Hospital exit and admission record analysis method and system based on big data analysis

Abstract

The invention belongs to the technical field of data analysis, and discloses a hospital exit and admission record analysis method and system based on big data analysis. The method comprises the following steps: in a cloud data center, constructing a hospital exit and admission record analysis engine by using an artificial intelligence algorithm according to historical hospital exit and admission record big data, and deploying the hospital exit and admission record analysis engine to a data server of each terminal; using the data server to collect real-time hospital exit and admission records of the user, and using the hospital exit and admission record analysis engine to analyze the real-time hospital exit and admission records to obtain a real-time analysis report, and storing the real-time analysis report to the cloud data center; and a user terminal is connected with the cloud data center, and real-time hospital exit and admission records and real-time analysis reports corresponding to the user are downloaded and visualized. The method solves the problems that in the prior art, the data dimension is single, feature extraction is shallow, analysis stays at the correlation level, causal insight is lacked, the analysis result applicability is poor, and intelligent decision support is lacked.

Description

Technical Field

[0001] The present invention belongs to the field of data analysis technology, and specifically relates to a method and system for analyzing admission and discharge records based on big data analysis. Background Art

[0002] As society accelerates its aging process, nursing homes, as important elderly care service institutions, face increasingly complex operational and management challenges. Admission and discharge records are crucial information for nursing home management, containing key data such as the residents' health status, care needs, cost settlement, and service evaluation. Traditional nursing home admission and discharge record management methods typically rely on paper documents or simple spreadsheets, which suffer from scattered information, untimely updates, weak analytical capabilities, and low data utilization.

[0003] The existing technology has the following defects: 1) Single data dimension and shallow feature extraction: Existing methods often focus on only a few explicit features, such as age, gender, and primary diagnosis, ignoring a wealth of potentially valuable information. For dynamically changing data (such as multiple medical visits, vital signs during hospitalization, medication adjustments, etc.), existing technologies either statically process the data, failing to capture complex, nonlinear temporal dependencies and pattern evolution. More critically, existing methods rarely effectively integrate user data with rich nursing knowledge (such as disease relationships, drug interactions, and treatment guidelines). This results in a lack of medical background support for analysis results, making it difficult to interpret the data and, even more so, unable to guide clinical practice. 2) Analysis remains at the correlation level, lacking causal insights: Most existing analyses can only reveal correlations between data. The lack of causal reasoning means that correlation cannot be distinguished from causation, which may lead to biased analytical conclusions and even mislead clinical decision-making and medical insurance policy-making. 3) Poor applicability of analysis results and lack of intelligent decision-making support: Many analysis results are only presented in the form of static reports or charts, with low information density, making them difficult to quickly understand and utilize. Nurses or managers need to spend a lot of time screening and interpreting massive amounts of information, which is inefficient; they also lack the ability to convert analysis results into specific and executable recommendations. Summary of the Invention

[0004] In order to solve the problems existing in the prior art, such as single data dimension, shallow feature extraction, analysis remaining at the correlation level, lack of causal insight, poor applicability of analysis results, and lack of intelligent decision support, the present invention aims to provide an admission and discharge record analysis method and system based on big data analysis.

[0005] The technical solution adopted in the present invention is: A method for analyzing admission and discharge records based on big data analysis includes the following steps: In the cloud data center, based on the big data of historical admission and discharge records, an admission and discharge record analysis engine is built using artificial intelligence algorithms and deployed to the data server of each terminal; Use the data server to collect users' real-time admission and discharge records, and use the admission and discharge record analysis engine to analyze the real-time admission and discharge records, obtain real-time analysis reports, and store them in the cloud data center; Use the user terminal to connect to the cloud data center to download and visualize the user's corresponding real-time admission and discharge records and real-time analysis reports.

[0006] Furthermore, the admission and discharge record analysis engine includes a dynamic multidimensional feature generation model, a nursing knowledge graph, a path mining and causal reasoning model, and an admission and discharge record analysis model that are connected in sequence.

[0007] Furthermore, the dynamic multidimensional feature generation model is constructed based on the Autoencoder-BERT-LSTM-GRU-AttentionKGEmb algorithm, and the dynamic multidimensional feature generation model is provided with a static feature extraction module constructed based on the Autoencoder algorithm, a dynamic feature generation module constructed based on the BERT-LSTM-GRU algorithm, and a knowledge graph embedding module constructed based on the AttentionKGEmb algorithm; The path mining and causal reasoning model is constructed based on the VAE-MTCD algorithm, and the path mining and causal reasoning model is provided with a path mining module constructed based on the VAE algorithm and a causal reasoning module constructed based on the MTCD algorithm; The admission and discharge record analysis model is constructed based on the MLP-MAPGRPO-cGAN algorithm, and the admission and discharge record analysis model is provided with an admission and discharge record analysis module constructed based on the MLP algorithm, an application decision generation module constructed based on the MAPGRPO algorithm, and an analysis report generation module constructed based on the cGAN algorithm.

