Artificial intelligence-based anorectal patient information sharing method and system

Abstract

According to the anorectal patient information sharing method and system based on artificial intelligence, the anorectal patient information and the anorectal patient information database are obtained, and the anorectal patient information and the multiple pieces of historical anorectal patient information are subjected to feature extraction based on the first convolution thread; the first convolution thread is obtained by training based on the description content of the target anorectal patient, and the second convolution thread is obtained by training based on the description content of the target anorectal patient; the cut data set of the number of the patient information is a data set obtained by cutting the number of the patient information on the anorectal information of the main example in the data set description direction; calculating a generality coefficient between the anorectal patient information feature and each historical anorectal patient information feature; according to the data sharing method and device, the patient interference data can be deleted, and it can be guaranteed that the data during sharing is accurate and reliable.

Description

Technical Field

[0001] This application relates to the technical field of data sharing. Specifically, it relates to a method and system for sharing anorectal patient information based on artificial intelligence. Background Art

[0002] The number of patient information in hospitals is extremely large. Patients need to show various reports to medical staff or medical staff need to access patient data through computers. However, when accessing information through computers, there are some problems, such as: data may be lost or uploaded incorrectly. In this way, the accuracy of the shared data is very poor, seriously affecting the judgment of medical staff. Therefore, there is an urgent need for a technical method to improve the above technical problems. Summary of the Invention

[0003] To improve the technical problems existing in the related art, this application provides a method and system for sharing anorectal patient information based on artificial intelligence.

[0004] In the first aspect, a method for sharing anorectal patient information based on artificial intelligence is provided. 1. A method for sharing anorectal patient information based on artificial intelligence, characterized in that the method includes:

[0005] Obtain anorectal patient information and an anorectal patient information database, where the anorectal patient information database includes a number of historical anorectal patient information;

[0006] Based on the first convolutional thread, perform feature extraction on the anorectal patient information and the number of historical anorectal patient information to obtain anorectal patient information features and a number of historical anorectal patient information features. The first convolutional thread is trained based on the target anorectal patient description content, and the target anorectal patient description content includes the first anorectal patient description content and the second anorectal patient description content. The first anorectal patient description content is a comparison result calculated based on the derivative data set of the main example anorectal information and the data set features of the specified potential example anorectal information. The second anorectal patient description content is the common coefficient distribution anorectal patient description content calculated based on the data set features of the main example anorectal information and the cut data set of the number of a number of patient information. The cut data set of the number of a number of patient information is a data set obtained by cutting the main example anorectal information in the data set description direction by the number of a number of patient information;

[0007] Calculate the common coefficient between the anorectal patient information feature and each of the historical anorectal patient information features;

[0008] Perform data sharing processing in combination with the common coefficient.

[0009] In this application, the training process of the first convolutional thread includes the following steps:

[0010] Obtain a number of first main example anorectal information and a number of specified potential example anorectal information, and perform dataset derivation on each of the first main example anorectal information to obtain a number of derived datasets, where the derived datasets include the first main example anorectal information and the datasets obtained by performing dataset derivation on the first main example anorectal information;

[0011] Based on the first convolutional thread, perform feature extraction on the derived datasets to obtain a splicing result, and based on the second convolutional thread, perform feature extraction on the number of specified potential example anorectal information to obtain a number of potential example anorectal information features;

[0012] Calculate a comparison result based on the splicing result and the potential example anorectal information features to obtain a first anorectal patient description content;

[0013] Perform shearing of the number of patient information in the dataset description direction on each of the first main example anorectal information to obtain a number of sheared datasets, and based on the first convolutional thread, perform feature extraction on the first main example anorectal information and the number of sheared datasets to obtain first main example anorectal information features and a number of sheared dataset features;

[0014] Calculate a common coefficient distribution anorectal patient description content based on the first main example anorectal information features, the number of sheared dataset features, and the common coefficient distribution relationship between the number of sheared datasets and the first main example anorectal information to obtain a second anorectal patient description content;

[0015] Optimize the coefficients of the first convolutional thread by combining the first anorectal patient description content and the second anorectal patient description content.

[0016] In this application, the method further includes:

[0017] Obtain a number of second main example anorectal information and the classification directory corresponding to each of the second main example anorectal information;

[0018] Based on the first convolutional thread, perform feature extraction on the second main example anorectal information to obtain second main example anorectal information features;

[0019] Classify the second main example anorectal information features based on a specified classification unit to obtain a classification prediction possibility value;

[0020] Calculate a classification anorectal patient description content based on the classification prediction possibility value and the classification directory to obtain a third anorectal patient description content;

[0021] Optimizing the coefficients of the first convolutional thread by combining the description content of the first anorectal patient and the description content of the second anorectal patient includes:

[0022] Optimizing the coefficients of the first convolutional thread by combining the description content of the first anorectal patient, the description content of the second anorectal patient, and the description content of the third anorectal patient.

[0023] In this application, deriving a dataset for each of the first major example anorectal information to obtain a number of derived datasets includes:

[0024] Deriving patient attributes from the first major example anorectal information to obtain a first local derived dataset;

[0025] Deriving treatment information from the first major example anorectal information to obtain a second local derived dataset;

[0026] Deriving treatment information from the first local derived dataset to obtain a third local derived dataset;

[0027] Determining a number of derived datasets by combining the first local derived dataset, the second local derived dataset, the third local derived dataset, and the first major example anorectal information.

[0028] In this application, before deriving patient attributes from the first major example anorectal information to obtain a first local derived dataset, it further includes:

[0029] Dividing the first major example anorectal information into a number of dataset segments, and extracting anorectal patient element information from each of the dataset segments to obtain a number of anorectal patient element information;

[0030] Deriving patient attributes from the first major example anorectal information to obtain a first local derived dataset includes:

[0031] Deriving patient attributes from the number of anorectal patient element information to obtain a number of first local derived anorectal patient element information;

[0032] Extracting features from the derived dataset based on the first convolutional thread to obtain a splicing result includes:

[0033] Performing feature extraction on a number of derivative anorectal patient element information based on a first convolutional thread to obtain a splicing result, where the number of derivative anorectal patient element information includes the number of anorectal patient element information, the number of first local derivative anorectal patient element information derived only based on patient attributes, the number of second local derivative anorectal patient element information derived only based on treatment information, and the number of third local derivative anorectal patient element information derived based on both patient attributes and treatment information.

[0034] In this application, the first convolutional thread includes a first unit and a second unit. The performing feature extraction on a number of derivative anorectal patient element information based on the first convolutional thread to obtain a splicing result includes:

[0035] Performing feature extraction on a number of derivative anorectal patient element information according to the first unit to obtain a number of feature points;

[0036] Performing feature splicing on the number of feature points according to the second unit to obtain a splicing result.

