AI-based multimodal medical information intelligent fusion processing system

By constructing a lookup popularity coordinate system and calculating lookup popularity coupled feature quantities, we can achieve refined quantitative evaluation and dynamic migration of existing sample data, solve the problem of mismatched storage resource configuration in multimodal medical information systems, and improve access efficiency and resource utilization.

CN123120348APending Publication Date: 2026-11-17GUILIN WOMEN & CHILDRENS HOSPITAL
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611271626.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing multimodal medical information intelligent fusion processing systems have shortcomings in storage resource optimization and data scheduling. They are difficult to conduct refined evaluation and dynamic migration based on search popularity, resulting in mismatched storage resource allocation and low access efficiency.

Method used

A coordinate system for browsing popularity is constructed. By calculating the coupled feature quantity of browsing popularity, a refined quantitative evaluation of the existing sample data is achieved. Based on the quantitative results, the samples are driven to dynamically migrate between hot and cold zones in the database. Combined with the real-time data generation, hot and cold data scheduling is performed to optimize the allocation of storage resources.

Benefits of technology

It improves the access response efficiency of high-frequency access to medical fusion data, makes reasonable use of high-speed storage resources, balances system storage costs and access performance, and dynamically adapts to data growth trends.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123120348A_ABST
    Figure CN123120348A_ABST
Patent Text Reader

Abstract

This invention discloses an AI-based intelligent fusion processing system for multimodal medical information, relating to the field of medical information. It addresses the problem of poor information processing efficiency in existing intelligent fusion processing systems. The system includes: acquiring a database for storing fused medical data, obtaining a medical storage database; performing historical lookup index analysis on the fused medical data contained in the medical storage database, obtaining a scatter plot of the lookup indexes corresponding to each fused medical data point, obtaining historical storage data; sorting the fused medical data by storage popularity based on the historical storage data; performing fusion processing on the existing medical information; obtaining processed existing information data based on the processing results; and performing real-time information processing on the real-time generated fused medical data based on the processed existing information data. This invention can improve the efficiency of intelligent fusion storage of multimodal medical information.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of medical information and involves database technology, specifically an AI-based multimodal medical information intelligent fusion processing system. Background Technology

[0002] Existing multimodal medical information intelligent fusion processing systems have the following shortcomings when performing medical information fusion processing: 1. Existing multimodal medical information intelligent fusion processing systems mostly lack a quantitative analysis mechanism for the browsing popularity of existing medical sample data during the medical information fusion processing process. It is difficult to conduct refined evaluation based on the characteristics of sample browsing behavior, and it is difficult to drive the dynamic migration and scheduling of samples between hot and cold storage areas based on the popularity evaluation results, which is not conducive to the optimal allocation of storage space resources. 2. Existing multimodal medical information intelligent fusion processing systems often struggle to adapt to the scale of real-time data generation in the cold and hot storage scheduling process. The data migration strategies for cold and hot areas lack real-time new data volume as a reference, which can easily lead to a mismatch between the allocation of cold and hot storage partition resources and the data growth trend.

[0003] To address this, we propose an AI-based intelligent fusion processing system for multimodal medical information. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this invention is to provide an AI-based intelligent fusion processing system for multimodal medical information, aiming to improve the efficiency of multimodal medical information fusion and storage.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an AI-based multimodal medical information intelligent fusion processing system, the specific working process of each module is as follows: Data acquisition module: acquires the database for medical data fusion and storage, obtains the medical storage database, performs historical access index analysis on the medical fusion data contained in the medical storage database, and obtains the scatter plot of access index corresponding to each medical fusion data based on the analysis results, thus obtaining the historical storage acquisition data; The existing information processing module sorts the medical fusion data by storage popularity based on historical stored and collected data, processes the existing medical information based on the sorting results, and obtains the processed existing information data based on the processing results. Real-time information processing module: Performs real-time information processing on the medical fusion data generated in real time based on the existing information processing data.

[0006] Furthermore, the historical stored data is acquired, specifically as follows: The medical data is collected from the databases used by medical institutions for integrated storage, resulting in a medical storage database. Collect the medical data that has been stored in the medical storage database to obtain multiple medical fusion data; Randomly select one sample of medical fusion data from the multiple collected medical fusion data, perform historical query index analysis on the sample fusion data, and obtain a scatter plot of query index based on the analysis results; Obtain the scatter plot of the query indicators corresponding to each medical fusion data point to obtain the historical stored and collected data.

