Clinical data acquisition and analysis system based on feature mining

Abstract

The invention relates to the field of intelligent analysis, and particularly discloses a clinical data acquisition and analysis system based on feature mining, which comprises a personnel registration module, a data acquisition module, a data auditing module, a data storage module and a data analysis module, and is characterized in that firstly, participant information is recorded through the personnel registration module; then clinical data are collected by using the data acquisition module, the information accuracy and integrity are ensured by using the data auditing module, the data passing the auditing can be properly stored in the data storage, and finally, the data analysis module processes and analyzes the stored data and extracts useful information to support scientific decisions. And the quality and reliability of clinical test data can be ensured.

Description

Technical Field

[0001] The present application relates to the field of intelligent analysis, and more specifically, to a clinical data collection and analysis system based on feature mining. Background Art

[0002] Clinical trials are crucial to verifying the efficacy and safety of new drugs. They mainly study the tolerance of drugs to the human body and understand the absorption, distribution and excretion of drugs in the human body through pharmacokinetics, thereby providing a basis for determining appropriate medication regimens. Therefore, in order to ensure the accuracy and effectiveness of clinical trials, the data collected during the trial must be strictly collected and analyzed.

[0003] Chinese patent CN117316366A proposes a clinical trial information data collection and analysis system and method, which first registers and verifies the clinical trial participant information, and then generates a data collection table and process based on the specific situation of the subject. After data collection, the system will perform a logical check to ensure data quality. Ultimately, all approved data will be used for analysis to assess whether the subject can continue to participate in the trial.

[0004] In the above patent, the collected clinical trial data is logically checked by using specific logic check statements (CHECK statements) to verify the validity of the data. However, although the method of using specific logic check statements can verify specific data structures and logical relationships, when the design of the clinical trial changes or new logic that needs to be verified appears, the check statements need to be manually updated, which has poor flexibility. In addition, although the traditional method of using specific logic check statements can verify the basic logical consistency of the data to a certain extent, it is limited by predefined rules and it is difficult to deeply understand the complex semantic relationships in the data. This method makes it difficult to capture the deep semantic information between the data and its context, such as implicit associations, dynamically changing trends, etc., resulting in an incomplete understanding of the data, which may miss some potentially important details or patterns.

[0005] Therefore, an optimized clinical data acquisition and analysis system is desired. Summary of the invention

[0006] In order to solve the above technical problems, this application is proposed. The embodiment of the present application provides a clinical data collection and analysis system based on feature mining, which includes: a personnel registration module, a data collection module, a data review module, a data storage module and a data analysis module. Specifically, the participant information is first recorded through the personnel registration module, and then the clinical data is collected using the data collection module. Then the data review module ensures the accuracy and completeness of the information, and the audited data will be properly stored in the data storage. Finally, the data analysis module will process and analyze the stored data, extract useful information to support scientific decision-making, so that the quality and reliability of clinical trial data can be ensured.

[0007] According to one aspect of the present application, a clinical data collection and analysis system based on feature mining is provided, which includes:

[0008] A personnel registration module is used to register the information of the subjects; a data collection module is used to collect the clinical trial information data of the subjects; a data review module is used to review and inspect the clinical trial information data to obtain clinical trial information data that has passed the review and inspection; a data storage module is used to store the clinical trial information data that has passed the review and inspection; a data analysis module is used to process and analyze the clinical trial information data that has passed the review and inspection;

[0009] Wherein, the data review module includes:

[0010] A medical common sense text description extraction unit, used to extract medical common sense text descriptions from an external knowledge base;

[0011] A clinical trial medical common sense semantic encoding unit, used to semantically encode the clinical trial information data and the medical common sense text description to obtain clinical trial information semantic features and medical common sense semantic features, and then perform full-text fine-grained semantic association to obtain clinical trial information-medical common sense semantic association features;

[0012] A clinical trial information-medical common sense semantic association unit, used to perform local semantic association significant enhancement on the clinical trial information-medical common sense semantic association feature to obtain a local significant representation of the clinical trial information-medical common sense semantic association;

[0013] The test result generating unit is used to obtain the test result based on the local significant representation of the clinical trial information-medical common sense semantic association, and the test result is used to indicate whether there is a logical error in the clinical trial information data.

[0014] In the above-mentioned clinical data collection and analysis system based on feature mining, the clinical trial medical common sense semantic encoding unit includes: a clinical trial information semantic encoding subunit, which is used to semantically encode the clinical trial information data to obtain a clinical trial information semantic encoding vector as the clinical trial information semantic feature; a medical common sense text description semantic encoding subunit, which is used to semantically encode the medical common sense text description to obtain a medical common sense semantic encoding vector as the medical common sense semantic feature; a clinical trial information-medical common sense semantic association subunit, which is used to input the clinical trial information semantic encoding vector and the medical common sense semantic encoding vector into a full-text fine-grained semantic association network to obtain a clinical trial information-medical common sense semantic association matrix as the clinical trial information-medical common sense semantic association feature; wherein, the clinical trial information-medical common sense semantic association subunit is used to multiply the clinical trial information semantic encoding vector and the transposed vector of the medical common sense semantic encoding vector to obtain a matrix as the clinical trial information-medical common sense semantic association matrix.

[0015] In the above-mentioned clinical data collection and analysis system based on feature mining, the clinical experiment information-medical common sense semantic association unit includes: a clinical experiment information-medical common sense feature decomposition subunit, which is used to perform feature decomposition on the clinical experiment information-medical common sense semantic association matrix to obtain a set of clinical experiment information-medical common sense local semantic association feature vectors; a clinical experiment information-medical common sense semantic distance association subunit, which is used to perform semantic distance span association on the set of clinical experiment information-medical common sense local semantic association feature vectors to obtain a clinical experiment information-medical common sense semantic association score modulation matrix; a clinical experiment information-medical common sense semantic enhancement subunit, which is used to perform local convolution significant feature extraction on the set of clinical experiment information-medical common sense local semantic association feature vectors and the clinical experiment information-medical common sense semantic association score modulation matrix to obtain a clinical experiment information-medical common sense semantic association local significant matrix as the clinical experiment information-medical common sense semantic association local significant representation.

[0016] In the above-mentioned clinical data collection and analysis system based on feature mining, the clinical experiment information-medical common sense semantic distance association subunit includes the clinical experiment information-medical common sense semantic distance association subunit, including: a clinical experiment information-medical common sense semantic association score secondary subunit, used to calculate the semantic association score between any two clinical experiment information-medical common sense local semantic association feature vectors in the set of the clinical experiment information-medical common sense local semantic association feature vectors to obtain a clinical experiment information-medical common sense local semantic association score matrix; a clinical experiment information-medical common sense attenuation modulation secondary subunit, used to attenuate and modulate the clinical experiment information-medical common sense local semantic association score matrix based on the distance between any two clinical experiment information-medical common sense local semantic association feature vectors in the set of the clinical experiment information-medical common sense local semantic association feature vectors to obtain the clinical experiment information-medical common sense semantic association score modulation matrix.

