Clinical test data storage device and storage method
By adopting a combination scheme of user-side, data management platform, data acquisition module, data analysis module and storage management module in clinical trial data storage devices, using RNN and random forest learning models, the complex management and data sharing efficiency of traditional storage devices are solved. The problem of low results enables efficient and safe management and sharing of clinical trial data.
Patent Information
- Application Number
- CN202410983720.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The management methods of traditional clinical trial data storage equipment are complex, the data sharing efficiency is low, and the integrity of the data cannot be effectively guaranteed, which affects the work efficiency of medical staff.
Using a combination of user-side, data management platform, data acquisition module, data analysis module and storage management module, using a hybrid storage method of local database and cloud database, data proofreading and analysis are carried out through RNN algorithm model and random forest learning model to generate Independent encoding to achieve efficient sharing of data.
It realizes flexible optimization of data management methods, improves data security and sharing efficiency, and ensures the integrity of clinical trial data and high-quality data sharing.
Smart Images

Figure CN118942599A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of clinical trial management, and in particular to a clinical trial data storage device and a storage method. Background Art
[0002] Clinical trial data refers to all information collected during the clinical trial about the subjects, trial design, implementation process and results. These data are essential for evaluating the safety and effectiveness of new drugs or treatments. The management and storage methods of clinical trial data directly affect the accuracy and reliability of the trial results. Therefore, it is crucial to choose an appropriate storage method. The traditional way of storing clinical trial data is paper storage, that is, all data are printed into paper documents and stored in folders or file cabinets in a certain order and classification. The advantage is that it is simple and easy to use, but the disadvantage is that it is easy to lose, damage, and inconvenient to retrieve and share. With the development of information technology, more and more clinical trial data are stored electronically. Electronic storage can be divided into local storage and cloud storage. Local storage stores clinical trial data on a local server or computer hard disk. The advantage is that the data security is high, but the disadvantage is that regular backup is required to prevent data loss. In addition, the data capacity of local storage is limited and may not meet the storage needs of a large amount of clinical trial data. Cloud storage stores clinical trial data on a cloud server and can be accessed and shared through the Internet. The advantages of cloud storage are large data capacity and easy retrieval and sharing, but the disadvantage is that the data security is relatively low and there may be a risk of data leakage. The storage method of clinical trial data needs to be comprehensively considered based on factors such as the scale of the trial, data volume, and safety requirements to ensure the security, integrity, and reliability of the clinical trial data.
[0003] At present, traditional clinical trial data storage devices use local databases and cloud databases for hybrid storage. Different types of data require different storage methods, and the management methods are complicated. It is easy to have problems such as medical record errors and sample confusion. During the data sharing process, the integrity of the clinical trial data cannot be effectively guaranteed, which seriously affects the work efficiency of medical staff. Summary of the invention
[0004] 1. Technical issues to be resolved
[0005] In view of the shortcomings of the prior art, the present invention provides a clinical trial data storage device and storage method, which has the advantages of flexible and optimized data management mode with high security, high data sharing efficiency and high quality, and solves the problems of complex management mode of traditional clinical trial data storage equipment, low data sharing efficiency and poor integrity.
