examination report, physical examination report information related to the consultation questions, so as to answer the consultation questions in combination with the physical examination report information. And extracting the content of the conclusion part of the physical examination report from the physical examination report so as to assist in solving the consultation problem.

For example, keyword extraction and feature extraction can be performed on the consultation questions and the physical examination reports respectively, and then physical examination reports related to the consultation questions are screened from the physical examination reports by means of keyword matching, feature matching and the like.

In some implementations, B1 above includes:

inputting the consultation questions and the physical examination report into a pre-trained information screening model to obtain physical examination report information related to the consultation questions; the information screening model is obtained through physical examination report information retrieval training.

For example, when the consultation problem is "doctor, i have done a physical examination recently, found that there is a micro-nodule in the lower left lung and a calcification lesion in the upper left lung, how do this return? Do i need to worry about? In the case of "inputting the consultation questions and the physical examination reports of physical examination mentioned in the consultation questions into the information screening model, the physical examination report fragments related to the lower left lung leaf micro-nodules mentioned in the consultation questions and the upper left lung leaf calcification foci can be obtained, for example: the physical examination report fragment may be [ chest CT pan: * X ","2, [ CT: lower left lung leaf micro nodule, please follow-up; on the left lung She Gaihua stove \n recommended respiratory medicine follow-up for 6-12 months. "].

The information screening model can be obtained by searching and training physical examination report information based on a neural network model. The information retrieval training is to enable the model to retrieve physical examination report information related to the input consultation questions from the input physical examination reports.

B2, determining second task description information according to the physical examination report information related to the consultation problem and the physical examination report conclusion, wherein the second task description information comprises description information of a problem solving task, and the problem solving task is used for solving the consultation problem based on the physical examination report information related to the consultation problem and the physical examination report conclusion;

Similar to the first task description information, the second task description information may be written by the relevant staff based on actual requirements, for example, in an alternative embodiment of the present application, the second task description information may be: now in a medical scenario, you play a specialized doctor, please give specific replies according to patient questions, combined with physical examination report conclusions, and physical examination report information related to the consultation questions.

And B3, combining at least the consultation problem, the second task description information, the physical examination report information related to the consultation problem and the physical examination report conclusion to obtain a second task instruction.

In this embodiment of the present application, the second task instruction is specifically configured to input the report parsing model, so that the report parsing model answers the consultation question based on the second task description information, the physical examination report information related to the consultation question, and the physical examination report conclusion in the second task instruction.

In another alternative embodiment of the present application, the second task instruction further includes a constraint specification for solving the consultation problem; wherein the constraint specification includes at least one of a reply request specification, an spam application specification, a fixed reply specification, and a forbidden reply specification.

For example, the reply claim specification may be: if the patient problem is specific to some abnormal items in the physical examination report, the patient problem needs to be read and then specifically replied; the spam application description may be: if the patient problem relates to the treatment scheme, the patient does not need to explain each abnormal item in reply, and directly explain whether treatment is needed and give treatment advice and reasons, but the patient must have a "under doctor's guidance" procedure; the fixed reply specification may be: if the patient problem is to read the whole physical examination report, for example, if you ask you to help me see the report/you ask me to see me severe case, then the patient is replied with "we have provided detailed report reading for you, please see report reading interface"; for example, the disabling reply specification may be: please not initiate an active inquiry like "i want to know about XXX", "you are not XX recently? ";

the following is an example of a second task hint instruction obtained by the above-described processing:

in the current medical scenario, you play a specialized doctor, please give specific replies according to the patient's problems, combining with the medical examination report conclusions and the relevant medical examination report fragments.

The following constraints on the reply are:

1. setting answers of ' I are your doctor ', ' hello ' I are xdoctor ' and the like, and giving more relative names by combining the patient name;

2. comprehensive reply is carried out on all contents in the user questions and the related physical examination report sheet fragments in combination with physical examination report conclusion:

1. if the patient problem is specific to some abnormal items in the physical examination report, the patient problem needs to be read and then specifically replied;

2. if the patient problem relates to the treatment scheme, the patient does not need to explain each abnormal item in reply, and directly explain whether treatment is needed and give treatment advice and reasons, but the patient must have a "under doctor's guidance" procedure;

