Perioperative period ultrasonic training platform for anesthesia major

Abstract

The invention discloses an anesthesia major perioperative period ultrasonic training platform, which relates to the technical field of medical education, and comprises a management module for information management and updating, a knowledge module for providing course resources and updating, an ultrasonic module for executing image acquisition and analysis, and an image storage module for classifying and storing images and videos. The scene module provides a virtual reality environment for training, the task module distributes tasks and supports a personalized learning mode, and the evaluation module optimizes course content through intelligent evaluation and adjusts a training scheme. The technical problem that theoretical teaching and practical operation are disjointed in the prior art is solved, and the technical effect of improving the training effect and the skill mastering accuracy is achieved by achieving deep fusion of theoretical teaching and practical operation.

Description

Technical Field

[0001] The present application relates to the field of medical education technology, and in particular to a perioperative ultrasound training platform for anesthesia professionals. Background Art

[0002] Perioperative ultrasound plays a vital role in anesthesia, especially during anesthesia, by monitoring the patient's physiological condition in real time to ensure the safety of anesthesia. Traditional ultrasound training methods mainly rely on theoretical teaching and limited practical opportunities, but in actual operation, there is a significant disconnect between theory and practice. Trainees often find it difficult to quickly apply the theoretical knowledge acquired in the classroom to practical operations, resulting in poor learning results. In addition, the scattered teaching resources, single teaching methods, and lack of timely and effective evaluation and feedback mechanisms in traditional training limit the learning progress and skill mastery accuracy of trainees. With the rise of emerging technologies such as artificial intelligence and virtual reality (VR), new ideas are provided to solve the above problems. However, existing technologies still fail to effectively combine theoretical teaching with practical operations, cannot meet personalized learning needs, and lack dynamic evaluation and feedback of trainees' real-time learning status. Therefore, how to achieve a deep integration of theory and practice and provide personalized learning paths through technical means has become a technical problem that needs to be solved in the current field of ultrasound training.

[0003] At the current stage, there is a technical problem in the relevant technologies that theoretical teaching is out of touch with practical operations. Summary of the invention

[0004] The present application solves the technical problem of the disconnection between theoretical teaching and practical operation in the prior art by providing a perioperative ultrasound training platform for anesthesia professionals.

[0005] This application provides an anesthesia professional perioperative ultrasound training platform, including:

[0006] Management module, used to manage, update and maintain the information within the target platform; knowledge module, used to provide learning resources, including course resource library and knowledge update unit; ultrasound module, used to perform ultrasound image acquisition and analysis, including ultrasound system and ultrasound image acquisition system; image and video storage module, used to classify and store ultrasound images and videos generated from ultrasound module and training operations; scene module, used to provide virtual or augmented reality environment for ultrasound learning and operation training; task module, used to distribute daily tasks and assessment tasks to users, including personal mode, group mode and customized mode; evaluation module, used to conduct training evaluation on users through intelligent evaluation system, and feed back the evaluation results to the course resource library of knowledge module for training adjustment.

[0007] The proposed anesthesia professional perioperative ultrasound training platform first manages and updates information through the management module, the knowledge module provides course resources and updates, the ultrasound module performs image acquisition and analysis, and the image storage module stores images and videos in categories. The scene module provides a virtual reality environment for training, the task module distributes tasks and supports personalized learning mode, and the evaluation module optimizes course content and adjusts training plans through intelligent evaluation. By realizing the deep integration of theoretical teaching and practical operation, the technical effect of improving training effect and the accuracy of skill mastery is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure are briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the platform according to the embodiment of the present application. It should be understood that the previous or following operations are not necessarily performed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0009] Figure 1 A schematic diagram of the structure of an anesthesia professional perioperative ultrasound training platform provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of the management module structure of an anesthesia professional perioperative ultrasound training platform provided in an embodiment of the present application.

[0011] Explanation of the reference numerals: management module 10 , knowledge module 20 , ultrasound module 30 , image and video storage module 40 , scene module 50 , task module 60 , evaluation module 70 . DETAILED DESCRIPTION

[0012] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0013] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0014] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, platform, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0015] The present application embodiment provides an anesthesia professional perioperative ultrasound training platform, such as Figure 1 As shown, the platform includes:

[0016] Management module 10 is used to manage, update and maintain the information in the target platform. Specifically, management module 10 is responsible for the management, update and maintenance of information in the target platform. In terms of user information management, it covers operations such as information verification storage during registration, identity authentication during login, and subsequent information management and update; it provides management personnel with login management with strict identity authentication, as well as platform maintenance and upgrade functions; in patient information management, it obtains information from legitimate data sources and encrypts and stores it, performs classification management and authority control; it also collaborates with cloud servers to store data, uses backup and other technologies to ensure security, and supports remote operation of the user end through a human-computer collaborative window, parses and forwards instructions, and ensures data security and communication stability.

[0017] In one possible implementation, Figure 2As shown, the management module 10 further includes: a user information management unit, which is used to register, log in, manage, update and maintain user information. Specifically, the user information management unit is a key link in the interaction between the platform and the user, and is fully responsible for matters related to user information. When a new user registers, it is necessary to fill in the name, contact information, user name, password and medical background and other related information on the interface provided by it. After submission, the unit will check the required items, format verification and query the uniqueness of the user name, encrypt the stored information after verification and notify the user of successful registration. When the user logs in, the unit matches the user name in the database, compares the password, and may also trigger additional verification such as verification code and biometrics. After successful login, the login time, IP address and other information are recorded. During the user's use of the platform, the unit updates the user status, learning progress and other information in real time. When the user needs to update information, such as contact information and password, the unit provides an entry and updates the database after strict verification. In addition, the unit also regularly maintains the database, cleans up invalid information, strengthens data encryption and authority management, ensures the security of user information, and lays a solid data foundation for the stable operation of the platform.

