A method and system for intelligent data classification and grading

Abstract

The present invention discloses a method and system for intelligent data classification and grading. The system obtains medical images to be processed and text data of corresponding cases, retrieves historical diagnosis and treatment data of each doctor in the corresponding department, obtains image labels and text labels based on the medical images and the text data respectively, fuses the image labels and the text labels to obtain hierarchical labels, retrieves the corresponding historical diagnosis and treatment data in the department for comparison based on the hierarchical labels, obtains a first recommendation index for the case, retrieves the corresponding historical diagnosis and treatment data of each doctor in the department for comparison, obtains a second recommendation index for the case, and analyzes the different comprehensive recommendation indexes of each doctor. The case is recommended and graded for different doctors according to the comprehensive recommendation index. The present invention has the characteristics of enhancing the accuracy of archiving diagnosis and treatment data and improving the utilization rate of diagnosis and treatment data.

Description

Technical Field

[0001] The present invention relates to the technical field of data classification and grading, and in particular to a method and system for intelligent data classification and grading. Background Art

[0002] In the modern medical environment, the collection, storage, processing and analysis of medical data have become particularly important. With the advancement of medical technology and the explosive growth of medical data, medical images have become an important basis for clinical diagnosis and treatment decisions. The complexity, diversity and massiveness of medical imaging data have brought a heavy burden on doctors. However, there are obvious deficiencies in the multi-source fusion and collation of medical images and case texts, which directly leads to confusion or errors in the archiving of medical data. In addition, the existing medical data processing system is also unable to provide personalized recommendations for doctors, resulting in inefficient use of medical data resources. Therefore, it is necessary to design an intelligent data classification and grading method and system to enhance the accuracy of medical data archiving and improve the utilization rate of medical data. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for intelligent data classification and grading to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a method and system for intelligent data classification and grading, which is executed in an intelligent data classification and grading method, and the operation steps of the method include:

[0005] Step S1: Obtain the medical images to be processed and the text data of the corresponding cases, retrieve the historical diagnosis and treatment data of each doctor in the corresponding department, and store the historical diagnosis and treatment data in a hierarchical structure. Each level is named with a different label, and each lower level is a further subdivision of the classification of the upper level.

[0006] Step S2: acquiring image labels and text labels based on the medical image and the text data respectively, fusing the image labels and the text labels to obtain hierarchical labels;

[0007] Step S3: based on the hierarchical label, retrieve the corresponding historical diagnosis and treatment data in the department for comparison to obtain the first recommendation index of the case, retrieve the corresponding historical diagnosis and treatment data of each doctor in the department for comparison to obtain the second recommendation index of the case;

[0008] Step S4: Based on the first recommendation index and the second recommendation index, different comprehensive recommendation indexes are obtained for each doctor, and different recommendations and graded archives are made for different doctors according to the comprehensive recommendation indexes.

[0009] According to the above technical solution, step S2 further includes the following steps:

[0010] Step S21: After preprocessing the medical image based on the medical image analysis terminal, the medical image is abnormally segmented by the U-Net algorithm to obtain a segmented image, the segmented image is analyzed for abnormal symptoms, a corresponding description text is generated according to the abnormal symptoms, an image feature vector is extracted according to the description text, and the image feature vector is matched with the image label according to the existing label stored in the hierarchical manner;

[0011] Step S22: After preprocessing the text data based on the text data analysis terminal, a text feature vector is extracted from the text data through a transformer model, and the text feature vector is matched with the text label according to the existing label stored in the hierarchical manner;

[0012] Step S23: Based on the hierarchical list of each label in the hierarchical storage, the image label and the text label are merged, duplicate labels are removed to obtain a unified label set, and the label set is organized into multi-level hierarchical labels according to the hierarchical relationship.

[0013] According to the above technical solution, step S21 further includes the following steps:

[0014] Step S211: classifying the medical image pixel by pixel based on the U-Net model, distinguishing normal and abnormal areas in the medical image, marking the abnormal areas according to their contours, and performing abnormal segmentation to obtain the segmented image;

[0015] Step S212: identifying and analyzing the abnormal symptoms of the segmented image through a deep learning model. In the abnormal segmentation process, generating a corresponding description text according to the identified abnormal symptoms, the description text including the abnormal description and the medical symptoms. The identification color of the abnormal area contour is adjusted according to the common proportion of the medical symptoms. The larger the proportion of the common situation, the lighter the identification color, and the smaller the proportion of the common situation, the darker the identification color.

[0016] Step S213: extracting an image feature vector based on the description text, matching the image feature vector with the image label according to the existing label, constructing a label feature space containing all the existing labels, matching the image feature vector with the existing label in the label feature space according to the similarity, obtaining a plurality of image labels, setting a similarity threshold, and when the image feature vector exceeds the similarity threshold, judging the case to be a new case and recommending the case to the doctors in the corresponding department of the case, constructing a new minimum label according to the image feature vector, performing hierarchical label analysis on the case and then archiving the case in a graded manner, and ending the comprehensive recommendation index analysis of the case.