[0017] In the above-mentioned clinical data collection and analysis system based on feature mining, the clinical experiment information-medical common sense semantic association score secondary sub-unit includes: a clinical experiment information-medical common sense semantic mapping tertiary sub-unit, which is used to map each clinical experiment information-medical common sense local semantic association feature vector in the set of clinical experiment information-medical common sense local semantic association feature vectors to the Poincare space to obtain a set of mapped clinical experiment information-medical common sense local semantic association feature vectors; a clinical experiment information-medical common sense semantic calculation tertiary sub-unit, which is used to calculate the local semantic association score between any two mapped clinical experiment information-medical common sense local semantic association feature vectors in the set of mapped clinical experiment information-medical common sense local semantic association feature vectors to obtain the clinical experiment information-medical common sense local semantic association score matrix composed of multiple clinical experiment information-medical common sense local semantic association score values.

[0018] In the above-mentioned clinical data collection and analysis system based on feature mining, the clinical experiment information-medical common sense semantic mapping three-level sub-unit is used to: after matrix-vector multiplication of the first weight matrix and the clinical experiment information-medical common sense local semantic association feature vector, the obtained clinical experiment information-medical common sense local semantic association weight vector is vector-matrix multiplied with the second weight matrix to obtain the mapped clinical experiment information-medical common sense local semantic association feature vector.

[0019] In the above-mentioned clinical data collection and analysis system based on feature mining, the clinical experiment information-medical common sense semantic calculation tertiary subunit is used to: extract the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector from the set of mapped clinical experiment information-medical common sense local semantic association feature vectors; calculate the positional difference between the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector to obtain the mapped clinical experiment information-medical common sense semantic difference vector, and calculate the square of the one-norm of the mapped clinical experiment information-medical common sense semantic difference vector to obtain the mapped clinical experiment information-medical common sense semantic difference norm value; calculate the difference between the constant one and the square of the one-norm of the i-th mapped clinical experiment information-medical common sense local semantic association feature vector to obtain the j-th mapped clinical experiment information-medical common sense semantic difference value; calculate the constant The difference between one and the square of the norm of one of the i-th mapped clinical experiment information-medical common sense local semantic association feature vector is used to obtain the j-th mapped clinical experiment information-medical common sense semantic difference value; the product between the i-th mapped clinical experiment information-medical common sense semantic difference value and the j-th mapped clinical experiment information-medical common sense semantic difference value is calculated to obtain the mapped clinical experiment information-medical common sense semantic product value; after dividing the mapped clinical experiment information-medical common sense semantic difference norm value by the mapped clinical experiment information-medical common sense semantic product value, the obtained value is first multiplied by a constant two and then added to a constant one to obtain the mapped clinical experiment information-medical common sense semantic measurement value; the mapped clinical experiment information-medical common sense semantic measurement value is input into the arccosh function to obtain the clinical experiment information-medical common sense local semantic association score value between the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector.

[0020] In the above-mentioned clinical data acquisition and analysis system based on feature mining, the clinical experiment information-medical common sense attenuation modulation secondary subunit is used to: calculate the distance between any two clinical experiment information-medical common sense local semantic association feature vectors to obtain a set of clinical experiment information-medical common sense distance values; calculate an exponential function with the natural constant e as the base, with each clinical experiment information-medical common sense distance value in the set of clinical experiment information-medical common sense distance values ​​as the exponent to obtain a set of clinical experiment information-medical common sense distance index values; multiply the set of clinical experiment information-medical common sense distance index values ​​by the attenuation modulation value by position to obtain a set of clinical experiment information-medical common sense distance weighted values; divide each clinical experiment information-medical common sense local semantic association score value in the clinical experiment information-medical common sense local semantic association score matrix by the corresponding clinical experiment information-medical common sense distance weighted value to obtain the clinical experiment information-medical common sense semantic association score modulation matrix composed of multiple clinical experiment information-medical common sense local semantic association score values.

[0021] In the above-mentioned clinical data collection and analysis system based on feature mining, the clinical experiment information-medical common sense empty semantic enhancer unit is used to: perform hollow convolution encoding on the clinical experiment information-medical common sense semantic association score modulation matrix to obtain a clinical experiment information-medical common sense semantic association score feature matrix; input the set of clinical experiment information-medical common sense local semantic association feature vectors and the clinical experiment information-medical common sense semantic association score feature matrix into the graph convolution encoding module to obtain a set of clinical experiment information-medical common sense local context association feature vectors; perform feature reshaping on the set of clinical experiment information-medical common sense local context association feature vectors to obtain the clinical experiment information-medical common sense semantic association local saliency matrix.

[0022] In the above-mentioned clinical data collection and analysis system based on feature mining, the test result generation unit is used to: input the clinical experiment information-medical common sense semantic association local significance matrix into the classifier-based logic test unit to obtain the test result, and the test result is used to indicate whether there is a logical error in the clinical trial information data.

[0023] Compared with the prior art, the present application provides a clinical data collection and analysis system based on feature mining, which obtains clinical trial information data, extracts medical common sense text descriptions from an external knowledge base, and uses deep learning data processing and encoding technology based on feature mining to semantically encode the clinical trial information data and the medical common sense text description, and then performs fine-grained semantic association between the semantically encoded clinical trial information semantic features and medical common sense semantic features, so as to intelligently judge whether there are logical errors in the clinical trial information data based on the representation obtained by local semantic enhancement of the clinical trial information-medical common sense semantic association features. In this way, through semantic association analysis, not only can the direct logical relationship within the data be captured, but also the implicit association between the data can be revealed, making the understanding of clinical trial information more comprehensive and in-depth. In addition, it can continuously adjust and optimize itself with the input of new data to adapt to changes in clinical trial design, without the need to frequently manually update the inspection rules, thereby improving the flexibility and intelligence of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0025] Figure 1 It is a block diagram of a clinical data collection and analysis system based on feature mining according to an embodiment of the present application.

[0026] Figure 2 It is a block diagram of a data review module in a clinical data collection and analysis system based on feature mining according to an embodiment of the present application.

[0027] Figure 3 This is a schematic diagram of data flow in a data review module in a clinical data collection and analysis system based on feature mining according to an embodiment of the present application.

