Anesthesia scheme generation system based on big data analysis

Abstract

The invention discloses an anesthesia scheme generation system based on big data analysis, and relates to the field of artificial intelligence, and the system comprises the steps: carrying out the feature extraction of text demands, and converting the text information into numerical vector representation; performing semantic analysis on the anesthesia demand data, extracting semantic information in a text, and displaying the extracted semantic information in a dynamic knowledge graph form of an entity-relationship-entity structure; constructing an association relationship between the demands; establishing indexes for the anesthesia demand data; based on a project reaction theory, constructing a subject ability value matrix and outputting the matrix as an anesthesia portrait; selecting collaborative filtering and a content-based recommendation algorithm to generate a personalized recommendation list; and the recommendation algorithm is optimized and improved in real time. Automatic classification and efficient retrieval of anesthesia demands are achieved through the dynamic knowledge graph, the classification accuracy and retrieval efficiency of the demands are improved, and accurate recommendation of anesthetists is achieved by constructing anesthesia portraits.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to an anesthesia plan generation system based on big data analysis. Background Art

[0002] With the rapid development of medical technology, the safety and effectiveness of surgical anesthesia are receiving increasing attention. Preoperative anesthesia visits are a crucial step in ensuring the smooth implementation of surgical anesthesia. They require doctors to fully understand the patient's health status, medical history, medication status, and other aspects in order to develop a personalized anesthesia plan. However, traditional preoperative anesthesia visit methods often rely on the doctor's experience and memory, and simply conduct a simple medical history inquiry and physical examination on the patient. This can lead to incomplete information and omission of important information, making it impossible to fully understand the patient's condition, including their physical function status, disease status, medication status, and other aspects, thereby better predicting their anesthesia risks.

[0003] Existing technologies often rely on simple push notifications or recommendations based on historical anesthesia behavior. This approach often fails to accurately meet the individual needs of anesthesia patients. The lack of effective correlation between anesthesia needs makes it difficult for anesthesia patients to develop a systematic knowledge system during the anesthesia process. Summary of the Invention

[0004] In order to solve the above technical problems, an anesthesia plan generation system based on big data analysis is provided. This technical solution solves the problems raised in the above background technology.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] Anesthesia plan generation system based on big data analysis, including:

[0007] a feature extraction module, wherein the feature extraction module collects at least one existing anesthesia requirement data and at least one anesthesia plan, wherein the anesthesia requirement data and the anesthesia plan are in a corresponding relationship, preprocesses the data, extracts features of text-based requirements, and converts the text information into a numerical vector representation;

[0008] A demand analysis module, which performs semantic analysis on anesthesia demand data, extracts semantic information from the text, and uses dependency syntax analysis to present the extracted semantic information in the form of a dynamic knowledge graph of an entity-relationship-entity structure. The semantic information includes entities, relationships, and attributes;

[0009] An association recognition module, which constructs association relationships between requirements based on semantic analysis results and uses the association relationships to achieve rapid positioning and mutual reference between requirements;

[0010] Anesthesia optimization module, which indexes various anesthesia requirement data, uses natural language processing technology to understand anesthesia intentions, and optimizes the anesthesia pathway;

[0011] A data processing module collects historical behavioral data of anesthesia protocols, records preference heat maps and error distribution multi-dimensional data, constructs a subject competency matrix based on item response theory, and outputs the matrix as an anesthesia profile;

[0012] A recommendation generation module, which generates a personalized recommendation list based on the anesthesia profile and anesthesia demand data by selecting collaborative filtering and content-based recommendation algorithms;

[0013] The plan production module optimizes and improves the recommendation algorithm in real time based on the effect of anesthesia feedback, and uses a personalized recommendation list to generate anesthesia plans.