[0017] According to the above technical solution, step S23 further includes the following steps:

[0018] Step S231: Based on the hierarchical list of each label in the hierarchical storage, the image label and the text label are merged, the image label and the text label are arranged in descending order according to the label relationship of the hierarchical relationship, and duplicate labels are removed to obtain a label path, and matching is performed one by one according to the height of the hierarchical list, and it is determined whether the label path is associated level by level along a continuous path in the hierarchical list. When the label path is discontinuous, the similarity matching is re-performed according to the corresponding image label and text label at the discontinuous node to obtain a unified label;

[0019] Step S232: when the label path is a continuous label path, the image label and the text label are merged into a label set of a continuous label path, and the label set constitutes the hierarchical label;

[0020] Step S233: When the label path is a plurality of continuous label paths, the case is judged to be a complication, and the image label and the text label path are merged into a label set of a plurality of continuous label paths, and the label set constitutes the hierarchical label.

[0021] According to the above technical solution, step S3 further includes the following steps:

[0022] Step S31: Based on the analysis results of the hierarchical labels, the number of historical diagnosis and treatment data in the department and the historical diagnosis and treatment data of all departments in the same layer with the minimum label in the hierarchical labels is counted, the minimum label of the case is matched with the historical diagnosis and treatment data of the department for similarity, and the first recommendation index is analyzed according to the matching results, wherein the calculation formula of the first recommendation index is:

[0023]

[0024] Where W1 represents the first recommendation index, T i represents the i-th minimum label in the hierarchical label set in the department, Indicates the historical diagnosis and treatment data of the department with label T i The number of associated cases, T j represents the total number of cases associated with all minimum labels in the historical diagnosis and treatment data of the department, Indicates the historical diagnosis and treatment data in the department with label T j The number of cases associated with Denotes the case with label T i The similarity of is in the range [0,1]. Indicates the label T i The weight coefficient of

[0025] Step S32: Based on the analysis results of the hierarchical labels, count the number of historical medical data of each doctor in the department and all personal historical medical data in the same layer with the minimum label in the hierarchical labels, perform similarity matching between the minimum label of the case and the personal historical medical data, and obtain the second recommendation index W2 according to the analysis process of step S31.

[0026] According to the above technical solution, step S4 further includes the following steps:

[0027] Step S41: Based on the analysis results of the first recommendation index and the second recommendation index, a comprehensive recommendation index of each doctor is calculated, wherein the calculation formula of the comprehensive recommendation index is:

[0028] W=αW1+βW2

[0029] Wherein, W represents the comprehensive recommendation index, α and β represent the coefficients for adjusting the first recommendation index and the second recommendation index in the comprehensive recommendation index respectively;

[0030] Step S42: Based on the comparison between the calculation result of the comprehensive recommendation index and the set threshold, the case is recommended differently to different doctors according to the comparison result; when the comprehensive recommendation index exceeds the threshold, the case is recommended to the corresponding doctor and push data is sent; when the comprehensive recommendation index exceeds 30% of the threshold, the case is stored in the recommendation list of the corresponding doctor according to the stratified label; when the comprehensive recommendation index does not exceed 30% of the threshold, the case is stored in the historical diagnosis and treatment data according to the stratified label.

[0031] According to the above technical solution, the system includes a medical data acquisition module, a hierarchical label acquisition module, a recommendation index analysis module, and a hierarchical archiving module:

[0032] The medical data acquisition module is used to obtain the medical images to be processed, the text data of the corresponding cases, and retrieve the historical diagnosis and treatment data of each doctor in the corresponding department. The historical diagnosis and treatment data is stored in a hierarchical structure, each level is named with a different label, and each lower level is a further subdivision of the classification of the upper level;

[0033] The hierarchical label acquisition module is used to respectively acquire image labels and text labels based on the medical image and the text data, and fuse the image labels and the text labels to obtain hierarchical labels;

[0034] The recommendation index analysis module is used to retrieve the corresponding historical diagnosis and treatment data in the department for comparison based on the hierarchical label to obtain the first recommendation index of the case, and retrieve the corresponding historical diagnosis and treatment data of each doctor in the department for comparison to obtain the second recommendation index of the case;

[0035] The hierarchical archiving module is used to obtain different comprehensive recommendation indexes for each doctor based on the first recommendation index and the second recommendation index. The case is recommended and graded for different doctors according to the comprehensive recommendation index.

[0036] According to the above technical solution, the recommendation index analysis module includes a first recommendation index analysis module, a second recommendation index analysis module, and a comprehensive recommendation index analysis module:

[0037] The first recommendation index analysis module is used to count the number of historical diagnosis and treatment data in the department and the historical diagnosis and treatment data of all departments in the same layer with the minimum label in the layered label based on the analysis result of the layered label, perform similarity matching between the minimum label of the case and the historical diagnosis and treatment data of the department, and analyze the first recommendation index according to the matching result, wherein the calculation formula of the first recommendation index is:

[0038]

[0039] Where W1 represents the first recommendation index, T i represents the i-th minimum label in the hierarchical label set in the department, Indicates the historical diagnosis and treatment data of the department with label T i The number of associated cases, T j represents the total number of cases associated with all minimum labels in the historical diagnosis and treatment data of the department, Indicates the historical diagnosis and treatment data in the department with label T j The number of cases associated with Denotes the case with label T i The similarity of is in the range of [0,1], indicating that Indicates the label T i The weight coefficient of

