A system for realizing object-oriented medical software pages based on feature engineering

Abstract

The present invention relates to the technical field of classification of medical information management programs, and discloses a system for realizing objectification of medical software pages based on feature engineering. The system for realizing objectification of medical software pages based on feature engineering adopts feature engineering technology to objectify the pages of hospital information management programs. The method of the present invention can allow the program to operate the hospital information management program like a person. If the hospital information management program is updated, most users can still quickly become familiar with the use of the new hospital information management program after simple thinking. Similarly, the system can still discover and record the changes in pages or functions that occur after the hospital information management program is updated, and quickly generate new solutions through the comparison function of feature engineering to provide a unified interface for the outside, while the outside can hardly perceive the changes in the system and still call the changed interface in the same way as calling the original interface.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information management program classification, and in particular to cross-platform information interaction, feature engineering, decision algorithms, automatic classification and clustering of data of medical information management programs, and specifically to a system for realizing objectification of medical software pages based on feature engineering. Background Art

[0002] With the rapid development of information technology, more and more hospitals in China are accelerating the overall construction of information platforms, HIS, EMR and other systems to improve the service level and core competitiveness of hospitals. Informatization not only improves the work efficiency of doctors, allowing doctors to have more time to serve patients, but also improves patient satisfaction and trust, and invisibly establishes the scientific and technological image of hospitals. At the same time, the information platform also makes communication between hospitals smoother and management more efficient.

[0003] The existing Chinese patent (publication number: CN112511591) proposes a method for implementing hospital interface data interaction, which is used for an agent service integration platform, including a configuration process and a data conversion process; the configuration process includes: S11, endpoint configuration, which is used to configure endpoint information, including interface address, namespace, belonging system, and whether standardized conversion is required; S12, agent service configuration, which is used to configure agent service information, including the belonging interface, method name, whether standardized conversion is required, and method parameters that encapsulate a specific interface service method; S13, data conversion configuration, which is used to configure the mapping configuration information of the business system dictionary and the standard dictionary of both parties in data conversion and the binding configuration information of the request message and the return message node data and the dictionary. Information; the data conversion process includes: S21, when receiving a request message for a proxy service of a business system, the request message for the proxy service is a proxy service message, first parsing the header parameters of the request message, querying and obtaining the requested proxy service information according to the header parameters, the mapping configuration information of the business system dictionaries and the standard dictionary of both parties of the data conversion, and the binding configuration information of the request message and the return message node data and the dictionary, as well as the corresponding endpoint information; S22, performing dictionary conversion on the request message according to the proxy service information, the mapping configuration information of the business system dictionaries and the standard dictionary of both parties of the data conversion, and the binding configuration information of the request message and the return message node data and the dictionary; S23, returning the message dictionary conversion result according to the corresponding endpoint information. The data conversion configuration includes: a proxy service message node business dictionary binding configuration, which is used to configure the business dictionary binding information of the proxy service message node, including configuring the dictionary of the node binding corresponding to the request system in the request message and the return message of the proxy service; an endpoint original interface message node business dictionary binding configuration, which is used to configure the business dictionary binding information of the endpoint original interface message node, including configuring the dictionary of the node binding corresponding to the request system in the original request and return message of the endpoint; and a mapping configuration of each business system dictionary and a standard dictionary, which is used to configure the mapping relationship between the dictionary of each business system and the standard dictionary. The proxy service message has the following unified format: a unified request address, located in the address bar of the request message, indicating the interface address of the proxy service integration platform; a unified method name, located in the method node of the request message, determined by the interface definition of the proxy service, indicating the name of the interface method to which the requested interface service method belongs; a request system code, located in the header of the request message, indicating the code of the business system that issues the request message; a service method transaction code, located in the header of the request message, indicating the proxy service identification parameter of the current proxy service; a request system identity code, located in the header of the request message, indicating the authentication basis of the request system; and service parameters, located in the body of the request message, indicating the service parameters of the requested interface service method.

