AI-based Traditional Chinese Medicine (TCM) standardized training data management and analysis platform

Abstract

This invention discloses an AI-based data management and analysis platform for TCM standardized residency training, belonging to the field of TCM standardized residency training data management technology. It addresses the problem of ineffective processing and analysis of acquired TCM standardized residency training data. This invention utilizes an AI comparison algorithm to perform curve data overlap comparison. This method can intuitively demonstrate the similarities and differences in key features between the standard residency training model data and the residency training model data. Machine learning algorithms are used to model the overlap comparison results, obtaining AI comparison model data for the residency training model data. This transforms the comparison results into a quantifiable model, facilitating objective evaluation of the residency training effect. Different quality ranges correspond to different learning plans, enabling targeted guidance based on the trainees' actual situations. Different learning paths are recommended based on the trainees' attribute tags, further meeting their personalized needs.

Description

Technical Field

[0001] This invention relates to the field of TCM standardized training data management technology, specifically a TCM standardized training data management and analysis platform based on AI artificial intelligence. Background Technology

[0002] Traditional Chinese Medicine (TCM) standardized training data management refers to the process of collecting, organizing, storing, analyzing, and maintaining various data generated during the standardized training (residence) of TCM practitioners.

[0003] Chinese patent CN111160857A discloses a hospital standardized residency training clinical teaching management system based on an internet platform. This system primarily uses an internet platform to pre-enter information on trainees and instructors for standardized residency training and generates standardized training plans for trainees. Information is collected, interacted with, and pushed through intelligent interactive terminals. Instructors and trainees only need to teach and learn according to the generated standardized training plans. Simultaneously, the system can retrieve assessment and attendance information from the previous standardized training session for the next time point. The entire standardized training plan, content, and training status are stored in the system for easy retrieval by the institution. While this invention achieves standardized and systematic hospital standardized residency training clinical teaching management through an internet-based cloud platform and intelligent interactive terminals, the following problems still exist in practical operation:

[0004] 1. The learning plan was not tailored to the students' actual situation, resulting in students being unable to intuitively understand their learning progress.

[0005] 2. The acquired standardized training data was not subjected to more accurate AI comparison, which resulted in the inability to manage trainees more conveniently.

[0006] 3. The model was not built using different methods based on the attributes of the standardized training data, resulting in poor performance of the constructed model.

[0007] 4. The data obtained from standardized residency training was not further classified and processed, which led to a decrease in data security. Summary of the Invention

[0008] The purpose of this invention is to provide an AI-based data management and analysis platform for standardized training in Traditional Chinese Medicine (TCM). This platform utilizes AI comparison algorithms to perform curve data overlap comparisons. This method can intuitively demonstrate the similarities and differences in key features between standard training model data and standardized training model data. Machine learning algorithms are used to model the overlap comparison results, resulting in AI-comparative modeling data for the standardized training model data. This transforms the comparison results into quantifiable models, facilitating objective evaluation of the training effectiveness. Different quality ranges correspond to different learning programs, enabling targeted guidance based on the trainees' actual situations. Recommending different learning paths based on trainees' attribute tags further meets their personalized needs and addresses problems in existing technologies.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A data management and analysis platform for standardized training in Traditional Chinese Medicine based on AI includes:

[0011] The standardized residency training data acquisition unit is used for:

[0012] The standardized training data is retrieved from the database, and the retrieved standardized training data is classified according to the attributes of the data. The standardized training data of each attribute is preprocessed in turn.

[0013] The standardized residency training data model building unit is used for:

[0014] The preprocessed standardized training data is used to build a model according to the data format, and the standardized training data after model building is labeled as standardized training model data;

[0015] Model building AI analysis unit, used for:

[0016] The data from the standardized residency training model and the data from the standardized residency training model were compared and analyzed using AI. The analysis results were then statistically classified, and the statistically classified analysis results were used as the target analysis data.

[0017] The data analysis and evaluation unit is used for:

[0018] The target analysis data is evaluated for quality according to the evaluation rules, which are retrieved from the database. Based on the quality evaluation results, a personalized learning plan is developed and displayed visually on the terminal.

[0019] Preferably, the standardized residency training data acquisition unit includes:

[0020] The attribute classification module is used for:

[0021] The standardized residency training data in the database includes basic trainee data, training data, assessment data, and health and safety data.

[0022] The basic data of trainees includes the name, gender, student ID, and major of TCM students; the training data includes learning progress records, including completed courses and practical tasks; the assessment data includes exam scores, including theoretical exams and practical operation assessments; the health and safety data includes trainees' health records during the training period and training-related safety incident records, including records of laboratory accidents and drug use errors.

[0023] The trainee basic data, training data, assessment data, and health and safety data are classified according to their data types. After classification, the trainee basic data, training data, assessment data, and health and safety data are obtained independently in the standardized training data.

[0024] Preferably, the standardized residency training data acquisition unit further includes:

[0025] The data preprocessing module is used for:

[0026] The trainees' basic data, training data, assessment data, and health and safety data were preprocessed separately.

[0027] Data preprocessing includes data cleaning, data transformation, and dataset partitioning;

[0028] Data cleaning involves processing trainees' basic data, training data, assessment data, and health and safety data for missing, outlier, and duplicate values.

[0029] After data cleaning, data transformation is performed, which includes feature scaling, feature encoding, feature selection, and dimensionality reduction.

[0030] After the data conversion is completed, the dataset is divided into training set, validation set and test set, which are the basic data of trainees, training data, assessment data and health and safety data after data conversion.