[0008] Furthermore, in the cloud data center, based on the big data of historical admission and discharge records, an admission and discharge record analysis engine is built using artificial intelligence algorithms and deployed to the data server of each terminal, including the following steps: In the cloud data center, historical admission and discharge record big data and nursing knowledge big data are collected and pre-processed to obtain a number of pre-processed historical admission and discharge record data and a number of pre-processed nursing knowledge data; Based on a number of pre-processed nursing knowledge data, a natural language processing algorithm is used to construct a nursing knowledge graph including a number of knowledge named entities and a number of knowledge entity relationships; Based on several pre-processed historical admission and discharge record data and nursing knowledge graphs, a deep learning algorithm is used to build a dynamic multidimensional feature generation model and obtain several historical dynamic multidimensional features; Based on several historical dynamic multidimensional features, generative machine learning and causal reasoning algorithms are used to build path mining and causal reasoning models, and obtain several historical causal reasoning results; Based on several historical causal reasoning results, a model for analyzing hospital admission and discharge records was constructed using generative machine learning and reinforcement learning algorithms. Integrate the dynamic multidimensional feature generation model, nursing knowledge graph, path mining and causal reasoning model, and admission and discharge record analysis model to obtain the initial admission and discharge record analysis engine; Using the federated learning mechanism, the initial admission and discharge record analysis engine is adjusted to obtain the final admission and discharge record analysis engine, and deployed to the data server of each terminal.

[0009] Furthermore, the federated learning mechanism is used to adjust the initial admission and discharge record analysis engine to obtain the final admission and discharge record analysis engine, which is then deployed to the data server of each terminal. The following steps are included: The initial hospital admission and discharge record analysis engine is enhanced with differential privacy to obtain an optimized hospital admission and discharge record analysis engine; Deploy a coordination server in the cloud data center, connect the coordination server to several terminal data servers, and deploy an optimized admission and discharge record analysis engine to each data server; On each data server, based on a number of collected training admission and discharge records, the corresponding optimized admission and discharge record analysis engine is trained to obtain a trained admission and discharge record analysis engine; Perform model compression / distillation on the trained admission and discharge record analysis engine to obtain a lightweight admission and discharge record analysis engine, extract the corresponding training model parameters, and send them to the coordination server; In the coordination server, all training model parameters are integrated to obtain comprehensive model parameters, and the comprehensive model parameters are sent to all data servers; On each data server, the trained admission and discharge record analysis engine is adjusted according to the comprehensive model parameters to obtain the final admission and discharge record analysis engine.

[0010] Furthermore, a data server is used to collect the user's real-time admission and discharge records, and an admission and discharge record analysis engine is used to analyze the real-time admission and discharge records to obtain a real-time analysis report, which is then stored in a cloud data center. The following steps are included: Use the data server to collect the user's real-time admission and discharge records, and pre-process them to obtain the pre-processed real-time admission and discharge records; Use the dynamic multi-dimensional feature generation model and nursing knowledge graph of the admission and discharge record analysis engine to extract the real-time dynamic features of the pre-processed real-time admission and discharge records; Based on real-time dynamic multi-dimensional features, the path mining and causal reasoning model of the admission and discharge record analysis engine is used to generate corresponding real-time causal reasoning results; Based on the real-time dynamic multi-dimensional features and real-time causal reasoning results, the hospital discharge record analysis model of the hospital discharge record analysis engine is used to generate corresponding real-time analysis reports; The user's real-time admission and discharge records and real-time analysis reports are uploaded to the cloud data center, and the blockchain network of the cloud data center is used to distribute the real-time admission and discharge records and real-time analysis reports.

[0011] Furthermore, the dynamic multidimensional feature generation model and nursing knowledge graph of the admission and discharge record analysis engine are used to extract the real-time dynamic features of the pre-processed real-time admission and discharge records, including the following steps: Input the pre-processed real-time admission and discharge records into the dynamic multi-dimensional feature generation model of the admission and discharge record analysis engine; Use the static feature extraction module of the dynamic multi-dimensional feature generation model to extract the real-time static features of the pre-processed real-time admission and discharge records; Based on the real-time static features, the dynamic feature generation module of the dynamic multi-dimensional feature generation model is used to extract the real-time dynamic features of the pre-processed real-time admission and discharge records; The knowledge graph embedding module of the dynamic multidimensional feature generation model is used to embed the nursing knowledge graph into the real-time static features to obtain real-time semantic association features, and then combined with the corresponding real-time dynamic features to obtain real-time dynamic multidimensional features.

[0012] Furthermore, based on the real-time dynamic multi-dimensional features, the path mining and causal reasoning model of the hospital admission and discharge record analysis engine is used to generate corresponding real-time causal reasoning results, including the following steps: Input real-time dynamic multi-dimensional features into the path mining and causal reasoning model of the hospital admission and discharge record analysis engine; Use the path mining module of the path mining and causal reasoning model to extract real-time nursing path information from real-time dynamic multidimensional features; According to the real-time dynamic multidimensional features and real-time nursing path information, the causal reasoning module of the path mining and causal reasoning model is used to generate the corresponding real-time causal reasoning results.