[0007] Furthermore, the scatter plot of the indicators is obtained as follows: Collect historical data retrieval operations of sample fusion data recorded in the medical storage database, and arbitrarily select a sample retrieval operation from the collected historical data retrieval operations; Perform a lookup sufficiency analysis on the sample lookup operations, and obtain the data lookup sufficiency corresponding to the sample lookup operations based on the analysis results; Perform adjacent operation access interval analysis on sample lookup operations, and obtain the inverse vectorized value of the access interval corresponding to the sample lookup operation based on the analysis results; Obtain the inverse vectorized value of the access interval and the data sufficiency corresponding to each historical data retrieval operation; Create a search popularity coordinate system, and mark the coordinate points of the access interval inverse vectorization value on the horizontal axis and the data search abundance on the vertical axis as search popularity scatter points in the search popularity coordinate system to obtain a scatter plot of search index corresponding to the sample fusion data.

[0008] Furthermore, the data sufficiency is obtained, as detailed below: The time period covered by the sample lookup operation is collected to obtain the sample data lookup period. The collected sample data lookup period is discretized into several data lookup time points. The amount of data looked up by the database business terminal at each data lookup time point is collected to obtain multiple sample data lookup amounts. The median of the obtained sample data lookup amounts is collected to obtain the median data sample lookup amount. The storage space occupied by the sample fusion data in the medical storage database is collected to obtain the sample data fusion storage volume. The ratio of the median data sampling and viewing volume to the sample data fusion storage volume is calculated to obtain the data viewing fullness corresponding to the sample viewing operation.

[0009] Furthermore, an analysis of the access interval between adjacent operations was performed on the sample lookup operation, as detailed below: Collect the operation time point corresponding to each historical data query operation to obtain multiple data query time points. Create a data query time axis with the current time as the end point of the time axis and mark each collected data query time point on the data query time axis. In the data retrieval timeline, the data retrieval time point corresponding to the sample retrieval operation is marked as the sample retrieval time point, and the operation time point that is adjacent to the sample retrieval time point is marked as the sample adjacent time point. If there is only one sample adjacent time point, the timeline interval between the sample adjacent time point and the sample retrieval time point is marked as the operation retrieval interval duration corresponding to the sample retrieval operation. If there are two sample adjacent time points, the average of the timeline intervals between the sample adjacent time points and each sample retrieval time point is calculated to obtain the operation retrieval interval duration corresponding to the sample retrieval operation. Repeat the process of collecting the operation interval duration corresponding to the sample query operation, obtain the operation interval duration corresponding to each historical data query operation, and calculate the average of the obtained operation interval durations to obtain the average query interval corresponding to the sample fusion data.

[0010] Furthermore, an analysis of the access interval between adjacent operations was performed on the sample lookup operation, as detailed below: Repeat the process of collecting the average lookup interval corresponding to the sample fusion data, obtain the average lookup interval corresponding to each piece of medical fusion data, compare the values ​​of the multiple average lookup intervals obtained, collect the interval duration of the smallest average lookup interval to obtain the first preset interval duration, and collect the interval duration of the largest average lookup interval to obtain the second preset interval duration. Calculate the difference between the operation query interval duration corresponding to the sample query operation and the second preset interval duration to obtain the first preset interval difference. Calculate the difference between the first preset interval duration and the second preset interval duration to obtain the second preset interval difference. Calculate the ratio of the first preset interval difference to the second preset interval difference to obtain the preset interval ratio. Create a preset interval feature ratio. Calculate the difference between the preset interval feature ratio and the preset interval ratio to obtain the inverse vectorized value of the access interval corresponding to the sample query operation.

[0011] Furthermore, the existing medical information will be integrated and processed as follows: Retrieve historical stored data, and based on the historical stored data, obtain a scatter plot of the query indicators corresponding to each medical fusion data point; In the acquired medical fusion data, arbitrarily select one sample fusion data, perform coordinate scatter analysis on the scatter plot of the corresponding search index of the sample fusion data, and obtain the search popularity coupling feature quantity corresponding to the sample fusion data based on the analysis results. The search popularity coupling feature quantity corresponding to each search popularity scatter point is obtained, and the average of the obtained multiple search popularity coupling feature quantities is calculated to obtain the comprehensive value of the popularity coupling feature corresponding to the sample fusion data. The comprehensive value of the popularity coupling feature corresponding to each medical fusion data is obtained.

