Medical data conversion method and device, equipment and medium

Abstract

The invention discloses a medical data conversion method and device, equipment and a medium. The method comprises the following steps: acquiring original medical data to be converted; extracting semantic granularity data in the original medical data to obtain a semantic data set; obtaining a resource parameter set contained in the fast medical operation resources; and based on the resource parameter set, performing mapping conversion on the semantic granularity data in the semantic data set to obtain target medical data.

Description

Technical Field

[0001] The present application relates to the field of medical data processing technology, and in particular to a medical data conversion method, device, equipment and medium. Background Art

[0002] In actual applications, different medical institutions and medical systems adopt different medical data management rules, which makes the structure and format of medical data corresponding to different medical institutions and medical systems different. However, the above management rules of medical data cannot meet the needs of medical information business interconnection and data intercommunication between different medical structures and medical systems. Summary of the invention

[0003] Based on the above technical problems, the embodiments of the present application provide a medical data conversion method, device, equipment and medium.

[0004] The technical solution provided by the embodiment of the present application is as follows:

[0005] The present application provides a method for converting medical data, the method comprising:

[0006] Obtaining raw medical data to be converted;

[0007] Extracting semantic granular data from the original medical data to obtain a semantic data set;

[0008] Get the resource parameter set contained in Fast Healthcare Interoperability Resources (FHIR);

[0009] Based on the resource parameter set, the semantic granular data in the semantic data set is mapped and converted to obtain target medical data.

[0010] In some embodiments, extracting the semantic granularity data from the original medical data to obtain a semantic data set includes:

[0011] Acquire medical object metadata associated with the original medical data;

[0012] Segmenting the original medical data based on the medical object metadata to obtain medical object attribute data;

[0013] The medical object attribute data is semantically segmented at a basic semantic unit granularity to obtain the semantic data set.

[0014] In some embodiments, mapping and converting the semantic granularity data in the semantic data set based on the resource parameter set to obtain target medical data includes:

[0015] Associatively identifying the semantic granular data and at least part of the resource parameters in the resource parameter set through a graphic mapping engine, and obtaining a mapping association relationship between the semantic granular data and at least part of the resource parameters;

[0016] Based on the mapping association relationship, the semantic granular data in the semantic data set is mapped and transformed to obtain the target medical data.

[0017] In some embodiments, mapping and converting the semantic granularity data in the semantic data set based on the mapping association relationship to obtain the target medical data includes:

[0018] determining a data structure of at least a portion of the resource parameters;

[0019] Through the trained conversion model, based on the mapping association relationship and the data structure, the semantic granularity data is converted to obtain the target medical data.

[0020] In some embodiments, the method further comprises:

[0021] Acquire sample data; wherein the sample data includes multiple types of initial medical data;

[0022] Acquire a sample mapping relationship between the sample data and the resource parameter set;

[0023] The sample data is converted based on the sample mapping relationship by using a conversion model to obtain a sample conversion result;

[0024] Acquire abnormal data contained in the sample conversion result;

[0025] The model parameters of the conversion model are adjusted based on the abnormal data, and the trained conversion model is obtained according to the corrected model parameters.

[0026] In some embodiments, adjusting the model parameters of the conversion model based on the abnormal data includes:

[0027] Obtaining k-th abnormal data; wherein the k-th abnormal data is abnormal data contained in the k-th conversion result output by the conversion model after the k-th parameter adjustment; k is an integer greater than or equal to 1;

[0028] Determine a kth abnormal type corresponding to the kth abnormal data;

[0029] Determine that data associated with the kth abnormal data in the sample data is the k+1th training data;

[0030] Based on the k+1th training data and the kth abnormality type, the model parameters of the conversion model after the kth parameter adjustment are adjusted to obtain the conversion model after the k+1th parameter adjustment.

[0031] In some embodiments, determining the kth abnormality type corresponding to the kth abnormal data includes:

[0032] Obtaining a set of exception types of the rapid medical operational resource;

[0033] Obtain kth abnormal description information of the kth abnormal data;

[0034] An exception type in the exception type set that matches the kth exception description information is determined to be the kth exception type.

[0035] In some embodiments, the obtaining of abnormal data included in the sample conversion result includes:

[0036] Obtain at least one configuration file;

[0037] The sample conversion result is detected based on the at least one configuration file, and data in the sample conversion result that does not meet the configuration rule included in the at least one configuration file is determined as the abnormal data.

[0038] In some embodiments, adjusting the model parameters of the conversion model based on the abnormal data, and obtaining the trained conversion model according to the adjusted model parameters, includes:

[0039] Obtaining the k+1th conversion result output by the conversion model after the k+1th parameter adjustment; wherein k is an integer greater than or equal to 1;

[0040] If the amount of the k+1th abnormal data included in the k+1th conversion result is less than or equal to the target threshold, it is determined that the conversion model after the k+1th parameter adjustment is the trained conversion model.

[0041] The present application also provides a medical data conversion device, the device comprising:

[0042] An acquisition module, used for acquiring original medical data to be converted;

[0043] An extraction module, used for extracting semantic granularity data from the original medical data to obtain a semantic data set;

[0044] The acquisition module is further used to acquire a resource parameter set included in the rapid medical operational resource;

[0045] A conversion module is used to map and convert the semantic granular data in the semantic data set based on the resource parameter set to obtain target medical data.

[0046] An embodiment of the present application also provides an electronic device, which includes a processor and a memory; a computer program is stored in the memory; when the computer program is executed by the processor, it can implement any of the medical data conversion methods described above.

[0047] An embodiment of the present application also provides a computer-readable storage medium, in which a computer program is stored; when the computer program is executed by a processor of an electronic device, the medical data conversion method as described above can be implemented.

[0048] The medical data conversion method provided in the embodiment of the present application obtains the original medical data to be converted, and obtains the semantic data set by extracting the semantic granularity data in the original medical data, so that the semantic data set can not only accurately reflect the medical information contained in the original medical data from the aspects of natural semantics or medical semantics, but also enables the semantic data set to accurately reflect the medical semantics represented by the original medical data in a causal or causal manner from the semantic granularity; and obtains the resource parameter set contained in FHIR, and based on the resource parameter set, maps and converts the semantic granularity data in the semantic data set, thereby realizing the control of the mapping and conversion direction of the semantic granularity data in the semantic data set. , and thus can realize the targeted mapping and conversion of the semantic granular data in the semantic data set; on the other hand, by mapping and converting the semantic granular data in the semantic data set based on the resource parameter set, the target medical data is obtained, so that the target medical data can not only comprehensively contain the medical objects and medical semantics represented by the original medical data, but also make the data format of the target medical data consistent with the data format of FHIR, thereby realizing the targeted, unified, automated and intelligent conversion of original medical data of any type and any source, and thus can meet the unified conversion requirements of multi-source heterogeneous medical data from different medical systems or medical institutions in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A flowchart of a medical data conversion method provided in an embodiment of the present application;

[0050] Figure 2 A schematic diagram of a process for obtaining a conversion model after the k+1th parameter adjustment provided in an embodiment of the present application;

[0051] Figure 3 A schematic diagram of the process of training the conversion model provided in the embodiment of the present application;

[0052] Figure 4 Another schematic diagram of a process for converting medical data provided in an embodiment of the present application;

[0053] Figure 5 A schematic diagram of the structure of a medical data conversion device provided in an embodiment of the present application;

[0054] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0056] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0057] In actual applications, different medical institutions and medical systems generate and store medical data according to different data management rules, resulting in various types of multi-source heterogeneous medical data, including the Health Level Seven (HL7) Version 2 (V2) and the Clinical Document Architecture (CDA). In view of the actual need for interoperability of medical information services, it is necessary to integrate the above multi-source heterogeneous medical data into standard unified data, taking global services as the starting point, and solving the problem of medical data not being interoperable between different medical structures and medical systems in the process of medical informationization, thereby realizing the interconnection and interoperability of medical data across systems and institutions.