[0013] Furthermore, based on the real-time dynamic multidimensional features and the real-time causal reasoning results, the hospital discharge record analysis model of the hospital discharge record analysis engine is used to generate a corresponding real-time analysis report, including the following steps: Inputting real-time dynamic multi-dimensional features and real-time causal reasoning results into the admission and discharge record analysis model of the admission and discharge record analysis engine; Based on the real-time dynamic multi-dimensional features, the admission and discharge record analysis module of the admission and discharge record analysis model is used to generate corresponding real-time admission and discharge record analysis results; Based on the real-time admission and discharge record analysis results and real-time causal reasoning results, the application decision generation module of the admission and discharge record analysis model is used to generate real-time application decisions for users; According to the real-time admission and discharge record analysis results, real-time causal reasoning results and real-time application decisions, the analysis report generation module of the admission and discharge record analysis model is used to generate the corresponding real-time analysis report.

[0014] A hospital admission and discharge record analysis system based on big data analysis is used to implement a hospital admission and discharge record analysis method. The system includes a cloud data center, several data servers, and several user terminals. The cloud data center is respectively communicated with the several data servers, and the several data servers are set in the corresponding terminals. The data servers are respectively communicated with several data systems in the terminals, and each user terminal is communicated with a data server within the communication range.

[0015] The beneficial effects of the present invention are: The present invention provides a method and system for analyzing hospital admission and discharge records based on big data analysis. The hospital admission and discharge record analysis engine, by combining a variety of artificial intelligence technologies, can not only extract static features such as basic information, diagnosis, nursing, and expenses of users, but also capture dynamic features such as hospitalization duration and changes in condition, and integrate nursing knowledge to form a rich and accurate multi-dimensional feature representation, overcoming the shortcomings of the existing technology of single and shallow features, making the characterization of users more comprehensive and in-depth; the hospital admission and discharge record analysis engine performs causal reasoning, which can go beyond simple correlation analysis and explore the causal relationship between diagnosis and treatment measures and user outcomes, greatly improving the interpretability of the analysis and the clinical guidance value, and providing a theoretical basis for optimizing nursing pathways; based on the extracted multi-dimensional features and causal reasoning results, specific and personalized applications are generated Decision-making recommendations (such as nursing adjustments and risk warnings) make up for the shortcomings of existing technologies that can only provide generalized information but cannot support individualized and precise decision-making, and directly empower nursing staff and managers; the analysis engine is deployed on the nursing home's data server, which can quickly collect real-time admission and discharge records and analyze them. At the same time, data sharing and result distribution are realized through the cloud platform, and user terminals can obtain analysis reports and visualization results in real time. This overcomes the problems of delayed updates and slow responses in traditional systems, and meets the needs of modern nursing for real-time monitoring and rapid decision-making; through the construction of cloud data centers, cross-terminal data aggregation and unified deployment of analysis engines are realized, which makes it possible to break data silos and conduct larger-scale collaborative data analysis, thereby improving the breadth and depth of analysis and generating insights that are difficult to obtain from a single terminal.

[0016] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of the admission and discharge record analysis method based on big data analysis in the present invention.

[0018] Figure 2 It is a structural block diagram of the admission and discharge record analysis system based on big data analysis in the present invention. DETAILED DESCRIPTION

[0019] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments.

