Intelligent indexing system and medical resource scheduling system using the intelligent indexing system

Abstract

Intelligent indexing system and medical resource scheduling system using the intelligent indexing system. The invention belongs to the field of medical information technology, and specifically relates to the field of medical resource scheduling technology. It solves the technical problem that the existing intelligent indexing system cannot perfectly adapt to the high performance requirements of medical resource scheduling. The intelligent indexing system includes a data acquisition module, a big data storage module, a data processing and integration module, and an intelligent indexing construction module. The medical resource scheduling system using the intelligent indexing system includes a data acquisition and preprocessing module, an intelligent analysis and prediction module, a medical resource allocation module, an intelligent monitoring and optimization module, and a cross-machine collaboration and information sharing module.

Description

Technical Field

[0001] The present invention belongs to the field of medical information technology, and specifically relates to the field of medical resource scheduling technology. Background Art

[0002] At present, problems such as uneven distribution of medical resources and inefficient scheduling still exist, especially in emergency medical treatment and daily medical resource management. Traditional medical resource scheduling methods rely on manual judgment and experience-based decision-making, which often makes it difficult to quickly and accurately match medical needs and resources. In recent years, the accumulation and analysis of medical big data has made it possible for intelligent scheduling of medical resources, but the existing medical big data application system still needs to be improved in scheduling accuracy and real-time performance.

[0003] In the process of implementing medical resource scheduling, it is necessary to collect information from multiple data sources, transmit data in the form of real-time data streams, and respond to data access needs. In this process, the cooperation of an intelligent indexing system is required. Due to the complexity of the data types involved in the medical resource scheduling process, the existing intelligent indexing system cannot perfectly adapt to the high performance requirements of medical resource scheduling. Summary of the invention

[0004] In order to solve the above problems, the present invention provides an intelligent indexing system and a medical resource scheduling system using the intelligent indexing system.

[0005] The intelligent indexing system includes a data acquisition module, a big data storage module, a data processing and integration module and an intelligent indexing construction module;

[0006] The data acquisition module is responsible for collecting information from various data sources and transmitting data in the form of real-time data streams;

[0007] The big data storage module is responsible for storing and managing the data transmitted by the data acquisition module and responding to data access requirements;

[0008] The data processing and integration module is responsible for cleaning, converting and integrating data from different sources and inputting the data into the intelligent index building module;

[0009] The intelligent index building module is responsible for automatically generating indexes according to data content and structure using distributed index building technology.

[0010] Furthermore, in the data acquisition module, data transmission in the form of real-time data stream is performed in the following manner:

[0011] S21, real-time capture of data streams in data sources through a distributed data collection framework;

[0012] S22, performing preliminary preprocessing on the captured data stream, including data format conversion, timestamp correction, outlier detection and filtering, and field completion;

[0013] S23, classifying and marking the data stream according to the business logic;

[0014] S24. Use a secure communication protocol to establish an encrypted data transmission channel, and transmit the encrypted data stream to the big data storage module.

[0015] Furthermore, in the big data storage module, a distributed file system is used to disperse and store data. The specific steps of the distributed file system when working are as follows:

[0016] S31. Data node storage: In a distributed file system, data files are divided into multiple data blocks and stored in each data node;

[0017] S32, Task decomposition and distribution: In the MapReduce process, a specific task will be decomposed into several subtasks, and these decomposed subtasks will be distributed to each data node;

[0018] S33, Task Scheduling and Management: The master node is responsible for managing and scheduling the tasks of each sub-node. The master node periodically receives the working status and results returned by the sub-nodes, and performs corresponding integration and processing;

[0019] S34, Hash table index: Hash table index uses hash function to map feature values ​​to specific storage locations, thereby achieving fast search. This indexing method is suitable for scenarios that require frequent search and update;

[0020] S35. Time series index: The time series index is sorted and stored according to the time sequence of the feature values, which facilitates time series analysis and prediction. This indexing method is suitable for scenarios where it is necessary to analyze the trend of feature values ​​changing over time.

[0021] Furthermore, when the data processing and integration module integrates data from different sources, data mapping and standardization techniques are used. Specifically, the data mapping and standardization techniques are:

[0022] S41. Map data from different sources and in different formats to establish a corresponding relationship between data fields; convert data formats to convert all data into a unified format that is recognizable and easy to process;

[0023] S42. After data mapping and format conversion, perform data standardization to ensure that all data follows unified specifications and standards;

[0024] S43. The standardized data are integrated into a centralized database that supports data retrieval and analysis.