[0031] After the dataset is divided, preprocessed student basic data, training data, assessment data, and health and safety data are obtained.

[0032] Preferably, the standardized residency training data model construction unit is further used for:

[0033] First, construct a table format for the trainees' basic data, training data, assessment data, and health and safety data;

[0034] The basic data of trainees was constructed into a table and then anonymized; the training data was presented in a time series format, which included training content, training time, training location, training method, trainer information, and trainee participation. The training content consisted of text data, image data, and video data.

[0035] Models are built using the student basic data, training data, assessment data, and health and safety data that have been constructed in tabular format.

[0036] Specifically, the basic student data was modeled using a clustering model; the training data was modeled using an analytic hierarchy process (AHP) model; the assessment data was modeled using a regression model; and the health and safety data was modeled using a time series model.

[0037] Once the trainee basic data, training data, assessment data, and health and safety data model are constructed, they will be uniformly labeled as standardized training model data.

[0038] Preferably, the model construction AI analysis unit includes:

[0039] The AI ​​comparative analysis module is used for:

[0040] Retrieve the standardized training model data from the database;

[0041] Standardize the format of standardized residency training model data and residency training model data;

[0042] Next, the key feature data of the standardized training model data and the standardized training model data are extracted. The key feature data are the indicator data of each attribute data in the standardized training model data and the standardized training model data.

[0043] The indicator data of the standardized residency training model data and the residency training model data were respectively converted into curve data;

[0044] After the curve data is converted, an AI comparison algorithm is used to compare the curve data overlap.

[0045] The overlapping comparison results are then modeled using machine learning algorithms, resulting in AI comparison modeling data for the standardized training model data.

[0046] Preferably, the model building AI analysis unit further includes:

[0047] The AI ​​statistical classification module is used for:

[0048] The AI ​​comparative modeling data was retrieved, and the attributes of the AI ​​comparative modeling data were confirmed.

[0049] Among them, the attributes of AI comparative modeling data are whether AI comparative modeling data belongs to basic student data, training data, assessment data, or health and safety data;

[0050] After the AI ​​comparative modeling data attributes are confirmed, attribute label fusion is performed.

[0051] After attribute tag fusion, target analysis data is obtained from the AI ​​comparative modeling data. The target analysis data consists of data from the non-overlapping areas in the AI ​​comparative modeling data.

[0052] Preferably, the data analysis and evaluation unit includes:

[0053] The standardized residency training data evaluation module is used for:

[0054] Import the target analysis data into the evaluation rules for quality assessment;

[0055] First, the target analysis data is converted into AI values, which are then converted into area data of non-overlapping regions in the target analysis data.

[0056] The quality range of the target analysis data is determined by the position of the area data values ​​in the evaluation rules.

[0057] The quality range includes excellent, good, pass, and fail.

[0058] The learning plan development module is used for:

[0059] The learning plan for students is determined based on the quality range of the target analysis data;

[0060] Different quality ranges are designed for different learning schemes, and different learning paths are recommended based on different attribute tags;

[0061] Finally, the learning plan and learning path will be transmitted to the terminal and displayed visually.

[0062] Students can view the learning plan and learning path after securely logging in on the terminal.

[0063] Preferably, the AI ​​statistical classification module performs attribute tag fusion after the AI ​​comparative modeling data attributes are confirmed, including:

[0064] Set up a sliding window, read the label information included in the AI ​​comparative modeling data based on the sliding window, and calculate the confidence level of the read label information:

[0065] ;

[0066] in, The confidence level of the read tag information; The average reading rate of the tag information included in the AI ​​comparative modeling data; The radius of the reading range; The rate at which AI-contrastive modeling data passes through the sliding window; The time interval for reading the label information included in the AI ​​comparison modeling data based on the sliding window;

[0067] The confidence level is compared with a preset confidence threshold, and labels with a confidence level greater than the preset confidence threshold are selected as valid labels;

[0068] Extract the feature vector of each valid label and perform matching; package the two matching valid labels as a set of fusion nodes;

[0069] The fusion nodes are normalized, and a fusion diagram is generated based on the normalized fusion nodes.

[0070] Based on the fusion diagram, determine the fusion priority information of the fusion nodes;

[0071] The attribute tags of the fusion nodes are fused according to the fusion priority information. During the fusion process, the fusion rate of each fusion node is monitored and a fusion progress observation chart is generated. The fusion progress observation chart is used to determine whether the attribute tag fusion is completed.

[0072] Preferably, it also includes: an assessment module, used to assess the learning status of trainees over t learning cycles based on standardized training data, obtain assessment values, and provide corresponding learning suggestions to trainees based on the assessment values;

[0073] The learning status of trainees over t learning cycles is evaluated based on the training data, resulting in evaluation values, including:

[0074] Determine the embedded function of the learning state evaluation model:

[0075] ;

[0076] in, This is the evaluation value of the learning status of the sample learners in the i-th learning cycle; The total number of cycles for calculating the evaluation values ​​of the learning state for the selected sample learners used to train the learning state evaluation model. The connection weights from the input layer to the output layer in the learning state evaluation model; The fitting coefficients of the learning state assessment model; For the sample students in the first The evaluation value of the learning status for each learning cycle; To be based on the total number of cycles Determined cluster centers; The maximum value among the selected cluster centers; The sensitivity threshold for evaluating the learning state model;

[0077] Input the training data of trainees over m learning cycles into the learning status assessment model, and output the assessment value of the user's learning status in each learning cycle to obtain the current learning cycle. And the previous m period ;

[0078] Calculate the evaluation value of the learner's learning status over t learning cycles. :

[0079] ;

[0080] in, For correction factor, ; For the students in the The evaluation value of the learning status for each learning cycle; For the students in the The evaluation value of the learning status for each learning cycle; For the students in the The evaluation value of the learning status for each learning cycle; For the students in the The evaluation value of the learning status for each learning cycle.

