A hospital intelligent financial management system based on multi-dimensional data fusion

Abstract

The present application provides a hospital intelligent financial management system based on multidimensional data fusion, which relates to the field of financial management technology. The multidimensional data obtained from multiple heterogeneous data sources are integrated to obtain a financial-related data warehouse; feature extraction is performed on the financial-related data warehouse to obtain a financial-related feature matrix, and the dimension link graph of each dimension in the financial-related feature matrix is ​​determined, and a process link graph related to hospital finance is constructed through the dimension link graphs of each dimension; the financial feature vectors of each dimension are extracted from the financial-related feature matrix, and the financial dimension entropy of each dimension is determined according to the corresponding financial feature vector, and the financial core feature sequence is screened out based on all the financial dimension entropies and the process link graph; and abnormal warning of hospital finance is performed based on the financial core feature sequence. The present application can mine the link relationship of multi-source heterogeneous data, and extract the core features therein for abnormality detection, so as to improve the early warning capability of the financial management system.

Description

Technical Field

[0001] The present application relates to the technical field of financial management, and more specifically, to an intelligent hospital financial management system based on multi-dimensional data fusion. Background Art

[0002] In the financial management process of modern hospitals, there are many types of data involving multiple dimensions, such as financial data, patient charging information, medical insurance settlement, drug procurement, equipment purchase, department budget, doctor salary, operating costs, etc. Traditional financial management methods usually rely on single-dimensional data analysis, which makes it difficult to fully grasp the hospital's capital flow, operational efficiency and cost control. The hospital intelligent financial management system based on multi-dimensional data fusion aims to build an efficient and accurate financial decision-making system by integrating and analyzing multiple heterogeneous data sources, so as to improve the hospital's capital management capabilities and realize intelligent and transparent financial operations.

[0003] However, although the hospital financial management model of multidimensional data fusion has broad application prospects, in the existing technology, the financial data of the hospital involves multiple business systems, and the data format and storage method are different. For example, patient charge data may be stored in the HIS system, while procurement data comes from the ERP system, and medical insurance settlement data is provided by the Medical Insurance Bureau. These data sources are highly heterogeneous. And in the process of multidimensional data fusion, it is necessary to establish the relationship between various financial dimensions. For example, how does the budget execution of the department affect the overall financial health of the hospital? Is the expenditure on drug procurement reasonable? How does the collection cycle of medical insurance settlement affect cash flow? These problems need to be based on complex data modeling technology. Therefore, how to mine the link relationship of multi-source heterogeneous data and extract the core features for anomaly detection to improve the early warning ability of the financial management system is a difficult problem faced by the industry. Summary of the invention

[0004] The present application provides a hospital intelligent financial management system based on multi-dimensional data fusion, which can mine the link relationship of multi-source heterogeneous data and extract the core features therein for anomaly detection, so as to improve the early warning capability of the financial management system.

[0005] The present application provides a hospital intelligent financial management system based on multi-dimensional data fusion, the management system comprising:

[0006] A data acquisition module is used to obtain multidimensional data related to hospital finance from multiple heterogeneous data sources, and integrate the multidimensional data to obtain a financial related data warehouse;

[0007] The spectrum construction module is used to extract features from the financial-related data warehouse to obtain a financial-related feature matrix, determine the dimensional link graph of each dimension in the financial-related feature matrix, and construct a process link spectrum related to hospital finance through the dimensional link graphs of each dimension;

[0008] The core screening module is used to extract the financial feature vectors of each dimension from the financial-related feature matrix, determine the financial dimension entropy of each dimension according to the corresponding financial feature vectors, and screen out the financial core feature sequence based on all the financial dimension entropy and the process link spectrum;

[0009] The anomaly warning module is used to perform anomaly warning on hospital finance based on the financial core feature sequence.

[0010] In this embodiment, the multi-dimensional data related to hospital finance includes financial data, patient data, medical service data, employee salary data, and drug procurement data.

[0011] In this embodiment, the data integration of the multi-dimensional data to obtain a financial-related data warehouse specifically includes:

[0012] Perform preprocessing on the multi-dimensional data to obtain preprocessed multi-dimensional data;

[0013] Perform semantic mapping on the preprocessed multi-dimensional data to obtain financial-related data for each dimension;

[0014] Construct a financial-related data warehouse through the financial-related data of all dimensions.