[0025] Furthermore, the distributed index construction technology is specifically as follows:

[0026] S51, distributed data sharding and hash mapping: first, the data is sharded using a distributed algorithm, and each data fragment is mapped using a hash function to generate a unique hash value, which will be used as an index key to locate the data fragment;

[0027] S52, metadata extraction and index entry construction: based on data sharding and hash mapping, extract metadata of each data fragment, including the size, storage location, timestamp and key information summary of the data fragment, and construct index entries based on the metadata. Each index entry contains a pointer to the data fragment and metadata information.

[0028] S53, distributed index structure design and optimization: based on the constructed index entries, a distributed index structure is designed to support efficient data retrieval and query. The distributed index structure is implemented using a distributed hash table or a distributed B-tree. At the same time, the index structure is optimized to balance the height of the tree, reduce the number of node accesses, and improve the query efficiency and stability of the index.

[0029] S54, index construction and consistency maintenance: Assign index entries to corresponding index nodes and establish associations between index nodes. At the same time, perform consistency checks and maintenance on the index to ensure the accuracy and completeness of the index. Regularly update and optimize the index to adapt to changes in data and query requirements.

[0030] The present invention provides a medical resource scheduling system using an intelligent indexing system, the system comprising a data acquisition and preprocessing module, an intelligent analysis and prediction module, a medical resource allocation module, an intelligent monitoring and optimization module, and a cross-machine collaboration and information sharing module;

[0031] The data acquisition and preprocessing module is responsible for integrating and optimizing diverse data sources from major medical institutions, public health databases, and personal health monitoring devices;

[0032] The intelligent analysis and prediction module is responsible for simulating the flow and allocation process of medical resources by building a network model, so as to provide a scientific basis for resource scheduling;

[0033] The intelligent monitoring and optimization module is responsible for real-time monitoring of the utilization efficiency of medical resources, patient satisfaction and medical quality indicators, automatically identifying bottleneck links through data analysis, and proposing optimization suggestions;

[0034] The cross-machine collaboration and information sharing module is responsible for realizing cross-institutional medical resource information sharing and collaborative operations.

[0035] Furthermore, the workflow of the data acquisition and preprocessing module is specifically as follows:

[0036] S71. Collect raw data;

[0037] S72. Use a multi-source data fusion engine to automatically identify and adapt data sources of different formats and standards;

[0038] S73. Encrypt sensitive data and use anonymization technology to protect patient privacy;

[0039] S74, storing the data in a distributed database and establishing a connection with an intelligent indexing system;

[0040] S75, intelligent indexing system locates relevant data based on user query requirements;

[0041] S76. Store the data in a designated location and make a backup.

[0042] Furthermore, the workflow of the intelligent analysis and prediction module is specifically as follows:

[0043] S81, introduce dynamic node attributes to build a network model framework;

[0044] S82. Design the priority of urgent requirements for the network model framework;

[0045] S83, setting network simulation parameters;

[0046] S84, associating the network simulation parameters with the historical data and setting them as dynamic adjustment data;

[0047] S85, introduce multi-objective optimization strategy;

[0048] S86, simulate the flow and allocation process of medical resources;

[0049] S87. Visual display of analysis report.

[0050] Furthermore, the intelligent monitoring and optimization module includes an intelligent scheduling platform, a real-time monitoring platform and an abnormal warning platform;

[0051] Intelligent scheduling platform: When a hospital encounters resource shortage or sudden changes in demand, a resource scheduling plan is automatically generated based on preset scheduling rules and priorities. The scheduling plan includes the reallocation of beds, temporary shift changes of medical staff, and emergency deployment of medical equipment. The system automatically adjusts the scheduling plan based on real-time data and network models.

[0052] Real-time monitoring platform: continuously tracks the hospital's operating status, including patient waiting time, bed occupancy rate, and medical equipment usage, and transmits data to the intelligent scheduling platform, which displays monitoring data in real time in the form of charts to help hospital managers intuitively understand the hospital's operating status;

[0053] Abnormal warning platform: If abnormal indicators are monitored, an early warning will be issued immediately to remind hospital managers to take emergency measures.

[0054] Furthermore, when the cross-machine collaboration and information sharing module realizes cross-institutional medical resource information sharing and collaborative operations, it obtains information collection and dissemination permissions through authorization from relevant institutions.

[0055] The beneficial effects of the intelligent indexing system described in the present invention are: data is stored in nodes through a distributed file system, and the accuracy and efficiency of the index are improved by indexing by situation, data mapping and standardization technology are used to integrate data from different sources to improve integration efficiency, distributed index construction technology is used to slice data, and index entries are constructed on this basis, so that each index entry contains a pointer to the data fragment and metadata information, thereby improving the accuracy of the index.

