Medical data processing method and device, electronic device and storage medium

By constructing the target disease co-occurrence matrix and semantic analysis, the disease representation data is generated, and the problems of high data labeling cost and differentiated doctors' writing habits in existing medical data processing methods are solved, and the accuracy of medical data processing and semantic analysis capabilities are improved. .

CN114613462BActive Publication Date: 2024-11-19PING AN TECH (SHENZHEN) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210234178.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-03-09
Publication Date
2024-11-19
Estimated Expiration
2042-03-09

AI Technical Summary

Technical Problem

The existing medical data processing methods rely on the named entity recognition model of neural networks, requiring a large amount of labeled training data. The professional nouns in the medical field are highly professional and have high labeling costs, which affects the accuracy of model training. Doctors' writing habits make entities different Identification of difficult contexts affects the accuracy of processing medical data.

Method used

A medical data processing method is proposed, by obtaining original medical data, performing word segmentation processing and constructing the target disease co-occurrence matrix, mapping the data to vector space, semantic analysis is performed using a pre-trained text understanding model, calculating the loss value of the disease phrase, and generating Condition characterization data to improve processing accuracy.

Benefits of technology

It improves the processing accuracy of medical data, can more accurately explain the corresponding disease characteristics of medical data, reduces dependence on labeled data, and enhances the semantic analysis ability of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114613462B_ABST
    Figure CN114613462B_ABST
Patent Text Reader

Abstract

The present application provides a medical data processing method and device, electronic device and storage medium, belonging to the field of artificial intelligence and digital medical technology. The method includes: obtaining the first medical data and the second medical data in the original medical data; performing word segmentation processing on the first medical data to obtain the disease character; constructing the target disease co-occurrence matrix according to the medical dictionary and the disease character; mapping the second medical data to the preset vector space to obtain the first text vector; splicing the first text vector with the preset random perturbation factor to obtain the second text vector; performing semantic analysis on the first text vector and the second text vector through the text understanding model to obtain the first text hidden feature and the second text hidden feature; calculating the loss value of the disease phrase through the disease feature, the first text hidden feature and the second text hidden feature, and obtaining the disease representation data according to the loss value. The present application can improve the accuracy of processing medical data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

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

[0002] At present, there are few data processing methods for the study of medical text understanding in the medical field. Commonly used medical data processing methods often need to rely on neural network named entity recognition models, which often require a large amount of labeled training data. However, the medical field data has a strong professionalism, high labeling costs, and a small number of sample medical data with high accuracy, which affects the accuracy of model training. At the same time, differences in doctors' writing habits will also make it difficult for named entity recognition models to classify and identify entities in context, affecting the accuracy of medical data processing and making it impossible to accurately interpret the disease characteristics corresponding to the medical data. Therefore, how to improve the accuracy of medical data processing has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The main purpose of the embodiments of the present application is to propose a medical data processing method and device, an electronic device and a storage medium, aiming to improve the accuracy of processing medical data.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a medical data processing method, the method comprising:

[0005] Acquire original medical data; wherein the original medical data includes first medical data and second medical data;

[0006] Performing word segmentation processing on the first medical data to obtain disease condition characters;

[0007] Constructing a target disease condition co-occurrence matrix according to a preset medical dictionary and the disease condition characters, wherein the target disease condition co-occurrence matrix includes disease condition phrases and disease condition features corresponding to the disease condition phrases;

[0008] Mapping the second medical data to a preset vector space to obtain a first text vector;

[0009] The first text vector is concatenated with a preset random disturbance factor to obtain a second text vector;

[0010] Performing semantic analysis on the first text vector through a pre-trained text understanding model to obtain a first text hidden feature, and performing semantic analysis on the second text vector through the text understanding model to obtain a second text hidden feature;

[0011] The loss value of the disease condition phrase is calculated by using the disease condition feature, the first text hidden feature, and the second text hidden feature, and the disease condition representation data is obtained according to the loss value.

[0012] In some embodiments, the step of performing word segmentation processing on the first medical data to obtain condition characters includes:

[0013] Performing word segmentation processing on the first medical data according to preset text category labels to obtain multiple condition label fields;

[0014] The condition label field is segmented to obtain the condition characters.

[0015] In some embodiments, the step of constructing a target disease co-occurrence matrix based on a preset medical dictionary and the disease character includes:

[0016] Traversing the medical dictionary according to the disease condition characters to obtain assignment parameters corresponding to a preset co-occurrence matrix;

[0017] Filling the preset co-occurrence matrix with elements according to the assigned parameters to obtain an initial disease co-occurrence matrix;

[0018] The initial disease condition co-occurrence matrix is ​​normalized to obtain the target disease condition co-occurrence matrix.

[0019] In some embodiments, the step of mapping the second medical data to a preset vector space to obtain a first text vector includes:

[0020] Obtaining characteristic dimension parameters of the vector space;

[0021] The second medical data is mapped to the vector space according to the feature dimension parameter to obtain the first text vector.

[0022] In some embodiments, the text understanding model includes an embedding layer, a convolutional layer, a pooling layer, and a fully connected layer. The steps of performing semantic analysis on the first text vector through the pre-trained text understanding model to obtain the first text hidden feature, and performing semantic analysis on the second text vector through the text understanding model to obtain the second text hidden feature include:

[0023] Performing word embedding processing on the first text vector through the embedding layer to obtain a first text embedding vector, and performing word embedding processing on the second text vector through the embedding layer to obtain a second text embedding vector;

[0024] Performing convolution processing on the first text embedding vector through the convolution layer to obtain a first text convolution vector, and performing convolution processing on the second text embedding vector through the convolution layer to obtain a second text convolution vector;

[0025] Performing pooling processing on the first text convolution vector through the pooling layer to obtain a first text pooling vector, and performing pooling processing on the second text convolution vector through the pooling layer to obtain a second text pooling vector;

[0026] The first text pooling vector is semantically analyzed by the prediction function of the fully connected layer to obtain the first text hidden feature, and the first text pooling vector is semantically analyzed by the prediction function to obtain the second text hidden feature.