[0015] In this embodiment, the feature extraction of the financial-related data warehouse to obtain a financial-related feature matrix specifically includes:

[0016] Extract features from the financial-related data of each dimension in the financial-related data warehouse respectively to obtain financial feature vectors of each dimension;

[0017] Construct a financial-related feature matrix based on the financial feature vectors of all dimensions.

[0018] In this embodiment, determining the dimensional link graph of each dimension in the financial-related feature matrix specifically includes:

[0019] Select a dimension from all dimensions of the financial-related feature matrix, and obtain the financial feature vector of the selected dimension;

[0020] Determine the feature link degree between each financial-related feature in the financial feature vector of the selected dimension;

[0021] Construct a dimensional link graph for the selected dimension based on the feature link degrees among various finance-related features, and then obtain the dimensional link graph for each dimension in the finance-related feature matrix.

[0022] In this embodiment, constructing a process link graph related to hospital finance through the dimensional link graphs of each dimension specifically includes:

[0023] Determine the key finance-related features of each dimension, and then determine the feature link degrees among the key finance-related features;

[0024] Connect the dimensional link graphs of each dimension according to the feature link degrees among the key finance-related features, and then obtain a process link graph related to hospital finance.

[0025] In this embodiment, determining the financial dimensional entropy of each dimension according to the corresponding financial feature vector specifically includes:

[0026] Obtain the scale factor of each dimension;

[0027] For the financial feature vectors of each dimension, obtain the weights corresponding to each finance-related feature in the financial feature vector;

[0028] Determine the financial dimensional entropy of the dimension corresponding to the financial feature vector through the weights corresponding to each finance-related feature and the corresponding scale factor, and then obtain the financial dimensional entropy of each dimension.

[0029] In this embodiment, screening out the financial core feature sequence based on all the financial dimensional entropies and the process link graph specifically includes:

[0030] For each finance-related feature in the finance-related feature matrix, determine the financial dimensional entropy of the dimension where the finance-related feature is located;

[0031] Extract all the feature link degrees corresponding to the finance-related feature in the process link graph;

[0032] Determine the financial core entropy of the finance-related feature through the financial dimensional entropy of the dimension where the finance-related feature is located and all the corresponding feature link degrees, and then obtain the financial core entropy of each finance-related feature in the finance-related feature matrix;

[0033] Select all the financial core features according to the financial core entropy of each finance-related feature;

[0034] Construct a financial core feature sequence according to all the financial core features.

[0035] In this embodiment, the abnormal warning of hospital finance based on the financial core feature sequence is to input the financial core feature sequence into an anomaly detection model for detection, and then obtain an abnormal warning report of hospital finance.

[0036] In this embodiment, the anomaly detection model is a long short-term memory network model.

[0037] The technical solutions provided by the disclosed embodiments of this application have the following beneficial effects:

[0038] The multi-dimensional data related to hospital finance is obtained from multiple heterogeneous data sources through a data collection module, and the multi-dimensional data is integrated to obtain a data warehouse related to finance; the feature extraction module extracts features from the data warehouse related to finance to obtain a feature matrix related to finance, determines the dimension link graph of each dimension in the feature matrix related to finance, and constructs a process link graph related to hospital finance through the dimension link graphs of each dimension; the core screening module extracts the financial feature vectors of each dimension from the feature matrix related to finance, determines the financial dimension entropy of each dimension according to the corresponding financial feature vectors, and screens out the financial core feature sequence based on all the financial dimension entropies and the process link graph; the abnormal warning module performs abnormal warning of hospital finance based on the financial core feature sequence.

[0039] It can be seen that in this application, first, feature extraction is performed on the data warehouse related to finance to obtain a feature matrix related to finance, and a process link graph related to hospital finance is constructed through the dimension link graph of each dimension, which can help comprehensively and systematically identify and understand the key links and their interrelationships in hospital financial management; then, by extracting the financial feature vectors of each dimension from the feature matrix related to finance and determining the financial dimension entropy of each dimension according to the corresponding financial feature vectors, the useful information amount of each dimension can be effectively quantified, and the financial core feature sequence is screened out based on the financial dimension entropy and the process link graph, which can help focus on the features that have the greatest impact on the financial situation, exclude redundant information, help accurately mine the link relationships in multi-source heterogeneous data, and extract the financial core features; finally, the abnormal warning of hospital finance is carried out based on the financial core feature sequence, which can effectively improve the intelligence and warning ability of the financial management system. By analyzing the financial core feature sequence, the system can real-time identify abnormal fluctuations and potential risks in financial indicators, thereby improving the warning ability of the financial management system.

