WO2022062353A1 - 医疗数据处理方法、装置、计算机设备和存储介质 - Google Patents

医疗数据处理方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2022062353A1
WO2022062353A1 PCT/CN2021/084350 CN2021084350W WO2022062353A1 WO 2022062353 A1 WO2022062353 A1 WO 2022062353A1 CN 2021084350 W CN2021084350 W CN 2021084350W WO 2022062353 A1 WO2022062353 A1 WO 2022062353A1
Authority
WO
WIPO (PCT)
Prior art keywords
disease
drug
target
data
initial
Prior art date
Application number
PCT/CN2021/084350
Other languages
English (en)
French (fr)
Inventor
陈思彤
王垂新
赵建双
Original Assignee
康键信息技术（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 康键信息技术（深圳）有限公司 filed Critical 康键信息技术（深圳）有限公司
Publication of WO2022062353A1 publication Critical patent/WO2022062353A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
    • G16H20/13—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients delivered from dispensers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367—Ontology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • the present application relates to the technical field of big data, and in particular, to a medical data processing method, apparatus, computer equipment and storage medium.
  • the inventor realizes that the recommended drugs are usually all the drugs corresponding to the target disease in the historical prescription data, and the doctor needs to re-screen manually before making recommendations, which makes the drug recommendation process less intelligent. , the processing efficiency is low.
  • a medical data processing method comprising:
  • Obtain pending medical data which includes the disease identifier of the target disease
  • the standard vector space is queried to determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the standard vector space is generated based on the knowledge graph of the relationship between diseases, symptoms and drugs , the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map;
  • the initial drug corresponding to the obtained correlation index satisfying the preset condition is the target drug corresponding to the target disease.
  • a medical data processing device comprising:
  • the pending medical data acquisition module is used to acquire the pending medical data, and the pending medical data includes the disease identifier of the target disease;
  • the query module is used to query the standard vector space based on the disease identification to determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the standard vector space is based on the association between diseases, symptoms and drugs
  • the knowledge map of the relationship is generated, and the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map;
  • the correlation index determination module has the ability to determine the correlation index between the target disease and each initial drug according to the disease feature vector and the corresponding multiple drug feature vectors;
  • the target drug determination module is used to obtain the initial drug corresponding to the correlation index satisfying the preset condition as the target drug corresponding to the target disease.
  • a computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:
  • Obtain pending medical data which includes the disease identifier of the target disease
  • the standard vector space is queried to determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the standard vector space is generated based on the knowledge graph of the relationship between diseases, symptoms and drugs , the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map;
  • the initial drug corresponding to the obtained correlation index satisfying the preset condition is the target drug corresponding to the target disease.
  • Obtain pending medical data which includes the disease identifier of the target disease
  • the standard vector space is queried to determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the standard vector space is generated based on the knowledge graph of the relationship between diseases, symptoms and drugs , the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map;
  • the initial drug corresponding to the obtained correlation index satisfying the preset condition is the target drug corresponding to the target disease.
  • This application can improve the accuracy of drug recommendation.
  • FIG. 1 is an application scenario diagram of a medical data processing method in one embodiment
  • FIG. 2 is a schematic flowchart of a medical data processing method in one embodiment
  • FIG. 3 is a schematic flowchart of a medical data processing method in another embodiment
  • FIG. 4 is a structural block diagram of a medical data processing apparatus in one embodiment
  • FIG. 5 is a diagram of the internal structure of a computer device in one embodiment.
  • the medical data processing method provided in this application can be applied to the application environment shown in FIG. 1 .
  • the terminal 102 communicates with the server 104 through the network.
  • the doctor inputs the medical data to be processed through the terminal 102 and sends it to the server 104 , and the medical data to be processed includes the disease identifier of the target disease.
  • the server 104 may query the standard vector space based on the disease identifier to determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the standard vector space is based on the disease
  • the knowledge map of the relationship between , symptoms and drugs is generated, and the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map.
  • the server 104 determines the correlation index between the target disease and each initial drug according to the disease feature vector and the corresponding plurality of drug feature vectors, and then obtains the initial drug corresponding to the correlation index satisfying the preset condition as the target corresponding to the target disease. drug.
  • the terminal 102 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices, and the server 104 can be implemented by an independent server or a server cluster composed of multiple servers.
  • a medical data processing method is provided, which is described by taking the method applied to the server in FIG. 1 as an example, including the following steps:
  • step S202 medical data to be processed is acquired, and the medical data to be processed includes the disease identifier of the target disease.
  • the medical data to be processed refers to data generated after a doctor conducts a consultation, for example, it may be online consultation result data and the like.
  • the medical data may include the target disease to be consulted and the symptoms corresponding to the target disease.
  • the medical data to be processed may also include a disease identifier corresponding to the target disease, for example, a disease name or an identifier such as a unique disease ID (Identity document) corresponding to the target disease in the online consultation system, such as a corresponding identifier for diabetes.
  • a disease identifier corresponding to the target disease
  • a disease name for example, a disease name or an identifier such as a unique disease ID (Identity document) corresponding to the target disease in the online consultation system, such as a corresponding identifier for diabetes.
  • the unique identification ID is TNB01 and so on.
  • the medical data to be processed may be data directly input by the doctor into the online consultation system based on their own diagnosis experience, or may also be the online consultation system after screening the corresponding disease symptoms in the online consultation system.
  • the results of big data statistics or pre-configured results are automatically obtained medical data.
  • the online consultation system can send a result confirmation request to the terminal before generating the final medical data , to request the doctor to determine whether the medical data to be processed is incorrect, and output it after the doctor confirms that it is correct.
  • Step S204 based on the disease identifier, query the standard vector space to determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the standard vector space is based on the association relationship between diseases, symptoms and drugs.
  • the standard vector space includes feature vectors corresponding to each disease and each drug in the knowledge graph.
  • the knowledge graph refers to a graph including diseases, symptoms, medicines, and the corresponding relationship among the three, and the knowledge graph associates the corresponding relationship between different diseases, corresponding symptoms, and the corresponding medicines that have been prescribed.
  • the standard vector space includes multiple feature vectors corresponding to diseases and drugs, and each feature vector reflects the relationship between diseases and drugs.
  • the standard vector space may refer to a medical entity vector space.
  • the server may construct a medical entity vector space in advance, and according to the corresponding disease identifier in the diagnosis result, find the disease feature vector corresponding to the target disease and the corresponding drug feature vector of the initial drug from the medical entity vector space.
  • the server may find multiple initial medicines corresponding to the target disease.
  • the corresponding medicine feature vector found may include medicines of multiple initial medicines such as medicine bottle A, medicine B, and medicine C. Feature vector.
  • Step S206 according to the disease feature vector and the corresponding multiple drug feature vectors, determine the correlation index between the target disease and each initial drug.
  • the correlation index refers to an index of the relationship between a drug and a disease. The higher the index value, the more related the drug and the disease, and the more suitable the drug is for the disease.
  • the server may calculate the disease feature vector of the target disease and the drug features of the multiple initial drugs respectively through Cosine similarity calculation and other methods.
  • a vector of correlation metrics can be shown in the following formula (1):
  • E disease represents the disease feature vector
  • E drug represents the drug feature vector of each initial drug
  • S represents the correlation index
  • step S208 the initial drug corresponding to the correlation index satisfying the preset condition is obtained as the target drug corresponding to the target disease.
  • the preset condition is a preset filter condition of the final correlation index, for example, the condition that the index value is the highest or the lowest.
  • the server may sort the calculated correlation indexes corresponding to the multiple initial drugs, and determine the initial drug with the highest index value from the sorted correlation indexes as the target drug corresponding to the target disease, and recommend to the terminal for feedback to the doctor through the terminal.
  • the server can also calculate the correlation between each initial medicine, and then recommend multiple initial medicines with strong correlation to the terminal.
  • the medical data to be processed includes the disease identifier of the target disease by acquiring the medical data to be processed, and then based on the disease identifier, the standard vector space is queried to determine the disease feature vector corresponding to the target disease and the corresponding target disease.
  • the drug feature vectors of multiple initial drugs The standard vector space is generated based on the knowledge map of the relationship between diseases, symptoms and drugs.
  • the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map, and further according to the disease feature vector and Corresponding to multiple drug feature vectors, the correlation index between the target disease and each initial drug is determined, and the initial drug corresponding to the correlation index satisfying the preset condition is obtained as the target drug corresponding to the target disease.
  • the correlation determination of the target drug based on the feature vector obtained from the pre-constructed medical entity vector space can reduce the amount of manual participation and improve the intelligence of drug recommendation compared with the recommendation of drugs directly based on the statistical results. level, which can improve the efficiency of data processing.
  • the constructed medical entity vector space can reflect the association information between diseases, symptoms and medicines, and then based on the medical entity vector space, the When the feature vector is used for drug recommendation, the accuracy of drug recommendation can be improved.
  • the generation method of the standard vector space may include: acquiring a knowledge graph corresponding to the association between diseases, symptoms and medicines; extracting features from the knowledge graph through a graph neural network model to obtain each disease in the corresponding knowledge graph
  • the disease feature vector and the drug feature vector corresponding to each drug, each disease feature vector and each drug feature vector include the relationship between the corresponding diseases, symptoms and drugs.
  • the server may acquire data related to diseases, symptoms, and medicines, and generate a knowledge graph based on the acquired data.
