Remote training platform, method and equipment for magnetocardiography and storage medium

Abstract

The invention relates to the technical field of medical training, and discloses a remote training platform, method and device for magnetocardiography and a storage medium. The remote training platform comprises a cloud server and a database; the cloud server comprises a training content management module and a test evaluation module; the database stores the training content of the magnetocardiography, user learning data and a test question bank; the training content management module is used for acquiring user learning data from a database, acquiring knowledge points extracted from a known magnetocardiography knowledge base and knowledge points extracted from cases, and constructing a knowledge graph, and edge weight values of user nodes and knowledge point nodes of the knowledge graph are used for indicating the mastery degree of the user on the knowledge points; and the training content management module is also used for generating personalized training courses based on the edge weight values of the user nodes and the knowledge point nodes. According to the invention, personalized training courses can be generated in a targeted manner according to the knowledge mastering condition of the user, precise training is realized, and the learning efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical training technology, and in particular to a remote training platform, method, device and storage medium for magnetocardiographs. Background Art

[0002] Magnetocardiography (MCG) is a device used to noninvasively detect the weak magnetic fields generated by cardiac electrical activity. It utilizes superconducting quantum interference device (SQUID) technology to achieve highly sensitive measurements. It demonstrates unique advantages in the diagnosis of cardiovascular diseases such as coronary microcirculatory disorders and arrhythmias. Currently, the prevalence of magnetocardiography is relatively low, necessitating specialized training for medical professionals in its operation and diagnosis.

[0003] However, traditional training methods rely on offline lectures or on-site hands-on instruction, which are time- and space-intensive. Existing remote training platforms are mostly general-purpose education systems that lack specialized content specific to magnetocardiographs (such as device principles, operating procedures, and clinical case studies), failing to meet the precise learning needs of medical practitioners. Most importantly, the training content provided by existing platforms is uniform, requiring users to select content based on their needs (e.g., their knowledge base), resulting in poor training effectiveness. Summary of the Invention

[0004] In view of this, the present invention provides a remote training platform, method, device and storage medium for magnetocardiographs to solve the problem that existing remote training platforms cannot formulate personalized training content for users, resulting in poor training effects.

[0005] In a first aspect, the present invention provides a remote training platform for magnetocardiographs, comprising: a cloud server and a database;

[0006] The cloud server includes a training content management module and a test evaluation module;

[0007] The database stores the training content, user learning data and test question bank of the magnetocardiograph; the user learning data includes learning behavior data and other data generated during the learning process;

[0008] The training content management module is used to obtain the user learning data from the database, and obtain knowledge points extracted from the known magnetic cardiomedical knowledge base and knowledge points extracted from case records, and construct a knowledge graph, wherein the knowledge graph includes user nodes and knowledge point nodes, and the edge weight values ​​between the user nodes and the knowledge point nodes are used to indicate the user's mastery of the corresponding knowledge points;

[0009] The training content management module is further configured to generate personalized training courses based on the edge weight values ​​between the user nodes and the knowledge point nodes.

[0010] In some optional embodiments, the database stores medical records;

[0011] The training content management module is specifically used for:

[0012] Extracting structured case data from the database and generating case knowledge point feature vectors;

[0013] Inputting the case knowledge point feature vector into the knowledge graph as an associated case library of the case knowledge point node;

[0014] Acquiring learning data of the user when studying the case;

[0015] Based on the learning data when the user studies the case, the edge weight value between the user node and the target knowledge point node is updated; wherein the target knowledge point node is the case knowledge point node corresponding to the case.

[0016] In some optional specific implementations, the training content management module is specifically used to:

[0017] The learning data of the user when studying the case is converted into sliding window statistical features, and then embedded into the note text using Word2Vec to generate a semantic vector;

[0018] Using a pre-trained 3D ResNet-50 model, extracting deep features of the medical images in the case; extracting features of the case text from the case; and obtaining a feature vector of the case knowledge points based on the deep features and features of the case text;

[0019] Encoding the semantic vector generated by the learning data into an N-dimensional hidden state, compressing the case knowledge point feature vector to N dimensions through a fully connected layer, performing attention weight calculation, and weighted summing to generate an N-dimensional cross-modal feature vector to represent the user-case association strength;

[0020] Based on the cross-modal feature vector, the knowledge graph is constructed or updated.