[0040] In summary, the technical solution adopted in this application can mine the link relationships of multi-source heterogeneous data, extract the core features therein for anomaly detection, and improve the warning ability of the financial management system. Description of the Drawings

[0041] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0042] Figure 1 is a module structure diagram of a hospital intelligent financial management system based on multi-dimensional data fusion provided by the present application;

[0043] Figure 2 is an exemplary flowchart for determining the dimension link diagram of each dimension in the financial-related feature matrix provided by the present application;

[0044] Figure 3 is an exemplary flowchart for determining the financial core feature sequence provided by the present application. Detailed implementation manners

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0046] The embodiments of the present application provide a hospital intelligent financial management system based on multi-dimensional data fusion. Its core is to obtain multi-dimensional data related to hospital finance from multiple heterogeneous data sources through a data collection module, integrate the multi-dimensional data to obtain a financial-related data warehouse; extract features from the financial-related data warehouse through a graph construction module to obtain a financial-related feature matrix, determine the dimension link diagram of each dimension in the financial-related feature matrix, and construct a process link graph related to hospital finance through the dimension link diagrams of each dimension; extract the financial feature vectors of each dimension in the financial-related feature matrix through a core screening module, determine the financial dimension entropy of each dimension according to the corresponding financial feature vectors, and screen out the financial core feature sequence based on all the financial dimension entropy and the process link graph; the anomaly warning module performs anomaly warning on hospital finance based on the financial core feature sequence. By adopting the above solution, the link relationship of multi-source heterogeneous data can be mined, and the core features can be extracted for anomaly detection to improve the warning ability of the financial management system.

[0047] To better understand the above technical solutions, the following will detail the above technical solutions in combination with the specification drawings and specific implementation manners. Refer to Figure 1As shown in the figure, this figure is a module structure diagram of a hospital intelligent financial management system based on multi-dimensional data fusion according to this embodiment of the present application. The management system includes: a data acquisition module 100, a graph construction module 200, a core screening module 300, and an anomaly warning module 400, which are described as follows:

[0048] The data acquisition module 100 is used to obtain multi-dimensional data related to hospital finance from multiple heterogeneous data sources, and perform data integration on the multi-dimensional data to obtain a data warehouse related to finance.

[0049] It should be noted that in this application, the multi-dimensional data related to hospital finance includes financial data, patient data, medical service data, employee salary data, and drug procurement data; specifically, in implementation, multi-dimensional data related to hospital finance can be obtained from multiple heterogeneous data sources. In this application, heterogeneous data sources include the hospital's HIS (Hospital Information System), ERP (Enterprise Resource Planning), LIS (Laboratory Information System), medical insurance settlement system, supply chain management system, and labor cost system, etc.

[0050] In this embodiment, the data integration of the multi-dimensional data to obtain a data warehouse related to finance can be specifically implemented in the following manner, that is:

[0051] Perform preprocessing on the multi-dimensional data to obtain preprocessed multi-dimensional data;

[0052] Perform semantic mapping on the preprocessed multi-dimensional data to obtain financial-related data for each dimension;

[0053] Construct a data warehouse related to finance through the financial-related data of all dimensions.

[0054] In specific implementation, first, multi-dimensional data can be preprocessed, that is, data cleaning and data format conversion are performed on the multi-dimensional data, so that the preprocessed multi-dimensional data can be obtained; then, semantic mapping can be performed on the preprocessed multi-dimensional data. Through semantic mapping, semantic relationships between data in each heterogeneous data source can be established, enabling data in different systems to be associated and aligned. For example, financial data → medical service data: By fields such as patient ID, department ID, and diagnosis and treatment items, financial data and medical service data are associated (such as "CT examination fee" matching "radiology department"). In the above way, financial-related data in each dimension can be obtained. The dimensions in this application include time dimension (financial analysis is performed by day, week, month, quarter, and year), department dimension (income, expenditure, and cost structure of different departments), medical insurance dimension (medical insurance payment ratio vs. out-of-pocket income), supply chain dimension (supplier procurement vs. drug use vs. expense expenditure), and labor cost dimension (doctor performance vs. labor cost vs. income contribution), etc.; finally, a financial-related data warehouse can be constructed through financial-related data in all dimensions, that is, financial-related data in all dimensions are stored in a database, so that a financial-related data warehouse can be obtained.