  • the server can extract the correspondence between diseases, medicines, and symptoms on the knowledge graph through a multi-relational graph neural network model, for example, respectively extract the symptoms and medicines corresponding to each disease The feature data between them, and based on the extracted feature data, a feature vector corresponding to each disease and each drug is generated, that is, a medical entity vector space is generated.
  • the graph neural network model may be pre-trained based on artificial intelligence.
  • the server may use the historical consultation data stored in the online question answering system database as the training set data, and generate the training set knowledge map.
  • the server annotates the knowledge graph to obtain the labeled knowledge graph of the training set.
  • the server inputs the knowledge graph of the training set into the initial graph neural network model constructed, and extracts features and generates corresponding feature vectors to train the initial graph neural network model.
  • the server can compare the corresponding relationship between the diseases, symptoms and medicines determined by each feature vector in the obtained medical entity vector space with the diseases, symptoms and medicines in the training set data. The correspondence between the drugs is compared, and the loss value is calculated.
  • the server can calculate the loss value of the model by defining a binary cross-entropy loss function, and the binary cross-entropy loss function formula is shown in formula (2):
  • the server may update the model parameters of the initial graph neural network model based on the calculated loss value, and iteratively process the initial graph neural network model to obtain a trained graph neural network model.
  • the knowledge graph of the correspondence between diseases, symptoms and medicines is obtained, and then features are extracted through the graph neural network model, and a medical entity vector space is constructed. Performing quantization processing is convenient for subsequent similarity calculation, which can improve the efficiency of data processing.
  • acquiring a knowledge graph corresponding to the association between diseases, symptoms, and medicines may include: acquiring preset medical consultation data; extracting data related to diseases, symptoms, and medicines from online medical consultation data Relevant initial target data; standardized preprocessing on initial target data to obtain standardized preprocessed target data; based on standardized preprocessed target data, a knowledge map of the association between diseases, symptoms and drugs is established.
  • the medical consultation data refers to online consultation data of doctors and patient users, which may include consultation conversations and consultation prescriptions finally generated by doctors based on online consultations.
  • the server may obtain online consultation data from historical data of the online consultation system, and then extract target data from the online consultation data according to preset keywords, for example, according to a preset disease Name, disease symptoms, and drug names, etc., and extract target data including diseases, symptoms, and drugs.
  • the server can standardize the extracted target data, for example, standardize the disease name, the drug name, and the format between the data, etc., to generate the target data after the standardized data.
  • the server may, based on the standardized target data, establish a knowledge map of the correspondence between each disease, each corresponding symptom, and each corresponding drug.
  • the construction of the knowledge map is generated based on the actual consultation data, so that the construction of the knowledge map can have a practical basis, and the accuracy of the constructed knowledge map can be improved.
  • performing standardized preprocessing on the initial target data to obtain the standardized preprocessed target data may include: obtaining a medical standard database; and extracting the target data to be converted based on a preset keyword Converting data; standardizing preprocessing and converting the data to be converted through the medical standard database to obtain the target data after standardization and preprocessing.
  • the medical standardization database refers to a database created based on industry standards.
  • the database records the correspondence between the standard names of various diseases and the common names used by doctors in practical applications, as well as the standard names of medicines and doctors in practical applications.
  • the medical standardization database can store the standard drug name "Amoxicillin” and the non-standard drug name "Amoxicillin”. Correspondence between the names "Amoxicillin” or "Amoxicillin”.
  • the server may perform standard conversion on the corresponding keywords in the target data according to the medical standard database, so as to obtain standardized target data.
  • the server may add non-standardized keywords to the corresponding standardized data based on the received association indication, thereby establishing a corresponding relationship.
  • the server may receive the standardized data adding instruction sent by the terminal to add the corresponding standardized data, and add the corresponding keyword to the corresponding standardized data to establish a corresponding relationship.
  • standardized target data can be obtained, so that the target data can be converted and generated based on the medical standard database, so that the standardized target data can be generated according to the same standard. , to improve the accuracy of subsequent knowledge graph establishment, thereby improving the accuracy of target drug recommendation.
  • the method may further include: acquiring prescription data corresponding to the target disease; , determine the weight index corresponding to each initial drug; based on the weight index and correlation index of each initial drug, obtain the final correlation index corresponding to each initial drug.
  • the server may acquire the doctor's prescribing data for the target disease, such as prescribing prescriptions, and generate weight indicators corresponding to each initial drug according to the acquired prescribing data.
  • the server may obtain the final correlation index corresponding to each initial drug according to the weight index and the correlation index of each initial drug.
  • the server may also multiply the weight index of each initial drug and the correlation index corresponding to each initial drug to generate a final correlation index corresponding to each initial drug.
  • acquiring the initial drug corresponding to the correlation index satisfying the preset condition as the target drug corresponding to the target disease may include: acquiring the initial drug corresponding to the final correlation index satisfying the preset condition as the corresponding target disease target drug.
  • the preset condition is a preset filter condition of the final correlation index, for example, the condition that the index value is the highest or the lowest.
  • the server may determine one or more initial medicines with the highest index value from among the plurality of initial medicines as the target medicine corresponding to the target disease according to the determined final correlation index, and recommend it to the terminal.
  • the server may sort a plurality of initial medicines according to the calculated final correlation index to obtain a plurality of sorted initial medicines. Then, the server selects the target medicine corresponding to the target disease from the sorted initial medicines.
  • the final correlation index corresponding to each initial drug is generated, and the target drug is determined and recommended, so that the drug can be Combined with the actual prescription data, the accuracy of drug recommendation can be improved.
  • the above method may further include: acquiring updated data on the database, where the updated data includes the correspondence between the target disease and each drug; and detecting whether there is a new drug corresponding to the target disease according to the updated data; When it is detected that there is a new drug corresponding to the target disease, the frequency of occurrence of the corresponding relationship between the target disease and the new drug in the updated data is counted, and when the frequency of occurrence is greater than the preset threshold, based on the target disease and the new drug Increase the correspondence between drugs and update the knowledge map.
  • the updated data refers to the obtained online real-time prescription data.
  • the server when the server acquires the online real-time prescription data, the server updates the database through the acquired online real-time prescription data, for example, updates the medicine used for a certain disease.
  • updating the database according to the update data may refer to adding, deleting or changing, for example, for a certain disease, adding a new drug, or deleting a corresponding existing drug, or changing a corresponding existing drug, etc. .
  • the server may also perform real-time detection according to the acquired update data to determine whether there is a new drug corresponding to the target disease.
  • the newly added drug here refers to a drug that does not correspond to the target disease in the existing prescription history, that is, the newly added drug is not used to treat the target disease.
  • the server when the server detects that there is a new drug corresponding to the target disease in the update data, the server may perform statistics on the occurrence frequency of the new drug corresponding to the target disease, for example, count the newly added drug in real time The number of times to apply to this target disease.
  • the server may determine the frequency of occurrence of the newly added drug based on a preset threshold, so as to determine whether the medication mode of the newly added drug reaches statistical significance for the target disease.
  • the server when the server determines that the frequency of occurrence of the correspondence between the target disease and the newly added drug is greater than a preset threshold, that is, when it is determined that the use of the newly added drug for the target disease is not an accidental cause, the server may determine that the The newly added drug is a new drug for treating the target disease. Continuing to refer to FIG. 3 , the server may update the knowledge map based on the correspondence between the target disease and the newly added drug.
  • the server may also perform real-time statistics on the association relationship between each drug and corresponding disease in the knowledge graph.
  • the knowledge graph may also be Update to make the obtained knowledge graph more accurate.
  • the knowledge graph is updated by combining the updated data, so that the knowledge graph collection online real-time prescription data is generated, the accuracy of the knowledge graph is improved, and the accuracy of the target drug determination can be improved, so as to improve the recommended medicines. accuracy.
  • the above method may further include: uploading at least one of the medical data, the medical entity vector space, and each correlation index to the blockchain, and storing it in a node of the blockchain.
  • Blockchain refers to a new application mode of computer technology such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information for verification. The validity of its information (anti-counterfeiting) and the generation of the next block.
  • the blockchain may include the underlying platform of the blockchain, the platform product service layer, and the application service layer.
  • the server can upload and store one or more data in the medical data, the medical entity vector space and each correlation index in the nodes of the blockchain to ensure the privacy and security of the data.
  • a medical data processing apparatus including: a medical data acquisition module 100 to be processed, a query module 200 , a correlation index determination module 300 and a target drug determination module 400 , wherein:
  • the to-be-processed medical data acquisition module 100 is configured to acquire the to-be-processed medical data, and the to-be-processed medical data includes the disease identifier of the target disease.
  • the query module 200 is used to query the standard vector space based on the disease identifier, and determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the knowledge map of the association relationship is generated, and the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map.
  • the correlation index determination module 300 has the ability to determine the correlation index between the target disease and each initial drug according to the disease feature vector and a plurality of corresponding drug feature vectors.
  • the target drug determination module 400 is configured to obtain the initial drug corresponding to the correlation index satisfying the preset condition as the target drug corresponding to the target disease.
  • the above-mentioned apparatus may further include:
  • the standard vector space generation module is used to generate the generation of standard vector spaces.
  • the standard vector space generation module may include:
  • the knowledge graph acquisition sub-module is used to obtain the knowledge graph of the relationship between diseases, symptoms and medicines.
