Medical resource management method and system based on AI model analysis

Abstract

The invention discloses a medical resource management method and system based on AI model analysis, and the method comprises the steps: collecting historical medical information resources and resource consumption data through a medical terminal, and converting the historical medical information resources and resource consumption data into pre-training text data in a unified text format, and enabling the pre-training text data to be used for constructing and training a BRET semantic analysis model; afterwards, collecting real-time demand data of each medical unit, converting the real-time demand data into demand word vectors by utilizing a semantic analysis model, clustering and grouping the medical units based on word vector similarity, and screening out an optimal demand word vector; in the next period, semantic analysis is conducted on the medical information resource big data collected in real time, and optimal medical information resource data are screened out in combination with the optimal demand word vectors. And finally, the optimized resource data is encrypted, uploaded and shared through the block chain network. According to the invention, accurate analysis of medical information resource demands can be effectively realized, reasonable utilization of medical information resources is realized, and information resource sharing efficiency among multiple medical units is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical data analysis, and more specifically, to a medical resource management method and system based on AI model analysis. Background Art

[0002] With the continuous deepening of medical information construction, the accumulation of massive medical data provides a rich data foundation for medical resource management and information service needs. However, there are many problems in the current medical resource management, such as uneven resource distribution, serious information island phenomenon, and great challenges in data security and privacy protection. In addition, in the traditional medical information resource sharing analysis, there is a lack of rational collection, analysis and statistics of medical information needs, and a lack of semantic analysis of information resource needs based on high precision, which leads to the inability to integrate and analyze the information resource needs of different medical units and the demand analysis of shared data.

[0003] Therefore, traditional medical resource management methods can no longer meet the needs of modern information-based medical systems, and an innovative management method and system are urgently needed to solve these problems. Summary of the invention

[0004] The present invention overcomes the defects of the prior art and proposes a medical resource management method and system based on AI model analysis.

[0005] The first aspect of the present invention provides a medical resource management method based on AI model analysis, comprising:

[0006] In a historical period of time, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals;

[0007] Convert the historical medical information resource data and the historical medical resource consumption data into a unified text format to form pre-trained text data, construct an AI-based BRET semantic analysis model, and import the pre-trained text data into the semantic analysis model for pre-training;

[0008] In a preset analysis cycle, the medical information demand data of multiple medical units are collected, and the text conversion and semantic analysis of the medical information demand data are performed through the semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through the similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the optimal demand word vector;

[0009] In the next preset cycle, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data;

[0010] A blockchain network is built based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network.

[0011] In this solution, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals within a historical period of time, specifically:

[0012] Set a historical time period;

[0013] In a historical period of time, historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals;

[0014] Each medical unit corresponds to a medical terminal, and data connections are established between the medical terminals through a local area network.

[0015] In this solution, the historical medical information resource data and the historical medical resource consumption data are converted into a unified text format to form pre-trained text data, an AI-based BRET semantic analysis model is constructed, and the pre-trained text data is imported into the semantic analysis model for pre-training, specifically:

[0016] Performing data cleaning and unified text format conversion on the historical medical information resource data and the historical medical resource consumption data to form pre-trained text data;

[0017] Build an AI-based BRET semantic analysis model;

[0018] Import the pre-trained text data into the semantic analysis model, segment the pre-trained text data using the WordPiece word segmentation algorithm to obtain phrase data;

[0019] The phrase data is input into the semantic analysis model for training. The input data is encoded through a multi-layer Transformer encoder, and the model parameters are adjusted based on the built-in loss function. During the back propagation process, the gradient descent method is used to minimize the loss function and continuously update the model parameters, finally obtaining the pre-trained semantic analysis model.

[0020] In this solution, within a preset analysis cycle, medical information demand data of multiple medical units are collected, and text conversion and semantic analysis are performed on the medical information demand data through a semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the preferred demand word vector, which is specifically:

[0021] Collect medical information demand data from multiple medical units within a preset analysis cycle;

[0022] Import the medical information demand data into the semantic analysis model for unified text format conversion to obtain demand text data;

[0023] The demand text data is segmented and input into the semantic analysis model for semantic analysis and word vectors are output, which are marked as demand word vectors.

[0024] Analyze the demand word vector of each medical unit based on multiple medical units;

[0025] A clustering model based on DBSCAN was constructed, with medical units as clustering units, the radiation radius of sample points set, and the demand word vectors of each medical unit as clustering sample data and imported into the clustering model for clustering. The clustering process performed data clustering through similarity analysis of demand word vectors. The similarity analysis process was combined with a semantic analysis model. The distance between the demand word vectors of two medical units was calculated through the semantic analysis model. The distance was calculated using the standard Euclidean distance, and the degree of similarity between the medical units was determined based on the size of the distance. Multiple cluster groups were formed through clustering.