[0028] Figure 4 It is a block diagram of a clinical trial medical common sense semantic encoding unit in a clinical data collection and analysis system based on feature mining according to an embodiment of the present application.

[0029] Figure 5 It is a block diagram of the clinical experiment information-medical common sense semantic association unit in the clinical data collection and analysis system based on feature mining according to an embodiment of the present application. DETAILED DESCRIPTION

[0030] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0031] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0032] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0033] Based on this, this application proposes a clinical data collection and analysis system based on feature mining. Figure 1 FIG. 1 is a block diagram of a clinical data collection and analysis system based on feature mining according to an embodiment of the present application. Figure 1 As shown, according to the clinical data collection and analysis system 100 based on feature mining according to the embodiment of the present application, it specifically includes: a personnel registration module 110, used to register the information of the subjects; a data collection module 120, used to collect the clinical trial information data of the subjects; a data review module 130, used to review and inspect the clinical trial information data to obtain the clinical trial information data that passes the review and inspection; a data storage module 140, used to store the clinical trial information data that passes the review and inspection; a data analysis module 150, used to process and analyze the clinical trial information data that passes the review and inspection.

[0034] Specifically, in the embodiment of the present application, the personnel registration module 110 is used to register the information of the subjects. It should be understood that the information of the subjects specifically includes but is not limited to the following: name, gender, age, address, contact information, emergency contact, and historical cases and physical examination reports. Among them, the physical examination report refers to the result report of various physical examinations conducted by the subjects before being certified as qualified participants. This information can be used to comprehensively evaluate the health status of the subjects to determine whether they are suitable for participating in clinical trials.

[0035] Specifically, in the embodiment of the present application, the data collection module 120 is used to collect clinical trial information data of the subjects. Accordingly, the data collected during the clinical trial are the health status and drug response of the subjects at different stages. For example, symptom scores and temperature measurements before and after treatment can be used for subsequent data review.

[0036] Correspondingly, in the data review module, the technical concept of the present application is to obtain clinical trial information data, extract medical common sense text descriptions from an external knowledge base, and use deep learning data processing and encoding technology based on feature mining to semantically encode the clinical trial information data and the medical common sense text description, and then perform fine-grained semantic association between the semantically encoded clinical trial information semantic features and medical common sense semantic features, so as to intelligently judge whether there are logical errors in the clinical trial information data based on the local semantic enhancement of the clinical trial information-medical common sense semantic association features. In this way, through semantic association analysis, not only can the direct logical relationship within the data be captured, but also the implicit association between the data can be revealed, making the understanding of clinical trial information more comprehensive and in-depth. In addition, it can continuously adjust and optimize itself with the input of new data to adapt to changes in clinical trial design, without the need to frequently manually update the inspection rules, thereby improving the flexibility and intelligence of the system.

[0037] Figure 2 It is a block diagram of a data review module in a clinical data collection and analysis system based on feature mining according to an embodiment of the present application. Figure 3 FIG. 1 is a schematic diagram of data flow in a data audit module in a clinical data collection and analysis system based on feature mining according to an embodiment of the present application. Figure 2 and Figure 3 As shown, the data review module 130 includes: a medical common sense text description extraction unit 131, which is used to extract medical common sense text descriptions from an external knowledge base; a clinical trial medical common sense semantic encoding unit 132, which is used to semantically encode the clinical trial information data and the medical common sense text description to obtain clinical trial information semantic features and medical common sense semantic features, and then perform full-text fine-grained semantic association to obtain clinical trial information-medical common sense semantic association features; a clinical trial information-medical common sense semantic association unit 133, which is used to significantly enhance the local semantic association of the clinical trial information-medical common sense semantic association features to obtain a local significant representation of the clinical trial information-medical common sense semantic association; a test result generation unit 134, which is used to obtain a test result based on the local significant representation of the clinical trial information-medical common sense semantic association, and the test result is used to indicate whether there is a logical error in the clinical trial information data.

[0038] In the embodiment of the present application, the medical common sense text description extraction unit 131 is used to extract medical common sense text descriptions from an external knowledge base. It should be understood that medical common sense text descriptions usually come from authoritative medical literature, textbooks, clinical guidelines or other forms of medical knowledge bases. These text descriptions provide knowledge of standard practices in the medical field, known disease characteristics, drug action mechanisms, etc. Based on this, in order to use medical common sense text descriptions as a reference standard to accurately process and analyze the clinical trial information data review process, and ensure that the data is not only numerically correct, but also conforms to medical logic, in the technical solution of the present application, clinical trial information data is obtained, and medical common sense text descriptions are extracted from an external knowledge base.

[0039] Specifically, in an embodiment of the present application, the clinical trial medical common sense semantic encoding unit 132 is used to semantically encode the clinical trial information data and the medical common sense text description to obtain clinical trial information semantic features and medical common sense semantic features, and then perform full-text fine-grained semantic association to obtain clinical trial information-medical common sense semantic association features. Figure 4 is a block diagram of a clinical trial medical common sense semantic encoding unit in a clinical data collection and analysis system based on feature mining according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 4 As shown, the clinical trial medical common sense semantic encoding unit 132 includes: a clinical trial information semantic encoding subunit 1321, which is used to semantically encode the clinical trial information data to obtain a clinical trial information semantic encoding vector as the clinical trial information semantic feature; a medical common sense text description semantic encoding subunit 1322, which is used to semantically encode the medical common sense text description to obtain a medical common sense semantic encoding vector as the medical common sense semantic feature; a clinical trial information-medical common sense semantic association subunit 1323, which is used to input the clinical trial information semantic encoding vector and the medical common sense semantic encoding vector into a full-text fine-grained semantic association network to obtain a clinical trial information-medical common sense semantic association matrix as the clinical trial information-medical common sense semantic association feature.

[0040] Specifically, the clinical trial information semantic encoding subunit 1321 is used to semantically encode the clinical trial information data to obtain a clinical trial information semantic encoding vector as the clinical trial information semantic feature. Accordingly, taking into account that the clinical trial information data contains rich semantic associations between contexts, and in order to more clearly understand and accurately analyze the contextual semantic meanings contained in the clinical trial information, in the technical solution of the present application, the clinical trial information data is semantically encoded to obtain a clinical trial information semantic encoding vector. In particular, in a specific embodiment of the present application, the clinical trial information data can be semantically encoded and contextually understood by using a semantic encoder based on the BERT model to obtain a clinical trial information semantic encoding vector containing more semantic information.