[0006] (II) Technical solution
[0007] To achieve the above object, the present invention provides the following technical solutions:
[0008] In a first aspect, a clinical trial data storage device is provided, comprising:
[0009] User terminal, data management platform, data collection module, data analysis module and storage management module;
[0010] The user terminal is used to upload and receive data;
[0011] The data management platform is used to store data and obtain historical data; the data management platform is provided with a local database and a cloud database; the local database caches clinical trial data for a fixed number of days, and each time the clinical trial data is cached, the local database will clear the historical cached clinical trial data; the cloud database extracts key words from the clinical trial data to permanently store the clinical trial data, and the key words include trial objectives, trial operations and clinical indicators; the user terminal and the data management platform are interconnected through a network;
[0012] The data acquisition module includes a data entry unit, a local data unit and a cloud data unit; the data entry unit collects an entry data set by connecting to a user terminal through a network, and the entry data set includes clinical trial data uploaded by all medical personnel through the user terminal and the time point of successful upload; the local data unit collects a local data set by connecting to a local database through a network, and the local data set includes clinical trial data cached in the local database according to a fixed number of days; the cloud data unit collects a cloud data set by connecting to a cloud database through a network, and the cloud data set includes clinical trial data permanently stored in the cloud database according to key words; the data acquisition module transmits the entry data set, the local data set and the cloud data set to the data analysis module through the network;
[0013] The data analysis module includes a content proofreading unit and an association analysis unit; the content proofreading unit is provided with an RNN algorithm model, the content proofreading unit inputs the input data set into the RNN algorithm model, combines the context calculation to generate a proofreading data set Jdsj, and transmits it to the user end or the association analysis unit through the network; the association analysis unit is provided with a random forest learning model RF, the association analysis unit substitutes the local data set and the cloud data set into the random forest learning model RF for training, and marks the trained random forest learning model as RF + The association analysis unit substitutes the input data set into the trained random forest learning model RF + , calculate and generate a standard data set Bzsj; the data analysis module combines the standard data set Bzsj and the trained random forest learning model RF through the network + Transmitting to a storage management module;
[0014] The storage management module generates corresponding independent codes in sequence according to all decision trees in the standard data group Bzsj, and synchronously transmits the independent codes to the user end and updates them to the local database and the cloud database, forming a data sharing closed loop.
[0015] Further: the expression of the input data set is to It corresponds to each clinical trial data uploaded by medical personnel through the user terminal and the time point of successful upload. t represents the time point when the user terminal successfully uploads the clinical trial data. The digital numbers 1 to n represent that there are n clinical trial data successfully uploaded by medical personnel through the user terminal. Each clinical trial data includes text information, image information, voice information and video information.
[0016] Further: the expression of the local data set is to They correspond to the clinical trial data cached in the local database according to a fixed number of days, the time point of successful caching, and the number of data cached successfully. bt represents the time point when the local database caches the clinical trial data successfully according to a fixed number of days. s represents the number of data cached successfully in the local database according to a fixed number of days. The digital numbers 1 to m represent that there are m copies of clinical trial data successfully cached in the cloud database according to a fixed number of days.
[0017] Further: the expression of the cloud data set is to They correspond in sequence to the clinical trial data permanently stored in each cloud database according to the key words and the time point of successful storage. yt represents the time point when the cloud database successfully permanently stores the clinical trial data according to the key words. g represents the key words included in the clinical trial data in the cloud database. The digital numbers 1 to z represent that there are z copies of clinical trial data permanently stored in the cloud database according to the key words.
[0018] Further: the calculation formula of the proofreading data set Jdsj is as follows:
[0019]
[0020] In the formula, Jdsj represents the proofreading data group, It means that the RNN algorithm model proofreads each clinical trial data uploaded by medical personnel through the user end in chronological order from early to late, and screens out erroneous data in the clinical trial data based on the context.
[0021] Further: when there is data in the proofreading data group Jdsj, it indicates that there is erroneous data in the clinical trial materials uploaded by the medical personnel through the user terminal. The content proofreading unit transmits the proofreading data group Jdsj to the user terminal through the network, waiting for the medical personnel to modify it into correct clinical trial materials through the user terminal. The user terminal transmits the modified correct clinical trial materials to the data entry unit, and the data entry unit adds the correct clinical trial materials to the input data set until there is no data in the proofreading data group Jdsj. When there is no data in the proofreading data group Jdsj, it indicates that there is erroneous data that does not exist in the clinical trial materials uploaded by the medical personnel through the user terminal. The content proofreading unit transmits the input data set to the association analysis unit through the network.
[0022] Further: The random forest learning model RF + The expression is as follows:
[0023]
[0024] In the formula, RF + Represents the random forest learning model after training. represents the i-th permanently stored clinical trial data in the cloud dataset, α represents the data weight when the random forest learning model RF is trained on the cloud dataset, It means that the z copies of permanently stored clinical trial data in the cloud dataset are distributed according to the α weight. represents the clinical trial data cached on the kth day in the local dataset, β represents the data weight when the random forest learning model RF is trained on the local dataset, Indicates that the m clinical trial data cached for a fixed number of days in the local dataset are all distributed according to the β weight, It means that the random forest learning model RF calculates the weighted average according to the α weight and the β weight, and distributes the cloud data set and the local data set for training and learning according to the weighted average.