[0081] Preferably, the standardized residency training data acquisition unit retrieves the standardized residency training data from the database, including:

[0082] Receive retrieval requests;

[0083] Based on the retrieval requirements, locate the storage node for the required standardized training data in the database, and retrieve the standardized training data from the database according to the storage node.

[0084] It also includes a retrieval and adjustment module, used for:

[0085] When locating the storage node for the required training data in the database based on the retrieval requirements, the calculation of the first occurrence during storage node location is performed. The probability of a second positioning error; ;

[0086] in, The first occurrence during storage node location The probability of a second positioning error; It is a composite symbol; The number of positioning steps during the positioning process; This refers to the classification error rate when distinguishing storage node categories during storage node location;

[0087] According to the occurrence of the first event during storage node location The probability of a second positioning error is used to calculate the positioning error of the storage node.

[0088] ;

[0089] in, This refers to the positioning error of the storage node; Total number of storage node categories; In order to occur the first Storage path difference when a second location error occurs;

[0090] The positioning error of the storage node is compared with the preset positioning error. If the positioning error is determined to be greater than the preset positioning error, a prompt message is generated to adjust the index information of the storage node in the database.

[0091] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0092] 1. The AI-based TCM standardized training data management and analysis platform provided by this invention achieves data standardization and normalization by organizing different types of data into tabular format. By selecting appropriate models based on data type and characteristics, it can better capture the inherent laws and relationships of the data. Through model construction on different types of data, it can achieve in-depth mining and analysis of standardized training data.

[0093] 2. The AI-based TCM standardized training data management and analysis platform provided by this invention uses AI comparison algorithms to compare overlapping curve data. This method can intuitively show the similarity and differences in key features between the standardized training model data and the standardized training model data. The overlapping comparison results are modeled using machine learning algorithms to obtain AI comparison modeling data of the standardized training model data. The comparison results can be transformed into a quantifiable model, which is convenient for objective evaluation and optimization of the standardized training effect.

[0094] 3. The AI-based TCM residency training data management and analysis platform provided by this invention offers different learning plans corresponding to different quality ranges, enabling targeted guidance based on the actual situation of trainees. Furthermore, it recommends different learning paths based on trainees' attribute tags, further meeting their personalized needs and improving learning outcomes. Attached Figure Description

[0095] Figure 1 This is a schematic diagram of the TCM residency training data management and analysis module of the present invention;

[0096] Figure 2 This is a schematic diagram of the TCM standardized training data management and analysis method of the present invention. Detailed Implementation

[0097] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0098] To address the issue of reduced data security caused by the lack of further data classification and processing after acquisition of residency training data in existing technologies, please refer to [link / reference needed]. Figure 1 and Figure 2 This embodiment provides the following technical solution:

[0099] A data management and analysis platform for standardized training in Traditional Chinese Medicine based on AI includes:

[0100] The standardized residency training data acquisition unit is used for:

[0101] The standardized training data is retrieved from the database, and the retrieved standardized training data is classified according to the attributes of the data. The standardized training data of each attribute is preprocessed in turn.

[0102] The standardized residency training data model building unit is used for:

[0103] The preprocessed standardized training data is used to build a model according to the data format, and the standardized training data after model building is labeled as standardized training model data;

[0104] Model building AI analysis unit, used for:

[0105] The data from the standardized residency training model and the data from the standardized residency training model were compared and analyzed using AI. The analysis results were then statistically classified, and the statistically classified analysis results were used as the target analysis data.

[0106] The data analysis and evaluation unit is used for:

[0107] The target analysis data is evaluated for quality according to the evaluation rules, which are retrieved from the database. Based on the quality evaluation results, a personalized learning plan is developed and displayed visually on the terminal.

[0108] Specifically, the standardized training data acquisition unit can improve the prediction accuracy and stability of the model; feature selection and dimensionality reduction can reduce the number of input features and lower the model's complexity; the standardized training data model building unit can build models on different data, enabling in-depth mining and analysis of the standardized training data; the model building AI analysis unit can dynamically adjust model parameters based on real-time data, improving the model's adaptability and accuracy; and the data analysis and evaluation unit can provide in-depth analysis and quality assessment, allowing teachers and administrators to gain a more comprehensive understanding of students' learning status and identify problems and shortcomings in teaching.

[0109] The standardized residency training data acquisition unit includes:

[0110] The attribute classification module is used for:

[0111] The standardized residency training data in the database includes basic trainee data, training data, assessment data, and health and safety data.

[0112] The basic data of trainees includes the name, gender, student ID, and major of TCM students; the training data includes learning progress records, including completed courses and practical tasks; the assessment data includes exam scores, including theoretical exams and practical operation assessments; the health and safety data includes trainees' health records during the training period and training-related safety incident records, including records of laboratory accidents and drug use errors.

[0113] The trainee basic data, training data, assessment data, and health and safety data are classified according to their data types. After classification, the trainee basic data, training data, assessment data, and health and safety data are obtained independently in the standardized training data.

[0114] The data preprocessing module is used for:

[0115] The trainees' basic data, training data, assessment data, and health and safety data were preprocessed separately.