[0041] Specifically, the medical common sense text description semantic encoding subunit 1322 is used to semantically encode the medical common sense text description to obtain a medical common sense semantic encoding vector as the medical common sense semantic feature. It should be understood that, considering that the medical common sense text description is composed of multiple medical common sense words or phrases, and each word contains semantic information of the context environment, therefore, in order to more accurately understand the contextual semantic meaning in the text description, in the technical solution of the present application, the medical common sense text description is semantically encoded to obtain a medical common sense semantic encoding vector. In particular, in a specific embodiment of the present application, the medical common sense text description is segmented and then input into a semantic encoding module containing a word embedding layer to divide an entire text into a local word sequence composed of multiple words, and the context information in each word is semantically associated and understood to obtain a comprehensive medical common sense semantic encoding vector of the entire text semantics.

[0042] Specifically, the clinical experiment information-medical common sense semantic association subunit 1323 is used to input the clinical experiment information semantic encoding vector and the medical common sense semantic encoding vector into the full-text fine-grained semantic association network to obtain the clinical experiment information-medical common sense semantic association matrix as the clinical experiment information-medical common sense semantic association feature. More specifically, in an embodiment of the present application, the clinical experiment information-medical common sense semantic association subunit 1323 is used to multiply the clinical experiment information semantic encoding vector and the transposed vector of the medical common sense semantic encoding vector to obtain a matrix as the clinical experiment information-medical common sense semantic association matrix. Accordingly, it is considered that the clinical experiment information semantic encoding vector and the medical common sense semantic encoding vector are only semantically associated. For example, if the semantic feature of the clinical data indicates that a certain drug has an undesirable side effect in a certain patient, and medical common sense indicates that the drug should not have such a side effect, then this can be marked as a potential logical error. In order to analyze the semantic association between the semantic representation and medical common sense in the clinical experiment data in a more detailed and fine-grained manner, so as to improve the accuracy and limitation of the data, in the technical solution of the present application, the clinical experiment information semantic encoding vector and the medical common sense semantic encoding vector are input into the full-text fine-grained semantic association network to capture the more complex semantic relationship between the two in a more fine-grained manner, and obtain the clinical experiment information-medical common sense semantic association matrix. In particular, in a specific embodiment of the present application, the matrix obtained by multiplying the transposed vector of the clinical experiment information semantic encoding vector and the medical common sense semantic encoding vector is used as the clinical experiment information-medical common sense semantic association matrix, so that each element in the obtained matrix represents the correlation or similarity between a certain feature in the clinical data and the corresponding feature in the medical common sense, so as to better quantify the degree of semantic similarity between the two sets of data, thereby helping to identify the part of the clinical data that is consistent with medical common sense.

[0043] In an embodiment of the present application, the clinical experiment information-medical common sense semantic association unit 133 is used to perform local semantic association significant enhancement on the clinical experiment information-medical common sense semantic association feature to obtain a local significant representation of the clinical experiment information-medical common sense semantic association. Figure 5 : is a block diagram of a clinical experimental information-medical common sense semantic association unit in a clinical data collection and analysis system based on feature mining according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 5As shown, the clinical experiment information-medical common sense semantic association unit 133 includes: a clinical experiment information-medical common sense feature decomposition subunit 1331, which is used to perform feature decomposition on the clinical experiment information-medical common sense semantic association matrix to obtain a set of clinical experiment information-medical common sense local semantic association feature vectors; a clinical experiment information-medical common sense semantic distance association subunit 1332, which is used to perform semantic distance span association on the set of clinical experiment information-medical common sense local semantic association feature vectors to obtain a clinical experiment information-medical common sense semantic association score modulation matrix; a clinical experiment information-medical common sense semantic enhancement subunit 1333, which is used to perform local convolution significant feature extraction on the set of clinical experiment information-medical common sense local semantic association feature vectors and the clinical experiment information-medical common sense semantic association score modulation matrix to obtain a clinical experiment information-medical common sense semantic association local significant matrix as the clinical experiment information-medical common sense semantic association local significant representation. It should be understood that, considering that the clinical experiment information-medical common sense semantic association matrix is ​​obtained through fine-grained interaction, there is a significant local semantic association relationship between the two, therefore, in order to further enhance the local significant features between the data of the two, so as to better detect and identify logical errors or abnormalities in clinical data. In the technical solution of the present application, the local semantic association of the clinical experiment information-medical common sense semantic association features is significantly enhanced to obtain the clinical experiment information-medical common sense semantic association local significant matrix as the clinical experiment information-medical common sense semantic association local significant representation.

[0044] Specifically, first, the clinical trial information-medical common sense semantic association matrix is ​​feature decomposed to simplify the complex relationship in a horizontal matrix into a set of feature vectors, each feature vector represents a local semantic association pattern, and a set of clinical trial information-medical common sense local semantic association feature vectors is obtained, which lays the foundation for the subsequent local significant enhancement of key features. In particular, in a specific embodiment of the present application, the clinical trial information-medical common sense semantic association matrix is ​​segmented based on row vectors or column vectors to obtain a set of clinical trial information-medical common sense local semantic association feature vectors.

[0045] Next, the set of the local semantic association feature vectors of the clinical experiment information-medical common sense is subjected to semantic distance span association to obtain the clinical experiment information-medical common sense semantic association score modulation matrix, so as to amplify and highlight the influence of these features through span association. Specifically, the set of the local semantic association feature vectors of the clinical experiment information-medical common sense is mapped to the Poincare space, and the Poincare distance between any two mapped features is used as the semantic score value between any two vectors to obtain the local semantic association score matrix of the clinical experiment information-medical common sense composed of multiple semantic association score values. It should be understood that the Poincare space is a non-Euclidean geometric space, which is particularly suitable for capturing data with a hierarchical structure. By mapping the feature vector to the Poincare space, the hierarchical relationship and complex dependency relationship in the clinical data can be better represented. In the Poincare space, the Poincare distance can better capture the semantic similarity and difference, that is, by calculating the Poincare distance, a more accurate semantic association score matrix can be obtained, thereby improving the semantic consistency between clinical data and medical common sense. Subsequently, the clinical experiment information-medical common sense local semantic association score matrix is ​​attenuated and modulated based on the distance between any two mapped features to highlight the significance of features with similar distances, and the clinical experiment information-medical common sense semantic association score modulation matrix is ​​obtained.

[0046] Then, the set of the clinical experiment information-medical common sense local semantic association feature vectors and the clinical experiment information-medical common sense semantic association score modulation matrix are subjected to local convolution significant feature extraction to obtain the clinical experiment information-medical common sense semantic association local significant matrix as the clinical experiment information-medical common sense semantic association local significant representation. Specifically, first, the void convolution coding is used to capture the association between each original local semantic feature, and then the void convolution coded features and the set of clinical experiment information-medical common sense local semantic association feature vectors are input into the graph convolution coding module to enhance the global semantic context association feature expression between the features to obtain the set of clinical experiment information-medical common sense local context association feature vectors. Finally, it is reshaped to obtain the final clinical experiment information-medical common sense semantic association local significant matrix.

