Pneumonia analysis and intelligent information recommendation method and system based on knowledge graph

Abstract

The present invention discloses a pneumonia analysis and intelligent information recommendation method and system based on a knowledge graph, based on the health big data of historical pneumonia patients and constructing a pneumonia knowledge graph, and obtaining target-related knowledge data related to pneumonia of target personnel; data classification is performed on the target-related knowledge data, and the classification process performs feasibility analysis of pneumonia pathology for each group of data, and multiple classification results are screened based on feasibility to obtain multiple groups of patient-related data; pneumonia TCM diagnosis data data are collected, and a pneumonia TCM diagnosis large model is constructed based on a preset language model, and multiple groups of associated data are imported into the large model for TCM diagnosis schemes, and multiple groups of diagnosis and treatment schemes are formed. The present invention recommends data in multiple dimensions based on the knowledge graph, and the diagnosis and treatment recommendation information has a better reference significance, which can greatly improve the medical auxiliary level of the system.

Description

Technical Field

[0001] The present invention relates to the field of knowledge graph analysis, and more specifically, to a knowledge graph-based pneumonia analysis and intelligent information recommendation method and system. Background Art

[0002] With the continuous development of medical big data technology, the demand for accurate diagnosis and personalized treatment of common diseases such as pneumonia is becoming increasingly urgent. Although the existing medical information system can collect and process a large amount of medical data, it still has many deficiencies in information integration, in-depth analysis and intelligent recommendation. Therefore, the development of a pneumonia analysis and intelligent information recommendation method based on knowledge graph is of great significance for the accuracy and efficiency of recommendation analysis of pneumonia diagnosis and treatment. Summary of the invention

[0003] The present invention overcomes the defects of the prior art and proposes a pneumonia analysis and intelligent information recommendation method and system based on knowledge graph.

[0004] The first aspect of the present invention provides a pneumonia analysis and intelligent information recommendation method based on a knowledge graph, comprising:

[0005] Collect health big data of historical pneumonia patients, perform text semantic analysis and entity relationship extraction based on NLP on the health big data, and construct a pneumonia knowledge graph;

[0006] Acquire pneumonia information of target personnel, perform triple information conversion based on the pneumonia information, and extract associated data through the pneumonia knowledge graph to obtain target associated knowledge data;

[0007] Based on the preset clustering model, the target associated knowledge data is classified. The feasibility of pneumonia pathology of each group of data is analyzed during the classification process. Multiple classification results are screened based on the feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are simulated and integrated with pneumonia patient information to form multiple groups of patient associated data.

[0008] Collect TCM diagnosis data on pneumonia, and build a large TCM diagnosis model for pneumonia based on the preset language model. Import multiple sets of related data into the large model for TCM diagnosis and form multiple sets of diagnosis and treatment plans.

[0009] In this solution, the health big data of historical pneumonia patients is collected, the health big data is subjected to NLP-based text semantic analysis and entity relationship extraction, and a pneumonia knowledge graph is constructed, specifically:

[0010] Based on the system database, obtain the health big data of historical pneumonia patients;

[0011] Perform data cleaning, data integration, and data redundancy removal preprocessing on the health big data;

[0012] The preprocessed health big data is converted into text format, and semantic analysis and triple data extraction are performed on the health big data based on the CNN semantic analysis model, wherein the triple data includes entity, attribute, and relationship data, and triple data is obtained;

[0013] A pneumonia knowledge graph is constructed based on the triple data.

[0014] In this solution, the health big data of historical pneumonia patients is collected, the health big data is subjected to NLP-based text semantic analysis and entity relationship extraction, and a pneumonia knowledge graph is constructed, which also includes:

[0015] Based on the AMIE algorithm, rule extraction and association rule learning are performed from the pneumonia knowledge graph to obtain new rule data;

[0016] Repeat the learning of association rules and the integration of new rule information until the new rule data reaches the preset data volume;

[0017] The new rule data is analyzed and evaluated for relationship and confidence calculated based on the pneumonia knowledge graph, and threshold screening is performed based on confidence to obtain the optimal relationship rules;

[0018] Import the optimal relationship rules into the pneumonia knowledge graph as new knowledge data.