[0021] In some optional specific embodiments, the cloud server further includes a case management module; the case management module is used to:

[0022] Obtain case text, perform semantic analysis on the case text based on a deep learning model, and extract target information, including symptoms, imaging features, and diagnostic conclusions;

[0023] Inputting the structured target information into an anomaly detection model, wherein the anomaly detection model is a self-supervised deep learning model based on contrastive learning;

[0024] Based on the output of the anomaly detection model, it is determined whether the case text is the target case text, and the target case text includes rare case text and misdiagnosed case text; the anomaly detection model automatically identifies the target case text by calculating the similarity between the target information of the case and the feature vector of the standard case library.

[0025] In some optional specific embodiments, the case management module is further used to:

[0026] Obtaining a medical knowledge graph, which is a pre-built general knowledge system based on existing medical theories, authoritative guidelines, industry standards, and academic literature. The nodes of the medical knowledge graph are determined based on magnetocardiograph principles and disease diagnosis standards. The nodes include diseases, symptoms, magnetocardiograph equipment, and / or magnetocardiograph technology, and edges represent logical relationships between nodes.

[0027] Based on the medical knowledge graph, the target information of the case text is verified for logical consistency, and the target information includes symptoms, imaging features and diagnostic conclusions.

[0028] In some optional specific embodiments, the cloud server further includes an expert review module, and the expert review module is used to obtain expert review opinions on the case;

[0029] The expert review module is embedded with an AI-assisted annotation tool to automatically generate review suggestions.

[0030] In some optional specific implementations, the remote training platform further includes: a user terminal;

[0031] The user terminal includes a remote training module, which is used to display links to various training contents in a classified manner. The training contents include at least one of the following: magnetocardiograph principles, magnetocardiograph operating specifications, and magnetocardiograph clinical case interpretations.

[0032] In a second aspect, the present invention provides a remote training method for a magnetocardiograph, comprising:

[0033] Obtain user learning data, as well as knowledge points extracted from a known magnetic cardiomedical knowledge base and knowledge points extracted from case histories, to construct or update a knowledge graph; wherein the knowledge graph includes user nodes and knowledge point nodes, and the edge weight values ​​between the user nodes and the knowledge point nodes are used to indicate the user's mastery of the corresponding knowledge points; the user learning data includes learning behavior data and other data generated during the learning process;

[0034] A personalized training course is generated based on the edge weight values ​​between the user node and the knowledge point node.

[0035] In a third aspect, the present invention provides a cloud server comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to thereby execute the remote training method for a magnetocardiograph according to the second aspect or any corresponding embodiment thereof.

[0036] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the remote training method for a magnetocardiograph according to the second aspect or any corresponding embodiment thereof.

[0037] In a fifth aspect, the present invention provides a computer program product comprising computer instructions for causing a computer to execute the remote training method for a magnetocardiograph according to the second aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in related technologies, the following briefly introduces the drawings required for use in the specific embodiments or related technical descriptions. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 is a schematic block diagram of the structure of a remote training platform for magnetocardiographs according to an embodiment of the present invention;

[0040] Figure 2 is a flow chart of a remote training method for a magnetocardiograph according to an embodiment of the present invention;

[0041] Figure 3 Schematic diagram of the hardware structure of the cloud server according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0043] The remote training platform provided by related technologies lacks a mechanism to track students' learning progress and knowledge mastery. Therefore, the training content it provides is unified. Training content that users have already learned and mastered well is still displayed in its original position. Customers need to filter the training content they currently need to learn based on their own knowledge mastery, which is inefficient.