  • the feature extraction sub-module is used to extract the features of the knowledge graph through the graph neural network model, and obtain the disease feature vector corresponding to each disease in the knowledge graph and the drug feature vector corresponding to each drug.
  • Each disease feature vector and each drug feature vector include Corresponding associations between diseases, symptoms, and medicines.
  • the knowledge graph acquisition sub-module may include:
  • the medical consultation data obtaining unit is used for obtaining preset medical consultation data.
  • the data extraction unit is used to extract initial target data related to diseases, symptoms and medicines from the online medical consultation data.
  • the normalization preprocessing unit is used to perform normalization preprocessing on the initial target data to obtain normalized preprocessed target data.
  • the knowledge graph establishment unit is used to establish a knowledge graph of the association relationship among diseases, symptoms and medicines based on the standardized preprocessed target data.
  • the normalization preprocessing unit may include:
  • the medical standard database acquisition subunit is used to acquire the medical standard database.
  • the to-be-converted data extraction subunit is configured to extract the to-be-converted data to be converted from the initial target data based on the preset keyword.
  • the standardized preprocessing conversion subunit is used to perform standardized preprocessing conversion on the data to be converted through the medical standard database, so as to obtain the standardized preprocessed target data.
  • the above-mentioned apparatus may further include:
  • the prescribing data acquisition module is used for the correlation index determination module 300 to obtain prescribing data corresponding to the target disease after determining the correlation index between the target disease and each initial drug according to the disease feature vector and the corresponding plurality of drug feature vectors.
  • the weight index determination module is used to determine the weight index corresponding to each initial medicine according to the prescription data.
  • the final correlation index determination module is used to obtain the final correlation index corresponding to each initial drug based on the weight index and the correlation index of each initial drug.
  • the target drug determination module 400 is configured to obtain the initial drug corresponding to the final correlation index satisfying the preset condition as the target drug corresponding to the target disease.
  • the above-mentioned apparatus may further include:
  • the more detailed data acquisition module is used to acquire updated data on the database, and the updated data includes the corresponding relationship between the target disease and each drug.
  • the detection module is used to detect whether there is a new drug corresponding to the target disease according to the updated data.
  • the knowledge map update module is used to count the occurrence frequency of the corresponding relationship between the target disease and the new drug in the updated data when it is detected that there is a new drug corresponding to the target disease, and when the occurrence frequency is greater than the preset threshold When , the knowledge graph is updated based on the correspondence between the target disease and the newly added drug.
  • the above-mentioned apparatus may further include:
  • the storage module is used to upload at least one of the medical data, the medical entity vector space and each correlation index to the blockchain, and store it in the nodes of the blockchain.
  • Each module in the above-mentioned medical data processing apparatus may be implemented in whole or in part by software, hardware and combinations thereof.
  • the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
  • a computer device is provided, and the computer device may be a server, and its internal structure diagram may be as shown in FIG. 5 .
  • the computer device includes a processor, memory, a network interface, and a database connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium, an internal memory.
  • the nonvolatile storage medium stores an operating system, a computer program, and a database.
  • the internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium.
  • the database of the computer equipment is used for storing data such as medical data, medical entity vector space and various correlation indicators.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer program when executed by a processor, implements a medical data processing method.
  • FIG. 5 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. Include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.
  • a computer device including a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program: acquiring medical data to be processed, the medical data to be processed includes a target disease Based on the disease identification, query the standard vector space to determine the disease feature vector corresponding to the target disease and the drug feature vectors of multiple initial drugs corresponding to the target disease.
  • the standard vector space is based on the association between diseases, symptoms and drugs.
  • the standard vector space includes the feature vectors of each disease and each drug in the corresponding knowledge map; according to the disease feature vector and the corresponding multiple drug feature vectors, the correlation index between the target disease and each initial drug is determined; The initial drug corresponding to the conditional correlation index is set as the target drug corresponding to the target disease.
  • the method for generating the standard vector space may include: acquiring a knowledge graph corresponding to the association between diseases, symptoms and medicines; performing feature extraction on the knowledge graph by using a graph neural network model, A disease feature vector corresponding to each disease in the knowledge map and a drug feature vector corresponding to each drug are obtained, and each disease feature vector and each drug feature vector include the relationship between the corresponding diseases, symptoms and drugs.
  • the acquisition of a knowledge graph corresponding to the association between diseases, symptoms and medicines may include: acquiring preset medical consultation data; extracting data from online medical consultations Generate initial target data related to diseases, symptoms, and drugs; standardize preprocessing on initial target data to obtain standardized preprocessed target data; establish associations between diseases, symptoms, and drugs based on standardized preprocessed target data knowledge graph.
  • the initial target data is standardized and preprocessed to obtain standardized preprocessed target data, which may include: obtaining a medical standard database;
  • the data to be converted is extracted from the data to be converted; the standardized preprocessing conversion is performed on the data to be converted through the medical standard database to obtain the target data after standardized preprocessing.
  • the processor when the processor executes the computer program, after determining the correlation index between the target disease and each initial drug according to the disease feature vector and the corresponding plurality of drug feature vectors, the following steps may be further implemented: obtaining the corresponding target Prescribing data for diseases; according to the prescribing data, determine the weight index corresponding to each initial drug; based on the weight index and correlation index of each initial drug, obtain the final correlation index corresponding to each initial drug.
  • the initial drug corresponding to the correlation index satisfying the preset condition is obtained as the target drug corresponding to the target disease, which may include: obtaining the final correlation index corresponding to the preset condition.
  • the initial drug is the target drug corresponding to the target disease.
  • the following steps may also be implemented: acquiring updated data on the database, where the updated data includes the correspondence between the target disease and each drug; and detecting whether there is a corresponding target according to the updated data New drugs for diseases; when it is detected that there are new drugs corresponding to the target diseases, the occurrence frequency of the corresponding relationship between the target diseases and the new drugs in the updated data is counted, and when the occurrence frequency is greater than the preset threshold , and update the knowledge map based on the correspondence between the target disease and the new drug.
  • the following steps may be further implemented: uploading at least one of the medical data, the medical entity vector space, and each correlation index to the blockchain, and storing it to a node of the blockchain middle.
  • a computer-readable storage medium may be volatile or non-volatile, and a computer program is stored thereon.
  • the computer program is executed by a processor The following steps are implemented: obtain the medical data to be processed, and the medical data to be processed includes the disease identifier of the target disease; based on the disease identifier, query the standard vector space to determine the disease feature vector corresponding to the target disease and the multiple initial medicines corresponding to the target disease.
  • Drug feature vector the standard vector space is generated based on the knowledge map of the relationship between diseases, symptoms and drugs.
  • the standard vector space includes the feature vectors corresponding to each disease and each drug in the knowledge map; according to the disease feature vector and the corresponding multiple drug features vector, determine the correlation index between the target disease and each initial drug; obtain the initial drug corresponding to the correlation index satisfying the preset condition as the target drug corresponding to the target disease.
  • the method for generating the standard vector space may include: acquiring a knowledge graph corresponding to the relationship between diseases, symptoms and medicines; extracting features from the knowledge graph through a graph neural network model , to obtain the disease feature vector corresponding to each disease in the knowledge map and the drug feature vector corresponding to each drug, and each disease feature vector and each drug feature vector include the relationship between the corresponding diseases, symptoms and drugs.
  • the acquisition of a knowledge graph corresponding to the association between diseases, symptoms and medicines may include: acquiring preset medical consultation data; Extract initial target data related to diseases, symptoms and medicines; standardize preprocessing on initial target data to obtain standard preprocessed target data; establish associations between diseases, symptoms and medicines based on standard preprocessed target data A knowledge graph of relationships.
  • the initial target data is standardized and preprocessed to obtain the standardized preprocessed target data, which may include: obtaining a medical standard database; The data to be converted is extracted from the data; standardized preprocessing is performed on the data to be converted through the medical standard database to obtain the target data after standardized preprocessing.
  • the following steps may also be implemented: obtaining the corresponding Prescribing data of the target disease; according to the prescribing data, determine the weight index corresponding to each initial drug; based on the weight index and correlation index of each initial drug, obtain the final correlation index corresponding to each initial drug.
  • the initial drug corresponding to the correlation index satisfying the preset condition is obtained as the target drug corresponding to the target disease, which may include: obtaining the final correlation index satisfying the preset condition.
  • the corresponding initial drug is the target drug corresponding to the target disease.
  • the following steps may also be implemented: acquiring update data on the database, where the update data includes the correspondence between the target disease and each drug; and detecting whether there is a correspondence according to the update data New drugs for the target disease; when it is detected that there are new drugs corresponding to the target disease, the frequency of occurrence of the corresponding relationship between the target disease and the new drug in the updated data is counted, and when the frequency of occurrence is greater than the preset threshold When , the knowledge graph is updated based on the correspondence between the target disease and the newly added drug.
  • the following steps may be further implemented: uploading at least one of the medical data, the medical entity vector space and each correlation index to the blockchain, and storing it in the blockchain in the node.
  • Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
  • SRAM static RAM
  • DRAM dynamic RAM
  • SDRAM synchronous DRAM
  • DDRSDRAM double data rate SDRAM
  • ESDRAM enhanced SDRAM
  • SLDRAM synchronous chain Road (Synchlink) DRAM
  • SLDRAM synchronous chain Road (Synchlink) DRAM
  • Rambus direct RAM
  • DRAM direct memory bus dynamic RAM
  • RDRAM memory bus dynamic RAM