[0019] In this solution, the target person's pneumonia information is obtained, triple information is converted based on the pneumonia information, and associated data is extracted through the pneumonia knowledge graph to obtain the target associated knowledge data, specifically:

[0020] Obtain pneumonia information of target personnel, perform semantic analysis and triple data conversion based on CNN semantic model, and form target triple data;

[0021] Perform vector transformation on the target entity in the target triple data and the entity data in the pneumonia knowledge graph based on the vector space model to obtain the first text vector and the second text vector respectively;

[0022] Calculating semantic similarity between the first text vector and the second text vector based on standard Euclidean distance, screening similarity data based on a threshold method, and obtaining a first similarity text vector and a second similarity text vector;

[0023] Acquire entity data corresponding to the first similarity text vector and the second similarity text vector and integrate the data to form associated entity data;

[0024] Based on the associative entity data, knowledge data is acquired from the pneumonia knowledge graph. The acquisition process is to locate the entity data from the knowledge graph based on the associative entity data, and traverse, query and count the relationship data, attribute data and associated entity data corresponding to the entity through the graph structure and the preset associated path distance to obtain the target associated knowledge data.

[0025] In this scheme, the target associated knowledge data is classified based on the preset clustering model, and the feasibility of pneumonia pathology of each group of data is analyzed during the classification process. Multiple classification results are screened based on feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are respectively simulated and integrated with pneumonia patient information to form multiple groups of patient associated data, specifically:

[0026] Construct a clustering model based on DBSCAN, import the target associated knowledge data into the clustering model for clustering and grouping, and obtain multiple initial groups;

[0027] The knowledge data in each initial group is subjected to feasibility analysis of pneumonia cases. The feasibility analysis process is to determine the data values ​​with pathological abnormalities in the data and mark them, calculate the proportion of abnormal value data, and obtain feasibility based on the proportion value;

[0028] The initial group whose feasibility meets expectations is marked as the preferred group, and the associated data in the initial group that does not meet expectations is integrated and the clustering grouping and feasibility calculation analysis and judgment are performed cyclically;

[0029] Finally, all the preferred groups are labeled and multiple groups of related data are obtained;

[0030] Collect the preset patient data format structure, convert and align the formats of multiple sets of related data, and form multiple sets of patient related data.

[0031] In this scheme, the data of pneumonia diagnosis in traditional Chinese medicine is collected, and a large model of pneumonia diagnosis in traditional Chinese medicine is constructed based on a preset language model. Multiple groups of related data are imported into the large model for traditional Chinese medicine diagnosis scheme, and multiple groups of diagnosis and treatment schemes are formed, specifically:

[0032] Based on the system database, collect TCM diagnosis data of pneumonia;

[0033] Construct a large model of TCM diagnosis of pneumonia based on the preset language model;

[0034] The data is cleaned and converted into text format, and then imported into the pneumonia TCM diagnosis model for training and learning;

[0035] Import multiple groups of patient-related data into the pneumonia TCM diagnosis model for diagnosis and treatment analysis and generate multiple groups of diagnosis and treatment plans;

[0036] The multiple groups of diagnosis and treatment plans are displayed through a preset terminal.

[0037] The second aspect of the present invention further provides a pneumonia analysis and intelligent information recommendation system based on a knowledge graph, the system comprising: a memory, a processor, the memory comprising a pneumonia analysis and intelligent information recommendation program based on a knowledge graph, the pneumonia analysis and intelligent information recommendation program based on a knowledge graph when executed by the processor implements the following steps:

[0038] Collect health big data of historical pneumonia patients, perform text semantic analysis and entity relationship extraction based on NLP on the health big data, and construct a pneumonia knowledge graph;

[0039] Acquire pneumonia information of target personnel, perform triple information conversion based on the pneumonia information, and extract associated data through the pneumonia knowledge graph to obtain target associated knowledge data;

[0040] Based on the preset clustering model, the target associated knowledge data is classified. The feasibility of pneumonia pathology of each group of data is analyzed during the classification process. Multiple classification results are screened based on the feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are simulated and integrated with pneumonia patient information to form multiple groups of patient associated data.