[0116] Data preprocessing includes data cleaning, data transformation, and dataset partitioning;

[0117] Data cleaning involves processing trainees' basic data, training data, assessment data, and health and safety data for missing, outlier, and duplicate values.

[0118] After data cleaning, data transformation is performed, which includes feature scaling, feature encoding, feature selection, and dimensionality reduction.

[0119] After the data conversion is completed, the dataset is divided into training set, validation set and test set, which are the basic data of trainees, training data, assessment data and health and safety data after data conversion.

[0120] After the dataset is divided, preprocessed student basic data, training data, assessment data, and health and safety data are obtained.

[0121] Specifically, the residency training data is first categorized using the attribute classification module. This categorization clearly organizes different types of data, making the data structure more orderly and easier to understand. This helps users or system administrators quickly locate the information they need, improving data processing efficiency. Categorization allows for the implementation of different security strategies for different data types. Basic student data may require encryption or access control to ensure data security and privacy. Training and assessment data may require sharing and viewing based on different permissions. The data preprocessing module preprocesses the residency training data. For missing data, methods such as deletion, imputation, or replacement can be selected based on the importance and degree of missing data. Outliers need to be identified first, and then a decision on whether to retain, correct, or delete them is made based on traditional Chinese medicine knowledge. Data cleaning and transformation eliminate noise and inconsistencies, improving data accuracy and reliability. Preprocessed data is more suitable for machine learning algorithms, improving model prediction accuracy and stability. Feature selection and dimensionality reduction reduce the number of input features, lowering model complexity and improving training and inference efficiency. Dataset partitioning allows for objective evaluation of model performance and comparison of different models.

[0122] To address the issue in existing technologies where different model building methods are not applied based on the attributes of residency training data, resulting in poor model presentation, please refer to [link to relevant documentation]. Figure 1 and Figure 2 This embodiment provides the following technical solution:

[0123] The standardized residency training data model building unit is also used for:

[0124] First, construct a table format for the trainees' basic data, training data, assessment data, and health and safety data;

[0125] The basic data of trainees was constructed into a table and then anonymized; the training data was presented in a time series format, which included training content, training time, training location, training method, trainer information, and trainee participation. The training content consisted of text data, image data, and video data.

[0126] Models are built using the student basic data, training data, assessment data, and health and safety data that have been constructed in tabular format.

[0127] Specifically, the basic student data was modeled using a clustering model; the training data was modeled using an analytic hierarchy process (AHP) model; the assessment data was modeled using a regression model; and the health and safety data was modeled using a time series model.

[0128] Once the trainee basic data, training data, assessment data, and health and safety data model are constructed, they will be uniformly labeled as standardized training model data.

[0129] Specifically, by organizing different types of data into tabular formats, data standardization and normalization are achieved. This helps ensure data consistency and accuracy, providing a solid foundation for subsequent data analysis and model building. Anonymizing trainee data protects their privacy and prevents data leakage and misuse. Selecting appropriate models based on data type and characteristics allows for better capture of inherent patterns and relationships within the data. Clustering models are suitable for analyzing the similarities and differences in trainee data; hierarchical analysis models help assess different factors in training data and their interrelationships; regression models can predict performance trends in assessment data; and time series models can analyze temporal changes and periodic patterns in health and safety data. By building models for different types of data, in-depth mining and analysis of standardized training data can be achieved. Unifying data from different sources and types into standardized training model data facilitates data integration and unified management. Furthermore, model-based data representation facilitates visualization, making data analysis results more intuitive and understandable.

[0130] To address the issue that existing technologies lack more precise AI comparison of acquired residency training data, thus hindering convenient trainee management, please refer to [link to relevant documentation]. Figure 1 and Figure 2 This embodiment provides the following technical solution:

[0131] The model building AI analysis unit includes:

[0132] The AI ​​comparative analysis module is used for:

[0133] Retrieve the standardized training model data from the database;

[0134] Standardize the format of standardized residency training model data and residency training model data;

[0135] Next, the key feature data of the standardized training model data and the standardized training model data are extracted. The key feature data are the indicator data of each attribute data in the standardized training model data and the standardized training model data.

[0136] The indicator data of the standardized residency training model data and the residency training model data were respectively converted into curve data;

[0137] After the curve data is converted, an AI comparison algorithm is used to compare the curve data overlap.

[0138] The overlapping comparison results are then modeled using machine learning algorithms, resulting in AI comparison modeling data for the standardized training model data.

[0139] The AI ​​statistical classification module is used for:

[0140] The AI ​​comparative modeling data was retrieved, and the attributes of the AI ​​comparative modeling data were confirmed.

[0141] Among them, the attributes of AI comparative modeling data are whether AI comparative modeling data belongs to basic student data, training data, assessment data, or health and safety data;

[0142] After the AI ​​comparative modeling data attributes are confirmed, attribute label fusion is performed.

[0143] After attribute tag fusion, target analysis data is obtained from the AI ​​comparative modeling data. The target analysis data consists of data from the non-overlapping areas in the AI ​​comparative modeling data.