[0040] The second recommendation index analysis module is used to count the number of all personal historical diagnosis and treatment data of each doctor in the department in the same layer as the minimum label in the hierarchical label based on the analysis result of the hierarchical label, perform similarity matching between the minimum label of the case and the personal historical diagnosis and treatment data, and obtain the second recommendation index W2 according to the analysis process of the first recommendation index analysis module;

[0041] The comprehensive recommendation index analysis module is used to calculate the comprehensive recommendation index of each doctor based on the analysis results of the first recommendation index and the second recommendation index, wherein the calculation formula of the comprehensive recommendation index is:

[0042] W=αW1+βW2

[0043] Wherein, W represents the comprehensive recommendation index, and α and β represent the coefficients for adjusting the first recommendation index and the second recommendation index in the comprehensive recommendation index, respectively.

[0044] According to the above technical solution, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the electronic device implements the method as claimed in any one of claims 1 to 6.

[0045] According to the above technical solution, a computer-readable storage medium is provided, characterized in that the computer-readable storage medium is used to store a computer program, and when the computer program is run on a computer, the computer executes the method as described in any one of claims 1 to 6.

[0046] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention, through the integration of medical images and text data, can more comprehensively extract case features, thereby obtaining more accurate image labels and text labels, and utilizes a hierarchical labeling system to merge image labels and text labels in the hierarchical storage, thereby realizing intelligent management and optimization of labels; based on hierarchical labels and historical diagnosis and treatment data, the present invention can calculate the first recommendation index and the second recommendation index of a case, thereby obtaining a comprehensive recommendation index; this personalized recommendation method not only improves the utilization efficiency of medical resources, but also archives cases in a hierarchical manner according to the comprehensive recommendation index, which not only facilitates doctors and managers to quickly search and review, but also provides rich data resources for subsequent medical research and data analysis, thereby enhancing the accuracy of archiving diagnosis and treatment data and improving the utilization rate of diagnosis and treatment data. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0048] Figure 1 The present invention is a flowchart of a method for intelligent data classification and grading according to an embodiment of the present invention.

[0049] Figure 2 It is a schematic diagram of a system for intelligent data classification and grading according to an embodiment of the present invention.

[0050] Figure 3 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] This embodiment can be applied to the scenario of case arrangement and recommendation. The method can be executed by an intelligent data classification and grading method and system provided by this embodiment. Figure 1 This is a flow chart of a method for intelligent data classification and grading according to an embodiment of the present invention, the method specifically comprises the following steps:

[0053] Step S1: Obtain the medical images to be processed and the text data of the corresponding cases, and retrieve the historical diagnosis and treatment data of each doctor in the corresponding department. The historical diagnosis and treatment data are stored in a hierarchical structure, each level is named with a different label, and each lower level is a further subdivision of the upper level classification.

[0054] Step S2: Obtain image labels and text labels based on the medical image and the text data respectively, fuse the image labels and the text labels to obtain hierarchical labels.

[0055] Step S3: Based on the hierarchical labels, the corresponding historical diagnosis and treatment data in the department are retrieved for comparison to obtain the first recommendation index of the case, and the corresponding historical diagnosis and treatment data of each doctor in the department are retrieved for comparison to obtain the second recommendation index of the case.

[0056] Step S4: Based on the first recommendation index and the second recommendation index, different comprehensive recommendation indexes are obtained for each doctor, and different recommendations and graded archives are made for different doctors according to the comprehensive recommendation indexes.

[0057] In a specific embodiment, diversified integration of medical information is achieved by comprehensively acquiring the medical images to be processed, case text data, and doctors' historical diagnosis and treatment data. These data not only cover the patient's direct medical records, but also include the doctor's past diagnosis and treatment experience and decisions. Among them, historical diagnosis and treatment data are stored in a hierarchical structure, and each level has a clear label. This structured storage method facilitates rapid retrieval and precise matching of data, laying the foundation for the efficient execution of subsequent steps.

[0058] In addition, through advanced label fusion technology, medical image labels and text labels are effectively integrated to generate hierarchical labels, which improves the consistency and completeness of the data and enables data analysis to go deeper into more specific details. The introduction of hierarchical labels enables the system to have a more comprehensive understanding of the characteristics of the case and the doctor's expertise. According to the hierarchical labels, the historical diagnosis and treatment data of the department and the doctor are retrieved for comparison, and the first and second recommendation indexes of the case are generated. These two indexes evaluate the case from the perspective of the department as a whole and the individual doctor, respectively, and generate comprehensive recommendation indexes for different doctors. Doctors are recommended and graded, which optimizes the allocation of medical resources and improves the efficiency and quality of diagnosis and treatment.

[0059] In summary, the system not only realizes comprehensive and efficient integration and analysis of medical data, but also archives cases in a hierarchical manner and provides doctors with personalized services. This not only improves the overall quality and efficiency of medical services, but also promotes the optimal allocation of medical resources and learning and communication among doctors. It has important practical application value and promotion significance.