[0020] Example 1: like Figure 1 As shown, this embodiment provides a method for analyzing admission and discharge records based on big data analysis, comprising the following steps: S1: In the cloud data center, based on the big data of historical admission and discharge records, an admission and discharge record analysis engine is built using artificial intelligence algorithms and deployed to the data server of each terminal; The hospital admission and discharge record analysis engine includes a dynamic multi-dimensional feature generation model, a nursing knowledge graph, a path mining and causal reasoning model, and an admission and discharge record analysis model that are connected in sequence; The dynamic multi-dimensional feature generation model is based on the Auto-Associative Neural Network (Autoencoder), Bidirectional Encoder Representations from Transformers (BERT), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Attention-based Knowledge Graph Embedding (AttentionKGEmb) algorithms. The model consists of a static feature extraction module based on the Autoencoder algorithm, a dynamic feature generation module based on the BERT-LSTM-GRU algorithm, and a knowledge graph embedding module based on the AttentionKGEmb algorithm. The path mining and causal reasoning model is built based on the variational autoencoder (VAE)-medical temporal causal discovery (MTCD) algorithm, and the path mining and causal reasoning model is equipped with a path mining module based on the VAE algorithm and a causal reasoning module based on the MTCD algorithm; The hospital admission and discharge record analysis model is built based on the Multi-Layer Perceptron (MLP)-Multi-Agent Parallel Group Relative Policy Optimization (MAPGRPO)-Conditional Generative Adversarial Network (cGAN) algorithm. The model includes an admission and discharge record analysis module based on the MLP algorithm, an application decision generation module based on the MAPGRPO algorithm, and an analysis report generation module based on the cGAN algorithm. The application decision generation module includes a user-level application decision generation layer, a coordination layer, and a system-level application decision generation layer connected in sequence. The user-level application decision generation layer is provided with a first set of multiple optimization objectives and a user-level intelligent agent. The system-level application decision generation layer is provided with a second set of multiple optimization objectives and a plurality of parallel system-level intelligent agents. In the cloud data center, based on the big data of historical admission and discharge records, an admission and discharge record analysis engine is built using artificial intelligence algorithms and deployed to the data server of each terminal. This includes the following steps: S1-1: In the cloud data center, historical admission and discharge record big data and nursing knowledge big data are collected from sources such as nursing home information systems, electronic medical record systems, and medical insurance systems. Pre-processing such as data cleaning, standardization, and format conversion is performed to obtain a number of pre-processed historical admission and discharge record data and a number of pre-processed nursing knowledge data. S1-2: Based on some pre-processed nursing knowledge data, use natural language processing algorithms to construct a nursing knowledge graph including some knowledge named entities and some knowledge entity relationships; S1-3: Based on some pre-processed historical admission and discharge record data and nursing knowledge graph, a deep learning algorithm is used to build a dynamic multidimensional feature generation model and obtain several historical dynamic multidimensional features; S1-4: Based on several historical dynamic multidimensional features, generative machine learning and causal reasoning algorithms are used to build a path mining and causal reasoning model, and obtain several historical causal reasoning results; S1-5: Based on several historical causal reasoning results, a model for analyzing hospital admission and discharge records was constructed using generative machine learning and reinforcement learning algorithms. S1-6: Integrate the dynamic multidimensional feature generation model, nursing knowledge graph, path mining and causal reasoning model, and admission and discharge record analysis model to obtain the initial admission and discharge record analysis engine; S1-7: Use the federated learning mechanism to adjust the initial admission and discharge record analysis engine to obtain the final admission and discharge record analysis engine, and deploy it to the data server of each terminal, including the following steps: S1-7-1: Perform differential privacy enhancement on the initial admission and discharge record analysis engine to obtain an optimized admission and discharge record analysis engine; S1-7-2: Deploy a coordination server in the cloud data center, connect the coordination server to several terminal data servers, and deploy the optimized admission and discharge record analysis engine to each data server; S1-7-3: On each data server, based on the collected training admission and discharge records, the corresponding optimized admission and discharge record analysis engine is trained to obtain a trained admission and discharge record analysis engine; S1-7-4: Perform model compression / distillation on the trained admission and discharge record analysis engine to obtain a lightweight admission and discharge record analysis engine, extract the corresponding training model parameters, and send them to the coordination server; S1-7-5: In the coordination server, all training model parameters are integrated to obtain comprehensive model parameters, and the comprehensive model parameters are sent to all data servers; S1-7-6: On each data server, adjust the trained admission and discharge record analysis engine according to the comprehensive model parameters to obtain the final admission and discharge record analysis engine; S2: Use the data server to collect the user's real-time admission and discharge records, and use the admission and discharge record analysis engine to analyze the real-time admission and discharge records to obtain a real-time analysis report, which is stored in the cloud data center. This includes the following steps: S2-1: Use the data server to collect the user's real-time admission and discharge records, and pre-process them to obtain the pre-processed real-time admission and discharge records; S2-2: Use the dynamic multidimensional feature generation model and nursing knowledge graph of the admission and discharge record analysis engine to extract the real-time dynamic features of the pre-processed real-time admission and discharge records, including the following steps: S2-2-1: Input the pre-processed real-time admission and discharge records into the dynamic multi-dimensional feature generation model of the admission and discharge record analysis engine; S2-2-2: Use the static feature extraction module of the dynamic multidimensional feature generation model to extract real-time static features such as user basic information (such as age segment, gender code, and medical insurance type weight), health status, nursing needs, diagnosis information (such as disease cluster labels and primary diagnosis severity), nursing information (such as surgery type code and medication frequency), cost information (such as average daily cost and reimbursement ratio), and time information (such as hospitalization duration and holiday marking) from pre-processed real-time admission and discharge records; S2-2-3: Based on the real-time static features, the dynamic feature generation module of the dynamic multi-dimensional feature generation model is used to extract the real-time dynamic features of the pre-processed real-time admission and discharge records; Specifically, the BRET processing of the dynamic feature generation module is used to process the real-time text data (such as medical record descriptions and nursing records) in the real-time admission and discharge records after preprocessing, and the real-time text semantic features are extracted. At the same time, based on the real-time text semantic features and real-time static features, a temporal LSTM-GRU dual-channel network is used to extract the real-time time series information (such as daily medication, test results, and expense flow) from the real-time structured data of the preprocessed real-time admission and discharge records. The dynamic change trends and key change rates (such as the change rate of diagnosis information, the change rate of nursing information, and the change rate of expense information) are captured. S2-2-4: Use the knowledge graph embedding module of the dynamic multidimensional feature generation model to embed the nursing knowledge graph containing several knowledge named entities such as diseases, symptoms, drugs, examinations, and nursing care, as well as several knowledge relationship entities, into real-time static features to obtain real-time semantic association features. Combined with the corresponding real-time dynamic features, real-time dynamic multidimensional features are obtained. S2-3: Based on the real-time dynamic multi-dimensional features, the path mining and causal reasoning model of the admission and discharge record analysis engine are used to generate corresponding real-time causal reasoning results, including the following steps: S2-3-1: Input real-time dynamic multi-dimensional features into the path mining and causal reasoning model of the hospital admission and