[0044] The embodiment of the present invention provides a remote training platform for magnetocardiographs, such as Figure 1 As shown, it includes a cloud server 101 and a database 102;

[0045] The cloud server includes a training content management module and a test evaluation module;

[0046] The database stores the training content of the magnetocardiograph, user learning data, test question bank and answer data; specifically, the user learning data includes learning behavior data and other data generated during the learning process, for example, answer data (specifically including answering speed, modification records, and final answers), relevant data of user notes, and training content viewing data (specifically including the training content viewed, viewing time, number of repeated viewings, knowledge content when viewing is paused, pause time, content of repeatedly viewed video segments, etc.).

[0047] In addition, the database can also store user login information. User login information includes user identification information and authentication information such as login password. Training content includes case studies.

[0048] The training content management module is used to retrieve the user's learning data from the database, as well as knowledge points extracted from the known magnetic cardiomedical knowledge base and from case histories. It then constructs a knowledge graph based on the user's learning data and knowledge points. The knowledge graph comprises user nodes and knowledge point nodes, with the edge weights between these nodes indicating the user's mastery of the corresponding knowledge point. Specifically, knowledge nodes primarily come from two sources: the known magnetic cardiomedical knowledge base and case histories. When deriving knowledge nodes from case histories, a pre-trained deep learning model such as BERT can be used to perform semantic parsing on the case text to extract key information such as symptoms, imaging features, and diagnostic conclusions. A knowledge graph is a graph network composed of nodes and edges. The detailed construction process (and update process, which is the same as the construction process) is described below. The training content management module is also used to generate personalized training courses based on the edge weights between the user nodes and knowledge point nodes. By quantifying the correlation between the user's knowledge status and knowledge points, a learning content sequence is generated that prioritizes weak points, provides logical connections, and is updated in real time. The knowledge graph here is a dynamic personalized graph based on user learning behavior, which is used for personalized training content recommendation.

[0049] Specifically, the training content of the magnetocardiograph includes the principle of magnetocardiograph, the operating specifications of magnetocardiograph, the interpretation of clinical cases of magnetocardiograph (i.e. analysis), etc. The smaller the edge weight value between the user node and the knowledge point node, the lower the corresponding user's mastery of the corresponding knowledge point. Therefore, the knowledge points (i.e. weak points) with low user mastery can be screened out according to the edge weight value. For example, the knowledge points corresponding to the knowledge point nodes with edge weight values ​​lower than the preset threshold can be screened out, and then the training content corresponding to these knowledge points can be screened out to form a personalized training course. For example, the link corresponding to the screened training content can be simply pushed to the homepage of the user terminal for display. Alternatively, the homepage of the user terminal has a recommendation column, which displays the link to the training content of the corresponding knowledge point in the order of the edge weight value from small to large.

[0050] The remote training platform for magnetocardiographs provided by the embodiment of the present invention characterizes the user's mastery of various knowledge points by constructing a dynamic knowledge graph that changes with the user's learning progress, thereby generating personalized training courses in a targeted manner. This solves the problem of rigid recommendations and inability to adapt to individual differences in traditional remote training platforms, realizes precise training, and improves learning efficiency.

[0051] Precision Training: Improve physicians' diagnostic capabilities through a closed-loop "learning - testing - practice" approach. "Practice" here refers to diagnostic simulation training based on real clinical cases (such as image feature recognition and diagnostic conclusion deduction). This utilizes real case material to enhance trainees' (i.e., users') diagnostic application capabilities using magnetocardiographs.

[0052] In some optional embodiments, the database further stores cases;

[0053] The training content management module is specifically used for:

[0054] Extracting structured case data from the database and generating case knowledge point feature vectors;

[0055] Inputting the case knowledge point feature vector into the knowledge graph as an associated case library of the case knowledge point node;

[0056] Acquiring learning data of the user when studying the case;

[0057] Based on the learning data of the user when studying the case, the edge weight value between the user node and the target knowledge point node is updated; wherein the target knowledge point node is the case knowledge point node corresponding to the case. In other words, when the user studies the case, the case interaction behavior data (such as annotation trajectory, learning time) can be collected in real time, that is, the learning data, and used to update the edge weight value between the user node and the corresponding knowledge point node.