[0144] Specifically, the AI ​​comparative analysis module performs AI analysis on the acquired standardized residency training data. After retrieving the standardized residency training model data from the database, it performs format unification processing with the residency training model data. This helps eliminate format differences between data, ensuring data consistency and comparability. Key feature data, i.e., indicator data for each attribute, is extracted from both the standardized and residency training model data. This highlights the key points of the data, reduces noise interference, and improves the accuracy and efficiency of subsequent analysis. The indicator data is converted into curve data, and the curve data overlap comparison is performed using an AI comparison algorithm. This method can intuitively display the similarities and differences in key features between the standardized and residency training model data. Curve comparison is not only easy for the human eye to recognize, but the algorithm can also accurately calculate the degree of difference between the two. Machine learning algorithms are used to model the overlap comparison results, obtaining AI comparative modeling data for the residency training model data. This step transforms the comparison results into a quantifiable model, facilitating objective evaluation and optimization of the residency training effect. AI-based comparative modeling can dynamically adjust model parameters based on real-time data, improving the model's adaptability and accuracy. The AI ​​statistical classification module confirms the attributes of the comparative training data, clarifying whether it belongs to basic student data, training data, assessment data, or health and safety data. This attribute confirmation helps to better understand the data's source and characteristics, providing clear guidance for subsequent analysis. By fusing the attribute labels of the AI ​​comparative modeling data, data from different sources can be integrated to form a more comprehensive and richer dataset. Non-overlapping areas in the AI ​​comparative modeling data are extracted as target analysis data, allowing for a more intuitive view of the areas for improvement in AI comparative modeling data with the same attributes.

[0145] To address the issue that existing technologies fail to develop more accurate learning plans based on students' actual situations, thus preventing students from intuitively understanding their learning progress, please refer to [link to relevant documentation]. Figure 1 and Figure 2 This embodiment provides the following technical solution:

[0146] The data analysis and evaluation unit includes:

[0147] The standardized residency training data evaluation module is used for:

[0148] Import the target analysis data into the evaluation rules for quality assessment;

[0149] First, the target analysis data is converted into AI values, which are then converted into area data of non-overlapping regions in the target analysis data.

[0150] The quality range of the target analysis data is determined by the position of the area data values ​​in the evaluation rules.

[0151] The quality range includes excellent, good, passable, and failable.

[0152] The learning plan development module is used for:

[0153] The learning plan for students is determined based on the quality range of the target analysis data;

[0154] Different quality ranges are designed for different learning schemes, and different learning paths are recommended based on different attribute tags;

[0155] Finally, the learning plan and learning path will be transmitted to the terminal and displayed visually.

[0156] Students can view the learning plan and learning path after securely logging in on the terminal.

[0157] Preferably, the AI ​​statistical classification module performs attribute tag fusion after the AI ​​comparative modeling data attributes are confirmed, including:

[0158] Set up a sliding window, read the label information included in the AI ​​comparative modeling data based on the sliding window, and calculate the confidence level of the read label information:

[0159] ;

[0160] in, The confidence level of the read tag information; The average reading rate of the tag information included in the AI ​​comparative modeling data; The radius of the reading range; The rate at which AI-contrastive modeling data passes through the sliding window; The time interval for reading the label information included in the AI ​​comparison modeling data based on the sliding window;

[0161] The confidence level is compared with a preset confidence threshold, and labels with a confidence level greater than the preset confidence threshold are selected as valid labels;

[0162] Extract the feature vector of each valid label and perform matching; package the two matching valid labels as a set of fusion nodes;

[0163] The fusion nodes are normalized, and a fusion diagram is generated based on the normalized fusion nodes.

[0164] Based on the fusion diagram, determine the fusion priority information of the fusion nodes;

[0165] The attribute tags of the fusion nodes are fused according to the fusion priority information. During the fusion process, the fusion rate of each fusion node is monitored and a fusion progress observation chart is generated. The fusion progress observation chart is used to determine whether the attribute tag fusion is completed.

[0166] The working principle of the above technical solution is as follows: In this embodiment, a sliding window is a data processing method that allows data to be processed by sliding within a fixed-size window. When processing label information included in AI contrastive modeling data, the sliding window is set up for reading according to the following steps: Determine the size of the sliding window: Based on the data and the processing task to be performed, determine an appropriate sliding window size, such as how many data points to process each time. Set the window start position: Initially, set the window at the beginning of the data. Read data: Read data of the sliding window size from the dataset. Process data: Perform corresponding label information processing on each read data point. Move the sliding window: As needed, slide the sliding window forward by a fixed step size to continue processing the next batch of data until the entire dataset has been processed. By setting and using a sliding window, large amounts of data can be processed effectively, and the dataset can be processed in batches as needed, reading label information in batches according to the reading efficiency.

[0167] In this embodiment, the confidence level of the read tag information is used to determine whether the read tag information is valid. The confidence level is compared with a preset confidence threshold, and tags with a confidence level greater than the preset confidence threshold are selected as valid tags to achieve tag filtering. Invalid tags are removed so that subsequent tag fusion is performed only on valid tags, reducing the amount of fusion data and improving the fusion rate.

[0168] In this embodiment, for each valid label, its feature vector is obtained through feature extraction technology. Two feature vectors are matched, and a match is indicated when the matching degree is greater than a preset matching degree threshold. The two matching valid labels are then packaged as a set of fusion nodes.

[0169] In this embodiment, the data volume of each fusion node is normalized to determine a normalized value, i.e., a weight coefficient. The fusion node with a larger data volume has a larger normalized value. The nodes are sorted from largest to smallest to serve as a fusion diagram and also to determine the fusion priority information of the fusion nodes.

[0170] In this embodiment, attribute tag fusion is performed on the fusion nodes according to the fusion priority information, which helps improve fusion efficiency. During the fusion process, the fusion rate of each fusion node is monitored, and a fusion progress observation graph is generated to facilitate the generation of information such as the fusion rate and fusion progress of each fusion node. The fusion progress observation graph is used to determine whether attribute tag fusion is complete.