discharge record analysis engine; S2-3-2: Path mining module using path mining and causal reasoning models extracts real-time nursing pathway information (such as the sequence from symptoms to precautions, and from precautions to nursing plan selection) from real-time dynamic multidimensional features, and can identify key decision points and possible abnormal patterns in the pathway; S2-3-3: Based on the real-time dynamic multidimensional features and real-time nursing pathway information, the causal reasoning module of the pathway mining and causal reasoning model is used to generate corresponding real-time causal reasoning results; S2-4: Based on the real-time dynamic multi-dimensional features and real-time causal reasoning results, the hospital discharge record analysis model of the hospital discharge record analysis engine is used to generate a corresponding real-time analysis report, including the following steps: S2-4-1: Input the real-time dynamic multi-dimensional features and real-time causal reasoning results into the admission and discharge record analysis model of the admission and discharge record analysis engine; S2-4-2: Based on the real-time dynamic multi-dimensional features, the admission and discharge record analysis module of the admission and discharge record analysis model is used to generate corresponding real-time admission and discharge record analysis results; Real-time admission and discharge record analysis results include real-time disease severity analysis results, real-time recovery progress analysis results, real-time nursing plan effectiveness analysis results, and real-time diagnostic accuracy analysis results; S2-4-3: Based on the real-time admission and discharge record analysis results and the real-time causal reasoning results, the application decision generation module of the admission and discharge record analysis model is used to generate the user's real-time application decision, including the following steps: The application decision generation module is constructed based on the Multi-Agent Parallel Group Related Policy Optimization (MAPGRPO) algorithm, and includes a user-level application decision generation layer, a coordination layer, and a system-level application decision generation layer connected in sequence. The user-level application decision generation layer is provided with a first set of multiple optimization objectives and user-level agents, and the system-level application decision generation layer is provided with a second set of multiple optimization objectives and a plurality of parallel system-level agents. The goals of the first multi-optimization objective set focus on the health and well-being of individual users and may include the following: Maximizing user health benefits: This is the primary goal, including increasing user survival rates, improving quality of life, and reducing complications and adverse events; Minimize user pain and discomfort: For example, reduce pain, discomfort and side effects during care and improve user comfort; Optimize user experience: for example, shorten hospital stays, increase nursing convenience, and improve communication between nursing staff; Maximize user satisfaction: meet user needs and expectations, and improve user trust and satisfaction with nursing services and the nursing team; Minimize the user's financial burden: For example, optimize care plans, reduce unnecessary examinations and care, and reduce the user's care costs; The second largest optimization objective set focuses on the overall operational efficiency and effectiveness of the nursing home, and may include the following objectives: Maximize nursing home operational efficiency: for example, improve bed utilization, optimize resource allocation, and shorten the average length of stay for users; Maximize the economic benefits of nursing homes: for example, increase nursing service revenue, control nursing costs, and improve the profitability of nursing homes; Improve nursing quality: for example, reduce nursing error rates, increase user satisfaction, and enhance the reputation of nursing homes; Optimize resource allocation: For example, rationally allocate nursing resources to avoid resource waste and improve resource utilization efficiency; Improve staff efficiency: For example, optimize staffing, improve work efficiency, and reduce staff fatigue and stress; S2-4-3-1: Based on the real-time admission and discharge record analysis results and the real-time causal reasoning results, based on the first multi-optimization goal set, use the user-level agent of the user-level application decision generation layer of the application decision generation module to generate application decisions and obtain real-time user-level application decisions; Real-time user-level application decisions include: Whether further examination or care is needed: For example, based on the user's symptoms and examination results, further imaging examinations or laboratory tests are recommended to confirm the diagnosis; Whether the care plan needs to be adjusted: For example, based on the user's condition changes and drug response, it is recommended to adjust the drug dosage or change the care plan; Whether risk warning is needed: For example, based on the user's risk prediction results, remind the caregiver to pay attention to possible complications or adverse events of the user and take preventive measures; Whether to arrange follow-up visits for the user: For example, based on the user's recovery, it is recommended to arrange regular follow-up visits for the user to monitor changes in the condition; S2-4-3-2: Through the coordination layer of the application decision generation module, the real-time admission and discharge record analysis results, real-time causal reasoning results, and real-time user-level application decisions are input into several parallel system-level intelligent agents in the system-level application decision generation layer; S2-4-3-3: Based on the real-time admission and discharge record analysis results, real-time causal reasoning results, and user-level application decisions, based on the second multi-optimization goal set, several system-level agents in the system-level application decision generation layer are used to generate application decisions and obtain several real-time system-level application decisions; Real-time system-level application decisions include: Real-time resource allocation strategies generated by the system-level agent corresponding to the resource allocation system: for example, optimizing bed allocation and equipment scheduling based on the user's condition and nursing home resources to improve resource utilization efficiency; Real-time staffing strategies generated by system-level agents corresponding to the staffing system: for example, optimizing staff scheduling and tool configuration based on user needs and staff expertise to improve the quality of nursing services; Real-time ward management strategies generated by the system-level agent corresponding to the ward management system: for example, optimizing ward cleaning and disinfection based on the user's condition and ward environment to improve user comfort and safety; Real-time nursing quality monitoring strategies generated by the system-level intelligent agent corresponding to the user monitoring system: for example, based on user risk prediction results and anomaly detection information, strengthen the monitoring and management of high-risk users to improve nursing quality; S5-5: Integrate user-level application decisions and several system-level application decisions to obtain the user's real-time application decision; S2-4-4: Based on the real-time admission and discharge record analysis results, real-time causal reasoning results, and real-time application decisions, use the analysis report generation module of the admission and discharge record analysis model to generate the corresponding real-time analysis report; S2-5: Upload the user's real-time admission and discharge records and real-time analysis reports to the cloud data center, and use the cloud data center's blockchain network to distribute and store the real-time admission and discharge records and real-time analysis reports; S3: Use the user terminal to connect to the cloud data center to download and visualize the user's real-time admission and discharge records and real-time analysis reports, including the following steps: S3-1: Use the user terminal to connect to the cloud data center and perform user login and security authentication. After passing, enter the download interface and transmit real-time user information to the cloud data center; S3-2: In the cloud data center, based on the real-time user information, the corresponding real-time admission and discharge records and real-time analysis reports are matched and extracted in the blockchain network. The real-time admission and discharge records and real-time analysis reports are encrypted using the private key in the asymmetric key to obtain the encrypted real-time admission and discharge records and encrypted real-time analysis reports. S3-3: The encrypted real-time admission and discharge records and the encrypted real-time analysis report are transmitted back to the user's user terminal. At the user terminal, the encrypted real-time admission and discharge records and the encrypted real-time analysis report are decrypted according to the public key in the asymmetric key to obtain the decrypted real-time admission and discharge records and the decrypted real-time analysis report. S3-4: Use the user terminal to visualize the decrypted real-time admission and discharge records and the decrypted real-time analysis report.