[0014] Preferably, the collecting of historical behavioral data of anesthesia protocols, recording preference heat maps and error distribution multi-dimensional data, constructing a subject competency matrix based on item response theory, and outputting the matrix as an anesthesia profile specifically includes:

[0015] Use data embedding technology to record the actual effect of the anesthesia plan during surgery;

[0016] Use heat map analysis tools to record the distribution of anesthetic regimen preferences during surgery and identify the focus of anesthesia concerns;

[0017] Collect error records of anesthesia protocols during practice or testing, including error types, error rates, and knowledge points involved;

[0018] Visualize the preference heat map to identify key areas and hot topics of anesthesia concern, and extract anesthesia interest points and anesthesia preference features based on the heat map analysis results;

[0019] Conduct statistical analysis on the error records of anesthesia protocols during practice or testing to identify weaknesses and error-prone areas of anesthesia protocols;

[0020] Based on the error distribution analysis results, determine the difficulty and discrimination of each knowledge point;

[0021] The ability value of the anesthesia plan was calculated based on its performance during surgery using the item response theory model;

[0022] The capability values of the anesthesia plan are organized according to the knowledge points, and a capability value matrix is constructed and output as an anesthesia profile, which corresponds to the anesthesia plan.

[0023] Preferably, the selecting collaborative filtering and content-based recommendation algorithms to generate a personalized recommendation list based on anesthesia profiles and anesthesia demand data specifically includes:

[0024] The cosine similarity formula was used to calculate the cosine similarity between anesthesia plans, anesthesia requirements, and anesthesia portraits.

[0025] Based on historical data, set a high similarity threshold;

[0026] Determine whether the cosine similarity between anesthesia plans, anesthesia requirements, or anesthesia portraits is higher than a high similarity threshold. If so, output the group of anesthesia plans, anesthesia requirements, or anesthesia portraits as similar anesthesia plans, anesthesia requirements, or anesthesia portraits; if not, do not output;

[0027] Based on the collaborative filtering algorithm, at least one recommendation list is generated from the perspectives of similar anesthesia plans, similar anesthesia requirements, and similar anesthesia portraits. In the recommendation list, similar anesthesia plans are divided into the same category to obtain anesthesia plan categories, similar anesthesia requirements are divided into the same category to obtain anesthesia requirement categories, and similar anesthesia portraits are divided into the same category to obtain anesthesia portrait categories.

[0028] According to the correspondence between anesthesia requirement data and anesthesia plans, establish a correspondence between anesthesia plans of the same category, anesthesia requirements of the same category, and anesthesia portraits of the same category;

[0029] The personalized recommendation list is re-ranked based on the historical behavioral data of anesthesia regimens and the popularity of anesthesia needs.

[0030] Preferably, the generating of the anesthesia plan using the personalized recommendation list specifically includes:

[0031] Acquire new anesthesia requirement data, calculate the cosine similarity between the new anesthesia requirement data and the anesthesia requirements in the recommendation list, and select the category of the anesthesia requirement data in the recommendation list that has the largest cosine similarity with the new anesthesia requirement data as the target category of the anesthesia requirement data;

[0032] The anesthesia plan category corresponding to the anesthesia requirement data target category is used as the anesthesia plan target category;

[0033] The anesthesia portrait with the largest capability value in the anesthesia portrait category corresponding to the anesthesia plan target category is used as the target anesthesia portrait, and the anesthesia plan corresponding to the target anesthesia portrait is used as the plan required by the new anesthesia requirement data;

[0034] The calculation of the ability value of the anesthesia portrait is as follows:

[0035] When the capability value matrix of the anesthesia portrait is a square matrix, the determinant of the capability value matrix is calculated as the capability value of the anesthesia portrait;

[0036] When the ability value matrix of the anesthesia portrait is not a square matrix, 1 is used as an element to supplement the ability value matrix of the anesthesia portrait into a square matrix and replace the original ability value matrix.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The automated classification and efficient retrieval of anesthesia needs can greatly improve the classification accuracy and retrieval efficiency of needs. The use of association relationships can achieve rapid positioning and mutual reference between needs, which helps anesthesiologists establish a systematic knowledge system during the anesthesia process and improve the anesthesia effect. Based on anesthesia portraits and anesthesia needs data, collaborative filtering and content-based recommendation algorithms are selected to generate personalized recommendation lists, which can achieve accurate recommendations for anesthesiologists and improve the utilization rate of anesthesia needs and the satisfaction of anesthesiologists. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flow chart of the anesthesia plan generation system based on big data analysis of the present invention;

[0040] Figure 2 A flow chart of the method for extracting features from text-based requirements and converting text information into a numerical vector representation;

[0041] Figure 3 A flow chart of the method for performing semantic analysis on anesthesia requirement data and extracting semantic information from text according to the present invention;

