CN113724858A

CN113724858A - Artificial intelligence-based disease examination item recommendation device, method and apparatus

Info

Publication number: CN113724858A
Application number: CN202111010221.3A
Authority: CN
Inventors: 杨怀忠
Original assignee: Ping An International Smart City Technology Co Ltd
Current assignee: Shenzhen Ping An Smart Healthcare Technology Co ltd
Priority date: 2021-08-31
Filing date: 2021-08-31
Publication date: 2021-11-30

Abstract

This application is applicable to the field of artificial intelligence technology, and provides a device, method and device for recommending disease inspection items based on artificial intelligence, wherein the device for recommending disease inspection items based on artificial intelligence The text information that matches the entity information in the disease knowledge map is extracted from the diagnosis data to obtain the consultation text information, and in the graph convolution layer of the inspection item recommendation model trained based on the disease knowledge map, it is determined that the consultation text Inquiry feature vector corresponding to the information. Afterwards, a target disease feature vector matching the consultation feature vector is determined in the first classifier of the check item recommendation model, and corresponding target disease check item information is determined according to the target disease feature vector. The embodiments of the present application can accurately recommend corresponding examination items for target patients in a digital medical scenario, thereby effectively assisting subsequent disease diagnosis and saving costs.

Description

Artificial intelligence-based disease examination item recommendation device, method and apparatus

Technical Field

The application relates to the technical field of artificial intelligence, in particular to disease examination item recommendation equipment, method and device based on artificial intelligence.

Background

Currently, in a medical scenario, when a patient is in a hospital visit, the patient is usually examined according to the corresponding examination items according to the guidance of a doctor, so that the doctor makes a corresponding diagnosis according to the examination results.

However, due to the many examination items, when some patients have rare symptoms, it is difficult for doctors to determine exactly which examination items are needed in advance. If there are too many redundant unnecessary examination items in the examination items that the doctor instructs the patient to perform, time delay and money waste may be caused. If the examination items instructed by the doctor to be performed by the patient lack some examination items closely related to the disease of the patient, the subsequent diagnosis may be inaccurate due to insufficient examination data. Therefore, how to accurately recommend the corresponding examination items to the patient is a problem that needs to be solved in the medical field.

Disclosure of Invention

In view of this, embodiments of the present application provide an artificial intelligence-based disease examination item recommendation apparatus, method, and device, so as to solve the problem in the prior art of how to accurately recommend a corresponding examination item for a patient.

A first aspect of an embodiment of the present application provides an artificial intelligence-based disease examination item recommendation apparatus, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where when the processor executes the computer program, the following steps are implemented:

acquiring inquiry data of a target patient;

extracting text information matched with entity information in a preset disease knowledge graph from the inquiry data to obtain inquiry text information; wherein the entity information in the disease knowledge graph comprises disease name information, disease symptom information and disease examination item information;

determining an inquiry characteristic vector corresponding to the inquiry text information in a graph volume layer of an inspection item recommendation model; the examination item recommendation model is obtained by training according to the disease knowledge graph;

determining a target disease feature vector matched with the inquiry feature vector in a first classifier of the inspection item recommendation model, and determining target disease inspection item information according to the target disease feature vector; wherein the target disease feature vector is a feature vector corresponding to a target disease; the target disease examination item information is the disease examination item information corresponding to the target disease.

Optionally, the processor, when executing the computer program, is further configured to:

acquiring preset medical data;

inputting the preset medical data into an entity information identification model for processing to obtain each entity information and the association probability among the entity information; the entity information identification model is a bidirectional cyclic neural network obtained by training based on medical sample data carrying entity information labels;

and constructing the disease knowledge graph according to the entity information and the association probability among the entity information.

Optionally, the entity information of the disease knowledge-graph further comprises medical guideline information, and correspondingly, the processor, when executing the computer program, is further configured to:

and determining target medical guideline information corresponding to the target disease feature vector.

Optionally, the entity information of the disease knowledge-graph further includes examination result information, and the processor, when executing the computer program, is further configured to:

acquiring target examination result information obtained by examining the target patient according to the target disease examination item information;

and inputting the inquiry text information and the examination result information into a disease decision module of the examination item recommendation model for processing to obtain diagnosis result information.

and updating the disease knowledge map and/or the inspection item recommendation model according to the inquiry text information, the target inspection result information and the diagnosis result information.

Optionally, the entity information of the disease knowledge map further comprises information of a treatment protocol, and the processor, when executing the computer program, is further configured to:

acquiring target disease description information of the target patient;

and inputting the target disease description information and the diagnosis result information into a treatment scheme recommendation module of the examination item recommendation model for processing to obtain a target treatment scheme.

and determining corresponding referral advice information according to the diagnosis result information.

