Intelligent medical platform data management method and system

Abstract

The invention relates to the field of intelligent data analysis, and provides an intelligent medical platform data management method and system, and the method comprises the steps: carrying out the semantic feature extraction of doctor inquiry text data on an intelligent medical platform through employing a data analysis technology based on artificial intelligence, so as to obtain inquiry semantic features, meanwhile, semantic feature extraction is conducted on patient reply text data on the intelligent medical platform to obtain symptom description semantic features, then the relation between the inquiry semantic features and the symptom description semantic features is established in the same semantic space, and finally an electronic medical record of the patient is obtained through a generator. Therefore, by automatically generating the electronic medical record, the accuracy of the data in the electronic medical record can be improved, and the standardization and consistency of the electronic medical record can be realized.

Description

Technical Field

[0001] The present application relates to the field of intelligent data analysis, and more specifically, to an intelligent medical platform data management method and system. Background Art

[0002] In the current medical field, traditional medical data management methods often rely on manual recording. Although this method can meet basic recording needs to a certain extent, its limitations are gradually revealed with the increase in medical needs and the popularization of remote consultation functions on medical platforms. The efficiency of manual recording is relatively low, and information omissions or errors are prone to occur during the recording process. These problems not only affect the accuracy of the data, but also bring considerable challenges to subsequent data analysis and utilization. Especially in the scenario of remote consultation, communication between medical workers and patients is mainly carried out through electronic devices, which requires that the recording and management of medical data must be more accurate and efficient. However, since different medical workers have different recording habits and formats, this personalized difference will lead to data inconsistency, which in turn affects data integration and analysis.

[0003] Therefore, an optimized intelligent medical platform data management solution is needed. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present application provides an intelligent medical platform data management method and system.

[0005] According to one aspect of the present application, a method for managing data of an intelligent medical platform is provided, which includes:

[0006] Obtain doctor inquiry text data and patient response text data on the intelligent medical platform;

[0007] Performing feature encoding on the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector;

[0008] An electronic medical record of the patient is generated based on the query-response semantic association feature vector.

[0009] According to another aspect of the present application, an intelligent medical platform data management system is provided, which includes:

[0010] Generate an electronic medical record-related data collection module to obtain doctor inquiry text data and patient response text data on the intelligent medical platform;

[0011] Generate an electronic medical record related data feature module, which is used to perform feature encoding on the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector;

[0012] The electronic medical record result generation module is used to generate the patient's electronic medical record based on the query-response semantic association feature vector.

[0013] This application has significant technical effects due to the adoption of the above technical solutions:

[0014] The intelligent medical platform data management method and system provided by the present application adopts data analysis technology based on artificial intelligence, extracts semantic features from doctor inquiry text data on the intelligent medical platform to obtain inquiry semantic features, and extracts semantic features from patient reply text data on the intelligent medical platform to obtain symptom description semantic features, and then establishes the relationship between inquiry semantic features and symptom description semantic features in the same semantic space, and finally obtains the patient's electronic medical record through a generator. In this way, by automatically generating electronic medical records, the accuracy of data in the electronic medical record can be improved, and it helps to achieve standardization and consistency of electronic medical records. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 Flow chart of an intelligent medical platform data management method according to an embodiment of the present application.

[0017] Figure 2 Schematic diagram of the architecture of the intelligent medical platform data management method according to the embodiment of the present application.

[0018] Figure 3 The present invention is a flowchart of performing feature encoding on the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector in the intelligent medical platform data management method according to an embodiment of the present application.

[0019] Figure 4 The present invention is a flowchart of generating a patient's electronic medical record based on the query-response semantic association feature vector in the intelligent medical platform data management method according to an embodiment of the present application.

[0020] Figure 5 This is a system block diagram of an intelligent medical platform data management system according to an embodiment of the present application.

[0021] Figure 6 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0023] Traditional medical data management relies on manual recording, which is inefficient and may cause information omissions or errors during the manual recording process. In addition, different medical workers have different recording habits and formats, which will lead to data inconsistency and make subsequent data analysis and utilization difficult. Therefore, an intelligent medical platform data management solution is expected.