[0060] In an embodiment of the present invention, the step S2 further includes the following steps:

[0061] Step S21: After preprocessing the medical image based on the medical image analysis terminal, the medical image is abnormally segmented by the U-Net algorithm to obtain a segmented image, the segmented image is analyzed for abnormal symptoms, a corresponding description text is generated according to the abnormal symptoms, an image feature vector is extracted according to the description text, and the image feature vector is matched with the image label according to the existing label stored in the hierarchical manner;

[0062] Step S22: After preprocessing the text data based on the text data analysis terminal, a text feature vector is extracted from the text data through a transformer model, and the text feature vector is matched with the text label according to the existing label stored in the hierarchical manner;

[0063] Step S23: Based on the hierarchical list of each label in the hierarchical storage, the image label and the text label are merged, duplicate labels are removed to obtain a unified label set, and the label set is organized into multi-level hierarchical labels according to the hierarchical relationship.

[0064] In an embodiment of the present invention, the step S21 further includes the following steps:

[0065] Step S211: classifying the medical image pixel by pixel based on the U-Net model, distinguishing normal and abnormal areas in the medical image, marking the abnormal areas according to their contours, and performing abnormal segmentation to obtain the segmented image;

[0066] Step S212: identifying and analyzing the abnormal symptoms of the segmented image through a deep learning model. In the abnormal segmentation process, generating a corresponding description text according to the identified abnormal symptoms, the description text including the abnormal description and the medical symptoms. The identification color of the abnormal area contour is adjusted according to the common proportion of the medical symptoms. The larger the proportion of the common situation, the lighter the identification color, and the smaller the proportion of the common situation, the darker the identification color.

[0067] Step S213: extracting an image feature vector based on the description text, matching the image feature vector with the image label according to the existing label, constructing a label feature space containing all the existing labels, matching the image feature vector with the existing label in the label feature space according to the similarity, obtaining a plurality of image labels, setting a similarity threshold, and when the image feature vector exceeds the similarity threshold, judging the case as a new case and recommending the case to each doctor in the corresponding department of the case, constructing a new minimum label according to the image feature vector, performing hierarchical label analysis on the case and then archiving the case in a graded manner, and ending the comprehensive recommendation index analysis of the case, wherein the minimum label is constructed based on the key features or attributes of the image feature vector, these features or attributes can more accurately reflect the characteristics of the case, can perform a more detailed hierarchical label analysis on the case, and add the newly constructed minimum label to the existing label set.

[0068] In an embodiment of the present invention, the step S23 further includes the following steps:

[0069] Step S231: Based on the hierarchical list of each label in the hierarchical storage, the image label and the text label are merged, the image label and the text label are arranged in descending order according to the label relationship of the hierarchical relationship, and duplicate labels are removed to obtain a label path, and matching is performed one by one according to the height of the hierarchical list, and it is determined whether the label path is associated level by level along a continuous path in the hierarchical list. When the label path is discontinuous, the similarity matching is re-performed according to the corresponding image label and text label at the discontinuous node to obtain a unified label;

[0070] Step S232: when the label path is a continuous label path, the image label and the text label are merged into a label set of a continuous label path, and the label set constitutes the hierarchical label;

[0071] Step S233: when the label path is a plurality of continuous label paths, the case is judged to be a complication, and the image label and the text label path are merged into a label set of a plurality of continuous label paths, and the label set constitutes the hierarchical label;

[0072] Exemplarily, the entire framework of the hierarchical list of hierarchical storage is already set, and the label names in the hierarchical list are set according to the hierarchical classification of the disease types, wherein the label names can be added with the minimum label according to the newly added disease types. When the image labels obtained are: bacterial pneumonia, simple parapneumonic pleural effusion, dry pleurisy stage, the text labels obtained are: respiratory system diseases, bacterial pneumonia, simple parapneumonic pleural effusion, the image labels and text labels are arranged in descending order according to the label relationship of the hierarchical relationship, and the label acquisition path is: respiratory system diseases->bacterial pneumonia- >Simple parapneumonic pleural effusion->Dry pleurisy stage, match the label path with the label name in the hierarchical list, and judge the continuity of the label path in the hierarchical list. When the path in the matched hierarchical list is: Respiratory diseases->Bacterial pneumonia->Simple parapneumonic pleural effusion->Dry pleurisy stage, the label path is judged to be a continuous label path. When there is only one path in the matched hierarchical list and there is a break or jump, the label path is judged to be a discontinuous label path. When there are multiple and continuous paths in the matched hierarchical list, it is judged to be multiple continuous label paths.

[0073] In an embodiment of the present invention, step S3 further includes the following steps:

[0074] Step S31: Based on the analysis results of the hierarchical labels, the number of historical diagnosis and treatment data in the department and the historical diagnosis and treatment data of all departments in the same layer with the minimum label in the hierarchical labels is counted, the minimum label of the case is matched with the historical diagnosis and treatment data of the department for similarity, and the first recommendation index is analyzed according to the matching results, wherein the calculation formula of the first recommendation index is:

[0075]

[0076] Where W1 represents the first recommendation index, T i represents the i-th minimum label in the hierarchical label set in the department, Indicates the historical diagnosis and treatment data of the department with label T i The number of associated cases, T j represents the total number of cases associated with all minimum labels in the historical diagnosis and treatment data of the department, Indicates the historical diagnosis and treatment data in the department with label T j The number of cases associated with Denotes the case with label T i The similarity of is in the range [0,1]. Indicates the label T i The weight coefficient of

[0077] Step S32: Based on the analysis results of the hierarchical labels, count the number of historical medical data of each doctor in the department and all personal historical medical data in the same layer with the minimum label in the hierarchical labels, perform similarity matching between the minimum label of the case and the personal historical medical data, and obtain the second recommendation index W2 according to the analysis process of step S31.