[0037] In this application, the performing feature splicing on the number of feature points according to the second unit to obtain a splicing result includes:

[0038] Based on performing clustering processing on the number of feature points, obtaining a first transition queue, a second transition queue, and a third transition queue;

[0039] Performing dimensionless simplification processing on the first transition queue and the second transition queue, and performing function processing on the dimensionless simplification processing result and the third transition queue to obtain a function processing result;

[0040] Calculating the depolarization result of the function processing result to obtain a depolarization processing result.

[0041] In this application, the performing feature extraction on the anorectal patient information and the number of historical anorectal patient information based on the first convolutional thread to obtain anorectal patient information features and a number of historical anorectal patient information features includes:

[0042] Extracting a number of anorectal patient information attributes from the anorectal patient information, and extracting a number of historical anorectal patient information attributes from each of the historical anorectal patient information;

[0043] Performing feature extraction on the number of anorectal patient information attributes according to the first unit to obtain a number of query feature points, and performing feature extraction on the number of historical anorectal patient information attributes according to the first unit to obtain a number of candidate feature points;

[0044] Performing feature splicing on the several query feature points according to the second unit to obtain the anorectal patient information features, and performing feature splicing on the several candidate feature points corresponding to each historical anorectal patient information according to the second unit to obtain several historical anorectal patient information features.

[0045] In this application, extracting several anorectal patient information attributes from the anorectal patient information, and extracting several historical anorectal patient information attributes from each of the historical anorectal patient information, including:

[0046] Dividing the anorectal patient information into several anorectal patient information segments, and dividing each of the historical anorectal patient information into several historical anorectal patient information segments;

[0047] Extracting one anorectal patient information attribute from each of the anorectal patient information segments to obtain several anorectal patient information attributes;

[0048] Extracting one historical anorectal patient information attribute from each of the historical anorectal patient information segments to obtain several historical anorectal patient information attributes corresponding to each of the historical anorectal patient information.

[0049] In this application, after optimizing the coefficients of the first convolutional thread by combining the first anorectal patient description content and the second anorectal patient description content, it further includes:

[0050] Obtaining the optimization coefficient of the thread coefficient;

[0051] Optimizing the coefficients of the second convolutional thread according to the optimized coefficients of the first convolutional thread and the optimization coefficient.

[0052] In this application, the combined common coefficient for data sharing processing includes:

[0053] Determining the data set in the several historical anorectal patient information whose common coefficient with the anorectal patient information is less than the specified target value as the first target data set;

[0054] Distributing the several first target data sets in descending order of the common coefficient with the anorectal patient information, and determining the specified number of data sets in the front as the second target data set;

[0055] Performing sharing processing on the several second target data.

[0056] In a second aspect, a sharing system for anorectal patient information based on artificial intelligence is provided, including a processor and a memory that communicate with each other, and the processor is configured to read and execute a computer program from the memory to implement the above method.

[0057] The method and system for sharing anorectal patient information based on artificial intelligence provided by the embodiments of the present application obtain anorectal patient information and an anorectal patient information database, where the anorectal patient information database includes a number of historical anorectal patient information; perform feature extraction on the anorectal patient information and a number of historical anorectal patient information based on a first convolutional thread to obtain anorectal patient information features and a number of historical anorectal patient information features. The first convolutional thread is trained based on target anorectal patient description content, and the target anorectal patient description content includes first anorectal patient description content and second anorectal patient description content. The first anorectal patient description content is a comparison result calculated based on the derivative data set of the main example anorectal information and the data set features of the specified potential example anorectal information. The second anorectal patient description content is the common coefficient distribution anorectal patient description content calculated based on the data set features of the main example anorectal information and the clipped data set of the number of a number of patient information. The clipped data set of the number of a number of patient information is a data set obtained by clipping the main example anorectal information in the data set description direction by the number of a number of patient information; calculate the common coefficient between the anorectal patient information features and each historical anorectal patient information feature; perform data sharing processing according to the common coefficient. In the present application, patient interference data can be deleted, and the data during sharing can be ensured to be accurate and reliable. Description of the Drawings

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0059] Figure 1 It is a flowchart of a method for sharing anorectal patient information based on artificial intelligence provided by the embodiments of the present application. Detailed Embodiments

[0060] In order to better understand the above technical solutions, the technical solutions of the present application will be described in detail below through the drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solutions of the present application, rather than limitations on the technical solutions of the present application. Without conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0061] Please refer to Figure 1 , which shows a method for sharing anorectal patient information based on artificial intelligence. The method may include the technical solutions described in the following steps 310-step 340.

[0062] Step 310: Obtain anorectal patient information and an anorectal patient information database.

[0063] Exemplarily, the anorectal patient information is data obtained from the current detection of a patient in a hospital, and the anorectal patient information database can be understood as sample data, which means a database composed of empirical results obtained from the detection of historical patients.

[0064] Step 320: Based on a first convolutional thread, perform feature extraction on the anorectal patient information and a number of historical anorectal patient information to obtain anorectal patient information features and a number of historical anorectal patient information features.

[0065] In a possible embodiment, the first convolutional thread can be an artificial intelligence thread for performing feature extraction on a data set. The first convolutional thread is trained based on target anorectal patient description content, and the target anorectal patient description content includes first anorectal patient description content and second anorectal patient description content. The first anorectal patient description content is a comparison result calculated based on the derived data set of the main example anorectal information and the data set features of the specified potential example anorectal information. The second anorectal patient description content is the common coefficient distribution anorectal patient description content calculated based on the data set features of the main example anorectal information and the clipped data set of the number of a number of patient information. The clipped data set of the number of a number of patient information is a data set obtained by clipping the main example anorectal information in the data set description direction by the number of a number of patient information.

[0066] Specifically, in one implementation, the training process of the first convolution thread includes the following steps: Step 410, obtain a number of first main example anorectal information and a number of specified potential example anorectal information, and perform dataset derivation on each first main example anorectal information to obtain a number of derived datasets; Step 420, based on the first convolution thread, perform feature extraction on the derived datasets to obtain a splicing result, and based on the second convolution thread, perform feature extraction on the number of specified potential example anorectal information to obtain a number of potential example anorectal information features; Step 430, calculate a comparison result based on the splicing result and the potential example anorectal information features to obtain the first anorectal patient description content; Step 440, perform shearing of the number of patient information in the dataset description direction on each first main example anorectal information to obtain a number of sheared datasets, and based on the first convolution thread, perform feature extraction on the first main example anorectal information and the number of sheared datasets to obtain the first main example anorectal information features and the number of sheared dataset features; Step 450, calculate the common coefficient distribution anorectal patient description content based on the first main example anorectal information features, the number of sheared dataset features, and the common coefficient distribution relationship between the number of sheared datasets and the first main example anorectal information to obtain the second anorectal patient description content; Step 460, optimize the coefficients of the first convolution thread according to the first anorectal patient description content and the second anorectal patient description content.