[0055] The graph construction module 200 is used to extract features from the financial-related data warehouse to obtain a financial-related feature matrix, determine the dimension link graph of each dimension in the financial-related feature matrix, and construct a process link graph related to hospital finance through the dimension link graphs of each dimension.

[0056] In this embodiment, the specific method for extracting features from the financial-related data warehouse to obtain a financial-related feature matrix can be as follows:

[0057] Extract features from the financial-related data in each dimension of the financial-related data warehouse respectively, and then obtain the financial feature vectors of each dimension;

[0058] Construct a financial-related feature matrix based on the financial feature vectors of all dimensions.

[0059] In specific implementation, first, financial-related features can be extracted from the financial-related data in each dimension of the financial-related data warehouse respectively, that is, through time series analysis and statistical methods to extract the financial-related data in each dimension of the financial-related data warehouse, so that multiple financial-related features can be obtained for each dimension. Thus, a vector composed of all the financial-related features corresponding to the dimension can be used as the financial feature vector. For example, the financial-related features that the feature vector of the department dimension can include are revenue ratio, labor cost ratio, bed occupancy rate, and equipment maintenance cost. Through the above method, the financial feature vectors of each dimension can be obtained; then, a financial-related feature matrix can be constructed based on the financial feature vectors of all dimensions, that is, using the dimension as the row of the matrix and filling the corresponding financial feature vector into the matrix, and the finally obtained matrix is used as the financial-related feature matrix.

[0060] Preferably, in this embodiment, with reference to Figure 2 as shown, this figure is an exemplary flowchart of determining the dimension link diagram of each dimension in the financial-related feature matrix in the embodiment of the present application. In this embodiment, the dimension link diagram of each dimension in the financial-related feature matrix can be specifically implemented by the following steps:

[0061] In step S21, select a dimension from all dimensions of the financial-related feature matrix and obtain the financial feature vector of the selected dimension;

[0062] In step S22, determine the feature link degree between each financial-related feature in the financial feature vector of the selected dimension;

[0063] In step S23, construct the dimension link diagram of the selected dimension according to the feature link degree between each financial-related feature, and then obtain the dimension link diagram of each dimension in the financial-related feature matrix.

[0064] In specific implementation, first, a dimension can be selected from all dimensions of the financial-related feature matrix and the corresponding financial feature vector of the selected dimension can be obtained; then, the feature link degree between each financial-related feature in the financial feature vector of the selected dimension can be determined. Among them, the feature link degree represents the degree of correlation between financial-related features. The Pearson correlation coefficient between financial-related features can be calculated and the calculation result can be used as the feature link degree between financial-related features. Through the above method, the feature link degree between each financial-related feature in the financial feature vector can be obtained; finally, the dimension link diagram of the selected dimension can be constructed according to the feature link degree between each financial-related feature, that is, connecting each financial-related feature in the financial feature vector of the selected dimension and using the corresponding feature link degree as the value of the edge, so that the dimension link diagram of the selected dimension can be obtained. Through the above method, the dimension link diagram of each dimension in the financial-related feature matrix can be obtained.

[0065] In this embodiment, to construct a process link map related to hospital finance through the dimension link maps of each dimension, the following method can be specifically adopted, that is:

[0066] Determine the key finance-related features of each dimension, and then determine the feature link degree between each key finance-related feature;

[0067] Connect the dimension link maps of each dimension according to the feature link degree between each key finance-related feature, and then obtain a process link map related to hospital finance.

[0068] In specific implementation, first, for the financial feature vectors of each dimension, calculate the total sum of the feature link degrees corresponding to each finance-related feature in the financial feature vector, and take the finance-related feature with the largest total sum of the feature link degrees as the key finance-related feature of the corresponding dimension; then, the feature link degree between each key finance-related feature can be calculated through the above-mentioned calculation method of the feature link degree; finally, the dimension link maps of each dimension can be connected according to the feature link degree between each key finance-related feature, that is, connect each key finance-related feature, and take the corresponding feature link degree as the value of the edge, so as to obtain a process link map related to hospital finance.