[0056] The beneficial effects of the medical resource scheduling system using the intelligent indexing system described in the present invention are: not only effectively improving the utilization rate and scheduling efficiency of medical resources, but also improving user experience and management effects through intelligent design, providing medical institutions with a more comprehensive and personalized resource scheduling solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic diagram of an intelligent indexing system in an embodiment of the present invention;

[0058] Figure 2 Schematic diagram of a medical resource scheduling system using an intelligent indexing system in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0060] Embodiment 1,

[0061] This embodiment provides an intelligent indexing system such as Figure 1 As shown, it includes a data acquisition module, a big data storage module, a data processing and integration module, and an intelligent index building module;

[0062] The data acquisition module is responsible for collecting information from various data sources and transmitting data in the form of real-time data streams;

[0063] The big data storage module is responsible for storing and managing the data transmitted by the data acquisition module and responding to data access requirements;

[0064] The data processing and integration module is responsible for cleaning, converting and integrating data from different sources and inputting the data into the intelligent index building module;

[0065] The intelligent index building module is responsible for automatically generating indexes according to data content and structure using distributed index building technology.

[0066] In the data acquisition module, data transmission in the form of real-time data stream is carried out in the following ways:

[0067] S21, real-time capture of data streams in data sources through a distributed data collection framework;

[0068] S22, performing preliminary preprocessing on the captured data stream, including data format conversion, timestamp correction, outlier detection and filtering, and field completion;

[0069] S23, classifying and marking the data stream according to the business logic;

[0070] S24. Use a secure communication protocol to establish an encrypted data transmission channel, and transmit the encrypted data stream to the big data storage module.

[0071] In the big data storage module, a distributed file system is used to disperse and store data. The specific steps of the distributed file system when working are as follows:

[0072] S31. Data node storage: In a distributed file system, data files are divided into multiple data blocks and stored in each data node;

[0073] S32, Task decomposition and distribution: In the MapReduce process, a specific task will be decomposed into several subtasks, and these decomposed subtasks will be distributed to each data node;

[0074] S33, Task Scheduling and Management: The master node is responsible for managing and scheduling the tasks of each sub-node. The master node periodically receives the working status and results returned by the sub-nodes, and performs corresponding integration and processing;

[0075] S34, Hash table index: Hash table index uses hash function to map feature values ​​to specific storage locations, thereby achieving fast search. This indexing method is suitable for scenarios that require frequent search and update;

[0076] S35. Time series index: The time series index is sorted and stored according to the time sequence of the feature values, which facilitates time series analysis and prediction. This indexing method is suitable for scenarios where it is necessary to analyze the trend of feature values ​​changing over time.

[0077] When the data processing and integration module integrates data from different sources, it uses data mapping and standardization technology, which is specifically:

[0078] S41. Map data from different sources and in different formats to establish a corresponding relationship between data fields; convert data formats to convert all data into a unified format that is recognizable and easy to process;

[0079] S42. After data mapping and format conversion, perform data standardization to ensure that all data follows unified specifications and standards;

[0080] S43. The standardized data are integrated into a centralized database that supports data retrieval and analysis.

[0081] The distributed index construction technology is specifically as follows:

[0082] S51, distributed data sharding and hash mapping: first, the data is sharded using a distributed algorithm, and each data fragment is mapped using a hash function to generate a unique hash value, which will be used as an index key to locate the data fragment;

[0083] S52, metadata extraction and index entry construction: based on data sharding and hash mapping, extract metadata of each data fragment, including the size, storage location, timestamp and key information summary of the data fragment, and construct index entries based on the metadata. Each index entry contains a pointer to the data fragment and metadata information.

[0084] S53, distributed index structure design and optimization: based on the constructed index entries, a distributed index structure is designed to support efficient data retrieval and query. The distributed index structure is implemented using a distributed hash table or a distributed B-tree. At the same time, the index structure is optimized to balance the height of the tree, reduce the number of node accesses, and improve the query efficiency and stability of the index.

[0085] S54, index construction and consistency maintenance: Assign index entries to corresponding index nodes and establish associations between index nodes. At the same time, perform consistency checks and maintenance on the index to ensure the accuracy and completeness of the index. Regularly update and optimize the index to adapt to changes in data and query requirements.

[0086] Embodiment 2,

[0087] This embodiment is a further limitation of the first embodiment. When the intelligent indexing system is applied to a medical resource scheduling system, the data it processes is medical data, and its specific data processing method is described.