[0026] Each cluster group included at least one medical unit;

[0027] Among multiple cluster groups, the group with the highest number of medical units is selected and marked as the preferred cluster group;

[0028] In the preferred clustering group, the corresponding demand word vectors of all medical units in the group are integrated to obtain the preferred demand word vector.

[0029] In this solution, in the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and the medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data, which is specifically:

[0030] In the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal;

[0031] Perform unified text formatting on medical information resource data and medical resource consumption data to form medical information resource big data;

[0032] Based on information type, the medical information resource big data is divided into multiple information text data sets;

[0033] Import the information text dataset into the semantic analysis model for word segmentation and semantic analysis, and output the word vector to obtain the current word vector;

[0034] Through the semantic analysis model, the current word vector and the preferred word vector are calculated based on the standard Euclidean distance similarity. If the similarity is greater than the preset similarity, the corresponding information text data set is used as the preferred text data set;

[0035] Analyze all information text data sets, and select all preferred text data sets for integration to obtain preferred medical information resource data.

[0036] In this solution, a blockchain network is constructed based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network, specifically:

[0037] The number of network nodes is set based on the number of all medical units, and the blockchain network is constructed according to the number of network nodes;

[0038] In the next preset period, the corresponding preferred medical information resource data is screened out based on all medical units;

[0039] Encrypting the preferred medical information resource data of a medical unit and generating a hash value;

[0040] The encrypted preferred medical information resource data and hash value are packaged to generate a data block, which is then uploaded to the blockchain network through a network node of a medical unit for data sharing.

[0041] The second aspect of the present invention further provides a medical resource management system based on AI model analysis, the system comprising: a memory, a processor, the memory comprising a medical resource management program based on AI model analysis, the medical resource management program based on AI model analysis when executed by the processor implements the following steps:

[0042] In a historical period of time, historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals;

[0043] Convert the historical medical information resource data and the historical medical resource consumption data into a unified text format to form pre-trained text data, construct an AI-based BRET semantic analysis model, and import the pre-trained text data into the semantic analysis model for pre-training;

[0044] In a preset analysis cycle, the medical information demand data of multiple medical units are collected, and the text conversion and semantic analysis of the medical information demand data are performed through the semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through the similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the optimal demand word vector;

[0045] In the next preset cycle, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data;

[0046] A blockchain network is built based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network.

[0047] The third aspect of the present invention also provides a computer-readable storage medium, which includes a medical resource management program based on AI model analysis. When the medical resource management program based on AI model analysis is executed by a processor, the steps of the medical resource management method based on AI model analysis as described in any one of the above items are implemented.

[0048] The present invention discloses a medical resource management method and system based on AI model analysis. First, historical medical information resources and resource consumption data are collected through medical terminals and converted into pre-trained text data in a unified text format for building and training a BRET semantic analysis model. Subsequently, the real-time demand data of each medical unit is collected, converted into demand word vectors using a semantic analysis model, and the medical units are clustered and grouped based on the similarity of the word vectors to screen out the optimal demand word vectors. In the next cycle, semantic analysis is performed on the medical information resource big data collected in real time, and the preferred medical information resource data is screened out in combination with the preferred demand word vectors. Finally, the preferred resource data is encrypted, uploaded and shared through a blockchain network. Through the present invention, it is possible to effectively realize accurate analysis of medical information resource needs, realize rational use of medical information resources, and improve the efficiency of information resource sharing among multiple medical units. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A flowchart of a medical resource management method based on AI model analysis of the present invention is shown;

[0050] Figure 2 A block diagram of a medical resource management system based on AI model analysis of the present invention is shown. DETAILED DESCRIPTION

[0051] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0053] Figure 1 A flow chart of a medical resource management method based on AI model analysis of the present invention is shown.

[0054] like Figure 1 As shown, the first aspect of the present invention provides a medical resource management method based on AI model analysis, comprising:

[0055] S102, collecting historical medical information resource data and historical medical resource consumption data of multiple medical units through medical terminals within a historical time period;

[0056] S104, converting the historical medical information resource data and the historical medical resource consumption data into a unified text format to form pre-trained text data, constructing an AI-based BRET semantic analysis model, and importing the pre-trained text data into the semantic analysis model for pre-training;

[0057] S106, within a preset analysis cycle, collect medical information demand data of multiple medical units, perform text conversion and semantic analysis on the medical information demand data through a semantic analysis model to form a demand word vector, cluster and group the medical units based on the demand word vector, and perform data clustering through similarity analysis of the demand word vectors to form multiple cluster groups, mark the group with the largest number of medical units in the cluster group, and select the corresponding word vector to obtain the optimal demand word vector;

[0058] S108, within the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and the medical resource consumption data are uniformly formatted to form medical information resource big data, and the semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data;

[0059] S110, building a blockchain network based on multiple medical units, packaging and encrypting the preferred medical information resource data of each medical unit and uploading it to the blockchain network.