[0058] In some optional specific implementations, the training content management module is specifically used to:

[0059] The learning data when the user studies the case is converted into sliding window statistical features, and then Word2Vec is used to embed the note text to generate a semantic vector. Specifically, the learning data can be time series data recorded in chronological order. In this embodiment, the features are first counted through a sliding window, and then Word2Vec is used to embed the note text to generate a semantic vector, thereby converting the text information in the learning data into a structured and computable semantic representation, so as to participate in feature fusion and knowledge modeling later.

[0060] Using a pre-trained 3D ResNet-50 model, we extract deep features from the medical images in the case and extract features from the case text. Based on these deep features and the features of the case text, we obtain a feature vector for the case knowledge points. Extracting deep features from medical images is intended to enhance multimodal understanding capabilities.

[0061] The semantic vector generated by the learning data is encoded into an N (N is a positive integer, for example, 128) dimensional hidden state, the case knowledge point feature vector is compressed to N dimensions through a fully connected layer, and then the attention weight calculation is performed, and the weighted summation generates an N-dimensional cross-modal feature vector. The N-dimensional cross-modal feature vector characterizes the user-case association strength, reflecting the user's understanding and application ability of the knowledge points (such as symptoms, imaging features, diagnostic conclusions, etc.) involved in a specific case. Specifically, the association strength reflects the user's understanding and application ability of the knowledge points (such as symptoms, imaging features, diagnostic conclusions, etc.) involved in a specific case. In this embodiment, the 128-dimensional dimensional design can not only retain enough feature details (such as different case types, user learning weaknesses), but also quantify the user-case association strength through vector distance (such as cosine similarity), providing data support for knowledge graph construction and personalized training course generation.

[0062] Specifically, the semantic vector can be encoded using LSTM. The case feature vector includes the deep features of the medical image and the features of the case text (including diagnosis, symptoms, etc.). Specifically, the deep features of the medical image in the case can be extracted according to the above method, and then the case text is encoded. Finally, the feature representation (feature vector) of the case can be obtained by fusion, that is, the case knowledge point feature vector (also called case feature vector).

[0063] Based on the cross-modal feature vector (i.e., the fused feature), the knowledge graph is constructed or updated. This reflects the user's knowledge mastery status in real time, and dynamically adjusts the edge weights of the knowledge point nodes in the knowledge graph. Specifically, the knowledge graph can be constructed and updated using a graph convolutional network (GCN).

[0064] After the knowledge graph is constructed for the first time, the knowledge graph can be updated using real-time user behavior data as the main trigger point, combined with periodic scanning, to ensure that the knowledge graph is updated synchronously with the user's learning progress, thereby accurately generating personalized training courses.

[0065] The present invention constructs a knowledge graph that reflects the user's knowledge mastery status. By integrating user learning data with clinical case features through a multimodal deep learning model, the knowledge graph is provided with quantified feature inputs. This knowledge graph is then dynamically updated in real time (as the user progresses) through a graph convolutional network (GCN) to update the user's knowledge mastery status. This facilitates the generation of personalized learning paths, improves the accuracy of recommended content, shortens training cycles, and reduces clinical operation errors.

[0066] In an embodiment of the present invention, a multimodal deep learning model can be used to achieve feature fusion of user learning data (user learning behavior data, note data, answers to questions, etc.) and medical cases. The multimodal deep learning model can also generate cross-modal feature vectors through LSTM encoding and attention mechanism to drive dynamic updates of the knowledge graph (such as adjusting the user-knowledge point edge weights).