[0018] The administrator login unit is used for the platform administrator of the target platform to log in, manage, maintain and upgrade. Specifically, the administrator login unit is of great significance to the stable and efficient operation of the target platform, and is the key entry point for various operations of the platform administrator. When logging in, the administrator needs to enter a complex combination password containing uppercase and lowercase letters, numbers and special characters and a specific account on the more secure login page. It is also necessary to pass multi-factor identity authentication, such as receiving a dynamic verification code, and may even use biometric technologies such as fingerprints and facial recognition. After a successful login, the administrator can perform all-round management on the management interface, view detailed information such as user registration time and login records, handle abnormal accounts, and review the learning resources and teaching materials of the management platform. In terms of maintenance operations, the administrator uses system monitoring tools to view the server CPU usage, memory usage and other operating status in real time, handle exceptions in a timely manner, and regularly back up data to off-site storage devices. When faced with a platform upgrade, the administrator first evaluates the upgrade content, performs simulation tests in a test environment, and uses the upgrade deployment tool to deploy the upgrade to the production environment in a grayscale release manner. The administrator first opens the new features to some users, collects feedback, and promotes them across the board after confirmation. After the upgrade is complete, the administrator tests again, handles problems encountered by users when using the new features, and ensures the smooth operation of the platform from all aspects.

[0019] The patient information management unit is used to obtain and manage patient information. Specifically, the patient information management unit is crucial in medical-related platforms, and its work revolves around the acquisition and management of patient information. In the acquisition stage, first clarify the legitimate data source, such as the cooperative hospital information system, obtain the patient's informed consent before acquisition, and collect it in a standardized format and standard through a secure data interface, covering basic information, medical history, ultrasound examination and other aspects. After collection, the data is stored in a specially constructed database. Relational or non-relational databases are selected according to data characteristics. Sensitive information is encrypted using encryption algorithms such as AES, and access rights are strictly set. When managing, classify the data to facilitate retrieval and query by disease type, examination time, etc., and update information in a timely manner, and regularly clean and optimize the database. In terms of use and sharing, it is used internally for training and teaching, following the usage specifications. Under the premise of compliance, it can be desensitized and shared with other institutions for collaboration, to help medical research, and to fully ensure the efficiency and security of patient information management, while taking into account privacy protection.

[0020] Cloud servers are used to store teaching and operation data. Specifically, as the data storage core of the training platform, cloud servers have a rigorous and comprehensive solution for the storage of teaching and operation data. Teaching data comes from the materials uploaded by teachers and the operation records of students. They are encrypted and transmitted through a special interface. After arrival, the teaching data is classified according to the subject, course, difficulty and other dimensions, and the operation data is classified according to the student identity, operation time and other dimensions. In terms of storage architecture, distributed storage technologies such as Ceph are used, combined with object storage and relational databases, and storage resources are reasonably allocated between SSD and HDD according to the importance and frequency of data use. In terms of data security, SSL / TLS protocol is used for transmission encryption, and AES is used for storage to encrypt sensitive data, and strict access rights are set. The backup strategy adopts a combination of full and incremental backup, off-site storage, and the establishment of a recovery mechanism. Rapid recovery is based on backup time and data loss, providing stable and reliable data support for the training platform in all aspects.

[0021] The human-computer collaboration window is used to realize remote operation on the user side. Specifically, the human-computer collaboration window is the key to realizing remote operation on the user side, and its operation process involves multiple closely related links. When the user starts remote operation, the window uses the TCP / IP protocol to establish a communication connection with the remote server, and verifies the user identity through the account password, and grants operation permissions according to the permission level. During the operation, the window collects user operation instructions in real time, converts and parses them, and uses data compression and caching technology to optimize transmission and transmit the instructions to the server. After receiving the instructions, the server calls the corresponding service resources to perform operations, such as adjusting equipment parameters in the ultrasound training platform, and then feeds back the operation results to the window. After the window parses and processes, it is intuitively displayed to the user, such as real-time display of ultrasound images. The SSL / TLS protocol is used to encrypt data throughout the process, and firewalls and intrusion detection systems are set up to ensure security and stability. In the face of abnormalities such as network interruptions, it can prompt in time and try to automatically recover, or provide troubleshooting suggestions, helping users to achieve convenient, efficient and safe remote operation in all aspects.

[0022] Knowledge module 20 is used to provide learning resources, including a course resource library and a knowledge update unit. Specifically, knowledge module 20, as the core supply module of learning resources, is operated in collaboration with the course resource library and the knowledge update unit. When building the course resource library, we widely collect theoretical knowledge such as medical textbooks, academic papers, and operation demonstration videos recorded by expert doctors, organize them according to knowledge systems such as basic theory and advanced operation, integrate them into the library using the learning management system, give resources unique identification and metadata tags, and use redundant storage and regular backup to ensure security. Subsequent regular reviews will be conducted to update and replace outdated content and optimize presentation methods based on new medical achievements, industry changes, and user feedback. The knowledge update unit is equipped with a professional team to monitor medical academic journals, industry conferences and other channels in real time, capture new knowledge, and exclude unreliable information through screening and evaluation. Reliable content is produced in the form of briefings, lecture videos, etc., and personalized push is made according to user learning history and preferences, so as to continuously provide users with rich, accurate, and timely learning resources to meet their diverse needs in perioperative ultrasound learning for anesthesia surgery.

[0023] In a possible implementation, the knowledge module 20 further includes: a theoretical teaching unit, which is used to analyze the evaluation results fed back by the evaluation module to obtain a personalized training plan. Specifically, the theoretical teaching unit receives multi-dimensional evaluation data covering theoretical test scores, classroom performance, homework completion, etc. from the evaluation module, establishes an evaluation file according to the individual classification of students, and records the time nodes. The evaluation results are mined using knowledge graph technology, and a framework sequence of knowledge points of the evaluation results is generated based on the logic of the knowledge system. Then, the association analysis is iterated in chronological order, the knowledge mastery at different stages is compared, and the reasons for the changes are explored. Finally, according to the analysis results, theoretical courses of weak knowledge are added to target the students' knowledge shortcomings and learning characteristics, tutoring and case explanations are arranged, learning resources are recommended, learning goals and time nodes are clarified, and timely adjustments are made based on learning progress and feedback to develop a personalized training plan.