Abstract

一种医疗数据处理方法、装置、计算机设备和存储介质，涉及大数据技术领域，所述方法包括：获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识（S202）；基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量（S204）；根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标（S206）；获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品（S208）。所述方法能够提升药品推荐的智能化水平。此外，还涉及区块链技术，医疗数据、所述医学实体向量空间以及各相关性指标均可存储于区块链中。

Description

医疗数据处理方法、装置、计算机设备和存储介质
本申请要求于2020年9月23日提交中国专利局、申请号为CN2020110096383，发明名称为“医疗数据处理方法、装置、计算机设备和存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及大数据技术领域，特别是涉及一种医疗数据处理方法、装置、计算机设备和存储介质。
背景技术
随着线的上问诊系统的发展，根据医生的线上问诊结果自动进行药品的推荐是现阶段医疗领域技术发展的趋向。
传统方式中，在医生给出诊断结果后，通常是通过历史开药数据的统计结果向患者推荐对应的药品，例如，针对诊断结果中确定的目标疾病，向用户推荐历史开药数据中对应该目标疾病的药品。
但是，在该种方式中，发明人意识到推荐的药品通常为历史开药数据中对应该目标疾病的所有的药品，还需要医生重新人工进行筛选后再进行推荐，从而使得药品推荐过程不够智能化，处理效率较低。
发明内容
一种医疗数据处理方法，所述方法包括：
获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识；
基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量；
根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标；
获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
一种医疗数据处理装置，所述装置包括：
待处理医疗数据获取模块，用于获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识；
查询模块，用于基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量；
相关性指标确定模块，拥有根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标；
目标药品确定模块，用于获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现以下步骤：
获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识；
基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对 应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量；
根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标；
获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现以下步骤：
获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识；
基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量；
根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标；
获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
本申请可以提升药品推荐的准确性。
附图说明
图1为一个实施例中医疗数据处理方法的应用场景图；
图2为一个实施例中医疗数据处理方法的流程示意图；
图3为另一个实施例中医疗数据处理方法的流程示意图；
图4为一个实施例中医疗数据处理装置的结构框图；
图5为一个实施例中计算机设备的内部结构图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本申请提供的医疗数据处理方法，可以应用于如图1所示的应用环境中。其中，终端102通过网络与服务器104进行通信。医生通过终端102输入待处理医疗数据，并发送至服务器104，待处理医疗数据包括目标疾病的疾病标识。服务器104在获取到待处理医疗数据后，可以基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量。进一步，服务器104根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标，然后获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。其中，终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备，服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。
在一个实施例中，如图2所示，提供了一种医疗数据处理方法，以该方法应用于图1中的服务器为例进行说明，包括以下步骤：
步骤S202，获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识。
其中，待处理医疗数据是指医生进行问诊后所生成的数据，例如，可以是线上问诊结果数据等。在本实施例中，医疗数据中可以包括问诊的目标疾病以及对应该目标疾病的症 状等。
在本实施例中，待处理医疗数据还可以包括对应目标疾病的疾病标识，例如，疾病名称或者是线上问诊系统中对应该目标疾病的唯一疾病ID(Identity document)等标识，如糖尿病对应唯一标识ID为TNB01等。
在本实施例中，待处理医疗数据可以是医生基于自身诊断经验直接输入至线上问诊系统的数据，或者也可以是在线上问诊系统中筛选对应的疾病症状后线上问诊系统根据大数据统计的结果或者是预先配置的结果自动得到的医疗数据。
具体地，当待处理医疗数据为线上问诊系统基于大数据统计的结果或者是预先配置的结果自动生成时，线上问诊系统可以在生成最终医疗数据之前，先发送结果确认请求至终端，以请求医生确定待处理医疗数据是否有误，并在医生确定无误后输出。
步骤S204，基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量。
其中，知识图谱是指包括疾病、症状、药品以及三者之间对应关系的图谱，知识图谱关联了不同的疾病、对应的症状以及开具过的对应的药品之间的对应关系。
标准向量空间包括多个对应疾病、药品的特征向量，各特征向量体现了疾病与药品之间的关系。具体地，标准向量空间可以是指医学实体向量空间。
在本实施例中，服务器可以预先构建医学实体向量空间，并根据诊断结果中对应的疾病标识，从医学实体向量空间中查找到对应目标疾病的疾病特征向量以及对应的初始药品的药品特征向量。
在本实施例中，服务器查到对应目标疾病的初始药品可以是多个，例如，对于糖尿病，查找到对应的药品特征向量可以包括药瓶A、药品B以及药品C等多个初始药品的药品特征向量。
步骤S206，根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标。
其中，相关性指标是指药品与疾病之间关联关系的指标，指标值越高，药品与疾病之间越相关，药品越适用于该疾病。
在本实施例中，服务器在获取到疾病特征向量以及所述多个初始药品的药品特征向量，可以通过Cosine相似度计算等方式，分别计算目标疾病的疾病特征向量以及多个初始药品的药品特征向量的相关性指标。具体地，Cosine计算公式可以如下公式(1)所示：
Figure PCTCN2021084350-appb-000001
其中，E 疾病表示疾病特征向量，E 药品表示各初始药品的药品特征向量，S表示相关性指标。
步骤S208，获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
其中，预设条件为预先设置的最终相关性指标的筛选条件，例如，指标值最高或者最低等条件。
在本实施例中，服务器可以对计算得到的多个初始药品对应的相关性指标进行排序，并从排序后的相关性指标中确定指标值最高的初始药品为对应目标疾病的目标药品，并推荐至终端，以通过终端反馈至医生。
在本实施例中，服务器最终推荐至终端的目标药品也可以是多个，例如，对应同一目标疾病配合使用的多个目标药品等。
具体地，服务器还可以计算各初始药品之间的相关性，然后推荐存在强相关性的多个 初始药品至终端。
上述医疗数据处理方法中，通过获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识，然后基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量，进一步根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标，并获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。从而，使得目标药品基于预先构建的医学实体向量空间得到的特征向量计算得到的相关性确定，相比于直接根据统计结果进行药品的推荐，减少了人工的参与量，提升了药品推荐的智能化水平，进而可以提升数据处理的效率。并且，由于医学实体向量空间基于疾病、症状以及药品之间对应关系的知识图谱生成，可以使得构建的医学实体向量空间体现了疾病、症状以及药品之间的关联信息，进而基于医学实体向量空间得到的特征向量进行药品推荐的时候，可以提升药品推荐的准确性。
在其中一个实施例中，标准向量空间的生成方式可以包括：获取对应疾病、症状以及药品之间关联关系的知识图谱；通过图神经网络模型对知识图谱进行特征提取，得到对应知识图谱中各疾病的疾病特征向量以及对应各药品的药品特征向量，各疾病特征向量以及各药品特征向量中包括对应的疾病、症状以及药品之间的关联关系。