[0069] It should be noted that extracting feature matrices related to finance from the finance-related data warehouse and constructing a process link map related to hospital finance through the dimension link map of each dimension can help to comprehensively and systematically identify and understand the key links and their interrelationships in hospital financial management. By revealing the link relationships in multi-source heterogeneous data, the core associations between each finance-related feature can be extracted, and then dynamic monitoring can be carried out through the process link map.

[0070] The core screening module 300 is used to extract the financial feature vectors of each dimension from the finance-related feature matrix, determine the financial dimension entropy of each dimension according to the corresponding financial feature vector, and screen out the financial core feature sequence based on all the financial dimension entropies and the process link map.

[0071] In specific implementation, the financial feature vectors of each dimension can be extracted from the finance-related feature matrix by means of traversal.

[0072] In this embodiment, the following method can be specifically adopted to determine the financial dimension entropy of each dimension according to the corresponding financial feature vector, that is:

[0073] Obtain the scale factor of each dimension;

[0074] For the financial feature vectors of each dimension, obtain the weights corresponding to each finance-related feature in the financial feature vector;

[0075] Determine the financial dimension entropy of the corresponding dimension of the financial feature vector through the weights corresponding to each financial-related feature and the corresponding scale factors, and then obtain the financial dimension entropy of each dimension.

[0076] When specifically implemented, first, the scale factor of each dimension can be obtained. The scale factor is a coefficient for normalizing between different financial-related features, used to ensure that different financial-related features can be compared on the same scale. The scale factor of each dimension can be set through expert knowledge or analysis of historical data. Then, for the financial feature vectors of each dimension, the weights corresponding to each financial-related feature in the financial feature vector can be obtained. The corresponding weights can be preset through historical experience. Finally, the financial dimension entropy of the corresponding dimension of the financial feature vector can be determined through the weights corresponding to each financial-related feature and the corresponding scale factors. Among them, the financial dimension entropy represents the information content related to finance contained in the corresponding dimension. Each financial-related feature can be multiplied by the corresponding weight respectively, and then the results of each multiplication can be multiplied by the scale factor, so that the sum of all the obtained results is used as the financial dimension entropy of the corresponding dimension of the financial feature vector. Through the above method, the financial dimension entropy of each dimension can be obtained.

[0077] Preferably, in this embodiment, refer to Figure 3 As shown, this figure is an exemplary flowchart for determining the financial core feature sequence in the embodiment of the present application. In this embodiment, screening out the financial core feature sequence based on all the financial dimension entropies and the process link map can be specifically implemented by the following steps:

[0078] In step S31, for each financial-related feature in the financial-related feature matrix, determine the financial dimension entropy of the dimension where the financial-related feature is located;

[0079] In step S32, extract all the feature link degrees corresponding to the financial-related feature in the process link map;

[0080] In step S33, determine the financial core entropy of the financial-related feature through the financial dimension entropy of the dimension where the financial-related feature is located and all the corresponding feature link degrees, and then obtain the financial core entropy of each financial-related feature in the financial-related feature matrix;

[0081] In step S34, screen out all the financial core features according to the financial core entropy of each financial-related feature;

[0082] In step S35, construct a financial core feature sequence according to all the financial core features.

[0083] In specific implementation, first, for each financial-related feature in the financial-related feature matrix, the financial dimension entropy of the dimension where the financial-related feature is located can be determined by traversing; then, all feature link degrees corresponding to the financial-related feature can be extracted from the process link graph by traversing, that is, the values of all edges connected to the financial-related feature; furthermore, the financial core entropy of the financial-related feature can be determined by the financial dimension entropy of the dimension where the financial-related feature is located and all corresponding feature link degrees. Among them, the financial core entropy represents the core degree of the financial-related feature in financial management. The product of the sum of all feature link degrees corresponding to the financial-related feature and the financial dimension entropy of the dimension where the financial-related feature is located can be used as the financial core entropy of the financial-related feature. In this way, the financial core entropy of each financial-related feature in the financial-related feature matrix can be obtained.