[0024] The simulation training unit is used to generate operation scenarios using ultrasound-guided puncture models and build a training operation scenario library. Specifically, the simulation training unit helps trainees improve their ultrasound-guided puncture operation capabilities through a series of rigorous processes. First, accurately select ultrasound-guided puncture models that are highly simulated, have realistic human tissue texture and acoustic characteristics from a large number of models, and reasonably layout and build them according to training needs, set different puncture paths and angles, and ensure good compatibility with ultrasound equipment. Next, based on common clinical cases and complex conditions, design a variety of operation scenarios covering different disease types, patient conditions and surgical needs, and use VR or AR technology to integrate virtual information with real models to enhance the sense of immersion in operations. Finally, digitally store the generated scenes in a special database, annotate with detailed indexes and labels, continue to pay attention to clinical progress, regularly update and improve the scenario library, and do a good job of data backup and security maintenance to create a high-quality simulation training environment for trainees.

[0025] The practical operation unit is used to use the knowledge base and artificial intelligence corpus to provide operation guidance to the user. Specifically, when the user performs ultrasound-related practical operations, the practical operation unit uses sensors and data acquisition interfaces to collect multi-dimensional data such as equipment parameter adjustments and operation actions in real time. After noise reduction and cleaning, the algorithm is used to extract key features, such as the texture and edge features of the ultrasound image, and the speed and acceleration of the operation action. Subsequently, the above features are matched and retrieved with the knowledge base stored in the form of semantic networks, ontology models, etc. and with efficient indexing to find relevant knowledge. At the same time, based on the artificial intelligence corpus trained with a large amount of text data, according to the knowledge base matching results, operation instructions are generated in natural language, and feedback is given to the user through voice or text, helping them to correct errors in a timely manner and improve their practical operation capabilities in all aspects.

[0026] In a possible implementation, the theoretical teaching unit further includes: an evaluation result sequence acquisition subunit, which is used to obtain the evaluation result sequence of the user within the preset time window. Specifically, the evaluation result sequence acquisition subunit is the cornerstone of the personalized learning analysis system, and its workflow revolves around the collection of multi-dimensional evaluation data. First, the training project or learning plan is investigated, and the length and start time of the preset time window are determined based on the information such as the learning cycle and course schedule, which can be fixed or dynamic. Then, the evaluation data source is sorted out, and the evaluation tools such as exams, homework, and classroom performance in the training platform are connected to collect data such as grades, completion status, and participation. At the same time, external data from third-party evaluation agencies or other systems are connected, such as industry certification grades, practical ability evaluation, etc. After that, for data from different sources, the interface program is used to extract them regularly or obtain them through FTP and API, and then clean them, handle format, missing values, etc., and finally integrate them according to the user ID to form a complete and accurate evaluation result sequence, which lays a solid data foundation for subsequent learning analysis and training program formulation.

[0027] The knowledge point extraction subunit is used to extract knowledge points from the evaluation result sequence using the knowledge graph to obtain the evaluation result knowledge point framework sequence. Specifically, the knowledge point extraction subunit is crucial in learning evaluation analysis. It uses the knowledge graph as the core tool to mine key knowledge points. First, a knowledge graph is constructed to widely collect domain knowledge from professional textbooks, academic papers and other materials, use ontology models and other methods to represent and model knowledge, build a knowledge logic architecture, and regularly pay attention to industry trends and update and optimize the graph. Then, the evaluation result sequence is preprocessed, noise and invalid data are cleaned, and the format is converted for subsequent analysis. After that, the evaluation results are matched based on the knowledge graph, and natural language processing and machine learning algorithms are used to find the corresponding relationship, extract and annotate knowledge points, and generate the evaluation result knowledge point framework sequence according to logical organization to reflect the user's knowledge mastery. Finally, a verification mechanism is established, and manual spot checks and cross-validation are used to ensure accuracy. The algorithm and graph are improved according to the feedback of the results, and the extraction quality is continuously improved to provide strong support for learning analysis and personalized training.

[0028] The iterative association analysis subunit is used to perform iterative association analysis on the evaluation result knowledge point framework sequence in the order from the front to the back, and obtain the iterative association evaluation result knowledge point framework sequence. Specifically, the iterative association analysis subunit plays a key role in the learning evaluation process. After receiving the evaluation result knowledge point framework sequence from the knowledge point extraction subunit, it first comprehensively combs, clarifies the timestamp, and determines the analysis dimensions, including the mastery of knowledge points, the strength of association, and the learning progress. Subsequently, a time series analysis is performed to longitudinally compare the mastery of each knowledge point at different times, draw a change curve, and predict the trend using the moving average method. At the same time, a knowledge point association analysis is carried out, an association network is constructed through co-occurrence analysis, and causal relationships are mined using Granger causality tests. As new evaluation results are generated, the analysis is dynamically updated, and the time series and association analysis results are integrated to form an iterative association evaluation result knowledge point framework sequence. Finally, the accuracy is verified by comparing with expert judgment, and the analysis method and process are optimized according to feedback from the teaching team, etc., to provide strong support for personalized and precise teaching.