[0041] Collect TCM diagnosis data on pneumonia, and build a large TCM diagnosis model for pneumonia based on the preset language model. Import multiple sets of related data into the large model for TCM diagnosis and form multiple sets of diagnosis and treatment plans.

[0042] The present invention discloses a pneumonia analysis and intelligent information recommendation method and system based on a knowledge graph, based on the health big data of historical pneumonia patients and constructing a pneumonia knowledge graph, and obtaining target-related knowledge data related to pneumonia of target personnel; data classification is performed on the target-related knowledge data, and the classification process performs feasibility analysis of pneumonia pathology for each group of data, and multiple classification results are screened based on feasibility to obtain multiple groups of patient-related data; pneumonia TCM diagnosis data data are collected, and a pneumonia TCM diagnosis large model is constructed based on a preset language model, and multiple groups of associated data are imported into the large model for TCM diagnosis schemes, and multiple groups of diagnosis and treatment schemes are formed. The present invention recommends data in multiple dimensions based on the knowledge graph, and the diagnosis and treatment recommendation information has a better reference significance, which can greatly improve the medical auxiliary level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flowchart of a knowledge graph-based pneumonia analysis and intelligent information recommendation method of the present invention is shown;

[0044] Figure 2 The flowchart of constructing the pneumonia knowledge graph of the present invention is shown;

[0045] Figure 3 A block diagram of a pneumonia analysis and intelligent information recommendation system based on a knowledge graph of the present invention is shown. DETAILED DESCRIPTION

[0046] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0048] Figure 1 A flow chart of a knowledge graph-based pneumonia analysis and intelligent information recommendation method of the present invention is shown.

[0049] like Figure 1 As shown, the first aspect of the present invention provides a pneumonia analysis and intelligent information recommendation method based on a knowledge graph, comprising:

[0050] S102, collecting health big data of historical pneumonia patients, performing text semantic analysis and entity relationship extraction based on NLP on the health big data, and constructing a pneumonia knowledge graph;

[0051] S104, obtaining pneumonia information of the target person, performing triple information conversion based on the pneumonia information, and extracting associated data through the pneumonia knowledge graph to obtain target associated knowledge data;

[0052] S106, based on the preset clustering model, classify the target associated knowledge data, perform feasibility analysis of pneumonia pathology for each group of data during the classification process, screen multiple classification results based on feasibility, and finally obtain multiple groups of associated data. Through the patient data format structure, simulate and integrate the pneumonia patient information of the multiple groups of associated data to form multiple groups of patient associated data;

[0053] S108, collect TCM diagnosis data of pneumonia, and build a TCM diagnosis model of pneumonia based on a preset language model, import multiple groups of related data into the big model for TCM diagnosis plan, and form multiple groups of diagnosis and treatment plans.

[0054] It should be noted that the health big data includes historical medical information, examination results, symptom descriptions, clinical data, pathological records and other data, which are used to describe the physical condition and pathological information of pneumonia patients.

[0055] Figure 2 A flowchart for constructing the pneumonia knowledge graph of the present invention is shown.

[0056] According to an embodiment of the present invention, the health big data of historical pneumonia patients is collected, the health big data is subjected to NLP-based text semantic analysis and entity relationship extraction, and a pneumonia knowledge graph is constructed, specifically:

[0057] S202, based on the system database, obtain the health big data of historical pneumonia patients;

[0058] S204, performing data cleaning, data integration, and data redundancy removal preprocessing on the health big data;

[0059] S206, converting the preprocessed health big data into a text format, and performing semantic analysis and triple data extraction on the health big data based on a CNN semantic analysis model, wherein the triple data includes entity, attribute, and relationship data, and obtaining triple data;

[0060] S208, constructing a pneumonia knowledge graph based on the triple data.