[0012] Furthermore, the existing medical information will be integrated and processed as follows: The available storage capacity of the hot data corresponding to the medical storage database is obtained to obtain the available storage quota for the hot zone; Based on the comprehensive quantity of heat coupling characteristics, the medical fusion data is labeled as T1 fusion data to Ta fusion data in descending order. The data volume of T1 fusion data and T2 fusion data is accumulated to obtain the M1 data accumulation. The data volume of T1 fusion data and T3 fusion data is accumulated to obtain the M2 data accumulation. And so on, the data volume of Ta-1 fusion data and Ta fusion data is accumulated to obtain the Ma-1 data accumulation. Calculate the difference between the accumulated data M1 and Ma-1 and the available storage quota of the hot zone, and mark the accumulated data corresponding to the smallest negative difference as the pre-selected accumulated data; The comprehensive value of heat coupling features of medical fusion data participating in the calculation of the cumulative amount of pre-selected data is collected to obtain the preset benchmark value of the viewing heat. Medical fusion data stored in the hot zone of the database and whose comprehensive value of heat coupling features is less than the preset benchmark value of the viewing heat is transferred to the cold zone of the database. Medical fusion data stored in the cold zone of the database and whose comprehensive value of heat coupling features is greater than or equal to the preset benchmark value of the viewing heat is transferred to the hot zone of the database. The medical fusion data stored in the database hot zone and the corresponding comprehensive quantity of heat coupling characteristics are set as stock information processing data.

[0013] Furthermore, the coupling features of the search popularity are obtained, as follows: Obtain the scatter plot of the query index. Select any query popularity scatter point in the scatter plot of the query index as the sample popularity scatter point. Draw a straight line perpendicular to the x-axis through the sample popularity scatter point to obtain the first coordinate characteristic line. Draw a straight line perpendicular to the y-axis through the sample popularity scatter point to obtain the second coordinate characteristic line. In the scatter plot of the index, the closed area formed by the x-axis, y-axis, first coordinate feature line and second coordinate feature line is obtained to obtain the viewing popularity projection area. The area value of the viewing popularity projection area is collected to obtain the viewing popularity projection area. In the scatter plot of the index, connect the sample popularity scatter points to the origin of the coordinate system to obtain the sample popularity scatter point connection line. Collect the length value of the sample popularity scatter point connection line to obtain the length of the popularity connection line. The length of the line connecting the search popularity and the projected area of ​​the search popularity are used to calculate the search popularity coupling feature corresponding to the sample popularity scatter points.

[0014] Furthermore, real-time information processing is performed on the medical fusion data, specifically as follows: Acquire existing information processing data, and based on the existing information processing data, obtain the medical fusion data stored in the database hot zone and the corresponding comprehensive quantity of hot coupling characteristics; The medical fusion data stored in the medical storage database at the current moment is collected to obtain multiple real-time fusion data. The data storage volume of the multiple real-time fusion data is accumulated to obtain the real-time new storage volume. Based on the comprehensive value of the heat coupling characteristics, the medical fusion data stored in the hot zone of the database are labeled as R1 fusion data to Rb fusion data in ascending order. The data volume of R1 fusion data to R2 fusion data is accumulated to obtain N1 cumulative data volume. The data volume of R1 fusion data to R3 fusion data is accumulated to obtain N2 cumulative data volume. And so on, the data volume of Rb-1 fusion data to Rb fusion data is accumulated to obtain Nb-1 cumulative data volume. Calculate the difference between the cumulative data volume from N1 to Na-1 and the real-time newly added storage volume. Mark the cumulative data volume corresponding to the smallest positive difference as the feature cumulative data volume. Collect the comprehensive value of the heat coupling feature of the medical fusion data involved in the calculation of the feature cumulative data volume. Mark the maximum comprehensive value of the heat coupling feature as the preset value of the hot zone migration. Medical fusion data stored in the hot zone of the database with a comprehensive value of heat coupling features less than or equal to the preset value for hot zone migration will be transferred to the cold zone of the database, while real-time fusion data will be stored in the hot zone of the database.

[0015] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention constructs a query popularity coordinate system and calculates query popularity coupling feature quantities to achieve a refined quantitative evaluation of the query popularity of existing sample fusion data. Based on the quantitative results, it can drive the dynamic migration of samples between hot and cold zones in the database, rationally allocate hot and cold storage resources, and improve the access response efficiency of high-frequency access to medical fusion data. 2. This invention combines the available storage quota in hot zones with the amount of real-time data generated to carry out hot and cold data scheduling constraints, dynamically adapting to the data growth trend of the system, maximizing the use of high-speed storage resources without exceeding the upper limit of hot zone storage capacity, and taking into account both system storage costs and access performance of multimodal medical data. Attached Figure Description

[0016] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0017] Figure 1 This is an overall system block diagram of the present invention; Figure 2 This is a schematic diagram of the data retrieval timeline of the present invention; Figure 3 This is a scatter plot of the reference indicators for this invention. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 Please see Figure 1 This invention provides a technical solution: an AI-based multimodal medical information intelligent fusion processing system, the specific working process of each module is as follows: The data acquisition module acquires the database for medical data fusion and storage, obtains the medical storage database, performs historical access index analysis on the medical fusion data contained in the medical storage database, and obtains a scatter plot of the access index corresponding to each medical fusion data based on the analysis results, thus obtaining the historical storage acquisition data. Specifically as follows: The medical data is collected from the databases used by medical institutions for integrated storage, resulting in a medical storage database. It should be noted here that: In this application, the medical unit referred to herein is a hospital, and the medical storage database referred to herein is specifically a collection of database systems used for persistent storage and management of various types of medical business data by medical institutions and medical platforms.