[0024] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks provides new solutions and solutions for data management of intelligent medical platforms.

[0025] Figure 1 Flow chart of an intelligent medical platform data management method according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of the intelligent medical platform data management method according to an embodiment of the present application. Figure 1 and Figure 2 As shown, the intelligent medical platform data management method according to the embodiment of the present application includes: S110, obtaining doctor's inquiry text data and patient's reply text data on the intelligent medical platform; S120, performing feature encoding on the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector; S130, generating the patient's electronic medical record based on the inquiry-reply semantic association feature vector.

[0026] Specifically, in the technical solution of the present application, the doctor's inquiry text data and the patient's reply text data are obtained on the intelligent medical platform. It should be understood that the doctor will first ask the patient's basic information, such as name, age, gender, contact information, past medical history, allergy history, etc. Then the doctor will ask the patient in detail about the current symptoms, including the time of symptom onset, duration, severity, location, nature (such as whether the pain is dull or sharp), accompanying symptoms (such as fever, cough, nausea, etc.), etc., which are key clues for diagnosing the disease. In addition, the doctor will also ask whether the patient has a history of similar diseases or other hereditary diseases in the family and the patient's previous treatment and medication, so as to understand the patient's disease evolution and treatment effect. The patient's reply text data is the patient's response to the doctor's inquiry, usually including the confirmation of basic information, and a detailed description of his own symptoms, including the specific manifestations of the symptoms, duration, impact on daily life, etc. The patient also needs to answer the doctor's family history and medical history according to his actual situation.

[0027] In general, the doctor's inquiry text data is one of the core contents of building a patient's electronic medical record. After word segmentation, semantic coding and association analysis, these data will be integrated into the electronic medical record. The key content in the patient's reply text directly reflects the patient's health status and changes in condition, and is an indispensable part of building an electronic medical record. After semantic coding and association analysis, these key contents will be integrated into the electronic medical record, and will be mutually verified with the doctor's inquiry text data to form a comprehensive description of the patient's health status. In particular, in an embodiment of the present application, one feasible way to obtain the doctor's inquiry text data and the patient's reply text data is to obtain them from the database of the intelligent medical platform.

[0028] Figure 3 The flowchart is a method for managing data on an intelligent medical platform according to an embodiment of the present application, in which the doctor's inquiry text data and the patient's reply text data are feature encoded to obtain a query-reply semantic association feature vector. Figure 3 As shown, the feature encoding of the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector includes: S121, performing semantic feature extraction on the doctor's inquiry text data to obtain an inquiry semantic feature vector; S122, performing semantic feature extraction on the patient's reply text data to obtain a symptom description semantic feature vector; S123, associating the inquiry semantic feature vector with the symptom description semantic feature vector to obtain the inquiry-reply semantic association feature vector.

[0029] The doctor consultation text data obtained from the intelligent medical platform is often a continuous sequence of strings. It is very difficult for computers to directly process such data. In order to reduce the difficulty of subsequent semantic processing of doctor consultation text, it is first necessary to perform word segmentation on the doctor inquiry text data. Different vocabulary combinations may have completely different meanings. Through word segmentation, it can be ensured that each vocabulary unit in the consultation text is correctly identified and processed, thereby improving the processing accuracy. The vocabulary units after word segmentation usually have clearer semantic meanings, which makes semantic analysis based on vocabulary units possible. Then, the doctor consultation text data after word segmentation is input into the inquiry text context encoder to obtain the inquiry semantic feature vector.