[0047] Specifically, the clinical experiment information-medical common sense semantic distance association subunit 1332 includes: a clinical experiment information-medical common sense semantic association score secondary subunit, which is used to calculate the semantic association score between any two clinical experiment information-medical common sense local semantic association feature vectors in the set of the clinical experiment information-medical common sense local semantic association feature vectors to obtain a clinical experiment information-medical common sense local semantic association score matrix; a clinical experiment information-medical common sense attenuation modulation secondary subunit, which is used to attenuate and modulate the clinical experiment information-medical common sense local semantic association score matrix based on the distance between any two clinical experiment information-medical common sense local semantic association feature vectors in the set of the clinical experiment information-medical common sense local semantic association feature vectors to obtain the clinical experiment information-medical common sense semantic association score modulation matrix.

[0048] Specifically, in an embodiment of the present application, the clinical experiment information-medical common sense semantic association score secondary sub-unit includes: a clinical experiment information-medical common sense semantic mapping tertiary sub-unit, which is used to map each clinical experiment information-medical common sense local semantic association feature vector in the set of clinical experiment information-medical common sense local semantic association feature vectors to the Poincare space to obtain a set of mapped clinical experiment information-medical common sense local semantic association feature vectors; a clinical experiment information-medical common sense semantic calculation tertiary sub-unit, which is used to calculate the local semantic association score between any two mapped clinical experiment information-medical common sense local semantic association feature vectors in the set of mapped clinical experiment information-medical common sense local semantic association feature vectors to obtain the clinical experiment information-medical common sense local semantic association score matrix composed of multiple clinical experiment information-medical common sense local semantic association score values.

[0049] More specifically, in an embodiment of the present application, the clinical experiment information-medical common sense semantic mapping third-level sub-unit is used to: after matrix-vector multiplication of the first weight matrix and the clinical experiment information-medical common sense local semantic association feature vector, the obtained clinical experiment information-medical common sense local semantic association weight vector is vector-matrix multiplied with the second weight matrix to obtain the mapped clinical experiment information-medical common sense local semantic association feature vector.

[0050] More specifically, in an embodiment of the present application, the clinical experiment information-medical common sense semantic calculation tertiary subunit is used to: extract the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector from the set of mapped clinical experiment information-medical common sense local semantic association feature vectors; calculate the positional difference between the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector to obtain the mapped clinical experiment information-medical common sense semantic difference vector, and calculate the square of the one-norm of the mapped clinical experiment information-medical common sense semantic difference vector to obtain the mapped clinical experiment information-medical common sense semantic difference norm value; calculate the difference between a constant one and the square of the one-norm of the i-th mapped clinical experiment information-medical common sense local semantic association feature vector to obtain the i-th mapped clinical experiment information-medical common sense semantic difference value; calculate the constant one and the The difference between the square of the norm of the jth mapped clinical experiment information-medical common sense local semantic association feature vector is used to obtain the jth mapped clinical experiment information-medical common sense semantic difference value; the product between the ith mapped clinical experiment information-medical common sense semantic difference value and the jth mapped clinical experiment information-medical common sense semantic difference value is calculated to obtain the mapped clinical experiment information-medical common sense semantic product value; after dividing the mapped clinical experiment information-medical common sense semantic difference norm value by the mapped clinical experiment information-medical common sense semantic product value, the obtained value is first multiplied by a constant two and then added to a constant one to obtain the mapped clinical experiment information-medical common sense semantic measurement value; the mapped clinical experiment information-medical common sense semantic measurement value is input into the arccosh function to obtain the clinical experiment information-medical common sense local semantic association score value between the ith mapped clinical experiment information-medical common sense local semantic association feature vector and the jth mapped clinical experiment information-medical common sense local semantic association feature vector.

[0051] Specifically, in an embodiment of the present application, the clinical experiment information-medical common sense attenuation modulation secondary subunit is used to: calculate the distance between any two clinical experiment information-medical common sense local semantic association feature vectors to obtain a set of clinical experiment information-medical common sense distance values; calculate an exponential function with the natural constant e as the base, with each clinical experiment information-medical common sense distance value in the set of clinical experiment information-medical common sense distance values ​​as the exponent to obtain a set of clinical experiment information-medical common sense distance index values; multiply the set of clinical experiment information-medical common sense distance index values ​​by the attenuation modulation value by position to obtain a set of clinical experiment information-medical common sense distance weighted values; divide each clinical experiment information-medical common sense local semantic association score value in the clinical experiment information-medical common sense local semantic association score matrix by the corresponding clinical experiment information-medical common sense distance weighted value to obtain the clinical experiment information-medical common sense semantic association score modulation matrix composed of multiple clinical experiment information-medical common sense local semantic association score values.

[0052] Specifically, the clinical experiment information-medical common sense empty semantic enhancer unit 1333 is used to: perform void convolution coding on the clinical experiment information-medical common sense semantic association score modulation matrix to obtain a clinical experiment information-medical common sense semantic association score feature matrix; input the set of clinical experiment information-medical common sense local semantic association feature vectors and the clinical experiment information-medical common sense semantic association score feature matrix into a graph convolution coding module to obtain a set of clinical experiment information-medical common sense local context association feature vectors; perform feature reshaping on the set of clinical experiment information-medical common sense local context association feature vectors to obtain the clinical experiment information-medical common sense semantic association local saliency matrix.

[0053] In the embodiment of the present application, specifically, the local semantic association of the clinical trial information-medical common sense semantic association matrix is ​​significantly enhanced, which can be expressed by the formula:

[0054] X=decouple(M a ) = {x1, x2, ..., x i , ..., x j , ..., x n}

[0055] h i =W1x i W2

[0056] H = {h1, h2, ..., h i , ..., h j , ..., h n}

[0057]

[0058]

[0059] T = AtrousConv(S')

[0060] X' = ​​GCN(X, T)

[0061] M a ' = reshape[X']