[0061] It should be noted that the pneumonia knowledge graph is a knowledge graph constructed based on existing historical pneumonia patient data. Through the knowledge graph, we can further conduct knowledge mining on the data to obtain effective and practical knowledge data, providing data support for subsequent data recommendations.

[0062] According to an embodiment of the present invention, the health big data of historical pneumonia patients is collected, the health big data is subjected to NLP-based text semantic analysis and entity relationship extraction, and a pneumonia knowledge graph is constructed, and further includes:

[0063] Based on the AMIE algorithm, rule extraction and association rule learning are performed from the pneumonia knowledge graph to obtain new rule data;

[0064] Repeat the learning of association rules and the integration of new rule information until the new rule data reaches the preset data volume;

[0065] The new rule data is analyzed and evaluated for relationship and confidence calculated based on the pneumonia knowledge graph, and threshold screening is performed based on confidence to obtain the optimal relationship rules;

[0066] Import the optimal relationship rules into the pneumonia knowledge graph as new knowledge data.

[0067] It should be noted that the AMIE algorithm is a rule-based reasoning analysis algorithm. The introduction of this algorithm can mine new knowledge from the knowledge graph. In the case of existing structural knowledge, knowledge training and learning can be carried out to build a complete logical knowledge graph, further improving the effectiveness of subsequent knowledge retrieval. Through the knowledge graph, effective association analysis of pneumonia patients can be carried out in big data. Compared with the simple comparison of basic information by traditional technology, the knowledge graph can learn knowledge under the graph structure constructed by big data and obtain new associated knowledge data, so that when acquiring target associated data, new knowledge data associated with the target pneumonia patient can be obtained in the form of data mining.

[0068] According to an embodiment of the present invention, the pneumonia information of the target person is obtained, triple information is converted based on the pneumonia information, and associated data is extracted through the pneumonia knowledge graph to obtain target associated knowledge data, specifically:

[0069] Obtain pneumonia information of target personnel, perform semantic analysis and triple data conversion based on CNN semantic model, and form target triple data;

[0070] Perform vector transformation on the target entity in the target triple data and the entity data in the pneumonia knowledge graph based on the vector space model to obtain the first text vector and the second text vector respectively;

[0071] Calculating semantic similarity between the first text vector and the second text vector based on standard Euclidean distance, screening similarity data based on a threshold method, and obtaining a first similarity text vector and a second similarity text vector;

[0072] Acquire entity data corresponding to the first similarity text vector and the second similarity text vector and integrate the data to form associated entity data;

[0073] Based on the associative entity data, knowledge data is acquired from the pneumonia knowledge graph. The acquisition process is to locate the entity data from the knowledge graph based on the associative entity data, and traverse, query and count the relationship data, attribute data and associated entity data corresponding to the entity through the graph structure and the preset associated path distance to obtain the target associated knowledge data.

[0074] It should be noted that, based on the calculation method of vector text, the entity data can be analyzed for distance, so as to screen out the associated knowledge data of similarity. In addition, based on the vector space model, the role and weight of the entity text in the overall knowledge structure can be analyzed, so as to analyze the associated data of the target personnel information in the knowledge graph. The traditional technology simply compares the symptom information of pneumonia patients with the relevant disease information in the database, and the mining ability of the knowledge data associated with pneumonia is poor and the similar data extraction ability is poor. Therefore, through the present invention, the associated patient data of the target pneumonia patient can be effectively improved, and the diagnostic analysis of multiple schemes can be further realized. The first similarity text vector and the second similarity text vector have the characteristics of high semantic similarity, based on which the knowledge data of the subsequent knowledge graph is screened, and data judgment is effectively performed. The preset associated path distance is the maximum path setting distance between the entity and the entity in the structure diagram, which is used to screen out the associated data.