[0042] Figure 4 A flow chart of the method for establishing association relationships between requirements of the present invention;

[0043] Figure 5 A flow chart of a method for indexing anesthesia requirement data according to the present invention;

[0044] Figure 6 The present invention constructs a subject capability value matrix and outputs it as a flow chart of the anesthesia profiling method;

[0045] Figure 7 A flow chart of the method for generating a personalized recommendation list by selecting collaborative filtering and content-based recommendation algorithms of the present invention;

[0046] Figure 8 This is a flow chart of the method for generating an anesthesia plan using a personalized recommendation list according to the present invention. DETAILED DESCRIPTION

[0047] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0048] Reference Figure 1 As shown in FIG, the anesthesia plan generation system based on big data analysis includes:

[0049] a feature extraction module, wherein the feature extraction module collects at least one existing anesthesia requirement data and at least one anesthesia plan, wherein the anesthesia requirement data and the anesthesia plan are in a corresponding relationship, preprocesses the data, extracts features of text-based requirements, and converts the text information into a numerical vector representation;

[0050] A demand analysis module, which performs semantic analysis on anesthesia demand data, extracts semantic information from the text, and uses dependency syntax analysis to present the extracted semantic information in the form of a dynamic knowledge graph of an entity-relationship-entity structure. The semantic information includes entities, relationships, and attributes;

[0051] An association recognition module, which constructs association relationships between requirements based on semantic analysis results and uses the association relationships to achieve rapid positioning and mutual reference between requirements;

[0052] Anesthesia optimization module, which indexes various anesthesia requirement data, uses natural language processing technology to understand anesthesia intentions, and optimizes the anesthesia pathway;

[0053] A data processing module collects historical behavioral data of anesthesia protocols, records preference heat maps and error distribution multi-dimensional data, constructs a subject competency matrix based on item response theory, and outputs the matrix as an anesthesia profile;

[0054] A recommendation generation module, which generates a personalized recommendation list based on the anesthesia profile and anesthesia demand data by selecting collaborative filtering and content-based recommendation algorithms;

[0055] The plan production module optimizes and improves the recommendation algorithm in real time based on the effect of anesthesia feedback, and uses a personalized recommendation list to generate anesthesia plans.

[0056] Reference Figure 2 As shown, at least one existing anesthesia requirement data and at least one anesthesia plan are collected. The anesthesia requirement data and the anesthesia plan are in a corresponding relationship. The data are preprocessed, and feature extraction is performed on the text requirements. The text information is converted into a numerical vector representation, which specifically includes:

[0057] Removing irrelevant characters from the anesthesia requirement data text, including web page tags, special symbols, and spaces;

[0058] Using a dictionary-based word segmentation method, the Chinese text of the anesthesia demand data is segmented into at least one independent word.

[0059] Traverse the entire data set, collect all the words that appear, build a vocabulary, and use each word in the vocabulary as a feature candidate;

[0060] Initialize a two-dimensional feature matrix, where rows represent text samples and columns represent words in the vocabulary. Traverse each text sample, count the number of times each word appears in the sample, and fill it into the feature matrix;

[0061] Based on the category distribution and the frequency of occurrence of the vocabulary in the entire data set, the expected frequency formula is used to calculate the expected frequency of each vocabulary in each category;

[0062] For each word, the chi-square value formula is used to calculate the chi-square value between its actual frequency and expected frequency in each category;

[0063] Arrange the chi-square values of all words from largest to smallest, determine the number of feature words n based on the requirements of the specific task and the limitations of computing requirements, and select the top n words with the largest chi-square values as feature words based on the sorting results;

[0064] Construct a bag-of-words model for text, where each word is represented as a one-hot vector with only one element being 1 and the rest being 0.

[0065] By adding the unique hot vectors of each word in the anesthesia requirement data text, a numerical vector representation of the text information is obtained.

[0066] The expected frequency formula is:

[0067]

[0068] Where, E ij is the expected frequency of word j in category i, N i is the number of texts in category i, T j is the number of occurrences of word j in the entire dataset, and N is the number of texts in the entire dataset;

[0069] The chi-square value formula is:

[0070]

[0071] Where K j is the chi-square value of vocabulary j, O ij is the actual frequency of word j in category i, and k is the number of categories.