在本实施例中，服务器可以获取与疾病、症状以及药品相关的数据，并基于获取的数据生成知识图谱。
在本实施例中，服务器在得到知识图谱后，可以通过多关系图神经网络模型进行知识图谱上的疾病、药品、症状之间对应关系的提取，例如，分别提取出各疾病对应的症状以及药品之间的特征数据，并基于提取到的特征数据生成对应各疾病以及各药品的特征向量，即生成医学实体向量空间。
在本实施例中，图神经网络模型可以基于人工智能预先训练完成的模型。具体地，服务器可以以线上问答系统数据库中存储的历史问诊数据作为训练集数据，并生成训练集知识图谱。
进一步，服务器对知识图谱进行标注，得到标注后的训练集知识图谱。
进一步，服务器将通过训练集知识图谱输入构建的初始图神经网络模型中，并进行特征的提取以及生成对应的特征向量，以对初始图神经网络模型进行训练。
在本实施例中，在图神经网络模型的训练过程中，服务器可以将得到的医学实体向量空间中各特征向量确定的疾病、症状以及药品之间的对应关系与训练集数据中疾病、症状以及药品之间的对应关系进行比较，并进行损失值的计算。
在本实施例中，服务器可以通过定义二元交叉熵损失函数进行模型损失值的计算，二元交叉熵损失函数公式如公式(2)所示：
Figure PCTCN2021084350-appb-000002
其中，y表示输入模型的数据，
Figure PCTCN2021084350-appb-000003
表示模型输出的结果。
在本实施例中，服务器可以基于计算得到的损失值对初始图神经网络模型的模型参数进行更新，并对初始图神经网络模型进行迭代处理，以得到训练完成的图神经网络模型。
上述实施例中，获取疾病、症状以及药品之间对应关系的知识图谱，然后通过图神经网络模型进行特征的提取，并构建医学实体向量空间，从而，图神经网络模型可以将非量化的图谱数据进行量化处理，便于后续相似性的计算，可以提升数据处理的效率。
在其中一个实施例中，获取对应疾病、症状以及药品之间关联关系的知识图谱，可以包括：获取预设的医疗问诊数据；从线上医疗问诊数据中提取出与疾病、症状以及药品相 关的初始目标数据；对初始目标数据进行标准化预处理，得到标准化预处理后的目标数据；基于标准化预处理后的目标数据，建立疾病、症状以及药品之间关联关系的知识图谱。
其中，医疗问诊数据是指医生以及病人用户的线上问诊数据，可以包括问诊对话以及医生基于线上问诊最终生成的问诊处方。
在本实施例中，服务器可以从线上问诊系统的历史数据中获取线上问诊数据，然后根据预设的关键词从线上问诊数据中提取出目标数据，例如根据预设的疾病名称、疾病症状以及药品名称等，提取出包含疾病、症状以及药品的目标数据。
进一步，服务器可以对提取的目标数据进行标准化数据，例如，对疾病名称、药品名称以及数据之间的格式等进行标准化数据，生成标准化数据后的目标数据。
在本实施例中，服务器在得到标准化的目标数据后，可以基于标准化的目标数据，建立各疾病、对应的各症状以及对应的各药品之间对应关系的知识图谱。
上述实施例中，通过获取医疗问诊数据并构建知识图谱，使得知识图谱的构建基于实际的问诊数据生成，可以使得知识图谱的构建具备实践依据，提升构建的知识图谱的准确性。
在其中一个实施例中，对初始目标数据进行标准化预处理，得到标准化预处理后的目标数据，可以包括：获取医学标准数据库；基于预设关键字，从初始目标数据中提取出待转换的待转换数据；通过医学标准数据库对待转换数据进行标准化预处理转换，得到标准化预处理后的目标数据。
其中，医学标准化数据库是指基于行业标准出创建的数据库，数据库中记载了各疾病的标准名称以及医生在实际应用的所用的通用名称之间的对应关系，以及药品的标准名称以及医生在实际应用的所用的通用名称之间的对应关系。例如，对于医生常用的药品名称“安莫西林”或“安默西林”，其标准药品名称为“阿莫西林”等，则医学标准化数据库可以存储有标准药品名称“阿莫西林”与非标准名称“安莫西林”或“安默西林”之间的对应关系。
在本实施例中，服务器可以根据医疗标准数据库对目标数据中对应的关键字进行标准转换，以得到标准化后的目标数据。
在本实施例中，对于某些关键字，在医疗标准数据库中可能不存在对应的标准数据或者不存在对应的关系，则可以通过人工手动的方式进行判定并进行关系的建立，关联对应的标准化数据。例如，当医疗标准数据库中存在对应的标准化数据时，则服务器可以基于接收的关联指示，将非标准化的关键字添加至对应标准化数据后，从而建立对应关系。当医疗标准数据库中不存在对应的标准化数据时，则服务器可以接收终端发送的标准化数据添加指令，以添加对应的标准化数据，并将对应的关键字添加至对应的标准化数据后，建立对应关系。
上述实施例中，通过基于获取的医疗标准数据库，并进行关键字的转换，从而得到标准化的目标数据，从而可以使得目标数据基于医疗标准数据库转换生成，使得标准化处理后的目标数据按照同一标准生成，提升后续知识图谱建立的准确性，进而提升目标药品推荐的准确性。
在其中一个实施例中，根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标之后，还可以包括：获取对应目标疾病的开药数据；根据开药数据，确定对应各初始药品的权重指标；基于各初始药品的权重指标以及相关性指标，得到对应各初始药品的最终相关性指标。
在本实施例中，服务器可以获取医生的对该目标疾病的开药数据，例如开药处方等，并根据获取的该开药数据，生成对应各初始药品的权重指标。
进一步，服务器可以根据各初始药品的权重指标以及相关性指标，得到对应各初始药品的最终相关性指标。
在本实施例中，服务器也可以将各初始药品的权重指标以及对应各初始药品的相关性指标进行相乘处理，生成对应各初始药品的最终相关性指标。
在本实施例中，获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品，可以包括：获取满足预设条件的最终相关性指标所对应的初始药品为对应目标疾病的目标药品。
如前所述，预设条件为预先设置的最终相关性指标的筛选条件，例如，指标值最高或者最低等条件。
具体地，服务器可以根据确定的最终相关性指标，从多个初始药品中确定指标值最高的一个或者多个初始药品为对应目标疾病的目标药品，并推荐至终端。
在本实施例中，继续参考图3，服务器可以根据计算的最终相关性指标，对多个初始药品进行排序，得到排序后的多个初始药品。然后服务器从排序后的各初始药品中筛选出对应目标疾病的目标药品。
上述实施例中，通过结合开药数据，并生成对应的权重指标，然后基于相关性指标以及权重指标，生成对应各初始药品的最终相关性指标并进行目标药品的确定以及推荐，从而可以使得药品的推荐结合实际的开药数据，可以提升药品推荐的准确性。
在其中一个实施例中，上述方法还可以包括：获取对数据库的更新数据，更新数据中包括目标疾病与各药品之间的对应关系；根据更新数据，检测是否存在对应目标疾病的新增药品；当检测到存在对应目标疾病的新增药品时，则对更新数据中目标疾病与新增药品之间的对应关系的出现频次进行统计，并在出现频次大于预设阈值时，基于目标疾病与新增药品之间的对应关系，对知识图谱进行更新。
其中，更新数据是指获取的线上实时开药数据。在本实施例中，服务器在获取到线上实时开药数据时，通过获取的线上实时开药数据，对数据库进行更新，例如，对某一疾病所使用的药品进行更新。
具体地，根据更新数据对数据库进行更新可以是指增加、删除或者是更改等，例如，对于某一疾病，增加新的药品，或者删除已经对应存在的药品，或者是更改已经对应存在的药品等。
在本实施例中，服务器也可以根据获取的更新数据，以进行实时检测，确定是否存在对应目标疾病的新增药品。
本领域技术人员可以理解的是，此处所述新增药品是指在已有开药历史中，未对应于该目标疾病的药品，即该新增药品未用于治疗该目标疾病。
在本实施例中，当服务器检测到更新数据中存在对应该目标疾病的新增药品时，则可以对该新增药品对应于该目标疾病的出现频次进行统计，例如，实时统计该新增药品应用于该目标疾病的次数。
进一步，服务器可以基于预设阈值对该新增药品对该出现频次进行判定，以确定对于该目标疾病该新增药品的用药方式是否达到统计学意义。
在本实施例中，当服务器确定目标疾病与新增药品之间的对应关系的出现频次大于预设阈值时，即确定该新增药品用于该目标疾病并非偶然原因时，则服务器可以确定该新增药品为治疗该目标疾病的新的药品，继续参考图3，服务器可以基于目标疾病与新增药品之间的对应关系，对知识图谱进行更新。
在本实施例中，服务器也可以对知识图谱中各药品以及对应的各疾病之间的关联关系进行实时统计，当确定某一药品长时间未用于治疗某一疾病时，也可以对知识图谱进行更新，使得得到的知识图谱更加准确。
上述实施例中，通过结合更新数据，对知识图谱进行更新，从而使得知识图谱集合线上实时开药数据生成，提升知识图谱的准确性，进而可以提升目标药品确定的准确性，以提升推荐的准确性。
在其中一个实施例中，上述方法还可以包括：将医疗数据、医学实体向量空间以及各相关性指标中的至少一个上传至区块链，并存储至区块链的节点中。
其中，区块链是指分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Block chain)，本质上是一个去中心化的数据库，是一串使用密码学方法相关联产生的数据块，每一个数据块中包含了一批次网络交易的信息，用于验证其信息的有效性(防伪)和生成下一个区块。
具体地，区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
在本实施例中，服务器可以将医疗数据、医学实体向量空间以及各相关性指标中的一个或者多个数据上传并存储于区块链的节点中，以保证数据的私密性和安全性。
上述实施例中，通过将医疗数据、医学实体向量空间以及各相关性指标中至少一个上传至区块链并存储于区块链的节点中，从而可以保障存储至区块链节点中数据的私密性，可以提升数据的安全性。
应该理解的是，虽然图2的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，图2中的至少一部分步骤可以包括多个子步骤或者多个阶段，这些子步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些子步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
在一个实施例中，如图4所示，提供了一种医疗数据处理装置，包括：待处理医疗数据获取模块100、查询模块200、相关性指标确定模块300和目标药品确定模块400，其中：
待处理医疗数据获取模块100，用于获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识。