[0067] In summary, the construction and updating of the knowledge graph in the embodiment of the present invention adopts a method that combines a multimodal deep learning model with a graph convolutional network (GCN): the multimodal model is responsible for feature extraction and fusion of multi-source data, and the GCN is responsible for constructing and updating the node weights and edge relationships of the knowledge graph based on the fused features. The two work together to achieve dynamic representation of the user's knowledge state. In other words, the core function of the multimodal deep learning model is to fuse user learning data with clinical case characteristics to provide quantitative feature input for the knowledge graph, while the construction and updating of the knowledge graph is achieved by the graph convolutional network (GCN).

[0068] In some optional specific implementations, the cloud server further includes a case management module; the case management module is configured to:

[0069] Obtain case text and perform semantic parsing on it using a deep learning model to extract target information, including symptoms, imaging features, and diagnostic conclusions. Specifically, the deep learning model mentioned above can be a pre-trained BERT model. The case text can be existing in the database or newly acquired.

[0070] Inputting the structured target information into an anomaly detection model, wherein the anomaly detection model is a self-supervised deep learning model based on contrastive learning;

[0071] Based on the output of the anomaly detection model, it is determined whether the case text is the target case text, and the target case text includes the text of high-value cases such as rare case text and misdiagnosed case text; the anomaly detection model automatically identifies the target case text by calculating the similarity between the target information of the case and the feature vector of the standard case library.

[0072] In the embodiments of the present invention, rare cases, misdiagnosed cases, and other high-value cases contain clinically uncommon or error-prone diagnostic scenarios, enriching the depth and diversity of training content, allowing trainees to be exposed to specific disease application scenarios and avoid repeated misdiagnoses. At the same time, they are used as key data to construct knowledge graphs to generate more targeted training courses. In addition, the selected rare cases, misdiagnosed cases, and other high-value cases can also serve as core material for test question banks, making the assessment closer to clinical scenarios, and ultimately building a "learning-testing-training" closed-loop system, improving the pertinence and effectiveness of training, and achieving the goal of precision training.

[0073] In some optional specific implementations, the case management module is further used to:

[0074] Obtaining a medical knowledge graph; the medical knowledge graph is a general knowledge system pre-constructed based on existing medical theories, authoritative guidelines, industry standards, and academic literature. The nodes of the medical knowledge graph are based on medical concepts determined by magnetocardiograph principles, disease diagnosis standards, etc. The nodes include diseases, symptoms, magnetocardiograph equipment, and / or magnetocardiograph technology, and edges represent logical relationships between nodes.

[0075] The medical knowledge graph is used to verify the logical consistency of the target information corresponding to the case text.

[0076] Specifically, the medical knowledge graph can be constructed during the initialization of the remote training platform. Edges in the medical knowledge graph can represent logical relationships, such as the causal relationship between symptoms and diseases, or the subordination of equipment and technology. The medical knowledge graph is primarily used to verify the logical consistency of target information extracted from case texts, such as symptoms, imaging features, and diagnostic conclusions. This ensures that case data conforms to medical logic and provides underlying support for the accuracy of training content.

[0077] Key fields (i.e., target information) from the case can be extracted first, and then logical consistency can be verified based on the key fields. Specifically, logical consistency can be verified through Natural Language Processing (NLP) technology.

[0078] In the embodiment of the present invention, the process of verifying the case text mainly includes:

[0079] First, target information such as symptoms, imaging features, and diagnostic conclusions is extracted from the case text using a deep learning model (such as a pre-trained BERT). This information is then converted into structured data (such as keyword segments). This information is then matched and verified with the medical concepts (such as nodes such as "magnetocardiography principle" and "coronary microcirculation disorder") and logical relationships between nodes (such as the "symptom-disease" causal relationship and the "diagnostic conclusion-imaging feature" correspondence) stored in the pre-built medical knowledge graph. For example, if the imaging feature of "normal magnetic field distribution on the magnetocardiogram" in the case text has no corresponding logical association with the diagnostic conclusion of "myocardial ischemia" in the medical knowledge graph, it is determined to be logically inconsistent. This ensures the accuracy of the case data and provides underlying support for the correctness of the training content.