[0021] Example 2: like Figure 2 As shown, this embodiment provides an admission and discharge record analysis system based on big data analysis, which is used to implement an admission and discharge record analysis method. The system includes a cloud data center, a plurality of data servers, and a plurality of user terminals. The cloud data center is respectively connected to the plurality of data servers in communication with each other. The plurality of data servers are provided in the corresponding terminals, and the data servers are respectively connected to the plurality of data systems in the terminals. Each user terminal is connected to the data server within the communication range. The cloud data center uses artificial intelligence algorithms to build an admission and discharge record analysis engine based on historical admission and discharge record big data and deploys it to the data server of each terminal; The data server is used to collect users' real-time admission and discharge records, and use the admission and discharge record analysis engine to analyze the real-time admission and discharge records, obtain real-time analysis reports, and store them in the cloud data center; User terminals are used to connect to the cloud data center to download and visualize the user's corresponding real-time admission and discharge records and real-time analysis reports.

[0022] The present invention provides a method and system for analyzing hospital admission and discharge records based on big data analysis. The hospital admission and discharge record analysis engine, by combining a variety of artificial intelligence technologies, can not only extract static features such as basic information, diagnosis, nursing, and expenses of users, but also capture dynamic features such as hospitalization duration and changes in condition, and integrate nursing knowledge to form a rich and accurate multi-dimensional feature representation, overcoming the shortcomings of the existing technology of single and shallow features, making the characterization of users more comprehensive and in-depth; the hospital admission and discharge record analysis engine performs causal reasoning, which can go beyond simple correlation analysis and explore the causal relationship between diagnosis and treatment measures and user outcomes, greatly improving the interpretability of the analysis and the clinical guidance value, and providing a theoretical basis for optimizing nursing pathways; based on the extracted multi-dimensional features and causal reasoning results, specific and personalized applications are generated Decision-making recommendations (such as nursing adjustments and risk warnings) make up for the shortcomings of existing technologies that can only provide generalized information but cannot support individualized and precise decision-making, and directly empower nursing staff and managers; the analysis engine is deployed on the nursing home's data server, which can quickly collect real-time admission and discharge records and analyze them. At the same time, data sharing and result distribution are realized through the cloud platform, and user terminals can obtain analysis reports and visualization results in real time. This overcomes the problems of delayed updates and slow responses in traditional systems, and meets the needs of modern nursing for real-time monitoring and rapid decision-making; through the construction of cloud data centers, cross-terminal data aggregation and unified deployment of analysis engines are realized, which makes it possible to break data silos and conduct larger-scale collaborative data analysis, thereby improving the breadth and depth of analysis and generating insights that are difficult to obtain from a single terminal.