[0058] In order to solve the above technical problems, some methods for integrating heterogeneous data are provided in the related technologies. However, these methods usually lack top-level design, and various methods are designed and managed separately, which leads to the single application of each integration method, almost no correlation between each other, and each integration method targets different data sources and data structures, and the data management modes and representations are also diverse. Therefore, the above methods themselves also have the characteristics of diversity and cannot achieve unified data conversion and management of multi-source heterogeneous medical data.

[0059] Based on the above problems, the embodiments of the present application provide a medical data conversion method, device, equipment and medium.

[0060] The medical data conversion method provided in the embodiment of the present application can be implemented by a processor of an electronic device. The above-mentioned processor can be at least one of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller (MCU), and a microprocessor.

[0061] Exemplarily, the electronic device may be a physical device or a virtual machine device.

[0062] Exemplarily, the electronic device may be a computer device.

[0063] Figure 1 A flowchart of a medical data conversion method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the process may include the following steps:

[0064] Step 101: Obtain original medical data to be converted.

[0065] In one implementation, the original medical data may include medical data resources generated by various medical institutions or medical systems through data management rules adopted by them.

[0066] In one embodiment, the original medical data may include medical data resources obtained from at least two medical institutions or medical systems. For example, the original medical data may include first medical data and second medical data, wherein the data format, data management method, and at least one of the medical semantics represented by the first medical data and the second medical data may be different. For another example, the first medical data may be HL7V2, and the second medical data may be CDA.

[0067] In one implementation, the original medical data may be obtained through a data transmission interface between at least two medical institutions or medical systems; illustratively, the data transmission interface may be a Representational State Transfer (REST).

[0068] Exemplarily, the RESTFUL interface can be used to implement the transmission of original medical data based on the Hyper Text Transfer Protocol (HTTP) between a medical institution or a medical system; and, through the RESTFUL interface, it can also be implemented to perform GET, POST, PUT, and DELETE operations on the original medical data resources of the medical institution or medical system, wherein GET is used to obtain medical data resources, POST is used to create new medical data resources in the data storage medium of the medical institution or medical system, PUT is used to update medical data resources, and DELETE is used to delete the medical data resources stored in the above-mentioned data storage medium.

[0069] Exemplarily, the medical data resources provided by the medical system or medical structure may include Extensible Markup Language (XML) data or JavaScript Object Notation (JSON) data.

[0070] For example, if the medical data resources provided by the medical system or medical structure are not XML data or JSON data, these medical data resources can be converted into XML data or JSON data, and then the above XML data or JSON data can be obtained through the RESTFUL interface.

[0071] Step 102: extract semantic granular data from the original medical data to obtain a semantic data set.

[0072] In one embodiment, semantic granularity data may include data that can represent complete semantic granularity, for example, semantic granularity data may include outpatient medical records in original medical data; accordingly, a semantic data set may include a set of semantic granularity data that can fully represent at least one medical operation or medical process, for example, semantic granularity data may represent the complete process of the first patient from registration to medical treatment.

[0073] In one embodiment, semantic granular data may include data that can characterize the granularity of basic medical objects. For example, semantic granular data may include the name and gender data of the second patient in the outpatient medical record contained in the original medical data; accordingly, the semantic data set may include data such as the department where the second patient is registered and a description of the second patient's symptoms.

[0074] In one implementation, the semantic data set may be obtained by:

[0075] The semantic granular data in the original medical data is extracted through the semantic extraction model, and the chronological relationship between each semantic granular data is integrated to obtain a semantic data set.

[0076] Natural language recognition is performed on the original medical data through natural language processing (NLP) to obtain a natural language recognition result, and the above natural language recognition result is corrected according to medical terminology, and then the correction result is determined as a semantic data set.

[0077] Step 103: Obtain a resource parameter set included in the rapid medical operational resource.

[0078] Among them, the full spelling of Fast Healthcare Interoperability Resources in English is Fast Healthcare Interoperability Resources, abbreviated as FHIR in English; among them, the FHIR standard provides how to exchange healthcare information between different computer systems. Through the implementation of these standards, healthcare information including clinical medical data and administrative data can be securely and stably provided to users or devices with data access rights.

[0079] FHIR also provides a collection of healthcare resources that express modular medical concepts. By exchanging these healthcare resources, data can be shared quickly and efficiently in healthcare and related processes. The above healthcare resources cover the basic elements of healthcare and a variety of clinical models; the above basic elements include patients, admissions, diagnostic reports, drugs, and problem lists.

[0080] In one embodiment, the resource parameters in the resource parameter set may include at least partially modular medical concepts included in FHIR.

[0081] In one embodiment, the resource parameters in the resource parameter set may include abstract concepts of basic elements of medical care; illustratively, the basic elements may include patients, admissions, diagnostic reports, medications, question lists, and interactive objects that interact with patients; wherein the interactive objects may include at least one of doctors, nurses, and diagnostic equipment.

[0082] In one implementation, the resource parameters in the resource parameter set may be embodied in a structured data format to meet the data processing requirements of electronic devices including computer devices.

[0083] In one implementation, the resource parameters in the resource parameter set may be presented in text form to facilitate maintenance and updating.

[0084] In one implementation, a resource parameter set may include a set of resource parameters and description data of the resource parameters; illustratively, the description data of the resource parameters is used to indicate the type of resource parameters, the data volume of the resource parameters, and the data structure of the resource parameters.

[0085] In one implementation, the resource parameter set may be obtained in any of the following ways:

[0086] A set of resource parameters is pre-set in FHIR. According to data conversion requirements, at least part of the resource parameters can be obtained from the above resource parameter set, and the obtained at least part of the resource parameters can be determined as a resource parameter set. Exemplarily, the data conversion requirements may include data conversion targets associated with type, time, and source, wherein the type-associated data conversion target may include converting medical data resources related to at least one of the patient's (patient's) gender category, age stage category, and patient category; the time-associated data conversion target may include converting medical data resources of at least one era, year, season, or month; the patient category-associated data conversion target may include converting medical data resources corresponding to patient categories such as pediatrics, oncology, and dermatology.

[0087] A resource parameter selection rule is determined based on the original medical data, and at least part of the resource parameters are selected from the resource parameters contained in the FHIR according to the above selection rule, and then the above at least part of the resource parameters are determined as a resource parameter set. For example, based on the gender category, age stage category and disease category of the patient (patient) contained in the original medical data, resource parameters corresponding to the gender category, age stage category and disease category of the patient (patient) can be selected from the resource parameters of the FHIR, and the above resource parameters are determined as a resource parameter set.

[0088] Step 104: Based on the resource parameter set, the semantic granular data in the semantic data set is mapped and converted to obtain target medical data.

[0089] In one implementation, the organization form and storage form of the data in the target medical data may be consistent with the organization form and storage form of the resource parameters in the resource parameter set.

[0090] In one embodiment, the target medical data may be obtained by any of the following methods:

[0091] Predetermine at least one type of conversion rule between semantic granular data and resource parameters in a resource parameter set, and map and convert the semantic granular data in the semantic data set one by one according to the conversion rule, so as to obtain target medical data.

[0092] The medical semantics of the mth resource parameter in the resource parameter set is obtained, and the medical semantics represented by the nth semantic granularity data in the semantic data set is obtained at the same time. If the medical semantics of the mth resource parameter matches the medical semantics represented by the nth semantic granularity data, then based on the data format of the mth resource parameter, the nth semantic granularity data is mapped and converted in the data format dimension to obtain the target medical data corresponding to the nth semantic granularity data; wherein m and n are both integers greater than or equal to 1.