[0062] Among them, M a is the clinical trial information-medical common sense semantic association matrix, decouple(·) is the feature decomposition operation, x1, x2, ..., x i , ..., x j , ..., x n is the first, second, ..., i-th, ..., j-th, ..., n-th clinical experiment information-medical common sense local semantic association feature vector in the set of clinical experiment information-medical common sense local semantic association feature vectors, X is the set of clinical experiment information-medical common sense local semantic association feature vectors, W1 and W2 are the first weight matrix and the second weight matrix respectively, H is the set of clinical experiment information-medical common sense local semantic association feature vectors after mapping, h1, h2, ..., h i , ..., h j , ..., h n are respectively the first, second, ..., ith, ..., jth, ..., nth mapped clinical experiment information-medical common sense local semantic association feature vectors in the set of mapped clinical experiment information-medical common sense local semantic association feature vectors, ||·|| 2 is the square of the one-norm of the vector, arccosh(·) is the inverse hyperbolic cosine function, d P (h i ,h j ) is used to calculate h i and h j The local semantic association score between clinical trial information and medical knowledge, S ij is the eigenvalue of each position in the local semantic association score matrix of clinical trial information-medical knowledge, d(x i , x j ) is x i and x j The clinical experimental information-medical knowledge distance value between them, α is the attenuation modulation value, S ij' is the eigenvalue of each position in the clinical trial information-medical common sense semantic association score modulation matrix, S' is the clinical trial information-medical common sense semantic association score modulation matrix, AtrousConv(·) is the void convolution coding operation, T is the clinical trial information-medical common sense semantic association score feature matrix, GCN(·,·) is the graph convolution coding operation, X' is the set of clinical trial information-medical common sense local context association feature vectors, reshape[·] is the shape reshaping operation, M a ' is the local significant matrix of the clinical trial information-medical common sense semantic association.

[0063] In the embodiment of the present application, the test result generating unit 134 is used to obtain the test result based on the local significant representation of the clinical trial information-medical common sense semantic association, and the test result is used to indicate whether there is a logical error in the clinical trial information data. Specifically, in the embodiment of the present application, the test result generating unit is used to: input the local significant matrix of the clinical trial information-medical common sense semantic association into the logic test unit based on the classifier to obtain the test result, and the test result is used to indicate whether there is a logical error in the clinical trial information data. That is, the clinical trial information-medical common sense semantic association matrix is ​​used to perform local semantic significant enhancement to obtain the local significant representation of the clinical trial information-medical common sense semantic association for classification processing, so as to intelligently judge whether there is a logical error in the clinical trial information data. In this way, through semantic association analysis, not only the direct logical relationship within the data can be captured, but also the implicit association between the data can be revealed, so that the understanding of the clinical trial information is more comprehensive and in-depth. In addition, it can continuously adjust and optimize itself with the input of new data to adapt to changes in clinical trial design, without frequent manual update of inspection rules, thereby improving the flexibility and intelligence of the system.

[0064] It should be understandable that when the clinical trial information semantic encoding vector and the medical common sense semantic encoding vector respectively represent the text semantic context encoding features of the medical common sense text description and the clinical trial information data, when the full-text fine-grained semantic associations and local semantic associations are significantly enhanced, the clinical trial information-medical common sense semantic association local significant matrix will also cause the diversity of full-text fine-grained semantic associations due to the differences in the fine-grained text expression, affecting the iterative consistency of the classification mapping and reducing the accuracy of the classification results.

[0065] Preferably, the method of inputting the clinical experiment information-medical common sense semantic association local significant vector into a logic verification unit based on a classifier to obtain a test result includes: flattening the features of the clinical experiment information-medical common sense semantic association local significant matrix to obtain a clinical experiment information-medical common sense semantic association local significant vector; determining the clinical experiment information-medical common sense semantic association probability value obtained by inputting the clinical experiment information-medical common sense semantic association local significant vector into a logic verification unit based on a classifier, wherein the clinical experiment information-medical common sense semantic association probability value indicates the probability of a logical error in the clinical trial information data; calculating the clinical experiment information-medical common sense semantic association local significant vector. The product of the first clinical experiment information-medical common sense semantic association feature value of the local significant vector and the clinical experiment information-medical common sense semantic association probability value is calculated to obtain the first clinical experiment information-medical common sense semantic association feature probability product; the product of the second clinical experiment information-medical common sense semantic association feature value of the local significant vector of the clinical experiment information-medical common sense semantic association and one minus the clinical experiment information-medical common sense semantic association probability value is calculated to obtain the second clinical experiment information-medical common sense semantic association feature inverse probability product; the first clinical experiment information-medical common sense semantic association feature value is calculated divided by one minus the difference between the clinical experiment information-medical common sense semantic association probability value. The quotient is calculated to obtain the inverse probability quotient of the first clinical experiment information-medical common sense semantic association feature; the quotient of the second clinical experiment information-medical common sense semantic association feature value and the clinical experiment information-medical common sense semantic association probability value is calculated to obtain the second clinical experiment information-medical common sense semantic association feature probability quotient; the product of the first clinical experiment information-medical common sense semantic association feature probability and the product of the second clinical experiment information-medical common sense semantic association feature inverse probability are added, and then the difference between the first clinical experiment information-medical common sense semantic association feature inverse probability quotient and the second clinical experiment information-medical common sense semantic association feature probability quotient is subtracted to obtain the clinical experiment information-medical common sense semantic association feature. Linked correction feature values; for each first clinical experiment information-medical common sense semantic association feature value and second clinical experiment information-medical common sense semantic association feature value of the clinical experiment information-medical common sense semantic association local significant vector, matrix multiplication of the clinical experiment information-medical common sense semantic association correction feature matrix composed of the clinical experiment information-medical common sense semantic association correction feature values ​​and the clinical experiment information-medical common sense semantic association local significant vector is performed to obtain a corrected clinical experiment information-medical common sense semantic association local significant vector; the corrected clinical experiment information-medical common sense semantic association local significant vector is input into a classifier-based logic test unit to obtain a test result.

[0066] The correction process of the clinical experiment information-medical common sense semantic association local significant vector V is expressed as follows:

[0067]

[0068] m i，j ∈M,v i ∈V,v j(j≠i) ∈V

[0069]

[0070] Where V is the local significant vector of the clinical trial information-medical common sense semantic association, v i and v j are the eigenvalues ​​of the i-th and j-th positions in the local significant vector of the clinical trial information-medical common sense semantic association, p is the probability that the clinical trial information data contains logical errors, m i，j represents the clinical experiment information-medical common sense semantic association correction feature value at the (i, j)th position in the clinical experiment information-medical common sense semantic association correction feature matrix, M is the clinical experiment information-medical common sense semantic association correction feature matrix, represents matrix multiplication, and V' represents the corrected local significant vector of clinical experiment information-medical common sense semantic association.