[0023] The present invention is not limited to the above optional embodiments. Anyone can derive various other forms of products based on the teachings of the present invention. The above specific embodiments should not be construed as limiting the scope of protection of the present invention. The scope of protection of the present invention shall be based on the scope defined in the claims, and the description can be used to interpret the claims.

Claims

1. A method for analyzing admission and discharge records based on big data analysis, characterized by: The steps include: In the cloud data center, based on the big data of historical admission and discharge records, an admission and discharge record analysis engine is built using artificial intelligence algorithms and deployed to the data server of each terminal; Use the data server to collect users' real-time admission and discharge records, and use the admission and discharge record analysis engine to analyze the real-time admission and discharge records, obtain real-time analysis reports, and store them in the cloud data center; Use the user terminal to connect to the cloud data center to download and visualize the user's corresponding real-time admission and discharge records and real-time analysis reports.

2. The method for analyzing admission and discharge records based on big data analysis according to claim 1, characterized in that: The admission and discharge record analysis engine includes a dynamic multidimensional feature generation model, a nursing knowledge graph, a path mining and causal reasoning model, and an admission and discharge record analysis model that are connected in sequence.

3. The method for analyzing admission and discharge records based on big data analysis according to claim 2, characterized in that: The dynamic multidimensional feature generation model is constructed based on the Autoencoder-BERT-LSTM-GRU-AttentionKGEmb algorithm, and the dynamic multidimensional feature generation model is provided with a static feature extraction module constructed based on the Autoencoder algorithm, a dynamic feature generation module constructed based on the BERT-LSTM-GRU algorithm, and a knowledge graph embedding module constructed based on the AttentionKGEmb algorithm; The path mining and causal reasoning model is constructed based on the VAE-MTCD algorithm, and the path mining and causal reasoning model is provided with a path mining module constructed based on the VAE algorithm and a causal reasoning module constructed based on the MTCD algorithm; The admission and discharge record analysis model is constructed based on the MLP-MAPGRPO-cGAN algorithm, and the admission and discharge record analysis model is provided with an admission and discharge record analysis module constructed based on the MLP algorithm, an application decision generation module constructed based on the MAPGRPO algorithm, and an analysis report generation module constructed based on the cGAN algorithm.

4. The method for analyzing admission and discharge records based on big data analysis according to claim 3, characterized in that: In the cloud data center, based on the big data of historical admission and discharge records, an admission and discharge record analysis engine is built using artificial intelligence algorithms and deployed to the data server of each terminal. This includes the following steps: In the cloud data center, historical admission and discharge record big data and nursing knowledge big data are collected and pre-processed to obtain a number of pre-processed historical admission and discharge record data and a number of pre-processed nursing knowledge data; Based on a number of pre-processed nursing knowledge data, a natural language processing algorithm is used to construct a nursing knowledge graph including a number of knowledge named entities and a number of knowledge entity relationships; Based on several pre-processed historical admission and discharge record data and nursing knowledge graphs, a deep learning algorithm is used to build a dynamic multidimensional feature generation model and obtain several historical dynamic multidimensional features; Based on several historical dynamic multidimensional features, generative machine learning and causal reasoning algorithms are used to build path mining and causal reasoning models, and obtain several historical causal reasoning results; Based on several historical causal reasoning results, a model for analyzing hospital admission and discharge records was constructed using generative machine learning and reinforcement learning algorithms. Integrate the dynamic multidimensional feature generation model, nursing knowledge graph, path mining and causal reasoning model, and admission and discharge record analysis model to obtain the initial admission and discharge record analysis engine; Using the federated learning mechanism, the initial admission and discharge record analysis engine is adjusted to obtain the final admission and discharge record analysis engine, and deployed to the data server of each terminal.

