A data system based on medical information

Abstract

The application relates to a medical information-based data system, comprising a medical data processing module configured to acquire medical data and preprocess the medical data to obtain a data format with a patient admission identifier as the core; a medical algorithm model layer configured to take the preprocessed medical data as input, extract examination information, image information, medication information, surgery information and symptom information in the medical data through corresponding algorithm models respectively, and take the information as input of a cache processing module; and an algorithm fusion module configured to disassemble data requirements obtained into combination categories formed by data types and business requirements, determine matched algorithm models according to the combination categories and a preset algorithm model routing table, call the cache processing module to obtain data processed by the selected algorithm models, determine data to be output according to the business requirements, and output the data after splicing.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a data system based on medical information. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] There are many types of medical information, including patient identity information, diagnosis and treatment information, symptom information, and drug information, etc. Multiple types are usually stored in the system in the form of text or images. Due to the different sensitivity, data type and purpose of various information, they are usually distributed in different subsystems, and the information format in each subsystem is different. Each type of information is suitable for different algorithms, and each algorithm produces different results in medical information, making each subsystem relatively independent, causing information island problem, resulting in the need to call multiple subsystems when applying information, making the server space occupy too high, and the running efficiency of the data system is reduced. SUMMARY

[0004] In order to solve the technical problems existing in the background art, the present application provides a data system based on medical information, which pre-processes medical information to obtain a standardized data format with patient admission identification as the core, extracts the required information according to the data type using the corresponding algorithm model, and outputs the information processed by the algorithm model after splicing according to the requirements received by the data system. According to the business requirements and medical data types, different algorithm models can be called for processing, thereby solving the information island problem, reducing the server space occupation of data, and reducing the complexity of data cleaning.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0006] The first aspect of the present application provides a data system based on medical information, comprising:

[0007] The medical data processing module is configured to obtain medical data and pre-process it to obtain a data format with patient admission identification as the core;

[0008] The medical algorithm model layer is configured to take the pre-processed medical data as input, extract the test information, image information, medication information, surgery information and symptom information in the medical data through the corresponding algorithm model, and take them as the input of the cache processing module;

[0009] The algorithm fusion module is configured to: according to the obtained data requirements, disassemble a combination category formed by data types and business requirements, determine a matched algorithm model according to the combination category and a preset algorithm model routing table, and call the cache processing module to obtain data processed by the selected algorithm model, determine data to be output according to the business requirements, and output the data after splicing.

[0010] The preprocessing includes:

[0011] Adapting to different types of databases and establishing connections between databases;

[0012] According to the type of data, filtering out useless information in the data based on regular expressions;

[0013] According to the standard reference table, convert the output into standardized terms and associate them with the patient admission identifier to obtain standardized processed data.

[0014] Obtain a data format with the patient admission identifier as the core, specifically: query the test information, image information, medication information, surgery information and symptom information through the patient admission identifier, and associate the queried information with the admission identifier as the core.

[0015] The medical algorithm model layer takes the preprocessed medical data as input, extracts the test information, image information, medication information, surgery information and symptom information in the medical data through corresponding algorithm models, and takes the extracted information as the input of the cache processing module; specifically:

[0016] Based on the recognition model, the disease information contained in the image information is obtained;

[0017] Based on the alignment algorithm, the test information of the patient is parsed;

[0018] Based on natural language processing and named entity recognition, the patient symptom information contained in the medical record is parsed;

[0019] Based on regular expression matching, the patient medication and surgery information is parsed.

[0020] The cache processing module temporarily stores the processing results of the algorithm model, specifically:

[0021] Receive data processed by each algorithm model;

[0022] According to the type of medical data, select a cache storage medium;

[0023] According to the usage frequency of the medical data, set the cache expiration time.

[0024] According to the obtained data requirements, the combination categories formed by the data types and the business requirements are disassembled; specifically, a coordinate system is established, the horizontal coordinate represents the data type, the vertical coordinate represents the business requirement type, and the point formed by the coordinate values represents the combination category formed by the data type and the business requirement.