[0060] According to an embodiment of the present invention, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through a medical terminal within a historical time period, specifically:

[0061] Set a historical time period;

[0062] In a historical period of time, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals;

[0063] Each medical unit corresponds to a medical terminal, and data connections are established between the medical terminals through a local area network.

[0064] It should be noted that medical information resource data specifically refers to information data with medical analysis value, including patient information, medical history information, diagnosis information, treatment plans, etc., and medical resource consumption data specifically refers to statistical data on resource consumption, including human resources, material resources, equipment, drugs, etc. Text formatting of medical resource consumption data means texting the corresponding statistical values ​​and statistical data type information so as to facilitate subsequent semantic analysis. For example, statistics on the type of a certain disease and the number of visits, patient registration statistics, workload statistics of doctors and nurses, frequency of equipment use, etc., will be processed into information text.

[0065] The historical time period is a longer data collection period set by the user.

[0066] According to an embodiment of the present invention, the historical medical information resource data and the historical medical resource consumption data are converted into a unified text format to form pre-trained text data, an AI-based BRET semantic analysis model is constructed, and the pre-trained text data is imported into the semantic analysis model for pre-training, specifically:

[0067] Performing data cleaning and unified text format conversion on the historical medical information resource data and the historical medical resource consumption data to form pre-trained text data;

[0068] Build an AI-based BRET semantic analysis model;

[0069] Import the pre-trained text data into the semantic analysis model, segment the pre-trained text data using the WordPiece word segmentation algorithm, and obtain phrase data;

[0070] The phrase data is input into the semantic analysis model for training. The input data is encoded through a multi-layer Transformer encoder, and the model parameters are adjusted based on the built-in loss function. During the back propagation process, the gradient descent method is used to minimize the loss function and continuously update the model parameters, finally obtaining the pre-trained semantic analysis model.

[0071] It should be noted that the BERT (Bidirectional Encoder Representations from Transformers) model is a deep learning model based on the Transformer architecture. It is an AI language model that learns and analyzes word vectors on large-scale texts through pre-training, and can capture the complex relationship between words. The present invention uses this model to perform effective semantic analysis and semantic similarity analysis on the information needs of medical institutions, realize accurate analysis of medical information resource needs, and subsequently realize the rational use of medical information resources. In the pre-trained semantic analysis model, the model is trained based on medical information resources, and semantic analysis of information resources is performed to form a large language model. Through this model, accurate semantic understanding of medical information can be effectively achieved. Subsequently, the model is used to perform effective semantic analysis of medical information needs. In the BRET semantic analysis model, the collected historical data (i.e., pre-trained text data) will be used as context data in the semantic analysis process.

[0072] It is worth mentioning that in traditional medical information resource query and analysis, there is a lack of rational collection, analysis and statistics of medical information needs, and a lack of highly precise semantic analysis of information resource needs. Therefore, it is impossible to conduct integrated analysis of the information resource needs of different medical units and analyze the needs for shared data.

[0073] According to an embodiment of the present invention, within a preset analysis cycle, medical information demand data of multiple medical units are collected, and text conversion and semantic analysis are performed on the medical information demand data through a semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the preferred demand word vector, which is specifically:

[0074] Collect medical information demand data from multiple medical units within a preset analysis cycle;

[0075] Import the medical information demand data into the semantic analysis model for unified text format conversion to obtain demand text data;

[0076] The demand text data is segmented and input into the semantic analysis model for semantic analysis and word vectors are output, which are marked as demand word vectors.

[0077] Analyze the demand word vector of each medical unit based on multiple medical units;

[0078] A clustering model based on DBSCAN was constructed, with medical units as clustering units, the radiation radius of sample points set, and the demand word vectors of each medical unit as clustering sample data and imported into the clustering model for clustering. The clustering process performed data clustering through similarity analysis of demand word vectors. The similarity analysis process was combined with a semantic analysis model. The distance between the demand word vectors of two medical units was calculated through the semantic analysis model. The distance was calculated using the standard Euclidean distance, and the degree of similarity between the medical units was determined based on the size of the distance. Multiple cluster groups were formed through clustering.

[0079] Each cluster group included at least one medical unit;

[0080] Among multiple cluster groups, the group with the highest number of medical units is selected and marked as the preferred cluster group;

[0081] In the preferred clustering group, the corresponding demand word vectors of all medical units in the group are integrated to obtain the preferred demand word vector.