[0020] The medical data already stored in the medical storage database is collected to obtain multiple medical fusion data.

[0021] It should be noted here that: In this application, the medical fusion data involved here specifically refers to all medical data generated during a single patient visit as the basic storage unit. Each piece of stored data corresponds to a set of various types of medical data generated during a single patient visit, which can be used by medical personnel to query medical visit data.

[0022] From the multiple medical fusion data collected, one sample of fusion data is randomly selected, and historical query index analysis is performed on the sample fusion data. Based on the analysis results, a scatter plot of the query index is obtained.

[0023] Specifically as follows: Collect historical data retrieval operations of sample fusion data recorded in the medical storage database, and arbitrarily select a sample retrieval operation from the collected historical data retrieval operations; It should be noted here that: In this application, the historical data retrieval operation specifically refers to the access behavior initiated by the business terminal corresponding to the sample query record to the medical storage database for persistently stored medical fusion data. The access behavior involved here includes, but is not limited to, retrieval operation, data loading and reading operation, and partial retrieval operation.

[0024] The time period covered by the sample lookup operation is collected to obtain the sample data lookup period. The collected sample data lookup period is discretized into several data lookup time points. The amount of data looked up by the database business terminal at each data lookup time point is collected to obtain multiple sample data lookup volumes. The median of the obtained sample data lookup volumes is collected to obtain the median data sample lookup volume. It should be noted here that: In this application, the database business terminal involved here specifically refers to various business application terminals that use medical storage databases and initiate data query / read / write requests. These business application terminals include, but are not limited to, clinical workstations, MDT multidisciplinary consultation platforms, and medical record archiving and query systems.

[0025] The storage space occupied by the sample fusion data in the medical storage database is collected to obtain the sample data fusion storage volume. The ratio of the median data sampling and viewing volume to the sample data fusion storage volume is calculated to obtain the data viewing fullness corresponding to the sample viewing operation. Please see Figure 2 Collect the operation time point corresponding to each historical data query operation to obtain multiple data query time points, create a data query time axis with the current time as the end point of the time axis, and mark each collected data query time point on the data query time axis respectively. It should be noted here that: In this application, the data access time points marked on the data access time axis include the data access time points corresponding to the sample access operations.

[0026] In the data retrieval timeline, the data retrieval time point corresponding to the sample retrieval operation is marked as the sample retrieval time point, and the operation time point that is adjacent to the sample retrieval time point is marked as the sample adjacent time point. If there is only one sample adjacent time point, the timeline interval between the sample adjacent time point and the sample retrieval time point is marked as the operation retrieval interval duration corresponding to the sample retrieval operation. If there are two sample adjacent time points, the average of the timeline intervals between the sample adjacent time points and each sample retrieval time point is calculated to obtain the operation retrieval interval duration corresponding to the sample retrieval operation. Repeat the process of collecting the operation interval duration corresponding to the sample query operation, obtain the operation interval duration corresponding to each historical data query operation, and calculate the average of the obtained operation interval durations to obtain the average query interval corresponding to the sample fusion data.

[0027] Repeat the process of collecting the average lookup interval corresponding to the sample fusion data, obtain the average lookup interval corresponding to each piece of medical fusion data, compare the values ​​of the multiple average lookup intervals obtained, collect the interval duration of the smallest average lookup interval to obtain the first preset interval duration, and collect the interval duration of the largest average lookup interval to obtain the second preset interval duration. Calculate the difference between the operation query interval duration corresponding to the sample query operation and the second preset interval duration to obtain the first preset interval difference. Calculate the difference between the first preset interval duration and the second preset interval duration to obtain the second preset interval difference. Calculate the ratio of the first preset interval difference to the second preset interval difference to obtain the preset interval ratio. Create a preset interval feature ratio. Calculate the difference between the preset interval feature ratio and the preset interval ratio to obtain the inverse vectorized value of the access interval corresponding to the sample query operation.

[0028] It should be noted here that: In this application, the access interval inverse vectorization value is the inverse vectorization index of the lookup interval duration. The larger the access interval inverse vectorization value, the shorter the lookup interval duration between the sample lookup operation and its adjacent lookup operation, indicating that the lookup behavior distribution of the sample fused data is more dense and the lookup is relatively frequent, which represents that the sample has a higher lookup popularity.