[0093] From the above, it can be seen that the medical data conversion method provided in the embodiment of the present application, after obtaining the original medical data to be converted, obtains a semantic data set by extracting the semantic granularity data in the original medical data, so that the semantic data set can not only accurately reflect the medical information contained in the original medical data from the aspects of natural semantics or medical semantics, but also enables the semantic data set to accurately reflect the medical semantics represented by the original medical data in a causal or causal manner from the semantic granularity; and, obtains the resource parameter set contained in FHIR, and based on the resource parameter set, maps and converts the semantic granularity data in the semantic data set, thereby realizing the mapping conversion direction of the semantic granularity data in the semantic data set. , thereby realizing targeted mapping and conversion of semantic granular data in the semantic data set; on the other hand, by mapping and converting the semantic granular data in the semantic data set based on the resource parameter set, the target medical data is obtained, so that the target medical data can not only comprehensively contain the medical objects and medical semantics represented by the original medical data, but also make the data format of the target medical data consistent with the data format of FHIR, thereby realizing targeted, unified, automated and intelligent conversion of original medical data of any type and any source, thereby meeting the unified conversion requirements of multi-source heterogeneous medical data from different medical systems or medical institutions in practical applications.

[0094] Based on the above embodiments, in the medical data conversion method provided in the embodiments of the present application, the semantic granularity data in the original medical data is extracted to obtain a semantic data set, which can be achieved by the following steps:

[0095] Step A1: Obtain medical object metadata associated with original medical data.

[0096] In one embodiment, medical object metadata can be used to describe the type of medical object contained in the original medical data, the name or identification of the medical object, the offset position of the medical object in the original medical data, the type and amount of each data element contained in the medical object, etc.

[0097] It should be noted that the medical object metadata corresponding to the original medical data from different sources may be different. For example, in the first original medical data obtained from the first medical institution, the field identifier of the patient name is patient name, while in the second original medical data obtained from the second medical institution, the field identifier of the patient name may be huanzhe nam.

[0098] In one embodiment, the medical data meta-object can be obtained by any of the following methods:

[0099] Determine the source institution or source system of the original medical data, obtain the medical data management rules of the source institution or source system, and extract the medical data meta-object from the original medical data according to the medical data management rules.

[0100] The data organization form of the original medical data is analyzed, and the medical data meta-object is determined based on the analysis results.

[0101] Step A2: segment the original medical data based on the medical object metadata to obtain medical object attribute data.

[0102] In one embodiment, the medical object attribute data may include descriptive data of the object attributes of the medical object or the values ​​of the object attributes; exemplarily, the descriptive data is used to describe the morphology and / or process of the medical object, for example, the descriptive data may include the patient status of a first patient in a first time period; exemplarily, the values ​​of the object attributes may be the attribute values ​​of at least one object attribute of the medical object, for example, the values ​​of the object attributes may include the height and weight of a second patient, etc.

[0103] In one embodiment, the medical object attribute data may include at least part of the attribute data of the medical objects in the original medical data, and may also include at least part of the attribute data of any medical object in the original medical data.

[0104] In one embodiment, the medical object attribute data may be obtained by any of the following methods:

[0105] Based on the offset of the mth medical object in the original medical data specified in the medical object metadata and the data amount of the object identification of the mth medical object, the offset of the mth medical object attribute data is determined; based on the data amount of the mth medical object attribute data specified in the medical object metadata and the offset of the mth medical object attribute data, the storage location of the mth medical object attribute data in the original medical data is determined; and then based on the above storage location, the original medical data is segmented to obtain the mth medical object attribute parameters.

[0106] Attribute data of at least one dimension of the m-th medical object is acquired, and the attribute data set of the at least one dimension is determined as the medical object attribute data of the m-th medical object.

[0107] Step A3: perform semantic segmentation on the medical object attribute data at the granularity of basic semantic units to obtain a semantic data set.

[0108] In one embodiment, the basic semantic unit granularity may include the minimum semantic unit granularity. For example, the basic semantic unit granularity may be the smallest unit of object data in the field of medical technology. For example, the smallest semantic unit may include name, gender, age, drug, etc.

[0109] In one implementation, the semantic data set may be obtained by any of the following methods:

[0110] A medical term set is obtained, and semantic segmentation of medical object attribute data at a basic semantic unit granularity is performed based on the data in the medical term set, thereby obtaining a semantic data set.

[0111] Through semantic recognition algorithms or semantic recognition models, data of basic semantic unit granularity in medical object attribute data are identified and segmented, and the set of basic semantic units obtained by identification and segmentation is determined as a semantic data set; for example, the semantic data set corresponding to the patient name may include patient and name, or huanzhe and nam.

[0112] In an embodiment of the present application, the semantic data set obtained by semantically segmenting the medical object attribute data at the granularity of basic semantic units can be recorded as the minimum source parameter, which is used to represent the original medical data represented at the granularity of basic semantic units.

[0113] From the above, it can be seen that in the medical data conversion method provided in the embodiment of the present application, the mutual correlation and storage location between the medical object data in the original medical data can be accurately and completely obtained through the medical object metadata associated with the original medical data; in this way, the original medical data is segmented based on the medical object metadata to obtain medical object attribute data, so that the medical object attribute data can comprehensively and accurately represent the medical information contained in the original medical data; on this basis, the medical object attribute data is semantically segmented at the granularity of basic semantic units, so that the semantic data set obtained through the above steps can more finely characterize and display the medical information carried in the original medical data from the granularity of basic semantic units.

[0114] Based on the above embodiments, in the medical data conversion method provided in the embodiments of the present application, based on the medical parameter set, the semantic granularity data in the semantic data set is mapped and converted to obtain the target medical data, which can be achieved by the following steps:

[0115] Step B1: Associatively identifying semantic granular data and at least part of resource parameters in a resource parameter set through a graphic mapping engine, and obtaining a mapping association relationship between the semantic granular data and at least part of the resource parameters.

[0116] In one implementation, the mapping association relationship may include a one-to-one mapping relationship between semantic granularity data and resource parameters in at least part of the resource parameters, for example, a mapping relationship exists between the mth semantic granularity data and the nth resource parameter.

[0117] In one embodiment, the above-mentioned mapping association relationship may include a many-to-one mapping relationship between semantic granularity data and resource parameters in at least part of the resource parameters. For example, the mth semantic granularity data and the nth semantic granularity data contained in the first original medical data both have a mapping relationship with the nth resource parameter.

[0118] In one implementation, the graph mapping engine may respectively identify the semantic granularity data and at least a portion of the resource parameters, and convert the semantic granularity data and at least a portion of the resource parameters into graph data, and then display the graph data respectively.

[0119] In one implementation, the graphic mapping engine may extract the matching relationship between the semantic granular data and at least a portion of the resource parameters during the process of identifying the semantic granular data and at least a portion of the resource parameters, and determine the matching relationship as the mapping association relationship.

[0120] In one implementation, the graphic mapping engine may respond to a user's selection of an association between the semantic granularity data and at least a portion of the resource parameters, and determine the result of the association selection as a mapping association relationship.

[0121] In one embodiment, at least part of the resource parameters may be determined in response to a user selection, and may also be selected from resource parameters included in FHIR based on a degree of match with the semantic granularity data.

[0122] In the related art, the method of graphical data processing is mainly used in the query field of heterogeneous data sources. For example, the user selects the query range through the front-end visual interface, and determines whether the selected data source has been loaded into the presto cluster according to the above query range. If the selected data source is not loaded into the presto cluster, the data source to be loaded is added to the presto cluster; then the plug-in required for the data source to be loaded is detected to be a built-in data source plug-in in the presto cluster. If the plug-in required for the data source to be loaded is a built-in data source plug-in in the presto cluster, the presto cluster dynamically loads the configuration file in the catalog folder; if the plug-in required for the data source to be loaded is not a built-in data source plug-in in the presto cluster, a plug-in for the data source to be loaded is created and built into the Presto cluster, and the configuration file in the catalog folder is loaded to complete the loading of the data source. This method configures different basic information of the data source, uses the operation of the graphical interface, dynamically loads the data source, and realizes the query of heterogeneous data sources based on presto.