[0078] In an embodiment of the present invention, the step S4 further includes the following steps:

[0079] Step S41: Based on the analysis results of the first recommendation index and the second recommendation index, a comprehensive recommendation index of each doctor is calculated, wherein the calculation formula of the comprehensive recommendation index is:

[0080] W=αW1+βW2

[0081] Wherein, W represents the comprehensive recommendation index, α and β represent the coefficients for adjusting the first recommendation index and the second recommendation index in the comprehensive recommendation index respectively;

[0082] Step S42: Based on the comparison between the calculation result of the comprehensive recommendation index and the set threshold, the case is recommended differently to different doctors according to the comparison result; when the comprehensive recommendation index exceeds the threshold, the case is recommended to the corresponding doctor and push data is sent; when the comprehensive recommendation index exceeds 30% of the threshold, the case is stored in the recommendation list of the corresponding doctor according to the stratified label; when the comprehensive recommendation index does not exceed 30% of the threshold, the case is stored in the historical diagnosis and treatment data according to the stratified label.

[0083] It should be noted that each doctor's historical diagnosis and treatment data and recommendation list are time-sensitive. As the doctor's work accumulates and the types of diagnosis and treatment cases are updated, when there are overlaps with subsequently added cases in the recommendation list, these overlapping cases will be automatically removed from the recommendation list. At the same time, based on the number of related cases added in the historical diagnosis and treatment data and the number of recommendations of these cases in the recommendation list, the weight of such cases for the second recommendation index is dynamically adjusted, and the information in the recommendation library and the recommendation weight are accurately adjusted according to the doctor's real-time historical case situation, thereby ensuring that the recommended cases received by the doctor are both in line with his or her professional field and can reflect the latest diagnosis and treatment trends and case types. In this way, doctors can screen, search and learn cases more efficiently, reduce unnecessary waste of time, and improve work efficiency and diagnosis and treatment quality.

[0084] The embodiment of the present invention provides a method and system for intelligent data classification and grading. Figure 2 A schematic diagram of the module composition of a system for intelligent data classification and grading provided in the second embodiment of the present invention is shown in FIG. Figure 2 As shown, the system includes a medical data acquisition module, a hierarchical label acquisition module, a recommendation index analysis module, and a hierarchical archiving module:

[0085] The medical data acquisition module is used to obtain the medical images to be processed, the text data of the corresponding cases, and retrieve the historical diagnosis and treatment data of each doctor in the corresponding department. The historical diagnosis and treatment data is stored in a hierarchical structure, each level is named with a different label, and each lower level is a further subdivision of the classification of the upper level;

[0086] The hierarchical label acquisition module is used to respectively acquire image labels and text labels based on the medical image and the text data, and fuse the image labels and the text labels to obtain hierarchical labels;

[0087] The recommendation index analysis module is used to retrieve the corresponding historical diagnosis and treatment data in the department for comparison based on the hierarchical label to obtain the first recommendation index of the case, and retrieve the corresponding historical diagnosis and treatment data of each doctor in the department for comparison to obtain the second recommendation index of the case;

[0088] The hierarchical archiving module is used to obtain different comprehensive recommendation indexes for each doctor based on the first recommendation index and the second recommendation index. The case is recommended and graded for different doctors according to the comprehensive recommendation index.

[0089] In some embodiments of the present invention, the layered label acquisition module includes an image label acquisition module, a text label acquisition module, and a label fusion module:

[0090] The image label acquisition module is used to perform abnormal segmentation on the medical image through the U-Net algorithm to obtain a segmented image after preprocessing the medical image based on the medical image analysis terminal, perform abnormal symptom analysis on the segmented image, generate corresponding description text according to the abnormal symptoms, extract image feature vectors according to the description text, and match the image feature vectors with the image labels according to the existing labels stored in the hierarchical manner;

[0091] The text label acquisition module is used to extract text feature vectors from the text data through a transformer model after preprocessing the text data based on the text data analysis terminal, and the text feature vectors are matched with the text labels according to the existing labels stored in the hierarchical manner;

[0092] The label fusion module is used to fuse the image label and the text label based on the hierarchical list of each label in the hierarchical storage, remove duplicate labels to obtain a unified label set, and organize the label set into multi-level hierarchical labels according to the hierarchical relationship.