[0080] In an embodiment of the present invention, a pre-constructed medical knowledge graph is used to verify whether the target information (symptoms, imaging features, diagnostic conclusions) extracted from the case text conforms to the logical associations and industry standards in the medical field, thereby ensuring the accuracy and reliability of the case data.

[0081] If there is a logical conflict between the target information of the case text and the medical knowledge graph, and the conflict is not caused by a real clinical misdiagnosis (such as data entry error, feature extraction bias), the case will be judged as an "invalid case" and will not be included in the high-value case database;

[0082] If the conflict is a logical contradiction of a real clinical misdiagnosis (such as the conflict between "normal magnetocardiogram" and "heart failure diagnosis" in the medical knowledge graph), the misdiagnosed case will still be retained because it can provide students with a typical error case reference and strengthen diagnostic logic training.

[0083] In some optional specific implementations, the cloud server also includes an expert review module, which is used to obtain expert review opinions on the case; specifically, the expert review module includes an expert review interface to facilitate experts to input review opinions and display expert review opinions.

[0084] The expert review module is embedded with AI-assisted annotation tools to automatically generate review suggestions, such as marking lesion areas and recommending similar cases, thereby improving the review efficiency and quality of magnetocardiograph training materials and enhancing the pertinence and scientific nature of the training content.

[0085] That is, the expert review module in this embodiment of the present invention incorporates a lesion annotation tool and similar case recommendation function. The purpose of annotating lesion areas and recommending similar cases is to improve the efficiency and quality of the review of magnetocardiograph training materials and strengthen the pertinence and scientific nature of the training content.

[0086] Expert review is only conducted on cases that have passed logical consistency verification and are of high value, ensuring that the review objects are both of educational significance (high value) and consistent with medical logic (verified), thereby guaranteeing the accuracy and practicality of the training materials.

[0087] In the embodiments of the present invention, a multi-level intelligent review process is used to overcome the bottleneck of low efficiency of manual review and the difficulty in comprehensively extracting multi-dimensional information such as symptoms, imaging features, and diagnostic conclusions when manually annotating case texts, thereby improving the efficiency of case review and the ability to screen high-value training cases.

[0088] In summary, in the embodiments of the present invention, by combining expert manual review and artificial intelligence assistance, multi-dimensional key information such as symptoms, imaging features and diagnostic conclusions in high-value cases is comprehensively extracted, which facilitates the construction and updating of knowledge graphs and the generation of new training content.

[0089] Furthermore, collected clinical cases can serve as a basis for the development and improvement of magnetocardiographs. Specifically, clinical cases can be collected through the case management module, and structured case texts can be stored in the case library. Finally, the structured cases in the case library can be used to feed back into the development of magnetocardiographs, shortening the algorithm iteration cycle.

[0090] Some optional remote training platforms further include: a user terminal;

[0091] The user terminal includes a remote training module, which is used to display links to various training contents in a classified manner. The training contents include at least one of the following: magnetocardiograph principles, magnetocardiograph operating specifications, and magnetocardiograph clinical case interpretations.

[0092] Specifically, the link may be displayed in the form of text, a picture, or a button. By clicking on the link, the user can obtain the corresponding training content from the database via the cloud server.

[0093] The training content can be in text, video, or PowerPoint format. The user terminal can be implemented as a web page, meaning that the remote training platform provided by the embodiment of the present invention is based on a browser / server (B / S) architecture. The user terminal can interact with the cloud server via the HTTP protocol. After the user clicks a link on the web page, an HTTP request is triggered. The cloud server verifies the user's permissions (such as registered user) and retrieves the corresponding training content from the database and returns it to the user terminal. The user terminal can also be implemented as an application, meaning that the remote training platform provided by the embodiment of the present invention can also be based on a client / server (C / S) architecture. If the user terminal is implemented as a web page, the training content returned by the cloud server can be encrypted, and the user terminal needs to decrypt the training content before displaying it. In addition, the remote training module of the user terminal can also provide a download button corresponding to each training content. After clicking the download button, the user can download the corresponding training content to local storage.