[0123] Obviously, the above method can only realize the query of heterogeneous data sources through a graphical data processing environment, while the graphic mapping engine in the embodiment of the present application can determine the mapping association relationship between semantic granularity data and at least part of the resource parameters. Therefore, the data processing function of the graphic mapping engine in the embodiment of the present application is more powerful.

[0124] Step B2: Based on the mapping association relationship, the semantic granular data in the semantic data set is mapped and transformed to obtain the target medical data.

[0125] In one embodiment, the target medical data may be obtained by any of the following methods:

[0126] Based on the mapping association relationship, the data in the semantic granularity data is directly mapped and converted into a format or type corresponding to the resource parameter to obtain a first conversion result, and the first conversion result is determined as the target medical data.

[0127] Based on the mapping association relationship, at least part of the data in the semantic data set is mapped and transformed to obtain a second transformation result, and the second transformation result and the semantic granular data excluding at least part of the above-mentioned data are determined as target medical data; wherein, at least part of the data in the semantic granular data may include data in the semantic data set that does not completely match at least part of the resource parameters.

[0128] Related technologies also provide methods for assembling data elements to form data groups, or extracting data elements to generate XML message instances. However, the above methods cannot realize the mutual conversion of multi-source heterogeneous medical data. In addition, the above solutions need to be developed through software code to query the data groups and data element information bound to XML elements from the annotation model according to the traversal path, and dynamically fill in data and dynamically increase or decrease the element structure according to the binding information, data groups and data element information defined by the annotation model.

[0129] In the medical data conversion method provided in the embodiment of the present application, the semantic granular data and at least part of the resource parameters in the resource parameter set are associated and identified through the graphic mapping engine to obtain the mapping association relationship between the semantic granular data and at least part of the resource parameters. In this way, not only can the data recognition and data association functions of the graphical tool be fully utilized, but also the workload caused by code development can be reduced, the software cost and difficulty of obtaining the mapping association relationship can be reduced, and the efficiency of determining the mapping association relationship can be improved; and, through the above-mentioned operation of the graphic mapping engine, a visual mapping association relationship determination scheme is provided; on this basis, the efficiency of mapping and converting the semantic granular data to obtain the target medical data based on the mapping association relationship can be improved.

[0130] Based on the above embodiments, in the medical data conversion method provided in the embodiments of the present application, based on the mapping association relationship, the semantic granularity data in the semantic data set is mapped and converted to obtain the target medical data, which can be achieved in the following ways:

[0131] Determine the data structure of at least part of the resource parameters; convert the semantic granularity data based on the mapping association relationship and the data structure through the trained conversion model to obtain the target medical data.

[0132] In one embodiment, the data structure of at least part of the resource parameters may include the data element types contained in any resource parameter in at least part of the resource parameters, the dependency relationships between various types of data elements, and the chronological position relationships of various types of data elements in the resource parameters.

[0133] In one implementation, at least part of the resource parameters may be identified through a trained conversion network, thereby determining a data structure of at least part of the resource parameters.

[0134] In one implementation, the trained conversion model may include a neural network model; illustratively, the neural network model may include a forward feedback (FF) network.

[0135] In one implementation, the trained conversion model may have at least one of the following functions:

[0136] A data structure that identifies medical semantics of semantically granular data and at least a portion of resource parameters.

[0137] The medical semantics possessed by the semantic granularity data are converted into data elements required for the data structure of at least part of the resource parameters.

[0138] The converted data elements are arranged according to the relative positions and interdependencies between the data elements contained in the data structure.

[0139] In one implementation, the target medical data may be obtained by any of the following methods:

[0140] Through the trained conversion model, the semantic granular data is identified to obtain the medical semantics of the semantic granular data, and the medical semantics is abstracted and segmented into data elements based on the data structure of the resource parameters in at least part of the resource parameters, so as to obtain the data elements contained in the above data structure, and then the dependency relationship between various data elements is constructed, and then the various data elements are integrated according to the dependency relationship, and finally the integration result is determined as the target medical data.

[0141] Through the trained conversion model, the relational data with deviations in the mapping association relationship is adjusted to obtain the adjusted mapping association relationship, and then based on the structural data, the data elements of the semantic granularity data contained in the adjusted mapping association relationship are abstracted to obtain the data elements of the type contained in the structural data, and based on the dependency contained in the structural data, the data elements of the type contained in the structural data are integrated, and then the integration result is determined as the target medical data.

[0142] From the above, it can be seen that the medical data conversion method provided in the embodiment of the present application can comprehensively and accurately obtain the types of data elements contained in at least part of the resource parameters and the dependencies between the data elements by determining the data structure of at least part of the resource parameters; on this basis, through the trained conversion model, based on the mapping association relationship and the data structure, the semantic granularity data is converted, so as to achieve targeted conversion of the semantic granularity data; and, with the advantage of the high efficiency of the trained conversion model in data processing, the efficiency of the conversion of the semantic granularity data can also be improved; at the same time, in the process of converting the semantic granularity data through the trained conversion model, adding the factor of mapping association relationship can realize the control of the conversion direction of the trained conversion model in the conversion of semantic granularity data, and can further improve the efficiency of the conversion of the semantic granularity data.

[0143] Based on the above embodiments, the medical data conversion method provided in the embodiments of the present application may further include the following steps:

[0144] Step C1: Obtain sample data.

[0145] Among them, the sample data includes various types of initial medical data.

[0146] In one embodiment, the sample data may include medical data from multiple medical systems or medical institutions.

[0147] In one implementation, the sample data may be acquired by the same method as that used in the aforementioned embodiments to acquire the original medical data.

[0148] In one implementation, the sample data may include a resource tag; illustratively, the resource tag may include a target resource parameter corresponding to the sample data; illustratively, the target resource parameter may include a resource parameter in FHIR obtained by pre-converting the sample data.

[0149] Step C2: Obtain a sample mapping relationship between sample data and a resource parameter set.

[0150] In one implementation, the sample mapping relationship can be obtained by processing the semantic granularity data and the resource parameter set included in the sample data through the graphic mapping engine provided in the aforementioned embodiment.

[0151] Step C3: convert the sample data based on the sample mapping relationship through the conversion model to obtain a sample conversion result.

[0152] In one implementation, the semantic granularity data in the sample data may be converted based on the sample mapping relationship and the data structure of the resource parameters by using the same method as the aforementioned embodiment, thereby obtaining a sample conversion result.

[0153] Step C4: Obtain abnormal data contained in the sample conversion result.

[0154] In one implementation, the abnormal data may include sample data that cannot be converted into a target format; illustratively, the target format may include a format of a resource parameter in FHIR, or may be a data format possessed by a target resource parameter.

[0155] In one implementation, the abnormal data may include data of samples that are not in the target format in the sample conversion results.

[0156] In one embodiment, abnormal data may be determined by any of the following methods:

[0157] The sample conversion result is matched with the target resource parameter. If the sample conversion result does not match the target resource parameter, the sample conversion result may be determined as abnormal data.

[0158] The sample conversion results are detected by using pre-set detection rules, and the sample conversion results that do not meet the above detection rules are determined as abnormal data.

[0159] Step C5: adjusting the model parameters of the conversion model based on the abnormal data, and obtaining a trained conversion model according to the adjusted model parameters.

[0160] In one implementation, the trained conversion model can be obtained by any of the following methods:

[0161] Obtain first data corresponding to abnormal data in the sample data, determine data features of the first data, obtain second data having the above data features, merge the first data and the second data to obtain merged data, so as to achieve the expansion of the number of sample data having the above data features, and then transform the merged data through the transformation model, so that the transformation model can adjust the transformation parameters associated with the transformation process of the merged data, and obtain the trained transformation model according to the adjusted transformation parameters.