[0075] The target-related knowledge data includes corresponding associated entities, attributes, relationships and other knowledge data, which are used to describe the associated data existing in the knowledge graph. Through this associated data, similar symptoms, cases and related patients of the target patients can be grasped, and it has certain additional knowledge data to describe the correlation between the data.

[0076] According to an embodiment of the present invention, based on a preset clustering model, data classification is performed on the target associated knowledge data, and the classification process performs feasibility analysis of pneumonia pathology for each group of data, and multiple classification results are screened based on feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are respectively simulated and integrated with pneumonia patient information to form multiple groups of patient associated data, specifically:

[0077] Construct a clustering model based on DBSCAN, import the target associated knowledge data into the clustering model for clustering and grouping, and obtain multiple initial groups;

[0078] The knowledge data in each initial group is subjected to feasibility analysis of pneumonia cases. The feasibility analysis process is to determine the data values ​​with pathological abnormalities in the data and mark them, calculate the proportion of abnormal value data, and obtain feasibility based on the proportion value;

[0079] The initial group whose feasibility meets expectations is marked as the preferred group, and the associated data in the initial group that does not meet expectations is integrated and the clustering grouping and feasibility calculation analysis and judgment are performed cyclically;

[0080] Finally, all the preferred groups are labeled and multiple groups of related data are obtained;

[0081] Collect the preset patient data format structure, convert and align the formats of multiple sets of related data, and form multiple sets of patient related data.

[0082] It should be noted that the feasibility is obtained based on the proportion value, that is, the reciprocal of the proportion value is equal to the feasibility. The larger the proportion value, the higher the abnormality, the lower the feasibility, which reflects that the grouped data is less consistent with the pathological logic, and needs to be regrouped to obtain a more realistic related patient situation. The preset patient data format structure is the relevant format of the patient entry system, including various medical indicators, data arrangement of basic information, file format, etc.

[0083] In obtaining the target associated knowledge data, since the associated data generally has a large amount of data, and the pneumonia-related knowledge data in the data may contain the conditions of multiple pneumonia patients, the present invention clusters and groups the associated data to achieve further data mining and form multiple patient conditions. The multiple patient conditions have certain similarities with the target patients, and data collection is performed based on data mining. The associated data obtained can be further recommended for diagnosis and treatment analysis, so as to recommend data in multiple dimensions based on the knowledge graph, and the diagnosis and treatment recommendation information has better reference significance. The medical auxiliary level of the system can be greatly improved. The data in the patient-associated data has certain data predictability.

[0084] According to an embodiment of the present invention, the data of TCM diagnosis of pneumonia is collected, and a TCM diagnosis model of pneumonia is constructed based on a preset language model, multiple groups of related data are imported into the large model for TCM diagnosis scheme, and multiple groups of diagnosis and treatment schemes are formed, specifically:

[0085] Based on the system database, collect TCM diagnosis data of pneumonia;

[0086] Construct a large model of TCM diagnosis of pneumonia based on the preset language model;

[0087] The data is cleaned and converted into text format, and then imported into the pneumonia TCM diagnosis model for training and learning;

[0088] Import multiple groups of patient-related data into the pneumonia TCM diagnosis model for diagnosis and treatment analysis and generate multiple groups of diagnosis and treatment plans;

[0089] The multiple groups of diagnosis and treatment plans are displayed through a preset terminal.

[0090] It should be noted that the preset language model is built based on basic language models such as GPT-4 and PALM 2, and is specifically set by the user for semantic analysis and medical judgment.