[0029] Repeat the process of collecting the inverse vectorized value of the access interval and the data access fullness corresponding to the sample lookup operation, and obtain the inverse vectorized value of the access interval and the data access fullness corresponding to each historical data lookup operation. Create a search popularity coordinate system, and mark the coordinate points of the access interval inverse vectorization value on the horizontal axis and the data search abundance on the vertical axis as search popularity scatter points in the search popularity coordinate system to obtain the search index scatter plot corresponding to the sample fusion data. Repeat the process of collecting scatter plots of the query indicators corresponding to the sample fusion data, and obtain the scatter plots of the query indicators corresponding to each medical fusion data to obtain the historical storage data. The existing information processing module sorts the medical fusion data by storage popularity based on historical stored and collected data, performs fusion processing on the existing medical information based on the sorting results, and obtains the existing information processing data based on the processing results.

[0030] Specifically as follows: Retrieve historical stored data, and based on the historical stored data, obtain a scatter plot of the query indicators corresponding to each medical fusion data point; In the acquired medical fusion data, arbitrarily select one sample fusion data, perform coordinate scatter analysis on the scatter plot of the corresponding search index of the sample fusion data, and obtain the search popularity coupling feature quantity corresponding to the sample fusion data based on the analysis results. Specifically as follows: Please see Figure 3 Obtain the scatter plot of the query index. Select any query popularity scatter point in the scatter plot of the query index as the sample popularity scatter point. Draw a straight line perpendicular to the x-axis through the sample popularity scatter point to obtain the first coordinate characteristic line. Draw a straight line perpendicular to the y-axis through the sample popularity scatter point to obtain the second coordinate characteristic line.

[0031] In the scatter plot of the index, the closed area formed by the x-axis, y-axis, first coordinate feature line and second coordinate feature line is obtained to obtain the viewing popularity projection area. The area value of the viewing popularity projection area is collected to obtain the viewing popularity projection area. In the scatter plot of the index, connect the sample popularity scatter points to the origin of the coordinate system to obtain the sample popularity scatter point connection line. Collect the length value of the sample popularity scatter point connection line to obtain the length of the popularity connection line. The length of the line connecting the search popularity and the projected area of ​​the search popularity are used to calculate the search popularity coupling feature corresponding to the sample popularity scatter points. The specific formula for calculating the coupling characteristic of the search popularity is as follows: ; Where Ohl is the coupling feature of the search popularity, Sty is the projected area of ​​the search popularity, and Cxd is the length of the search popularity connection line. It should be noted here that: In this application, the following test data exists during the collection of the heat coupling feature quantity: ; It should be noted here that: In this application, the projected area Sty represents the overall clustering scale of the search points, and the length of the search heat connection line Cxd represents the upper limit of the heat extreme value of the search behavior. Taking the square root of the area Sty realizes the transformation from two-dimensional area to one-dimensional scale, completing the dimensional matching of the two geometric indicators. The multiplication of the two achieves feature synergistic coupling. This method requires the clustering scale and the heat extreme value to be at a high level simultaneously in order to obtain a higher search heat coupling feature quantity. It can effectively distinguish between continuous and stable search hotspots and pulse-like temporary search behaviors with only single-point peaks, avoiding false judgments of heat inflated by relying solely on the connection length. At the same time, the calculation is simple and adaptable to the engineering calculation requirements of batch cold and hot screening of medical fusion data.

[0032] Repeat the process of obtaining the search popularity coupling feature corresponding to the sample popularity scatter points, obtain the search popularity coupling feature corresponding to each search popularity scatter point, and calculate the average of the obtained multiple search popularity coupling feature values ​​to obtain the comprehensive value of the popularity coupling feature corresponding to the sample fusion data. Repeat the process of collecting the comprehensive value of the thermal coupling features corresponding to the sample fusion data, and obtain the comprehensive value of the thermal coupling features corresponding to each medical fusion data.

[0033] The available storage capacity of the hot data corresponding to the medical storage database is obtained to obtain the available storage quota for the hot zone; It should be noted here that: In this application, the available storage quota for the hot zone is specifically the upper limit of writable storage space in the hot storage area specifically allocated for storing hot data within the medical storage database. The medical storage database pre-divides into two types of storage areas: hot zone and cold zone. The hot zone is a high-speed storage medium with a higher read / write response speed and is only used to store sample fusion data with high access frequency and requiring fast access. The available storage quota for the hot zone serves as a capacity constraint threshold. When the total capacity of the hot data to be migrated approaches the quota, the system selects and eliminates sample fusion data with low access frequency based on the comprehensive value of the hot coupling characteristics, thereby realizing dynamic scheduling of hot and cold data.