[0004] There is also a Chinese patent (publication number: CN112463765A) that proposes a medical data management method based on a big data architecture, building a data center dataLake data lake to manage all collected data, analysis process data, analysis models and analysis result data, and job scheduling, supporting the smooth analysis process of all data on the entire platform; building a big data collection platform, which is based on an open source heterogeneous data synchronization engine, creates data interfaces for different data types, expands different read and write plug-ins, and stores the collected data in a big data center after pre-processing and classification; building a big data computing platform, which uses a hybrid approach to store the collected data. The data is stored in a storage mode, and the Spark computing engine is used to provide data analysis and mining for the stored data; a big data service platform is built, and the big data service platform uniformly manages and monitors the information of the big data center, and the information of the big data center includes the total number of databases, the total number of data, the total capacity, the overall information of the basic table and the synthetic table; a big data exchange platform is built, and the big data exchange platform uniformly manages the information exchange between the hospital's big data center and the in-hospital business system and other information platforms outside the hospital; a big data application platform is built, and the big data application platform is used to directly query the information of the big data center, and the analysis results are presented in a variety of different charts. The big data collection platform is also used to: support the collection of data sources of different types of databases including Mysql, Postgresql, Clickhouse, Oracle, and Sqlserver; support writing the collected data to hive, Mysql, Postgresql, Clickhouse, Oracle, and Sqlserver; support the configuration of the read and write speed to avoid the read and write tasks occupying too many server resources and affecting business applications; support the specified data source of the collection through the visual interface, preview the data table by checking the data table to be extracted, and lock the data range to be extracted by checking the data field under the specified data table.

[0005] The above similar methods have great deficiencies when applied to commercial hospital information management programs, i.e. hospital management software products: currently popular similar products also use simple image recognition technology or color recognition technology to implement scripts that simulate human operations. The interfaces provided by existing hospital management software, i.e. management methods, are not standardized and universal. Not only are the interfaces of various hospital management software not unified, but each time a new system is faced or the original system is upgraded, a new set of scripts must be rebuilt, and the sample sets of colors and images also need to be replaced manually, which involves relatively cumbersome and time-consuming workloads. In addition, current similar products cannot overcome the problem of interface failure caused by changes in the working environment of hospital information management programs.

[0006] Therefore, to address these problems, we proposed a system based on feature engineering to objectify medical software pages. Summary of the invention

[0007] The purpose of the present invention is to propose a system for realizing the objectification of medical software pages based on feature engineering, which is mainly used to connect with medical internal management information software and database systems, and provide interface services for various medical application software projects such as connecting various information management systems of the hospital and realizing seamless information interaction within the hospital. The invention can objectify the pages of hospital information management programs, and automatically convert a series of functions such as browsing, searching, editing, saving, importing and exporting in the hospital information management programs into interfaces for calls by acquisition software, etc.

[0008] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a system for realizing objectification of medical software pages based on feature engineering, a functional module for realizing objectification of hospital information management program pages by feature engineering, comprising the following steps:

[0009] S1: pre-processing of hospital information management procedures;

[0010] Provide configuration information for the target medical information management program, extract preliminary features, and mark the main functional points;

[0011] S2: Operational training of hospital information management procedures;

[0012] Start by opening the program, logging in, entering the target page, and clicking the target button;

[0013] S3: System generator for objectifying hospital information management program pages;

[0014] S4: Medical information management program operation tools;

[0015] S5: Toolset;

[0016] a. PC event API;

[0017] b. PC event listener;

[0018] c. PC screen capture;

[0019] d. Application domain positioning;

[0020] e. Application event scanning;

[0021] f. Feature extraction;

[0022] S6: Log system;

[0023] a. Record the training process;

[0024] b. Record the operation process.

[0025] Preferably, the system is composed of a feature library, an API mapping table, an application management system, a log system, a database system, and a control center, and the system operation includes the following steps:

[0026] S1: The control center sends instructions to the application management system;

[0027] S2: the control center sends log data to the arrival log;

[0028] S3: The control center interacts with the DB system for data;

[0029] S4: The control center performs API matching with the API mapping system;

[0030] S5: The feature library performs feature verification on the control center.

[0031] Preferably, the system deployment process of the object-oriented hospital information management program page includes the following steps:

[0032] S1: An operator sends an operation instruction to the hospital information management program for objectification. The operator can be a person or a program.

[0033] S2: The objectification of the hospital information management program can map the operation of the target program into an operation interface through interaction with the hospital information management system;

[0034] S3: The objectification of the hospital information management program can reach the operator through the operation interface;

[0035] S4: The hospital information management program feeds back the execution result to the hospital information management program object.