[0091] According to an embodiment of the present invention, it also includes:

[0092] Based on the time dimension, the target personnel's pneumonia information is divided into periodic data to obtain multi-period pneumonia data;

[0093] Serializing the multi-cycle pneumonia data based on the time dimension to form target serialized data;

[0094] Based on the basic information of the target person, similarity data is screened from the pneumonia knowledge graph, and the retrieved patient data is obtained;

[0095] Tracing patients and obtaining information based on searching patient data and system database to obtain basic information of similar patients;

[0096] Obtain the periodic pneumonia data of similar patients through the basic information of similar patients, and perform time serialization on the periodic pneumonia data to form similar serialized data;

[0097] Based on the LSTM prediction algorithm, the target serialized data and the similar serialized data are predicted respectively to form the first prediction data and the second prediction data;

[0098] Based on the standard Euclidean distance, the difference between the first prediction data and the second prediction data is calculated and analyzed, and the second prediction data that is highly similar to the first prediction data is screened out through the threshold method and marked as high-similarity prediction data;

[0099] Based on the high similarity prediction data, the patient is traced and marked to obtain the recommended patient information, and the corresponding diagnosis and treatment process information is obtained based on the recommended patient information;

[0100] Extract data values ​​from the first predicted data and the multi-cycle pneumonia data and draw a data change curve based on the time dimension to obtain a lesion data curve graph;

[0101] Draw the numerical change curve of the multi-cycle pneumonia data of the recommended patients and the corresponding predicted data based on the time dimension to form a similar lesion data curve chart;

[0102] Generate a second recommended treatment plan for the target person through the diagnosis and treatment process information;

[0103] The lesion data curve graph, similar lesion data curve graph and the second recommended diagnosis and treatment plan are simultaneously sent to a preset terminal device for display.

[0104] It should be noted that the second recommended diagnosis and treatment plan is obtained based on the analysis of predictive lesion data. The plan is extracted by analyzing similar patients based on the predicted lesion data. Combined with the visualization curve, it can further assist doctors in diagnostic analysis. The analysis angle can be based on similar lesion analysis of historical patients. The basic information of the target person includes the patient's gender, age, occupation and other basic information. Based on the basic information, similar personnel are screened from the knowledge graph. The multi-cycle pneumonia data is the periodic health monitoring data obtained by dividing the time period based on all pneumonia information data of all target persons, including various medical indicator information, from which the pathological rules can be obtained. The similar patients generally include multiple people. Each similar patient corresponds to an independent second prediction data. The pneumonia prediction analysis of the target person corresponds to the first prediction data.

[0105] Figure 3 A block diagram of a pneumonia analysis and intelligent information recommendation system based on a knowledge graph of the present invention is shown.

[0106] The second aspect of the present invention further provides a pneumonia analysis and intelligent information recommendation system 3 based on a knowledge graph, the system comprising: a memory 31, a processor 32, the memory comprising a pneumonia analysis and intelligent information recommendation program based on a knowledge graph, the pneumonia analysis and intelligent information recommendation program based on a knowledge graph when executed by the processor implements the following steps:

[0107] Collect health big data of historical pneumonia patients, perform text semantic analysis and entity relationship extraction based on NLP on the health big data, and construct a pneumonia knowledge graph;

[0108] Acquire pneumonia information of target personnel, perform triple information conversion based on the pneumonia information, and extract associated data through the pneumonia knowledge graph to obtain target associated knowledge data;

[0109] Based on the preset clustering model, the target associated knowledge data is classified. The feasibility of pneumonia pathology of each group of data is analyzed during the classification process. Multiple classification results are screened based on the feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are simulated and integrated with pneumonia patient information to form multiple groups of patient associated data.

[0110] Collect TCM diagnosis data on pneumonia, and build a large TCM diagnosis model for pneumonia based on the preset language model. Import multiple sets of related data into the large model for TCM diagnosis and form multiple sets of diagnosis and treatment plans.

[0111] It should be noted that the health big data includes historical medical information, examination results, symptom descriptions, clinical data, pathological records and other data, which are used to describe the physical condition and pathological information of pneumonia patients.