[0162] Based on the abnormality type represented by the abnormal data, a conversion unit related to the abnormality type in the conversion model is determined, and then the conversion parameters of the conversion unit are adjusted, and a trained conversion model is obtained based on the conversion unit after the conversion parameters are adjusted.

[0163] From the above, it can be seen that in the medical data conversion method provided in the embodiment of the present application, the acquired sample data includes multiple types of initial medical data, so that the sample data can comprehensively and fully reflect the diversity and differences of the initial medical data generated from different sources and according to different data management rules; and, obtaining the sample mapping relationship between the sample data domain resource parameter sets can accelerate the efficiency of converting the sample data through the conversion model, thereby obtaining the sample conversion result; at the same time, after obtaining the abnormal data contained in the sample conversion result, by adjusting the model parameters of the conversion model based on the abnormal data, the conversion model can be trained in a targeted manner, thereby shortening the time to obtain the trained conversion model and improving the training efficiency of the conversion model.

[0164] Based on the above embodiments, in the medical data conversion method provided in the embodiments of the present application, the model parameters of the conversion model are adjusted based on the abnormal data, which can be achieved by Figure 2 The process shown is implemented as follows, Figure 2 A flow chart of obtaining a conversion model after the k+1th parameter adjustment provided in an embodiment of the present application is as follows: Figure 2As shown, the process may include the following steps:

[0165] Step 201: Obtain the kth abnormal data.

[0166] The k-th abnormal data is the abnormal data contained in the k-th conversion result output by the conversion model after the k-th parameter adjustment; k is an integer greater than or equal to 1.

[0167] In one implementation, the number of data in the k-th abnormal data may be at least two.

[0168] Step 202: Determine the kth abnormal type corresponding to the kth abnormal data.

[0169] In one implementation, the exception type included in the kth exception type may be at least one.

[0170] In one implementation, the kth exception type may include an exception type in which the sample data included in the kth conversion result is not converted, and may also include an exception type in which the kth conversion result is not a target data resource.

[0171] In one implementation, the kth abnormality type may be determined by any of the following methods:

[0172] The kth exception type is determined according to the data structure of the kth exception data. For example, if the data structure of the kth exception data is not the data structure of the target resource parameter, the kth exception type can be determined as an exception type of data structure error.

[0173] The kth exception type is determined according to the degree of matching between the kth exception data and the target resource parameters. For example, if the kth exception data does not completely match the target resource parameters, the kth exception type can be determined as: an exception type of data conversion failure.

[0174] Step 203: Determine the data associated with the kth abnormal data in the sample data as the k+1th training data.

[0175] In one embodiment, the data in the sample data may carry a data identifier, and the above data identifier may remain unchanged during the process of the conversion model processing the sample data. That is, the kth abnormal data may also carry a data identifier. Therefore, the kth data corresponding to the kth abnormal data can be determined from the sample data through the data identifier carried by the kth abnormal data, and the kth data can be determined as the k+1th training data; wherein the above data identifier may include the serial number of the data in the sample data.

[0176] Step 204: Based on the k+1th training data and the kth abnormality type, adjust the model parameters of the conversion model after the kth parameter adjustment to obtain the conversion model after the k+1th parameter adjustment.

[0177] In one implementation, the input data of the conversion model after the kth parameter adjustment can be expressed by formula (1):

[0178] w(k+1)=w(x(k),F k ) (1)

[0179] Where w(k+1) is the input data of the conversion model after the kth parameter adjustment, F k It may be the kth abnormal type, and x(k) may be at least part of the data associated with the kth abnormal type in the sample data, that is, the k+1th training data.

[0180] In one implementation, the conversion model after the k+1th parameter adjustment can be obtained by:

[0181] Based on the kth abnormality type, in the process of processing the k+1th training data through the conversion model after the kth parameter adjustment, the model parameters for converting the kth type of abnormal data contained in the conversion model after the kth parameter adjustment are adjusted, so as to obtain the conversion model after the k+1th parameter adjustment.

[0182] From the above, it can be seen that in the medical data conversion method provided in the embodiment of the present application, by obtaining the kth abnormal data contained in the kth conversion result output by the conversion model after the kth parameter adjustment, the conversion efficiency of the conversion model after the kth parameter adjustment for the sample data can be accurately determined; and, by determining the kth abnormal type corresponding to the kth abnormal data, the abnormal representation corresponding to the kth abnormal data can be accurately determined; at the same time, after determining that the data associated with the kth abnormal data in the sample data is the k+1th training data, based on the k+1th training data and the kth abnormal type, the model parameters of the conversion model after the kth parameter adjustment are adjusted to obtain the conversion model after the k+1th parameter adjustment, thereby realizing targeted adjustment of the model parameters of the conversion model after the kth parameter adjustment, thereby improving the training speed of the conversion model and shortening the training time.

[0183] Based on the above embodiments, in the medical data conversion method provided in the embodiments of the present application, determining the kth abnormal type corresponding to the kth abnormal data can be achieved by the following steps:

[0184] Step D1: Get the FHIR exception type set.

[0185] In one implementation, the exception type set may include a set of pre-defined, standardized exception types in the FHIR system.

[0186] Step D2: Obtain the kth abnormal description information of the kth abnormal data.

[0187] In one implementation, the k-th abnormal description information may include at least part of the rules in the pre-set medical data standardization implementation specification that the k-th abnormal data does not satisfy.

[0188] In one embodiment, the kth exception description information can characterize the degree of mismatch between the kth exception data and the target resource parameters from at least one dimension; exemplarily, the kth exception description information may include that the kth exception data lacks at least one data element contained in the target resource parameters, the dependency relationship between the data elements in the kth exception data does not match the dependency relationship between the data elements in the target resource parameters, the kth exception data contains data elements other than the data elements specified by the target resource parameters, etc.

[0189] In one implementation, the kth exception description information may be obtained in any of the following ways:

[0190] The k-th abnormal data is detected according to a preset medical data standardization implementation specification, and at least part of the rules included in the medical data standardization implementation specification that are not satisfied by the k-th abnormal data are determined as the k-th abnormal description information.

[0191] The data elements contained in the kth abnormal data, the types of each data element, and the interdependence between the data elements are analyzed, and the degree of differentiation between the above data and the data of the same dimension in the target resource parameters is determined as the kth abnormal description information.

[0192] Step D3: Determine the exception type in the exception type set that matches the kth exception description information as the kth exception type.

[0193] In one implementation, traversal matching may be performed based on the kth exception description information and the exception types in the exception type set, and the exception type in the exception type set that matches the kth exception description information is determined as the kth exception type.

[0194] In one implementation, the number of exception types included in the kth exception type may be at least two.

[0195] As can be seen from the above, in the medical data conversion method provided in the embodiment of the present application, after obtaining the exception type set of FHIR and the kth exception description information of the kth exception data, the exception type in the exception type set that matches the kth exception description information is determined to be the kth exception type. In this way, since the exception type set is a set of exception types of FHIR, the difference between the kth exception type and the resource parameters contained in the FHIR can be improved through the above operation, thereby improving the accuracy of the kth exception type.

[0196] Based on the above embodiments, in the medical data conversion method provided in the embodiments of the present application, obtaining abnormal data contained in the sample data conversion result can be achieved by the following steps:

[0197] Step E1: Obtain at least one configuration file.

[0198] In one implementation, the configuration file may include an implementation guide for medical data conversion; illustratively, the implementation guide may include implementation specifications for medical data standardization established by a medical institution.

[0199] In one implementation, if the implementation guide is not obtained or does not exist, a custom configuration file may be obtained and the custom configuration file may be determined as the above configuration file.

[0200] In one implementation, if the custom configuration file is not obtained or does not exist, a default configuration file may be obtained, and the default configuration file may be determined as the above configuration file.