[0029] The training program identification subunit is used to identify the training program for the knowledge point framework sequence of the iterative association evaluation result, and obtain the personalized training program for the user. Specifically, the training program identification subunit is the key execution link of the personalized training system. It customizes exclusive training programs for users based on the knowledge point framework sequence of the iterative association evaluation result. First, in-depth research is conducted on the training needs under different learning scenarios, and a training program template library is constructed according to categories such as basic consolidation and ability improvement, covering training objectives, course arrangements and other contents, and regularly updated according to feedback and industry changes. Next, the user's knowledge point mastery is analyzed, and their learning style and preferences are understood in combination with questionnaires and behavioral analysis. Subsequently, the template library is preliminarily matched with the program based on these characteristics, and the most suitable template is selected by comprehensively analyzing factors such as resource availability and time compatibility. Then, the course content is adjusted according to the user's knowledge shortcomings, and the progress and teaching methods are optimized in combination with their learning habits, and a personalized program containing details such as learning goals, content, and time is generated. Finally, before implementation, experts are invited to evaluate and communicate with users to verify the feasibility of the program, and the effect is continuously tracked during implementation, and dynamically optimized based on feedback to meet the diverse learning needs of users.

[0030] In a possible implementation, the training program identification subunit further includes: a knowledge point framework extraction micro-unit, which is used to extract the first evaluation result knowledge point framework and the second evaluation result knowledge point framework located at the first and second positions in the evaluation result knowledge point framework sequence. Specifically, the knowledge point framework extraction micro-unit first establishes a connection with the database or data transmission channel storing the evaluation result knowledge point framework sequence, calls the API to obtain read permissions by configuring the correct parameters, and verifies the data format, and converts it if it does not match. After preparation, the position identifiers located at the first and second positions are quickly identified based on the index structure, and corresponding methods are used for different data structures to completely extract the framework containing knowledge points, associations, attributes and other contents from the sequence, and mark them as the first and second evaluation result knowledge point frameworks respectively. After that, they are encapsulated into specific data objects or collections, metadata is added, and then passed to the next micro-unit through a preset channel, such as the knowledge point framework sequence acquisition micro-unit, to promote subsequent analysis processes.

[0031] The knowledge point framework sequence acquisition micro-unit is used to perform iterative association analysis on the first evaluation result knowledge point framework and the second evaluation result knowledge point framework to obtain the first stage iterative association evaluation result knowledge point framework sequence. Specifically, after the knowledge point framework sequence acquisition micro-unit receives the first and second evaluation result knowledge point frameworks from the knowledge point framework extraction micro-unit, it first organizes the data, constructs the knowledge points and association relationships into a knowledge graph prototype, and sets the analysis dimensions such as importance, logical association strength, and practical application relevance. Then, the first round of association analysis is carried out to sort out the logical relationship of the knowledge points, such as the causal relationship between the anatomical structure and the physiological function in medicine, and establish a preliminary association model represented by a graph model, where the nodes are knowledge points and the edge weights represent the association strength. In subsequent iterations, new data and rules such as actual cases and discipline development trends are introduced to dynamically adjust the association strength. After multiple rounds of iterations, the determined knowledge point association relationships are integrated to generate the first stage iterative association evaluation result knowledge point framework sequence, which clearly presents each knowledge point and its association after in-depth analysis, providing a basis for subsequent learning analysis and teaching decisions.

[0032] The result knowledge point framework extraction micro-unit is used to extract the third evaluation result knowledge point framework located in the third position in the evaluation result knowledge point framework sequence again. Specifically, the result knowledge point framework extraction micro-unit first establishes a connection with the data source storing the evaluation result knowledge point framework sequence. Whether it is a database, a file system or a distributed storage platform, the corresponding parameters must be configured, and then the data environment is initialized, the data integrity and consistency are checked, and the necessary dependent libraries are loaded. Then, according to the sequence indexing rules, the third position identifier is accurately located. If the sequence contains interference information, the screening conditions are set to exclude it, and only the knowledge point framework related data is retained. After determining the position, according to the data storage format, the corresponding parsing tool is used to completely extract the third evaluation result knowledge point framework, and then it is converted into a common format, and metadata such as extraction time and source identification are added for packaging. Finally, the output is passed to the next micro-unit responsible for association analysis, providing a data basis for subsequent knowledge integration analysis.

[0033] The iterative association analysis micro-unit is used to perform iterative association analysis on the knowledge point framework sequence of the first-stage iterative association evaluation result and the knowledge point framework of the third evaluation result, and obtain the knowledge point framework sequence of the second-stage iterative association evaluation result. Specifically, the iterative association analysis micro-unit receives the knowledge point framework sequence of the first-stage iterative association evaluation result from the knowledge point framework sequence acquisition micro-unit, receives the knowledge point framework sequence of the third evaluation result from the result knowledge point framework extraction micro-unit, integrates the two in a unified data structure, and then performs standardization processing, unifies the naming of knowledge points, and normalizes the association strength. The association matrix is ​​constructed with the processed data, and the association rules are mined using algorithms such as Apriori, such as discovering the association between the mastery of function knowledge and the ability to draw coordinates. In subsequent iterations, the matrix is ​​updated according to the rules, the association strength value is increased or decreased, and new data such as teaching cases and student feedback are introduced and adjusted continuously. After multiple rounds of iterations, when the matrix and rules are stable, the integration results generate the knowledge point framework sequence of the second-stage iterative association evaluation result, presenting the deep association between knowledge points, providing support for teaching resource configuration and personalized learning plans, and finally the output is passed to the subsequent micro-units.

[0034] The iterative association evaluation result knowledge point sequence acquisition micro-unit is used to perform iterative association analysis on the remaining evaluation result knowledge point frameworks in the evaluation result knowledge point framework sequence based on the second stage iterative association evaluation result knowledge point framework sequence, and obtain the iterative association evaluation result knowledge point framework sequence. Specifically, the iterative association evaluation result knowledge point sequence acquisition micro-unit first receives the second stage iterative association evaluation result knowledge point framework sequence from the iterative association analysis micro-unit, and obtains the remaining evaluation result knowledge point frameworks, organizes the data according to the tree structure, and sets analysis parameters such as calculating the association strength with cosine similarity and taking the change of association strength less than the threshold as the termination condition. Then, for each remaining knowledge point framework, association analysis is performed with the existing sequence knowledge points in turn, the logical connection is judged, and the preliminary results are recorded in the temporary association relationship table, indicating the association strength and type. Then, multiple rounds of iterations are performed, and the table is updated according to the new results and feedback, and the potential association pattern is mined and added using the frequent item set mining algorithm. After the iteration meets the termination condition, the association relationship is integrated to generate the final sequence, and its integrity and consistency are verified, and finally output to provide learning path recommendations for the intelligent teaching system, or as a data source for the knowledge graph.