[0030] The essential structure of the inquiry text context encoder in this application is a first context encoder based on a transformer including an embedding layer. Those skilled in the art should understand that the embedding layer is a commonly used technique in natural language processing for mapping vocabulary into a high-dimensional vector space. Each vector (i.e., an embedding vector) in this vector space represents the semantic information of the corresponding vocabulary. The purpose of the embedding layer is to capture the semantic similarity between vocabulary so that the vocabulary corresponding to similar vectors in the vector space is also semantically similar. In the doctor's inquiry text data, the embedding layer can convert each word after segmentation into an embedding vector of fixed dimension, providing a basis for subsequent processing. The core of the transformer structure lies in the self-attention mechanism, which allows the model to take into account other words in the entire input sequence when processing each word, thereby capturing long-distance dependencies in the input sequence. In the doctor's inquiry text data, the transformer structure can process the text sequence after segmentation, capture the contextual information in the doctor's inquiry through the self-attention mechanism, and then generate a more accurate inquiry semantic feature vector. In the doctor's inquiry text data, the converter structure can process the keyword embedding vector sequence obtained by the embedding layer after word segmentation, and then capture the contextual information in the input word embedding vector sequence through the self-attention mechanism, thereby generating a more accurate inquiry semantic feature vector.

[0031] Using the converter structure to process the doctor's inquiry text data after word segmentation can solve the problem of gradient vanishing or gradient exploding that traditional recurrent neural network or long short-term memory neural network models encounter when processing long sequences. This allows the converter to capture long-distance dependencies when processing long texts. In addition, the converter structure can realize parallel computing and improve the computing efficiency of the model.

[0032] In the technical solution of the present application, after the doctor's inquiry text data is segmented, a possible implementation method for obtaining an inquiry semantic feature vector through an inquiry text context encoder may be: performing word segmentation processing on the doctor's inquiry text data to obtain a plurality of doctor's inquiry text content keywords; passing the plurality of doctor's inquiry text content keywords through an embedding layer to convert each of the plurality of doctor's inquiry text content words into a doctor's inquiry text content keyword embedding vector to obtain a plurality of doctor's inquiry text content keyword embedding vectors; arranging the plurality of doctor's inquiry text content keyword embedding vectors into a doctor's inquiry content input vector; converting the doctor's inquiry content input vector into a doctor's inquiry content query vector and a doctor's inquiry content key vector respectively through a learnable embedding matrix; calculating the The product of the query vector of the doctor's inquiry content and the transposed vector of the doctor's inquiry content key vector is used to obtain the doctor's inquiry content self-attention association matrix; the doctor's inquiry content self-attention association matrix is ​​normalized to obtain the doctor's inquiry content standardized self-attention association matrix; the doctor's inquiry content standardized self-attention association matrix is ​​input into the Softmax activation function for activation to obtain the doctor's inquiry content self-attention feature matrix; the doctor's inquiry content self-attention feature matrix is ​​multiplied with each doctor's inquiry text content keyword embedding vector in the multiple doctor's inquiry text content keyword embedding vectors as the value vector to obtain multiple doctor's inquiry content context feature vectors; the multiple doctor's inquiry content context feature vectors are cascaded to obtain the inquiry semantic feature vector.

[0033] In the medical field, patient reply texts often contain rich symptom description information. In order to split this information into easy-to-process vocabulary units, the patient reply text data can be segmented. In particular, in the embodiments of the present application, a trained segmentation model, such as jieba segmentation, HanLP, etc., is generally used to segment the patient reply text. Considering that the embedding layer can map the segmented vocabulary units to a high-dimensional vector space, the semantic information of the corresponding vocabulary can be obtained. In order to deeply understand the meaning of each word and the contextual relationship between words, the attention mechanism of the converter can be used to perform context semantic encoding on multiple keyword embedding vectors that pass through the embedding layer, so as to accurately understand the complete semantics of the patient's symptom description. That is, the patient reply text data after segmentation is passed through the symptom description text context semantic encoder to obtain the symptom description semantic feature vector. The present application includes a second context encoder based on the embedding layer. Through the processing of the encoder, the patient reply text segmentation result is converted into a symptom description semantic feature vector, which not only contains the semantic information of the key words in the patient reply text, but also incorporates the context dependency, providing strong support for the subsequent generation of electronic medical records.