[0027] In some embodiments, the step of calculating the loss value of the condition phrase by using the condition feature, the first text hidden feature, and the second text hidden feature, and obtaining the condition representation data according to the loss value includes:

[0028] Calculating the loss value according to the disease condition feature, the first text hidden feature, the second text hidden feature and a preset loss function;

[0029] According to the loss value, calculating the model loss value of the text understanding model;

[0030] The disease condition phrases are screened according to the model loss value to obtain the disease condition representation data.

[0031] In some embodiments, before the steps of performing semantic analysis on the first text vector by using a pre-trained text understanding model to obtain a first text hidden feature, and performing semantic analysis on the second text vector by using the text understanding model to obtain a second text hidden feature, the method further includes pre-training the text understanding model, specifically including:

[0032] Acquire sample medical data; wherein the sample medical data is text data;

[0033] Inputting the sample medical data into the text understanding model;

[0034] Extracting features from the sample medical data using the text understanding model to obtain sample disease characteristics;

[0035] Concatenating the sample disease condition characteristic with the random disturbance factor to obtain a sample disease condition characteristic vector;

[0036] Calculating the noise standard deviation between the sample disease feature vectors through the loss function of the text understanding model;

[0037] The loss function is optimized according to the noise standard deviation to update the text understanding model.

[0038] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a medical data processing device, the device comprising:

[0039] A data acquisition module, used to acquire original medical data; wherein the original medical data includes first medical data and second medical data;

[0040] A word segmentation module, used for performing word segmentation processing on the first medical data to obtain disease condition characters;

[0041] A matrix construction module, used to construct a target disease co-occurrence matrix according to a preset medical dictionary and the disease characters, wherein the target disease co-occurrence matrix includes disease phrases and disease features corresponding to the disease phrases;

[0042] A mapping module, used for mapping the second medical data to a preset vector space to obtain a first text vector;

[0043] A concatenation module, used for concatenating the first text vector with a preset random disturbance factor to obtain a second text vector;

[0044] A semantic analysis module, configured to perform semantic analysis on the first text vector through a pre-trained text understanding model to obtain a first text hidden feature, and perform semantic analysis on the second text vector through the text understanding model to obtain a second text hidden feature;

[0045] The calculation module is used to calculate the loss value of the disease condition phrase through the disease condition feature, the first text hidden feature and the second text hidden feature, and obtain the disease condition representation data according to the loss value.

[0046] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program, when executed by the processor, implements the method described in the first aspect above.

[0047] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a storage medium, which is a computer-readable storage medium used for computer-readable storage, and the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method described in the first aspect above.

[0048] The medical data processing method and device, electronic device and storage medium proposed in the present application obtain original medical data; wherein the original medical data includes first medical data and second medical data. Then, the first medical data is segmented to obtain disease characters, and a target disease co-occurrence matrix is ​​constructed according to a preset medical dictionary and disease characters, wherein the target disease co-occurrence matrix includes disease phrases and disease features corresponding to disease phrases, and the first medical data can be processed at the single word level to determine the fields that may constitute the phrases, thereby achieving a rough division of disease phrases. Then, the second medical data is mapped to a preset vector space to obtain a first text vector, the first text vector is concatenated with a preset random perturbation factor to obtain a second text vector, and the first text vector and the second text vector are semantically analyzed respectively by a pre-trained text understanding model to obtain the first text hidden features and the second text hidden features, so that the semantic features of the sentences in the second medical data can be better captured, thereby determining the importance of different characters and different word segments in the entire sentence. Finally, the loss value of the disease condition phrase is calculated through the disease condition characteristics, the first text hidden features, and the second text hidden features, and the disease condition representation data is obtained based on the loss value, which can improve the accuracy of processing medical data and enable the disease condition characteristics corresponding to the medical data to be more accurately explained through the disease condition representation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a flow chart of a medical data processing method provided in an embodiment of the present application;

[0050] Figure 2 yes Figure 1 Flow chart of step S102 in FIG.

[0051] Figure 3 yes Figure 1 Flow chart of step S103 in FIG.

[0052] Figure 4 yes Figure 1 Flow chart of step S104 in FIG.

[0053] Figure 5 is another flow chart of the medical data processing method provided by an embodiment of the present application;

[0054] Figure 6 yes Figure 1 Flow chart of step S106 in FIG.

[0055] Figure 7 yes Figure 1 Flow chart of step S107 in FIG.

[0056] Figure 8 is a structural schematic diagram of a medical data processing device provided in an embodiment of the present application;

[0057] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. 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.

[0059] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0061] First, some nouns involved in this application are analyzed:

[0062] Artificial intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. AI is a branch of computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing and expert systems. AI can simulate the information process of human consciousness and thinking. AI is also a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0063] Natural language processing (NLP): NLP uses computers to process, understand and apply human languages ​​(such as Chinese, English, etc.). NLP is a branch of artificial intelligence and an interdisciplinary subject between computer science and linguistics. It is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, and text understanding. Natural language processing is often used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.

[0064] Information Extraction (NER): A text processing technology that extracts specified types of entity, relationship, event and other factual information from natural language text and forms structured data output. Information extraction is a technology that extracts specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and chapters. Text information is composed of some small specific units, such as characters, words, phrases, sentences, paragraphs, or a combination of these specific units. Extracting noun phrases, names, place names, etc. from text data are all text information extraction. Of course, the information extracted by text information extraction technology can be various types of information.

[0065] Encoder: It converts the input sequence into a vector of fixed length.