[0036] Preferably, the operating principle of the page logic parser is as follows:

[0037] S1: The page logic parsing engine obtains the feature library, page, and event, and then issues events, verifies events, and verifies pages.

[0038] S2: The page logic parsing engine can modify the page logic tree; the internal system of the page logic tree refers to the page logic parsing engine;

[0039] S3: The page logic parsing engine compares pages and features and performs feature aggregation at the same time;

[0040] S4: After comparing the pages, page fitting is performed, and after comparing the features, new features are generated.

[0041] Preferably, the characteristic principle includes the following process:

[0042] S1: After the features are stored, they enter the feature recognition neural network, followed by the feature data compressor;

[0043] S2: The feature data compressor then enters the standard neural network, which performs analysis to obtain the standard neural network feature parameters;

[0044] S3: Standard neural network feature parameters are stored in the feature library through indexes;

[0045] S4: The feature extractor finally obtains the features.

[0046] Preferably, the page logic diagram data includes the following process:

[0047] S1: The initial node implements the initial interface of the PC through the initialization process;

[0048] S2: The initial interface on the PC side then performs operations to realize page a and page b;

[0049] S3: The initial interface of the PC then performs operations, opens the interface of the target program, operates the program, and then opens page 1, page 2, page 3, and page 4;

[0050] S4: Page 1 in the S3 process may return to the interface of the target program;

[0051] S5: Page 1 in the S3 process can be opened to page 5 and page 6.

[0052] Preferably, the construction tool set comprises the following steps:

[0053] a. Encapsulated event API: including mouse and keyboard events;

[0054] b. Encapsulate event monitoring API to monitor device events;

[0055] c. Device screen acquisition interface: save data frame by frame;

[0056] d. Application domain positioning: By comparing the original window and the window of the application, the position of the application on the screen is obtained by subtracting the difference;

[0057] e. Application event scanning: Scan the page with the mouse, record the place where the mouse changes on the page, and verify whether the location is the event triggering location;

[0058] f. Feature extraction: When a small amount of changes occur between frames, these small changes can generally be used as effective features;

[0059] g. Feature library construction;

[0060] h. Build other tools: database tools, neural network structure templates, etc.

[0061] Preferably, the preprocessing of the hospital information management system in the S1 process includes the following steps:

[0062] a. Encapsulate the opening and closing methods of the hospital information management system;

[0063] b. Write the login account information of the hospital information management system into the configuration file;

[0064] c. Annotate the main functional points and features of the hospital information management software and save them in csv file;

[0065] d. Preset a binary classification convolutional neural network;

[0066] f. Use the labeled csv files to train multiple binary classification convolutional neural networks, and import the trained neural networks into the feature library;

[0067] e. Build the page graph data structure and initialize it.

[0068] Preferably, the hospital information management program page objectification program includes the following steps:

[0069] a. Enable the page logic parser. The built-in engine of the page logic parser drives the event API to start the hospital information management system on the PC;

[0070] b. After the hospital information system is started, the page logic parser monitors the screen changes through the screen capture module and obtains the position of the application on the screen through the application domain positioning tool;

[0071] c. The page logic parser writes the application startup success event into the page logic graph;

[0072] d. The page logic parser controls the mouse to scan the application domain through the event API. If a page change is found, the page change is recorded and the changed part of the page is recorded as a preliminary feature. The preliminary feature can be determined by the feature score (feature verification) to determine whether it becomes a new feature access feature library. The changed page position is also recorded as a possible click event. Event verification can be used to determine whether it is a valid event. If it is a valid event, the page or PC background will change after clicking;

[0073] e. When a new event is generated, we believe that the page has undergone an essential change and has entered a new state of the application. We identify the changed page as a new page and record it in the page logic graph. The event that generates the new page is recorded as a directed line from the previous graph node to this graph node.

[0074] f. Repeat this process. When a loop is encountered, jump out of the loop and take a new branch. If a loop cannot be jumped out, close the hospital information management system program, restart it, and return to the origin of the page logic diagram;

[0075] g. Save the trained page logic diagram and the generated new features.

[0076] Preferably, the page objectification program is generated using the already generated page logic diagram and feature library.