[0072] Reference Figure 3As shown in the figure, semantic analysis is performed on the anesthesia demand data to extract semantic information from the text. Dependency syntactic analysis is used to present the extracted semantic information in the form of a dynamic knowledge graph with an entity-relationship-entity structure. Specifically, the following are included:

[0073] Using a dependency syntax analysis tool to perform dependency syntax analysis on the preprocessed text to reveal dependency relationships between words, including subject-verb relationships, verb-object relationships, and attributive-predicate relationships;

[0074] The output of dependency syntax analysis is a tree structure representation, where each node represents a word and each edge represents the dependency relationship between words;

[0075] In the dependency parsing tree, identify noun phrases and nouns as entity candidates, and confirm the accuracy of the entities based on part-of-speech tags and context information;

[0076] Store the extracted entities in the entity library;

[0077] In the dependency parsing tree, verb phrases and verbs are identified as relation candidates, and the subject and object of the verb are analyzed to determine the semantic relationship between entities;

[0078] Verify and correct the extracted relationships based on contextual information and domain knowledge;

[0079] Attribute extraction is achieved by analyzing the attributives and modifiers in the dependency syntax tree;

[0080] Create an empty knowledge graph, define the graph structure and the attributes of nodes and edges, set the attribute fields of entities and relationships. The attribute fields of entities include name, type, and description, and the attribute fields of relationships include relationship type and confidence.

[0081] Add the extracted entities and relationships to the knowledge graph, create nodes for each entity, set the node attributes, create edges for each pair of entities with relationships, and set the edge attributes;

[0082] Based on the addition of new text, new entities and relationships are extracted in real time, the knowledge graph is updated, new nodes and edges are added, and the properties of existing nodes and edges are modified;

[0083] The knowledge graph is presented as a graphical interface, with nodes represented as circles and edges as curves, and different types of entities and relationships are distinguished by different colors, shapes and sizes.

[0084] Syntactic analysis is one of the key technologies in natural language processing. It is the process of analyzing the input text sentences to obtain the syntactic structure of the sentences. On the one hand, analyzing the syntactic structure is the inherent requirement of language understanding. Syntactic analysis is an important part of language understanding. On the other hand, it also provides support for other natural language processing tasks. Dependency analysis is used to identify the interdependence between words in a sentence. Dependency syntactic analysis belongs to shallow syntactic analysis.

[0085] Reference Figure 4 As shown, based on the semantic analysis results, the association relationship between requirements is constructed, and the association relationship is used to achieve rapid positioning and mutual reference between requirements. Specifically, it includes:

[0086] Define core relationships between requirements, including equivalence relationships, hierarchical relationships, association relationships, temporal relationships, causal relationships, component relationships, and attribute relationships;

[0087] Based on the semantic analysis results, identify the relationship between requirements;

[0088] Use a graph data structure to represent the relationships between requirements. Nodes in the graph represent requirements, and edges represent relationships. Weights are assigned to each edge based on the strength and type of the relationship.

[0089] Design an anesthesia interface to search for requirements by entity, attribute, or relationship. When an anesthesia requirement query is submitted, traverse and search in the association graph based on the query conditions.

[0090] Display the query results in the form of a network graph, highlighting the nodes and edges that match the query conditions;

[0091] When anesthesia needs to reference a requirement, the system finds at least one other requirement related to the requirement based on the association graph;

[0092] Track and manage reference relationships between requirements, and update the reference relationship diagram in real time when the content or location of requirements changes.

[0093] An anesthesia-friendly query interface is designed to allow anesthesia to search by entity name, attribute or relationship type, and parse the query conditions entered by anesthesia into graph traversal query statements. When anesthesia needs to submit a query, the system performs a depth-first search or breadth-first search in the graph based on the parsed query conditions, and filters out nodes and edges that meet the conditions according to the query conditions.

[0094] Reference Figure 5 As shown in the figure, indexing of each anesthesia requirement data and using natural language processing technology to understand anesthesia intentions and optimize the anesthesia pathway specifically include:

[0095] Initialize the index structure and design a data structure to store the inverted index, where the key is the word and the value is a list of requirements where the word appears;

[0096] Each requirement is represented in the list as a unique identifier of the requirement;

[0097] For each requirement, traverse the text content after word segmentation;

[0098] Determine whether each word already exists in the index. If so, add the unique identifier of the requirement to the document list of the word. If not, create a new entry and use the unique identifier of the requirement as the first element of the document list of the entry.