[0082] It should be noted that the unified text format conversion is based on the preset text format standard to perform data standardization conversion, which is conducive to the unified data analysis of the information system. Medical information demand data specifically refers to the demand of medical units for information data, generally for query, statistics, calculation and other needs. For example, it is necessary to query the incidence rate of a certain disease and patient information of medical units within a certain area. In the similarity between medical units determined by the distance, the greater the distance, the smaller the similarity, which is inversely proportional. In an embodiment of the present invention, an AI-based semantic analysis model is adopted, and the context of the semantic analysis model is constructed through pre-training of historical data. The semantic model has the function of semantically analyzing the medical information resource needs of multiple medical units. Furthermore, the information resource needs of different medical units are clustered through a clustering model. The clustering process is clustered based on semantic similarity analysis. Therefore, the semantic analysis model is combined in the clustering process, and the clustering process of the original clustering model is improved at the same time, to realize the medical unit demand analysis based on semantic analysis, improve the accuracy of the medical information resource demand analysis of medical units, and effectively classify the information resources in the subsequent process to screen out the preferred medical information resource data. In the process of resource sharing, the blockchain is introduced to encrypt and share the screened data, and the preferred medical information resource data is shared on the chain to the overall network, so as to realize the precise sharing of high-demand shared data on the chain and effectively improve the security of the data sharing process.

[0083] According to an embodiment of the present invention, in the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and the medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data, which is specifically:

[0084] In the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal;

[0085] Perform unified text formatting on medical information resource data and medical resource consumption data to form medical information resource big data;

[0086] Based on information type, the medical information resource big data is divided into multiple information text data sets;

[0087] Import the information text dataset into the semantic analysis model for word segmentation and semantic analysis, and output the word vector to obtain the current word vector;

[0088] Through the semantic analysis model, the current word vector and the preferred word vector are calculated based on the standard Euclidean distance similarity. If the similarity is greater than the preset similarity, the corresponding information text data set is used as the preferred text data set;

[0089] Analyze all information text data sets, and select all preferred text data sets for integration to obtain preferred medical information resource data.

[0090] It should be noted that the preferred medical information resource data is an effectively screened information resource with a high information utilization rate. Therefore, it will be encrypted and shared as data of the blockchain network in the future. The multiple types of medical information resources based on information type can be divided by user-defined types. Each medical unit corresponds to a preferred medical information resource data.

[0091] According to an embodiment of the present invention, the blockchain network is constructed based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network, specifically:

[0092] The number of network nodes is set based on the number of all medical units, and the blockchain network is constructed according to the number of network nodes;

[0093] In the next preset period, the corresponding preferred medical information resource data is screened out based on all medical units;

[0094] Encrypting the preferred medical information resource data of a medical unit and generating a hash value;

[0095] The encrypted preferred medical information resource data and hash value are packaged to generate a data block, which is then uploaded to the blockchain network through a network node of a medical unit for data sharing.

[0096] It should be noted that the encryption may adopt an asymmetric encryption algorithm, and the hash value is used to verify the integrity and uniqueness of the data.

[0097] Through the present invention, efficient data sharing between different medical units can be achieved. At the same time, valuable medical information data can be stored through blockchain, which improves the security of information sharing and provides a safe and fast sharing solution for the information analysis needs of different medical units. In addition, the present invention records high-frequency and high-correlation information demand analysis based on a clustering model through periodic information analysis demand collection and corresponding demand semantic analysis, and packages high-utilization information data and sends it to the blockchain network, effectively improving data utilization and data sharing effectiveness, while meeting the security of medical data, effectively promoting medical informatization and providing high-quality system services.

[0098] According to an embodiment of the present invention, it also includes:

[0099] Mark a current medical unit;

[0100] Within a preset analysis cycle, the information resource acquisition record of the current medical unit is obtained, and the corresponding information acquisition party medical unit is marked as an information acquisition unit through the information resource acquisition record;

[0101] Establish information dependency mapping between current medical units and information acquisition units, and record the mapping relationship;

[0102] Analyze all medical units and store the corresponding mapping relationships in a graph structure to obtain a dependency mapping graph;

[0103] In the dependency mapping graph, each graph node is a medical unit, and if there is an edge between graph nodes, the edge indicates that there is a dependency relationship between the two graph nodes;

[0104] In the dependency map, all medical units are counted, and medical units with information dependencies and more than three are marked and grouped together to obtain a shared group. In the structure of the dependency map, the nodes in the shared group form a closed loop.

[0105] A shared local area network is built based on the network nodes of the medical units in the sharing group. When the medical units in the sharing group issue information resource retrieval requirements, information resource retrieval is performed through the shared local area network first, and secondly, through the blockchain network.