[0025] According to the combination category and the preset algorithm model routing table, a matched algorithm model is determined; specifically, according to the obtained combination category and the preset algorithm model routing table, common information and requirement points are obtained by classifying the business requirements, a matched algorithm model is selected according to the common information and the corresponding requirement points, and a matched algorithm is determined.

[0026] The data processed by the selected algorithm model is obtained by calling the cache processing module, and when the required data does not exist in the cache processing module, the data is generated by calling the algorithm model layer.

[0027] The security management module further includes a desensitization service submodule and a verification service submodule, and the data output by the algorithm model is authenticated and tamper-proof verified by the verification service submodule; before the medical data module provides data to the algorithm model, the data is desensitized by the desensitization service submodule.

[0028] Compared with the prior art, the above one or more technical solutions have the following beneficial effects:

[0029] 1. The medical information is preprocessed to obtain a standardized data format with a patient admission identifier as the core, the required information is extracted by using a corresponding algorithm model according to the data type, and the information processed by the algorithm model is output after being spliced according to the requirements received by the data system, different algorithm models can be called for processing according to the business requirements and the medical data types, thereby solving the information island problem, reducing the data occupation in the server space, and improving the operation efficiency of the data system.

[0030] 2. The cache data processing module can select different storage media according to different types of medical data, the various structured, semi-structured and unstructured data such as image data, examination data and medical order data involved in the medical industry correspond to different storage media, the structured data and the unstructured data are stored separately, the structured data is stored in a relational storage medium, and the data value can be mined by using a large number of mature analysis tools.

[0031] 3. The algorithm fusion module can call different algorithm models for processing of medical data according to business requirements and medical data types, and can fully play the advantages of various algorithms.

[0032] 4. The security management module can ensure the security of medical data, enhance the protection of patient privacy data, and perform authentication and tamper-proof verification before outputting data using an algorithm model to avoid unauthorized application use. Before providing data to the algorithm model, the medical data module calls the desensitization service for desensitization to avoid patient privacy data entering the model, isolate patient privacy data from the algorithm model, and avoid patient privacy data leakage. BRIEF DESCRIPTION OF DRAWINGS

[0033] The accompanying drawings, which form a part of this description, are included to provide further understanding of the application, and are incorporated in and constitute a part of this application. The illustrative embodiments of the application and their description serve to explain the application. It can not specifically set forth all embodiments of the application as upon such development. Details can vary substantially without departing from the basic principles of the application.

[0034] Fig. 1 is a schematic diagram of the system function architecture provided by one or more embodiments of the application;

[0035] Fig. 2 is a workflow diagram of the algorithm fusion module provided by one or more embodiments of the application. DETAILED DESCRIPTION

[0036] The application will be further described with reference to the drawings and examples.

[0037] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0038] It should be noted that the terms used herein are only intended to describe specific embodiments and are not intended to limit exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should be understood that when the terms "comprise" and / or "include" are used in the specification, they refer to the presence of a feature, step, operation, device, component and / or combination thereof.

[0039] There are various algorithms for processing different types of data in the prior art of medical information, such as machine learning (ML), robotic process automation (RPA), computer vision (CV) and natural language processing (NLP), etc. Different types of medical information can obtain the required results through different algorithms.

[0040] For example, patient identity information can be extracted from medical record data to obtain age, gender, and other information through natural language processing, and patient medical history, current medical history, and other information can also be obtained.

[0041] A single algorithm can only play a role in a single type of medical information, and the information processing in the system needs to aggregate the results of algorithm processing of all medical information of the patient from admission to discharge. The inventors found that in the current data system for medical information, data is saved in multiple subsystems according to different data types and uses, the information formats in the subsystems are different, the algorithms suitable for each type of information are inconsistent, and the results generated by each algorithm in the medical information are also inconsistent, making each subsystem relatively independent, resulting in the need to call multiple subsystems when applying information, resulting in high server space occupation and high complexity of data cleaning.