[0077] The technical solution of the present invention has the following beneficial effects:

[0078] The system for realizing objectification of medical software pages based on feature engineering adopts feature engineering technology to objectify the pages of the hospital information management program. The method of the present invention can allow the program to operate the hospital information management program just like a person. If the hospital information management program is updated, most users (users) can still quickly become familiar with the new hospital information management program after simple thinking. Similarly, the system can still discover and record the changes in pages or functions that occur after the hospital information management program is updated, and quickly generate new solutions through the comparison function of feature engineering to provide a unified interface for the outside world, while the outside world can hardly perceive the changes in the system and still call the changed interface in the same way as calling the original interface. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 The system function diagram of the object-oriented hospital information management program page of the present invention;

[0080] Figure 2 The system deployment application diagram for objectifying the hospital information management program page of the present invention;

[0081] Figure 3 It is a training flow chart of the present invention;

[0082] Figure 4 This is a schematic diagram of the principle of the page logic parser of the present invention;

[0083] Figure 5 It is a schematic diagram of the principle of the feature library of the present invention;

[0084] Figure 6 This is the data structure diagram of the page logic diagram of the present invention. DETAILED DESCRIPTION

[0085] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 scope of protection of the present invention.

[0086] Embodiment 1

[0087] See also Figure 1-Figure 6 , a system for realizing objectification of medical software pages based on feature engineering, a functional module for realizing objectification of hospital information management program pages by feature engineering, comprising the following steps:

[0088] S1: pre-processing of hospital information management procedures;

[0089] Provide configuration information for the target medical information management program, extract preliminary features, and mark the main functional points;

[0090] S2: Operational training of hospital information management procedures;

[0091] Start by opening the program, logging in, entering the target page, and clicking the target button;

[0092] S3: System generator for objectifying hospital information management program pages;

[0093] S4: Medical information management program operation tools;

[0094] S5: Toolset;

[0095] a. PC event API;

[0096] b. PC event listener;

[0097] c. PC screen capture;

[0098] d. Application domain positioning;

[0099] e. Application event scanning;

[0100] f. Feature extraction;

[0101] S6: Log system;

[0102] a. Record the training process;

[0103] b. Record the operation process.

[0104] The system consists of a feature library, an API mapping table, an application management system, a log system, a database system, and a control center. The system operation includes the following steps:

[0105] S1: The control center sends instructions to the application management system;

[0106] S2: The control center sends log data to the arrival log;

[0107] S3: The control center interacts with the DB system for data;

[0108] S4: The control center matches the API with the API mapping system;

[0109] S5: The feature library performs feature verification on the control center.

[0110] The system deployment process of object-oriented hospital information management program page includes the following steps:

[0111] S1: The operator sends the operation instruction to the hospital information management program for objectification. The operator can be a person or a program.

[0112] S2: The objectification of hospital information management program can map the operations of the target program into the operation interface through interaction with hospital information management;

[0113] S3: The objectification of hospital information management programs can reach operators through the operation interface;

[0114] S4: The hospital information management program feeds back the execution results to the hospital information management program for objectification.

[0115] The page logic parser operates as follows:

[0116] S1: The page logic parsing engine obtains the feature library, page, and event, and then issues events, verifies events, and verifies pages.

[0117] S2: The page logic parsing engine can modify the page logic tree; the internal system of the page logic tree refers to the page logic parsing engine;

[0118] S3: The page logic parsing engine compares pages and features and performs feature aggregation at the same time;

[0119] S4: After comparing the pages, page fitting is performed, and after comparing the features, new features are generated.

[0120] The characteristic principle includes the following processes:

[0121] S1: After the features are stored, they enter the feature recognition neural network, followed by the feature data compressor;

[0122] S2: The feature data compressor then enters the standard neural network, which performs analysis to obtain the standard neural network feature parameters;

[0123] S3: Standard neural network feature parameters are stored in the feature library through indexes;

[0124] S4: The feature extractor finally obtains the features.

[0125] The page logic diagram data includes the following processes:

[0126] S1: The initial node implements the initial interface of the PC through the initialization process;

[0127] S2: The initial interface on the PC side then performs operations to realize page a and page b;

[0128] S3: The initial interface of the PC then performs operations, opens the interface of the target program, operates the program, and then opens page 1, page 2, page 3, and page 4;

[0129] S4: Page 1 in the S3 process can return to the interface of the target program;

[0130] S5: Page 1 in the S3 process can be opened to page 5 and page 6.