[0067] In step 410, the first main example anorectal information is a dataset sample for training the first convolution thread, and a number of first main example anorectal information can form a training set. The number of first main example anorectal information can include datasets of multiple dataset types, so that the trained first convolution thread can be applicable to general scenarios.

[0068] The specified potential example anorectal information can be a dataset that is different from the first main example anorectal information in content and is used to assist the first main example anorectal information in training the first convolution thread. When a dataset is different from the first main example anorectal information, this dataset can be used as the specified potential example anorectal information.

[0069] The specified potential example anorectal information can also be a dataset that has a relatively small association with the dataset type characteristics of the first main example anorectal information.

[0070] The specified potential example anorectal information can be obtained from the potential example anorectal information library. The potential example anorectal information library can be pre-built based on a number of first main example anorectal information and contains a large number of specified potential example anorectal information.

[0071] Based on a number of first main example anorectal information, a potential example anorectal information database can be built to obtain in advance a data set with content differences from the number of first main example anorectal information, and build a potential example anorectal information database. When performing the first convolutional thread training, a number of specified potential example anorectal information can be randomly extracted from the potential example anorectal information database and a number of first main example anorectal information to train the first convolutional thread.

[0072] Building a potential example anorectal information database based on a number of first main example anorectal information can also build a potential example anorectal information database based on the data set types of a number of first main example anorectal information. When the number of first main example anorectal information contains data sets of multiple data set types, data sets of multiple data set types can be obtained to build a potential example anorectal information database. When performing the first convolutional thread training, a number of specified potential example anorectal information of other data set types can be obtained from the potential example anorectal information database based on the data set type of each first main example anorectal information to train the first convolutional thread.

[0073] Performing data set derivation on each first main example anorectal information can obtain a number of derived data sets. The derived data sets include the first main example anorectal information and the data sets obtained by performing data set derivation on the first main example anorectal information. Performing data set derivation on the first main example anorectal information can increase the number of training samples for training the first convolutional thread. The derived data sets obtained by derivation based on the first main example anorectal information can be used as duplicate data sets of the first main example anorectal information to train the first convolutional thread.

[0074] In one implementation, performing data set derivation on each first main example anorectal information to obtain a number of derived data sets, including: performing patient attribute derivation on the first main example anorectal information to obtain a first local derived data set; performing treatment information derivation on the first main example anorectal information to obtain a second local derived data set; performing treatment information derivation on the first local derived data set to obtain a third local derived data set; determining a number of derived data sets based on the first local derived data set, the second local derived data set, the third local derived data set, and the first main example anorectal information.

[0075] When performing data set derivation on the first main example anorectal information, patient attribute derivation and treatment information derivation can also be performed on the first main example anorectal information at the same time. That is to say, the first local derived data set obtained by patient attribute derivation can be further subjected to treatment information derivation to obtain a third local derived data set.

[0076] Derivation of the dataset is achieved from several aspects through patient attribute derivation and treatment information derivation, improving the comprehensiveness of the derived dataset. For a number of derived datasets obtained from the first major example of anorectal information, they belong to duplicate datasets. Therefore, using the derived datasets obtained through various methods to train the first convolutional thread can improve the generalization and robustness of the thread in feature extraction for duplicate datasets, and thus improve the accuracy of feature extraction by the first convolutional thread.

[0077] The above implementation manner performs patient attribute derivation and treatment information derivation based on the complete first major example of anorectal information. In another implementation manner, before performing patient attribute derivation on the first major example of anorectal information to obtain the first local derived dataset, it further includes: dividing the first major example of anorectal information into several dataset segments, and extracting anorectal patient element information from each dataset segment to obtain several anorectal patient element information.

[0078] The data volume of the first major example of anorectal information is relatively large, and sometimes the difference in anorectal patient element information for several consecutive ones is relatively small. Therefore, the first major example of anorectal information can be divided into several dataset segments, and several anorectal patient element information can be extracted from the several dataset segments to represent the entire first major example of anorectal information. When performing dataset derivation, derivation can be performed based on the several extracted anorectal patient element information.

[0079] After extracting anorectal patient element information from the first major example of anorectal information, when performing dataset derivation, derivation can be performed based on the anorectal patient element information. Therefore, based on the several extracted anorectal patient element information, performing patient attribute derivation on the first major example of anorectal information to obtain the first local derived dataset includes: performing patient attribute derivation on the several anorectal patient element information to obtain several first local derived anorectal patient element information.

[0080] When performing patient attribute derivation on the several anorectal patient element information, the number of deleted frames can be the same as the number of added frames, so that the number of the several first local derived anorectal patient element information after patient attribute derivation is the same as the number of the several anorectal patient element information, to ensure that after feature extraction based on the several anorectal patient element information and the several first local derived anorectal patient element information, the corresponding feature dimensions of their splicing results are the same.

[0081] Derive treatment information from the first partial derivative dataset to obtain a third partial derivative dataset, including: Derive treatment information from several first partial derivative anorectal patient element information derived from patient attributes to obtain several third partial derivative anorectal patient element information. That is to say, derive patient attributes and treatment information from several anorectal patient element information simultaneously to obtain several third partial derivative anorectal patient element information.

[0082] Divide the first main example anorectal information into multiple segments and then extract several for dataset derivation, which realizes reducing the number of anorectal patient element information of the first main example anorectal information and improving the efficiency of dataset derivation without affecting the dataset derivation result.

[0083] In step 420, first perform feature extraction on the derivative dataset based on the first convolutional thread to obtain a splicing result.

[0084] In an implementation manner of dataset derivation in step 410, several can be extracted from the first main example anorectal information frame for patient attribute derivation to obtain several first partial derivative anorectal patient element information, perform treatment information derivation to obtain several second partial derivative anorectal patient element information, and simultaneously perform patient attribute derivation and treatment information derivation to obtain several third partial derivative anorectal patient element information. Therefore, perform feature extraction on the derivative dataset based on the first convolutional thread to obtain a splicing result, including: Perform feature extraction on several derivative anorectal patient element information based on the first convolutional thread to obtain a splicing result. The several derivative anorectal patient element information includes several anorectal patient element information, several first partial derivative anorectal patient element information derived only based on patient attributes, several second partial derivative anorectal patient element information derived only based on treatment information, and several third partial derivative anorectal patient element information derived simultaneously based on patient attributes and treatment information.

[0085] Compared with directly performing feature extraction on the complete first main example anorectal information, performing feature extraction on several derivative anorectal patient element information can improve the feature extraction efficiency.