A second aspect of the embodiments of the present application provides a disease inspection item recommendation method based on artificial intelligence, including:

acquiring inquiry data of a target patient;

determining a target disease feature vector matched with the inquiry feature vector in a first classifier of the inspection item recommendation model, and determining target disease inspection item information according to the target disease feature vector; wherein the target disease feature vector is a feature vector which integrates the entity information corresponding to the target disease in the disease knowledge base; the target disease examination item information is the disease examination item information matched with the target disease feature vector.

A third aspect of an embodiment of the present application provides an artificial intelligence-based disease examination item recommendation apparatus, including:

the inquiry data acquisition unit is used for acquiring inquiry data of a target patient;

the inquiry text information determining unit is used for extracting text information matched with entity information in a preset disease knowledge map from the inquiry data to obtain inquiry text information; wherein the entity information in the disease knowledge graph comprises disease name information, disease symptom information and disease examination item information;

the inquiry characteristic vector determining unit is used for determining inquiry characteristic vectors corresponding to the inquiry text information in the graph volume layer of the inspection item recommendation model; the examination item recommendation model is obtained by training according to the disease knowledge graph;

the target disease examination item determining unit is used for determining a target disease feature vector matched with the inquiry feature vector in a first classifier of the examination item recommendation model and determining target disease examination item information according to the target disease feature vector; wherein the target disease feature vector is a feature vector which integrates the entity information corresponding to the target disease in the disease knowledge base; the target disease examination item information is the disease examination item information matched with the target disease feature vector.

A fourth aspect of embodiments of the present application provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the steps of the artificial intelligence based disease inspection item recommendation method provided by the second aspect.

A fifth aspect of the embodiments of the present application provides a computer program product, which, when running on an artificial intelligence based disease inspection item recommendation apparatus, causes the artificial intelligence based disease inspection item recommendation apparatus to execute the steps of the artificial intelligence based disease inspection item recommendation method according to the second aspect.

The disease examination item recommendation device based on artificial intelligence has the following beneficial effects: in the embodiment of the application, according to the inquiry data of a target patient, text information matched with entity information in a disease knowledge graph is extracted from the inquiry data to obtain inquiry text information, and inquiry characteristic vectors corresponding to the inquiry text information are determined in a graph volume layer of an examination item recommendation model obtained based on disease knowledge graph training. And then, determining a target disease feature vector matched with the inquiry feature vector in a first classifier of the examination item recommendation model, and determining corresponding target disease examination item information according to the target disease feature vector. According to the preset disease knowledge graph and the examination item recommendation model obtained based on the disease knowledge graph training, the corresponding target disease examination item information can be accurately determined, so that the corresponding examination item can be accurately recommended for the target patient, the subsequent disease diagnosis can be accurately and effectively assisted, the accuracy of the subsequent diagnosis is improved, the money waste caused by the fact that the patient conducts too many redundant and invalid examination items can be avoided, and the labor cost and the material cost are saved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings used in the embodiments or the description of the prior art will be briefly described below.

Fig. 1 is a block diagram of an artificial intelligence-based disease examination item recommendation device according to an embodiment of the present application.

FIG. 2 is a flowchart of an implementation of a disease examination item recommendation method based on artificial intelligence according to an embodiment of the present application;

fig. 3 is a block diagram of a disease examination item recommendation device based on artificial intelligence according to an embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

In order to explain the technical solution described in the present application, the following description will be given by way of specific examples.

It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It is also to be understood that the terminology used in the description of the present application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in the specification of the present application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

In addition, in the description of the present application, the terms "first", "second", and the like are used only for distinguishing the description, and are not intended to indicate or imply relative importance.

At present, in a medical scene, due to the variety of examination items available, when symptoms of some patients are rare, it is difficult for a doctor to accurately determine in advance which examination items need to be performed by the patient. If there are too many redundant unnecessary examination items in the examination items that the doctor instructs the patient to perform, time delay and money waste may be caused. If the examination items instructed by the doctor to be performed by the patient lack some examination items closely related to the disease of the patient, the subsequent diagnosis may be inaccurate due to insufficient examination data. Therefore, how to accurately recommend the corresponding examination items to the patient is a problem that needs to be solved in the medical field.

In order to solve the above technical problem, an embodiment of the present application provides an artificial intelligence-based disease examination item recommendation apparatus, method and device, including: according to the inquiry data of the target patient, extracting text information matched with the entity information in the disease knowledge graph from the inquiry data to obtain inquiry text information, and determining inquiry characteristic vectors corresponding to the inquiry text information in a graph volume layer of an examination item recommendation model obtained based on disease knowledge graph training. And then, determining a target disease feature vector matched with the inquiry feature vector in a first classifier of the examination item recommendation model, and determining corresponding target disease examination item information according to the target disease feature vector.

According to the preset disease knowledge graph and the examination item recommendation model obtained based on the disease knowledge graph training, the corresponding target disease examination item information can be accurately determined, so that the corresponding examination item can be accurately recommended for the target patient, the subsequent disease diagnosis can be accurately and effectively assisted, the accuracy of the subsequent diagnosis is improved, and the money waste caused by the fact that the patient conducts too many redundant and invalid examination items can be avoided.