[0084] In addition, in specific implementation, all financial core features can be screened out according to the financial core entropy of each financial-related feature; first, a threshold can be set based on historical experience and data analysis; then, the financial core entropy of each financial-related feature can be compared with the set threshold. If the financial core entropy of the financial-related feature exceeds the set threshold, it means that the financial-related feature has a greater impact on the financial system and should usually be regarded as a core feature, that is, a financial core feature. All financial core features can be screened out through the comparison process; finally, a financial core feature sequence can be constructed according to all financial core features, that is, the sequence composed of all financial core features is used as the financial core feature sequence. This financial core feature sequence represents the most important feature set for the hospital financial management system, can effectively reflect the hospital's financial status and operation conditions, and the financial core feature sequence is an important basis for subsequent anomaly detection, early warning systems, and decision support systems.

[0085] It should be noted that by extracting the financial feature vectors of each dimension in the financial-related feature matrix and determining the financial dimension entropy of each dimension according to the corresponding financial feature vectors, the useful information amount of each dimension can be effectively quantified. Screening out the financial core feature sequence based on the financial dimension entropy and the process link graph can help focus on the features that have the greatest impact on the financial situation, exclude redundant information, help accurately mine the link relationships in multi-source heterogeneous data, and extract the financial core features.

[0086] The anomaly warning module 400 is used to perform anomaly warning on the hospital's finances according to the financial core feature sequence.

[0087] In this embodiment, performing anomaly warning on the hospital's finances according to the financial core feature sequence is to input the financial core feature sequence into an anomaly detection model for detection, and then obtain an anomaly warning report on the hospital's finances. It should be noted that in this application, the anomaly detection model is a long short-term memory network model.

[0088] In specific implementation, the extracted financial core feature sequence can be input into an anomaly detection model, namely the LSTM model. The LSTM model will automatically learn the long-term dependencies and patterns in the financial core feature sequence, including the trends of financial indicators over time, periodic fluctuations, and abnormal patterns. The memory units of the LSTM can capture these time dependencies and make predictions about future financial situations. By learning historical financial data, the LSTM can identify potential anomalies in future data, thereby generating an anomaly warning report. For example, the anomaly warning report will indicate the abnormal points in the financial data, possible reasons, and potential risks. The report can also provide targeted suggestions to help the hospital's financial management team make timely adjustments or take necessary countermeasures.

[0089] It should be noted that performing anomaly warning for hospital finances based on the financial core feature sequence can effectively improve the intelligence and warning capabilities of the financial management system. By analyzing the financial core feature sequence, the system can real-time identify abnormal fluctuations and potential risks in financial indicators, thereby enhancing the warning capabilities of the financial management system.

[0090] Thus, it can be seen that in this application, first, feature extraction is performed on the financial-related data warehouse to obtain a financial-related feature matrix, and a process link graph is constructed through the dimension link graph of each dimension to help comprehensively and systematically identify and understand the key links and their interrelationships in hospital financial management; then, by extracting the financial feature vectors of each dimension from the financial-related feature matrix and determining the financial dimension entropy of each dimension according to the corresponding financial feature vectors, the useful information amount of each dimension can be effectively quantified. Based on the financial dimension entropy and the process link graph, the financial core feature sequence is screened out, which can help focus on the features that have the greatest impact on the financial situation, exclude redundant information, help accurately mine the link relationships in multi-source heterogeneous data, and extract the financial core features; finally, performing anomaly warning for hospital finances based on the financial core feature sequence can effectively improve the intelligence and warning capabilities of the financial management system. By analyzing the financial core feature sequence, the system can real-time identify abnormal fluctuations and potential risks in financial indicators, thereby enhancing the warning capabilities of the financial management system.

[0091] In summary, the technical solution adopted in this application can mine the link relationships of multi-source heterogeneous data, extract the core features therein for anomaly detection, so as to improve the warning capabilities of the financial management system.

[0092] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can also be implemented. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0093] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM), or other optical disc memories, magnetic disc memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0094] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent in such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the existence of another identical element in the process, method, commodity or device comprising the element.