5. The method for analyzing admission and discharge records based on big data analysis according to claim 4, characterized in that: Using a federated learning mechanism, the initial admission and discharge record analysis engine was adjusted to obtain the final admission and discharge record analysis engine, which was then deployed to the data server of each terminal. This process includes the following steps: The initial hospital admission and discharge record analysis engine is enhanced with differential privacy to obtain an optimized hospital admission and discharge record analysis engine; Deploy a coordination server in the cloud data center, connect the coordination server to several terminal data servers, and deploy an optimized admission and discharge record analysis engine to each data server; On each data server, based on a number of collected training admission and discharge records, the corresponding optimized admission and discharge record analysis engine is trained to obtain a trained admission and discharge record analysis engine; Perform model compression / distillation on the trained admission and discharge record analysis engine to obtain a lightweight admission and discharge record analysis engine, extract the corresponding training model parameters, and send them to the coordination server; In the coordination server, all training model parameters are integrated to obtain comprehensive model parameters, and the comprehensive model parameters are sent to all data servers; On each data server, the trained admission and discharge record analysis engine is adjusted according to the comprehensive model parameters to obtain the final admission and discharge record analysis engine.

6. The method for analyzing admission and discharge records based on big data analysis according to claim 5, characterized in that: Use the data server to collect users' real-time admission and discharge records, and use the admission and discharge record analysis engine to analyze the real-time admission and discharge records, obtain real-time analysis reports, and store them in the cloud data center. The steps include the following: Use the data server to collect the user's real-time admission and discharge records, and pre-process them to obtain the pre-processed real-time admission and discharge records; Use the dynamic multi-dimensional feature generation model and nursing knowledge graph of the admission and discharge record analysis engine to extract the real-time dynamic features of the pre-processed real-time admission and discharge records; Based on real-time dynamic multi-dimensional features, the path mining and causal reasoning model of the admission and discharge record analysis engine is used to generate corresponding real-time causal reasoning results; Based on the real-time dynamic multi-dimensional features and real-time causal reasoning results, the hospital discharge record analysis model of the hospital discharge record analysis engine is used to generate corresponding real-time analysis reports; The user's real-time admission and discharge records and real-time analysis reports are uploaded to the cloud data center, and the blockchain network of the cloud data center is used to distribute the real-time admission and discharge records and real-time analysis reports.

7. The method for analyzing admission and discharge records based on big data analysis according to claim 6, characterized in that: Using the dynamic multidimensional feature generation model and nursing knowledge graph of the admission and discharge record analysis engine, we extract the real-time dynamic features of the pre-processed real-time admission and discharge records, including the following steps: Input the pre-processed real-time admission and discharge records into the dynamic multi-dimensional feature generation model of the admission and discharge record analysis engine; Use the static feature extraction module of the dynamic multi-dimensional feature generation model to extract the real-time static features of the pre-processed real-time admission and discharge records; Based on the real-time static features, the dynamic feature generation module of the dynamic multi-dimensional feature generation model is used to extract the real-time dynamic features of the pre-processed real-time admission and discharge records; The knowledge graph embedding module of the dynamic multidimensional feature generation model is used to embed the nursing knowledge graph into the real-time static features to obtain real-time semantic association features, and then combined with the corresponding real-time dynamic features to obtain real-time dynamic multidimensional features.

8. The method for analyzing admission and discharge records based on big data analysis according to claim 7, characterized in that: Based on the real-time dynamic multi-dimensional features, the path mining and causal reasoning model of the admission and discharge record analysis engine is used to generate corresponding real-time causal reasoning results, including the following steps: Input real-time dynamic multi-dimensional features into the path mining and causal reasoning model of the hospital admission and discharge record analysis engine; Use the path mining module of the path mining and causal reasoning model to extract real-time nursing path information from real-time dynamic multidimensional features; According to the real-time dynamic multidimensional features and real-time nursing path information, the causal reasoning module of the path mining and causal reasoning model is used to generate the corresponding real-time causal reasoning results.

9. The method for analyzing admission and discharge records based on big data analysis according to claim 8, characterized in that: Based on the real-time dynamic multidimensional features and real-time causal reasoning results, the hospital discharge record analysis model of the hospital discharge record analysis engine is used to generate the corresponding real-time analysis report, including the following steps: Inputting real-time dynamic multi-dimensional features and real-time causal reasoning results into the admission and discharge record analysis model of the admission and discharge record analysis engine; Based on the real-time dynamic multi-dimensional features, the admission and discharge record analysis module of the admission and discharge record analysis model is used to generate corresponding real-time admission and discharge record analysis results; Based on the real-time admission and discharge record analysis results and real-time causal reasoning results, the application decision generation module of the admission and discharge record analysis model is used to generate real-time application decisions for users; According to the real-time admission and discharge record analysis results, real-time causal reasoning results and real-time application decisions, the analysis report generation module of the admission and discharge record analysis model is used to generate the corresponding real-time analysis report.

10. A hospital admission and discharge record analysis system based on big data analysis, used to implement the hospital admission and discharge record analysis method according to any one of claims 1 to 9, characterized in that: The system includes a cloud data center, several data servers and several user terminals. The cloud data center is communicated with the several data servers respectively. Several of the data servers are set in corresponding terminals, and the data servers are communicated with several data systems in the terminals respectively. Each of the user terminals is communicated with the data servers within the communication range.

Images