[0112] According to an embodiment of the present invention, the health big data of historical pneumonia patients is collected, the health big data is subjected to NLP-based text semantic analysis and entity relationship extraction, and a pneumonia knowledge graph is constructed, specifically:

[0113] Based on the system database, obtain the health big data of historical pneumonia patients;

[0114] Perform data cleaning, data integration, and data redundancy removal preprocessing on the health big data;

[0115] The preprocessed health big data is converted into text format, and semantic analysis and triple data extraction are performed on the health big data based on the CNN semantic analysis model, wherein the triple data includes entity, attribute, and relationship data, and triple data is obtained;

[0116] A pneumonia knowledge graph is constructed based on the triple data.

[0117] It should be noted that the pneumonia knowledge graph is a knowledge graph constructed based on existing historical pneumonia patient data. Through the knowledge graph, we can further conduct knowledge mining on the data to obtain effective and practical knowledge data, providing data support for subsequent data recommendations.

[0118] According to an embodiment of the present invention, the health big data of historical pneumonia patients is collected, the health big data is subjected to NLP-based text semantic analysis and entity relationship extraction, and a pneumonia knowledge graph is constructed, and further includes:

[0119] Based on the AMIE algorithm, rule extraction and association rule learning are performed from the pneumonia knowledge graph to obtain new rule data;

[0120] Repeat the learning of association rules and the integration of new rule information until the new rule data reaches the preset data volume;

[0121] The new rule data is analyzed and evaluated for relationship and confidence calculated based on the pneumonia knowledge graph, and threshold screening is performed based on confidence to obtain the optimal relationship rules;

[0122] Import the optimal relationship rules into the pneumonia knowledge graph as new knowledge data.

[0123] It should be noted that the AMIE algorithm is a rule-based reasoning analysis algorithm. The introduction of this algorithm can mine new knowledge from the knowledge graph. In the case of existing structural knowledge, knowledge training and learning can be carried out to build a complete logical knowledge graph, further improving the effectiveness of subsequent knowledge retrieval. Through the knowledge graph, effective association analysis of pneumonia patients can be carried out in big data. Compared with the simple comparison of basic information by traditional technology, the knowledge graph can learn knowledge under the graph structure constructed by big data and obtain new associated knowledge data, so that when acquiring target associated data, new knowledge data associated with the target pneumonia patient can be obtained in the form of data mining.

[0124] According to an embodiment of the present invention, the pneumonia information of the target person is obtained, triple information is converted based on the pneumonia information, and associated data is extracted through the pneumonia knowledge graph to obtain target associated knowledge data, specifically:

[0125] Obtain pneumonia information of target personnel, perform semantic analysis and triple data conversion based on CNN semantic model, and form target triple data;

[0126] Perform vector transformation on the target entity in the target triple data and the entity data in the pneumonia knowledge graph based on the vector space model to obtain the first text vector and the second text vector respectively;

[0127] Calculating semantic similarity between the first text vector and the second text vector based on standard Euclidean distance, screening similarity data based on a threshold method, and obtaining a first similarity text vector and a second similarity text vector;

[0128] Acquire entity data corresponding to the first similarity text vector and the second similarity text vector and integrate the data to form associated entity data;

[0129] Based on the associative entity data, knowledge data is acquired from the pneumonia knowledge graph. The acquisition process is to locate the entity data from the knowledge graph based on the associative entity data, and traverse, query and count the relationship data, attribute data and associated entity data corresponding to the entity through the graph structure and the preset associated path distance to obtain the target associated knowledge data.