[0086] In an implementation manner, the first convolutional thread includes a first unit and a second unit. Therefore, perform feature extraction on several derivative anorectal patient element information based on the first convolutional thread to obtain a splicing result, including: Perform feature extraction on several derivative anorectal patient element information based on the first unit to obtain several feature points; perform feature splicing on the several feature points based on the second unit to obtain a splicing result.

[0087] After obtaining a number of feature points, the number of feature points can be input into the second unit. The second unit is used to perform feature splicing on the number of feature points to obtain the splicing result corresponding to the derived dataset. The second unit can be a multi-layer deep learning thread.

[0088] In one implementation, based on the second unit performing feature splicing on a number of feature points to obtain a splicing result, it includes: performing clustering processing on a number of feature points to obtain a first transition queue, a second transition queue, and a third transition queue; performing dimensionless simplification processing on the first transition queue and the second transition queue, and performing function processing on the dimensionless simplification processing result and the third transition queue to obtain a function processing result; calculating the depolarization result of the function processing result to obtain a depolarization processing result.

[0089] In another implementation of dataset derivation in step 410, a number of derived datasets are determined according to the first local derived dataset, the second local derived dataset, the third local derived dataset, and the first main example anorectal information.

[0090] In another implementation, based on the first convolutional thread performing feature extraction on the derived dataset to obtain a splicing result, it includes: dividing the derived dataset into a number of derived dataset segments, and extracting a number of anorectal patient element information images from the number of derived dataset segments; performing frame-level feature extraction on the number of anorectal patient element information images based on the first neural network to obtain a splicing result.

[0091] The difference between this implementation and the previous implementation is that in the previous implementation, frame extraction is performed before dataset derivation, dataset derivation is performed based on the extracted number of anorectal patient element information, and finally feature extraction is performed based on the number of derived anorectal patient element information. In this implementation, frame extraction is performed after obtaining the derived dataset. The method of extracting a number of anorectal patient element information images from a number of derived dataset segments is the same as the method of dividing the first main example anorectal information into a number of dataset segments and extracting anorectal patient element information from each dataset segment in the previous embodiment, and will not be elaborated here. Extracting frames from the derived dataset for feature extraction can reduce the amount of features and improve the efficiency of feature extraction.

[0092] In addition to obtaining a splicing result by performing feature extraction on the derived dataset based on the first convolutional thread, in step 420, it further includes: performing feature extraction on a number of specified potential example anorectal information based on the second convolutional thread to obtain a number of potential example anorectal information features.

[0093] The second convolution thread can be a convolution thread with the same structure as the first convolution thread, and is used to extract features from the specified potential example anorectal information. Since the specified potential example anorectal information is a dataset that is different from the first main example anorectal information, in order to avoid the specified potential example anorectal information affecting the feature extraction of the first main example anorectal information, the first main example anorectal information and the specified potential example anorectal information can be subjected to feature extraction using different convolution threads.

[0094] Since the second convolution thread can have the same structure as the first convolution thread, the way the second convolution thread extracts features from the specified potential example anorectal information is the same as the way the first convolution thread extracts features from the derived dataset, which will not be elaborated here. The dimension of the features of the potential example anorectal information obtained by the second convolution thread for feature extraction is the same as the dimension of the splicing result.

[0095] In step 430, a comparison result is calculated based on the splicing result and the features of the potential example anorectal information to obtain the first description content of the anorectal patient.

[0096] In one implementation, calculating a comparison result based on the splicing result and the features of the potential example anorectal information to obtain the first description content of the anorectal patient includes: for each first main example anorectal information, randomly extracting two from the corresponding several splicing results to form a positive sample pair; calculating the first training commonality coefficient between the two splicing results in the positive sample pair; determining a target splicing result from the two splicing results in the positive sample pair, and calculating several second training commonality coefficients between the target splicing result and several features of the potential example anorectal information; calculating a comparison result based on the first training commonality coefficient and the several second training commonality coefficients to obtain the first description content of the anorectal patient.

[0097] The derived dataset includes the first main example anorectal information and the dataset derived from the first main example anorectal information. That is to say, each first main example anorectal information has corresponding several derived datasets, and the several derived datasets belong to the same dataset and have a very high feature commonality coefficient.

[0098] In the foregoing one implementation, the several derived datasets include the first main example anorectal information, the first local derived dataset obtained by deriving patient attributes from the first main example anorectal information, the second local derived dataset obtained by deriving treatment information from the first main example anorectal information, and the third local derived dataset obtained by simultaneously deriving patient attributes and treatment information from the first main example anorectal information. Therefore, the splicing result includes the first main example anorectal information feature, the first local splicing result, the second local splicing result, and the third local splicing result.

[0099] In step 440, first, for each first main example anorectal information, a number of patient information numbers are sheared in the direction of the dataset description to obtain a number of sheared datasets.

[0100] After obtaining a number of sheared datasets, feature extraction can be performed on the first main example anorectal information and the number of sheared datasets based on the first convolutional thread to obtain the first main example anorectal information features and the number of sheared dataset features.

[0101] In step 450, based on the first main example anorectal information features, the number of sheared dataset features, and the common coefficient distribution relationship between the number of sheared datasets and the first main example anorectal information, the common coefficient distribution anorectal patient description content is calculated to obtain the second anorectal patient description content.

[0102] The common coefficient distribution anorectal patient description content can enable the first convolutional thread to learn the common coefficient distribution relationship of the features, so that the first convolutional thread can extract dataset features with closer differences for more similar datasets.

[0103] In one implementation, based on the first main example anorectal information features, the number of sheared dataset features, and the common coefficient distribution relationship between the number of sheared datasets and the first main example anorectal information, the common coefficient distribution anorectal patient description content is calculated to obtain the second anorectal patient description content, including: combining the first main example anorectal information features and the number of sheared dataset features in pairs to obtain a number of feature comparison groups; calculating the third training common coefficient between the two dataset features in each feature comparison group; calculating the common coefficient distribution anorectal patient description content according to the number of third training common coefficients and the common coefficient distribution relationship between the number of sheared datasets and the first main example anorectal information to obtain the second anorectal patient description content.

[0104] For each feature comparison group, calculate the third training common coefficient between the two dataset features therein. The method of calculating the third training common coefficient can be the same as the methods of calculating the first training common coefficient and the second training common coefficient in the foregoing embodiments, and will not be elaborated here.

[0105] The common coefficient distribution between the number of sheared datasets and the first main example anorectal information can be that the larger the number of patient information in the sheared dataset, the higher the common coefficient distribution with the first main example anorectal information, which is also the goal to be achieved by the third training common coefficient. That is to say, the higher the third training common coefficient between the dataset features corresponding to the datasets with closer patient information numbers.