[0093] In some embodiments of the present invention, the abnormal region segmentation submodule, the abnormal symptom analysis submodule, and the new label construction submodule are included:

[0094] The abnormal region segmentation submodule is used to classify the medical image pixel by pixel based on the U-Net model, distinguish normal and abnormal regions in the medical image, mark them according to the contours of the abnormal regions, and perform the abnormal segmentation to obtain the segmented image;

[0095] The abnormal symptom analysis submodule is used to identify and analyze the abnormal symptoms of the segmented image through a deep learning model. In the abnormal segmentation process, a corresponding description text is generated according to the identified abnormal symptoms, and the description text includes an abnormal description and medical symptoms. The identification color of the abnormal area contour is adjusted according to the common situation proportion of the medical symptoms. The larger the common situation proportion, the lighter the identification color, and the smaller the common situation proportion, the darker the identification color;

[0096] The new label construction submodule is used to extract an image feature vector based on the description text, match the image feature vector with the image label according to the existing label, construct a label feature space containing all the existing labels, match the image feature vector with the existing label in the label feature space according to the similarity, obtain multiple image labels, set a similarity threshold, and when the image feature vector exceeds the similarity threshold, judge that the case is a new case and recommend the case to the doctors in the corresponding department of the case, construct a new minimum label according to the image feature vector, perform hierarchical label analysis on the case and then archive it in a graded manner, and end the comprehensive recommendation index analysis of the case.

[0097] In some embodiments of the present invention, the recommendation index analysis module includes a first recommendation index analysis module, a second recommendation index analysis module, and a comprehensive recommendation index analysis module:

[0098] The first recommendation index analysis module is used to count the number of historical diagnosis and treatment data in the department and the historical diagnosis and treatment data of all departments in the same layer with the minimum label in the layered label based on the analysis result of the layered label, perform similarity matching between the minimum label of the case and the historical diagnosis and treatment data of the department, and analyze the first recommendation index according to the matching result, wherein the calculation formula of the first recommendation index is:

[0099]

[0100] Where W1 represents the first recommendation index, T i represents the i-th minimum label in the hierarchical label set in the department, Indicates the historical diagnosis and treatment data of the department with label T i The number of associated cases, T j represents the total number of cases associated with all minimum labels in the historical diagnosis and treatment data of the department, Indicates the historical diagnosis and treatment data in the department with label T j The number of cases associated with Denotes the case with label T i The similarity of is in the range of [0,1], indicating that Indicates the label T i The weight coefficient of

[0101] The second recommendation index analysis module is used to count the number of all personal historical diagnosis and treatment data of each doctor in the department in the same layer as the minimum label in the hierarchical label based on the analysis result of the hierarchical label, perform similarity matching between the minimum label of the case and the personal historical diagnosis and treatment data, and obtain the second recommendation index W2 according to the analysis process of the first recommendation index analysis module;

[0102] The comprehensive recommendation index analysis module is used to calculate the comprehensive recommendation index of each doctor based on the analysis results of the first recommendation index and the second recommendation index, wherein the calculation formula of the comprehensive recommendation index is:

[0103] W=αW1+βW2

[0104] Wherein, W represents the comprehensive recommendation index, and α and β represent the coefficients for adjusting the first recommendation index and the second recommendation index in the comprehensive recommendation index, respectively.

[0105] It should be noted that, although several units or subunits of the device are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided into multiple units to be embodied.

[0106] Based on the same inventive concept as the above-mentioned method embodiment, an electronic device is also provided in an embodiment of the present invention, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the electronic device implements the control method in the above-mentioned embodiment.

[0107] In one embodiment, the electronic device may be a server. In this embodiment, the structure of the electronic device may be as follows: Figure 3 As shown, it includes a memory 2001 , a communication module 2003 and one or more processors 2002 .

[0108] The memory 2001 is used to store computer programs executed by the processor 2002. The memory 2001 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and programs required for running the instant messaging function, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0109] The memory 2001 may be a volatile memory, such as a random-access memory (RAM); the memory 2001 may also be a non-volatile memory, such as a read-only memory, a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD); or the memory 2001 may be any other medium that can be used to carry or store a desired computer program in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 2001 may be a combination of the above memories.

[0110] The processor 2002 may include one or more central processing units (CPU) or a digital processing unit, etc. The processor 2002 is used to implement the above-mentioned audio data processing method when calling the computer program stored in the memory 2001 .

[0111] The communication module 2003 is used to communicate with terminal devices and other servers.

[0112] The specific connection medium between the memory 2001, the communication module 2003 and the processor 2002 is not limited in the embodiment of the present invention. Figure 3 In the embodiment, the memory 2001 and the processor 2002 are connected via a bus 2004. The bus 2004 is Figure 3 The arrows in the figure are used to describe the connection between other components, which is only for illustrative purposes and is not intended to be limiting. Bus 2004 can be divided into address bus, data bus, control bus, etc. For ease of description, Figure 3 The diagram shows that only one arrow is used, but this does not mean that there is only one bus or only one type of bus.