[0035] In a possible implementation, the theoretical teaching unit further includes: an inner product mapping subunit, which is used to perform inner product mapping on the first evaluation result knowledge point framework and the second evaluation result knowledge point framework to obtain a first-stage framework similarity set. Specifically, the inner product mapping subunit receives the first and second evaluation result knowledge point frameworks from other related units, and the knowledge points in the frameworks have different attributes. In order to perform inner product mapping, a quantitative method such as a word vector model is used to convert the knowledge points into vectors. Next, each knowledge point vector of the first framework is paired with a corresponding or logically related vector in the second framework, and each pair of vectors is calculated according to the inner product formula. Calculate the similarity and collect the calculation results to form the first-stage framework similarity set containing the similarity values ​​of each pair of knowledge points. Finally, output the set to the subsequent unit with metadata such as knowledge point identifiers. At the same time, optimize its own processing process based on subsequent feedback, re-examine the vectorization method and pairing strategy, and improve the accuracy of similarity calculation.

[0036] The similarity set screening subunit is used to screen the first-stage framework similarity set to obtain the first-stage screening framework similarity set. Specifically, before performing the screening operation, the similarity set screening subunit will clarify the screening target according to the purpose of knowledge analysis and the application scenario, such as screening high-similarity knowledge point pairs for the intelligent learning system to assist students in learning, or screening data reflecting the knowledge system for the knowledge graph construction. Based on this, the screening criteria are determined. The common method is to set a similarity threshold, such as setting it to 0.75, and the importance weight of the knowledge point can also be considered. At the same time, the list or dictionary used to store the screening results is initialized. Then, each knowledge point pair and its similarity value in the first-stage framework similarity set are read in turn, and judged according to the screening criteria. When judging only by the threshold, the similarity value is compared with the threshold. If it is greater than or equal to, the condition is met. The results that meet the conditions are stored in the initialized storage structure. After the screening is completed, the sampling verification results are checked to ensure that there is no error in the screening. Finally, the verified first-stage screening framework similarity set is output to the next processing unit. When outputting, the data format is guaranteed to be compatible and the screening process description is attached.

[0037] The normalization processing subunit is used to normalize the similarity set of the first stage screening framework using the similarity normalization formula, and embed the processing result into the initially empty matrix to obtain the first stage iterative correlation matrix. Specifically, the normalization processing subunit receives the first stage screening framework similarity set from the similarity set screening subunit, assuming that the normalization formula is used Next, create an n×n initially empty matrix based on the number of knowledge points involved in the set, with rows and columns corresponding to knowledge points. Then, read the similarity values ​​lim(x i ,y i ), calculate the numerator according to the formula With the denominator Get the normalized value Nor[lim(x i ,y i )]. Finally, the filling position of the normalized value in the matrix is ​​determined and filled according to the corresponding relationship of the knowledge points, generating the first-stage iterative association matrix that fully displays the association strength of the knowledge points, providing key data for subsequent knowledge analysis and application.

[0038] The convolution operation subunit is used to perform convolution operation on the first-stage iterative association matrix and the second evaluation result knowledge point framework to obtain the first-stage iterative association evaluation result knowledge point framework sequence. Specifically, the convolution operation subunit obtains the first-stage iterative association matrix from the normalization processing subunit, extracts the micro-unit from the knowledge point framework to obtain the second evaluation result knowledge point framework, checks and adjusts the dimensions of the two to make them compatible, such as reshaping the knowledge point framework vector set, filling or cutting the matrix. According to the analysis objectives and data characteristics, a convolution kernel of appropriate size and initial weight is selected, such as selecting a 3×3 convolution kernel for mining local strong associations. The convolution kernel slides on the two data according to the step size, multiplies the corresponding area elements each time it slides and accumulates and sums them to obtain the convolution result value, which is filled into the intermediate result matrix. After normalization, threshold filtering and other processing of the matrix, it is converted into the required format, expanded by rows or columns to form a sequence, and output to subsequent micro-units to provide key data for knowledge integration.

[0039] In a possible implementation, the convolution operation subunit further includes: a preset similarity threshold acquisition microunit, which is used to obtain a preset similarity threshold, and use the preset similarity threshold to screen the first stage framework similarity set to obtain the first stage screening framework similarity set. Specifically, the preset similarity threshold acquisition microunit first determines the preset similarity threshold through statistical analysis based on historical data (such as analyzing the association between students' mastery of knowledge points and similarity on the online education platform) and expert experience setting (such as medical experts based on clinical diagnosis needs), stores it in the form of a database or configuration file, and then reads it in a corresponding manner. Then, the first stage framework similarity set is received from the inner product mapping subunit, and the set contains the similarity values ​​of corresponding knowledge points in the knowledge point frameworks of the two evaluation results. After that, each similarity value in the set and its corresponding knowledge point pair are traversed in turn, and the similarity value is compared with the preset threshold. If it is greater than the threshold, the knowledge point pair and the similarity value are retained and stored in a new data structure to form the first stage screening framework similarity set. Before output, the results are checked for integrity and format correctness. If they are correct, the collection is passed to the subsequent processing unit, and additional information such as the threshold and time used for filtering can also be attached.