[0106] In order to calculate the content description of anorectal patients for the commonality coefficient distribution more intuitively, in one implementation, the content description of anorectal patients for the commonality coefficient distribution is calculated according to the relationship between a number of third training commonality coefficients and the commonality coefficient distribution of a number of clipped data sets and the first main example anorectal information, obtaining the second content description of anorectal patients, including: generating a commonality coefficient queue according to a number of third training commonality coefficients; calculating the content description of anorectal patients for the commonality coefficient distribution based on the commonality coefficient queue and the commonality coefficient distribution relationship between a number of clipped data sets and the first main example anorectal information, obtaining the second content description of anorectal patients.

[0107] The commonality coefficient queue can intuitively represent the relationship between the characteristics of the clipped data sets of the number of patient information and the characteristics of the first main example anorectal information. The number of rows and columns of the commonality coefficient queue is equal to the sum of the number of characteristics of the first main example anorectal information and the characteristics of the clipped data set. Each row and each column represent a data set characteristic, and each queue value represents the third training commonality coefficient between the two data set characteristics at the corresponding position.

[0108] The relationship between the commonality coefficient distribution of a number of clipped data sets and the first main example anorectal information should be that the closer the number of patient information is, the higher the commonality coefficient distribution. Therefore, when calculating the content description of anorectal patients for the commonality coefficient distribution based on the commonality coefficient queue and the commonality coefficient distribution relationship between a number of clipped data sets and the first main example anorectal information, the calculation can be based on the commonality coefficient distribution in the commonality coefficient queue.

[0109] Using the commonality coefficient queue to calculate the content description of anorectal patients for the commonality coefficient distribution can intuitively and accurately obtain the commonality coefficient distribution relationship between a number of clipped data sets and the first main example anorectal information, and calculate the content description of anorectal patients for the commonality coefficient distribution between queue values based on the commonality coefficient distribution relationship, which is beneficial to improving the efficiency and accuracy of calculating the content description of anorectal patients for the commonality coefficient distribution.

[0110] In another implementation, the training process of the first convolution thread further includes: obtaining a number of second main example anorectal information and the corresponding classification directory for each second main example anorectal information; extracting the characteristics of the second main example anorectal information based on the first convolution thread, obtaining the characteristics of the second main example anorectal information; classifying the characteristics of the second main example anorectal information based on the specified classification unit, obtaining the classification prediction possibility value; calculating the content description of classified anorectal patients based on the classification prediction possibility value and the classification directory, obtaining the third content description of anorectal patients.

[0111] A number of second major example anorectal information are dataset samples carrying classification catalogs. A number of second major example anorectal information may have an intersection with a number of first dataset samples, or may be major example anorectal information that is completely different from a number of first major example anorectal information. A number of second major example anorectal information can obtain a dataset carrying a classification catalog from a publicly available dataset website.

[0112] In step 460, the coefficients of the first convolution thread are optimized according to the first anorectal patient description content and the second anorectal patient description content.

[0113] Calculating the total anorectal patient description content can also assign different weight coefficients to the first anorectal patient description content and the second anorectal patient description content, calculate the weighted sum between the first anorectal patient description content and the second anorectal patient description content based on the weight coefficients, and obtain the total anorectal patient description content. The weight coefficients indicate the influence of the first anorectal patient description content and the second anorectal patient description content on the training of the first convolution thread. Since the first anorectal patient description content trains the first convolution thread's feature extraction ability for distinguishing duplicate datasets and similar datasets, and the second anorectal patient description content trains the first convolution thread's ability for the common coefficient distribution of a number of similar datasets. Therefore, different weight coefficients can be assigned to the first anorectal patient description content and the second anorectal patient description content based on the training bias of the first convolution thread in actual applications.

[0114] In the foregoing implementation manner, in addition to the first anorectal patient description content and the second anorectal patient description content, a third anorectal patient description content is also calculated based on a number of second major example anorectal information carrying a catalog. Based on this, optimizing the coefficients of the first convolution thread according to the first anorectal patient description content and the second anorectal patient description content includes: optimizing the coefficients of the first convolution thread according to the first anorectal patient description content, the second anorectal patient description content, and the third anorectal patient description content.

[0115] In an implementation manner, after optimizing the coefficients of the first convolution thread according to the first anorectal patient description content and the second anorectal patient description content, the training process of the first convolution thread further includes: obtaining the optimization coefficient of the thread coefficient; optimizing the coefficients of the second convolution thread based on the optimized coefficients of the first convolution thread and the optimization coefficient.

[0116] After optimizing the first convolution thread, the coefficients of the second convolution thread can also be optimized according to the coefficients of the first convolution thread, so that the second convolution thread can achieve a more accurate extraction effect when extracting features of specified potential example anorectal information, thereby improving the accuracy of using potential example anorectal information to assist the training of the first convolution thread.

[0117] The second convolution thread has the same network result as the first convolution thread. Therefore, the second convolution thread can be optimized based on the optimization of the coefficients in the first convolution thread. The optimization can be carried out based on the previous gradient optimization direction of the second convolution thread and the optimization method of the gradient in the first convolution thread.

[0118] Optimizing the coefficients in the second convolution thread based on the optimization can enable the second convolution thread to converge quickly to achieve the best feature extraction effect. The potential example anorectal information features obtained based on the second convolution thread are becoming more and more accurate, which can then accelerate the training of the first convolution thread.

[0119] Using the first convolution thread to extract features from the derived dataset and using the second neural network to extract features from the specified potential example anorectal information can avoid the influence of the feature extraction of the potential example anorectal information on the feature extraction of the derived dataset, which is beneficial to improving the accuracy of the training of the first convolution thread. Training the first convolution thread based on the comparison result between the splicing result and the potential example anorectal information features can enable the first convolution thread to have the ability to distinguish between duplicate datasets and similar datasets when performing feature extraction; training the first convolution thread based on the common coefficient distribution anorectal patient description content between several clipped datasets and the first main example anorectal information can enable the first convolution thread to have the ability to extract different dataset features based on the common coefficient distribution between datasets, and for datasets with higher common coefficients, it can extract more similar dataset features. Therefore, the training process of the first convolution thread proposed in the embodiments of the present disclosure is beneficial to improving the accuracy of the first convolution thread in performing feature extraction.

[0120] After training the first convolution thread, the first convolution thread can be used to extract features from anorectal patient information and several historical anorectal patient information.

[0121] In the foregoing implementation manner, the first convolution thread includes a first unit and a second unit. Based on this, when using the first convolution thread to extract features from anorectal patient information and several historical anorectal patient information, obtaining anorectal patient information features and several historical anorectal patient information features includes: extracting several anorectal patient information attributes from the anorectal patient information, and extracting several historical anorectal patient information attributes from each historical anorectal patient information; extracting several query feature points based on the first unit from the several anorectal patient information attributes, and extracting several candidate feature points based on the first unit from the several historical anorectal patient information attributes; performing feature splicing on the several query feature points based on the second unit to obtain anorectal patient information features, and performing feature splicing on the several candidate feature points corresponding to each historical anorectal patient information based on the second unit to obtain several historical anorectal patient information features.