[0099] Identify the query intent of anesthesia based on machine learning methods;

[0100] Rewrite or expand the query text based on the anesthesia intent and entity recognition results;

[0101] The indexes in the anesthesia requirement database are sorted and filtered, the similarity between the anesthesia requirement and the query text is calculated, and the similarity is sorted according to the similarity.

[0102] Use a pre-trained machine learning model to analyze the user’s query text and identify the intent behind it, which includes information queries (the user wants to obtain specific information), recommendation requests (the user hopes to receive recommendations for related needs), or problem solving (the user is looking for a solution to a specific problem).

[0103] Reference Figure 6 As shown, historical behavioral data of anesthesia protocols are collected, preference heat maps and error distribution multi-dimensional data are recorded, and based on item response theory, a subject competency matrix is constructed and output as an anesthesia profile. Specifically, it includes:

[0104] Use data embedding technology to record the actual effect of the anesthesia plan during surgery;

[0105] Use heat map analysis tools to record the distribution of anesthetic regimen preferences during surgery and identify the focus of anesthesia concerns;

[0106] Collect error records of anesthesia protocols during practice or testing, including error types, error rates, and knowledge points involved;

[0107] Visualize the preference heat map to identify key areas and hot topics of anesthesia concern, and extract anesthesia interest points and anesthesia preference features based on the heat map analysis results;

[0108] Conduct statistical analysis on the error records of anesthesia protocols during practice or testing to identify weaknesses and error-prone areas of anesthesia protocols;

[0109] Based on the error distribution analysis results, determine the difficulty and discrimination of each knowledge point;

[0110] The ability value of the anesthesia plan was calculated based on its performance during surgery using the item response theory model;

[0111] The capability values of the anesthesia plan are organized according to the knowledge points, and a capability value matrix is constructed and output as an anesthesia profile, which corresponds to the anesthesia plan.

[0112] It can capture and record various anesthesia operations during surgery to ensure that anesthesia behaviors can be fully captured.

[0113] Reference Figure 7 As shown in the figure, based on anesthesia profile and anesthesia demand data, collaborative filtering and content-based recommendation algorithms are selected to generate a personalized recommendation list, specifically including:

[0114] The cosine similarity formula was used to calculate the cosine similarity between anesthesia plans, anesthesia requirements, and anesthesia portraits.

[0115] Based on historical data, set a high similarity threshold;

[0116] Determine whether the cosine similarity between anesthesia plans, anesthesia requirements, or anesthesia portraits is higher than a high similarity threshold. If so, output the group of anesthesia plans, anesthesia requirements, or anesthesia portraits as similar anesthesia plans, anesthesia requirements, or anesthesia portraits; if not, do not output;

[0117] Based on the collaborative filtering algorithm, at least one recommendation list is generated from the perspectives of similar anesthesia plans, similar anesthesia requirements, and similar anesthesia portraits. In the recommendation list, similar anesthesia plans are divided into the same category to obtain anesthesia plan categories, similar anesthesia requirements are divided into the same category to obtain anesthesia requirement categories, and similar anesthesia portraits are divided into the same category to obtain anesthesia portrait categories.

[0118] According to the correspondence between anesthesia requirement data and anesthesia plans, establish a correspondence between anesthesia plans of the same category, anesthesia requirements of the same category, and anesthesia portraits of the same category;

[0119] The personalized recommendation list is re-ranked based on the historical behavioral data of anesthesia regimens and the popularity of anesthesia needs.

[0120] The cosine similarity formula is:

[0121]

[0122] Where C os is the cosine similarity between anesthesia, between anesthesia requirements, and between anesthesia portraits and anesthesia requirement features, are the eigenvectors of the anesthesia subject capability matrix or the numerical vectors of the anesthesia requirement data text information, are the modulus lengths of the corresponding vectors.