[0042] Therefore, the following embodiments give a data system based on medical information, which pre-processes medical information to obtain a standardized data format with patient admission identification as the core, extracts the required information according to the data type using the corresponding algorithm model, and outputs the information processed by the algorithm model after splicing according to the requirements received by the data system. Different algorithm models can be called for processing according to business requirements and medical data types, thereby solving the information island problem, reducing the server space occupation of data, and reducing the complexity of data cleaning.

[0043] Embodiment one:

[0044] As shown in Figs. 1-2 A data system based on medical information includes:

[0045] The medical data processing module is configured to obtain and preprocess medical data to obtain a data format with patient admission identification as the core;

[0046] The medical algorithm model layer is configured to take the preprocessed medical data as input, extract the test information, image information, medication information, surgery information and symptom information in the medical data through the corresponding algorithm model, and take them as the input of the cache processing module;

[0047] The algorithm fusion module is configured to: according to the obtained data requirements, decompose into combination categories formed by data types and business requirements, determine the matching algorithm model according to the combination category and the preset algorithm model routing table, and call the cache processing module to obtain the data processed by the selected algorithm model; according to the combination category, determine the data to be associated, and output the data associated with the patient admission identification.

[0048] Specifically:

[0049] The medical data processing module obtains various data in the medical system, and after standardization processing, forms a data form that is easy to process and centered on the patient; specifically:

[0050] I. Adapt to different types of databases;

[0051] Obtain the type information of the database, and select a matching connection mode based on the connection mode package according to the obtained database type information, establish a connection between the databases, and complete the adaptation.

[0052] Various data in the medical system are usually saved in the database of the body or in the cloud platform, and the embodiment establishes data communication between the local database and the cloud platform.

[0053] I. Clean various types of medical text data;

[0054] Filter out useless information in the data based on regular expressions, for example, the zip code of the patient's location, special symbols that cannot be recognized, etc.

[0055] According to the type of data, the regular expression is adjusted multiple times for filtering until it meets the requirements of the medical algorithm model layer interface.

[0056] III. Standardize various types of medical text data formats; specifically:

[0057] Loop input different types of medical data, read the medical term standard comparison table matched with the input type;

[0058] According to the standard comparison table, convert and output standardized terms and associate them with the patient's admission identifier to obtain standardized processed data.

[0059] IV. Establish a data form centered on the patient's admission identifier;

[0060] Through the patient's admission identifier, query the test information, image information, medication information, surgery information, and symptom information, and after querying, associate (assemble) the queried information with the admission identifier as the core.

[0061] The medical algorithm model layer receives the processed standardized data, loads multiple types of algorithm models, and processes different medical texts, and the processed medical texts are input to the cache processing module, specifically:

[0062] I. Receive the standardized data centered on the patient's admission identifier;

[0063] II. Process the data;

[0064] The image conclusion information uses LSTM+CRF (an existing recognition model) to identify the disease information contained in the conclusion.

[0065] The test information uses galechurch (an alignment algorithm) for information alignment to parse the patient's standard test item information.

[0066] Medical record information uses NLP+NER (natural language processing + named entity recognition) for entity recognition to parse the patient symptom information contained in the medical record.

[0067] Medication + surgery information uses regular expressions for matching to parse patient medication and surgery information.

[0068] In this embodiment, the algorithm corresponding to each data type is pre-set.

[0069] III outputs the processed data as the input of the cache processing module;

[0070] Using different algorithm models, disease information, test item information, symptom information, medication information and surgery information are extracted from the data centered on the patient admission identifier as output. This part of information is associated with the patient admission identifier and can be queried according to the patient admission identifier.