[0034] Next, the inquiry semantic feature vector and the symptom description semantic feature vector are associated to obtain the inquiry-response semantic association feature vector. It should be understood that inquiry and symptom description are two core parts in doctor-patient communication. Inquiries are usually initiated by doctors to obtain information such as the patient's specific symptoms and medical history; while symptom descriptions are direct feedback from patients on their health status. Associating the semantic feature vectors of the two can more accurately understand the patient's actual condition and reduce information bias caused by unclear expressions or misunderstandings. And by associating the semantic features of inquiry and symptom description, a closer semantic association network can be constructed. This association is not limited to literal matching, but also includes semantic logical relationships and contextual connections, which helps to better understand the deep meaning of the conversation. The association results of inquiry and symptom description can be directly reflected in the electronic medical record, making the medical record content more complete and accurate.

[0035] In an embodiment of the present application, a possible implementation method of associating the inquiry semantic feature vector with the symptom description semantic feature vector to obtain the inquiry-response semantic association feature vector may be: associating the inquiry semantic feature vector with the symptom description semantic feature vector using the following association fusion formula to obtain the inquiry-response semantic association feature vector; wherein the association fusion formula is:

[0036]

[0037] Among them, V co is the query-response semantic association feature vector, V a is the query semantic feature vector, V b A semantic feature vector describing the symptom, represents the addition of the elements at corresponding positions of the inquiry semantic feature vector and the symptom description semantic feature vector, and α and β are weighted parameters for controlling the balance between the inquiry semantic feature vector and the symptom description semantic feature vector in the inquiry-response semantic association feature vector.

[0038] Figure 4 The flowchart of generating the electronic medical record of a patient based on the query-response semantic association feature vector in the intelligent medical platform data management method according to the embodiment of the present application is as follows. Figure 4 As shown, the generation of the patient's electronic medical record based on the query-response semantic association feature vector includes: S131, optimizing the query-response semantic association feature vector based on the class probability distribution of the learning predictive matching to obtain an optimized query-response semantic association feature vector; S132, passing the optimized query-response semantic association feature vector through an electronic medical record generator to obtain a generation result, and the generation result is the patient's electronic medical record.

[0039] In particular, in the technical solution of the present application, on the intelligent medical platform, when the inquiry-response semantic association feature vector is processed by the electronic medical record generator to generate the patient's electronic medical record, there may be a problem of category harmony deviation, which affects the generation iteration effect. Specifically, if the text data of the doctor's inquiry and the patient's reply is incomplete or erroneous, it will directly affect the accuracy of word segmentation and context encoding, and then affect the quality of the feature vector and the generation of the electronic medical record. The word segmentation process may cause information loss due to insufficient recognition of professional terms, abbreviations or new words. The context encoder may not be able to fully understand the complex semantic relationships in the text. If the predetermined class label of the electronic medical record generator does not completely match the actual electronic medical record structure, or the class label setting is too broad or narrow, the generated electronic medical record may not meet the actual needs. In order to improve the accuracy and iterative effect of electronic medical record generation, in the technical solution of the present application, the inquiry-response semantic association feature vector is optimized based on the class probability distribution of learning predictive matching to obtain an optimized inquiry-response semantic association feature vector.

[0040] Specifically, in an embodiment of the present application, the query-response semantic association feature vector is optimized based on the class probability distribution of the learning predictive matching to obtain an optimized query-response semantic association feature vector, including: performing probabilistic activation based on the Sigmoid function on the query-response semantic association feature vector to obtain a probabilistic query-response semantic association feature vector; inputting the query-response semantic association feature vector into the electronic medical record generator to obtain an initial class label probability value; calculating the difference between the feature value of each position in the probabilistic query-response semantic association feature vector and the initial class label probability value to obtain a difference feature vector, and passing the difference feature vector through a mask function to obtain an initial weight vector for viewpoint metric prediction; calculating Calculate the autocorrelation matrix of the viewpoint metric prediction initial weight vector, divide the autocorrelation matrix of the viewpoint metric prediction initial weight vector by the square root of the length of the viewpoint metric prediction initial weight vector, and then multiply it by the viewpoint metric prediction initial weight vector through a softmax function to obtain the viewpoint metric prediction attention-oriented weight vector; perform positional point multiplication on the viewpoint metric prediction attention-oriented weight vector and the query-response semantic association feature vector to obtain a first weighted query-response semantic association feature vector; multiply a predetermined weight hyperparameter by the query-response semantic association feature vector, and then add the first weighted query-response semantic association feature vector by position to obtain the optimized query-response semantic association feature vector.