[0131] Embodiment 2

[0132] As an embodiment of the present invention, please refer to Figure 1-6 The difference between this implementation and the first implementation is that: constructing the tool set includes the following steps:

[0133] a. Encapsulated event API: including mouse and keyboard events;

[0134] b. Encapsulate event monitoring API to monitor device events;

[0135] c. Device screen acquisition interface: save data frame by frame;

[0136] d. Application domain positioning: By comparing the original window and the window of the application, the position of the application on the screen is obtained by subtracting the difference;

[0137] e. Application event scanning: Scan the page with the mouse, record the place where the mouse changes on the page, and verify whether the location is the event triggering location;

[0138] f. Feature extraction: When a small amount of changes occur between frames, these small changes can generally be used as effective features;

[0139] g. Feature library construction;

[0140] h. Build other tools: database tools, neural network structure templates, etc.

[0141] The preprocessing of hospital information management system includes the following steps:

[0142] a. Encapsulate the opening and closing methods of the hospital information management system;

[0143] b. Write the login account information of the hospital information management system into the configuration file;

[0144] c. Annotate the main functional points and features of the hospital information management software and save them in csv file;

[0145] d. Preset a binary classification convolutional neural network;

[0146] f. Use the labeled csv files to train multiple binary classification convolutional neural networks, and import the trained neural networks into the feature library;

[0147] e. Build the page graph data structure and initialize it.

[0148] The hospital information management program page objectification process includes the following steps:

[0149] a. Enable the page logic parser. The built-in engine of the page logic parser drives the event API to start the hospital information management system on the PC;

[0150] b. After the hospital information system is started, the page logic parser monitors the screen changes through the screen capture module and obtains the position of the application on the screen through the application domain positioning tool;

[0151] c. The page logic parser writes the application startup success event into the page logic graph;

[0152] d. The page logic parser controls the mouse to scan the application domain through the event API. If a page change is found, the page change is recorded and the changed part of the page is recorded as a preliminary feature. The preliminary feature can be determined by the feature score (feature verification) to determine whether it becomes a new feature access feature library. The changed page position is also recorded as a possible click event. Event verification can be used to determine whether it is a valid event. If it is a valid event, the page or PC background will change after clicking;

[0153] e. When a new event is generated, we believe that the page has undergone an essential change and has entered a new state of the application. We identify the changed page as a new page and record it in the page logic graph. The event that generates the new page is recorded as a directed line from the previous graph node to this graph node.

[0154] f. Repeat this process. When a loop is encountered, jump out of the loop and take a new branch. If a loop cannot be jumped out, close the hospital information management system program, restart it, and return to the origin of the page logic diagram;

[0155] g. Save the trained page logic diagram and the generated new features, and generate a page object program using the generated page logic diagram and feature library.

[0156] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A system for realizing object-oriented medical software pages based on feature engineering, characterized in that: The feature engineering realizes the functional module of object-oriented hospital information management program page, including the following steps: S1: pre-processing of hospital information management procedures; Provide configuration information for the target medical information management program, extract preliminary features, and mark the main functional points; S2: Operational training of hospital information management procedures; Start by opening the program, logging in, entering the target page, and clicking the target button; S3: System generator for objectifying hospital information management program pages; S4: Medical information management program operation tools; S5: Build toolset; a. PC event API; b. PC event listener; c. PC screen capture; d. Application domain positioning; e. Application event scanning; f. Feature extraction; S6: Log system; a. Record the training process; b. Record the operation process; The objectification of the hospital information management program page in the S3 process includes the following steps: a. Enable the page logic parser. The built-in engine of the page logic parser drives the event API to start the hospital information management system on the PC; b. After the hospital information management system is started, the page logic parser monitors the screen changes through the screen capture module and obtains the position of the application on the screen through the application domain positioning tool; c. The page logic parser writes the application startup success event into the page logic graph; d. The page logic parser controls the mouse to scan the application domain through the event API. If a page change is found, the page change is recorded and the changed page part is recorded as a preliminary feature. The preliminary feature can be determined by the feature score to determine whether it becomes a new feature to access the feature library. The changed page position is also recorded as a possible click event. Through event verification, it can be determined whether it is a valid event. If it is a valid event, the page or PC background will change after clicking; e. When a new event is generated, it is considered that the page has undergone an essential change and has entered a new state of the application. The changed page is determined as a new page and recorded in the page logic graph. The event that generates the new page is recorded as a directed line from the previous graph node to the current graph node. f. Repeat this process. When a loop is encountered, jump out of the loop and take a new branch. If a loop cannot be jumped out, close the hospital information management system program, restart it, and return to the origin of the page logic diagram; g. Save the trained page logic diagram and the generated new features; The operating principle of the page logic parser is as follows: S1: The page logic parsing engine obtains the feature library, page, and event, and then issues events, verifies events, and verifies pages; S2: The page logic parsing engine can modify the page logic tree; the internal system of the page logic tree refers to the page logic parsing engine; S3: The page logic parsing engine compares pages and features and performs feature aggregation at the same time; S4: After comparing the pages, page fitting is performed, and after comparing the features, new features are generated.