[0025] Further: The calculation formula of the standard data set Bzsj is as follows:
[0026]
[0027] In the formula, Bzsj represents the standard data set, It indicates the clinical trial data uploaded by the jth medical staff through the user terminal in the input data set. Represents the random forest learning model RF after training + At the same time, n sets of clinical trial data uploaded by medical personnel through the user terminal are collected to build multiple groups of decision trees containing the same clinical trial category.
[0028] Further: The random forest learning model RF + For each decision tree in the standard data set Bzsj, predictions are made independently, and a single independent code is generated. The clinical trial data obtained with the independent code does not overlap with any data in the local data set and the cloud data set.
[0029] The second party offers a method for storing clinical trial data, including:
[0030] Step 1: The data collection module arranges the initial clinical trial data uploaded by the user end from the earliest to the latest time point according to the successful upload time to form an input data set, and transmits it to the data analysis module through the network;
[0031] Step 2: The data analysis module is provided with an RNN algorithm model, which combines the context of the input data set to calculate and generate the proofreading data set Jdsj; if there is data in the proofreading data set Jdsj, it is transmitted to the user end through the network to wait for modification; if there is no data in the proofreading data set Jdsj, the association analysis is continued according to the input data set;
[0032] Step 3: The data analysis module is equipped with a random forest learning model RF. The clinical trial data cached in the local database for a fixed number of days and the clinical trial data permanently stored in the cloud database according to the key words are substituted into the random forest learning model RF for training, and the trained random forest learning model is marked as RF. + ;
[0033] Step 4: The data analysis module substitutes the input data set into the trained random forest learning model RF + , calculate and generate the standard data set Bzsj, and construct multiple groups of decision trees containing the same clinical trial categories;
[0034] Step 5: The storage management module uses the random forest learning model RF + For each group of decision trees in the standard data set Bzsj, predictions are made independently, and a single independent code is generated and transmitted to the user end, and then updated to the local database and the cloud database to complete the data sharing process.
[0035] Compared with the prior art, the present invention provides a clinical trial data storage device and storage method, which has the following beneficial effects:
[0036] 1. The present invention uploads and receives data through the user end, and the data management platform is used to store data and obtain historical data. The user end and the data management platform are interconnected through a network. The data management platform is provided with a local database and a cloud database. The local database caches clinical trial data according to a fixed number of days, and each time the clinical trial data is cached, the local database will clear the historical cached clinical trial data. The cloud database extracts key words from the clinical trial data to permanently store the clinical trial data. The key words include test objectives, test operations and clinical indicators. The data acquisition module classifies and collects the input data set, the local data set and the cloud data set and transmits them to Data analysis module: The data analysis module is provided with an RNN algorithm model, which combines the context of the input data set to calculate and generate the proofreading data set Jdsj. If there is data in the proofreading data set Jdsj, it is transmitted to the user end through the network for modification. If there is no data in the proofreading data set Jdsj, the association analysis is continued according to the input data set. The data analysis module is provided with a random forest learning model RF, which substitutes the clinical trial data cached in the local database for a fixed number of days and the clinical trial data permanently stored in the cloud database according to the key words into the random forest learning model RF for training, and marks the trained random forest learning model as RF + , compatible with local databases and cloud databases to the greatest extent, and flexibly optimizes data management methods with high security.
[0037] 2. The present invention substitutes the input data set into the trained random forest learning model RF through the data analysis module + , calculate and generate the standard data set Bzsj, build multiple groups of decision trees containing the same clinical trial category, and the storage management module uses the random forest learning model RF + For each decision tree in the standard data set Bzsj, predictions are made independently. The predictions generate a single independent code that is transmitted to the user end and updated to the local database and cloud database to complete the data sharing process with high efficiency and quality. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Figure 1 A schematic diagram of the structure of a clinical trial data storage device provided by an embodiment of the present invention;
[0039] Figure 2 A schematic diagram of a flow chart of a clinical trial data storage method provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0040] 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.