[0041] In an embodiment of the present application, another expression of performing class probability distribution optimization based on learning predictive matching on the query-response semantic association feature vector to obtain an optimized query-response semantic association feature vector may be: processing the query-response semantic association feature vector using the following formula to obtain the optimized query-response semantic association feature vector; wherein the formula is:

[0042] V cp =sigmoid(V co )

[0043]

[0044] Among them, V co represents the query-response semantic association feature vector, sigmoid represents the activation function, V cp represents the probabilistic query-response semantic association feature vector, P represents the initial class label probability value, represents positional subtraction, mask represents the mask function, x represents the variable of the input mask function, θ represents the predetermined threshold, V e represents the initial weight vector for viewpoint metric prediction, represents matrix multiplication, T represents the transpose of the vector, L represents the length of the initial weight vector for viewpoint metric prediction, Softmax represents the normalized exponential function, V ea represents the viewpoint metric prediction attention-oriented weight vector, α represents the predetermined weight hyperparameter, It means point multiplication by position. Indicates adding by position, V co ' represents the optimized query-response semantic association feature vector.

[0045] That is, in the technical solution of the present application, the inquiry-response semantic association feature vector is optimized for class probability distribution based on learning predictive matching, which first implements probabilistic activation based on the Sigmoid function on the inquiry-response semantic association feature vector to obtain a probabilistic inquiry-response semantic association feature vector, and inputs the inquiry-response semantic association feature vector into an electronic medical record generator to obtain an initial class label probability value, and initializes the viewpoint metric prediction contribution of each position in the canonical probabilistic inquiry-response semantic association feature vector to a predetermined class probability label through inverse error analysis and masking.

[0046] Then, the viewpoint space of the reverse optimization is expanded by predicting the autocorrelation attention representation of the initial weight vector with the viewpoint metric, and using its sub-dimensional consistency as the attention-guided weight, the reverse consistency attention mechanism of the sub-dimensional distribution is applied to the query-response semantic association feature vector. In this way, the forward feature manifold and the backward class probability distribution are learned and predicted based on the viewpoint metric prediction of post-machine learning, so that the optimized query-response semantic association feature vector has a relatively good distribution consistency with respect to the class probability distribution of the electronic medical record generator, so as to improve its generation accuracy by the electronic medical record generator.

[0047] Finally, the optimized query-response semantic association feature vector is passed through the electronic medical record generator to obtain a generated result, which is the patient's electronic medical record. It should be understood that the optimized query-response semantic association feature vector already contains rich semantic information and contextual relationships. Passing this information to the electronic medical record generator can ensure that the generated electronic medical record content accurately reflects the patient's actual situation and reduces errors caused by human factors. The electronic medical record generator generally adopts a generator structure based on an automatic codec. The generated electronic medical record records the patient's medical history, symptoms, diagnosis results, treatment plan and other information in detail, provides doctors with comprehensive medical information, and helps doctors make accurate diagnosis and treatment decisions. In this way, electronic medical records with consistent format and easy analysis can be automatically generated, and machine-generated medical records reduce errors and omissions caused by human factors and improve the accuracy of data.

[0048] In summary, the data management method of the intelligent medical platform based on the embodiment of the present application is explained, which adopts data analysis technology based on artificial intelligence, extracts semantic features from the doctor's inquiry text data on the intelligent medical platform to obtain inquiry semantic features, and extracts semantic features from the patient's reply text data on the intelligent medical platform to obtain symptom description semantic features, and then establishes the relationship between the inquiry semantic features and the symptom description semantic features in the same semantic space, and finally obtains the patient's electronic medical record through the generator. In this way, by automatically generating electronic medical records, the accuracy of the data in the electronic medical record can be improved, and it helps to achieve the standardization and consistency of the electronic medical record.