[0066] Long Short-Term Memory (LSTM): It is a time-recurrent neural network specially designed to solve the long-term dependency problem of general RNN (recurrent neural network). All RNNs have a chain form of repeated neural network modules. In standard RNN, this repeated structural module has only a very simple structure, such as a tanh layer. LSTM is a type of neural network containing LSTM blocks or other. In literature or other materials, LSTM blocks may be described as intelligent network units because they can remember values ​​of indefinite lengths of time. There is a gate in the block that can determine whether the input is important enough to be remembered and whether it can be output.

[0067] Bi-directional Long Short-Term Memory (Bi-LSTM): It is composed of forward LSTM and backward LSTM. It is often used to model contextual information in natural language processing tasks. Bi-LSTM combines the information of the input sequence in both the forward and backward directions based on LSTM. For the output at time t, the forward LSTM layer has the information at time t and before in the input sequence, while the backward LSTM layer has the information at time t and after in the input sequence. The output of the forward LSTM layer at time t is denoted as, and the output of the backward LSTM layer at time t is denoted as. The vectors output by the two LSTM layers can be processed by addition, average, or concatenation.

[0068] With the improvement of economic level, people will inevitably pay more attention to their health status and have higher and higher requirements for medical services. Existing medical services are limited by various factors such as resources and management, and it is difficult to meet people's growing needs. Intelligent medical care is becoming more and more important, and making full use of the knowledge in medical texts can accelerate the process of intelligent medical care.

[0069] At present, there are few studies on text understanding in the medical field. Commonly used data processing methods for medical texts often need to rely on neural network named entity recognition models, which often require a large amount of labeled training data. However, the specific nouns in medical data are highly professional and the labeling cost is high, resulting in less accurate labeled data and a lack of large-scale medical text data sets, which affects the accuracy of model training. At the same time, due to the large differences in doctors' writing habits, current entity recognition models are difficult to classify entities in context and identify medical entities, which affects the accuracy of processing medical data. Therefore, how to improve the accuracy of processing medical data has become a technical problem that needs to be solved urgently.

[0070] Based on this, the embodiments of the present application provide a medical data processing method and device, an electronic device and a storage medium, aiming to improve the accuracy of understanding medical texts.

[0071] The medical data processing method and device, electronic device and storage medium provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the medical data processing method in the embodiments of the present application is described.

[0072] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0073] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0074] The embodiments of the present application can realize the analysis of the patient's historical condition based on medical cloud technology. Among them, medical cloud refers to the use of "cloud computing" to create a medical and health service cloud platform based on new technologies such as cloud computing, mobile technology, multimedia, 4G communication, big data, and the Internet of Things, combined with medical technology, to achieve the sharing of medical resources and the expansion of medical scope. Because of the application and combination of cloud computing technology, the medical cloud improves the efficiency of medical institutions and facilitates residents to seek medical treatment. For example, the current hospital appointment registration, electronic medical records, medical insurance, etc. are all the products of the combination of cloud computing and the medical field. The medical cloud also has the advantages of data security, information sharing, dynamic expansion, and overall layout.

[0075] The medical data processing method provided in the embodiment of the present application relates to the field of artificial intelligence and digital medical technology. The medical data processing method provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or can be configured as a server cluster or a distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the medical data processing method, etc., but is not limited to the above forms.

[0076] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0077] Figure 1 is an optional flowchart of the medical data processing method provided in the embodiment of the present application. Figure 1 The medical data processing method may include but is not limited to steps S101 to S107.

[0078] Step S101, obtaining original medical data; wherein the original medical data includes first medical data and second medical data;

[0079] Step S102, performing word segmentation processing on the first medical data to obtain disease condition characters;

[0080] Step S103, constructing a target disease co-occurrence matrix according to a preset medical dictionary and disease characters, wherein the target disease co-occurrence matrix includes disease phrases and disease features corresponding to the disease phrases;

[0081] Step S104, mapping the second medical data to a preset vector space to obtain a first text vector;

[0082] Step S105, concatenating the first text vector with a preset random disturbance factor to obtain a second text vector;

[0083] Step S106, performing semantic analysis on the first text vector through a pre-trained text understanding model to obtain a first text hidden feature, and performing semantic analysis on the second text vector through the text understanding model to obtain a second text hidden feature;

[0084] Step S107, calculating the loss value of the disease condition phrase through the disease condition feature, the first text hidden feature, and the second text hidden feature, and obtaining the disease condition representation data according to the loss value.

[0085] In steps S101 to S107 of the embodiment of the present application, by performing word segmentation processing on the first medical data, obtaining the disease condition characters, and constructing the target disease condition co-occurrence matrix according to the preset medical dictionary and the disease condition characters, the first medical data can be processed at the single word level, the fields that may constitute the phrases can be determined, and the rough division of the disease condition phrases can be achieved. The first text vector and the second text vector are semantically analyzed respectively by the pre-trained text understanding model to obtain the first text hidden features and the second text hidden features, which can better capture the sentence semantic features in the second medical data, thereby determining the importance of different characters and different word segments in the entire sentence. Finally, the loss value of the disease condition phrase is calculated through the disease condition features, the first text hidden features, and the second text hidden features, and the disease condition representation data is obtained according to the loss value, which can improve the accuracy of processing medical data, so that users can more accurately interpret the disease condition features corresponding to the medical data through the disease condition representation data.

[0086] In step S101 of some embodiments, a web crawler can be written to crawl data in a targeted manner after setting the data source to obtain the original medical data. It should be noted that the original medical data includes the first medical data and the second medical data, and the first medical data and the second medical data include the patient's basic information (such as gender, age, chief complaint, etc.), past medical history, medical records, medication records, and condition analysis data, etc. The first medical data and the second medical data can be the same medical data or different medical data. In order to improve data processing efficiency, the first medical data and the second medical data in the embodiment of the present application are the same medical data.