[0034] Based on the comprehensive quantity of heat coupling characteristics, the medical fusion data is labeled as T1 fusion data to Ta fusion data in descending order. The data volume of T1 fusion data and T2 fusion data is accumulated to obtain the M1 data accumulation. The data volume of T1 fusion data and T3 fusion data is accumulated to obtain the M2 data accumulation. And so on, the data volume of Ta-1 fusion data and Ta fusion data is accumulated to obtain the Ma-1 data accumulation. It should be noted here that: In this application, T1 to Ta and M1 to Ma-1 are all marking symbols.

[0035] Calculate the difference between the accumulated data M1 and Ma-1 and the available storage quota of the hot zone, and mark the accumulated data corresponding to the smallest negative difference as the pre-selected accumulated data.

[0036] It should be noted here that: In this application, the minimum negative difference value involved here is specifically the difference between each data accumulation amount and the available storage quota of the hot zone, and the largest (closest to 0) difference among all differences. The pre-selected data accumulation amount corresponding to this difference represents that when the corresponding number of high-hot samples are included in the hot zone storage, the space occupied will not exceed the available storage quota of the hot zone, and the hot zone storage space can be fully utilized to the maximum extent. If a larger data accumulation amount is selected, the total storage capacity will exceed the hot zone quota limit.

[0037] The comprehensive value of heat coupling features of medical fusion data participating in the calculation of the cumulative amount of pre-selected data is collected to obtain the preset benchmark value of the viewing heat. Medical fusion data stored in the hot zone of the database and whose comprehensive value of heat coupling features is less than the preset benchmark value of the viewing heat is transferred to the cold zone of the database. Medical fusion data stored in the cold zone of the database and whose comprehensive value of heat coupling features is greater than or equal to the preset benchmark value of the viewing heat is transferred to the hot zone of the database.

[0038] It should be noted here that: In this application, the database hot zone and the database cold zone are two independent storage areas divided into medical storage databases. The database hot zone uses high read / write performance storage media to store sample fusion data with high comprehensive value of hot coupling features and high access frequency, ensuring fast retrieval response of multimodal medical data. The database cold zone uses low cost and large capacity storage media to store sample fusion data with low access frequency and low long-term access frequency. The system dynamically completes the migration of hot and cold data based on the sample access frequency, achieving a balance between access latency and storage cost.

[0039] The medical fusion data stored in the database hot zone and the corresponding comprehensive quantity of hot coupling characteristics are set as stock information processing data; The real-time information processing module performs real-time information processing on the medical fusion data generated in real time based on the existing information processing data. Specifically as follows: Acquire existing information processing data, and based on the existing information processing data, obtain the medical fusion data stored in the database hot zone and the corresponding comprehensive quantity of hot coupling characteristics; The medical fusion data stored in the medical storage database at the current moment is collected to obtain multiple real-time fusion data. The data storage volume of the multiple real-time fusion data is accumulated to obtain the real-time increase in storage volume.

[0040] Based on the comprehensive value of the heat coupling characteristics, the medical fusion data stored in the hot zone of the database are labeled as R1 fusion data to Rb fusion data in ascending order. The data volume of R1 fusion data to R2 fusion data is accumulated to obtain N1 cumulative data volume. The data volume of R1 fusion data to R3 fusion data is accumulated to obtain N2 cumulative data volume. And so on, the data volume of Rb-1 fusion data to Rb fusion data is accumulated to obtain Nb-1 cumulative data volume. Calculate the difference between the cumulative data volume from N1 to Na-1 and the real-time newly added storage volume. Mark the cumulative data volume corresponding to the smallest positive difference as the feature cumulative data volume. Collect the comprehensive value of the heat coupling feature of the medical fusion data involved in the calculation of the feature cumulative data volume. Mark the maximum comprehensive value of the heat coupling feature as the preset value of the hot zone migration. Medical fusion data stored in the hot zone of the database with a comprehensive value of heat coupling features less than or equal to the preset value for hot zone migration will be transferred to the cold zone of the database, and real-time fusion data will be stored in the hot zone of the database.

[0041] This invention constructs a lookup popularity coordinate system and calculates lookup popularity coupling feature quantities to achieve a refined quantitative assessment of the lookup popularity of existing sample fusion data. Based on the quantitative results, it can drive the dynamic migration of samples between hot and cold zones in the database, rationally allocate hot and cold storage resources, and improve the access response efficiency of high-frequency access to medical fusion data.

[0042] Furthermore, this invention combines the available storage quota in hot zones with the amount of real-time data generated to carry out hot and cold data scheduling constraints, dynamically adapting to the data growth trend of the system, maximizing the use of high-speed storage resources without exceeding the upper limit of hot zone storage capacity, and taking into account both system storage costs and access performance of multimodal medical data.