[0049] Figure 5 : is a system block diagram of an intelligent medical platform data management system according to an embodiment of the present application. Figure 5As shown, according to the embodiment of the present application, the intelligent medical platform data management system 100 includes: a module 110 for generating electronic medical record related data collection, which is used to obtain doctor's inquiry text data and patient's reply text data on the intelligent medical platform; a module 120 for generating electronic medical record related data features, which is used to perform feature encoding on the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector; an electronic medical record result generation module 130, which is used to generate the patient's electronic medical record based on the inquiry-reply semantic association feature vector.

[0050] Here, those skilled in the art can understand that the specific functions and operations of each unit and module in the above-mentioned intelligent medical platform data management system 100 have been referred to above. Figures 1 to 4 The method has been introduced in detail in the description of the intelligent medical platform data management method, and therefore, its repeated description will be omitted.

[0051] In summary, the intelligent medical platform data management system 100 based on the embodiment of the present application is explained, which adopts data analysis technology based on artificial intelligence, extracts semantic features from the doctor's inquiry text data on the intelligent medical platform to obtain inquiry semantic features, and extracts semantic features from the patient's reply text data on the intelligent medical platform to obtain symptom description semantic features, and then establishes the relationship between the inquiry semantic features and the symptom description semantic features in the same semantic space, and finally obtains the patient's electronic medical record through the generator. In this way, by automatically generating electronic medical records, the accuracy of the data in the electronic medical record can be improved, and it helps to achieve the standardization and consistency of the electronic medical record.

[0052] As described above, the intelligent medical platform data management system 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for intelligent medical platform data management. In one example, the intelligent medical platform data management system 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the intelligent medical platform data management system 100 can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the intelligent medical platform data management system 100 can also be one of the many hardware modules of the wireless terminal.

[0053] Alternatively, in another example, the intelligent medical platform data management system 100 and the wireless terminal may also be separate devices, and the intelligent medical platform data management system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0054] Below, reference Figure 6 To describe an electronic device according to an embodiment of the present application. Figure 6 1 is a block diagram of an electronic device according to an embodiment of the present application. For example, the electronic device 10 can be provided as a server or a terminal device. Figure 6 The electronic device 10 includes a processing component 11, which further includes one or more processors, and a memory resource represented by a memory 12 for storing instructions executable by the processing component 11, such as an application. The application stored in the memory 12 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 11 is configured to execute instructions to perform the above method.

[0055] The electronic device 10 may further include a power supply component 13 configured to perform power management of the electronic device 10, a wired or wireless network interface 14 configured to connect the electronic device 10 to a network, and an input / output (I / O) interface 15. The electronic device 10 may operate based on an operating system stored in the memory 12, such as Windows Server™, Mac OSX™, Unix™, Linux™, FreeBSD™ or the like.

[0056] In an embodiment of the present application, a computer-readable storage medium is also provided, such as a memory 12 including computer program instructions. The computer program instructions can be executed by a processing component 11 of an electronic device 10 to complete the above method.

[0057] The present application may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer program instructions for causing a processor to implement various aspects of the present application.

[0058] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0059] The computer program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer program instructions from the network and forwards the computer program instructions for storage in a computer-readable storage medium in each computing / processing device.

[0060] The computer program instructions for performing the operation of the present application can be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The computer program instructions can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer, partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet). In some embodiments, by using the state information of the computer program instructions to personalize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA) or a programmable logic array (PLA), the electronic circuit can execute the computer program instructions, thereby realizing various aspects of the present application.

[0061] Various aspects of the present application are described herein with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer program instructions.

[0062] These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0063] Computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

Claims

1. A method for managing data on an intelligent medical platform, characterized in that: include: Obtain doctor inquiry text data and patient response text data on the intelligent medical platform; Performing feature encoding on the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector; An electronic medical record of the patient is generated based on the query-response semantic association feature vector.

2. The intelligent medical platform data management method according to claim 1, characterized in that: The doctor's inquiry text data and the patient's reply text data are encoded with semantic features to obtain an inquiry-reply semantic association feature vector, including: Performing semantic feature extraction on the doctor's inquiry text data to obtain an inquiry semantic feature vector; Performing semantic feature extraction on the patient response text data to obtain a symptom description semantic feature vector; The inquiry semantic feature vector and the symptom description semantic feature vector are associated to obtain the inquiry-response semantic association feature vector.