[0087] In a possible implementation, the above-mentioned original medical data may be an electronic healthcare record, which is an electronic personal health record, including a series of electronic records with preservation value such as medical records, electrocardiograms, and medical images.

[0088] See also Figure 2 In some embodiments, step S102 may include but is not limited to steps S201 to S202:

[0089] Step S201, performing word segmentation processing on the first medical data according to a preset text category label to obtain a plurality of condition label fields;

[0090] Step S202, segmenting the condition label field to obtain condition characters.

[0091] In step S201 of some embodiments, the first medical data is segmented by a pre-trained Jieba segmenter to obtain multiple disease label fields. When using the Jieba segmenter for segmentation, a directed acyclic graph corresponding to the first medical data is first generated by comparing the preset dictionary in the Jieba segmenter, and then the shortest path on the directed acyclic graph is found according to the preset selection mode, the preset dictionary and the text category label, and the first medical data is intercepted according to the shortest path, or the first medical data is directly intercepted to obtain multiple disease label segments.

[0092] In step S202 of some embodiments, the condition label field is segmented by a decoder and part-of-speech category, field length, etc. in a preset U-Net network to obtain a plurality of condition characters.

[0093] Through the above steps S201 to S202, the first medical data can be segmented into single disease condition characters. For example, "patient, 63 years old, chest pain for 2 days" is processed as: "patient", "patient", "6", "3", "years old", "chest", "pain", "2", "days".

[0094] See also Figure 3 In some embodiments, step S103 may include but is not limited to steps S301 to S303:

[0095] Step S301, traversing the medical dictionary according to the disease character to obtain the assignment parameters corresponding to the preset co-occurrence matrix;

[0096] Step S302, filling the preset co-occurrence matrix with elements according to the assigned parameters to obtain an initial disease co-occurrence matrix;

[0097] Step S303: normalize the initial disease co-occurrence matrix to obtain a target disease co-occurrence matrix.

[0098] In step S301 of some embodiments, the preset co-occurrence matrix is ​​an all-zero matrix M, and the pre-constructed medical dictionary may include medical names, medical terms, non-medical names, folk names, international medical terms, etc. related to various medical pathologies. By traversing this medical dictionary according to the disease condition characters, medical specific names can be listed, and the entity features contained in the disease condition characters can be identified. The entity features may include the above-mentioned medical names, medical terms, non-medical names, folk names, international medical terms, modifiers, time information and other entity words in multiple dimensions related to medical pathologies. Assuming that the size of the medical dictionary is w, then the size of the preset co-occurrence matrix is ​​w×w, and then the medical dictionary is traversed according to the disease condition characters to obtain the assignment parameters corresponding to the preset co-occurrence matrix. It should be explained that the index of the medical dictionary is all the disease condition characters of all the first medical data, and the vector representation corresponding to the disease condition character obtained by the index is obtained.

[0099] It should be explained that, when the above preset co-occurrence matrix is ​​initialized, all elements are assigned parameters of 0; after traversing all the disease characters, the matrix element a of the preset co-occurrence matrix is ij It represents the number of times a condition phrase consisting of the i-th condition character and the j-th condition character appearing in sequence appears in all the first medical data.

[0100] For example, suppose we have two pieces of first medical data: "patient 63 years old, chest pain for 2 days" and "patient male, cough for 2 weeks", then the disease characters obtained from these two pieces of first medical data are: "patient", "patient", "6", "3", "years old", "chest", "pain", "2", "days", "male", "cough", "cough", "2", "weeks", and each disease character will get an initialized vector representation. For another example, if the disease character of the input first medical data is "chest pain for 3 days, hypertension for 1 year", the assignment parameter corresponding to the preset co-occurrence matrix is ​​1.

[0101] In step S302 of some embodiments, the preset co-occurrence matrix is ​​filled with elements according to the assigned parameters, and each assigned parameter is added to the corresponding position in the preset co-occurrence matrix to obtain an initial disease co-occurrence matrix.

[0102] For example, if the condition characters of the first medical data input are "chest pain for 3 days, hypertension for 1 year", the assignment parameter corresponding to the preset co-occurrence matrix is ​​1, and 1 is added to the corresponding position of the preset co-occurrence matrix, that is, first find out which column "chest" is in the preset co-occurrence matrix, then see which row "pain" is in, and then add 1 to the value of the corresponding position; then find out which column "pain" is in and which row "3" is in, and then add 1 to the corresponding position of the preset co-occurrence matrix.

[0103] In step S303 of some embodiments, the initial disease co-occurrence matrix is ​​normalized to [0, 1] to obtain the target disease co-occurrence matrix, that is, the initial disease co-occurrence matrix is ​​processed according to column normalization to obtain the target disease co-occurrence matrix. Specifically, each column of the initial disease co-occurrence matrix is ​​summed, and then the elements of the current column are divided by the sum of the current column. This process is equivalent to calculating the probability that the disease character in the current column and all other disease characters form a phrase.

[0104] For example, the column corresponding to the word "chest" appears 10 times in the row "stuffy", 39 times in the row "pain", 1 time in the row "abdomen", and 0 times in other rows. After column normalization, the probability of "chest" and "stuffy" forming a phrase becomes 0.2, the probability of "chest" and "pain" forming a phrase becomes 0.78, and the probability of "chest" and "abdomen" forming a phrase becomes 0.02. In this regard, a probability threshold can be set in advance to determine whether all situations with probabilities greater than the probability threshold can form a disease phrase that meets the requirements. For example, if the probability threshold is set to 0.1, the words starting with "chest" can be obtained as "chest tightness" and "chest pain", while "chest and abdomen" are eliminated.