The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.

Disease examination project recommendation equipment based on artificial intelligence that this application embodiment relates to can be applied to in the wisdom medical scene to promote the construction in wisdom city.

Illustratively, the disease examination item recommendation device based on artificial intelligence can be applied to an intelligent diagnosis scene of rare diseases. Rare diseases refer to those diseases with a low prevalence, first rare patients and second more rare physicians with the ability to diagnose and treat rare diseases. The treatment time of many rare diseases is long, for example, the misdiagnosis rate of Kennedy disease reaches 71 percent, and the average diagnosis period is 5 years. The biggest difficulty in diagnosing rare diseases is that the awareness of doctors is too low, resulting in high misdiagnosis rate and long diagnosis time of rare diseases. In the disease knowledge map of the embodiments of the present application, entity information of rare diseases may be included. Illustratively, the rare diseases can include the progressive freezing disease (also called amyotrophic lateral sclerosis), and the disease knowledge map can contain name information of the progressive freezing disease, symptom information such as muscle twitch, stiffness or hypodynamia corresponding to the progressive freezing disease, and disease examination item information of the amyotrophic lateral sclerosis pathogeny gene identification corresponding to the progressive freezing disease. Through the disease examination item recommendation device based on artificial intelligence, the corresponding rare disease examination item information can be accurately recommended for the target patient with rare disease suspected symptoms according to the inquiry data of the target patient, so that the target patient with rare disease suspected symptoms can be accurately examined in time. In some embodiments, the diagnosis result of the target patient can be further obtained according to the examination data, so as to accurately confirm whether the target patient has a rare disease, and if the target patient has the rare disease, a corresponding treatment scheme can be accurately provided through the decision module, or a referral suggestion can be provided, so that the diagnosis efficiency and accuracy of the rare disease are improved, and the damage to the target patient due to delayed treatment is avoided.

The first embodiment is as follows:

fig. 1 is a schematic diagram of an artificial intelligence-based disease examination item recommendation apparatus according to an embodiment of the present application. The artificial intelligence-based disease examination item recommendation device of the embodiment of the application includes, but is not limited to, a server, a computer, a mobile terminal and the like in a digital medical system.

The artificial intelligence based disease examination item recommendation device may include, but is not limited to, a processor 10, a memory 11. Those skilled in the art will appreciate that FIG. 1 is merely an example of an artificial intelligence based disease examination item recommendation device 1 and does not constitute a limitation of the artificial intelligence based disease examination item recommendation device 1 and may include more or fewer components than shown, or combine certain components, or different components, e.g., the artificial intelligence based disease examination item recommendation device may also include input-output devices, network access devices, buses, etc.

The Processor 10 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The memory 11 may be an internal storage unit of the artificial intelligence based disease inspection item recommendation device 1, such as a hard disk or a memory of the artificial intelligence based disease inspection item recommendation device 1. The memory 11 may also be an external storage device of the artificial intelligence based disease inspection item recommendation device 1, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, provided on the artificial intelligence based disease inspection item recommendation device 1. Further, the memory 11 may also include both an internal storage unit and an external storage device of the artificial intelligence based disease inspection item recommendation device 1. The memory 11 is used for storing the computer program and other programs and data required by the artificial intelligence based disease examination item recommendation apparatus. The memory 11 may also be used to temporarily store data that has been output or is to be output.

Illustratively, the computer program 12 may be partitioned into one or more modules/units, which are stored in the memory 11 and executed by the processor 10 to accomplish the present application. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions for describing the execution process of the computer program 12 in the artificial intelligence based disease examination item recommendation device 1.

When the processor 10 executes the computer program 12, the following steps S101 to S104 are implemented:

step S101: acquiring inquiry data of a target patient;

in the embodiment of the present application, the target patient is a patient currently being referred to, and the inquiry data of the target patient may include basic information such as age and sex of the target patient, chief complaint information of the target patient, historical medical record information of the target patient, family history information, and preliminary description information of a doctor on a disease condition of the target patient.

In one embodiment, the voice information of the target patient and the voice information of the doctor can be acquired through a voice acquisition module, and text information corresponding to the voice information is obtained through a voice recognition module and is used as the inquiry data of the target patient.

In another embodiment, historical medical record information of the target patient can be called from a medical record database, basic description information of the current situation of the target patient, which is input by a doctor after the inquiry, is obtained, and the inquiry data of the target patient is generated by combining the historical medical record information and the basic description information.

Step S102: extracting text information matched with entity information in a preset disease knowledge graph from the inquiry data to obtain inquiry text information; wherein the entity information in the disease knowledge graph comprises disease name information, disease symptom information and disease examination item information.