[0041] Since traditional clinical trial data storage devices use local databases and cloud databases for hybrid storage, different types of data require different storage methods, and the management methods are complicated. It is easy to have problems such as medical record errors and sample confusion. During the data sharing process, the integrity of the clinical trial data cannot be effectively guaranteed, which seriously affects the work efficiency of medical staff. Therefore, a clinical trial data storage device and storage method are provided.
[0042] First, see Figure 1 , a clinical trial data storage device, the device comprising:
[0043] The user end and the data management platform. The user end is used to upload and receive data. The user end can be specifically divided into a mobile end and a web end, which is convenient for medical personnel to upload and receive data more conveniently. The data management platform is used to store data and obtain historical data, store newly uploaded clinical trial data and obtain independent codes that have formed decision trees. The user end and the data management platform are interconnected through the network. The clinical trial data storage device includes a data acquisition module, a data analysis module and a storage management module;
[0044] The data management platform is equipped with a local database and a cloud database. The local database caches clinical trial data for a fixed number of days, and each time the clinical trial data is cached, the local database will clear the historical cached clinical trial data. The local database has limited memory capacity and is mainly used to back up some data. When the data is subsequently queried, the response speed and query efficiency are high, meeting the needs of medical personnel to quickly query recent clinical trial data. The cloud database extracts key words from the clinical trial data to permanently store the clinical trial data. The cloud database has a large memory capacity, and permanent storage and non-tamperability are the key to ensuring data integrity. Key words include trial objectives, trial operations, and clinical indicators.
[0045] The data collection module includes a data entry unit, a local data unit, and a cloud data unit. The data entry unit collects and enters data sets through a network connection to the user end. The data set includes all clinical trial data uploaded by medical personnel through the user end and the time point of successful upload. The expression of the data set is to Each clinical trial data uploaded by medical personnel through the user end and the time point of successful upload are the initial data. The data needs to be proofread and modified according to the initial data later. t represents the time point when the user end successfully uploads the clinical trial data. The digital numbers 1 to n represent that there are n clinical trial data successfully uploaded by medical personnel through the user end. Each clinical trial data includes text information, image information, voice information and video information. The clinical trial data has continuity. The clinical trial data after the first storage has an independent code. Subsequent medical personnel can directly add new content through the independent code without having to re-query and find the data and then add it.
[0046] The local data unit collects the local data set through the network connection to the local database. The local data set includes the clinical trial data cached in the local database according to a fixed number of days. The expression of the local data set is: to It corresponds to the clinical trial data cached in the local database according to a fixed number of days, the time point of successful caching, and the number of data cached successfully. For example, it can be set to update the local database every day, every month, or every year according to the actual test needs. bt represents the time point when the local database caches the clinical trial data successfully according to a fixed number of days, s represents the number of data cached successfully according to a fixed number of days in the local database, and the digital numbers 1 to m represent that there are m copies of clinical trial data cached successfully according to a fixed number of days in the cloud database.