[0123] Reference Figure 8 As shown in Figure 2, the generation of anesthesia plans using a personalized recommendation list specifically includes:

[0124] Acquire new anesthesia requirement data, calculate the cosine similarity between the new anesthesia requirement data and the anesthesia requirements in the recommendation list, and select the category of the anesthesia requirement data in the recommendation list that has the largest cosine similarity with the new anesthesia requirement data as the target category of the anesthesia requirement data;

[0125] The anesthesia plan category corresponding to the anesthesia requirement data target category is used as the anesthesia plan target category;

[0126] The anesthesia portrait with the largest capability value in the anesthesia portrait category corresponding to the anesthesia plan target category is used as the target anesthesia portrait, and the anesthesia plan corresponding to the target anesthesia portrait is used as the plan required by the new anesthesia requirement data;

[0127] The calculation of the ability value of the anesthesia portrait is as follows:

[0128] When the capability value matrix of the anesthesia portrait is a square matrix, the determinant of the capability value matrix is calculated as the capability value of the anesthesia portrait;

[0129] When the ability value matrix of the anesthesia portrait is not a square matrix, 1 is used as an element to supplement the ability value matrix of the anesthesia portrait into a square matrix and replace the original ability value matrix.

[0130] When screening solutions, solutions are screened based on similarity, and solutions that are close to the needs are selected as alternatives. However, there are multiple alternative solutions, so further selection is required. When selecting, the solution corresponding to the anesthesia profile with the largest ability value is selected as the final solution. The effect of this solution is the better solution among the optional solutions.

[0131] Furthermore, the present solution also proposes a computer-readable storage medium on which a computer-readable program is stored. When the computer-readable program is called, the above-mentioned anesthesia plan generation system based on big data analysis is executed.

[0132] It is understandable that the storage medium may be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid state disk (SSD).

[0133] In summary, the advantages of the present invention are: realizing automatic classification and efficient retrieval of anesthesia needs, greatly improving the classification accuracy and retrieval efficiency of needs, utilizing association relationships to realize rapid positioning and mutual reference between needs, helping anesthesiologists to establish a systematic knowledge system during the anesthesia process and improve the anesthesia effect, and generating personalized recommendation lists based on the selection of collaborative filtering and content-based recommendation algorithms based on anesthesia portraits and anesthesia needs data, which can achieve accurate recommendations for anesthesiologists and improve the utilization rate of anesthesia needs and the satisfaction of anesthesiologists.

[0134] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. An anesthesia plan generation system based on big data analysis, characterized by: include: a feature extraction module, wherein the feature extraction module collects at least one existing anesthesia requirement data and at least one anesthesia plan, wherein the anesthesia requirement data and the anesthesia plan are in a corresponding relationship, preprocesses the data, extracts features of text-based requirements, and converts the text information into a numerical vector representation; A demand analysis module, which performs semantic analysis on anesthesia demand data, extracts semantic information from the text, and uses dependency syntax analysis to present the extracted semantic information in the form of a dynamic knowledge graph of an entity-relationship-entity structure. The semantic information includes entities, relationships, and attributes; An association recognition module, which constructs association relationships between requirements based on semantic analysis results and uses the association relationships to achieve rapid positioning and mutual reference between requirements; Anesthesia optimization module, which indexes various anesthesia requirement data, uses natural language processing technology to understand anesthesia intentions, and optimizes the anesthesia pathway; A data processing module collects historical behavioral data of anesthesia protocols, records preference heat maps and error distribution multi-dimensional data, constructs a subject competency matrix based on item response theory, and outputs the matrix as an anesthesia profile; A recommendation generation module, which generates a personalized recommendation list based on the anesthesia profile and anesthesia demand data by selecting collaborative filtering and content-based recommendation algorithms; The plan production module optimizes and improves the recommendation algorithm in real time based on the effect of anesthesia feedback, and uses a personalized recommendation list to generate anesthesia plans.