In the embodiment of the application, the preset disease knowledge graph is a knowledge graph which is constructed in advance and comprises entity information such as disease name information, disease symptom information and disease examination item information of various diseases (including rare diseases and common diseases). In some embodiments, the preset disease knowledge map may further include medical guideline information, case information, treatment protocol information, drug information, population information, etc. corresponding to the disease. In some embodiments, the preset disease knowledge map can be constructed in advance by professionals according to business modeling, knowledge acquisition, knowledge fusion, knowledge storage and knowledge calculation modes.

After the inquiry data is acquired, each text message matched with the entity information in the disease knowledge map can be extracted from the inquiry data according to the inquiry data and a preset disease knowledge map, and the text messages are combined into inquiry text messages.

In one embodiment, the inquiry data may be input into a pre-trained keyword extraction model for keyword extraction processing, so as to obtain target text information matched with each entity information. And generating inquiry text information according to the target text information. The keyword extraction model is a neural network model obtained by training in advance according to an inquiry data sample carrying an entity information label.

Step S103: determining an inquiry characteristic vector corresponding to the inquiry text information in a graph volume layer of an inspection item recommendation model; and the examination item recommendation model is obtained by training according to the disease knowledge graph.

In the embodiment of the application, the inspection item recommendation model is a neural network model obtained by training in advance according to the disease knowledge graph. The inspection item recommendation model includes at least a graph convolution layer and a first classifier. Since the disease knowledge graph is specifically a graph data, each entity information of the disease knowledge graph is a node in the graph data, the association relationship between each entity information can be represented by a connection between two nodes in the graph data (which may be referred to as an edge of the graph data), and the graph convolution algorithm of the graph convolution layer is an algorithm capable of accurately learning the feature information of the graph data, the information of the disease knowledge graph can be accurately learned and used by the graph convolution layer.

Specifically, the above-mentioned inquiry text message is actually a set of information including a plurality of entity information, so the inquiry text message can be actually regarded as one piece of graph data, and the inquiry text message is input into a graph volume layer to perform a graph volume operation, so that the features of the inquiry text message can be extracted to generate corresponding feature vectors. The feature vector obtained by the inquiry text information through the graph convolution operation is called an inquiry feature vector.

In one embodiment, the graph convolutional layer comprises a first sub-graph convolutional layer and a second sub-graph convolutional layer. After the inquiry text information is obtained, the inquiry text information can be converted into graph data, the node information of the graph data is each entity information contained in the inquiry text information, and the edge of the graph data is the incidence relation between each entity information; and determining a node characteristic matrix X and an adjacent matrix A corresponding to the inquiry text message according to the graph data corresponding to the inquiry text message. Inputting the node characteristic matrix X and the adjacent matrix A into a first subgraph convolution layer for preliminary graph convolution processing to obtain a first matrix H₁. The processing of the first graph convolutional layer can be represented by the following formula:

wherein the activation function relu (x) max (0, x);

the normalized Laplace matrix, which is the adjacency matrix A, can be according to the equation

Calculating and converting to obtain (D is a degree matrix of the adjacent matrix A); w₀The weight parameters of the first sub-graph convolution layer determined through learning training.

Then, the first matrix H is applied₁And inputting a second sub-graph convolution layer to perform graph convolution processing again to obtain a node fusion characteristic matrix Z of the graph data. The processing of the second sub-graph convolutional layer can be represented by the following formula:

wherein, W₁Second subgraph determined for learning trainingWeight parameters of convolutional layers.

And after obtaining the node fusion characteristic matrix corresponding to the inquiry text information, converting the node fusion characteristic matrix into a One-dimensional vector form through One-Hot Encoding (One-Hot Encoding), and obtaining an inquiry characteristic vector.

In one embodiment, the atlas layer is specifically an attention-integrated atlas layer, which can perform attention weighting based on the association between each entity information contained in the inquiry text information, so as to obtain a more accurate inquiry feature vector.

Step S104: determining a target disease feature vector matched with the inquiry feature vector in a first classifier of the inspection item recommendation model, and determining target disease inspection item information according to the target disease feature vector; wherein the target disease feature vector is a feature vector corresponding to a target disease; the target disease examination item information is the disease examination item information corresponding to the target disease.

In this embodiment of the application, the first classifier of the inspection item recommendation model may be a softmax classifier (a logistic regression classifier) obtained by training, a Support Vector Machine (SVM) classifier, or the like. The first classifier prestores a preset number of disease feature vectors generated in advance, and each disease feature vector represents feature information of one disease respectively. Specifically, the disease knowledge map comprises a preset number of diseases; when the training of the inspection item recommendation model is finished, for each disease in the disease knowledge graph, a corresponding node feature matrix and an adjacent matrix can be constructed according to entity information such as disease name information, disease symptom information and disease inspection item information corresponding to the disease and the incidence relation among the entity information, and a node fusion feature matrix corresponding to the disease is obtained through graph convolution layer processing; and carrying out one-hot encoding treatment on the node fusion characteristic matrix corresponding to the disease to obtain a disease characteristic vector corresponding to the disease.