[0040] In a possible implementation, the normalization processing subunit further includes: a similarity normalization formula microunit, which is used for the similarity normalization formula: Among them, Nor[lim(x i ,y i )] is the normalized value corresponding to the i-th first-stage screening framework similarity in the first-stage screening framework similarity set, e is the base of the natural logarithm, m is the total number of first-stage screening framework similarities in the first-stage screening framework similarity set, lim(x i ,y i) is the first stage screening framework similarity between the i-th knowledge point in the first evaluation result knowledge point framework and the i-th knowledge point at the same position in the second evaluation result knowledge point framework. Specifically, the similarity normalization formula micro-unit receives the first stage screening framework similarity set from the similarity set screening sub-unit, and specifies the formula The meaning of each parameter here is that e is the natural logarithm base, m is the total number of similarities in the set, Nor[lim(x i ,y i )] is a normalized value. Next, read each similarity value lim(x i ,y i ), calculate the numerator first Then perform the same exponential operation on all similarity values ​​and add them up to the denominator, and divide the numerator by the denominator to calculate the normalized value. Finally, organize all normalized values ​​into a new set and output them to subsequent micro-units to provide data for building the association matrix. It can also be used in the knowledge graph to determine the edge weights and the intelligent recommendation system to make accurate recommendations.

[0041] In a possible implementation, the practical operation unit further includes: a real-time operation information acquisition subunit, which is used to extract data from the user's real-time operation and obtain real-time operation information. Specifically, the real-time operation information acquisition subunit first analyzes the user operation scenario, such as the scenario of APP remote control and direct operation of the control panel in the smart home appliance control system, and defines the data collection boundary accordingly. Then, hardware sensors (such as smart watch acceleration sensors, smart car steering wheel angle sensors) and software interfaces (mobile applications use operating system APIs, web applications use JavaScript event monitoring) are selected for data collection. When the user operates, real-time monitoring is performed and data is recorded in a format containing timestamps, operation types, and operation objects, such as e-commerce APPs recording relevant information about users clicking on the purchase button. After collection, the erroneous and duplicate data are cleaned up, converted into a unified format (such as JSON), and then sent to the feature extraction subunit through network transmission protocols such as HTTP and MQTT.

[0042] The feature extraction subunit is used to extract features from the real-time operation information and obtain real-time operation features. Specifically, the feature extraction subunit receives real-time operation information mainly in JSON format and containing various operation details from the real-time operation information acquisition subunit, first cleans and removes noise and erroneous data, and normalizes data of different dimensions and value ranges. Then, according to the operation type and business needs, a suitable method is selected. The TF-IDF algorithm can be used for text operations, and the SIFT algorithm can be selected for image operations. The intelligent customer service system can also be combined with a word vector model. Then, features are extracted according to the selected method, such as performing time series analysis on the music playback application operation, and then combining and screening features to remove redundancy and retain features that have a great effect on business goals. Finally, the features are organized into a feature matrix and output to the real-time operation feature matching subunit. The data is guaranteed to be complete and accurate during output, and a description of the feature extraction process is attached.

[0043] The real-time operation feature matching subunit is used to match the real-time operation feature in the knowledge base to obtain a matching knowledge set. Specifically, when the real-time operation feature matching subunit is working, it first organizes the knowledge base, classifies it by subject field and knowledge type, and constructs an inverted index with keywords, subject terms, etc. as index items to improve search efficiency. Then, the real-time operation feature presented in the form of feature vectors, etc. is received from the feature extraction subunit, and it is converted and normalized to make it consistent with the knowledge feature representation of the knowledge base. Then, a matching algorithm is selected based on the features and the characteristics of the knowledge base, such as the cosine similarity algorithm for text and the Euclidean distance algorithm for numerical types, and the operation features are matched and calculated one by one with the knowledge base knowledge items. Finally, a similarity threshold is set, and knowledge items greater than the threshold are screened out, and they are sorted from high to low according to the similarity value. Combined with the update time and frequency of use optimization, a complete and accurate matching knowledge set with meta information is output to the real-time voice prompt information acquisition subunit.

[0044] The real-time voice prompt information acquisition subunit is used to use the artificial intelligence corpus to perform voice prompt matching on the matching knowledge set, obtain real-time voice prompt information, and provide operation guidance to the user based on the real-time voice prompt information. Specifically, the real-time voice prompt information acquisition subunit first receives the matching knowledge set from the real-time operation feature matching subunit, and prepares the classified, labeled and preprocessed artificial intelligence corpus. Then, the knowledge items are semantically analyzed, key information is extracted, template matching is performed in the corpus, and the best matching template is selected after calculating the similarity. The template is then converted into text and optimized, and a voice signal is generated using speech synthesis technology. Finally, voice prompts are played to the user through speakers, headphones and other devices, and the volume is ensured to be moderate and the sound quality is clear during playback. User operation feedback is also continuously monitored. If the user operation is completed, subsequent guidance is provided. If the operation is incorrect, the prompt is repeated or adjusted. At the same time, the prompt content and method are optimized according to user feedback to improve the operation guidance effect and user experience.

[0045] The ultrasound module 30 is used to perform ultrasound image acquisition and analysis, including an ultrasound system and an ultrasound image acquisition system. Specifically, when the ultrasound module 30 performs ultrasound image acquisition and analysis, the preparation stage first checks and debugs the host and probe of the ultrasound system to ensure that the power supply is stable, the probe is firmly connected and not damaged, and the gain, TGC and other parameters of the ultrasound image acquisition system are debugged and the storage device is checked; the patient's basic information and medical history are collected, and the patient is guided to prepare the examination site and adjust the appropriate body position. In the acquisition stage, after applying the coupling agent to the examination site, the probe is placed, and multi-layer ultrasound images are obtained using different scanning methods, which are frozen in time and stored in DICOM format. If multi-modality is supported, elastic imaging, contrast ultrasound and other acquisitions can also be performed. In the analysis stage, the image is first denoised and enhanced for preprocessing, and then the doctor manually observes and identifies the characteristics of the lesion, and CAD technology can also be used to automatically extract classification features for auxiliary diagnosis. In the subsequent processing stage, an examination report containing patient information, image description, and diagnostic opinions is generated according to the analysis results, and the image and report are stored in the PACS system and shared on demand to facilitate remote diagnosis.