2. According to claim 1, a system for realizing object-oriented medical software pages based on feature engineering, characterized in that: The system consists of a feature library, an API mapping table, an application management system, a log system, a database system, and a control center. The following steps are involved: S1: The control center sends instructions to the application management system; S2: The control center sends log data to the log system; S3: The control center interacts with the database system for data; S4: The control center performs API matching with the API mapping table; S5: The feature library performs feature verification on the control center.

3. A system for realizing object-oriented medical software page based on feature engineering according to claim 2, characterized in that: The system deployment process of the object-oriented hospital information management program page includes the following steps: S1: An operator sends an operation instruction to the hospital information management program for objectification. The operator can be a person or a program. S2: The hospital information management can be objectified with the hospital information management program to map the operation of the target program into an operation interface through interaction; S3: The objectification of the hospital information management program can reach the operator through the operation interface; S4: The hospital information management program feeds back the execution result to the hospital information management program object.

4. According to claim 1, a system for realizing object-oriented medical software page based on feature engineering, characterized in that: Feature extraction The following processes are included: S1: After the features are stored, they enter the feature recognition neural network, followed by the feature data compressor; S2: The feature data compressor then enters the standard neural network, which performs analysis to obtain the standard neural network feature parameters; S3: Standard neural network feature parameters are stored in the feature library through indexes; S4: The feature extractor finally obtains the features.

5. According to claim 1, a system for realizing object-oriented medical software page based on feature engineering, characterized in that: The page logic diagram data includes the following processes: S1: The initial node implements the initial interface of the PC through the initialization process; S2: The initial interface on the PC side then performs operations to realize page a and page b; S3: The initial interface of the PC then performs operations, opens the interface of the target program, operates the program, and then opens page 1, page 2, page 3, and page 4; S4: Page 1 in the S3 process may return to the interface of the target program; S5: Page 1 in the S3 process can be opened to page 5 and page 6.

6. The system for realizing object-oriented medical software page based on feature engineering according to claim 1, characterized in that: The S5 process to build the toolset includes the following steps: a. Encapsulated event API: including mouse and keyboard events; b. Encapsulate event monitoring API to monitor device events; c. Device screen acquisition interface: save data frame by frame; d. Application domain positioning: By comparing the original window and the window of the application, the position of the application on the screen is obtained by subtracting the difference; e. Application event scanning: Scan the page with the mouse, record the place where the mouse changes on the page, and verify whether the location is the event triggering location; f. Feature extraction: When a small amount of changes occur on the page between frames, these small changes can be used as effective features; g. Feature library construction; h. Build other tools: database tools, neural network structure templates.

7. The system for realizing object-oriented medical software page based on feature engineering according to claim 1, characterized in that: The hospital information management program preprocessing in S1 process includes the following steps: a. Encapsulate the opening and closing methods of the hospital information management system; b. Write the login account information of the hospital information management system into the configuration file; c. Annotate the main functional points and features of the hospital information management software and save them in csv file; d. Preset a binary classification convolutional neural network; f. Use the labeled csv files to train multiple binary classification convolutional neural networks, and import the trained neural networks into the feature library; e. Build the page graph data structure and initialize it.

8. The system for realizing object-oriented medical software page based on feature engineering according to claim 1, characterized in that: Generate a page object program using the already generated page logic diagram and feature library.

Images