[0122] In one embodiment, a number of anorectal patient information attributes are extracted from the anorectal patient information, and a number of historical anorectal patient information attributes are extracted from each piece of historical anorectal patient information, including: dividing the anorectal patient information into a number of anorectal patient information segments, and dividing each piece of historical anorectal patient information into a number of historical anorectal patient information segments; extracting one anorectal patient information attribute from each anorectal patient information segment to obtain a number of anorectal patient information attributes; and extracting one historical anorectal patient information attribute from each historical anorectal patient information segment to obtain a number of historical anorectal patient information attributes corresponding to each piece of historical anorectal patient information.

[0123] Similar to the way of extracting frames from the first main example anorectal information in the foregoing embodiment, the anorectal patient information can be first divided into a number of anorectal patient information segments, and one is extracted from each anorectal patient information segment, so as to obtain a number of anorectal patient information attributes. The historical anorectal patient information is divided into a number of historical anorectal patient information segments, and one is extracted from each historical anorectal patient information segment, so as to obtain a number of historical anorectal patient information attributes.

[0124] After extracting a number of anorectal patient information attributes and a number of historical anorectal patient information attributes, feature extraction can be performed on the number of anorectal patient information attributes based on the first unit, and feature extraction can be performed on the number of historical anorectal patient information attributes. The process of feature extraction by the first unit has been described in detail in the training process of the foregoing first convolutional thread, and will not be elaborated here.

[0125] Step 330: Calculate the commonality coefficient between the anorectal patient information feature and each historical anorectal patient information feature.

[0126] In step 330, similar to the process of calculating the first training commonality coefficient in the training process of the foregoing first convolutional thread, methods such as Euclidean difference, Manhattan difference, and cosine commonality coefficient can be used to calculate the commonality coefficient between the anorectal patient information feature and each historical anorectal patient information feature, and will not be elaborated here.

[0127] Step 340: Perform data sharing processing according to the commonality coefficient.

[0128] In another embodiment, performing data sharing processing according to the commonality coefficient includes: determining that the data set with a commonality coefficient less than the specified target value between the number of historical anorectal patient information and the anorectal patient information is the first target data set; distributing the number of first target data sets in descending order of the commonality coefficient with the anorectal patient information, and determining the specified number of data sets in the front as the second target data set; and performing sharing processing on the number of second target data.

[0129] Since some historical anorectal patient information may contain data sets that are duplicates of the content of anorectal patient information, sharing these duplicate data with medical objects will affect the perception of medical objects. Therefore, a specified target value can be used to determine whether the historical anorectal patient information is a duplicate of the anorectal patient information.

[0130] This is another flowchart of the anorectal patient information sharing method based on artificial intelligence provided by the present disclosure. The method specifically includes the following steps:

[0131] Step 1010: Obtain anorectal patient information and several pieces of historical anorectal patient information.

[0132] Several pieces of historical anorectal patient information are data sets that can be recommended to medical objects and can be obtained from the data set library in the background of the data sharing processing application. The purpose of data sharing processing is to find similar historical anorectal patient information from several pieces of historical anorectal patient information for comparison.

[0133] Step 1020: Extract several anorectal patient information attributes from the anorectal patient information, and extract several historical anorectal patient information attributes from each piece of historical anorectal patient information.

[0134] Extracting several anorectal patient information attributes from the anorectal patient information can divide the anorectal patient information into several anorectal patient information segments, and extract one anorectal patient information attribute from each anorectal patient information segment to obtain several anorectal patient information attributes.

[0135] Similarly, extracting several historical anorectal patient information attributes from the historical anorectal patient information can divide each piece of historical anorectal patient information into several historical anorectal patient information segments, and extract one historical anorectal patient information attribute from each historical anorectal patient information segment to obtain several historical anorectal patient information attributes corresponding to each piece of historical anorectal patient information. The number of divisions of the anorectal patient information and the historical anorectal patient information can be the same.

[0136] Step 1030: Use the first unit in the first convolutional thread to extract features from several anorectal patient information attributes to obtain several query feature points, and extract features from several historical anorectal patient information attributes to obtain several candidate feature points; use the second unit to splice the features of several query feature points to obtain anorectal patient information features, and splice the features of several historical anorectal patient information attributes to obtain historical anorectal patient information features.

[0137] Step 1031: Obtain a number of first main example anorectal information and a number of specified potential example anorectal information. Divide the first main example anorectal information into a number of data set segments, and extract anorectal patient element information from each data set segment to obtain a number of anorectal patient element information. Perform patient attribute derivation and treatment information derivation on the number of anorectal patient element information to obtain a number of derived anorectal patient element information.

[0138] The first main example anorectal information is a data set sample for training the first convolutional thread, and a number of first main example anorectal information can form a training set. The first main example anorectal information can be obtained from a publicly available data set database or from a number of data sharing processing applications. The specified potential example anorectal information can be a data set that is different from the first main example anorectal information in content and is used to assist the first main example anorectal information in training the first convolutional thread.

[0139] Divide the first main example anorectal information into a number of data set segments, and extract a number of anorectal patient element information from the number of data set segments to represent the entire first main example anorectal information.

[0140] The number of derived anorectal patient element information includes a number of anorectal patient element information extracted from the first main example anorectal information, a number of anorectal patient element information that only performs patient attribute derivation, a number of anorectal patient element information that only performs treatment information derivation, and a number of anorectal patient element information that simultaneously performs patient attribute derivation and treatment information derivation.

[0141] Step 1032: Based on the first unit in the first convolutional thread, perform feature extraction on the number of derived anorectal patient element information to obtain a number of feature points; based on the second unit, perform feature splicing on the number of feature points to obtain a splicing result; based on the second convolutional thread, perform feature extraction on the number of specified potential example anorectal information to obtain a number of potential example anorectal information features.

[0142] The second convolutional thread can be a convolutional thread with the same structure as the first convolutional thread and is used to perform feature extraction on the specified potential example anorectal information. Since the second convolutional thread can have the same structure as the first convolutional thread, the method for the second convolutional thread to perform feature extraction on the specified potential example anorectal information is the same as the method for the first convolutional thread to perform feature extraction on the derived data set.

[0143] Step 1033: Calculate a comparison result based on the splicing result and the potential example anorectal information features to obtain the first anorectal patient description content.

[0144] Step 1034: Perform shearing on each first major example anorectal information in the dataset description direction for a number of patient information, obtaining a number of sheared datasets, and based on the first convolutional thread, perform feature extraction on the first major example anorectal information and the number of sheared datasets to obtain the first major example anorectal information features and the number of sheared dataset features.