3. if the patient problem is to read the whole physical examination report, for example, if you ask you to help me see the report/you ask me to see me severe case, then the patient is replied with "we have provided detailed report reading for you, please see report reading interface";

3. the patients need to be clearly replied, professionally and executable, and the patients need to be clearly replied as much as possible, besides asking for complicated medication schemes/avoiding operations (such as how to take the medicine in the late stage of gastric cancer/how to do not need to operate the medicine in the present situation), the general disease patients can ask for the medication to give the patients some kinds of medicines and the list of medicines;

4. The information of the reply content is required to be ensured not to conflict with the relevant physical examination report sheet fragments and the physical examination report sheet conclusion, if the relevant physical examination report sheet fragments are not available, the context is combined for normal reply;

6. please not initiate an active inquiry like "i want to know about XXX", "you are not XX recently? ";

the following is a physical examination report conclusion: { physical examination report conclusion }

The following are relevant physical examination report fragments:

[ "chest CT plain scan: * X ","2, [ CT: lower left lung leaf micro nodule, please follow-up; on the left lung She Gaihua stove \n recommended respiratory medicine follow-up for 6-12 months. "]

The following are user questions:

doctor, i have recently done a physical examination and found that there are micro-nodules in the lower left lung and calcification lesions in the upper left lung, how do this return? Do i need to worry about?

And replacing the physical examination report conclusion obtained after the physical examination report is structured into the instruction, so that a complete second task prompt instruction can be obtained.

S106, inputting the second task prompt instruction into the report analysis model to obtain a solution result of the consultation problem.

In the embodiment of the application, the report analysis model not only has the capability of reading the physical examination report, but also has the capability of answering the consultation questions based on the second task instruction. Specifically, the ability of the report parsing model to answer the consultation questions is obtained based on a fine tuning training of the report parsing, and specifically, the fine tuning training refers to a task training of the report parsing model to execute a task of the report parsing in the fine tuning process mentioned in step S103.

Based on the above processing, the technical solution provided by the embodiment of the present application not only can read the physical examination report of the user, but also can solve the problem about the physical examination report provided by the user. In the answering process, the physical examination report information related to the problem can be screened out from the physical examination report, and the problem of the user can be accurately and correctly solved based on the physical examination report information and the physical examination report conclusion. In the multiple rounds of questions and answers of the user, the questions and answers of the user in each round are answered according to the scheme, so that the multiple rounds of questions and answers of the user experience can be improved.

Exemplary apparatus

Correspondingly, the embodiment of the application also provides a data processing device, as shown in fig. 2, which comprises:

an information processing unit 201 for acquiring physical examination information from the physical examination report;

an instruction generating unit 202, configured to generate a first task prompt instruction that includes at least the physical examination information and first task description information, where the first task description information includes description information of a physical examination report interpretation task, and the physical examination report interpretation task is configured to interpret the physical examination report based on the physical examination information;

a report interpretation unit 203, configured to input the first task prompting instruction into a pre-trained report analysis model, to obtain an interpretation result of the physical examination report;

In an optional embodiment of the present application, the acquiring physical examination information from the physical examination report includes:

or,

In an optional embodiment of the present application, generating a first task suggestion instruction at least including the physical examination information and the first task description information includes:

In an optional embodiment of the present application, the physical examination report interpretation template includes a description of hidden danger of disease, a description of abnormal items of physical examination, and a description of items to be focused on;

In an alternative embodiment of the present application, the report interpretation unit is further configured to:

acquiring consultation questions aiming at the physical examination report;

In an optional embodiment of the present application, the generating a second task suggestion instruction that includes at least the consultation problem and second task description information includes:

In an optional embodiment of the present application, the screening, according to the consultation problem, physical examination report information related to the consultation problem from the physical examination report includes:

In an alternative embodiment of the present application, the second task instruction further includes a constraint specification for solving the consultation problem;

The data processing device provided in this embodiment belongs to the same application conception as the data processing method provided in the foregoing embodiments of the present application, and may execute the data processing method provided in any of the foregoing embodiments of the present application, and has a functional module and beneficial effects corresponding to executing the data processing method. Technical details not described in detail in this embodiment may be referred to the specific processing content of the data processing method provided in the foregoing embodiment of the present application, and will not be described herein again.