[0047] The cloud data unit collects cloud data sets through a network connection to the cloud database. The cloud data sets include clinical trial data permanently stored in the cloud database according to key words. The expression of the cloud data set is: to Corresponding to each copy of the clinical trial data permanently stored in the cloud database according to the key words and the time point of successful storage, yt represents the time point when the cloud database successfully permanently stores the clinical trial data according to the key words, g represents the key words included in the clinical trial data in the cloud database, and the digital numbers 1 to z represent that there are Z copies of the clinical trial data permanently stored in the cloud database according to the key words;
[0048] The data collection module transmits the input data set, the local data set and the cloud data set to the data analysis module through the network;
[0049] The data analysis module includes a content proofreading unit and a correlation analysis unit. The content proofreading unit is provided with an RNN algorithm model. The content proofreading unit inputs the input data set into the RNN algorithm model and generates a proofreading data group Jdsj by combining the context. The calculation formula is as follows:
[0050]
[0051] In the formula, Jdsj represents the proofreading data group, It means that the RNN algorithm model proofreads each clinical trial data uploaded by medical personnel through the user end in chronological order from early to late, screens out erroneous data in the clinical trial data based on the contextual relationship, and proofreads the initial data to avoid problems such as medical record errors and sample confusion;
[0052] When there is data in the proofreading data group Jdsj, it indicates that there is erroneous data in the clinical trial materials uploaded by the medical personnel through the user terminal, for example, repeated uploaded medical record content or confused sample labels. The content proofreading unit transmits the proofreading data group Jdsj to the user terminal through the network, waiting for the medical personnel to modify it to the correct clinical trial materials through the user terminal. The user terminal transmits the modified correct clinical trial materials to the data entry unit. The data entry unit adds the correct clinical trial materials to the input data set until there is no data in the proofreading data group Jdsj. When there is no data in the proofreading data group Jdsj, for example, updated content related to medical records or clearly labeled sample labels, it indicates that there is erroneous data that does not exist in the clinical trial materials uploaded by the medical personnel through the user terminal. The content proofreading unit transmits the input data set to the association analysis unit through the network.
[0053] The association analysis unit is provided with a random forest learning model RF. The association analysis unit substitutes the local data set and the cloud data set into the random forest learning model RF for training, and marks the trained random forest learning model as RF. + , which is expressed as follows:
[0054]
[0055] In the formula, RF + Represents the random forest learning model after training. represents the i-th permanently stored clinical trial data in the cloud dataset, α represents the data weight when the random forest learning model RF is trained on the cloud dataset, It means that the z copies of permanently stored clinical trial data in the cloud dataset are distributed according to the α weight. represents the clinical trial data cached on the kth day in the local dataset, β represents the data weight when the random forest learning model RF is trained on the local dataset, Indicates that the m clinical trial data cached for a fixed number of days in the local dataset are all distributed according to the β weight, It means that the random forest learning model RF calculates the weighted average according to the α weight and the β weight, and distributes the cloud data set and the local data set for training and learning according to the weighted average. It is compatible with the cloud database and the local database to the greatest extent, and the data management method is flexible and optimized with high security.
[0056] The association analysis unit substitutes the input data set into the trained random forest learning model RF + , calculate and generate the standard data set Bzsj, the calculation formula is as follows:
[0057]
[0058] In the formula, Bzsj represents the standard data set, It indicates the clinical trial data uploaded by the jth medical staff through the user terminal in the input data set. Represents the random forest learning model RF after training + At the same time, n sets of clinical trial data uploaded by medical personnel through the user end are collected to build multiple groups of decision trees containing the same clinical trial category. Each group of decision trees corresponds to an independent clinical trial project.
[0059] The data analysis module combines the standard data set Bzsj and the trained random forest learning model RF through the network + Transmitting to a storage management module;
[0060] The storage management module generates corresponding independent codes according to all decision trees in the standard data set Bzsj, and the random forest learning model RF + For each decision tree in the standard data set Bzsj, predictions are made independently, and a single independent code is generated. The clinical trial data with the independent code does not overlap with any data in the local data set and the cloud data set. The independent code is synchronously transmitted to the user end and updated to the local database and the cloud database, forming a closed loop of data sharing with high efficiency and quality.
[0061] Second, see Figure 2 , provides a clinical trial data storage method, the method comprising:
[0062] Step 1: Medical personnel upload the initial clinical trial data through the user terminal. The data collection module arranges the data from the earliest to the latest time points according to the successful upload, and transmits it to the data analysis module through the network;
[0063] Step 2: The data analysis module is provided with an RNN algorithm model, which combines the context of the input data set to generate the proofreading data set Jdsj. If there is data in the proofreading data set Jdsj, it is transmitted to the user end through the network to wait for modification. If there is no data in the proofreading data set Jdsj, the association analysis is continued according to the input data set.