[0043] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementation methods. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The present invention is limited only to the technical solutions and their full scope and equivalents.

Claims

1. An AI-based multimodal medical information intelligent fusion processing system, characterized in that, include: Data acquisition module: acquires the database for medical data fusion and storage, obtains the medical storage database, performs historical access index analysis on the medical fusion data contained in the medical storage database, obtains the scatter plot of the access index corresponding to each medical fusion data, and obtains the historical storage acquisition data; The existing information processing module sorts the medical fusion data by storage popularity based on historical stored and collected data, and then processes the existing medical information based on the sorting results to obtain the existing information processing data. Real-time information processing module: Performs real-time information processing on the medical fusion data generated in real time based on the existing information processing data.

2. The AI-based multimodal medical information intelligent fusion processing system according to claim 1, characterized in that, The historical stored data is retrieved as follows: The medical data is collected from the databases used by medical institutions for integrated storage, resulting in a medical storage database. Collect the medical data that has been stored in the medical storage database to obtain multiple medical fusion data; Randomly select one sample of medical fusion data from the multiple collected medical fusion data, perform historical query index analysis on the sample fusion data, and obtain a scatter plot of query index based on the analysis results; Obtain the scatter plot of the query indicators corresponding to each medical fusion data point to get the historical stored and collected data.

3. The AI-based multimodal medical information intelligent fusion processing system according to claim 2, characterized in that, The scatter plot of the query indicator is obtained as follows: Collect historical data retrieval operations of sample fusion data recorded in the medical storage database, and select a sample retrieval operation from the collected historical data retrieval operations; Perform a lookup sufficiency analysis on the sample lookup operations to obtain the data lookup sufficiency corresponding to the sample lookup operations; Perform adjacent operation access interval analysis on sample lookup operations to obtain the inverse vectorized value of the access interval corresponding to the sample lookup operation; Obtain the inverse vectorized value of the access interval and the data sufficiency corresponding to each historical data retrieval operation; Create a coordinate system for browsing popularity, and mark the coordinate points with the horizontal axis representing the inverse vectorized value of the access interval and the vertical axis representing the data browsing abundance as browsing popularity scatter points to obtain a scatter plot of browsing indicators.

4. The AI-based multimodal medical information intelligent fusion processing system according to claim 3, characterized in that, The data sufficiency is obtained as follows: The time period covered by the sample lookup operation is collected to obtain the sample data lookup period. The collected sample data lookup period is discretized into several data lookup time points. The amount of data looked up by the database business terminal at each data lookup time point is collected to obtain multiple sample data lookup amounts. The median of the obtained sample data lookup amounts is collected to obtain the median data sample lookup amount. The storage space occupied by the sample fusion data in the medical storage database is collected to obtain the sample data fusion storage volume. The median data sampling access volume and the sample data fusion storage volume are calculated to obtain the data access sufficiency.

5. The AI-based multimodal medical information intelligent fusion processing system according to claim 3, characterized in that, The analysis of adjacent operation access intervals for sample lookup operations is as follows: Collect the operation time point corresponding to each historical data query operation to obtain multiple data query time points. Create a data query time axis with the current time as the end point of the time axis and mark each collected data query time point on the data query time axis. The data retrieval time point corresponding to the sample retrieval operation is marked as the sample retrieval time point. The operation time point that is adjacent to the sample retrieval time point is marked as the sample adjacent time point. If there is only one sample adjacent time point, the time axis interval between the sample adjacent time point and the sample retrieval time point is marked as the operation retrieval interval duration corresponding to the sample retrieval operation. If there are two sample adjacent time points, the average time interval between the sample adjacent time points and each sample retrieval time point is calculated to obtain the operation retrieval interval duration corresponding to the sample retrieval operation.

6. The AI-based multimodal medical information intelligent fusion processing system according to claim 5, characterized in that, The analysis of adjacent operation access intervals for sample lookup operations is as follows: Obtain the operation interval duration corresponding to each historical data query operation, and calculate the average of the multiple operation interval durations to obtain the average query interval corresponding to the sample fusion data. The average lookup interval corresponding to each piece of medical fusion data is obtained, and the values ​​of multiple average lookup intervals are compared. The minimum average lookup interval is collected to obtain the first preset interval duration, and the maximum average lookup interval is collected to obtain the second preset interval duration. The average lookup interval is normalized based on the first preset interval duration and the second preset interval duration to obtain the inverse vectorized value of the access interval corresponding to the sample lookup operation.