[0088] The function of the data collection module is to efficiently collect information from multiple data sources. The technical means to achieve this function is to use API interface and real-time data stream processing technology to ensure the comprehensiveness and timeliness of the data.

[0089] Real-time data stream processing technology: Data is transmitted in the form of streams, and timeliness is a major problem. If there is no timeliness, the delay will be too serious. The method includes:

[0090] Step 1: Use a distributed data collection framework to capture medical big data streams from multiple sources such as hospital information systems (HIS), electronic medical record systems (EMR), and medical device interfaces in real time;

[0091] Step 2: Perform preliminary preprocessing on the captured data stream, including data format conversion, timestamp correction, outlier detection and filtering, and necessary field completion to ensure data accuracy and consistency;

[0092] Step 3: Classify and mark the data stream according to business logic, such as basic patient information, medical records, examination results, etc., to facilitate subsequent analysis;

[0093] Step 4: Use a secure communication protocol (such as TLS / SSL) to establish an encrypted data transmission channel, and securely transmit the encrypted data stream to the data center of the medical resource intelligent scheduling system;

[0094] Step 5: Decrypt the received encrypted data stream in the data center to restore the original format of the data and provide the original data for subsequent analysis;

[0095] Step 6: Design and implement a feature extraction algorithm to extract medical features from the decrypted data stream, quantify them and convert them into a format suitable for machine learning and analysis, and build a fast search and time series index to improve query and analysis efficiency.

[0096] Furthermore, a preferred method is proposed, wherein step 2 includes: intelligently identifying outliers in the data stream, using machine learning algorithms (such as isolation forests) to automatically detect and mark abnormal data, such as abnormally high vital sign values, unreasonable drug dosages, etc.; interpolating missing data, and making reasonable estimates based on contextual information (such as data at adjacent time points, average data of similar patients) to reduce the impact of missing data on subsequent analysis.

[0097] The function of the big data storage module is to store and manage massive amounts of data. The technical means to achieve this function is to use a distributed file system to disperse the storage of data and use multiple nodes to improve storage capacity and reliability. At the same time, combined with a composite big data architecture, query performance is optimized. The two work together to ensure that the big data storage module can handle PB-level data storage needs and quickly respond to data access requests through data sharding and redundant backup mechanisms.

[0098] The structure of a distributed file system is described as follows: A distributed file system is a system that stores data on multiple servers. It aims to improve the reliability and availability of data and enable fast access and expansion of data. The system includes the following modules:

[0099] 1. Data nodes: In a distributed file system, data files are divided into multiple data blocks and stored in each data node. Each data node not only stores data blocks from multiple files, but also stores multiple copies of these data blocks to ensure data redundancy and the accuracy of subsequent operations.

[0100] 2. Storage strategy: In order to improve storage efficiency and data security, distributed file systems usually adopt specific storage strategies. For example, the storage locations of data blocks and replicas are carefully planned to balance the load, improve access speed, and enhance fault tolerance.

[0101] 3. Task decomposition and distribution: In the MapReduce process, a specific task will be decomposed into several small tasks, and these decomposed tasks will be distributed to each data node.

[0102] 4. Task Scheduling and Management: The master node is responsible for managing and scheduling the tasks of each sub-node. It will periodically receive the working status and results returned by the sub-nodes, and perform corresponding integration and processing.

[0103] 5. Hash table index: Hash table index uses hash function to map feature values ​​to specific storage locations, thereby achieving fast search. This indexing method is suitable for scenarios that require frequent search and update.

[0104] 6. Time series index: The time series index is sorted and stored according to the time sequence of the feature values, which is convenient for time series analysis and prediction. This indexing method is suitable for scenarios where it is necessary to analyze the trend of feature values ​​over time.

[0105] The function of the data processing and integration module is to clean, convert and integrate data from different sources. The technical means to achieve this function is the data cleaning algorithm to remove duplication, correct erroneous data, and integrate the data into a unified format through data mapping and standardization technology, providing a high-quality data foundation for subsequent analysis.

[0106] The data mapping and normalization techniques are described as follows:

[0107] Step 1: The system maps medical data from different sources and formats, that is, establishes a correspondence between data fields. For example, the disease diagnosis codes of different hospitals are uniformly mapped to the International Classification of Diseases (ICD), and the drug names are mapped to the generic names or chemical names in the standard drug database. At the same time, the data format is converted to a unified format that the system can recognize and easily process, such as JSON or XML, to facilitate subsequent data integration and analysis.