[0105] Furthermore, when constructing a disease phrase according to the target disease co-occurrence matrix, the column characters of the target disease co-occurrence matrix are set as the first level and the row characters as the second level. When reconstructing the disease phrase, the row characters and column characters corresponding to the mark position 1 are extracted, and the column characters are placed in the first position and the row characters in the second position to form the disease phrase. By extracting features of the disease phrase through a sequence classifier or the like, the disease features corresponding to the disease phrase can be easily obtained.

[0106] See also Figure 4 In some embodiments, step S104 may include but is not limited to steps S401 to S402:

[0107] Step S401, obtaining characteristic dimension parameters of the vector space;

[0108] Step S402: Map the second medical data to a vector space according to the feature dimension parameter to obtain a first text vector.

[0109] In step S401 of some embodiments, a feature dimension parameter of the vector space is obtained, and the feature dimension parameter can be set according to actual needs. For example, the feature dimension parameter can be 256, 512, etc., but is not limited thereto.

[0110] In step S402 of some embodiments, the second medical data is mapped from the semantic space to the vector space once or multiple times through the MLP network, and the second medical data is mapped to a pre-set vector space to obtain a first text vector x=[x1, x2, ..., xi ,…,x n ], so that the dimension of the first text vector is the same as the feature dimension parameter. The MLP network includes a linear layer, a ReLu activation function and a linear layer.

[0111] In step S105 of some embodiments, when the first text vector is concatenated with the preset random perturbation factor, first, the preset random perturbation factor is word-embedded to obtain a random perturbation vector, and then the first text vector is concatenated with the random perturbation vector, that is, the first text vector and the random perturbation vector are vector-added to obtain a second text vector x′=x+∈, wherein the preset random perturbation factor can be represented by a vector as ∈, ∈~N(0,r 2 I), the larger the r, the greater the first text vector x=[x1,x2,…,x i ,…,x n ]The more information that needs to be lost, the less important the first text vector is.

[0112] See also Figure 5 In some embodiments, before step S106, the medical data processing method further includes pre-training a text understanding model, specifically including steps S501 to S506:

[0113] Step S501, obtaining sample medical data; wherein the sample medical data is text data;

[0114] Step S502, inputting sample medical data into a text understanding model;

[0115] Step S503, extracting features from the sample medical data through the text understanding model to obtain sample disease characteristics;

[0116] Step S504, concatenating the sample disease characteristics with the random disturbance factor to obtain a sample disease characteristic vector;

[0117] Step S505, calculating the noise standard deviation between the sample disease feature vectors through the loss function of the text understanding model;

[0118] Step S506, optimizing the loss function according to the noise standard deviation to update the text understanding model.

[0119] In step S501 of some embodiments, a web crawler can be written to crawl data in a targeted manner after setting the data source to obtain sample medical data. It should be noted that the sample medical data is text data, including the patient's basic information, past medical history, medical records, medication records, and condition analysis data.

[0120] In step S502 of some embodiments, the sample medical data is input into a text understanding model, wherein the text understanding model is a TextCNN model, and the text understanding model includes an embedding layer, a convolutional layer, a fully connected layer, and a transformer layer.

[0121] In step S503 of some embodiments, feature extraction is performed on the sample medical data through the convolutional layer of the text understanding model to capture the semantic features of the sample medical data and obtain the sample disease characteristics.

[0122] In step S504 of some embodiments, when the sample disease feature and the random disturbance factor are concatenated, first, the preset random disturbance factor is word-embedded to obtain a random disturbance vector ∈, and then the sample disease feature z is concatenated with the random disturbance vector ∈, that is, the sample disease feature and the random disturbance vector are vector-added to obtain a sample disease feature vector z′=z+∈, wherein the preset random disturbance factor can be represented by a vector as ∈, ∈~N(0,r 2 I), the random disturbance factor follows a normal distribution N, where N is 0 as the mean and r 2 I is the standard deviation of a normal distribution.

[0123] In step S505 of some embodiments, when calculating the noise standard deviation r between the sample disease feature vectors through the loss function of the text understanding model, the loss function adding the random perturbation factor can be defined as shown in formula (1):

[0124]

[0125] Among them, s is the sample disease characteristic z=[z1,z2,…,z i ,…,z n ] The corresponding hidden state vector after classification processing, H(z′ i |s) is used to approximately reflect the sample condition feature z corresponding to the i-th sample condition character during forward propagation i The amount of semantic information lost by s in .

[0126] In addition, in order to improve the accuracy of data processing, after adding the random perturbation factor, it is necessary to ensure that the original vector representation is not affected as much as possible. Therefore, the mean square error loss function is used to constrain the random perturbation factor, which is defined as shown in formula (2):

[0127]

[0128] Among them, ′ is the sample disease feature vector obtained after adding the random perturbation factor to the sample disease feature, T(z′) is the hidden layer state vector corresponding to the sample disease feature vector z′ after classification processing, and s is the hidden layer state vector corresponding to the sample disease feature z after classification processing. ∈ is the expected value. Since the random disturbance factor follows a normal distribution, E ∈ Specifically, it refers to the expected value of the mean square error of the hidden layer vector when random disturbance factors and random disturbance factors are added when ∈ obeys the normal distribution.

[0129] In step S506 of some embodiments, when optimizing the loss function according to the noise standard deviation r, the loss function value loss of the loss function is first calculated, and then the loss function value is back-propagated using the gradient descent method, and the loss function value is fed back to the text understanding model, and the model parameters of the text understanding model are modified, and the above process is repeated until the loss function value meets the preset iteration condition, wherein the preset iteration condition is that the number of iterations can reach the preset value, or the variance of the loss function is less than the preset threshold. When the loss function value meets the preset iteration condition, the back propagation can be stopped, and the final model parameters are used as the final model parameters to complete the update of the text understanding model.