7. The AI-based multimodal medical information intelligent fusion processing system according to claim 1, characterized in that, The existing medical information will be integrated and processed as follows: Retrieve historical stored data, and based on the historical stored data, obtain the scatter plot of the corresponding query indicators for each medical fusion data point; In the acquired medical fusion data, arbitrarily select one sample fusion data, perform coordinate scatter analysis on the scatter plot of the corresponding search index of the sample fusion data, and obtain the search popularity coupling feature quantity corresponding to the sample fusion data. Obtain the lookup popularity coupling feature corresponding to each lookup popularity scatter point, and calculate the multiple lookup popularity coupling feature values ​​to obtain the comprehensive value of the popularity coupling feature corresponding to the sample fusion data. Obtain the comprehensive value of the popularity coupling feature corresponding to each medical fusion data.

8. The AI-based multimodal medical information intelligent fusion processing system according to claim 7, characterized in that, The existing medical information will be integrated and processed as follows: The available storage capacity of the hot data corresponding to the medical storage database is obtained to obtain the available storage quota for the hot zone; Based on the comprehensive quantity of heat coupling characteristics, the medical fusion data are labeled as T1 fusion data to Ta fusion data in descending order. The data volume of T1 fusion data and T2 fusion data are accumulated to obtain the M1 data accumulation. Similarly, the data volume of Ta-1 fusion data and Ta fusion data are accumulated to obtain the Ma-1 data accumulation. Calculate the difference between the accumulated data M1 and Ma-1 and the available storage quota of the hot zone, and mark the accumulated data corresponding to the smallest negative difference as the pre-selected accumulated data; The comprehensive value of heat coupling features of medical fusion data participating in the calculation of the cumulative amount of pre-selected data is collected to obtain the preset benchmark value of the viewing heat. Medical fusion data stored in the hot zone of the database and whose comprehensive value of heat coupling features is less than the preset benchmark value of the viewing heat is transferred to the cold zone of the database. Medical fusion data stored in the cold zone of the database and whose comprehensive value of heat coupling features is greater than or equal to the preset benchmark value of the viewing heat is transferred to the hot zone of the database. The medical fusion data stored in the database hot zone and the corresponding comprehensive quantity of heat coupling characteristics are set as stock information processing data.

9. The AI-based multimodal medical information intelligent fusion processing system according to claim 7, characterized in that, The heat coupling feature quantity is obtained as follows: Obtain the scatter plot of the query index. Select any query popularity scatter point in the scatter plot of the query index as the sample popularity scatter point. Draw a straight line perpendicular to the x-axis through the sample popularity scatter point to obtain the first coordinate characteristic line. Draw a straight line perpendicular to the y-axis through the sample popularity scatter point to obtain the second coordinate characteristic line. In the scatter plot of the index, the closed area formed by the x-axis, y-axis, first coordinate feature line and second coordinate feature line is obtained to obtain the viewing popularity projection area. The area value of the viewing popularity projection area is collected to obtain the viewing popularity projection area. In the scatter plot of the index, connect the sample popularity scatter points to the origin of the coordinate system to obtain the sample popularity scatter point connection line. Collect the length value of the sample popularity scatter point connection line to obtain the length of the popularity connection line. The length of the connection line and the projected area of ​​the search popularity are used to calculate the coupling characteristic of the search popularity.

10. The AI-based multimodal medical information intelligent fusion processing system according to claim 1, characterized in that, Real-time information processing of medical fusion data is performed as follows: Acquire existing information processing data, and based on the existing information processing data, obtain the medical fusion data stored in the database hot zone and the corresponding comprehensive quantity of hot coupling characteristics; The medical fusion data stored in the medical storage database at the current moment is collected to obtain multiple real-time fusion data. The data storage volume of the multiple real-time fusion data is accumulated to obtain the real-time new storage volume. Based on the comprehensive value of the heat coupling characteristics, the medical fusion data stored in the hot zone of the database are labeled as R1 fusion data to Rb fusion data in ascending order. The data volume of R1 fusion data to R2 fusion data is accumulated to obtain N1 cumulative data volume. Similarly, the data volume of Rb-1 fusion data to Rb fusion data is accumulated to obtain Nb-1 cumulative data volume. Calculate the difference between the cumulative data volume from N1 to Na-1 and the real-time newly added storage volume. Mark the cumulative data volume corresponding to the smallest positive difference as the feature cumulative data volume. Collect the comprehensive value of the heat coupling feature of the medical fusion data involved in the calculation of the feature cumulative data volume. Mark the maximum comprehensive value of the heat coupling feature as the preset value of the hot zone migration. Medical fusion data stored in the hot zone of the database with a comprehensive value of heat coupling features less than or equal to the preset value for hot zone migration will be transferred to the cold zone of the database, and real-time fusion data will be stored in the hot zone of the database.