[0108] Step 2: After data mapping and format conversion, the system performs data standardization to ensure that all data follows unified specifications and standards. This includes standardization of data values ​​(such as age standardization to age groups), standardization of time formats (such as dates unified to YYYY-MM-DD format), and unit conversion (such as blood pressure values ​​from mmHg to kPa). In addition, the system will also perform data integrity checks and logical consistency checks to ensure that the data is logically reasonable and self-consistent.

[0109] Step 3: The standardized data is integrated into a centralized database that supports efficient data retrieval and analysis. During the integration process, the system will consider the timeliness, relevance, and importance of the data, and build a data warehouse or data lake so that subsequent intelligent scheduling algorithms can quickly access and utilize the data. At the same time, the data will be encrypted and stored in accordance with security and privacy protection requirements to ensure data security and compliance.

[0110] Furthermore, a preferred method is proposed, wherein step 1 includes: performing standardized mapping of medical terms, viewing and maintaining a comprehensive dictionary of medical terms to ensure that the same medical concept in different contexts can be correctly identified and uniformly expressed; performing time zone conversion and standardized processing on timestamp data to ensure the temporal consistency of cross-regional medical data, facilitating time series analysis and comparison.

[0111] The function of the intelligent index construction module is to automatically generate efficient indexes based on data content and structure. The technical means to achieve this function is to use natural language processing (NLP) technology to understand text content, combine machine learning algorithms to analyze data features, and automatically generate keyword indexes, semantic indexes, inverted indexes and other index types. At the same time, distributed index construction technology is used to accelerate the index generation process and improve index accuracy and query efficiency.

[0112] The distributed index construction technology is described as follows:

[0113] Step 1: Distributed data sharding and hash mapping: First, the collected medical big data is sharded using a distributed algorithm to ensure that each data fragment is of moderate size for distributed storage and processing. Next, a hash function is used to map each data fragment to generate a unique hash value, which will be used as the index key to quickly locate the data fragment. The hash mapping process ensures uniform distribution and fast access to data.

[0114] Furthermore, a preferred method is proposed, wherein the distributed data sharding and hash mapping in step 1 include: optimizing the shard size of medical big data to ensure that the shard size can meet the storage and processing requirements while avoiding the performance bottleneck caused by overly large shards; using efficient hash functions and hash conflict resolution strategies to reduce the occurrence of hash collisions and improve the accuracy and efficiency of hash mapping. In addition, the uniqueness and scalability of hash values ​​must also be considered to adapt to the growth of data volume and changes in query requirements.

[0115] Step 2: Metadata extraction and index entry construction: Based on data sharding and hash mapping, the metadata of each data fragment is extracted, including the size, storage location, timestamp, and key information summary of the data fragment. Then, index entries are constructed based on the metadata. Each index entry contains a pointer to the data fragment (i.e., hash value) and metadata information. This step provides the necessary information basis for subsequent index queries.

[0116] Step 3: Design and optimize the distributed index structure: Based on the extracted index entries, design the distributed index structure to support efficient data retrieval and query. This structure needs to fully consider the distribution characteristics of the data, the query mode and the system performance requirements, and is usually implemented using data structures such as distributed hash tables (DHT) and distributed B-trees. At the same time, the index structure is optimized, such as balancing the height of the tree and reducing the number of node accesses, to improve the query efficiency and stability of the index.

[0117] Step 4: Index construction and consistency maintenance: On the distributed storage system, the index is gradually constructed according to the designed index structure. This includes assigning index entries to corresponding index nodes and establishing associations between index nodes. At the same time, the index consistency is checked and maintained to ensure the accuracy and completeness of the index. This step also includes regular updates and optimizations of the index to adapt to changes in data and query requirements.

[0118] Embodiment 3,

[0119] This embodiment provides a medical resource scheduling system using an intelligent indexing system, such as Figure 2FIG. 1 is a schematic diagram of the system, which includes a data acquisition and preprocessing module, an intelligent analysis and prediction module, a medical resource allocation module, an intelligent monitoring and optimization module, and a cross-machine collaboration and information sharing module;

[0120] The data acquisition and preprocessing module is responsible for integrating and optimizing diverse data sources from major medical institutions, public health databases, and personal health monitoring devices;

[0121] The intelligent analysis and prediction module is responsible for simulating the flow and allocation process of medical resources by building a network model, so as to provide a scientific basis for resource scheduling;

[0122] The intelligent monitoring and optimization module is responsible for real-time monitoring of the utilization efficiency of medical resources, patient satisfaction and medical quality indicators, automatically identifying bottleneck links through data analysis, and proposing optimization suggestions;

[0123] The cross-machine collaboration and information sharing module is responsible for realizing cross-institutional medical resource information sharing and collaborative operations.