[0106] It should be noted that the present invention uses a graph structure to store the information dependency relationship between medical units, and can efficiently count, analyze and screen the relationship. The edges between the graph nodes are directed edges, which record the dependency and dependence relationship between the two nodes. The medical units with information dependencies and more than three in number are marked and grouped together. Specifically, multiple consecutive nodes can be connected by edges in the dependency mapping graph for marking. For example, there are three graph nodes, representing three medical units A, B, and C. The relationship between ABC includes: A depends on B, B depends on C, and C depends on A. The dependency relationship between the three forms a closed loop. Therefore, the three medical units A, B, and C can constitute a sharing group.

[0107] Information resource acquisition records refer to the records of data acquired by medical units based on the sharing process during information demand analysis.

[0108] It is worth mentioning that the incidence of certain diseases is regional. Furthermore, different medical units have different needs for information resources such as illness. For example, in a certain medical unit, the number of patients with a certain disease increases dramatically. Therefore, it is necessary to retrieve information data of related disease types. It is necessary to query the corresponding disease information or the incidence rate in other units, patient medical records, etc. to improve disease prevention and control capabilities. For example, the current medical resources of medical units are tight, and it is necessary to retrieve and query the medical resource statistics of other medical units to better optimize resource allocation. Therefore, the information resource needs of different medical units are different, and there is a certain dependence in a certain period of time.

[0109] In a shared LAN, the information resource data collected by medical units in real time can be directly shared in the LAN, which can effectively improve sharing efficiency.

[0110] In an embodiment of the present invention, based on the analysis cycle, the information dependency relationship of the medical units is saved in the form of a graph structure, and node analysis is performed in the graph. A sharing group is formed for multiple medical units that meet the corresponding dependency relationships. In the real-time sharing process of multiple medical units, the medical units can give priority to information search and screening through the sharing group based on information demand analysis to meet the real-time sharing needs. When the sharing group cannot meet the information needs, a large range of multi-network node information can be obtained through the blockchain network, thereby improving the efficiency of information resource sharing between multiple medical units and the efficiency of completing information retrieval and matching tasks.

[0111] Figure 2 A block diagram of a medical resource management system based on AI model analysis of the present invention is shown.

[0112] The second aspect of the present invention further provides a medical resource management system 2 based on AI model analysis, the system comprising: a memory 21, a processor 22, the memory comprising a medical resource management program based on AI model analysis, the medical resource management program based on AI model analysis when executed by the processor implements the following steps:

[0113] In a historical period of time, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals;

[0114] Convert the historical medical information resource data and the historical medical resource consumption data into a unified text format to form pre-trained text data, construct an AI-based BRET semantic analysis model, and import the pre-trained text data into the semantic analysis model for pre-training;

[0115] In a preset analysis cycle, the medical information demand data of multiple medical units are collected, and the text conversion and semantic analysis of the medical information demand data are performed through the semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through the similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the optimal demand word vector;

[0116] In the next preset cycle, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data;

[0117] A blockchain network is built based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network.

[0118] According to an embodiment of the present invention, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through a medical terminal within a historical time period, specifically:

[0119] Set a historical time period;

[0120] In a historical period of time, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals;

[0121] Each medical unit corresponds to a medical terminal, and data connections are established between the medical terminals through a local area network.

[0122] It should be noted that medical information resource data specifically refers to information data with medical analysis value, including patient information, medical history information, diagnosis information, treatment plans, etc., and medical resource consumption data specifically refers to statistical data on resource consumption, including human resources, material resources, equipment, drugs, etc. Text formatting of medical resource consumption data means texting the corresponding statistical values ​​and statistical data type information so as to facilitate subsequent semantic analysis. For example, statistics on the type of a certain disease and the number of visits, patient registration statistics, workload statistics of doctors and nurses, frequency of equipment use, etc., will be processed into information text.

[0123] The historical time period is a longer data collection period set by the user.

[0124] According to an embodiment of the present invention, the historical medical information resource data and the historical medical resource consumption data are converted into a unified text format to form pre-trained text data, an AI-based BRET semantic analysis model is constructed, and the pre-trained text data is imported into the semantic analysis model for pre-training, specifically:

[0125] Performing data cleaning and unified text format conversion on the historical medical information resource data and the historical medical resource consumption data to form pre-trained text data;

[0126] Build an AI-based BRET semantic analysis model;

[0127] Import the pre-trained text data into the semantic analysis model, segment the pre-trained text data using the WordPiece word segmentation algorithm to obtain phrase data;

[0128] The phrase data is input into the semantic analysis model for training. The input data is encoded through a multi-layer Transformer encoder, and the model parameters are adjusted based on the built-in loss function. During the back propagation process, the gradient descent method is used to minimize the loss function and continuously update the model parameters, finally obtaining the pre-trained semantic analysis model.