Similarity calculation (for example, cosine similarity calculation) is performed on the current inquiry feature vector generated in step S103 and each pre-stored disease feature vector in the first classifier, so as to determine the similarity between the inquiry feature vector and each pre-stored disease feature vector. And then, determining the disease feature vector with the highest similarity with the inquiry feature vector as a target disease feature vector, wherein the target disease feature vector is the feature vector fusing entity information corresponding to the target disease in the disease knowledge map.

After the target disease feature vector is determined, the entity information of the target disease corresponding to the target disease feature vector can be determined, and the disease examination item information of the target disease is obtained from the entity information as the target disease examination item information.

After the target disease examination item information is obtained, the target disease examination item information can be output through screen display, voice broadcast, character printing and the like so as to recommend the current target patient to carry out examination of the target examination item.

In the embodiment of the application, according to the inquiry data of a target patient, text information matched with entity information in a disease knowledge graph is extracted from the inquiry data to obtain inquiry text information, and inquiry characteristic vectors corresponding to the inquiry text information are determined in a graph volume layer of an examination item recommendation model obtained based on disease knowledge graph training. And then, determining a target disease feature vector matched with the inquiry feature vector in a first classifier of the examination item recommendation model, and determining corresponding target disease examination item information according to the target disease feature vector. According to the preset disease knowledge graph and the examination item recommendation model obtained based on the disease knowledge graph training, the corresponding target disease examination item information can be accurately determined, so that the corresponding examination item can be accurately recommended for the target patient, the subsequent disease diagnosis can be accurately and effectively assisted, the accuracy of the subsequent diagnosis is improved, and the money waste caused by the fact that the patient conducts too many redundant and invalid examination items can be avoided.

Optionally, when the processor executes the computer program, specifically before the step S101, the processor executes the following steps:

acquiring preset medical data;

The preset medical data of the embodiment of the application includes, but is not limited to, literature data of pharmacology, genetics, pathology and the like, and may also include a large amount of patient medical record data stored in a hospital database. In one embodiment, the preset medical data may be directly obtained from a preset database. In another embodiment, the preset medical data can be obtained by data capturing from a designated medical information website through a crawler tool.

The entity information identification model in the embodiment of the application is a Bidirectional Recurrent Neural Network (BRNN) obtained by training medical sample data carrying an entity information tag, the entity information tag can be specifically obtained from a medical word stock built by professionals in advance, and the medical sample data entity information tag can be labeled by the professionals and also can be labeled by a preset tag labeling tool.

The acquired preset medical data is data mixed with a plurality of entity information, and the preset medical data is input into the entity information identification model for processing, so that each piece of entity information which is divided independently can be obtained. And, the entity information recognition model can also calculate the association probability between the divided individual entity information. For example, if two entity information appear on the same webpage or the same document more frequently, the association probability between the two entity information is higher.

In one embodiment, the BRNN may be a Bidirectional Long-Short Term Memory (BiLSTM) network. Through the BilSTM pair, the associated information of the context of the preset medical data in two directions can be analyzed simultaneously, so that the accuracy of entity information extraction is improved. Meanwhile, an attention mechanism is added in the BilSTM network, so that different weights are given to each data segment in the preset medical data when the preset medical data are processed, the data segment with higher weight can be subjected to entity information extraction, and the accuracy of entity information identification is further improved.

After obtaining each entity information and the association probability between each entity information, each entity information can be used as a node, and the association probability between each entity information is used as the edge connection probability between nodes, so that the disease knowledge graph is constructed.

In the embodiment of the application, the preset medical data can be acquired and processed through the entity information identification model to obtain the association probability between each entity information and each entity information, so that the corresponding disease knowledge graph can be automatically and accurately constructed, the subsequent disease examination item recommendation equipment can accurately determine the target disease examination item information according to the disease knowledge graph, and accurately recommend the corresponding disease examination item to the target patient.

Optionally, the entity information of the disease knowledge graph further includes medical guideline information, and correspondingly, the processor, when executing the computer program, is further configured to:

In the embodiment of the application, the medical guideline information can be directly information describing diagnosis guidance, treatment guidance and nursing suggestion of diseases. Alternatively, the medical guideline information may also be the name of an authoritative guideline document corresponding to the disease or download link information. In a disease knowledge map, there is corresponding medical guideline information for each disease.

After the processor executes the above step S104 to determine the target disease feature vector currently matching with the inquiry feature vector, the disease name information of the target disease can be determined according to the target disease feature vector. And then, inquiring a disease knowledge graph according to the disease name information of the target disease, and indexing medical guideline information which has an association relation with the target disease and is used as target medical guideline information. After the target medical guide information is obtained, the information can be fed back to a doctor who diagnoses the target patient in a screen text output or voice broadcast mode.

In the embodiment of the application, the entity information of the disease knowledge graph further comprises medical guideline information, and after the artificial intelligence-based disease examination item recommendation device determines the target disease feature vector, the corresponding target medical guideline information can be determined according to the target disease feature vector, so that auxiliary guidance information can be provided for doctors, and the diagnosis accuracy is improved.