3. The intelligent medical platform data management method according to claim 2, characterized in that: The semantic feature extraction is performed on the doctor's inquiry text data to obtain an inquiry semantic feature vector, including: segmenting the doctor's inquiry text data and passing it through an inquiry text context encoder to obtain the inquiry semantic feature vector.

4. The intelligent medical platform data management method according to claim 3, characterized in that: The semantic feature extraction is performed on the patient response text data to obtain a symptom description semantic feature vector, including: after the patient response text data is segmented, it is passed through a symptom description text context semantic encoder to obtain the symptom description semantic feature vector.

5. The intelligent medical platform data management method according to claim 4, characterized in that: The query text context encoder is a first transformer-based context encoder including an embedding layer, and the symptom description text context semantic encoder is a second transformer-based context encoder including an embedding layer.

6. The intelligent medical platform data management method according to claim 5, characterized in that: Generating the patient's electronic medical record based on the query-response semantic association feature vector includes: Optimizing the query-response semantic association feature vector based on the class probability distribution of the learning predictive matching to obtain an optimized query-response semantic association feature vector; The optimized query-response semantic association feature vector is passed through an electronic medical record generator to obtain a generated result, which is the electronic medical record of the patient.

7. The intelligent medical platform data management method according to claim 6, characterized in that: The query-response semantic association feature vector is optimized based on the class probability distribution of the learning predictive matching to obtain an optimized query-response semantic association feature vector, comprising: Performing probabilistic activation based on a Sigmoid function on the inquiry-response semantic association feature vector to obtain a probabilistic inquiry-response semantic association feature vector; Inputting the query-response semantic association feature vector into the electronic medical record generator to obtain an initial class label probability value; Calculating the difference between the feature value of each position in the probabilistic query-response semantic association feature vector and the initial class label probability value to obtain a difference feature vector, and passing the difference feature vector through a mask function to obtain an initial weight vector for viewpoint metric prediction; Calculating the autocorrelation matrix of the viewpoint metric prediction initial weight vector, dividing the autocorrelation matrix of the viewpoint metric prediction initial weight vector by the square root of the length of the viewpoint metric prediction initial weight vector, and then multiplying it by the viewpoint metric prediction initial weight vector through a softmax function to obtain a viewpoint metric prediction attention-oriented weight vector; Performing positional point multiplication of the viewpoint metric predicted attention-oriented weight vector and the query-response semantic association feature vector to obtain a first weighted query-response semantic association feature vector; The predetermined weight hyperparameter is multiplied by the query-response semantic association feature vector, and then the resultant is added to the first weighted query-response semantic association feature vector by position to obtain the optimized query-response semantic association feature vector.

8. An intelligent medical platform data management system, characterized in that: include: Generate an electronic medical record-related data collection module to obtain doctor inquiry text data and patient response text data on the intelligent medical platform; Generate an electronic medical record related data feature module, which is used to perform feature encoding on the doctor's inquiry text data and the patient's reply text data to obtain an inquiry-reply semantic association feature vector; The electronic medical record result generation module is used to generate the patient's electronic medical record based on the query-response semantic association feature vector.

9. The intelligent medical platform data management system according to claim 8, characterized in that: The module for generating electronic medical record-related data features includes: A doctor's inquiry text semantic feature extraction unit, used for extracting semantic features from the doctor's inquiry text data to obtain an inquiry semantic feature vector; A patient reply text semantic feature extraction unit, used for extracting semantic features from the patient reply text data to obtain a symptom description semantic feature vector; The inquiry-response semantic association feature generating unit is used to associate the inquiry semantic feature vector with the symptom description semantic feature vector to obtain the inquiry-response semantic association feature vector.

10. The intelligent medical platform data management system according to claim 9, characterized in that: The electronic medical record result generating module comprises: A feature optimization unit, configured to optimize the query-response semantic association feature vector based on the class probability distribution of the learning predictive matching to obtain an optimized query-response semantic association feature vector; The feature parsing unit is used to pass the optimized query-response semantic association feature vector through an electronic medical record generator to obtain a generation result, and the generation result is the electronic medical record of the patient.

Images