The functions implemented by the above information processing unit 201, instruction generating unit 202, and report interpreting unit 203 may be implemented by the same or different processors, respectively, and the embodiments of the present application are not limited.

It will be appreciated that the elements of the above apparatus may be implemented in the form of processor-invoked software. For example, the device includes a processor, where the processor is connected to a memory, and the memory stores instructions, and the processor invokes the instructions stored in the memory to implement any of the methods above or to implement functions of each unit of the device, where the processor may be a general-purpose processor, such as a CPU or a microprocessor, and the memory may be a memory within the device or a memory outside the device. Alternatively, the units in the apparatus may be implemented in the form of hardware circuits, and the functions of some or all of the units may be implemented by designing hardware circuits, which may be understood as one or more processors; for example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are implemented by designing the logic relationships of the elements in the circuit; for another example, in another implementation, the hardware circuit may be implemented by a PLD, for example, an FPGA may include a large number of logic gates, and the connection relationship between the logic gates is configured by a configuration file, so as to implement the functions of some or all of the above units. All units of the above device may be realized in the form of processor calling software, or in the form of hardware circuits, or in part in the form of processor calling software, and in the rest in the form of hardware circuits.

In the embodiment of the application, the processor is a circuit with signal processing capability, and in one implementation, the processor may be a circuit with instruction reading and running capability, such as a CPU, a microprocessor, a GPU, or a DSP, etc.; in another implementation, the processor may implement a function through a logical relationship of hardware circuitry that is fixed or reconfigurable, e.g., a hardware circuit implemented by the processor as an ASIC or PLD, such as an FPGA, or the like. In the reconfigurable hardware circuit, the processor loads the configuration document, and the process of implementing the configuration of the hardware circuit may be understood as a process of loading instructions by the processor to implement the functions of some or all of the above units. Furthermore, a hardware circuit designed for artificial intelligence may be provided, which may be understood as an ASIC, such as NPU, TPU, DPU, etc.

It will be seen that each of the units in the above apparatus may be one or more processors (or processing circuits) configured to implement the above method, for example: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

Furthermore, the units in the above apparatus may be integrated together in whole or in part, or may be implemented independently. In one implementation, these units are integrated together and implemented in the form of an SOC. The SOC may include at least one processor for implementing any of the methods above or for implementing the functions of the units of the apparatus, where the at least one processor may be of different types, including, for example, a CPU and an FPGA, a CPU and an artificial intelligence processor, a CPU and a GPU, and the like.

Exemplary electronic device

Another embodiment of the present application further proposes an electronic device, referring to fig. 3, including:

a memory 200 and a processor 210;

wherein the memory 200 is connected to the processor 210, and is used for storing a program;

the processor 210 is configured to implement the data processing method disclosed in any one of the foregoing embodiments by executing the program stored in the memory 200.

Specifically, the electronic device may further include: a bus, a communication interface 220, an input device 230, and an output device 240.

The processor 210, the memory 200, the communication interface 220, the input device 230, and the output device 240 are interconnected by a bus. Wherein:

a bus may comprise a path that communicates information between components of a computer system.

Processor 210 may be a general-purpose processor such as a general-purpose Central Processing Unit (CPU), microprocessor, etc., or may be an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs in accordance with aspects of the present invention. But may also be a Digital Signal Processor (DSP), application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components.

Processor 210 may include a main processor, and may also include a baseband chip, modem, and the like.

The memory 200 stores programs for implementing the technical scheme of the present invention, and may also store an operating system and other key services. In particular, the program may include program code including computer-operating instructions. More specifically, the memory 200 may include read-only memory (ROM), other types of static storage devices that may store static information and instructions, random access memory (random access memory, RAM), other types of dynamic storage devices that may store information and instructions, disk storage, flash, and the like.

The input device 230 may include means for receiving data and information entered by a user, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor, among others.

Output device 240 may include means, such as a display screen, printer, speakers, etc., that allow information to be output to a user.

The communication interface 220 may include devices using any transceiver or the like for communicating with other devices or communication networks, such as ethernet, radio Access Network (RAN), wireless Local Area Network (WLAN), etc.

Processor 210 executes programs stored in memory 200 and invokes other devices that may be used to implement various steps of any of the data processing methods provided in the above-described embodiments of the present application.