In the embodiment of the present application, the entity information of the disease knowledge graph further includes examination result information, that is, for each disease in the disease knowledge graph, the entity information associated with the disease name information further includes examination result information of an examination item corresponding to the disease, and the examination result information in the disease knowledge graph can provide important reference information for the examination of the current patient.

After the target disease examination item information is determined and the target patient is recommended to perform the examination of the corresponding target disease examination item, examination result information obtained by the target patient performing the examination of the corresponding target disease examination item according to the target disease examination item information is acquired and is called target examination result information.

In one embodiment, the artificial intelligence based disease examination item recommendation device can establish communication connection with the examination devices corresponding to the respective examination items of the hospital. After acquiring the inquiry data of the target patient and determining the corresponding target disease examination item information for the target patient, the address table of the pre-stored examination device can be queried according to the target disease examination item information, and the target address information of the target examination device corresponding to the target disease examination item required by the target patient at present is determined. And sending the unique identification information (such as the diagnosis and treatment number of the hospital or the identification number of the patient) of the target patient to be checked to the target checking equipment according to the target address information. And then, when the target examination equipment detects that the target patient enters the unique identification information to examine the target disease examination item, sending the target examination result information of the target patient obtained by the target examination equipment to the artificial intelligence-based disease examination item recommendation equipment, so that the artificial intelligence-based disease examination item recommendation equipment can obtain the target examination result information.

After the target examination result information is obtained, the artificial intelligence-based disease examination item recommendation device inputs the examination result information and the inquiry text information obtained in the step S102 into a disease decision module of the examination item recommendation model for processing, so as to obtain corresponding diagnosis result information. The diagnosis result information may include information indicating whether the target patient has a disease, and may further include name information of the disease from which the target patient has.

In one embodiment, the disease decision module includes a graph convolution layer and a second classifier. And inputting the target graph data into a graph volume layer of a disease decision module for graph volume processing according to target graph data which is composed of inquiry text information and target examination result information and contains all entity information, and obtaining a target symptom feature vector corresponding to the current target patient. Then, inputting the target symptom feature vector into a second classifier, and calculating the similarity between the target symptom feature vector and the symptom feature vector corresponding to each disease pre-stored in the second classifier; if the similarity between the target symptom characteristic vector and each prestored symptom characteristic vector is smaller than the preset similarity, the information indicating that the target patient does not suffer from the target disease currently is output as a diagnosis result. If symptom feature vectors with the similarity degree larger than or equal to the preset similarity degree exist in the symptom feature vectors, the target patient is judged to have the disease, disease name information corresponding to the symptom feature vectors with the similarity degree larger than or equal to the preset similarity degree of the target symptom feature vectors is determined as name information of the disease currently suffered by the target patient, and the disease name information is output as diagnosis result information of the target patient.

In the embodiment of the application, after the target disease examination item information corresponding to the target patient is determined, the target examination result information obtained by examining the target patient according to the target disease examination item information can be obtained, and the inquiry text information and the target examination result information are processed by the disease decision module of the examination item recommendation model to obtain the diagnosis result information, so that accurate diagnosis reference information can be provided for a doctor.

In one embodiment, after the artificial intelligence-based disease examination item recommendation device obtains corresponding diagnosis result information through the disease decision module, a group of entity information with a mutual correlation relationship can be generated according to the obtained inquiry text information, the target examination result information and the diagnosis result information, and the group of entity information is added to a preset disease knowledge graph to obtain an updated disease knowledge graph.

In another embodiment, after obtaining the corresponding diagnosis result information through the disease decision module, the artificial intelligence-based disease examination item recommendation device may obtain the final diagnosis result information of the target patient (the exact diagnosis result information obtained by combining professional analysis and judgment of a doctor), form a test set with the inquiry text information, the target examination result information and the diagnosis result information, form a verification set with the inquiry text information, the target examination result information and the final diagnosis result information of the target patient, continue to perform iterative training on the examination item recommendation model, and further update the model parameters of the examination item recommendation model, thereby obtaining an updated examination item recommendation model.

In the embodiment of the application, after the diagnosis result information is obtained, the disease knowledge map and/or the examination item recommendation model can be automatically updated according to the inquiry text information, the target examination result information and the diagnosis result information, so that the performance of the artificial intelligence-based disease examination item recommendation equipment can be automatically optimized, and the accuracy of subsequent disease examination item recommendation is improved.

acquiring target disease description information of the target patient;

In the embodiment of the application, after the current diagnosis result information of the target patient is obtained and a diagnosis reference is provided for a doctor, the information describing the state of an illness of the target patient by the doctor in combination with the judgment of the doctor is obtained and is called as target state of illness description information. In one embodiment, the target condition description information dictated by the physician may be entered by the voice capture module. In another embodiment, the description information filled in by the doctor in the designated electronic form can be obtained to obtain the target disease description information.