[0130] See also Figure 6 In some embodiments, step S106 includes but is not limited to steps S601 to S604:

[0131] Step S601, performing word embedding processing on the first text vector through the embedding layer to obtain the first text embedding vector, and performing word embedding processing on the second text vector through the embedding layer to obtain the second text embedding vector;

[0132] Step S602, performing convolution processing on the first text embedding vector through a convolution layer to obtain a first text convolution vector, and performing convolution processing on the second text embedding vector through a convolution layer to obtain a second text convolution vector;

[0133] Step S603, performing pooling processing on the first text convolution vector through a pooling layer to obtain a first text pooling vector, and performing pooling processing on the second text convolution vector through a pooling layer to obtain a second text pooling vector;

[0134] Step S604, performing semantic analysis on the first text pooling vector through the prediction function of the fully connected layer to obtain the first text hidden feature, and performing semantic analysis on the first text pooling vector through the prediction function to obtain the second text hidden feature.

[0135] In step S601 of some embodiments, the first text vector and the second text vector are respectively subjected to word embedding processing by a long short-term memory algorithm (LSTM algorithm), a bidirectional long short-term memory algorithm (Li-LSTM algorithm) or a transformer algorithm of an embedding layer to obtain a first text embedding vector and a second text embedding vector. For example, the first text vector is subjected to word embedding processing by the Li-LSTM algorithm in a sequence from left to right and from right to left to obtain a first text embedding vector, and the second text vector is subjected to word embedding processing by the Li-LSTM algorithm in a sequence from left to right and from right to left to obtain a second text embedding vector.

[0136] In step S602 of some embodiments, the first text embedding vector is convolved through a convolution layer to capture the semantic features of the first text embedding vector and obtain the first text convolution vector; and the second text embedding vector is convolved through a convolution layer to capture the semantic features of the second text embedding vector and obtain the second text convolution vector.

[0137] In step S603 of some embodiments, the first text convolution vector is subjected to maximum pooling or average pooling through a pooling layer to obtain a first text pooling vector; and the second text convolution vector is subjected to maximum pooling or average pooling through a pooling layer to obtain a second text pooling vector.

[0138] In step S604 of some embodiments, the prediction function of the fully connected layer can be a softmax function or other functions without limitation. Taking the softmax function as an example, when the first text pooling vector is semantically analyzed by the softmax function, a probability distribution is created on each reference text category label for the first text pooling vector by the softmax function, and the relevant probability value of each reference text category is obtained, and the reference text category with the largest relevant probability value is selected as the first text hidden feature; when the second text pooling vector is semantically analyzed by the softmax function, a probability distribution is created on each reference text category label for the second text pooling vector by the softmax function, and the relevant probability value of each reference text category is obtained, and the reference text category with the largest relevant probability value is selected as the second text hidden feature.

[0139] See also Figure 7 In some embodiments, step S107 may also include but is not limited to steps S701 to S703:

[0140] Step S701, calculating a loss value according to the disease feature, the first text hidden feature, the second text hidden feature and a preset loss function;

[0141] Step S702, calculating a model loss value of the text understanding model according to the loss value;

[0142] Step S703, screening the disease phrases according to the model loss value to obtain disease representation data.

[0143] In step S701 of some embodiments, first, according to the noise standard deviation r obtained from the aforementioned steps S505 to S506, the loss function L1 of adding random disturbance factors to the disease feature, the first text hidden feature, and the second text hidden feature is calculated, and is defined as shown in formula (3):

[0144]

[0145] Where s is the first text hidden feature X = [X1, X2, ..., X i ,…,X n ] The corresponding hidden state vector after classification processing, H(X′ i |s) is used to approximately reflect the first text hidden feature X corresponding to the i-th disease character during forward propagation i The amount of semantic information lost by s in .

[0146] Furthermore, in order to improve the accuracy of data processing, after adding the random perturbation factor, it is necessary to ensure that the original vector representation is not affected as much as possible. Therefore, the mean square error loss function is used. The random disturbance factor is constrained and defined as shown in formula (4):

[0147]

[0148] Where X is the second text hidden feature obtained by adding a random perturbation factor to the first text hidden feature, T(X′) is the hidden layer state vector corresponding to the second text hidden feature X′ after classification processing, and s is the hidden layer state vector corresponding to the first text hidden feature X after classification processing. ∈ is the expected value. Since the random disturbance factor follows a normal distribution, E ∈ Specifically, it refers to the expected value of the mean square error of the hidden layer vector when random disturbance factors and random disturbance factors are added when ∈ obeys the normal distribution.

[0149] Since only the above two loss functions (L1, ) is used for calculation, and only the importance of a single disease character can be obtained. However, most of the original medical data are medical terms and common words. If the importance of several characters in a medical word is too different, it will lead to a large deviation in the interpretation of the medical text. Therefore, this application also calculates the loss value according to the disease feature, the first text hidden feature, the second text hidden feature and the preset vocabulary loss function. The word-level interpretation is promoted to the vocabulary-level interpretation. Specifically, the vocabulary loss function can be expressed as shown in formula (5):

[0150]

[0151] Among them, M (i+1)i Represents the corresponding element of the disease character in the target disease co-occurrence matrix. (i+1)i =1, it means that the two corresponding disease characters are likely to belong to the same word, and the random perturbation factors added to them should also be close.

[0152] In step S702 of some embodiments, the model loss value of the text understanding model is calculated according to the loss value. When , it can be calculated according to formula (6), which can be expressed as follows:

[0153]

[0154] Among them, α and β are the hyperparameters of the text understanding model, and α and β represent exist The proportion of .

[0155] In step S703 of some embodiments, when screening the disease phrases according to the model loss value, disease phrases in the target disease co-occurrence matrix that can make the model loss value fall within the preset loss value range are selected, and these disease phrases that meet the requirements are combined to obtain disease representation data.

[0156] The above steps S701 to S703 solve the problem that interpreting medical data only at the single-word level will lead to very different interpretations of the importance of several disease characters in the same disease phrase, which can improve the accuracy of processing medical data and enable the disease characteristics corresponding to the medical data to be more accurately interpreted through the disease representation data.