[0094] In some optional specific implementations, the user terminal is further configured to:

[0095] Obtain the answer data entered by the user for the test question;

[0096] Encrypting the answer data and the user's identification information to obtain encrypted data;

[0097] The encrypted data is transmitted to the cloud server.

[0098] After receiving the encrypted data (including the encrypted answer data), the cloud server decrypts it and stores it in the database.

[0099] Specifically, the test questions can be questions extracted from the test question bank, and the question types include but are not limited to multiple-choice questions, true-or-false questions, and short-answer questions. The cloud server can also include a test evaluation module, and the test questions can be specifically extracted by the test evaluation module. When extracting test questions, they can be extracted based on the user's learning progress: the test evaluation module obtains the edge weight values ​​of the user node and the knowledge point node in the knowledge graph in real time, and preferentially extracts test questions corresponding to knowledge points with edge weights less than a threshold (such as 0.4).

[0100] The encryption algorithm may be, for example, the AES-256 (Advanced Encryption Standard, 256-bit key) algorithm, which is a symmetric encryption algorithm.

[0101] In summary, the embodiments of the present invention safeguard user information security through data encryption transmission and permission management mechanisms. Furthermore, the remote training platform provided by the embodiments of the present invention not only offers various magnetocardiograph training content but also provides an instant testing function, enabling real-time assessment of the user's knowledge level. This allows for the generation of targeted, personalized training courses, improving user learning efficiency.

[0102] According to an embodiment of the present invention, an embodiment of a remote training method for a magnetocardiograph is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of executable computer instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0103] In this embodiment, a remote training method for magnetocardiographs is provided, which can be used on the above-mentioned cloud server. Figure 2 is a flow chart of a remote training method for a magnetocardiograph according to an embodiment of the present invention, such as Figure 2 As shown, the process includes the following steps:

[0104] Step S201: Acquire user learning data, knowledge points extracted from a known magnetic cardiomedical knowledge base, and knowledge points extracted from case histories to construct or update a knowledge graph. The knowledge graph includes user nodes and knowledge point nodes, with edge weights between these nodes indicating the user's mastery of the corresponding knowledge point. The user learning data includes learning behavior data and other data generated during the learning process. For detailed technical solutions and corresponding technical effects, please refer to the aforementioned remote training platform embodiment and will not be repeated here.

[0105] Step S202: Generate a personalized training course based on the edge weight value between the user node and the knowledge point node. Detailed technical solutions and corresponding technical effects can be found in the above-mentioned remote training platform embodiment, which will not be described in detail here.

[0106] See also Figure 3 , Figure 3 This is a schematic diagram of the structure of a cloud server provided by an optional embodiment of the present invention. Figure 3As shown, the cloud server includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components utilize different buses to communicate with each other and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the cloud server, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple cloud servers can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 3 A processor 10 is taken as an example.

[0107] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0108] The memory 20 stores instructions that can be executed by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0109] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created based on the use of the cloud server, etc. In addition, the memory 20 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 optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the cloud server 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.

[0110] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0111] The cloud server also includes a communication interface 30 for the cloud server to communicate with other devices or communication networks.

[0112] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0113] A portion of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0114] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A remote training platform for magnetocardiographs, characterized in that: include: Cloud servers and databases; The cloud server includes a training content management module and a test evaluation module; The database stores the training content, user learning data and test question bank of the magnetocardiograph; the user learning data includes learning behavior data and other data generated during the learning process; The training content management module is used to obtain the user learning data from the database, and obtain knowledge points extracted from the known magnetic cardiomedical knowledge base and knowledge points extracted from case records, and construct a knowledge graph, wherein the knowledge graph includes user nodes and knowledge point nodes, and the edge weight values ​​between the user nodes and the knowledge point nodes are used to indicate the user's mastery of the corresponding knowledge points; The training content management module is further configured to generate personalized training courses based on the edge weight values ​​between the user nodes and the knowledge point nodes.