After the target disease description information of the target patient is acquired, the target disease description information and the diagnosis result information are input into a treatment scheme recommending module of the examination item recommending model together for processing, and a target treatment scheme suitable for the target patient is obtained.

In one embodiment, the treatment plan recommendation module includes a graph convolutional layer and a third classifier. And inputting the graph data which is composed of the target disease description information and the diagnosis result information and contains a plurality of entity information into a graph convolution layer of the treatment scheme recommendation module for graph convolution processing to obtain a target disease characteristic vector. And then inputting the target disease condition characteristic vector into a third classifier, and determining a treatment scheme matched with the target disease condition characteristic vector as a target treatment scheme.

In the embodiment of the application, after the diagnosis result information is determined, the target illness state description information of the target patient by the doctor can be acquired, and the target illness state description information can sufficiently represent the illness state of the target patient, so that after the target illness state description information and the diagnosis result information are analyzed and processed by the treatment scheme recommending module of the examination item recommending model, the target treatment scheme can be accurately recommended for the target patient, and the doctor can be effectively assisted in performing subsequent treatment on the target patient.

In this embodiment of the application, after obtaining the diagnosis result information, the processor may query the current hospital disease case or the current hospital doctor professional field information stored in the memory or the third-party database, and send referral advice information to the target patient if the query determines that the current hospital does not have the diagnosis and treatment experience of the target disease described in the diagnosis result information or does not have a professional doctor who is good at treating the target disease.

Further, the artificial intelligence based disease examination item recommendation apparatus may search for a medical experience with the target disease or other hospitals of the medical professionals based on the name information of the target disease contained in the diagnosis result information, and output the searched name information and/or address information of the target hospital as referral advice information.

In the embodiment of the application, after the diagnosis result information is obtained, the corresponding referral suggestion information can be automatically determined, so that the referral suggestion can be timely provided for the target patient when the current hospital cannot perform subsequent diagnosis and treatment on the target patient, and the delay of disease treatment of the target patient is avoided.

determining information of the target medicine according to the diagnosis result information; the target drug is a drug for treating a target disease;

and inquiring a target pharmacy with the target medicine, and sending the address information of the target pharmacy to the terminal equipment of the target patient.

In the embodiment of the application, the information of the target medicine can be automatically determined according to the diagnosis result information, and the address information of the target pharmacy capable of purchasing the target medicine is sent to the terminal device of the target patient, so that the target patient can conveniently purchase the target medicine, and particularly when the target medicine is a rare medicine which is difficult to purchase, the target patient can be effectively provided with convenient information.

Or sending the information of the target medicine to an automatic medicine dispensing device of a hospital to instruct the automatic medicine dispensing device to provide the target medicine for the target patient according to the information of the target medicine.

In the embodiment of the application, the information of the target medicine can be automatically sent to the automatic medicine dispensing device, so that the target medicine can be automatically provided for the target patient, and the medicine can be accurately and efficiently provided for the target patient under the condition of saving labor cost.

Example two:

referring to fig. 2, fig. 2 is a flowchart illustrating an implementation of a method for recommending disease inspection items based on artificial intelligence according to an embodiment of the present application, where an execution subject of the method is the disease inspection item recommendation apparatus based on artificial intelligence as illustrated in fig. 1. The artificial intelligence-based disease examination item recommendation method shown in fig. 2 is detailed as follows:

in S201, the inquiry data of the target patient is acquired.

In S202, extracting text information matched with entity information in a preset disease knowledge graph from the inquiry data to obtain inquiry text information; wherein the entity information in the disease knowledge graph comprises disease name information, disease symptom information and disease examination item information.

In S203, determining an inquiry feature vector corresponding to the inquiry text information in a graph volume layer of an inspection item recommendation model; and the examination item recommendation model is obtained by training according to the disease knowledge graph.

In S204, determining a target disease feature vector matching the inquiry feature vector in a first classifier of the examination item recommendation model, and determining target disease examination item information according to the target disease feature vector; wherein the target disease feature vector is a feature vector corresponding to a target disease; the target disease examination item information is the disease examination item information corresponding to the target disease.

Optionally, before the step S201, the method further includes:

s2001: acquiring preset medical data;

s2002: inputting the preset medical data into an entity information identification model for processing to obtain each entity information and the association probability among the entity information; the entity information identification model is a bidirectional cyclic neural network obtained by training based on medical sample data carrying entity information labels;

s2003: and constructing the disease knowledge graph according to the entity information and the association probability among the entity information.

Optionally, the entity information of the disease knowledge graph further includes medical guideline information, and correspondingly, after the step S204, further includes:

Optionally, the entity information of the disease knowledge graph further includes examination result information, and correspondingly, after the step S204, further includes:

Optionally, after the disease decision module that inputs the inquiry text information and the examination result information into the examination item recommendation model for processing to obtain diagnosis result information, the method further includes:

acquiring target disease description information of the target patient;

It should be noted that, since the implementation steps of the method embodiment and the artificial intelligence based disease examination item recommendation device in the previous embodiment are based on the same concept, specific functions and technical effects thereof may be referred to in the description of the first embodiment, and are not described herein again.