[0157] The medical data processing method of the embodiment of the present application obtains original medical data; wherein the original medical data includes first medical data and second medical data. Then, the first medical data is segmented to obtain disease characters, and a target disease co-occurrence matrix is ​​constructed according to a preset medical dictionary and disease characters, wherein the target disease co-occurrence matrix includes disease phrases and disease features corresponding to the disease phrases, and the first medical data can be processed at the single word level to determine the fields that may constitute the phrases, thereby achieving a rough division of the disease phrases. Then, the second medical data is mapped to a preset vector space to obtain a first text vector, the first text vector is concatenated with a preset random perturbation factor to obtain a second text vector, and the first text vector and the second text vector are semantically analyzed respectively by a pre-trained text understanding model to obtain the first text hidden features and the second text hidden features. In this way, the semantic features of the sentences in the second medical data can be better captured, thereby determining the importance of different characters and different word segments in the entire sentence. Finally, the loss value of the disease condition phrase is calculated through the disease condition characteristics, the first text hidden features, and the second text hidden features, and the disease condition representation data is obtained based on the loss value, which can improve the accuracy of processing medical data and enable the disease condition characteristics corresponding to the medical data to be more accurately explained through the disease condition representation data.

[0158] See also Figure 8 The present application also provides a medical data processing device that can implement the above-mentioned medical data processing method. The device includes:

[0159] The data acquisition module 801 is used to acquire original medical data; wherein the original medical data includes first medical data and second medical data;

[0160] The word segmentation module 802 is used to perform word segmentation processing on the first medical data to obtain disease condition characters;

[0161] A matrix construction module 803 is used to construct a target disease co-occurrence matrix according to a preset medical dictionary and disease characters, wherein the target disease co-occurrence matrix includes disease phrases and disease features corresponding to the disease phrases;

[0162] A mapping module 804, used for mapping the second medical data to a preset vector space to obtain a first text vector;

[0163] A concatenation module 805 is used to concatenate the first text vector with a preset random disturbance factor to obtain a second text vector;

[0164] A semantic analysis module 806, configured to perform semantic analysis on the first text vector using a pre-trained text understanding model to obtain a first text hidden feature, and perform semantic analysis on the second text vector using the text understanding model to obtain a second text hidden feature;

[0165] The calculation module 807 is used to calculate the loss value of the disease condition phrase through the disease condition feature, the first text hidden feature and the second text hidden feature, and obtain the disease condition representation data according to the loss value.

[0166] The specific implementation of the medical data processing device is basically the same as the specific implementation of the above-mentioned medical data processing method, and will not be repeated here.

[0167] The specific implementation of the medical data processing device is basically the same as the specific implementation of the above-mentioned medical data processing method, and will not be repeated here.

[0168] The embodiment of the present application also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned medical data processing method. The electronic device can be any intelligent terminal including a tablet computer, a vehicle-mounted computer, etc.

[0169] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0170] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0171] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the medical data processing method of the embodiment of this application;

[0172] Input / output interface 903, used to implement information input and output;

[0173] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized by wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);

[0174] A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0175] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0176] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium used for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the above-mentioned medical data processing method.

[0177] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0178] The medical data processing method, medical data processing device, electronic device and storage medium proposed in the present application obtain original medical data; wherein the original medical data includes first medical data and second medical data. Then, the first medical data is segmented to obtain disease characters, and a target disease co-occurrence matrix is ​​constructed according to a preset medical dictionary and disease characters, wherein the target disease co-occurrence matrix includes disease phrases and disease characteristics corresponding to disease phrases, and the first medical data can be processed at the single word level to determine the fields that may constitute the phrases, thereby achieving a rough division of disease phrases. Then, the second medical data is mapped to a preset vector space to obtain a first text vector, the first text vector is concatenated with a preset random perturbation factor to obtain a second text vector, and the first text vector and the second text vector are semantically analyzed respectively by a pre-trained text understanding model to obtain the first text hidden features and the second text hidden features, so that the semantic features of the sentences in the second medical data can be better captured, thereby determining the importance of different characters and different word segments in the entire sentence. Finally, the loss value of the disease condition phrase is calculated through the disease condition characteristics, the first text hidden features, and the second text hidden features, and the disease condition representation data is obtained based on the loss value, which can improve the accuracy of processing medical data and enable the disease condition characteristics corresponding to the medical data to be more accurately explained through the disease condition representation data.

[0179] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0180] It can be understood by those skilled in the art that Figure 1-7 The technical solutions shown in the figure do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown in the figure, or a combination of certain steps, or different steps.

[0181] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0182] Those skilled in the art will appreciate that all or some of the steps, systems, and functional modules / units in the medical data processing method disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof.