[0124] Data collection and preprocessing module: This module is designed to integrate and optimize data flows from a wide range of medical ecosystems, covering diverse data sources from major medical institutions, public health databases, and personal health monitoring devices. The medical big data processed is not limited to patient medical records, diagnosis and treatment details, medical resource allocation status, and medical equipment operation logs, but also extends to emerging health monitoring parameters to form a comprehensive and in-depth data set. The technology used in this module is not only simple distributed storage and cloud computing technology, but also blockchain processing is added on this basis. The specific data processing process is as follows:

[0125] S71. Collect raw data;

[0126] S72. Use a multi-source data fusion engine to automatically identify and adapt data sources of different formats and standards;

[0127] S73. Encrypt sensitive data and use anonymization technology to protect patient privacy;

[0128] S74, storing the data in a distributed database and establishing a connection with an intelligent indexing system;

[0129] S75, intelligent indexing system locates relevant data based on user query requirements;

[0130] S76. Store the data in a designated location and make a backup.

[0131] Intelligent analysis and prediction module: This module simulates the flow and allocation of medical resources by building a complex network model, providing a scientific basis for resource scheduling. First, a network model is built based on the types and distribution characteristics of medical resources. The innovation lies in the introduction of dynamic node attributes, such as changes in resource volume over time, priority of emergency needs, etc., to make the model closer to reality and provide a basis for subsequent simulations; an adaptive learning mechanism is designed so that the model can automatically optimize parameter settings and improve prediction accuracy based on feedback from simulation results and actual scheduling effects.

[0132] The specific workflow of the intelligent analysis and prediction module is as follows:

[0133] S81, introduce dynamic node attributes to build a network model framework;

[0134] S82. Design the priority of urgent requirements for the network model framework;

[0135] S83, setting network simulation parameters;

[0136] S84, associating the network simulation parameters with the historical data and setting them as dynamic adjustment data;

[0137] S85, introduce multi-objective optimization strategy;

[0138] S86, simulate the flow and allocation process of medical resources;

[0139] S87. Visual display of analysis report.

[0140] The methods and formulas or calculation steps used by the intelligent analysis and prediction module are as follows.

[0141] 1. Convolutional Neural Network (CNN)

[0142] principle:

[0143] The convolution operation is the core of CNN, which extracts local features of input data by sliding the convolution kernel on the input data, performing local connections and weighted summation. The convolution kernel is a small matrix whose weights remain unchanged during the sliding process, which helps to reduce the number of model parameters and reduce computational complexity. The activation function is used to introduce nonlinearity so that the neural network can learn more complex features. The pooling layer is used to downsample the output of the convolution layer to reduce the number of parameters and improve computational efficiency.

[0144] formula:

[0145]

[0146] Among them, W outPut is the width of the output feature map; W input is the width of the input image; W filteris the width of the convolution kernel (filter); P is the size of the padding, that is, the number of border pixel layers added to the edge of the input image. If the Same mode is used, P will be automatically adjusted according to the convolution kernel size and stride, so that the output feature map has the same size as the input image (or at least one of the height / width is the same); if the Valid mode is used, P = 0. S is the stride of the convolution kernel, that is, the number of pixels that the convolution kernel moves on the input image each time.

[0147] 2. Bayesian Network

[0148] principle:

[0149] Bayesian networks, also known as "belief networks" or formally as "directed acyclic graph models", use directed acyclic graphs (DAGs) as a tool to depict and analyze complex networks of relationships between features. The arrows in the figure connect two nodes, which symbolize that there is a causal relationship between two random variables or that they are not conditionally independent. The nodes themselves represent various random variables, including but not limited to observable variables, hidden variables, and unknown parameters. These variables are usually labeled {x_1, x_2, ..., x_n}. Bayesian networks usually use matrices to describe relationship networks.

[0150] 3. Decision Tree

[0151] principle:

[0152] The basic construction strategy of decision trees relies on a binary recursive partitioning principle, that is, each iteration ensures that the constructed decision tree nodes have two branches, thus naturally forming a binary tree structure. If a certain attribute has q different possible values, when this attribute is used as the partitioning basis to generate two branches, theoretically there are 2 (q-1)-1 This is because one of the q values ​​is selected as the dividing point to divide the data set into two, and all possible combinations of dividing points (except the case of no division) constitute these segmentation methods.