2. The remote training platform according to claim 1, characterized in that: The database stores cases; The training content management module is specifically used for: Extracting structured case data from the database and generating case knowledge point feature vectors; Inputting the case knowledge point feature vector into the knowledge graph as an associated case library of the case knowledge point node; Acquiring learning data of the user when studying the case; Based on the learning data when the user studies the case, the edge weight value between the user node and the target knowledge point node is updated; wherein the target knowledge point node is the case knowledge point node corresponding to the case.

3. The remote training platform according to claim 2, characterized in that: The training content management module is specifically used for: The learning data of the user when studying the case is converted into sliding window statistical features, and then embedded into the note text using Word2Vec to generate a semantic vector; Using a pre-trained 3D ResNet-50 model, extracting deep features of the medical images in the case; extracting features of the case text from the case; and obtaining a feature vector of the case knowledge points based on the deep features and features of the case text; Encoding the semantic vector generated by the learning data into an N-dimensional hidden state, compressing the case knowledge point feature vector to N dimensions through a fully connected layer, performing attention weight calculation, and weighted summing to generate an N-dimensional cross-modal feature vector to represent the user-case association strength; Based on the cross-modal feature vector, the knowledge graph is constructed or updated.

4. The remote training platform according to any one of claims 1 to 3, characterized in that: The cloud server further includes a case management module; the case management module is used to: Obtain case text, perform semantic analysis on the case text based on a deep learning model, and extract target information, including symptoms, imaging features, and diagnostic conclusions; Inputting the structured target information into an anomaly detection model, wherein the anomaly detection model is a self-supervised deep learning model based on contrastive learning; Based on the output of the anomaly detection model, it is determined whether the case text is the target case text, and the target case text includes rare case text and misdiagnosed case text; the anomaly detection model automatically identifies the target case text by calculating the similarity between the target information of the case and the feature vector of the standard case library.

5. The remote training platform according to claim 4, characterized in that: The case management module is further used to: Obtaining a medical knowledge graph, which is a pre-constructed general knowledge system based on existing medical theories, authoritative guidelines, industry standards, and academic literature. The nodes of the medical knowledge graph are determined based on magnetocardiograph principles and disease diagnosis standards. The nodes of the medical knowledge graph include diseases, symptoms, magnetocardiograph equipment, and / or magnetocardiograph technology. The edges of the medical knowledge graph represent logical relationships between nodes. Based on the medical knowledge graph, the target information of the case text is verified for logical consistency, and the target information includes symptoms, imaging features and diagnostic conclusions.

6. The remote training platform according to claim 1, characterized in that: The cloud server also includes an expert review module, which is used to obtain expert review opinions on the case; The expert review module is embedded with an AI-assisted annotation tool to automatically generate review suggestions.

7. The remote training platform according to claim 1, characterized in that: Also includes: User terminal; The user terminal includes a remote training module, which is used to display links to various training contents in a classified manner. The training contents include at least one of the following: magnetocardiograph principles, magnetocardiograph operating specifications, and magnetocardiograph clinical case interpretations.

8. A remote training method for magnetocardiographs, characterized in that: include: Obtain user learning data, as well as knowledge points extracted from a known magnetic cardiomedical knowledge base and knowledge points extracted from case histories, to construct or update a knowledge graph; wherein the knowledge graph includes user nodes and knowledge point nodes, and the edge weight values ​​between the user nodes and the knowledge point nodes are used to indicate the user's mastery of the corresponding knowledge points; the user learning data includes learning behavior data and other data generated during the learning process; A personalized training course is generated based on the edge weight values ​​between the user node and the knowledge point node.

9. A cloud server, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the remote training method for the magnetocardiograph according to claim 8 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the remote training method for a magnetocardiograph according to claim 8.

Images