[0113] Based on the same inventive concept as the above method embodiment, an embodiment of the present invention also provides a computer-readable storage medium, the computer-readable storage medium is used to store a computer program, when the computer program is run on a computer, the electronic device implements the control method in the above embodiment. The computer-readable storage medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0114] Based on the same inventive concept as the above-mentioned method embodiment, an embodiment of the present invention further provides a computer program product, the computer program product including a computer program, when the program product is run on an electronic device, the computer program is used to make the electronic device execute the steps of the control method according to various exemplary embodiments of the present invention described above in this specification. The program product may adopt any combination of one or more readable media. These computer program commands may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the commands executed by the processor of the computer or other programmable data processing device generate commands for implementing the control method in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0115] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

Claims

1. A method for intelligent data classification and grading, characterized by: Obtain the medical images to be processed, the text data of the corresponding cases, and retrieve the historical diagnosis and treatment data of each doctor in the corresponding department. The historical diagnosis and treatment data is stored in a hierarchical structure, each level is named with a different label, and each lower level is a further subdivision of the classification of the upper level; Acquire image labels and text labels based on the medical image and the text data, respectively, and fuse the image labels and the text labels to obtain hierarchical labels; Based on the hierarchical label, the corresponding historical diagnosis and treatment data in the department are retrieved for comparison to obtain a first recommendation index for the case, and the corresponding historical diagnosis and treatment data of each doctor in the department are retrieved for comparison to obtain a second recommendation index for the case; Based on the first recommendation index and the second recommendation index, different comprehensive recommendation indexes are obtained for each doctor, and the case is recommended and graded for different doctors according to the comprehensive recommendation index.

2. The method for intelligent data classification and grading according to claim 1, characterized in that: The obtaining of image labels and text labels based on the medical image and the text data, respectively, and fusing the image labels and the text labels to obtain hierarchical labels includes: After preprocessing the medical image based on the medical image analysis terminal, the medical image is abnormally segmented by the U-Net algorithm to obtain a segmented image, the segmented image is analyzed for abnormal symptoms, a corresponding description text is generated according to the abnormal symptoms, an image feature vector is extracted according to the description text, and the image feature vector is matched with the image label according to the existing label stored in the hierarchy; After the text data is preprocessed by the text data analysis terminal, a text feature vector is extracted from the text data through a transformer model, and the text feature vector is matched with the text label according to the existing label stored in the hierarchical manner; The image label and the text label are merged based on the hierarchical list of each label in the hierarchical storage, repeated labels are removed to obtain a unified label set, and the label set is organized into multi-level hierarchical labels according to the hierarchical relationship.

3. The method for intelligent data classification and grading according to claim 2, characterized in that: The medical image is segmented abnormally by a U-Net algorithm to obtain a segmented image, the segmented image is analyzed for abnormal symptoms, a corresponding description text is generated according to the abnormal symptoms, an image feature vector is extracted according to the description text, and the image feature vector is matched with the image label according to the existing label stored in a hierarchical manner, including: Based on the U-Net model, the medical image is classified pixel by pixel, normal and abnormal areas in the medical image are distinguished, the abnormal areas are marked according to their contours, and the abnormal area is segmented to obtain the segmented image; Identify and analyze the abnormal symptoms of the segmented image through a deep learning model. During the abnormal segmentation process, generate a corresponding description text according to the identified abnormal symptoms, the description text includes an abnormal description and medical symptoms, and adjust the identification color of the abnormal area contour according to the common proportion of the medical symptoms. The larger the proportion of the common situation, the lighter the identification color, and the smaller the proportion of the common situation, the darker the identification color; An image feature vector is extracted based on the description text, the image feature vector is matched with the image label according to the existing label, a label feature space containing all the existing labels is constructed, the image feature vector is matched with the existing label in the label feature space according to the similarity, a plurality of image labels are obtained, a similarity threshold is set, when the image feature vector exceeds the similarity threshold, the case is judged to be a new case and the case is recommended to each doctor in the corresponding department of the case, a new minimum label is constructed according to the image feature vector, the case is subjected to hierarchical label analysis and then graded archiving is performed, and the comprehensive recommendation index analysis of the case is ended.

4. The method for intelligent data classification and grading according to claim 3, characterized in that: The step of fusing the image label and the text label based on the hierarchical list of each label in the hierarchical storage, removing duplicate labels to obtain a unified label set, and organizing the label set into multi-level hierarchical labels according to the hierarchical relationship includes: Based on the hierarchical list of each label in the hierarchical storage, the image label and the text label are merged, the image label and the text label are arranged in descending order according to the label relationship of the hierarchical relationship, and duplicate labels are removed to obtain a label path, and the label paths are matched one by one according to the height of the hierarchical list, and it is determined whether the label path is associated step by step along a continuous path in the hierarchical list. When the label path is discontinuous, the similarity matching is re-performed according to the corresponding image label and text label at the discontinuous node to obtain a unified label; When the label path is a continuous label path, the image label and the text label are merged into a label set of a continuous label path, and the label set constitutes the hierarchical label; When the label path is a plurality of continuous label paths, the case is judged to be a complication, and the image label and the text label path are fused into a label set of a plurality of continuous label paths, and the label set constitutes the hierarchical label.