[0071] Therefore, the present application is aimed at the region-boundary integral relationship of the high-dimensional feature manifold of the local significant matrix of the clinical trial information-medical common sense semantic association, and the probability distribution of the feature set of the local significant matrix of the clinical trial information-medical common sense semantic association is defined as a manifold constraint boundary to approximate the high-dimensional manifold of the feature set of the local significant matrix of the clinical trial information-medical common sense semantic association, which is composed of eigenvalue pairs. The single-connected region representation is used to avoid the mapping ambiguity of the diversified feature representation of the local significant matrix of the clinical trial information-medical common sense semantic association to the local manifold representation in the high-dimensional feature class convergence space, and improve the execution iteration consistency of each local feature distribution in the mapping task to improve the feature classification convergence effect and improve the accuracy of the test results obtained by the logic test unit based on the classifier. In this way, through semantic association analysis, not only can the direct logical relationship within the data be captured, but also the implicit association between the data can be revealed, so that the understanding of the clinical trial information is more comprehensive and in-depth. In addition, it can continuously adjust and optimize itself with the input of new data to adapt to changes in clinical trial design without frequent manual update of inspection rules, thereby improving the flexibility and intelligence of the system.

[0072] Therefore, the data review module 130 is explained, which obtains clinical trial information data, extracts medical common sense text descriptions from an external knowledge base, and uses deep learning data processing and encoding technology based on feature mining to semantically encode the clinical trial information data and the medical common sense text description, and then performs fine-grained semantic association between the semantically encoded clinical trial information semantic features and medical common sense semantic features, so as to intelligently judge whether there are logical errors in the clinical trial information data based on the representation obtained by local semantic enhancement of the clinical trial information-medical common sense semantic association features. In this way, through semantic association analysis, not only can the direct logical relationship within the data be captured, but also the implicit association between the data can be revealed, making the understanding of clinical trial information more comprehensive and in-depth. In addition, it can continuously adjust and optimize itself with the input of new data to adapt to changes in clinical trial design, without the need to frequently manually update the inspection rules, thereby improving the flexibility and intelligence of the system.

[0073] Specifically, in an embodiment of the present application, the data storage module 140 is used to store the clinical trial information data that has passed the audit and inspection. It should be understood that the stored clinical trial information data that has passed the audit and inspection mainly includes detailed information and related data of the subjects during the trial. In particular, in a specific embodiment of the present application, these data can be stored by using EpiInfo, which has powerful statistical analysis functions and can convert the data format into the data format of other commonly used databases or statistical analysis software, such as dBASE, SAS, SPSS, etc., for further data processing and analysis. The specific steps can be referred to CN117316366A. Since this step is not the focus of this application, it will not be introduced in detail.

[0074] Specifically, in the embodiment of the present application, the data analysis module 150 is used to process and analyze the clinical trial information data that has passed the audit inspection. That is, through the analysis of clinical trial data, researchers can evaluate the therapeutic effect of new drugs on target diseases, including the dose-effect relationship of drugs, the best medication regimen, etc., so as to ensure the effectiveness and safety of drug clinical trials.

[0075] In summary, a clinical data collection and analysis system 100 based on feature mining according to an embodiment of the present application is explained, which includes: a personnel registration module, a data collection module, a data review module, a data storage module and a data analysis module. Specifically, the participant information is first recorded through the personnel registration module, and then the clinical data is collected using the data collection module. Then the data review module ensures the accuracy and completeness of the information, and the data that has passed the review will be properly stored in the data storage. Finally, the data analysis module will process and analyze the stored data to extract useful information to support scientific decision-making. In this way, the quality and reliability of the clinical trial data can be ensured.

[0076] As described above, the clinical data collection and analysis system 100 based on feature mining according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a clinical data collection and analysis algorithm based on feature mining. In a possible implementation, the clinical data collection and analysis system 100 based on feature mining according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the clinical data collection and analysis system 100 based on feature mining can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the clinical data collection and analysis system 100 based on feature mining can also be one of the many hardware modules of the wireless terminal.

[0077] Alternatively, in another example, the feature mining-based clinical data acquisition and analysis system 100 and the wireless terminal may also be separate devices, and the feature mining-based clinical data acquisition and analysis system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0078] The above descriptions of various implementations of the present disclosure are exemplary and not exhaustive. The disclosure is not limited to the disclosed implementations, and many modifications and variations are obvious to those of ordinary skill in the art without departing from the scope and spirit of the described implementations. The terms used herein are selected to best explain the principles of the implementations, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the various practical methods disclosed herein.

Claims

1. A clinical data collection and analysis system based on feature mining, characterized in that: include: Personnel registration module, used to register the information of test subjects; A data collection module, used to collect clinical trial information data of the subjects; A data review module, used to review and inspect the clinical trial information data to obtain clinical trial information data that passes the review and inspection; A data storage module, used to store the clinical trial information data that has passed the review and inspection; A data analysis module, used to process and analyze the clinical trial information data that has passed the review and inspection; Wherein, the data review module includes: A medical common sense text description extraction unit, used to extract medical common sense text descriptions from an external knowledge base; A clinical trial medical common sense semantic encoding unit, used to semantically encode the clinical trial information data and the medical common sense text description to obtain clinical trial information semantic features and medical common sense semantic features, and then perform full-text fine-grained semantic association to obtain clinical trial information-medical common sense semantic association features; A clinical trial information-medical common sense semantic association unit, used to perform local semantic association significant enhancement on the clinical trial information-medical common sense semantic association feature to obtain a local significant representation of the clinical trial information-medical common sense semantic association; The test result generating unit is used to obtain the test result based on the local significant representation of the clinical trial information-medical common sense semantic association, and the test result is used to indicate whether there is a logical error in the clinical trial information data.

2. The clinical data collection and analysis system based on feature mining according to claim 1, characterized in that: The clinical trial medical common sense semantic coding unit includes: A clinical trial information semantic encoding subunit, used for semantically encoding the clinical trial information data to obtain a clinical trial information semantic encoding vector as the clinical trial information semantic feature; A medical common sense text description semantic coding subunit is used to perform semantic coding on the medical common sense text description to obtain a medical common sense semantic coding vector as the medical common sense semantic feature; A clinical trial information-medical common sense semantic association subunit, used for inputting the clinical trial information semantic encoding vector and the medical common sense semantic encoding vector into a full-text fine-grained semantic association network to obtain a clinical trial information-medical common sense semantic association matrix as the clinical trial information-medical common sense semantic association feature; Among them, the clinical experiment information-medical common sense semantic association subunit is used to obtain a matrix by multiplying the clinical experiment information semantic encoding vector and the transposed vector of the medical common sense semantic encoding vector as the clinical experiment information-medical common sense semantic association matrix.