[0046] The image and video storage module 40 is used to classify and store the ultrasound images and videos generated from the ultrasound module and training operations. Specifically, before storing ultrasound images and videos, the image and video storage module 40 first prepares hardware resources, selects storage devices such as hard disk arrays, NAS or SAN according to needs, and builds a high-speed and stable storage network; then configures the software system, installs storage management software, and formulates rules for classification by source (ultrasound module acquisition, training operation generation) and data type (ultrasound image, ultrasound video). When storing, first receive and verify the integrity of the data, then classify and annotate according to the rules, add patient information and other annotation content, and then store it in the corresponding device and create an index. After storage, regularly back up data to a tape library or cloud storage to prevent loss, clean up expired useless data, and optimize storage devices; use encryption technology and strict access control permissions to ensure data security and privacy.

[0047] The scene module 50 is used to provide a virtual or augmented reality environment for ultrasound learning and operation training. Specifically, when the scene module 50 provides a virtual or augmented reality environment for ultrasound learning and operation training, it first conducts demand research, communicates with ultrasound teaching institutions and medical professionals, clarifies the pain points and needs of current learning and training, and then evaluates and selects appropriate virtual or augmented reality technologies, such as Unity_3D, Unreal_Engine development engines, and Microsoft_HoloLens, Magic_Leap device development toolkits. Then, content construction is carried out, high-precision and diversified 3D models of human organs are constructed based on medical imaging data, algorithms are developed to simulate ultrasound imaging and superimpose them on the models, and various training scenes such as simulated hospital examination rooms are created. Then, the functions are realized, natural interaction methods such as handles and gesture recognition are designed, real-time operation feedback and diagnostic suggestions are provided, and multi-person collaborative learning is supported. Finally, target users such as medical students and interns are invited to try it out, feedback is collected and continuously optimized to improve the training effect.

[0048] Task module 60 is used to distribute daily tasks and assessment tasks to users, including personal mode, group mode and customized mode. Specifically, before distributing tasks, task module 60 first builds a task library that includes ultrasound knowledge and operation tasks, and is labeled by difficulty and field classification, and collects information such as user identity and skill level through questionnaires, tests, etc. In personal mode, tasks are matched according to user information, such as assigning basic theory multiple-choice questions and operation video learning tasks to novice medical students, tracking the completion progress and quality in real time, and adjusting the difficulty and type of tasks according to the results. In group mode, reasonable grouping is carried out according to factors such as skill level and learning goals to ensure that members' skills complement each other, and collaborative tasks such as complex case diagnosis and analysis reports are assigned to the group. The group performance is evaluated and feedback is given from multiple aspects such as discussion activity and communication efficiency. In customized mode, in-depth communication is first conducted with users to understand specific needs, such as training for specific disease diagnosis or equipment operation, and exclusive tasks are customized accordingly. Regular return visits are made to collect feedback and optimize to meet user expectations.

[0049] The evaluation module 70 is used to conduct training evaluation on users through an intelligent evaluation system, and to feed back the evaluation results to the course resource library of the knowledge module for training adjustment. Specifically, when the evaluation module 70 performs training evaluation, it first plans to collect the task completion data of users in the task module, such as the accuracy of answering questions, the standardization of operations, and the training behavior data of the scenario module, such as the accuracy of probe operation, etc., and determines the collection method. At the same time, it builds an intelligent evaluation system that integrates algorithm models such as decision trees and neural networks to complete the preliminary training debugging. During the training, the system collects user behavior data in real time, comprehensively analyzes the user's theoretical knowledge and operational skills performance, and locates advantages and disadvantages. At the end of the evaluation, a report containing a comprehensive score, results of each dimension, and comparative analysis is generated, and the results are fed back to the course resource library of the knowledge module. The course resource library updates the course content accordingly, invites experts to review to ensure accuracy and effectiveness, and adjusts the course structure sequence to plan personalized learning paths for different users, such as recommending special training content for users who are weak in abdominal ultrasound examinations.

[0050] In a possible implementation, the evaluation module 70 further includes: a voice module for language recognition and interaction, including platform background music, operation music and voice prompts. Specifically, when the voice module is started, it automatically detects and adapts audio devices such as microphones and speakers, loads the voice recognition engine and interaction framework, selects a language model according to the scene, and designs interaction rules. During operation, the microphone collects voice signals in real time according to parameters such as the appropriate sampling rate, and inputs them into the voice recognition engine after noise reduction and filtering preprocessing, converts them into text, and then analyzes the semantics through natural language processing technology to understand the user's intention. Subsequently, a response strategy is formulated according to the interaction rules, instructions are executed, and results are fed back. At the same time, the module will play background music and operation music according to the scene and user settings, and convert the system replies into voice prompts using speech synthesis technology, which are played to the user through the speaker after parameter setting.

[0051] The embodiment of the present application adopts information management and updating through the management module, the knowledge module provides course resources and updates, the ultrasound module performs image acquisition and analysis, and the image storage module classifies and stores images and videos. The scene module provides a virtual reality environment for training, the task module distributes tasks and supports personalized learning mode, and the evaluation module optimizes course content and adjusts the training plan through intelligent evaluation, achieving the technical effect of improving the training effect and the accuracy of skill mastery by realizing the deep integration of theoretical teaching and practical operation.