[0183] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0184] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0185] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0186] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0187] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0188] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the medical data processing method of each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0189] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A medical data processing method, characterized in that: The method comprises: Acquire original medical data; wherein the original medical data includes first medical data and second medical data; Performing word segmentation processing on the first medical data to obtain disease condition characters; Constructing a target disease condition co-occurrence matrix according to a preset medical dictionary and the disease condition characters, wherein the target disease condition co-occurrence matrix includes disease condition phrases and disease condition features corresponding to the disease condition phrases; Mapping the second medical data to a preset vector space to obtain a first text vector; The first text vector is concatenated with a preset random disturbance factor to obtain a second text vector; Performing semantic analysis on the first text vector through a pre-trained text understanding model to obtain a first text hidden feature, and performing semantic analysis on the second text vector through the text understanding model to obtain a second text hidden feature; Calculate the loss value of the condition phrase by using the condition feature, the first text hidden feature, and the second text hidden feature, and obtain condition representation data according to the loss value; Constructing a target disease co-occurrence matrix according to a preset medical dictionary and the disease character, including: Traversing the medical dictionary according to the disease character to obtain the assignment parameters corresponding to the preset co-occurrence matrix; filling the preset co-occurrence matrix with elements according to the assignment parameters to obtain the initial disease co-occurrence matrix; normalizing the initial disease co-occurrence matrix to obtain the target disease co-occurrence matrix; The text understanding model includes an embedding layer, a convolution layer, a pooling layer, and a fully connected layer. The first text vector is semantically analyzed by the pre-trained text understanding model to obtain a first text hidden feature, and the second text vector is semantically analyzed by the text understanding model to obtain a second text hidden feature, including: Performing word embedding processing on the first text vector through the embedding layer to obtain a first text embedding vector, and performing word embedding processing on the second text vector through the embedding layer to obtain a second text embedding vector; Performing convolution processing on the first text embedding vector through the convolution layer to obtain a first text convolution vector, and performing convolution processing on the second text embedding vector through the convolution layer to obtain a second text convolution vector; Performing pooling processing on the first text convolution vector through the pooling layer to obtain a first text pooling vector, and performing pooling processing on the second text convolution vector through the pooling layer to obtain a second text pooling vector; The first text pooling vector is semantically analyzed by the prediction function of the fully connected layer to obtain the first text hidden feature, and the first text pooling vector is semantically analyzed by the prediction function to obtain the second text hidden feature.

2. The medical data processing method according to claim 1, characterized in that: The step of performing word segmentation processing on the first medical data to obtain condition characters includes: Performing word segmentation processing on the first medical data according to preset text category labels to obtain multiple condition label fields; The condition label field is segmented to obtain the condition characters.

3. The medical data processing method according to claim 1, characterized in that: The step of mapping the second medical data to a preset vector space to obtain a first text vector includes: Obtaining characteristic dimension parameters of the vector space; The second medical data is mapped to the vector space according to the feature dimension parameter to obtain the first text vector.

4. The medical data processing method according to any one of claims 1 to 3, characterized in that: The step of calculating the loss value of the condition phrase by using the condition feature, the first text hidden feature, and the second text hidden feature, and obtaining the condition representation data according to the loss value includes: Calculating the loss value according to the disease condition feature, the first text hidden feature, the second text hidden feature and a preset loss function; According to the loss value, calculating the model loss value of the text understanding model; The disease condition phrases are screened according to the model loss value to obtain the disease condition representation data.

5. The medical data processing method according to any one of claims 1 to 3, characterized in that: Before the steps of performing semantic analysis on the first text vector by using a pre-trained text understanding model to obtain a first text hidden feature, and performing semantic analysis on the second text vector by using the text understanding model to obtain a second text hidden feature, the method further includes pre-training the text understanding model, specifically including: Acquire sample medical data; wherein the sample medical data is text data; Inputting the sample medical data into the text understanding model; Extracting features from the sample medical data using the text understanding model to obtain sample disease characteristics; Concatenating the sample disease condition characteristic with the random disturbance factor to obtain a sample disease condition characteristic vector; Calculating the noise standard deviation between the sample disease feature vectors through the loss function of the text understanding model; The loss function is optimized according to the noise standard deviation to update the text understanding model.

6. A medical data processing device, characterized in that: The device comprises: A data acquisition module, used to acquire original medical data; wherein the original medical data includes first medical data and second medical data; A word segmentation module, used for performing word segmentation processing on the first medical data to obtain disease condition characters; A matrix construction module, used to construct a target disease co-occurrence matrix according to a preset medical dictionary and the disease characters, wherein the target disease co-occurrence matrix includes disease phrases and disease features corresponding to the disease phrases; A mapping module, used for mapping the second medical data to a preset vector space to obtain a first text vector; A concatenation module, used for concatenating the first text vector with a preset random disturbance factor to obtain a second text vector; A semantic analysis module, configured to perform semantic analysis on the first text vector through a pre-trained text understanding model to obtain a first text hidden feature, and perform semantic analysis on the second text vector through the text understanding model to obtain a second text hidden feature; A calculation module, used to calculate the loss value of the condition phrase according to the condition feature, the first text hidden feature, and the second text hidden feature, and obtain condition representation data according to the loss value; Constructing a target disease co-occurrence matrix according to a preset medical dictionary and the disease character, including: Traversing the medical dictionary according to the disease character to obtain the assignment parameters corresponding to the preset co-occurrence matrix; filling the preset co-occurrence matrix with elements according to the assignment parameters to obtain the initial disease co-occurrence matrix; normalizing the initial disease co-occurrence matrix to obtain the target disease co-occurrence matrix; The text understanding model includes an embedding layer, a convolution layer, a pooling layer, and a fully connected layer. The first text vector is semantically analyzed by the pre-trained text understanding model to obtain a first text hidden feature, and the second text vector is semantically analyzed by the text understanding model to obtain a second text hidden feature, including: The first text vector is word-embedded by the embedding layer to obtain a first text embedding vector, and the second text vector is word-embedded by the embedding layer to obtain a second text embedding vector; the first text embedding vector is convolved by the convolution layer to obtain a first text convolution vector, and the second text embedding vector is convolved by the convolution layer to obtain a second text convolution vector; the first text convolution vector is pooled by the pooling layer to obtain a first text pooling vector, and the second text convolution vector is pooled by the pooling layer to obtain a second text pooling vector; the first text pooling vector is semantically analyzed by the prediction function of the fully connected layer to obtain the first text hidden feature, and the first text pooling vector is semantically analyzed by the prediction function to obtain the second text hidden feature.

7. An electronic device, characterized in that: The electronic device includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the medical data processing method as described in any one of claims 1 to 5 are realized.

8. A storage medium, the storage medium being a computer-readable storage medium, used for computer-readable storage, characterized in that: The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the medical data processing method described in any one of claims 1 to 5.