[0145] After obtaining a number of sheared datasets, the first major example anorectal information and the number of sheared datasets can be input into the first neural network for feature extraction to obtain the first major example anorectal information features and the number of sheared dataset features.

[0146] Step 1035: Calculate the common coefficient distribution anorectal patient description content based on the first major example anorectal information features, the number of sheared dataset features, and the common coefficient distribution relationship between the number of sheared datasets and the first major example anorectal information, obtaining the second anorectal patient description content.

[0147] Step 1036: Obtain the second major example anorectal information and the corresponding classification catalog, perform feature extraction on the second major example anorectal information based on the first convolutional thread to obtain the second major example anorectal information features, and classify the second major example anorectal information features based on the specified classification network to obtain the classification prediction possibility value.

[0148] After performing feature extraction on the second major example anorectal information based on the first convolutional thread to obtain the second major example anorectal information features, classify the second major example anorectal information features based on the specified classification network to obtain the classification prediction possibility value. The specified classification unit can be a multi-layer artificial intelligence thread, for example, a CNN network or a Transformer thread. The classification prediction possibility value can represent the probability that the second major example anorectal information is predicted to be classified into the classification catalog.

[0149] Step 1037: Calculate the classified anorectal patient description content based on the classification prediction possibility value and the classification catalog, obtaining the third anorectal patient description content.

[0150] Step 1038: Optimize the coefficients of the first convolutional thread according to the first anorectal patient description content, the second anorectal patient description content, and the third anorectal patient description content.

[0151] When optimizing the coefficients of the first convolutional thread according to the first anorectal patient description content, the second anorectal patient description content, and the third anorectal patient description content, the first anorectal patient description content, the second anorectal patient description content, and the third anorectal patient description content can be added together to obtain the total anorectal patient description content.

[0152] Step 1039: Optimize the coefficients in the second convolution thread based on the coefficients optimized by the first convolution thread and the optimization coefficients.

[0153] After optimizing the first convolution thread, the coefficients of the second convolution thread can also be optimized according to the coefficients of the first convolution thread. The second convolution thread has the same network structure as the first convolution thread, so the second convolution thread can be optimized based on the optimization of the coefficients in the first convolution thread. The optimization can be carried out based on the previous gradient optimization direction of the second convolution thread and the gradient optimization method in the first convolution thread. The optimization coefficient can indicate the degree of optimization of the coefficients in the second convolution thread. The larger the optimization coefficient, the greater the influence of the previous gradient optimization direction of the second convolution thread on the current gradient optimization.

[0154] Step 1040: Calculate the commonality coefficients between the anorectal patient information features and the historical anorectal patient information features corresponding to each historical anorectal patient information.

[0155] Step 1050: Determine the historical anorectal patient information with a commonality coefficient less than the specified target value with the anorectal patient information as the candidate recommendation dataset, distribute the candidate recommendation dataset in descending order of the commonality coefficient with the anorectal patient information, and share the top predetermined number of data processed to medical objects.

[0156] On the above basis, an anorectal patient information sharing device based on artificial intelligence is provided. The device includes:

[0157] A data acquisition module, configured to acquire anorectal patient information and an anorectal patient information database, where the anorectal patient information database includes a number of historical anorectal patient information;

[0158] A feature extraction module, configured to extract features from the anorectal patient information and the number of historical anorectal patient information based on the first convolution thread, to obtain anorectal patient information features and a number of historical anorectal patient information features. The first convolution thread is trained based on the target anorectal patient description content, and the target anorectal patient description content includes the first anorectal patient description content and the second anorectal patient description content. The first anorectal patient description content is a comparison result calculated based on the derivative dataset of the main example anorectal information and the dataset features of the specified potential example anorectal information. The second anorectal patient description content is the commonality coefficient distribution anorectal patient description content calculated based on the dataset features of the main example anorectal information and the shear dataset of the number of a number of patient information. The shear dataset of the number of a number of patient information is a dataset obtained by shearing the main example anorectal information in the dataset description direction by a number of patient information;

[0159] A commonality coefficient calculation module for calculating the commonality coefficient between the anorectal patient information features and each of the historical anorectal patient information features;

[0160] A data sharing module for performing data sharing processing in combination with the commonality coefficient.

[0161] On the above basis, an anorectal patient information sharing system based on artificial intelligence is shown, including a processor and a memory that communicate with each other. The processor is configured to read and execute a computer program from the memory to implement the above method.

[0162] On the above basis, a computer-readable storage medium is also provided, on which a computer program stored thereon implements the above method when running.

[0163] In summary, based on the above solution, anorectal patient information and an anorectal patient information database are obtained. The anorectal patient information database includes a number of historical anorectal patient information. Feature extraction is performed on the anorectal patient information and a number of historical anorectal patient information based on the first convolutional thread to obtain anorectal patient information features and a number of historical anorectal patient information features. The first convolutional thread is trained based on the target anorectal patient description content, and the target anorectal patient description content includes the first anorectal patient description content and the second anorectal patient description content. The first anorectal patient description content is a comparison result calculated based on the derived data set of the main example anorectal information and the data set features of the specified potential example anorectal information. The second anorectal patient description content is the commonality coefficient distribution anorectal patient description content calculated based on the data set features of the main example anorectal information and the clipped data set of the number of a number of patient information. The clipped data set of the number of a number of patient information is a data set obtained by clipping the main example anorectal information in the data set description direction by the number of a number of patient information. Calculate the commonality coefficient between the anorectal patient information features and each historical anorectal patient information feature; perform data sharing processing according to the commonality coefficient. In this application, patient interference data can be deleted, and the data during sharing can be ensured to be accurate and reliable.

[0164] It should be understood that the systems and their modules shown above can be implemented in various ways. For example, in some embodiments, the systems and their modules can be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art can understand that the above methods and systems can be implemented using computer-executable instructions and / or included in processor control code. For example, such code is provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The systems and their modules of the present application can be implemented not only by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips and transistors, or programmable hardware devices such as field programmable gate arrays and programmable logic devices, but also by software executed by various types of processors, or by a combination of the above hardware circuits and software (e.g., firmware).

[0165] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the possible beneficial effects can be any one or several combinations of the above, or any other possible beneficial effects that can be obtained.

Claims

1. An artificial intelligence-based information sharing method for anorectal patients, characterized in that, The method includes: Obtaining anorectal patient information and an anorectal patient information database, where the anorectal patient information database includes a number of historical anorectal patient information; Performing feature extraction on the anorectal patient information and the number of historical anorectal patient information based on a first convolution thread to obtain anorectal patient information features and a number of historical anorectal patient information features. The first convolution thread is trained based on target anorectal patient description content, and the target anorectal patient description content includes first anorectal patient description content and second anorectal patient description content. The first anorectal patient description content is a comparison result calculated based on the derived dataset of the main example anorectal information and the dataset features of the specified potential example anorectal information. The second anorectal patient description content is the common coefficient distribution anorectal patient description content calculated based on the dataset features of the main example anorectal information and the clipped dataset of the number of patient information. The clipped dataset of the number of patient information is a dataset obtained by clipping the main example anorectal information in the dataset description direction by the number of patient information; Calculating the common coefficient between the anorectal patient information features and each of the historical anorectal patient information features; Performing data sharing processing in combination with the common coefficient.