[0129] It should be noted that the BERT (Bidirectional Encoder Representations from Transformers) model is a deep learning model based on the Transformer architecture. It is an AI language model that learns and analyzes word vectors on large-scale texts through pre-training, and can capture the complex relationship between words. The present invention uses this model to perform effective semantic analysis and semantic similarity analysis on the information needs of medical institutions, realize accurate analysis of medical information resource needs, and subsequently realize the rational use of medical information resources. In the pre-trained semantic analysis model, the model is trained based on medical information resources, and semantic analysis of information resources is performed to form a large language model. Through this model, accurate semantic understanding of medical information can be effectively achieved. Subsequently, the model is used to perform effective semantic analysis of medical information needs. In the BRET semantic analysis model, the collected historical data (i.e., pre-trained text data) will be used as context data in the semantic analysis process.

[0130] It is worth mentioning that in traditional medical information resource query and analysis, there is a lack of rational collection, analysis and statistics of medical information needs, and a lack of highly precise semantic analysis of information resource needs. Therefore, it is impossible to conduct integrated analysis of the information resource needs of different medical units and analyze the needs for shared data.

[0131] According to an embodiment of the present invention, within a preset analysis cycle, medical information demand data of multiple medical units are collected, and text conversion and semantic analysis are performed on the medical information demand data through a semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the preferred demand word vector, which is specifically:

[0132] Collect medical information demand data from multiple medical units within a preset analysis cycle;

[0133] Import the medical information demand data into the semantic analysis model for unified text format conversion to obtain demand text data;

[0134] The demand text data is segmented and input into the semantic analysis model for semantic analysis and word vectors are output, which are marked as demand word vectors.

[0135] Analyze the demand word vector of each medical unit based on multiple medical units;

[0136] A clustering model based on DBSCAN was constructed, with medical units as clustering units, the radiation radius of sample points set, and the demand word vectors of each medical unit as clustering sample data and imported into the clustering model for clustering. The clustering process performed data clustering through similarity analysis of demand word vectors. The similarity analysis process was combined with a semantic analysis model. The distance between the demand word vectors of two medical units was calculated through the semantic analysis model. The distance was calculated using the standard Euclidean distance, and the degree of similarity between the medical units was determined based on the size of the distance. Multiple cluster groups were formed through clustering.

[0137] Each cluster group included at least one medical unit;

[0138] Among multiple cluster groups, the group with the highest number of medical units is selected and marked as the preferred cluster group;

[0139] In the preferred clustering group, the corresponding demand word vectors of all medical units in the group are integrated to obtain the preferred demand word vector.

[0140] It should be noted that the unified text format conversion is based on the preset text format standard to perform data standardization conversion, which is conducive to the unified data analysis of the information system. Medical information demand data specifically refers to the demand of medical units for information data, generally for query, statistics, calculation and other needs. For example, it is necessary to query the incidence rate of a certain disease and patient information of medical units within a certain area. In the similarity between medical units determined by the distance, the greater the distance, the smaller the similarity, which is inversely proportional. In an embodiment of the present invention, an AI-based semantic analysis model is adopted, and the context of the semantic analysis model is constructed through pre-training of historical data. The semantic model has the function of semantically analyzing the medical information resource needs of multiple medical units. Furthermore, the information resource needs of different medical units are clustered through a clustering model. The clustering process is clustered based on semantic similarity analysis. Therefore, the semantic analysis model is combined in the clustering process, and the clustering process of the original clustering model is improved at the same time, to realize the medical unit demand analysis based on semantic analysis, improve the accuracy of the medical information resource demand analysis of medical units, and effectively classify the information resources in the subsequent process to screen out the preferred medical information resource data. In the process of resource sharing, the blockchain is introduced to encrypt and share the screened data, and the preferred medical information resource data is shared on the chain to the overall network, so as to realize the precise sharing of high-demand shared data on the chain and effectively improve the security of the data sharing process.

[0141] According to an embodiment of the present invention, in the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and the medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data, which is specifically:

[0142] In the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal;

[0143] Perform unified text formatting on medical information resource data and medical resource consumption data to form medical information resource big data;

[0144] Based on information type, the medical information resource big data is divided into multiple information text data sets;

[0145] Import the information text dataset into the semantic analysis model for word segmentation and semantic analysis, and output the word vector to obtain the current word vector;

[0146] Through the semantic analysis model, the current word vector and the preferred word vector are calculated based on the standard Euclidean distance similarity. If the similarity is greater than the preset similarity, the corresponding information text data set is used as the preferred text data set;

[0147] Analyze all information text data sets, and select all preferred text data sets for integration to obtain preferred medical information resource data.

[0148] It should be noted that the preferred medical information resource data is an effectively screened information resource with a high information utilization rate. Therefore, it will be encrypted and shared as data of the blockchain network in the future. The multiple types of medical information resources based on information type can be divided by user-defined types. Each medical unit corresponds to a preferred medical information resource data.