It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

Example three:

fig. 3 is a schematic structural diagram illustrating an artificial intelligence-based disease examination item recommendation apparatus according to an embodiment of the present application. For convenience of explanation, only the portions related to the embodiments of the present application are shown:

the disease examination item recommendation device based on artificial intelligence comprises: an inquiry data acquisition unit 31, an inquiry text information determination unit 32, an inquiry feature vector determination unit 33, and a target disease examination item determination unit 34. Wherein:

an inquiry data acquiring unit 31 for acquiring inquiry data of the target patient.

An inquiry text information determining unit 32, configured to extract text information matched with entity information in a preset disease knowledge graph from the inquiry data to obtain inquiry text information; wherein the entity information in the disease knowledge graph comprises disease name information, disease symptom information and disease examination item information.

An inquiry feature vector determining unit 33, configured to determine an inquiry feature vector corresponding to the inquiry text information in a graph volume layer of the inspection item recommendation model; and the examination item recommendation model is obtained by training according to the disease knowledge graph.

A target disease examination item determining unit 34, configured to determine a target disease feature vector matching the inquiry feature vector in the first classifier of the examination item recommendation model, and determine target disease examination item information according to the target disease feature vector; wherein the target disease feature vector is a feature vector which integrates the entity information corresponding to the target disease in the disease knowledge base; the target disease examination item information is the disease examination item information matched with the target disease feature vector.

Optionally, the artificial intelligence based disease examination item recommendation apparatus further includes:

the disease knowledge map construction unit is used for acquiring preset medical data; inputting the preset medical data into an entity information identification model for processing to obtain each entity information and the association probability among the entity information; the entity information identification model is a bidirectional cyclic neural network obtained by training based on medical sample data carrying entity information labels; and constructing the disease knowledge graph according to the entity information and the association probability among the entity information.

Optionally, the entity information of the disease knowledge graph further includes medical guideline information, and correspondingly, the artificial intelligence based disease examination item recommendation apparatus further includes:

and the medical guideline information determining unit is used for determining target medical guideline information corresponding to the target disease feature vector.

Optionally, the entity information of the disease knowledge graph further includes examination result information, and correspondingly, the artificial intelligence based disease examination item recommendation apparatus further includes:

a diagnosis result information determining unit for acquiring target examination result information obtained by examining the target patient according to the target disease examination item information; and inputting the inquiry text information and the examination result information into a disease decision module of the examination item recommendation model for processing to obtain diagnosis result information.

a treatment plan determination unit for acquiring target disease description information of the target patient;

and the referral suggestion unit is used for determining corresponding referral suggestion information according to the diagnosis result information.

It should be noted that, for the information interaction, execution process, and other contents between the above-mentioned devices/units, the specific functions and technical effects thereof are based on the same concept as those of the first embodiment, and specific reference may be made to the related description in the first embodiment, and no further description is given here.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. 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.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus/electronic device and method may be implemented in other ways. For example, the above-described apparatus/electronic device embodiments are merely illustrative, and for example, the division of the modules or units is only one logical division, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow in the method of the embodiments described above can be realized by a computer program, which can be stored in a computer-readable storage medium and can realize the steps of the embodiments of the methods described above when the computer program is executed by a processor. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.

The above-mentioned embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present application and are intended to be included within the scope of the present application.

Claims

1. An artificial intelligence based disease examination item recommendation device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that when the computer program is executed by the processor, the following steps are implemented:

acquiring inquiry data of a target patient;

2. The artificial intelligence based disease inspection item recommendation device of claim 1, wherein the processor, when executing the computer program, is further configured to:

acquiring preset medical data;

3. The artificial intelligence based disease examination item recommendation device of claim 1, wherein the entity information of the disease knowledge-graph further comprises medical guideline information, and correspondingly, the processor, when executing the computer program, is further configured to:

4. The artificial intelligence based disease examination item recommendation apparatus of any one of claims 1 to 3, wherein the entity information of the disease knowledge graph further comprises examination result information, and the processor, when executing the computer program, is further configured to:

5. The artificial intelligence based disease inspection item recommendation device of claim 4, wherein the processor, when executing the computer program, is further configured to:

6. The artificial intelligence based disease examination item recommendation device of claim 4, wherein the entity information of the disease knowledge-graph further comprises information of a treatment protocol, the processor, when executing the computer program, is further configured to:

acquiring target disease description information of the target patient;

7. The artificial intelligence based disease inspection item recommendation device of claim 4, wherein the processor, when executing the computer program, is further configured to:

8. A disease examination item recommendation method based on artificial intelligence is characterized by comprising the following steps:

acquiring inquiry data of a target patient;

9. An artificial intelligence-based disease examination item recommendation device, comprising:

10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method as claimed in claim 8.