2. The method according to claim 1, wherein The training process of the first convolution thread includes the following steps: Obtaining a number of first main example anorectal information and a number of specified potential example anorectal information, and performing dataset derivation on each of the first main example anorectal information to obtain a number of derived datasets. The derived datasets include the first main example anorectal information and the datasets obtained by performing dataset derivation on the first main example anorectal information; Performing feature extraction on the derived datasets based on a first convolution thread to obtain a splicing result, and performing feature extraction on the number of specified potential example anorectal information based on a second convolution thread to obtain a number of potential example anorectal information features; Calculating a comparison result based on the splicing result and the potential example anorectal information features to obtain the first anorectal patient description content; Clipping each of the first main example anorectal information in the dataset description direction by the number of patient information to obtain a number of clipped datasets, and performing feature extraction on the first main example anorectal information and the number of clipped datasets based on the first convolution thread to obtain first main example anorectal information features and a number of clipped dataset features; Calculating the common coefficient distribution anorectal patient description content based on the first main example anorectal information features, the number of clipped dataset features, and the common coefficient distribution relationship between the number of clipped datasets and the first main example anorectal information to obtain the second anorectal patient description content; Optimizing the coefficients of the first convolution thread by combining the first anorectal patient description content and the second anorectal patient description content.

3. The method according to claim 2, wherein The method further includes: Obtaining a number of second main example anorectal information and the classification directory corresponding to each of the second main example anorectal information; Extract features from the second main example anorectal information according to the first convolutional thread to obtain the features of the second main example anorectal information; Classify the features of the second main example anorectal information based on a specified classification unit to obtain a classification prediction probability value; Calculate the classified anorectal patient description content according to the classification prediction probability value and the classification directory to obtain the third anorectal patient description content; The optimization of the coefficients of the first convolutional thread by combining the first anorectal patient description content and the second anorectal patient description content includes: Optimize the coefficients of the first convolutional thread by combining the first anorectal patient description content, the second anorectal patient description content, and the third anorectal patient description content.

4. The method according to claim 2, wherein The derivation of the dataset for each of the first main example anorectal information to obtain a number of derived datasets includes: Derive patient attributes from the first main example anorectal information to obtain a first local derived dataset; Derive treatment information from the first main example anorectal information to obtain a second local derived dataset; Derive treatment information from the first local derived dataset to obtain a third local derived dataset; Determine a number of derived datasets by combining the first local derived dataset, the second local derived dataset, the third local derived dataset, and the first main example anorectal information.

5. The method according to claim 4, wherein Before deriving the patient attributes from the first main example anorectal information to obtain the first local derived dataset, it further includes: Divide the first main example anorectal information into a number of dataset segments, and extract anorectal patient element information from each of the dataset segments to obtain a number of anorectal patient element information; The derivation of the patient attributes from the first main example anorectal information to obtain the first local derived dataset includes: Derive patient attributes from the number of anorectal patient element information to obtain a number of first local derived anorectal patient element information; The feature extraction of the derived dataset based on the first convolutional thread to obtain a splicing result includes: Extract features from a number of derived anorectal patient element information based on the first convolutional thread to obtain a splicing result, where the number of derived anorectal patient element information includes the number of anorectal patient element information, the number of first local derived anorectal patient element information derived only based on patient attributes, the number of second local derived anorectal patient element information derived only based on treatment information, and the number of third local derived anorectal patient element information derived based on both patient attributes and treatment information.

6. The method according to claim 5, wherein The first convolutional thread includes a first unit and a second unit. The feature extraction of a number of derived anorectal patient element information based on the first convolutional thread to obtain a splicing result includes: Extract features from a number of derived anorectal patient element information according to the first unit to obtain a number of feature points; Perform feature splicing on the number of feature points according to the second unit to obtain a splicing result; Among them, the performing feature splicing on the number of feature points according to the second unit to obtain a splicing result includes: Based on clustering the several feature points, a first transition queue, a second transition queue, and a third transition queue are obtained; Perform dimensionless simplification processing on the first transition queue and the second transition queue, and perform function processing on the dimensionless simplification processing result and the third transition queue to obtain a function processing result; Calculate the depolarization result of the function processing result to obtain a depolarization processing result.

7. The method according to claim 6, wherein The feature extraction of the anorectal patient information and the several historical anorectal patient information based on the first convolution thread to obtain anorectal patient information features and several historical anorectal patient information features includes: Extract several anorectal patient information attributes from the anorectal patient information, and extract several historical anorectal patient information attributes from each of the historical anorectal patient information; Extract several query feature points based on the first unit from the several anorectal patient information attributes, and extract several candidate feature points based on the first unit from the several historical anorectal patient information attributes; Perform feature splicing on the several query feature points based on the second unit to obtain anorectal patient information features, and perform feature splicing on the several candidate feature points corresponding to each historical anorectal patient information based on the second unit to obtain several historical anorectal patient information features; Among them, the extraction of several anorectal patient information attributes from the anorectal patient information and the extraction of several historical anorectal patient information attributes from each of the historical anorectal patient information includes: Divide the anorectal patient information into several anorectal patient information segments, and divide each of the historical anorectal patient information into several historical anorectal patient information segments; Extract one anorectal patient information attribute from each of the anorectal patient information segments to obtain several anorectal patient information attributes; Extract one historical anorectal patient information attribute from each of the historical anorectal patient information segments to obtain several historical anorectal patient information attributes corresponding to each of the historical anorectal patient information.

8. The method according to claim 2, characterized in that, After optimizing the coefficients of the first convolution thread by combining the first anorectal patient description content and the second anorectal patient description content, it further includes: Obtain the optimization coefficient of the thread coefficient; Optimize the coefficients of the second convolution thread according to the optimized coefficients of the first convolution thread and the optimization coefficient.

9. The method according to claim 1, characterized in that The data sharing process by combining the common coefficients includes: Determine the data sets in the several historical anorectal patient information whose common coefficients with the anorectal patient information are less than the specified target value as the first target data sets; Distribute the several first target data sets in descending order of the common coefficients with the anorectal patient information, and determine the specified number of data sets in the front as the second target data sets; Perform sharing processing on the several second target data.

10. An anorectal patient information sharing system based on artificial intelligence, characterized in that, It includes a processor and a memory that communicate with each other. The processor is used to read and execute a computer program from the memory to implement the method according to any one of claims 1-9.

Images