[0149] According to an embodiment of the present invention, the blockchain network is constructed based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network, specifically:

[0150] The number of network nodes is set based on the number of all medical units, and the blockchain network is constructed according to the number of network nodes;

[0151] In the next preset period, the corresponding preferred medical information resource data is screened out based on all medical units;

[0152] Encrypting the preferred medical information resource data of a medical unit and generating a hash value;

[0153] The encrypted preferred medical information resource data and hash value are packaged to generate a data block, which is then uploaded to the blockchain network through a network node of a medical unit for data sharing.

[0154] It should be noted that the encryption may adopt an asymmetric encryption algorithm, and the hash value is used to verify the integrity and uniqueness of the data.

[0155] Through the present invention, efficient data sharing between different medical units can be achieved. At the same time, valuable medical information data can be stored through blockchain, which improves the security of information sharing and provides a safe and fast sharing solution for the information analysis needs of different medical units. In addition, the present invention records high-frequency and high-correlation information demand analysis based on a clustering model through periodic information analysis demand collection and corresponding demand semantic analysis, and packages high-utilization information data and sends it to the blockchain network, effectively improving data utilization and data sharing effectiveness, while meeting the security of medical data, effectively promoting medical informatization and providing high-quality system services.

[0156] The third aspect of the present invention also provides a computer-readable storage medium, which includes a medical resource management program based on AI model analysis. When the medical resource management program based on AI model analysis is executed by a processor, the steps of the medical resource management method based on AI model analysis as described in any one of the above items are implemented.

[0157] The present invention discloses a medical resource management method and system based on AI model analysis. First, historical medical information resources and resource consumption data are collected through medical terminals and converted into pre-trained text data in a unified text format for building and training a BRET semantic analysis model. Subsequently, the real-time demand data of each medical unit is collected, converted into demand word vectors using a semantic analysis model, and the medical units are clustered and grouped based on the similarity of the word vectors to screen out the optimal demand word vectors. In the next cycle, semantic analysis is performed on the medical information resource big data collected in real time, and the preferred medical information resource data is screened out in combination with the preferred demand word vectors. Finally, the preferred resource data is encrypted, uploaded and shared through a blockchain network. Through the present invention, it is possible to effectively realize accurate analysis of medical information resource needs, realize rational use of medical information resources, and improve the efficiency of information resource sharing among multiple medical units.

[0158] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0159] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0160] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0161] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0162] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0163] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A medical resource management method based on AI model analysis, characterized in that: include: In a historical period of time, historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals; Convert the historical medical information resource data and the historical medical resource consumption data into a unified text format to form pre-trained text data, construct an AI-based BRET semantic analysis model, and import the pre-trained text data into the semantic analysis model for pre-training; In a preset analysis cycle, the medical information demand data of multiple medical units are collected, and the text conversion and semantic analysis of the medical information demand data are performed through the semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through the similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the optimal demand word vector; In the next preset cycle, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data; A blockchain network is built based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network.

2. According to claim 1, a medical resource management method based on AI model analysis is characterized in that: In a historical time period, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through the medical terminal, specifically: Set a historical time period; In a historical period of time, historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals; Each medical unit corresponds to a medical terminal, and data connections are established between the medical terminals through a local area network.

3. A medical resource management method based on AI model analysis according to claim 2, characterized in that: The historical medical information resource data and the historical medical resource consumption data are converted into a unified text format to form pre-trained text data, an AI-based BRET semantic analysis model is constructed, and the pre-trained text data is imported into the semantic analysis model for pre-training, specifically: Performing data cleaning and unified text format conversion on the historical medical information resource data and the historical medical resource consumption data to form pre-trained text data; Build an AI-based BRET semantic analysis model; Import the pre-trained text data into the semantic analysis model, segment the pre-trained text data using the WordPiece word segmentation algorithm to obtain phrase data; The phrase data is input into the semantic analysis model for training. The input data is encoded through a multi-layer Transformer encoder, and the model parameters are adjusted based on the built-in loss function. During the back propagation process, the gradient descent method is used to minimize the loss function and continuously update the model parameters, finally obtaining the pre-trained semantic analysis model.

4. According to claim 3, a medical resource management method based on AI model analysis is characterized in that: In the preset analysis cycle, medical information demand data of multiple medical units are collected, and text conversion and semantic analysis are performed on the medical information demand data through a semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the preferred demand word vector, which is specifically: Collect medical information demand data from multiple medical units within a preset analysis cycle; Import the medical information demand data into the semantic analysis model for unified text format conversion to obtain demand text data; The demand text data is segmented and input into the semantic analysis model for semantic analysis and word vectors are output, which are marked as demand word vectors. Analyze the demand word vector of each medical unit based on multiple medical units; A clustering model based on DBSCAN was constructed, with medical units as clustering units, the radiation radius of sample points set, and the demand word vectors of each medical unit as clustering sample data and imported into the clustering model for clustering. The clustering process performed data clustering through similarity analysis of demand word vectors. The similarity analysis process was combined with a semantic analysis model. The distance between the demand word vectors of two medical units was calculated through the semantic analysis model. The distance was calculated using the standard Euclidean distance, and the degree of similarity between the medical units was determined based on the size of the distance. Multiple cluster groups were formed through clustering. Each cluster group included at least one medical unit; Among multiple cluster groups, the group with the highest number of medical units is selected and marked as the preferred cluster group; In the preferred clustering group, the corresponding demand word vectors of all medical units in the group are integrated to obtain the preferred demand word vector.