[0071] The cache processing module temporarily stores the processing results of the algorithm model, specifically:

[0072] I receives the data processed by each algorithm model;

[0073] II selects a cache storage medium according to the medical data type;

[0074] There are some objective selection conditions for medical data. Disease data, test data and medication information belong to structured data, which will be stored in a relational database. Medical orders, examinations and symptom information belong to unstructured data, which will be stored in a document type database. In the medical field, the type of this part of data is fixed.

[0075] III sets the cache expiration time according to the frequency of use of medical data.

[0076] The algorithm fusion module receives the requirements of the medical system and disassembles it. According to the requirements, it calls the data processed by each algorithm model temporarily stored in the cache processing module, and fuses the medical text processed by each algorithm model through the algorithm fusion module to output the fused data, thereby forming an intelligent portrait centered on the patient; Specifically:

[0077] I receives the medical system request;

[0078] II disassembles the business requirements in the medical system request to obtain the required data type. The data type and business requirements form a combination category;

[0079] The embodiment adopts quadrant decomposition method, the horizontal coordinate represents data type (inspection information A, examination information B, medication information C, operation information D, symptom information E), the vertical coordinate represents business requirement type (inspection requirement 01, examination requirement 02, medication requirement 03, operation requirement 04, symptom requirement 05), and the combination category formed by the data type and the business requirement is represented by the coordinate value of the point in the coordinate system.

[0080] For example, the inspection information contains medication requirement, forming A03 combination category, and the operation information contains symptom requirement, forming D05 combination category.

[0081] III. According to the obtained combination category and the acquired algorithm model routing table, the similar requirement points are classified, the common information and requirement points are found out, the matching algorithm model is selected according to the common information and the corresponding requirement points, and thus the most suitable algorithm for the information and requirement is selected.

[0082] In the embodiment, the algorithm model routing table is a self-defined maintenance comparison table. For example, the combination category is D05 combination category, indicating that the current combination category contains structured data operation information and unstructured data symptom information, and the comparison table returns operation information parsed by using regular expression and symptom information parsed by using natural language processing and named entity recognition.

[0083] In the embodiment, the common information refers to the type of data, for example, the part of information in the combination category belonging to structured data.

[0084] In the embodiment, the requirement point refers to the output information meeting the business requirement.

[0085] In the embodiment, when the algorithm model is selected, the experience table is selected according to the setting, and if the experience table cannot be acquired, a prompt is popped up, and the required algorithm model is selected by manual intervention.

[0086] IV. The cache processing module is called to acquire the data processed by the algorithm model.

[0087] When the required data does not exist in the cache processing module, the data is generated by calling the algorithm model layer.

[0088] V. According to the combination category formed by the data type and the business requirement, the business requirement data splicing list is acquired, and the data to be spliced (associated) is determined.

[0089] VI. The data processed by the splicing (association) algorithm module is spliced (associated).

[0090] VII. The spliced (associated) data is output.

[0091] In order to ensure the safety of medical data and enhance the protection of patient privacy data, the security management module is added in the embodiment, which includes two sub-modules of desensitization service and verification service. The data output by the algorithm model is authenticated and tamper-proof verified before use, so as to avoid unauthorized application. The desensitization service is called before the medical data module provides data for the algorithm model, so as to avoid patient privacy data from entering the model, isolate patient privacy data from the algorithm model, and avoid patient privacy data leakage.

[0092] The verification service submodule is configured to:

[0093] I. Receive the verification information transmitted by the algorithm fusion module;

[0094] II. Call the verification service for authentication and tamper-proof verification.

[0095] The desensitization service submodule is configured to:

[0096] I. Receive medical data;

[0097] II. Process patient privacy information;

[0098] III. Output desensitized medical data.

[0099] The above system solves the problems of inconsistent storage media of different types of medical systems, large differences between different types of data, and high degree of data discretization. With standardization, the medical data text is unified, the inconsistent informationization manufacturers in current hospitals are solved, the relative independence of each system is solved, the algorithm is unified, the information island problem is solved, the relatively unified data type is provided for the algorithm model, and the complexity of data cleaning is reduced.