2. The anesthesia plan generation system based on big data analysis according to claim 1, characterized in that: The collecting of at least one existing anesthesia requirement data and at least one anesthesia plan, wherein the anesthesia requirement data and the anesthesia plan are in a corresponding relationship, preprocessing the data, performing feature extraction on text-based requirements, and converting the text information into a numerical vector representation specifically includes: Removing irrelevant characters from the anesthesia requirement data text, including web page tags, special symbols, and spaces; Using a dictionary-based word segmentation method, the Chinese text of the anesthesia demand data is segmented into at least one independent word. Traverse the entire data set, collect all the words that appear, build a vocabulary, and use each word in the vocabulary as a feature candidate; Initialize a two-dimensional feature matrix, where rows represent text samples and columns represent words in the vocabulary. Traverse each text sample, count the number of times each word appears in the sample, and fill it into the feature matrix; Based on the category distribution and the frequency of occurrence of the vocabulary in the entire data set, the expected frequency formula is used to calculate the expected frequency of each vocabulary in each category; For each word, the chi-square value formula is used to calculate the chi-square value between its actual frequency and expected frequency in each category; Arrange the chi-square values of all words from largest to smallest, determine the number of feature words n based on the requirements of the specific task and the limitations of computing requirements, and select the top n words with the largest chi-square values as feature words based on the sorting results; Construct a bag-of-words model for text, where each word is represented as a one-hot vector with only one element being 1 and the rest being 0. By adding the unique hot vectors of each word in the anesthesia requirement data text, a numerical vector representation of the text information is obtained.

3. The anesthesia plan generation system based on big data analysis according to claim 2, characterized in that: The semantic analysis of anesthesia demand data is performed to extract semantic information from the text, and the extracted semantic information is presented in the form of a dynamic knowledge graph of entity-relationship-entity structure using dependency syntax analysis. Specifically, the following steps are performed: Using a dependency syntax analysis tool to perform dependency syntax analysis on the preprocessed text to reveal dependency relationships between words, including subject-verb relationships, verb-object relationships, and attributive-predicate relationships; The output of dependency syntax analysis is a tree structure representation, where each node represents a word and each edge represents the dependency relationship between words; In the dependency parsing tree, identify noun phrases and nouns as entity candidates, and confirm the accuracy of the entities based on part-of-speech tags and context information; Store the extracted entities in the entity library; In the dependency parsing tree, verb phrases and verbs are identified as relation candidates, and the subject and object of the verb are analyzed to determine the semantic relationship between entities; Verify and correct the extracted relationships based on contextual information and domain knowledge; Attribute extraction is achieved by analyzing the attributives and modifiers in the dependency syntax tree; Create an empty knowledge graph, define the graph structure and the attributes of nodes and edges, set the attribute fields of entities and relationships. The attribute fields of entities include name, type, and description, and the attribute fields of relationships include relationship type and confidence. Add the extracted entities and relationships to the knowledge graph, create nodes for each entity, set the node attributes, create edges for each pair of entities with relationships, and set the edge attributes; Based on the addition of new text, new entities and relationships are extracted in real time, the knowledge graph is updated, new nodes and edges are added, and the properties of existing nodes and edges are modified; The knowledge graph is presented as a graphical interface, with nodes represented as circles and edges as curves, and different types of entities and relationships are distinguished by different colors, shapes and sizes.

4. The anesthesia plan generation system based on big data analysis according to claim 3, characterized in that: The steps of constructing association relationships between requirements based on semantic analysis results and utilizing the association relationships to achieve rapid location and mutual reference between requirements specifically include: Define core relationships between requirements, including equivalence relationships, hierarchical relationships, association relationships, temporal relationships, causal relationships, component relationships, and attribute relationships; Based on the semantic analysis results, identify the relationship between requirements; Use a graph data structure to represent the relationships between requirements. Nodes in the graph represent requirements, and edges represent relationships. Weights are assigned to each edge based on the strength and type of the relationship. Design an anesthesia interface to search for requirements by entity, attribute, or relationship. When an anesthesia requirement query is submitted, traverse and search in the association graph based on the query conditions. Display the query results in the form of a network graph, highlighting the nodes and edges that match the query conditions; When anesthesia needs to reference a requirement, the system finds at least one other requirement related to the requirement based on the association graph; Track and manage reference relationships between requirements, and update the reference relationship diagram in real time when the content or location of requirements changes.

5. The anesthesia plan generation system based on big data analysis according to claim 4, characterized in that: The aforementioned indexing of various anesthesia requirement data, understanding of anesthesia intentions using natural language processing technology, and optimization of anesthesia pathways specifically include: Initialize the index structure and design a data structure to store the inverted index, where the key is the word and the value is a list of requirements where the word appears; Each requirement is represented in the list as a unique identifier of the requirement; For each requirement, traverse the text content after word segmentation; Determine whether each word already exists in the index. If so, add the unique identifier of the requirement to the document list of the word. If not, create a new entry and use the unique identifier of the requirement as the first element of the document list of the entry. Identify the query intent of anesthesia based on machine learning methods; Rewrite or expand the query text based on the anesthesia intent and entity recognition results; The indexes in the anesthesia requirement database are sorted and filtered, the similarity between the anesthesia requirement and the query text is calculated, and the similarity is sorted according to the similarity.