[0052] Although the present application makes various references to certain modules in the platform according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be implemented; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0053] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. An ultrasound training platform for perioperative anesthesia surgery, characterized in that: The platform includes: Management module, used to manage, update and maintain the information in the target platform; The knowledge module is used to provide learning resources, including the course resource library and knowledge update unit; An ultrasound module, used for performing ultrasound image acquisition and analysis, including an ultrasound system and an ultrasound image acquisition system; An image and video storage module, used for classifying and storing ultrasound images and videos generated from the ultrasound module and training operations; A scenario module for providing a virtual or augmented reality environment for ultrasound learning and operation training; Task module, used to distribute daily tasks and assessment tasks to users, including individual mode, group mode and customized mode; The evaluation module is used to conduct training evaluation on users through an intelligent evaluation system and feed the evaluation results back to the course resource library of the knowledge module for training adjustments.

2. The perioperative ultrasound training platform for anesthesia surgery according to claim 1, characterized in that: The management module comprises: User information management unit, used for user information registration, login, management, update and maintenance; An administrator login unit, used for the platform administrator of the target platform to log in, manage, maintain and upgrade; Patient data management unit, used to obtain and manage patient information; Cloud servers, used to store teaching and operation data; The human-computer collaboration window is used to realize remote operation on the user side.

3. The perioperative ultrasound training platform for anesthesia surgery according to claim 1, characterized in that: The course resource library includes: Theoretical teaching unit, used to analyze the evaluation results fed back by the evaluation module to obtain a personalized training plan; A simulation training unit is used to generate operation scenarios using an ultrasound-guided puncture model and build a training operation scenario library; The practical operation unit is used to provide operational guidance to users using the knowledge base and artificial intelligence corpus.

4. The perioperative ultrasound training platform for anesthesia surgery as claimed in claim 3, characterized in that: The theoretical teaching units include: An evaluation result sequence acquisition subunit is used to acquire a user's evaluation result sequence within a preset time window; A knowledge point extraction subunit is used to extract knowledge points from the evaluation result sequence using the knowledge graph to obtain a knowledge point framework sequence of the evaluation result; An iterative association analysis subunit, used to perform iterative association analysis on the evaluation result knowledge point framework sequence in a time-ordered order to obtain an iterative association evaluation result knowledge point framework sequence; The training program identification subunit is used to identify the training program for the knowledge point framework sequence of the iterative association evaluation result to obtain the user's personalized training program.

5. The perioperative ultrasound training platform for anesthesia surgery as claimed in claim 4, characterized in that: The theoretical teaching unit also includes: A knowledge point framework extraction micro-unit, used to extract the first evaluation result knowledge point framework and the second evaluation result knowledge point framework located at the first and second positions in the evaluation result knowledge point framework sequence; A knowledge point framework sequence acquisition micro-unit is used to perform iterative association analysis on the first evaluation result knowledge point framework and the second evaluation result knowledge point framework to obtain a first-stage iterative association evaluation result knowledge point framework sequence; A result knowledge point framework extraction micro-unit is used to extract again a third evaluation result knowledge point framework located at the third position in the evaluation result knowledge point framework sequence; Iterative association analysis micro-unit, used to perform iterative association analysis on the first-stage iterative association evaluation result knowledge point framework sequence and the third evaluation result knowledge point framework sequence to obtain the second-stage iterative association evaluation result knowledge point framework sequence; The iterative association evaluation result knowledge point sequence acquisition micro-unit is used to perform iterative association analysis on the remaining evaluation result knowledge point frames in the evaluation result knowledge point frame sequence based on the second-stage iterative association evaluation result knowledge point frame sequence to obtain the iterative association evaluation result knowledge point frame sequence.

6. The perioperative ultrasound training platform for anesthesia surgery as claimed in claim 5, characterized in that: The theoretical teaching unit also includes: An inner product mapping subunit, used to perform inner product mapping on the first evaluation result knowledge point framework and the second evaluation result knowledge point framework to obtain a first stage framework similarity set; A similarity set screening subunit is used to screen the first-stage framework similarity set to obtain a first-stage screened framework similarity set; A normalization processing subunit, used to normalize the similarity set of the first stage screening framework using a similarity normalization formula, and embed the processing result into an initially empty matrix to obtain a first stage iterative correlation matrix; The convolution operation subunit is used to perform a convolution operation on the first stage iterative association matrix and the second evaluation result knowledge point framework to obtain the first stage iterative association evaluation result knowledge point framework sequence.

7. The perioperative ultrasound training platform for anesthesia surgery according to claim 6, characterized in that: The theoretical teaching unit also includes: The preset similarity threshold acquisition micro-unit is used to obtain the preset similarity threshold, and use the preset similarity threshold to screen the first-stage framework similarity set to obtain the first-stage screened framework similarity set.

8. The perioperative ultrasound training platform for anesthesia surgery as claimed in claim 7, characterized in that: The similarity normalization formula is: Among them, Nor[lim(x i ,y i )] is the normalized value corresponding to the i-th first-stage screening framework similarity in the first-stage screening framework similarity set, e is the base of the natural logarithm, m is the total number of first-stage screening framework similarities in the first-stage screening framework similarity set, lim(x i ,y i ) is the first-stage screening framework similarity between the ith knowledge point in the first evaluation result knowledge point framework and the ith knowledge point at the same position in the second evaluation result knowledge point framework.

9. The perioperative ultrasound training platform for anesthesia surgery as claimed in claim 3, characterized in that: The practical operation unit includes: A real-time operation information acquisition subunit is used to extract data of the user's real-time operation and obtain real-time operation information; A feature extraction subunit, used to extract features from the real-time operation information to obtain real-time operation features; A real-time operation feature matching subunit, used to match the real-time operation feature in the knowledge base to obtain a matching knowledge set; The real-time voice prompt information acquisition subunit is used to use the artificial intelligence corpus to perform voice prompt matching on the matching knowledge set, obtain real-time voice prompt information, and provide operation guidance to the user based on the real-time voice prompt information.

10. The perioperative ultrasound training platform for anesthesia surgery according to claim 1, characterized in that: Also includes: The voice module is used for language recognition and interaction, including platform background music, operation music and voice prompts.

Images