[0201] In one implementation, the configuration file may include at least one of an implementation guide, a custom configuration file, and a default configuration file.

[0202] In one implementation, a configuration file may include multiple configuration rules; illustratively, the configuration rules may indicate the type and quantity of data elements included in FHIR type resource parameters, the mutual setting method between each data element, whether each data element is a mandatory type in the resource parameter, etc.

[0203] Step E2: Detect the sample conversion results based on at least one configuration file, and determine the data in the sample conversion results that does not satisfy the configuration rules included in the at least one configuration file as abnormal data.

[0204] In one embodiment, abnormal data may be determined by any of the following methods:

[0205] Based on any one of the implementation guide, the default configuration file or the custom configuration file, the data in the sample conversion result is detected, and the data in the sample conversion result that does not meet the configuration rules contained in any one of the above files is determined as abnormal data.

[0206] Based on the implementation guide, the default mode file, and the custom configuration file, the data in the sample conversion result is detected in turn, and the data in the sample conversion result that does not meet the configuration rules contained in the above files is determined as abnormal data.

[0207] From the above, it can be seen that the medical data conversion method provided in the embodiment of the present application, after obtaining at least one configuration file, detects the sample conversion results based on at least one configuration file, and determines the data in the sample conversion results that does not meet the configuration rules contained in at least one configuration file as abnormal data. In this way, in the above process of detecting the sample conversion results, by adjusting or replacing the number or type of configuration files, it is possible to achieve diversified and flexible detection of the sample conversion results; and through the above method, the accuracy of abnormal data can be improved.

[0208] Based on the above embodiments, in the medical data conversion method provided in the embodiments of the present application, the model parameters of the conversion model are adjusted based on the abnormal data, and the trained conversion model is obtained according to the adjusted model parameters, which can be achieved in the following ways:

[0209] Obtain the k+1th conversion result output by the conversion model after the k+1th parameter adjustment; if the amount of k+1th abnormal data contained in the k+1th conversion result is less than the target threshold, determine that the conversion model after the k+1th parameter adjustment is the trained conversion model.

[0210] Here, k is an integer greater than or equal to 1.

[0211] Exemplarily, if the amount of k+1th abnormal data contained in the k+1th conversion result is greater than the target threshold, the k+2th training data and the k+1th abnormal type can be obtained by the method provided in the aforementioned embodiment, and based on the k+1th abnormal type and the k+2th training data, the model parameters of the conversion model after the k+1th parameter adjustment can be further adjusted.

[0212] In one embodiment, the target threshold may be determined in advance based on data conversion requirements; illustratively, the data conversion requirements may include parameter requirements such as the target success rate of the original medical data conversion, which is not limited in this embodiment of the present application.

[0213] In one embodiment, the target threshold may be zero.

[0214] From the above, it can be seen that in the medical data conversion method provided by the embodiment of the present application, if the k+1th conversion result output by the conversion model after the k+1th parameter adjustment contains the k+1th abnormal data whose data volume is less than or equal to the target threshold, then the conversion model after the k+1th parameter adjustment is determined to be the conversion model that has been trained. In this way, after each parameter adjustment is completed, by judging in real time whether the amount of abnormal data is less than or equal to the target threshold, the training process can be controlled in real time; and by adjusting the target threshold, the conversion model can be trained in a diversified and flexible manner, so that the trained conversion model can meet the diverse medical data conversion needs.

[0215] Figure 3 A flow chart of a training conversion model provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, the process may include the following steps:

[0216] Step 301, start.

[0217] Step 302: Transmit source data.

[0218] Exemplarily, the source data may be the sample data in the aforementioned embodiment.

[0219] Exemplarily, the sample data may be transmitted via RESTFUL.

[0220] Step 303: extract and split the minimum parameter unit in the source data by using the cosine similarity method.

[0221] Exemplarily, the minimum parameter unit in the source data may include the semantic granularity data in the aforementioned embodiment.

[0222] Step 304: extract and split the minimum parameter unit in the target FHIR data by using the cosine similarity method.

[0223] Exemplarily, the target FHIR data may include a set of target resource parameters in FHIR that have a correct conversion relationship with the source data.

[0224] Exemplarily, the minimum parameter unit in the FHIR data may include the target resource parameter in the aforementioned embodiment.

[0225] Exemplarily, the order of step 303 and step 304 can be adjusted or performed in parallel, which is not limited in the embodiment of the present application.

[0226] Exemplarily, after step 304 is completed, either step 305 or step 306 may be selected for execution.

[0227] Step 305: Generate XML code through structured filling.

[0228] Exemplarily, a structured filling operation may be performed by the graph mapping engine provided in the aforementioned embodiment to generate XML code, thereby establishing a sample mapping relationship between the minimum parameter unit in the source data and the minimum parameter unit in the FHIR data.

[0229] Step 306: Enter the XML code to generate a sample mapping relationship.

[0230] Exemplarily, the sample mapping relationship may also be generated by XML code development.

[0231] Step 307: Convert to FHIR format.

[0232] Exemplarily, the minimum parameter unit in the source data may be converted by the conversion model after the k-th parameter adjustment, thereby obtaining the k-th sample conversion result in the FHIR format.

[0233] Step 308: Check the sample conversion result.

[0234] Exemplarily, the sample conversion result may be checked by any one of the implementation guide, the custom configuration file, and the default configuration file to obtain the kth abnormal data.

[0235] Step 309: Get the exception type F k .

[0236] For example, the exception description information of the kth exception data can be matched by the exception type set in FHIR to obtain the kth exception type F k .

[0237] Step 310: Set the preconditions for the k+1th training and perform the k+1th conversion.

[0238] Exemplarily, the precondition of the k+1th training may include input data of the k+1th training.

[0239] Exemplarily, the input data for the k+1th training may be as shown in formula (1) in the aforementioned embodiment, and may include the kth abnormal type, and at least part of the data in the sample data associated with the kth abnormal data, namely, the k+1th training data.

[0240] Exemplarily, based on the kth abnormality type and at least part of the data associated with the kth abnormal data in the sample data, the model parameters of the training model after the kth parameter adjustment can be adjusted to obtain the training model after the k+1th parameter adjustment.

[0241] Step 311: If the model converges, switch to the monitoring state.

[0242] Exemplarily, if the model has not switched to the converged state, it may not switch to the monitoring state.

[0243] Exemplarily, model convergence may include the number of abnormal data included in the sample conversion results output by the training model after the k+1th parameter adjustment being less than or equal to the target threshold.

[0244] Exemplarily, the monitoring state may include a state in which a trained conversion model is waiting for input of raw medical data to be converted.

[0245] Step 312: Automatically generate an Application Programming Interface (API) or a Software Development Kit (SDK).

[0246] Exemplarily, the steps of transmitting source data, extracting and splitting source data into minimum parameter units, and the trained conversion model can be packaged into an SDK, through which a unified and automated conversion process of any original medical data can be achieved.

[0247] Exemplarily, the interface for transmitting source data and outputting FHIR resource parameters of the trained conversion model can be encapsulated into a unified API, and an API calling method is provided so that in scenarios where the original medical data needs to be converted, the above API can be called to achieve flexible conversion of the original medical data.

[0248] Step 313, end.

[0249] From the above process, it can be known that in the process of training the conversion model in the embodiment of the present application, the precondition for the next training can be set according to the abnormal type corresponding to the abnormal data in the sample conversion result output by the conversion model each time, and the targeted adjustment and control of the training direction of the conversion model can be achieved, thereby improving the data conversion accuracy of the conversion model after training, and solving the technical problems of low data resource matching rate and poor conversion accuracy in FHIR during medical data conversion in related technologies. It can also shorten the training cycle of the conversion model and improve the training efficiency; and, by automatically generating APIs or SDKs, it can realize flexible conversion of original medical data in a variety of scenarios, thereby breaking through the limitations of regional, network and operating system barriers.