5. A medical resource management method based on AI model analysis according to claim 4, characterized in that: In the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and the medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data, which is specifically: In the next preset period, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal; Perform unified text formatting on medical information resource data and medical resource consumption data to form medical information resource big data; Based on information type, the medical information resource big data is divided into multiple information text data sets; Import the information text dataset into the semantic analysis model for word segmentation and semantic analysis, and output the word vector to obtain the current word vector; Through the semantic analysis model, the current word vector and the preferred word vector are calculated based on the standard Euclidean distance similarity. If the similarity is greater than the preset similarity, the corresponding information text data set is used as the preferred text data set; Analyze all information text data sets, and select all preferred text data sets for integration to obtain preferred medical information resource data.

6. A medical resource management method based on AI model analysis according to claim 5, characterized in that: The blockchain network is constructed based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network, specifically: The number of network nodes is set based on the number of all medical units, and the blockchain network is constructed according to the number of network nodes; In the next preset period, the corresponding preferred medical information resource data is screened out based on all medical units; Encrypting the preferred medical information resource data of a medical unit and generating a hash value; The encrypted preferred medical information resource data and hash value are packaged to generate a data block, which is then uploaded to the blockchain network through a network node of a medical unit for data sharing.

7. A medical resource management system based on AI model analysis, characterized in that: The system includes: a memory and a processor, wherein the memory includes a medical resource management program based on AI model analysis, and when the medical resource management program based on AI model analysis is executed by the processor, the following steps are implemented: In a historical period of time, historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals; Convert the historical medical information resource data and the historical medical resource consumption data into a unified text format to form pre-trained text data, construct an AI-based BRET semantic analysis model, and import the pre-trained text data into the semantic analysis model for pre-training; In a preset analysis cycle, the medical information demand data of multiple medical units are collected, and the text conversion and semantic analysis of the medical information demand data are performed through the semantic analysis model to form a demand word vector. The medical units are clustered and grouped based on the demand word vector. The clustering process performs data clustering through the similarity analysis of the demand word vector to form multiple cluster groups. The group with the largest number of medical units in the cluster group is marked, and the corresponding word vector is screened to obtain the optimal demand word vector; In the next preset cycle, the medical information resource data and medical resource consumption data of the medical unit are collected through the medical terminal, and the medical information resource data and medical resource consumption data are uniformly formatted to form medical information resource big data. The semantic analysis model is combined with the preferred demand word vector to perform semantic analysis and resource data screening on the medical information resource big data to obtain the preferred medical information resource data; A blockchain network is built based on multiple medical units, and the preferred medical information resource data of each medical unit is packaged, encrypted and uploaded to the blockchain network.

8. A medical resource management system based on AI model analysis according to claim 7, characterized in that: In a historical time period, the historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through the medical terminal, specifically: Set a historical time period; In a historical period of time, historical medical information resource data and historical medical resource consumption data of multiple medical units are collected through medical terminals; Each medical unit corresponds to a medical terminal, and data connections are established between the medical terminals through a local area network.

9. A medical resource management system based on AI model analysis according to claim 7, characterized in that: The historical medical information resource data and the historical medical resource consumption data are converted into a unified text format to form pre-trained text data, an AI-based BRET semantic analysis model is constructed, and the pre-trained text data is imported into the semantic analysis model for pre-training, specifically: Performing data cleaning and unified text format conversion on the historical medical information resource data and the historical medical resource consumption data to form pre-trained text data; Build an AI-based BRET semantic analysis model; Import the pre-trained text data into the semantic analysis model, segment the pre-trained text data using the WordPiece word segmentation algorithm to obtain phrase data; The phrase data is input into the semantic analysis model for training. The input data is encoded through a multi-layer Transformer encoder, and the model parameters are adjusted based on the built-in loss function. During the back propagation process, the gradient descent method is used to minimize the loss function and continuously update the model parameters, finally obtaining the pre-trained semantic analysis model.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a medical resource management program based on AI model analysis. When the medical resource management program based on AI model analysis is executed by a processor, the steps of the medical resource management method based on AI model analysis as described in any one of claims 1 to 6 are implemented.

Images