[0064] Step 3: The data analysis module is equipped with a random forest learning model RF. The clinical trial data cached in the local database for a fixed number of days and the clinical trial data permanently stored in the cloud database according to the key words are substituted into the random forest learning model RF for training, and the trained random forest learning model is marked as RF. + , it is compatible with local databases and cloud databases to the greatest extent, and it can flexibly optimize data management methods with high security;
[0065] Step 4: The data analysis module substitutes the input data set into the trained random forest learning model RF + , calculate and generate the standard data set Bzsj, and construct multiple groups of decision trees containing the same clinical trial categories;
[0066] Step 5: The storage management module uses the random forest learning model RF + For each decision tree in the standard data set BzSj, predictions are made independently. The predictions generate a single independent code that is transmitted to the user end and updated to the local database and cloud database to complete the data sharing process with high efficiency and quality.
[0067] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A clinical trial data storage device, characterized in that: include: User terminal, data management platform, data collection module, data analysis module and storage management module; The user terminal is used to upload and receive data; The data management platform is used to store data and obtain historical data; the data management platform is provided with a local database and a cloud database; the local database caches clinical trial data for a fixed number of days, and each time the clinical trial data is cached, the local database will clear the historical cached clinical trial data; the cloud database extracts key words from the clinical trial data to permanently store the clinical trial data, and the key words include trial objectives, trial operations and clinical indicators; the user terminal and the data management platform are interconnected through a network; The data acquisition module includes a data entry unit, a local data unit and a cloud data unit; the data entry unit collects an entry data set by connecting to a user terminal through a network, and the entry data set includes clinical trial data uploaded by all medical personnel through the user terminal and the time point of successful upload; the local data unit collects a local data set by connecting to a local database through a network, and the local data set includes clinical trial data cached in the local database according to a fixed number of days; the cloud data unit collects a cloud data set by connecting to a cloud database through a network, and the cloud data set includes clinical trial data permanently stored in the cloud database according to key words; the data acquisition module transmits the entry data set, the local data set and the cloud data set to the data analysis module through the network; The data analysis module includes a content proofreading unit and an association analysis unit; the content proofreading unit is provided with an RNN algorithm model, the content proofreading unit inputs the input data set into the RNN algorithm model, combines the context calculation to generate a proofreading data set Jdsj, and transmits it to the user end or the association analysis unit through the network; the association analysis unit is provided with a random forest learning model RF, the association analysis unit substitutes the local data set and the cloud data set into the random forest learning model RF for training, and marks the trained random forest learning model as RF + The association analysis unit substitutes the input data set into the trained random forest learning model RF + , calculate and generate a standard data set Bzsj; the data analysis module combines the standard data set Bzsj and the trained random forest learning model RF through the network + Transmitting to a storage management module; The storage management module generates corresponding independent codes in sequence according to all decision trees in the standard data group Bzsj, and synchronously transmits the independent codes to the user end and updates them to the local database and the cloud database, forming a data sharing closed loop.
2. A clinical trial data storage device according to claim 1, characterized in that: The expression of the input data set is to It corresponds to each clinical trial data uploaded by medical personnel through the user terminal and the time point of successful upload. t represents the time point when the user terminal successfully uploads the clinical trial data. The digital numbers 1 to n represent that there are n clinical trial data successfully uploaded by medical personnel through the user terminal. Each clinical trial data includes text information, image information, voice information and video information.
3. A clinical trial data storage device according to claim 2, characterized in that: The expression of the local data set is: to They correspond to the clinical trial data cached in the local database according to a fixed number of days, the time point of successful caching, and the number of data cached successfully. bt represents the time point when the local database caches the clinical trial data successfully according to a fixed number of days. s represents the number of data cached successfully in the local database according to a fixed number of days. The digital numbers 1 to m represent that there are m copies of clinical trial data successfully cached in the cloud database according to a fixed number of days.
4. A clinical trial data storage device according to claim 3, characterized in that: The expression of the cloud dataset is: to They correspond in sequence to the clinical trial data permanently stored in each cloud database according to the key words and the time point of successful storage. yt represents the time point when the cloud database successfully permanently stores the clinical trial data according to the key words. g represents the key words included in the clinical trial data in the cloud database. The digital numbers 1 to z represent that there are z copies of clinical trial data permanently stored in the cloud database according to the key words.