[0100] The cache data processing module can select different storage media according to different types of medical data. The current artificial intelligence platform in the medical industry does not consider selecting different storage media according to the type of medical data to store the algorithm processing result. The PACS image data, examination data, medical order data and various structured, semi-structured and unstructured data involved in the medical industry need to be saved in different storage media. According to the type of data, different types of storage media are selected, which can prevent server storage space abuse, reduce storage cost, separate structured data and unstructured data, and store structured data in a relational storage medium. This type of data can be analyzed by using a large number of mature analysis tools to realize data insight and data value mining.

[0101] The algorithm fusion module calls different models according to business requirements and medical data types to process medical data, which can fully play the advantages of various artificial intelligence algorithms.

[0102] The above merely provides the preferred embodiments of the present application, and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the principles and technical scope of the present application shall fall into the scope of the present application.

Claims

1. A data system based on medical information, characterized in that, Comprise: medical data processing module, configured to: obtain medical data and pre-processing, get the data format with patient admission identification as the core; medical algorithm model layer, configured to: with pre-processed medical data as input, through the corresponding algorithm model respectively extracts the test information, image information, medication information, surgery information and symptom information in the medical data, as the input of the cache processing module; algorithm fusion module, configured to: according to the data requirements obtained, disassembled into data type and business demand formed combination category, according to the combination category and the preset algorithm model routing table, determine the matching algorithm model, and call the cache processing module to obtain the data processed by the selected algorithm model, according to the business requirements to determine the data to be output, after splicing output; Wherein, the medical algorithm model layer takes the pre-processed medical data as input, and extracts the test information, image information, medication information, surgery information and symptom information in the medical data through the corresponding algorithm model, as the input of the cache processing module; comprising: based on the recognition model to obtain the disease information contained in the image information; based on the alignment algorithm, the test information of the patient is parsed; based on natural language processing and named entity recognition, the patient's symptom information contained in the medical record is parsed; based on regular expression matching, the patient's medication and surgery information is parsed; Wherein, according to the combination category and the preset algorithm model routing table, determine the matching algorithm model; Specifically: according to the combination category obtained and the preset algorithm model routing table, the business requirements are classified to get common information and demand points, and the matching algorithm model is selected according to the common information and the corresponding demand points, and the matching algorithm model is determined.

2. A data system based on medical information as claimed in claim 1, characterized in that, The pre-processing, comprising: adapt to different types of database and establish the connection between the databases; based on regular expression, filter out the useless information in the data according to the type of data; according to the standard comparison table, convert the output to standardized terminology and associate with the patient admission identification, get the standardized processed data.

3. A data system based on medical information according to claim 1, characterized in that, Get the data format with patient admission identification as the core, specifically: through the patient admission identification, query the test information, image information, medication information, surgery information and symptom information, and the information obtained by core correlation query.

4. A data system based on medical information according to claim 1, characterized in that, The cache processing module temporarily stores the processing results of the algorithm model, specifically: receive the data processed by each algorithm model; according to the type of medical data, select the cache storage medium; according to the use frequency of medical data, set the cache expiration time.

5. A data system based on medical information according to claim 1, characterized in that, According to the data requirements obtained, disassembled into data type and business demand formed combination category; Specifically: establish a coordinate system, the horizontal coordinate represents the data type, the vertical coordinate represents the business demand type, and the point formed by the coordinate value represents the combination category formed by the data type and business demand.

6. A data system based on medical information according to claim 1, characterized in that, Call the cache processing module to obtain the data processed by the selected algorithm model, when the required data does not exist in the cache processing module, generate data by calling the algorithm model.

7. A data system based on medical information according to claim 1, characterized in that, Also has the security management module, including the desensitization service sub module and the verification service sub module.

8. A data system based on medical information according to claim 7, characterized in that, The data output by the algorithm model is authenticated and tamper-proof verified by the verification service submodule in advance; and the medical data module is desensitized by the desensitization service submodule in advance before being provided to the algorithm model.

Images