5. The method of intelligent data classification and grading according to claim 4, characterized in that: The retrieving corresponding historical diagnosis and treatment data in the department based on the hierarchical label for comparison to obtain the first recommendation index of the case, and retrieving corresponding historical diagnosis and treatment data of each doctor in the department for comparison to obtain the second recommendation index of the case, includes: Based on the analysis results of the hierarchical labels, the number of historical diagnosis and treatment data in the department and the historical diagnosis and treatment data of all departments in the same layer with the minimum label in the hierarchical labels is counted, the minimum label of the case is matched with the historical diagnosis and treatment data of the department for similarity, and the first recommendation index is analyzed according to the matching results, wherein the calculation formula of the first recommendation index is: Where W1 represents the first recommendation index, T i represents the i-th minimum label in the hierarchical label set in the department, Indicates the historical diagnosis and treatment data of the department with label T i The number of associated cases, T j represents the total number of cases associated with all minimum labels in the historical diagnosis and treatment data of the department, Indicates the historical diagnosis and treatment data in the department with label T j The number of cases associated with Denotes the case with label T i The similarity of is in the range of [0, 1]. Indicates the label T i The weight coefficient of Based on the analysis results of the hierarchical labels, the number of historical medical data of each doctor in the department and all personal historical medical data in the same layer with the minimum label in the hierarchical labels is counted, and the minimum label of the case is matched with the personal historical medical data for similarity. According to the analysis process of the first recommendation index, the second recommendation index W2 is obtained.

6. The method for intelligent data classification and grading according to claim 5, characterized in that: The obtaining of different comprehensive recommendation indexes for different doctors based on the first recommendation index and the second recommendation index, and making different recommendations and graded filings for different doctors according to the comprehensive recommendation indexes for the cases, includes: Based on the analysis results of the first recommendation index and the second recommendation index, a comprehensive recommendation index of each doctor is calculated, wherein the calculation formula of the comprehensive recommendation index is: W=αW1+βW2 Wherein, W represents the comprehensive recommendation index, α and β represent the coefficients for adjusting the first recommendation index and the second recommendation index in the comprehensive recommendation index respectively; Based on the comparison between the calculation result of the comprehensive recommendation index and the set threshold, the case is recommended differently to different doctors according to the comparison result. When the comprehensive recommendation index exceeds the threshold, the case is recommended to the corresponding doctor and push data is sent. When the comprehensive recommendation index exceeds 30% of the threshold, the case is stored in the recommendation list of the corresponding doctor according to the stratified label. When the comprehensive recommendation index does not exceed 30% of the threshold, the case is stored in the historical diagnosis and treatment data according to the stratified label.

7. An intelligent data classification and grading system, characterized by: The system includes a medical data acquisition module, a hierarchical label acquisition module, a recommendation index analysis module, and a hierarchical archiving module: The medical data acquisition module is used to obtain the medical images to be processed, the text data of the corresponding cases, and retrieve the historical diagnosis and treatment data of each doctor in the corresponding department. The historical diagnosis and treatment data is stored in a hierarchical structure, each level is named with a different label, and each lower level is a further subdivision of the classification of the upper level; The hierarchical label acquisition module is used to respectively acquire image labels and text labels based on the medical image and the text data, and fuse the image labels and the text labels to obtain hierarchical labels; The recommendation index analysis module is used to retrieve the corresponding historical diagnosis and treatment data in the department for comparison based on the hierarchical label to obtain the first recommendation index of the case, and retrieve the corresponding historical diagnosis and treatment data of each doctor in the department for comparison to obtain the second recommendation index of the case; The hierarchical archiving module is used to obtain different comprehensive recommendation indexes for each doctor based on the first recommendation index and the second recommendation index. The case is recommended and graded for different doctors according to the comprehensive recommendation index.

8. The intelligent data classification and grading system according to claim 7, characterized in that: The recommendation index analysis module includes a first recommendation index analysis module, a second recommendation index analysis module, and a comprehensive recommendation index analysis module: The first recommendation index analysis module is used to count the number of historical diagnosis and treatment data in the department and the historical diagnosis and treatment data of all departments in the same layer with the minimum label in the layered label based on the analysis result of the layered label, perform similarity matching between the minimum label of the case and the historical diagnosis and treatment data of the department, and analyze the first recommendation index according to the matching result, wherein the calculation formula of the first recommendation index is: Where W1 represents the first recommendation index, T i represents the i-th minimum label in the hierarchical label set in the department, Indicates the historical diagnosis and treatment data of the department with label T i The number of associated cases, T j represents the total number of cases associated with all minimum labels in the historical diagnosis and treatment data of the department, Indicates the historical diagnosis and treatment data in the department with label T j The number of cases associated with Denotes the case with label T i The similarity of is [0, 1], which means, Indicates the label T i The weight coefficient of The second recommendation index analysis module is used to count the number of all personal historical diagnosis and treatment data of each doctor in the department in the same layer as the minimum label in the hierarchical label based on the analysis result of the hierarchical label, perform similarity matching between the minimum label of the case and the personal historical diagnosis and treatment data, and obtain the second recommendation index W2 according to the analysis process of the first recommendation index analysis module; The comprehensive recommendation index analysis module is used to calculate the comprehensive recommendation index of each doctor based on the analysis results of the first recommendation index and the second recommendation index, wherein the calculation formula of the comprehensive recommendation index is: W=αW1+βW2 Wherein, W represents the comprehensive recommendation index, and α and β represent the coefficients for adjusting the first recommendation index and the second recommendation index in the comprehensive recommendation index, respectively.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the electronic device implements the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store a computer program, and when the computer program is executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 6.

Images