[0171] The beneficial effects of the above technical solution are as follows: A sliding window is set up, and the label information included in the AI ​​comparison modeling data is read based on the sliding window to achieve rapid reading of label information. Based on the calculation of the confidence of the read label information, effective labels are selected, and the effective labels are fused. During the fusion process, the fusion priority information is determined, the fusion rate and the fusion progress information of each fusion node are understood, which facilitates comprehensive monitoring of the label fusion process and improves the accuracy of label fusion.

[0172] Preferably, it also includes: an assessment module, used to assess the learning status of trainees over t learning cycles based on standardized training data, obtain assessment values, and provide corresponding learning suggestions to trainees based on the assessment values;

[0173] The learning status of trainees over t learning cycles is evaluated based on the training data, resulting in evaluation values, including:

[0174] Determine the embedded function of the learning state evaluation model:

[0175] ;

[0176] in, This is the evaluation value of the learning status of the sample learners in the i-th learning cycle; The total number of cycles for calculating the evaluation values ​​of the learning state for the selected sample learners used to train the learning state evaluation model. The connection weights from the input layer to the output layer in the learning state evaluation model; The fitting coefficients of the learning state assessment model; For the sample students in the first The evaluation value of the learning status for each learning cycle; To be based on the total number of cycles Determined cluster centers; The maximum value among the selected cluster centers; The sensitivity threshold for evaluating the learning state model;

[0177] Input the training data of trainees over m learning cycles into the learning status assessment model, and output the assessment value of the user's learning status in each learning cycle to obtain the current learning cycle. And the previous m period ;

[0178] Calculate the evaluation value of the learner's learning status over t learning cycles. :

[0179] ;

[0180] in, For correction factor, ; For the students in the The evaluation value of the learning status for each learning cycle; For the students in the The evaluation value of the learning status for each learning cycle; For the students in the The evaluation value of the learning status for each learning cycle; For the students in the The evaluation value of the learning status for each learning cycle.

[0181] The working principle and beneficial effects of the above technical solution are as follows: Based on the standardized residency training data, the learning status of trainees over t learning cycles is evaluated to obtain evaluation values. These values ​​characterize the trainees' learning status over t learning cycles, allowing teachers to clearly understand each trainee's learning status and provide corresponding learning suggestions, as well as adjust teaching methods and content, thereby improving teaching effectiveness. When evaluating the learning status of trainees over t learning cycles using standardized residency training data to obtain evaluation values, a learning status evaluation model is first constructed based on the sample trainees' data. The embedded function of the learning status evaluation model is then determined, where... To be based on the total number of cycles The determined cluster centers are n / 4. The training data of trainees over m learning cycles are input into the learning status assessment model, which outputs the assessment value of the user's learning status in each learning cycle. This allows for a clear understanding of the trainee's learning status assessment value in each learning cycle. The model also calculates the trainee's learning status assessment value over t learning cycles, meaning it can also understand the trainee's learning status assessment value at any stage (including several learning cycles). This facilitates a comprehensive and accurate understanding of the trainee's learning status, making the given learning suggestions more accurate.

[0182] Preferably, the standardized residency training data acquisition unit retrieves the standardized residency training data from the database, including:

[0183] Receive retrieval requests;

[0184] Based on the retrieval requirements, locate the storage node for the required standardized training data in the database, and retrieve the standardized training data from the database according to the storage node.

[0185] It also includes a retrieval and adjustment module, used for:

[0186] When locating the storage node for the required training data in the database based on the retrieval requirements, the calculation of the first occurrence during storage node location is performed. The probability of a second positioning error;

[0187] ;

[0188] in, The first occurrence during storage node location The probability of a second positioning error; It is a composite symbol; The number of positioning steps during the positioning process; This refers to the classification error rate when distinguishing storage node categories during storage node location;

[0189] According to the occurrence of the first event during storage node location The probability of a second positioning error is used to calculate the positioning error of the storage node.

[0190] ;

[0191] in, This refers to the positioning error of the storage node; Total number of storage node categories; In order to occur the first Storage path difference when a second location error occurs;

[0192] The positioning error of the storage node is compared with the preset positioning error. If the positioning error is determined to be greater than the preset positioning error, a prompt message is generated to adjust the index information of the storage node in the database.

[0193] The working principle and beneficial effects of the above technical solution are as follows: Based on the retrieval requirements, the storage node for the required standardized residency training data in the database is located, and the data is retrieved from the database based on the storage node; thus achieving accurate retrieval of standardized residency training data from the database. When locating the storage node for the required standardized residency training data in the database according to the retrieval requirements, the information for locating the storage node is detected. First, the occurrence of the [missing information] during storage node location is calculated. The probability of a second location error is calculated. Then, the location error of the storage node is calculated and compared with the preset location error. If the location error is greater than the preset location error, it means that the index of the current data retrieved from the database is not accurate enough and the corresponding standardized training data cannot be determined. Therefore, a prompt message is generated to adjust the index information of the storage node in the database. The index information of the storage node in the database is updated and corrected to facilitate the accurate determination of the corresponding standardized training data according to the retrieval requirements in the future.