[0250] Figure 4 Another schematic diagram of a process for converting medical data provided in an embodiment of the present application.

[0251] It should be noted that Figure 4 Can be Figure 3 The subsequent process, that is, Figure 4 Can be Figure 3 The generated API or SDK application process. Figure 4 In the present invention, the conversion process of medical data can be realized through a business integration platform, a heterogeneous data graphical conversion system and a hospital side; wherein, the business integration platform is used to provide an API interface and an SDK, the heterogeneous data graphical conversion system may include the graphical mapping engine in the aforementioned embodiment and a trained conversion model, the hospital side can be used to provide original medical data, and can also receive data output by the heterogeneous data graphical conversion system and perform specified operations on the data.

[0252] like Figure 4 As shown, the process may include the following steps:

[0253] Step 401: Create a conversion project.

[0254] Exemplarily, after the heterogeneous data graphical conversion system establishes the conversion project, it can notify the hospital side so that the hospital side can execute step 402.

[0255] Step 402: Send a registration API interface request.

[0256] Exemplarily, the hospital sends a registration API interface request, and after the request is received by the heterogeneous data graphical conversion system, the original medical data to be converted can be sent to the heterogeneous data graphical conversion system.

[0257] Step 403: Receive request parameters for unified management.

[0258] Exemplarily, after the heterogeneous data graphical conversion system receives the API interface request sent by the hospital, the registration request can be enabled.

[0259] Step 404: Check whether the interface registration is complete.

[0260] Exemplarily, if the heterogeneous data graphical conversion system detects that the interface registration is not completed, step 402 is executed; if it is detected that the interface registration is completed, step 405 is executed.

[0261] Step 405: hospital data resources and FHIR resources are mapped.

[0262] Exemplarily, the hospital data resource may include the original medical data in the aforementioned embodiment, and the FHIR resource may include the resource parameter set in the aforementioned embodiment.

[0263] For example, the connection mapping between hospital data resources and FHIR resources can be automatically realized by a graphic mapping engine, or the connection mapping operation can be performed manually by professional technicians to obtain the association mapping relationship in the above embodiment.

[0264] Step 406: Determine whether to fully map.

[0265] Exemplarily, the heterogeneous data graphical conversion system can determine whether all hospital data resources are mapped to FHIR resources. If not, the hospital side can be notified to execute step 407. If yes, step 408 can be executed.

[0266] Step 407: The hospital modifies the fields that cannot be mapped.

[0267] Exemplarily, the hospital side may obtain the unmappable fields and modify them based on the medical data management rules, and step 402 may be executed after the modification is completed.

[0268] Step 408: Start conversion.

[0269] Exemplarily, the heterogeneous data graphical conversion system can convert hospital data resources through the conversion model trained in the aforementioned embodiment to obtain target medical data.

[0270] Step 409: Determine whether the conversion is successful.

[0271] Exemplarily, the heterogeneous data graphical conversion system can determine whether the hospital data resources are converted successfully. If the conversion is not successful, step 410 is executed; if the conversion is completed, step 411 is executed.

[0272] Step 410: Determine whether the hospital data resources are abnormal.

[0273] Exemplarily, if the hospital determines that the hospital data resources are abnormal, step 407 is executed; if the hospital data resources are normal, step 408 is executed to continue converting the hospital data resources.

[0274] Step 411: Convert to FHIR interface, and the permission configuration is the same as the original hospital interface.

[0275] Exemplarily, after receiving the message sent by the heterogeneous data graphical conversion system that the conversion of hospital data resources is completed, the business integration platform can generate a FHIR interface for accessing the target medical data and configure the access rights of the interface to be the same as the access rights of the original hospital interface.

[0276] Exemplarily, the hospital original interface may include an API provided by the hospital for accessing hospital data resources.

[0277] Step 412, end.

[0278] From the above, it can be seen that in the medical data conversion method provided in the embodiment of the present application, with the help of the graphic mapping engine and the trained conversion model in the heterogeneous data graphical conversion system, it is possible to achieve efficient, accurate, reliable and high-success-rate real-time conversion of hospital data resources provided by the hospital side, thereby reducing the workload of converting medical data through code writing and debugging, and improving the flexibility of medical data conversion; and, based on the above characteristics, the heterogeneous data graphical conversion system can be customized and developed on-site according to actual data conversion needs, thereby increasing the flexibility of data conversion project docking, reducing the cost for developers to learn different data standards, and reducing the difficulty of secondary development and multiple development of data conversion projects, thereby providing a visual, unified and highly flexible medical data conversion system, as well as a medical data conversion system that can be flexibly executed and can be implemented at any time, which can meet the actual needs of interconnection between multi-source heterogeneous medical data across institutions and systems.

[0279] Moreover, the medical data conversion method provided in the embodiment of the present application can quickly and intelligently convert various heterogeneous medical data from multiple sources into FHIR data, thereby reducing the workload of manual conversion and verification of medical data; experimental data show that the data conversion accuracy of the medical data conversion method provided in the embodiment of the present application is greater than or equal to 95%. Therefore, the medical data conversion method provided in the embodiment of the present application can realize the rapid interconnection and intercommunication of medical data, greatly improving the implementation speed of the construction of the medical data community.

[0280] Based on the above embodiments, the present application also provides a medical data conversion device. Figure 5 A schematic diagram of the structure of the medical data conversion device provided in the embodiment of the present application is shown in FIG. Figure 5 As shown, the device may include:

[0281] An acquisition module 501 is used to acquire original medical data to be converted;

[0282] An extraction module 502 is used to extract semantic granularity data from the original medical data to obtain a semantic data set;

[0283] The acquisition module 501 is also used to acquire a resource parameter set included in the rapid medical operational resource;

[0284] The conversion module 503 is used to map and convert the semantic granular data in the semantic data set based on the resource parameter set to obtain the target medical data.

[0285] In some embodiments, the acquisition module 501 is used to acquire medical object metadata associated with the original medical data;

[0286] The extraction module 502 is used to segment the original medical data based on the medical object metadata to obtain medical object attribute data; and perform semantic segmentation on the medical object attribute data at the granularity of basic semantic units to obtain a semantic data set.

[0287] In some embodiments, the conversion module 503 is used to associate and identify semantic granular data and at least part of the resource parameters in the resource parameter set through a graphic mapping engine to obtain a mapping association relationship between the semantic granular data and at least part of the resource parameters; based on the mapping association relationship, the semantic granular data in the semantic data set is mapped and converted to obtain target medical data.

[0288] In some embodiments, the medical data conversion device further includes a determination module for determining a data structure of at least part of the resource parameters.

[0289] The conversion module 503 is used to convert the semantic granularity data through the trained conversion model based on the mapping association relationship and data structure to obtain the target medical data.

[0290] In some embodiments, the acquisition module 501 is used to acquire sample data; acquire a sample mapping relationship between the sample data and a resource parameter set; wherein the sample data includes multiple types of initial medical data;

[0291] The medical data conversion device also includes a training module, which is used to convert the sample data based on the sample mapping relationship through the conversion model to obtain the sample conversion result; obtain the abnormal data contained in the sample conversion result; adjust the model parameters of the conversion model based on the abnormal data, and obtain the trained conversion model according to the adjusted model parameters.

[0292] In some embodiments, the acquisition module 501 is used to acquire the kth abnormal data; wherein the kth abnormal data is the abnormal data contained in the kth conversion result output by the conversion model after the kth parameter adjustment; k is an integer greater than or equal to 1;

[0293] A determination module is used to determine the kth abnormal type corresponding to the kth abnormal data; determine that the data associated with the kth abnormal data in the sample data is the k+1th training data;

[0294] The training module is used to adjust the model parameters of the conversion model after the kth parameter adjustment based on the k+1th training data and the kth abnormality type to obtain the conversion model after the k+1th parameter adjustment.