查询模块200，用于基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量。
相关性指标确定模块300，拥有根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标。
目标药品确定模块400，用于获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
在其中一个实施例中，上述装置还可以包括：
标准向量空间生成模块，用于生成标准向量空间的生成。
在本实例中，标准向量空间生成模块可以包括：
知识图谱获取子模块，用于获取对应疾病、症状以及药品之间关联关系的知识图谱。
特征提取子模块，用于通过图神经网络模型对知识图谱进行特征提取，得到对应知识图谱中各疾病的疾病特征向量以及对应各药品的药品特征向量，各疾病特征向量以及各药品特征向量中包括对应的疾病、症状以及药品之间的关联关系。
在其中一个实施例中，知识图谱获取子模块可以包括：
医疗问诊数据获取单元，用于获取预设的医疗问诊数据。
数据提取单元，用于从线上医疗问诊数据中提取出与疾病、症状以及药品相关的初始目标数据。
标准化预处理单元，用于对初始目标数据进行标准化预处理，得到标准化预处理后的目标数据。
知识图谱建立单元，用于基于标准化预处理后的目标数据，建立疾病、症状以及药品之间关联关系的知识图谱。
在其中一个实施例中，标准化预处理单元可以包括：
医学标准数据库获取子单元，用于获取医学标准数据库。
待转换数据提取子单元，用于基于预设关键字，从初始目标数据中提取出待转换的待转换数据。
标准化预处理转换子单元，用于通过医学标准数据库对待转换数据进行标准化预处理转换，得到标准化预处理后的目标数据。
在其中一个实施例中，上述装置还可以包括：
开药数据获取模块，用于相关性指标确定模块300根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标之后，获取对应目标疾病的开药数据。
权重指标确定模块，用于根据开药数据，确定对应各初始药品的权重指标。
最终相关性指标确定模块，用于基于各初始药品的权重指标以及相关性指标，得到对应各初始药品的最终相关性指标。
在本实施例中，目标药品确定模块400用于获取满足预设条件的最终相关性指标所对应的初始药品为对应目标疾病的目标药品。
在其中一个实施例中，上述装置还可以包括：
更细数据获取模块，用于获取对数据库的更新数据，更新数据中包括目标疾病与各药品之间的对应关系。
检测模块，用于根据更新数据，检测是否存在对应目标疾病的新增药品。
知识图谱更新模块，用于当检测到存在对应目标疾病的新增药品时，则对更新数据中目标疾病与新增药品之间的对应关系的出现频次进行统计，并在出现频次大于预设阈值时，基于目标疾病与新增药品之间的对应关系，对知识图谱进行更新。
在其中一个实施例中，上述装置还可以包括：
存储模块，用于将医疗数据、医学实体向量空间以及各相关性指标中的至少一个上传至区块链，并存储至区块链的节点中。
关于医疗数据处理装置的具体限定可以参见上文中对于医疗数据处理方法的限定，在此不再赘述。上述医疗数据处理装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于计算机设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中，提供了一种计算机设备，该计算机设备可以是服务器，其内部结构图可以如图5所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的数据库用于存储医疗数据、医学实体向量空间以及各相关性指标等数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种医疗数据处理方法。
本领域技术人员可以理解，图5中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
在一个实施例中，提供了一种计算机设备，包括存储器和处理器，该存储器存储有计算机程序，该处理器执行计算机程序时实现以下步骤：获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识；基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中 各疾病以及各药品的特征向量；根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标；获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
在其中一个实施例中，处理器执行计算机程序时实现标准向量空间的生成方式可以包括：获取对应疾病、症状以及药品之间关联关系的知识图谱；通过图神经网络模型对知识图谱进行特征提取，得到对应知识图谱中各疾病的疾病特征向量以及对应各药品的药品特征向量，各疾病特征向量以及各药品特征向量中包括对应的疾病、症状以及药品之间的关联关系。
在其中一个实施例中，处理器执行计算机程序时实现获取对应疾病、症状以及药品之间关联关系的知识图谱，可以包括：获取预设的医疗问诊数据；从线上医疗问诊数据中提取出与疾病、症状以及药品相关的初始目标数据；对初始目标数据进行标准化预处理，得到标准化预处理后的目标数据；基于标准化预处理后的目标数据，建立疾病、症状以及药品之间关联关系的知识图谱。
在其中一个实施例中，处理器执行计算机程序时实现对初始目标数据进行标准化预处理，得到标准化预处理后的目标数据，可以包括：获取医学标准数据库；基于预设关键字，从初始目标数据中提取出待转换的待转换数据；通过医学标准数据库对待转换数据进行标准化预处理转换，得到标准化预处理后的目标数据。
在其中一个实施例中，处理器执行计算机程序时实现根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标之后，还可以实现以下步骤包括：获取对应目标疾病的开药数据；根据开药数据，确定对应各初始药品的权重指标；基于各初始药品的权重指标以及相关性指标，得到对应各初始药品的最终相关性指标。
在本实施例中，处理器执行计算机程序时实现获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品，可以包括：获取满足预设条件的最终相关性指标所对应的初始药品为对应目标疾病的目标药品。
在其中一个实施例中，处理器执行计算机程序时还可以实现以下步骤：获取对数据库的更新数据，更新数据中包括目标疾病与各药品之间的对应关系；根据更新数据，检测是否存在对应目标疾病的新增药品；当检测到存在对应目标疾病的新增药品时，则对更新数据中目标疾病与新增药品之间的对应关系的出现频次进行统计，并在出现频次大于预设阈值时，基于目标疾病与新增药品之间的对应关系，对知识图谱进行更新。
在其中一个实施例中，处理器执行计算机程序时还可以实现以下步骤：将医疗数据、医学实体向量空间以及各相关性指标中的至少一个上传至区块链，并存储至区块链的节点中。
在一个实施例中，提供了一种计算机可读存储介质，计算机可读存储介质可以是易失性的，也可以是非易失性的，其上存储有计算机程序，计算机程序被处理器执行时实现以下步骤：获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识；基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量；根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标；获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
在其中一个实施例中，计算机程序被处理器执行时实现标准向量空间的生成方式可以包括：获取对应疾病、症状以及药品之间关联关系的知识图谱；通过图神经网络模型对知识图谱进行特征提取，得到对应知识图谱中各疾病的疾病特征向量以及对应各药品的药品特征向量，各疾病特征向量以及各药品特征向量中包括对应的疾病、症状以及药品之间的关联关系。
在其中一个实施例中，计算机程序被处理器执行时实现获取对应疾病、症状以及药品之间关联关系的知识图谱，可以包括：获取预设的医疗问诊数据；从线上医疗问诊数据中提取出与疾病、症状以及药品相关的初始目标数据；对初始目标数据进行标准化预处理，得到标准化预处理后的目标数据；基于标准化预处理后的目标数据，建立疾病、症状以及药品之间关联关系的知识图谱。
在其中一个实施例中，计算机程序被处理器执行时实现对初始目标数据进行标准化预处理，得到标准化预处理后的目标数据，可以包括：获取医学标准数据库；基于预设关键字，从初始目标数据中提取出待转换的待转换数据；通过医学标准数据库对待转换数据进行标准化预处理转换，得到标准化预处理后的目标数据。
在其中一个实施例中，计算机程序被处理器执行时实现根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标之后，还可以实现以下步骤包括：获取对应目标疾病的开药数据；根据开药数据，确定对应各初始药品的权重指标；基于各初始药品的权重指标以及相关性指标，得到对应各初始药品的最终相关性指标。
在本实施例中，计算机程序被处理器执行时实现获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品，可以包括：获取满足预设条件的最终相关性指标所对应的初始药品为对应目标疾病的目标药品。
在其中一个实施例中，计算机程序被处理器执行时还可以实现以下步骤：获取对数据库的更新数据，更新数据中包括目标疾病与各药品之间的对应关系；根据更新数据，检测是否存在对应目标疾病的新增药品；当检测到存在对应目标疾病的新增药品时，则对更新数据中目标疾病与新增药品之间的对应关系的出现频次进行统计，并在出现频次大于预设阈值时，基于目标疾病与新增药品之间的对应关系，对知识图谱进行更新。
在其中一个实施例中，计算机程序被处理器执行时还可以实现以下步骤：将医疗数据、医学实体向量空间以及各相关性指标中的至少一个上传至区块链，并存储至区块链的节点中。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种医疗数据处理方法，其中，所述方法包括：
    获取待处理医疗数据，所述待处理医疗数据包括目标疾病的疾病标识；
    基于所述疾病标识，对标准向量空间进行查询，确定对应所述目标疾病的疾病特征向量以及对应所述目标疾病的多个初始药品的药品特征向量，所述标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，所述标准向量空间包括对应所述知识图谱中各疾病以及各药品的特征向量；