[0194] Specifically, the standardized residency training data assessment module uses AI numerical conversion to transform the target analysis data, using the area data of non-overlapping regions as the assessment basis. This accurately reflects the trainees' performance and level during the residency training process. Based on the position of the area data in the assessment rules, the quality range of the target analysis data can be determined, including different levels such as excellent, good, passable, and fail. This provides a reliable foundation for subsequent learning plan development. The learning plan development module creates personalized learning plans for trainees based on the quality range of the target analysis data. Different quality ranges correspond to different learning plans, enabling targeted guidance based on the trainees' actual situations. Simultaneously, different learning paths are recommended based on the trainees' attribute tags, further meeting their personalized needs and improving learning effectiveness. The learning plans and paths are visualized on the terminal, allowing trainees to intuitively understand their learning progress and plans. Furthermore, secure login is required to view the data, ensuring data security and privacy. This convenient operation improves the trainees' learning experience and helps enhance their learning motivation. Through in-depth analysis and quality assessment of the residency training data, teachers and administrators can gain a more comprehensive understanding of the trainees' learning status and identify problems and shortcomings in teaching. This helps teachers adjust their teaching strategies, improve their teaching methods, and enhance teaching quality. At the same time, personalized learning plans and pathways can better meet students' needs, improving student satisfaction and learning outcomes.

[0195] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0196] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An AI artificial intelligence-based traditional Chinese medicine residency training data management and analysis platform, characterized in that, The application relates to a rule training data acquisition unit for: Retrieving rule training data from a database, and performing attribute classification on the retrieved rule training data according to the attributes of the data, and sequentially performing data preprocessing on the rule training data of each attribute; A rule training data model construction unit for: Performing model construction on the rule training data after data preprocessing according to the format of the data, and labeling the rule training data after model construction as rule training model data; A model construction AI analysis unit for: Performing data AI comparison and analysis on the rule training model data and rule training standard model data, and performing statistical classification on the analysis results, and standardizing the analysis results after statistical classification as target analysis data; An analysis data evaluation unit for: Performing quality evaluation on the target analysis data according to evaluation rules, wherein the evaluation rules are retrieved from a database, and a personalized learning scheme is formulated according to the quality evaluation results, and the personalized learning scheme is visually displayed on a terminal; The model construction AI analysis unit comprises: An AI comparison and analysis module for: Retrieving rule training standard model data from a database; Uniformly formatting the rule training standard model data and the rule training model data; Extracting key feature data from the rule training standard model data and the rule training model data, wherein the key feature data is index data of each attribute data in the rule training standard model data and the rule training model data; Respectively converting the index data of the rule training standard model data and the rule training model data into curve data; Performing curve data overlap comparison through an AI comparison algorithm after curve data conversion; Modeling the overlap comparison results through a machine learning algorithm, and obtaining AI comparison modeling data of the rule training model data after modeling; An AI statistical classification module for: Retrieving the AI comparison modeling data, and confirming the attributes of the AI comparison modeling data; Wherein the attributes of the AI comparison modeling data are whether the AI comparison modeling data belongs to student basic data, training data, examination data or health and safety data; Performing attribute label fusion after attribute confirmation of the AI comparison modeling data; Obtaining target analysis data of the AI comparison modeling data after attribute label fusion, wherein the target analysis data is data in a non-overlapping region in the AI comparison modeling data; The AI statistical classification module performs attribute label fusion after attribute confirmation of the AI comparison modeling data, including: Setting a sliding window, reading label information included in the AI comparison modeling data based on the sliding window, and calculating the confidence of the read label information: Wherein C is the confidence of the read label information; G is the average reading rate of the label information included in the AI comparison modeling data; R is the radius of the reading range; V is the speed of the AI comparison modeling data through the sliding window; T is the time interval for reading the label information included in the AI comparison modeling data based on the sliding window; Comparing the confidence with a preset confidence threshold, and screening labels with a confidence greater than the preset confidence threshold as effective labels; Extracting feature vectors of each effective label, and performing matching; packaging two matched effective labels as a group of fusion nodes; ​ The fusion nodes are normalized, and a fusion sketch is generated according to the normalized fusion nodes; According to the fusion sketch, the fusion priority information of the fusion nodes is determined; According to the fusion priority information, the attribute label fusion is performed on the fusion nodes, the fusion rate of each fusion node is monitored during the fusion process, and a fusion progress observation graph is generated; and whether the attribute label fusion is completed is judged according to the fusion progress observation graph; The method for performing attribute label fusion comprises: A sliding window is set, and the size of the sliding window is determined to read the label information in the data set; The starting position of the window is set, and data of the size of the sliding window is read from the data set; The read data is processed for label information; The confidence of the label information is judged, the confidence is compared with a preset confidence threshold, and effective labels with a confidence greater than the preset threshold are screened out; Feature extraction is performed on the effective labels to obtain a feature vector, and the feature vector is matched; When the matching degree is greater than a preset matching degree threshold, the matched labels are packaged as fusion nodes; The data of the fusion nodes is normalized to determine a weight coefficient, and the fusion priority information is determined according to the weight coefficient sorting; Based on the fusion priority information, the attribute label fusion is performed, the fusion rate is monitored, and a fusion progress observation graph is generated.

2. The AI artificial intelligence-based traditional Chinese medicine residency training data management and analysis platform of claim 1, characterized in that: The rule training data acquisition unit comprises: The attribute classification module is configured to: The rule training data in the database comprises student basic data, training data, examination data and health and safety data; The student basic data is the name, gender, student ID and major of a student of traditional Chinese medicine; the training data is learning progress records, including completed courses and practical tasks; the examination data is examination scores, including theoretical examination and practical operation examination; and the health and safety data is health records of the student during the training and training-related safety event records, which are laboratory accident records and drug use error records; The student basic data, the training data, the examination data and the health and safety data are classified according to data types, and independent student basic data, training data, examination data and health and safety data are obtained after classification.