3. The clinical data collection and analysis system based on feature mining according to claim 2 is characterized in that: The clinical trial information-medical common sense semantic association unit includes: A clinical trial information-medical common sense feature decomposition subunit, used for performing feature decomposition on the clinical trial information-medical common sense semantic association matrix to obtain a set of clinical trial information-medical common sense local semantic association feature vectors; A clinical trial information-medical common sense semantic distance association subunit, used to perform semantic distance span association on the set of clinical trial information-medical common sense local semantic association feature vectors to obtain a clinical trial information-medical common sense semantic association score modulation matrix; The clinical experiment information-medical common sense semantic enhancement subunit is used to perform local convolution significant feature extraction on the set of clinical experiment information-medical common sense local semantic association feature vectors and the clinical experiment information-medical common sense semantic association score modulation matrix to obtain a clinical experiment information-medical common sense semantic association local significant matrix as the clinical experiment information-medical common sense semantic association local significant representation.

4. The clinical data collection and analysis system based on feature mining according to claim 3 is characterized in that: The clinical trial information-medical common sense semantic distance association subunit includes: A clinical trial information-medical common sense semantic association score secondary subunit, used to calculate the semantic association score between any two clinical trial information-medical common sense local semantic association feature vectors in the set of clinical trial information-medical common sense local semantic association feature vectors to obtain a clinical trial information-medical common sense local semantic association score matrix; The clinical experiment information-medical common sense attenuation modulation secondary sub-unit is used to attenuate and modulate the clinical experiment information-medical common sense local semantic association score matrix based on the distance between any two clinical experiment information-medical common sense local semantic association feature vectors in the set of clinical experiment information-medical common sense local semantic association feature vectors to obtain the clinical experiment information-medical common sense semantic association score modulation matrix.

5. The clinical data collection and analysis system based on feature mining according to claim 4 is characterized in that: The clinical trial information-medical common sense semantic association score secondary subunit includes: A clinical experiment information-medical common sense semantic mapping tertiary subunit, used to map each clinical experiment information-medical common sense local semantic association feature vector in the set of clinical experiment information-medical common sense local semantic association feature vectors to the Poincare space to obtain a set of mapped clinical experiment information-medical common sense local semantic association feature vectors; The third-level sub-unit of clinical experiment information-medical common sense semantic calculation is used to calculate the local semantic association score between any two mapped clinical experiment information-medical common sense local semantic association feature vectors in the set of mapped clinical experiment information-medical common sense local semantic association feature vectors to obtain the clinical experiment information-medical common sense local semantic association score matrix composed of multiple clinical experiment information-medical common sense local semantic association score values.

6. The clinical data collection and analysis system based on feature mining according to claim 5 is characterized in that: The clinical experiment information-medical common sense semantic mapping third-level sub-unit is used to: after matrix-vector multiplication of the first weight matrix and the clinical experiment information-medical common sense local semantic association feature vector, the obtained clinical experiment information-medical common sense local semantic association weight vector is vector-matrix multiplied with the second weight matrix to obtain the mapped clinical experiment information-medical common sense local semantic association feature vector.

7. The clinical data collection and analysis system based on feature mining according to claim 6 is characterized in that: The clinical trial information-medical common sense semantic computing third-level subunit is used for: Extracting the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector from the set of mapped clinical experiment information-medical common sense local semantic association feature vectors; Calculate the positional difference between the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector to obtain a mapped clinical experiment information-medical common sense semantic difference vector, and calculate the square of a norm of the mapped clinical experiment information-medical common sense semantic difference vector to obtain a mapped clinical experiment information-medical common sense semantic difference norm value; Calculate the difference between the constant one and the square of the one-norm of the i-th mapped clinical experiment information-medical common sense local semantic association feature vector to obtain the i-th mapped clinical experiment information-medical common sense semantic difference; Calculate the difference between the constant one and the square of the one-norm of the j-th mapped clinical experiment information-medical common sense local semantic association feature vector to obtain the j-th mapped clinical experiment information-medical common sense semantic difference; Calculate the product of the i-th mapped clinical experiment information-medical common sense semantic difference value and the j-th mapped clinical experiment information-medical common sense semantic difference value to obtain a mapped clinical experiment information-medical common sense semantic product value; After dividing the mapped clinical experiment information-medical common sense semantic difference norm value by the mapped clinical experiment information-medical common sense semantic product value, the obtained value is first multiplied by a constant two and then added to a constant one to obtain a mapped clinical experiment information-medical common sense semantic measurement value; The mapped clinical experiment information-medical common sense semantic measurement value is input into the arccosh function to obtain the clinical experiment information-medical common sense local semantic association score value between the i-th mapped clinical experiment information-medical common sense local semantic association feature vector and the j-th mapped clinical experiment information-medical common sense local semantic association feature vector.

8. The clinical data collection and analysis system based on feature mining according to claim 7 is characterized in that: The clinical trial information-medical knowledge attenuation modulation secondary subunit is used for: Calculating the distance between any two clinical experiment information-medical common sense local semantic association feature vectors to obtain a set of clinical experiment information-medical common sense distance values; Calculate an exponential function with the natural constant e as the base, with each clinical experiment information-medical common sense distance value in the set of clinical experiment information-medical common sense distance values ​​as the exponent to obtain a set of clinical experiment information-medical common sense distance index values; Multiplying the set of clinical experiment information-medical common sense distance index values ​​by the attenuation modulation value by position to obtain a set of clinical experiment information-medical common sense distance weighted values; Each clinical experiment information-medical common sense local semantic association score value in the clinical experiment information-medical common sense local semantic association score matrix is ​​divided by the corresponding clinical experiment information-medical common sense distance weighted value to obtain the clinical experiment information-medical common sense semantic association score modulation matrix composed of multiple clinical experiment information-medical common sense local semantic association score values.

9. The clinical data collection and analysis system based on feature mining according to claim 8, characterized in that: The clinical trial information-medical common sense empty semantic enhancement subunit is used for: Performing dilated convolution coding on the clinical experiment information-medical common sense semantic association score modulation matrix to obtain a clinical experiment information-medical common sense semantic association score feature matrix; Inputting the set of clinical experiment information-medical common sense local semantic association feature vectors and the clinical experiment information-medical common sense semantic association score feature matrix into a graph convolutional coding module to obtain a set of clinical experiment information-medical common sense local context association feature vectors; The set of the clinical experiment information-medical common sense local context association feature vectors is reshaped to obtain the clinical experiment information-medical common sense semantic association local saliency matrix.

10. The clinical data collection and analysis system based on feature mining according to claim 9, characterized in that: The test result generating unit is used to input the clinical trial information-medical common sense semantic association local significant matrix into a classifier-based logic test unit to obtain the test result, and the test result is used to indicate whether there is a logical error in the clinical trial information data.

Images