    根据所述疾病特征向量以及对应的多个药品特征向量，确定所述目标疾病与各所述初始药品的相关性指标；
    获取满足预设条件的相关性指标所对应的初始药品为对应所述目标疾病的目标药品。
  2. 根据权利要求1所述的方法，其中，所述标准向量空间的生成方式包括：
    获取对应疾病、症状以及药品之间关联关系的知识图谱；
    通过图神经网络模型对所述知识图谱进行特征提取，得到对应所述知识图谱中各疾病的疾病特征向量以及对应各药品的药品特征向量，各所述疾病特征向量以及各所述药品特征向量中包括对应的疾病、症状以及药品之间的关联关系。
  3. 根据权利要求2所述的方法，其中，所述获取对应疾病、症状以及药品之间关联关系的知识图谱，包括：
    获取预设的医疗问诊数据；
    从所述线上医疗问诊数据中提取出与疾病、症状以及药品相关的初始目标数据；
    对所述初始目标数据进行标准化预处理，得到标准化预处理后的目标数据；
    基于所述标准化预处理后的目标数据，建立疾病、症状以及药品之间关联关系的知识图谱。
  4. 根据权利要求3所述的方法，其中，所述对所述初始目标数据进行标准化预处理，得到标准化预处理后的目标数据，包括：
    获取医学标准数据库；
    基于预设关键字，从所述初始目标数据中提取出待转换的待转换数据；
    通过所述医学标准数据库对所述待转换数据进行标准化预处理转换，得到标准化预处理后的目标数据。
  5. 根据权利要求1至4任一项所述的方法，其中，所述根据所述疾病特征向量以及对应的多个药品特征向量，确定所述目标疾病与各所述初始药品的相关性指标之后，还包括：
    获取对应所述目标疾病的开药数据；
    根据所述开药数据，确定对应各所述初始药品的权重指标；
    基于各所述初始药品的权重指标以及相关性指标，得到对应各所述初始药品的最终相关性指标；
    所述获取满足预设条件的相关性指标所对应的初始药品为对应所述目标疾病的目标药品，包括：
    获取满足预设条件的最终相关性指标所对应的初始药品为对应所述目标疾病的目标药品。
  6. 根据权利要求1至4任一项所述的方法，其中，还包括：
    获取对数据库的更新数据，所述更新数据中包括目标疾病与各药品之间的对应关系；
    根据所述更新数据，检测是否存在对应所述目标疾病的新增药品；
    当检测到存在对应所述目标疾病的新增药品时，则对所述更新数据中所述目标疾病与所述新增药品之间的对应关系的出现频次进行统计，并在所述出现频次大于预设阈值时， 基于所述目标疾病与所述新增药品之间的对应关系，对所述知识图谱进行更新。
  7. 根据权利要求1至4任一项所述的方法，其中，所述方法还包括：
    将所述医疗数据、所述医学实体向量空间以及各所述相关性指标中的至少一个上传至区块链，并存储至区块链的节点中。
  8. 一种医疗数据处理装置，其中，所述装置包括：
    待处理医疗数据获取模块，用于获取待处理医疗数据，所述待处理医疗数据包括目标疾病的疾病标识；
    查询模块，用于基于所述疾病标识，对标准向量空间进行查询，确定对应所述目标疾病的疾病特征向量以及对应所述目标疾病的多个初始药品的药品特征向量，所述标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，所述标准向量空间包括对应所述知识图谱中各疾病以及各药品的特征向量；
    相关性指标确定模块，拥有根据所述疾病特征向量以及对应的多个药品特征向量，确定所述目标疾病与各所述初始药品的相关性指标；
    目标药品确定模块，用于获取满足预设条件的相关性指标所对应的初始药品为对应所述目标疾病的目标药品。
  9. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其中，所述处理器执行所述计算机程序时实现如下步骤：
    获取待处理医疗数据，待处理医疗数据包括目标疾病的疾病标识；
    基于疾病标识，对标准向量空间进行查询，确定对应目标疾病的疾病特征向量以及对应目标疾病的多个初始药品的药品特征向量，标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，标准向量空间包括对应知识图谱中各疾病以及各药品的特征向量；
    根据疾病特征向量以及对应的多个药品特征向量，确定目标疾病与各初始药品的相关性指标；
    获取满足预设条件的相关性指标所对应的初始药品为对应目标疾病的目标药品。
  10. 根据权利要求9所述的计算机设备，其中，所述标准向量空间的生成方式包括：
    获取对应疾病、症状以及药品之间关联关系的知识图谱；
    通过图神经网络模型对所述知识图谱进行特征提取，得到对应所述知识图谱中各疾病的疾病特征向量以及对应各药品的药品特征向量，各所述疾病特征向量以及各所述药品特征向量中包括对应的疾病、症状以及药品之间的关联关系。
  11. 根据权利要求10所述的计算机设备，其中，所述获取对应疾病、症状以及药品之间关联关系的知识图谱，包括：
    获取预设的医疗问诊数据；
    从所述线上医疗问诊数据中提取出与疾病、症状以及药品相关的初始目标数据；
    对所述初始目标数据进行标准化预处理，得到标准化预处理后的目标数据；
    基于所述标准化预处理后的目标数据，建立疾病、症状以及药品之间关联关系的知识图谱。
  12. 根据权利要求11所述的计算机设备，其中，所述对所述初始目标数据进行标准化预处理，得到标准化预处理后的目标数据，包括：
    获取医学标准数据库；
    基于预设关键字，从所述初始目标数据中提取出待转换的待转换数据；
    通过所述医学标准数据库对所述待转换数据进行标准化预处理转换，得到标准化预处理后的目标数据。
  13. 根据权利要求9至12任一项所述的计算机设备，其中，所述根据所述疾病特征向量以及对应的多个药品特征向量，确定所述目标疾病与各所述初始药品的相关性指标之 后，所述处理器执行所述计算机程序时还实现如下步骤：
    获取对应所述目标疾病的开药数据；
    根据所述开药数据，确定对应各所述初始药品的权重指标；
    基于各所述初始药品的权重指标以及相关性指标，得到对应各所述初始药品的最终相关性指标；
    所述获取满足预设条件的相关性指标所对应的初始药品为对应所述目标疾病的目标药品，包括：
    获取满足预设条件的最终相关性指标所对应的初始药品为对应所述目标疾病的目标药品。
  14. 根据权利要求9至12任一项所述的计算机设备，其中，所述处理器执行所述计算机程序时还实现如下步骤：
    获取对数据库的更新数据，所述更新数据中包括目标疾病与各药品之间的对应关系；
    根据所述更新数据，检测是否存在对应所述目标疾病的新增药品；
    当检测到存在对应所述目标疾病的新增药品时，则对所述更新数据中所述目标疾病与所述新增药品之间的对应关系的出现频次进行统计，并在所述出现频次大于预设阈值时，基于所述目标疾病与所述新增药品之间的对应关系，对所述知识图谱进行更新。
  15. 根据权利要求9至12任一项所述的计算机设备，其中，所述处理器执行所述计算机程序时还实现如下步骤：
    将所述医疗数据、所述医学实体向量空间以及各所述相关性指标中的至少一个上传至区块链，并存储至区块链的节点中。
  16. 一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现如下步骤：
    获取待处理医疗数据，所述待处理医疗数据包括目标疾病的疾病标识；
    基于所述疾病标识，对标准向量空间进行查询，确定对应所述目标疾病的疾病特征向量以及对应所述目标疾病的多个初始药品的药品特征向量，所述标准向量空间基于疾病、症状以及药品之间关联关系的知识图谱生成，所述标准向量空间包括对应所述知识图谱中各疾病以及各药品的特征向量；
    根据所述疾病特征向量以及对应的多个药品特征向量，确定所述目标疾病与各所述初始药品的相关性指标；
    获取满足预设条件的相关性指标所对应的初始药品为对应所述目标疾病的目标药品。
  17. 根据权利要求16所述的计算机可读存储介质，其中，所述标准向量空间的生成方式包括：
    获取对应疾病、症状以及药品之间关联关系的知识图谱；
    通过图神经网络模型对所述知识图谱进行特征提取，得到对应所述知识图谱中各疾病的疾病特征向量以及对应各药品的药品特征向量，各所述疾病特征向量以及各所述药品特征向量中包括对应的疾病、症状以及药品之间的关联关系。
  18. 根据权利要求17所述的计算机可读存储介质，其中，所述获取对应疾病、症状以及药品之间关联关系的知识图谱，包括：
    获取预设的医疗问诊数据；
    从所述线上医疗问诊数据中提取出与疾病、症状以及药品相关的初始目标数据；
    对所述初始目标数据进行标准化预处理，得到标准化预处理后的目标数据；
    基于所述标准化预处理后的目标数据，建立疾病、症状以及药品之间关联关系的知识图谱。
  19. 根据权利要求18所述的计算机可读存储介质，其中，所述对所述初始目标数据进行标准化预处理，得到标准化预处理后的目标数据，包括：
    获取医学标准数据库；
    基于预设关键字，从所述初始目标数据中提取出待转换的待转换数据；
    通过所述医学标准数据库对所述待转换数据进行标准化预处理转换，得到标准化预处理后的目标数据。
  20. 根据权利要求16至19任一项所述的计算机可读存储介质，其中，所述根据所述疾病特征向量以及对应的多个药品特征向量，确定所述目标疾病与各所述初始药品的相关性指标之后，所述计算机程序被处理器执行时还实现如下步骤：
    获取对应所述目标疾病的开药数据；
    根据所述开药数据，确定对应各所述初始药品的权重指标；
    基于各所述初始药品的权重指标以及相关性指标，得到对应各所述初始药品的最终相关性指标；
    所述获取满足预设条件的相关性指标所对应的初始药品为对应所述目标疾病的目标药品，包括：
    获取满足预设条件的最终相关性指标所对应的初始药品为对应所述目标疾病的目标药品。
PCT/CN2021/084350 2020-09-23 2021-03-31 医疗数据处理方法、装置、计算机设备和存储介质 WO2022062353A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011009638.3 2020-09-23
CN202011009638.3A CN112151141A (zh) 2020-09-23 2020-09-23 医疗数据处理方法、装置、计算机设备和存储介质