Claims

1. A hospital intelligent financial management system based on multidimensional data fusion, characterized in that: The management system comprises: A data acquisition module is used to obtain multidimensional data related to hospital finance from multiple heterogeneous data sources, and integrate the multidimensional data to obtain a financial related data warehouse; A graph construction module is used to extract features from the finance-related data warehouse to obtain a finance-related feature matrix, determine a dimension link graph of each dimension in the finance-related feature matrix, and construct a process link graph related to hospital finance through the dimension link graphs of each dimension; A core screening module, used to extract financial feature vectors of each dimension from the financial-related feature matrix, determine the financial dimension entropy of each dimension according to the corresponding financial feature vector, and screen out a financial core feature sequence based on all financial dimension entropies and the process link map; An abnormal warning module is used to provide abnormal warning of hospital finances based on the financial core feature sequence; Wherein, determining the dimension link diagram of each dimension in the financial-related feature matrix specifically includes: Selecting a dimension from all dimensions of the financial-related feature matrix, and obtaining a financial feature vector of the selected dimension; Determining the feature linkage degree between each financial-related feature in the financial feature vector of the selected dimension, wherein the Pearson correlation coefficient between the financial-related features is calculated, and the calculation result is used as the feature linkage degree between the financial-related features; Constructing a dimension link graph of the selected dimension according to the feature link degrees between the various financial-related features, and then obtaining a dimension link graph of each dimension in the financial-related feature matrix, wherein the various financial-related features in the financial feature vector of the selected dimension are connected, and the corresponding feature link degrees are used as edge values ​​to obtain the dimension link graph of the selected dimension; Among them, the process link map related to hospital finance is constructed through the dimension link map of each dimension, including: Determine the key financial-related features of each dimension, and then determine the feature linkage degree between each key financial-related feature, wherein the feature linkage degree represents the degree of correlation and linkage between financial-related features; The dimension link diagrams of each dimension are connected according to the characteristic link degree between each key financial-related characteristic, thereby obtaining the process link diagram related to hospital finance; Among them, determining the financial dimension entropy of each dimension according to the corresponding financial feature vector specifically includes: Obtain the scale factor of each dimension, where the scale factor is a coefficient for normalizing different financial-related characteristics; For the financial feature vectors of each dimension, obtain the weight corresponding to each financial-related feature in the financial feature vector; The financial dimension entropy of the dimension corresponding to the financial feature vector is determined by the weight corresponding to each financial-related feature and the corresponding scale factor, and then the financial dimension entropy of each dimension is obtained, wherein the financial dimension entropy represents the financial-related information content contained in the corresponding dimension, and each financial-related feature is multiplied by the corresponding weight, and each multiplication result is multiplied by the scale factor, and the sum of all the obtained results is taken as the financial dimension entropy of the dimension corresponding to the financial feature vector; The financial core feature sequences screened out based on all financial dimension entropies and the process link graph specifically include: For each financial-related feature in the financial-related feature matrix, determining the financial dimension entropy of the dimension where the financial-related feature is located; Extracting all feature link degrees corresponding to the financial-related features in the process link graph; The financial core entropy of the financial related feature is determined by the financial dimension entropy of the dimension where the financial related feature is located and the corresponding link degree of all features, and then the financial core entropy of each financial related feature in the financial related feature matrix is ​​obtained, wherein the financial core entropy represents the core degree of the financial related feature in financial management, and the product of the sum of all feature link degrees corresponding to the financial related feature and the financial dimension entropy of the dimension where the financial related feature is located is taken as the financial core entropy of the financial related feature; All financial core features are screened out based on the financial core entropy of each financial-related feature; Construct a financial core feature sequence based on all financial core features; Among them, the abnormal warning of hospital finance based on the financial core feature sequence is to input the financial core feature sequence into the anomaly detection model for detection, and then obtain the abnormal warning report of hospital finance, wherein the anomaly detection model is a long short-term memory network model.

2. The hospital intelligent financial management system based on multidimensional data fusion as claimed in claim 1, characterized in that: The multi-dimensional data related to hospital finance includes financial data, patient data, medical service data, employee salary data and drug purchase data.

3. The hospital intelligent financial management system based on multidimensional data fusion as claimed in claim 1, characterized in that: The multi-dimensional data is integrated to obtain a financial related data warehouse, which specifically includes: Preprocessing the multidimensional data to obtain preprocessed multidimensional data; Perform semantic mapping on the pre-processed multi-dimensional data to obtain financial-related data of each dimension; Build a financial-related data warehouse using financial-related data of all dimensions.

4. The hospital intelligent financial management system based on multidimensional data fusion as claimed in claim 1, characterized in that: The feature extraction is performed on the financial related data warehouse to obtain the financial related feature matrix, which specifically includes: Extracting features of the financial-related data of each dimension in the financial-related data warehouse respectively, and then obtaining financial feature vectors of each dimension; Construct a financial-related feature matrix based on the financial feature vectors of all dimensions.

Images