3. The AI artificial intelligence-based traditional Chinese medicine residency training data management and analysis platform of claim 2, characterized in that: The rule training data acquisition unit further comprises: The data preprocessing module is configured to: The student basic data, the training data, the examination data and the health and safety data are respectively preprocessed; Data preprocessing comprises data cleaning, data conversion and data set division; The data cleaning is to process the student basic data, the training data, the examination data and the health and safety data for missing values, abnormal values and repeated values; After data cleaning, data conversion is performed, which comprises feature scaling, feature encoding, feature selection and dimension reduction; After data conversion, data set division is performed, which comprises dividing the student basic data, the training data, the examination data and the health and safety data after data conversion into a training set, a validation set and a test set; After data set division, the student basic data, the training data, the examination data and the health and safety data after preprocessing are obtained.

4. The AI artificial intelligence-based traditional Chinese medicine residency training data management and analysis platform of claim 3, characterized in that: The rule training data model construction unit is further configured to: The basic data of the trainees, the training data, the assessment data and the health and safety data are first constructed in a table format; The basic data of the trainees is desensitized after being constructed in a table format; the table of the training data is time series data, the time series data includes training content, training time, training location, training method, trainer information and trainee participation, and the training content is text data, image data and video data; The basic data of the trainees, the training data, the assessment data and the health and safety data constructed in a table format are subjected to model construction; The basic data of the trainees is subjected to model construction through a clustering model; the training data is subjected to model construction through an analytic hierarchy process model; the assessment data is subjected to model construction through a regression model; and the health and safety data is subjected to model construction through a time series model; The basic data of the trainees, the training data, the assessment data and the health and safety data are subjected to model construction, and are uniformly labeled as the rule training model data.

5. The AI artificial intelligence-based traditional Chinese medicine residency training data management and analysis platform of claim 4, characterized in that: The analysis data evaluation unit comprises: The rule training data evaluation module is configured to: Import the target analysis data into the evaluation rules for quality evaluation; The target analysis data is first subjected to AI numerical conversion, wherein the AI numerical conversion is the area data of the non-overlapping region in the target analysis data; The quality range of the target analysis data is determined according to the position of the area data in the evaluation rules; The quality range includes excellent, good, pass and fail; The learning scheme formulation module is configured to: Determine the learning scheme of the trainee according to the quality range of the target analysis data; Different quality ranges correspond to different learning schemes, and different learning paths are recommended according to different attribute labels; The learning scheme and the learning path are finally transmitted to the terminal for visual display; The trainee can view the learning scheme and the learning path after logging in to the terminal.

6. The AI artificial intelligence-based traditional Chinese medicine residency training data management and analysis platform of claim 5, characterized in that: Further comprising: The evaluation module is configured to evaluate the learning state of the trainee in t learning cycles according to the rule training data, and obtain an evaluation value, and give corresponding learning suggestions to the trainee according to the evaluation value; The evaluation module is configured to evaluate the learning state of the trainee in t learning cycles according to the rule training data, and obtain an evaluation value, including: Determine the embedded function of the learning state evaluation model: wherein f(i) is the evaluation value of the learning state of the sample student in the i th learning period; n is the total number of periods for which the evaluation value of the learning state of the sample student is calculated for training the learning state evaluation model; k i is the connection weight from the input layer to the output layer in the learning state evaluation model; β is the fitting coefficient of the learning state evaluation model; X i is the evaluation value of the learning state of the sample student in the i th learning period; C i is the cluster center determined according to the total number of periods n; C max is the maximum value between the selected cluster centers; Z is the sensitivity threshold of the learning state evaluation model; Input the rule training data of the trainee in m learning cycles into the learning state evaluation model, output the evaluation value of the learning state of the user in each learning cycle, and obtain f(t) of the current learning cycle and f(t-1), f(t-2), …, f(t-m) of the previous m periods; Calculate the evaluation value F(t) of the learning state of the trainee in t learning cycles: wherein, is a correction factor, f(t) is an evaluation value of the learning state of the student at the tth learning cycle; f(t-1) is an evaluation value of the learning state of the student at the (t-1)th learning cycle; f(t-2) is an evaluation value of the learning state of the student at the (t-2)th learning cycle; and f(t-m) is an evaluation value of the learning state of the student at the (t-m)th learning cycle.

7. The AI artificial intelligence-based traditional Chinese medicine residency training data management and analysis platform of claim 6, characterized in that: The rule training data acquisition unit acquires the rule training data from the database, including: Receive the acquisition demand; According to the acquisition demand, the storage node of the required rule training data in the database is located, and the rule training data is acquired from the database according to the storage node; Further comprising an acquisition adjustment module configured to: When the storage node is located according to the acquisition demand, the probability of the i-th positioning error of the storage node is calculated; where p i is the probability of the i-th positioning error when locating the storage node; C is a combination symbol; b is the number of positioning steps when locating; and λ is the classification error rate when distinguishing the storage node category when locating the storage node. According to the probability of the i-th positioning error of the storage node, the positioning error of the storage node is calculated; Wherein, W is the positioning error of the storage node; Z is the total number of storage node categories; D is the storage path difference value when the i th positioning error occurs; The positioning error of the storage node is compared with the preset positioning error, and when it is determined that the positioning error is greater than the preset positioning error, the prompt information for adjusting the index information of the storage node in the database is generated.

Images