Publications (1)

Publication Number Publication Date
WO2022062353A1 true WO2022062353A1 (zh) 2022-03-31

Family

ID=73898013

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/084350 WO2022062353A1 (zh) 2020-09-23 2021-03-31 医疗数据处理方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN112151141A (zh)
WO (1) WO2022062353A1 (zh)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112151141A (zh) * 2020-09-23 2020-12-29 康键信息技术（深圳）有限公司 医疗数据处理方法、装置、计算机设备和存储介质
CN112951446A (zh) * 2021-04-16 2021-06-11 平安科技（深圳）有限公司 基于医药图谱的药品查询方法、装置、设备及存储介质
CN113434692B (zh) * 2021-06-22 2023-08-01 上海交通大学医学院附属仁济医院 图神经网络模型构建、诊疗方案推荐方法、系统及设备
CN113779274B (zh) * 2021-09-18 2024-04-05 深圳平安医疗健康科技服务有限公司 指标模拟仿真方法、装置、计算机设备及存储介质
CN114792571B (zh) * 2022-03-09 2023-03-07 广州方舟信息科技有限公司 药品信息推送方法、装置、服务器及计算机可读存储介质
CN114974501A (zh) * 2022-06-16 2022-08-30 康键信息技术（深圳）有限公司 基于人工智能的药品推荐方法、装置、设备及存储介质
CN116108906A (zh) * 2023-04-06 2023-05-12 北京亚信数据有限公司 疾病药品关系映射模型训练及相关推荐、检测方法和装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109658208A (zh) * 2019-01-15 2019-04-19 京东方科技集团股份有限公司 药品的推荐方法、装置、介质和电子设备
CN110223751A (zh) * 2019-05-16 2019-09-10 平安科技（深圳）有限公司 基于医疗知识图谱的处方评价方法、系统及计算机设备
CN110428910A (zh) * 2019-06-18 2019-11-08 浙江大学 临床用药适应症分析系统、方法、计算机设备和存储介质
CN110752035A (zh) * 2019-09-06 2020-02-04 深圳壹账通智能科技有限公司 健康数据处理方法、装置、计算机设备及存储介质
US20200245913A1 (en) * 2018-11-29 2020-08-06 January, Inc. Systems, methods, and devices for biophysical modeling and response prediction
CN111639190A (zh) * 2020-04-30 2020-09-08 南京理工大学 医疗知识图谱构建方法
CN112151141A (zh) * 2020-09-23 2020-12-29 康键信息技术（深圳）有限公司 医疗数据处理方法、装置、计算机设备和存储介质

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200245913A1 (en) * 2018-11-29 2020-08-06 January, Inc. Systems, methods, and devices for biophysical modeling and response prediction
CN109658208A (zh) * 2019-01-15 2019-04-19 京东方科技集团股份有限公司 药品的推荐方法、装置、介质和电子设备
CN110223751A (zh) * 2019-05-16 2019-09-10 平安科技（深圳）有限公司 基于医疗知识图谱的处方评价方法、系统及计算机设备
CN110428910A (zh) * 2019-06-18 2019-11-08 浙江大学 临床用药适应症分析系统、方法、计算机设备和存储介质
CN110752035A (zh) * 2019-09-06 2020-02-04 深圳壹账通智能科技有限公司 健康数据处理方法、装置、计算机设备及存储介质
CN111639190A (zh) * 2020-04-30 2020-09-08 南京理工大学 医疗知识图谱构建方法
CN112151141A (zh) * 2020-09-23 2020-12-29 康键信息技术（深圳）有限公司 医疗数据处理方法、装置、计算机设备和存储介质

Also Published As

Publication number Publication date
CN112151141A (zh) 2020-12-29

Similar Documents

Publication Publication Date Title
WO2022062353A1 (zh) 医疗数据处理方法、装置、计算机设备和存储介质
WO2021073277A1 (zh) 一种个性化精准用药推荐方法及装置
CN111696675B (zh) 基于物联网数据的用户数据分类方法、装置及计算机设备
WO2020034873A1 (zh) 数据标准化处理方法、装置、计算机设备和存储介质
CN109147198B (zh) 自助售药方法和系统
US20200050949A1 (en) Digital assistant platform
WO2021159759A1 (zh) 电子病历结构化方法、装置、计算机设备和存储介质
CN112016318B (zh) 基于解释模型的分诊信息推荐方法、装置、设备及介质
WO2021121187A1 (zh) 基于分词文本的电子病例查重方法、装置、计算机设备
CN112035741B (zh) 基于用户体检数据的预约方法、装置、设备及存储介质
WO2019060882A1 (en) CLINICAL DECISION-MAKING MANAGEMENT SYSTEM FOR ENDOCRINOLOGY OF REPRODUCTION AND INFERTILITY
WO2021151327A1 (zh) 分诊数据处理方法、装置、设备及介质
US20220044809A1 (en) Systems and methods for using deep learning to generate acuity scores for critically ill or injured patients
US20160188834A1 (en) Determination of patient-appropriate post-acute care settings
CN110837565B (zh) 实现药品推荐的模型训练装置及计算机设备
CN111882005B (zh) 数据类别确定方法及装置、终端设备、存储介质
CN112447300A (zh) 基于图神经网络的医疗查询方法、装置、计算机设备及存储介质
CN112035611A (zh) 目标用户推荐方法、装置、计算机设备和存储介质
CN112017745A (zh) 决策信息推荐、药物信息推荐方法、装置、设备及介质
CN112447270A (zh) 一种用药推荐方法、装置、设备及存储介质
WO2020248844A1 (zh) 测试对象的寿命预估方法、装置、设备及介质
WO2020132918A1 (zh) 药品预测方法、装置、计算机设备及存储介质
CN111933302A (zh) 药物推荐方法、装置、计算机设备及存储介质
CN111221853A (zh) 药品清单异常检测方法、装置、计算机设备和存储介质
CN111147474A (zh) 数据加密传输方法、装置、计算机设备和存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21870742

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 28/06/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 21870742

Country of ref document: EP

Kind code of ref document: A1