6. The anesthesia plan generation system based on big data analysis according to claim 5, characterized in that: The collection of historical behavioral data on anesthesia protocols, recording of preference heat maps and error distribution multi-dimensional data, and construction of a subject competency matrix based on item response theory and output as an anesthesia profile specifically include: Use data embedding technology to record the actual effect of the anesthesia plan during surgery; Use heat map analysis tools to record the distribution of anesthetic regimen preferences during surgery and identify the focus of anesthesia concerns; Collect error records of anesthesia protocols during practice or testing, including error types, error rates, and knowledge points involved; Visualize the preference heat map to identify key areas and hot topics of anesthesia concern, and extract anesthesia interest points and anesthesia preference features based on the heat map analysis results; Conduct statistical analysis on the error records of anesthesia protocols during practice or testing to identify weaknesses and error-prone areas of anesthesia protocols; Based on the error distribution analysis results, determine the difficulty and discrimination of each knowledge point; The ability value of the anesthesia plan was calculated based on its performance during surgery using the item response theory model; The capability values of the anesthesia plan are organized according to the knowledge points, and a capability value matrix is constructed and output as an anesthesia profile, which corresponds to the anesthesia plan.

7. The anesthesia plan generation system based on big data analysis according to claim 6, characterized in that: The process of selecting collaborative filtering and content-based recommendation algorithms to generate a personalized recommendation list based on anesthesia profiles and anesthesia demand data specifically includes: The cosine similarity formula was used to calculate the cosine similarity between anesthesia plans, anesthesia requirements, and anesthesia portraits. Based on historical data, set a high similarity threshold; Determine whether the cosine similarity between anesthesia plans, anesthesia requirements, or anesthesia portraits is higher than a high similarity threshold. If so, output the group of anesthesia plans, anesthesia requirements, or anesthesia portraits as similar anesthesia plans, anesthesia requirements, or anesthesia portraits; if not, do not output; Based on the collaborative filtering algorithm, at least one recommendation list is generated from the perspectives of similar anesthesia plans, similar anesthesia requirements, and similar anesthesia portraits. In the recommendation list, similar anesthesia plans are divided into the same category to obtain anesthesia plan categories, similar anesthesia requirements are divided into the same category to obtain anesthesia requirement categories, and similar anesthesia portraits are divided into the same category to obtain anesthesia portrait categories. According to the correspondence between anesthesia requirement data and anesthesia plans, establish a correspondence between anesthesia plans of the same category, anesthesia requirements of the same category, and anesthesia portraits of the same category; The personalized recommendation list is re-ranked based on the historical behavioral data of anesthesia regimens and the popularity of anesthesia needs.

8. The anesthesia plan generation system based on big data analysis according to claim 7, characterized in that: The generation of an anesthesia plan using the personalized recommendation list specifically includes: Acquire new anesthesia requirement data, calculate the cosine similarity between the new anesthesia requirement data and the anesthesia requirements in the recommendation list, and select the category of the anesthesia requirement data in the recommendation list that has the largest cosine similarity with the new anesthesia requirement data as the target category of the anesthesia requirement data; The anesthesia plan category corresponding to the anesthesia requirement data target category is used as the anesthesia plan target category; The anesthesia portrait with the largest capability value in the anesthesia portrait category corresponding to the anesthesia plan target category is used as the target anesthesia portrait, and the anesthesia plan corresponding to the target anesthesia portrait is used as the plan required by the new anesthesia requirement data; The calculation of the ability value of the anesthesia portrait is as follows: When the capability value matrix of the anesthesia portrait is a square matrix, the determinant of the capability value matrix is calculated as the capability value of the anesthesia portrait; When the ability value matrix of the anesthesia portrait is not a square matrix, 1 is used as an element to supplement the ability value matrix of the anesthesia portrait into a square matrix and replace the original ability value matrix.

Images