[0153] 4. Feature Engineering_PCA

[0154] principle:

[0155] PCA is a widely used dimensionality reduction technique that can convert multiple features (variables) into a few uncorrelated principal components while retaining as much important information as possible. The first principal component explains the largest variance in the data, and the subsequent principal components explain the largest part of the remaining variance in turn, and are orthogonal to each other (i.e., uncorrelated). PCA can not only reduce the dimension of the data, simplify the complexity of the model, and avoid overfitting, but also reveal the hidden structure in the data and promote a deeper understanding of the data. When dealing with high-dimensional data sets, PCA has become an indispensable tool, providing a more efficient and effective feature set for subsequent model training. From a mathematical point of view, PCA is a matrix decomposition algorithm; from a physical point of view, it is a linear transformation on a linear space.

[0156]

[0157] where p i is a row vector representing the i-th basis, a j is a column vector, representing the jth original data record. Using the idea of ​​combining numbers and shapes to solve this problem, we can get the following rectangular coordinate system.

[0158] The intelligent monitoring and optimization module includes an intelligent scheduling platform, a real-time monitoring platform and an abnormal warning platform;

[0159] Intelligent scheduling platform: When a hospital encounters resource shortage or sudden changes in demand, a resource scheduling plan is automatically generated based on preset scheduling rules and priorities. The scheduling plan includes the reallocation of beds, temporary shift changes of medical staff, and emergency deployment of medical equipment. The system automatically adjusts the scheduling plan based on real-time data and network models.

[0160] Real-time monitoring platform: continuously tracks the hospital's operating status, including patient waiting time, bed occupancy rate, and medical equipment usage, and transmits data to the intelligent scheduling platform, which displays monitoring data in real time in the form of charts to help hospital managers intuitively understand the hospital's operating status;

[0161] Abnormal warning platform: If abnormal indicators are monitored, an early warning will be issued immediately to remind hospital managers to take emergency measures.

[0162] The cross-machine collaboration and information sharing module realizes the information sharing and collaborative operation of medical resources across institutions. It obtains the information collection and dissemination authority through the authorization of relevant institutions, breaks the information barriers between traditional medical institutions, and realizes the information sharing and collaborative operation of medical resources across institutions. This not only promotes the rational use of medical resources, but also enhances the interoperability between medical institutions, providing patients with a seamless medical service experience.

Claims

1. Intelligent indexing system, characterized in that: The system includes a data acquisition module, a big data storage module, a data processing and integration module and an intelligent index construction module; The data acquisition module is responsible for collecting information from various data sources and transmitting data in the form of real-time data streams; The big data storage module is responsible for storing and managing the data transmitted by the data acquisition module and responding to data access requirements; The data processing and integration module is responsible for cleaning, converting and integrating data from different sources and inputting the data into the intelligent index building module; The intelligent index building module is responsible for automatically generating indexes according to data content and structure using distributed index building technology.

2. The intelligent indexing system according to claim 1, characterized in that: In the data acquisition module, data transmission in the form of real-time data stream is carried out in the following ways: S21, real-time capture of data streams in data sources through a distributed data collection framework; S22, performing preliminary preprocessing on the captured data stream, including data format conversion, timestamp correction, outlier detection and filtering, and field completion; S23, classifying and marking the data stream according to the business logic; S24. Use a secure communication protocol to establish an encrypted data transmission channel, and transmit the encrypted data stream to the big data storage module.

3. The intelligent indexing system according to claim 2, characterized in that: In the big data storage module, a distributed file system is used to disperse and store data. The specific steps of the distributed file system when working are as follows: S31. Data node storage: In a distributed file system, data files are divided into multiple data blocks and stored in each data node; S32, Task decomposition and distribution: In the MapReduce process, a specific task will be decomposed into several subtasks, and these decomposed subtasks will be distributed to each data node; S33, Task Scheduling and Management: The master node is responsible for managing and scheduling the tasks of each sub-node. The master node periodically receives the working status and results returned by the sub-nodes, and performs corresponding integration and processing; S34, Hash table index: Hash table index uses hash function to map feature values ​​to specific storage locations, thereby achieving fast search. This indexing method is suitable for scenarios that require frequent search and update; S35. Time series index: The time series index is sorted and stored according to the time sequence of the feature values, which facilitates time series analysis and prediction. This indexing method is suitable for scenarios where it is necessary to analyze the trend of feature values ​​changing over time.