The embodiment of the application also provides a chip, which comprises a processor and a data interface, wherein the processor reads and runs a program stored in a memory through the data interface so as to execute the data processing method described in any embodiment, and the specific processing process and the beneficial effects thereof can be described by referring to the embodiment of the data processing method.

Exemplary computer program product and storage Medium

In addition to the methods and apparatus described above, embodiments of the present application may also be a computer program product comprising computer program instructions which, when executed by a processor, cause the processor to perform the steps in a data processing method according to various embodiments of the present application described in any of the embodiments of the present application.

The computer program product may write program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server.

Furthermore, embodiments of the present application may also be a storage medium having stored thereon a computer program that is executed by a processor to perform the steps in the data processing method according to the various embodiments of the present application described in any of the above embodiments of the present specification, and specifically may implement the following steps:

s101, acquiring physical examination information from a physical examination report;

s102, generating a first task prompt instruction at least comprising the physical examination information and first task description information, wherein the first task description information comprises description information of a physical examination report interpretation task, and the physical examination report interpretation task is used for interpreting the physical examination report based on the physical examination information;

s103, inputting the first task prompt instruction into a pre-trained report analysis model to obtain an interpretation result of the physical examination report;

For the foregoing method embodiments, for simplicity of explanation, the methodologies are shown as a series of acts, but one of ordinary skill in the art will appreciate that the present application is not limited by the order of acts described, as some acts may, in accordance with the present application, occur in other orders or concurrently. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required in the present application.

It should be noted that, in the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described as different from other embodiments, and identical and similar parts between the embodiments are all enough to be referred to each other. For the apparatus class embodiments, the description is relatively simple as it is substantially similar to the method embodiments, and reference is made to the description of the method embodiments for relevant points.

The steps in the method of each embodiment of the application can be sequentially adjusted, combined and deleted according to actual needs, and the technical features described in each embodiment can be replaced or combined.

The modules and sub-modules in the device and the terminal of the embodiments of the present application may be combined, divided, and deleted according to actual needs.

In the embodiments provided in the present application, it should be understood that the disclosed terminal, apparatus and method may be implemented in other manners. For example, the above-described terminal embodiments are merely illustrative, and for example, the division of modules or sub-modules is merely a logical function division, and there may be other manners of division in actual implementation, for example, multiple sub-modules or modules may be combined or integrated into another module, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or modules, which may be in electrical, mechanical, or other forms.

The modules or sub-modules illustrated as separate components may or may not be physically separate, and components that are modules or sub-modules may or may not be physical modules or sub-modules, i.e., may be located in one place, or may be distributed over multiple network modules or sub-modules. Some or all of the modules or sub-modules may be selected according to actual needs to achieve the purpose of the embodiment.

In addition, each functional module or sub-module in each embodiment of the present application may be integrated in one processing module, or each module or sub-module may exist alone physically, or two or more modules or sub-modules may be integrated in one module. The integrated modules or sub-modules may be implemented in hardware or in software functional modules or sub-modules.

Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software unit executed by a processor, or in a combination of the two. The software elements may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

Finally, it is further noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of data processing, comprising:

acquiring physical examination information from a physical examination report;

2. The method of claim 1, wherein the obtaining physical examination information from the physical examination report comprises:

or,

3. The method of claim 1, wherein generating a first task hint instruction comprising at least the physical examination information and first task description information comprises:

4. The method of claim 3, wherein the physical examination report interpretation template comprises a description of a disease risk, a description of physical examination abnormal items, and a description of items of interest;

5. The method according to any one of claims 1 to 4, further comprising:

acquiring consultation questions aiming at the physical examination report;

6. The method of claim 5 wherein generating a second task suggestion instruction that includes at least the consultation problem and second task description information includes:

7. The method of claim 6, wherein the screening the medical examination report information related to the counseling questions from the medical examination report according to the counseling questions comprises:

8. The method of claim 5, wherein the second task instruction further comprises a constraint specification that solves the consultation problem;

9. A data processing apparatus, comprising:

10. An electronic device comprising a memory and a processor;

the memory is connected with the processor and used for storing programs;

The processor is configured to implement the data processing method according to any one of claims 1 to 8 by running a program in the memory.

11. A storage medium having stored thereon a computer program which, when executed by a processor, implements the data processing method according to any of claims 1 to 8.