5. A clinical trial data storage device according to claim 4, characterized in that: The calculation formula of the proofreading data set Jdsj is as follows: In the formula, Jdsj represents the proofreading data group, It means that the RNN algorithm model proofreads each clinical trial data uploaded by medical personnel through the user end in chronological order from early to late, and screens out erroneous data in the clinical trial data based on the context.
6. A clinical trial data storage device according to claim 5, characterized in that: When there is data in the proofreading data group Jdsj, it indicates that there is erroneous data in the clinical trial materials uploaded by the medical personnel through the user terminal. The content proofreading unit transmits the proofreading data group Jdsj to the user terminal through the network, waiting for the medical personnel to modify it into correct clinical trial materials through the user terminal. The user terminal transmits the modified correct clinical trial materials to the data entry unit, and the data entry unit adds the correct clinical trial materials to the entry data set until there is no data in the proofreading data group Jdsj. When there is no data in the proofreading data group Jdsj, it indicates that there is erroneous data in the clinical trial materials uploaded by the medical personnel through the user terminal. The content proofreading unit transmits the entry data set to the association analysis unit through the network.
7. A clinical trial data storage device according to claim 6, characterized in that: The random forest learning model RF + The expression is as follows: In the formula, RF + Represents the random forest learning model after training. represents the i-th permanently stored clinical trial data in the cloud dataset, α represents the data weight when the random forest learning model RF is trained on the cloud dataset, It means that the z copies of permanently stored clinical trial data in the cloud dataset are distributed according to the α weight. represents the clinical trial data cached on the kth day in the local dataset, β represents the data weight when the random forest learning model RF is trained on the local dataset, Indicates that the m clinical trial data cached for a fixed number of days in the local dataset are all distributed according to the β weight, It means that the random forest learning model RF calculates the weighted average according to the α weight and the β weight, and distributes the cloud data set and the local data set for training and learning according to the weighted average.
8. A clinical trial data storage device according to claim 7, characterized in that: The calculation formula of the standard data set Bzsj is as follows: In the formula, Bzsj represents the standard data set, represents the clinical trial data uploaded by the jth medical staff through the user terminal in the input data set. Represents the random forest learning model RF after training + At the same time, n sets of clinical trial data uploaded by medical personnel through the user terminal are collected to build multiple groups of decision trees containing the same clinical trial category.
9. A clinical trial data storage device according to claim 8, characterized in that: The random forest learning model RF + For each decision tree in the standard data set Bzsj, predictions are made independently, and a single independent code is generated. The clinical trial data obtained with the independent code does not overlap with any data in the local data set and the cloud data set.
10. A clinical trial data storage method, characterized in that: include: Step 1: The data collection module arranges the initial clinical trial data uploaded by the user end from the earliest to the latest time point according to the successful upload time to form an input data set, and transmits it to the data analysis module through the network; Step 2: The data analysis module is provided with an RNN algorithm model, which combines the context of the input data set to calculate and generate the proofreading data set Jdsj; if there is data in the proofreading data set Jdsj, it is transmitted to the user end through the network to wait for modification; if there is no data in the proofreading data set Jdsj, the association analysis is continued according to the input data set; Step 3: The data analysis module is equipped with a random forest learning model RF. The clinical trial data cached in the local database for a fixed number of days and the clinical trial data permanently stored in the cloud database according to the key words are substituted into the random forest learning model RF for training, and the trained random forest learning model is marked as RF. + ; Step 4: The data analysis module substitutes the input data set into the trained random forest learning model RF + , calculate and generate the standard data set Bzsj, and construct multiple groups of decision trees containing the same clinical trial categories; Step 5: The storage management module uses the random forest learning model RF + For each group of decision trees in the standard data set Bzsj, predictions are made independently, and a single independent code is generated and transmitted to the user end, and then updated to the local database and the cloud database to complete the data sharing process.