4. The intelligent indexing system according to claim 3, characterized in that: When the data processing and integration module integrates data from different sources, it uses data mapping and standardization technology, which is specifically: S41. Map data from different sources and in different formats to establish a corresponding relationship between data fields; convert data formats to convert all data into a unified format that is recognizable and easy to process; S42. After data mapping and format conversion, perform data standardization to ensure that all data follows unified specifications and standards; S43. The standardized data are integrated into a centralized database that supports data retrieval and analysis.

5. The intelligent indexing system according to claim 4, characterized in that: The distributed index construction technology is specifically as follows: S51, distributed data sharding and hash mapping: first, the data is sharded using a distributed algorithm, and each data fragment is mapped using a hash function to generate a unique hash value, which will be used as an index key to locate the data fragment; S52, metadata extraction and index entry construction: based on data sharding and hash mapping, metadata of each data fragment is extracted, including the size, storage location, timestamp and key information summary of the data fragment, and index entries are constructed based on the metadata. Each index entry includes a pointer to the data fragment and metadata information; S53, distributed index structure design and optimization: based on the constructed index entries, a distributed index structure is designed to support efficient data retrieval and query. The distributed index structure is implemented using a distributed hash table or a distributed B-tree. At the same time, the index structure is optimized to balance the height of the tree, reduce the number of node accesses, and improve the query efficiency and stability of the index. S54, index construction and consistency maintenance: Assign index entries to corresponding index nodes and establish associations between index nodes. At the same time, perform consistency checks and maintenance on the index to ensure the accuracy and completeness of the index. Regularly update and optimize the index to adapt to changes in data and query requirements.

6. A medical resource scheduling system using an intelligent indexing system, characterized in that: The system includes a data acquisition and preprocessing module, an intelligent analysis and prediction module, a medical resource allocation module, an intelligent monitoring and optimization module, and a cross-machine collaboration and information sharing module; The data acquisition and preprocessing module is responsible for integrating and optimizing diverse data sources from major medical institutions, public health databases, and personal health monitoring devices; The intelligent analysis and prediction module is responsible for simulating the flow and allocation process of medical resources by building a network model, so as to provide a scientific basis for resource scheduling; The intelligent monitoring and optimization module is responsible for real-time monitoring of the utilization efficiency of medical resources, patient satisfaction and medical quality indicators, automatically identifying bottleneck links through data analysis, and proposing optimization suggestions; The cross-machine collaboration and information sharing module is responsible for realizing cross-institutional medical resource information sharing and collaborative operations.

7. The medical resource scheduling system using the intelligent indexing system according to claim 6, characterized in that: The workflow of the data acquisition and preprocessing module is specifically as follows: S71. Collect raw data; S72. Use a multi-source data fusion engine to automatically identify and adapt data sources of different formats and standards; S73. Encrypt sensitive data and use anonymization technology to protect patient privacy; S74, storing the data in a distributed database and establishing a connection with an intelligent indexing system; S75, intelligent indexing system locates relevant data based on user query requirements; S76. Store the data in a designated location and make a backup.

8. The medical resource scheduling system using the intelligent indexing system according to claim 7, characterized in that: The specific workflow of the intelligent analysis and prediction module is as follows: S81, introduce dynamic node attributes to build a network model framework; S82. Design the priority of urgent requirements of the network model framework; S83, setting network simulation parameters; S84, associating the network simulation parameters with the historical data and setting them as dynamic adjustment data; S85, introduce multi-objective optimization strategy; S86, simulate the flow and allocation process of medical resources; S87. Visual display of analysis report.

9. The medical resource scheduling system using the intelligent indexing system according to claim 8, characterized in that: The intelligent monitoring and optimization module includes an intelligent scheduling platform, a real-time monitoring platform and an abnormal warning platform; Intelligent scheduling platform: When a hospital encounters resource shortage or sudden changes in demand, a resource scheduling plan is automatically generated based on preset scheduling rules and priorities. The scheduling plan includes the reallocation of beds, temporary shift changes of medical staff, and emergency deployment of medical equipment. The system automatically adjusts the scheduling plan based on real-time data and network models. Real-time monitoring platform: continuously tracks the hospital's operating status, including patient waiting time, bed occupancy rate, and medical equipment usage, and transmits data to the intelligent scheduling platform, which displays monitoring data in real time in the form of charts to help hospital managers intuitively understand the hospital's operating status; Abnormal warning platform: If abnormal indicators are monitored, an early warning will be issued immediately to remind hospital managers to take emergency measures.

10. The medical resource scheduling system using the intelligent indexing system according to claim 9, characterized in that: The cross-machine collaboration and information sharing module realizes cross-institutional medical resource information sharing and collaborative operations, and obtains information collection and dissemination permissions through authorization from relevant institutions.

Images