[0295] In some embodiments, the acquisition module 501 is used to acquire an abnormality type set of a rapid medical operational resource; acquire the kth abnormality description information of the kth abnormal data;

[0296] The determination module is used to determine that the exception type matching the kth exception description information in the exception type set is the kth exception type.

[0297] In some embodiments, the acquisition module 501 is used to acquire at least one configuration file;

[0298] The determination module is used to detect the sample conversion results based on at least one configuration file, and determine the data in the sample conversion results that does not meet the configuration rules included in the at least one configuration file as abnormal data.

[0299] In some embodiments, the acquisition module 501 is used to obtain the k+1th conversion result output by the conversion model after the k+1th parameter adjustment; wherein k is an integer greater than or equal to 1;

[0300] A determination module is used to determine that the conversion model after the k+1th parameter adjustment is the trained conversion model if the amount of the k+1th abnormal data included in the k+1th conversion result is less than or equal to the target threshold.

[0301] The medical data conversion device provided in the embodiment of the present application obtains the original medical data to be converted by extracting the semantic granularity data in the original medical data to obtain a semantic data set, so that the semantic data set can not only accurately reflect the medical information contained in the original medical data from the aspects of natural semantics or medical semantics, but also enable the semantic data set to accurately reflect the medical semantics represented by the original medical data in a causal or causal manner from the semantic granularity; and obtain the resource parameter set contained in FHIR, and based on the resource parameter set, map and convert the semantic granularity data in the semantic data set, thereby realizing the control of the mapping conversion direction of the semantic granularity data in the semantic data set, and then realizing the semantic data On the other hand, by mapping and converting the semantic granular data in the semantic data set based on the resource parameter set to obtain the target medical data, the target medical data can not only comprehensively contain the medical objects and medical semantics represented by the original medical data, but also make the data format of the target medical data consistent with the data format of FHIR, thereby realizing the targeted, unified, automated and intelligent conversion of original medical data of any type and any source, and thus meeting the unified conversion requirements of multi-source heterogeneous medical data from different medical systems or medical institutions in practical applications. Based on the above embodiments, the embodiments of the present application also provide an electronic device, Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application, such as Figure 6As shown, the electronic device may include a processor 601 and a memory 602; wherein the memory 602 stores a computer program, and when the computer program is executed by the processor 601, it can implement the medical data conversion method provided in any of the previous embodiments.

[0302] The above-mentioned processor can be at least one of ASIC, DSP, DSPD, PLD, FPGA, CPU, controller, microcontroller, and microprocessor.

[0303] The above-mentioned memory can be a volatile memory (volatile memory), such as a random access memory (Random Access Memory, RAM); or a non-volatile memory (non-volatile memory), such as a read-only memory (Read-Only Memory, ROM), flash memory, a hard disk drive (Hard Disk Drive, HDD) or a solid state drive (SSD); or a combination of the above types of memory, and provide instructions and data to the processor.

[0304] Based on the foregoing embodiments, an embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor of an electronic device, the medical data conversion method provided in any of the previous embodiments can be implemented.

[0305] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.

[0306] The methods disclosed in the various method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0307] The features disclosed in the various product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0308] The features disclosed in the various method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0309] It should be noted that the above-mentioned computer-readable storage medium can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory (Flash Memory), a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM) and other memories; it can also be various electronic devices including one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0310] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0311] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0312] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus necessary general hardware nodes, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0313] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0314] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0315] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0316] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A medical data conversion method, characterized in that: The method comprises: Obtaining raw medical data to be converted; Extracting semantic granular data from the original medical data to obtain a semantic data set; Get the resource parameter set contained in the rapid medical operational resource; Based on the resource parameter set, the semantic granular data in the semantic data set is mapped and converted to obtain target medical data.

2. The method according to claim 1, characterized in that The extracting of semantic granular data from the original medical data to obtain a semantic data set includes: Acquire medical object metadata associated with the original medical data; Segmenting the original medical data based on the medical object metadata to obtain medical object attribute data; The medical object attribute data is semantically segmented at a basic semantic unit granularity to obtain the semantic data set.

3. The method according to claim 1, characterized in that: The mapping and converting of the semantic granularity data in the semantic data set based on the resource parameter set to obtain target medical data includes: Associatively identifying the semantic granular data and at least part of the resource parameters in the resource parameter set through a graphic mapping engine, and obtaining a mapping association relationship between the semantic granular data and at least part of the resource parameters; Based on the mapping association relationship, the semantic granular data in the semantic data set is mapped and transformed to obtain the target medical data.

4. The method according to claim 3, characterized in that: The mapping conversion of the semantic granular data in the semantic data set based on the mapping association relationship to obtain the target medical data includes: determining a data structure of at least a portion of the resource parameters; Through the trained conversion model, based on the mapping association relationship and the data structure, the semantic granularity data is converted to obtain the target medical data.

5. The method according to claim 4, characterized in that The method further comprises: Acquire sample data; wherein the sample data includes multiple types of initial medical data; Acquire a sample mapping relationship between the sample data and the resource parameter set; The sample data is converted based on the sample mapping relationship by using a conversion model to obtain a sample conversion result; Acquire abnormal data contained in the sample conversion result; The model parameters of the conversion model are adjusted based on the abnormal data, and the trained conversion model is obtained according to the adjusted model parameters.

6. The method according to claim 5, characterized in that The adjusting the model parameters of the conversion model based on the abnormal data includes: Obtaining k-th abnormal data; wherein the k-th abnormal data is abnormal data contained in the k-th conversion result output by the conversion model after the k-th parameter adjustment; k is an integer greater than or equal to 1; Determine a kth abnormal type corresponding to the kth abnormal data; Determine that data associated with the kth abnormal data in the sample data is the k+1th training data; Based on the k+1th training data and the kth abnormality type, the model parameters of the conversion model after the kth parameter adjustment are adjusted to obtain the conversion model after the k+1th parameter adjustment.

7. The method according to claim 6, characterized in that The determining the kth abnormality type corresponding to the kth abnormal data includes: Obtaining a set of exception types of the rapid medical operational resource; Obtain kth abnormal description information of the kth abnormal data; An exception type in the exception type set that matches the kth exception description information is determined to be the kth exception type.

8. The method according to claim 5, characterized in that The obtaining of abnormal data contained in the sample conversion result includes: Obtain at least one configuration file; The sample conversion result is detected based on the at least one configuration file, and data in the sample conversion result that does not satisfy the configuration rule included in the at least one configuration file is determined as the abnormal data.

9. The method according to claim 5, characterized in that The adjusting the model parameters of the conversion model based on the abnormal data, and obtaining the trained conversion model according to the adjusted model parameters, includes: Obtaining the k+1th conversion result output by the conversion model after the k+1th parameter adjustment; wherein k is an integer greater than or equal to 1; If the amount of the k+1th abnormal data included in the k+1th conversion result is less than or equal to the target threshold, it is determined that the conversion model after the k+1th parameter adjustment is the trained conversion model.

10. A medical data conversion device, characterized in that: The device comprises: An acquisition module, used for acquiring original medical data to be converted; An extraction module, used for extracting semantic granularity data from the original medical data to obtain a semantic data set; The acquisition module is further used to acquire a resource parameter set included in the rapid medical operational resource; A conversion module is used to map and convert the semantic granular data in the semantic data set based on the resource parameter set to obtain target medical data.

11. An electronic device, characterized in that: The electronic device includes a processor and a memory; a computer program is stored in the memory; when the computer program is executed by the processor, the medical data conversion method as described in any one of claims 1 to 9 can be implemented.

12. A computer-readable storage medium, characterized in that: The storage medium stores a computer program; when the computer program is executed by a processor of an electronic device, it can implement the medical data conversion method as described in any one of claims 1 to 9.

Images