[0130] It should be noted that, based on the calculation method of vector text, the entity data can be analyzed for distance, so as to screen out the associated knowledge data of similarity. In addition, based on the vector space model, the role and weight of the entity text in the overall knowledge structure can be analyzed, so as to analyze the associated data of the target personnel information in the knowledge graph. The traditional technology simply compares the symptom information of pneumonia patients with the relevant disease information in the database, and the mining ability of the knowledge data associated with pneumonia is poor and the similar data extraction ability is poor. Therefore, through the present invention, the associated patient data of the target pneumonia patient can be effectively improved, and the diagnostic analysis of multiple schemes can be further realized. The first similarity text vector and the second similarity text vector have the characteristics of high semantic similarity, based on which the knowledge data of the subsequent knowledge graph is screened, and data judgment is effectively performed. The preset associated path distance is the maximum path setting distance between the entity and the entity in the structure diagram, which is used to screen out the associated data.

[0131] The target-related knowledge data includes corresponding associated entities, attributes, relationships and other knowledge data, which are used to describe the associated data existing in the knowledge graph. Through this associated data, similar symptoms, cases and related patients of the target patients can be grasped, and it has certain additional knowledge data to describe the correlation between the data.

[0132] According to an embodiment of the present invention, based on a preset clustering model, data classification is performed on the target associated knowledge data, and the classification process performs feasibility analysis of pneumonia pathology for each group of data, and multiple classification results are screened based on feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are respectively simulated and integrated with pneumonia patient information to form multiple groups of patient associated data, specifically:

[0133] Construct a clustering model based on DBSCAN, import the target associated knowledge data into the clustering model for clustering and grouping, and obtain multiple initial groups;

[0134] The knowledge data in each initial group is subjected to feasibility analysis of pneumonia cases. The feasibility analysis process is to determine the data values ​​with pathological abnormalities in the data and mark them, calculate the proportion of abnormal value data, and obtain feasibility based on the proportion value;

[0135] The initial group whose feasibility meets expectations is marked as the preferred group, and the associated data in the initial group that does not meet expectations is integrated and the clustering grouping and feasibility calculation analysis and judgment are performed cyclically;

[0136] Finally, all the preferred groups are labeled and multiple groups of related data are obtained;

[0137] Collect the preset patient data format structure, convert and align the formats of multiple sets of related data, and form multiple sets of patient related data.

[0138] It should be noted that the feasibility is obtained based on the proportion value, that is, the reciprocal of the proportion value is equal to the feasibility. The larger the proportion value, the higher the abnormality, the lower the feasibility, which reflects that the grouped data is less consistent with the pathological logic, and needs to be regrouped to obtain a more realistic related patient situation. The preset patient data format structure is the relevant format of the patient entry system, including various medical indicators, data arrangement of basic information, file format, etc.

[0139] In obtaining the target associated knowledge data, since the associated data generally has a large amount of data, and the pneumonia-related knowledge data in the data may contain the conditions of multiple pneumonia patients, the present invention clusters and groups the associated data to achieve further data mining and form multiple patient conditions. The multiple patient conditions have certain similarities with the target patients, and data collection is performed based on data mining. The associated data obtained can be further recommended for diagnosis and treatment analysis, so as to recommend data in multiple dimensions based on the knowledge graph, and the diagnosis and treatment recommendation information has better reference significance. The medical auxiliary level of the system can be greatly improved. The data in the patient-associated data has certain data predictability.

[0140] According to an embodiment of the present invention, the data of TCM diagnosis of pneumonia is collected, and a TCM diagnosis model of pneumonia is constructed based on a preset language model, multiple groups of related data are imported into the large model for TCM diagnosis scheme, and multiple groups of diagnosis and treatment schemes are formed, specifically:

[0141] Based on the system database, collect TCM diagnosis data of pneumonia;

[0142] Construct a large model of TCM diagnosis of pneumonia based on the preset language model;

[0143] The data is cleaned and converted into text format, and then imported into the pneumonia TCM diagnosis model for training and learning;

[0144] Import multiple groups of patient-related data into the pneumonia TCM diagnosis model for diagnosis and treatment analysis and generate multiple groups of diagnosis and treatment plans;

[0145] The multiple groups of diagnosis and treatment plans are displayed through a preset terminal.

[0146] It should be noted that the preset language model is built based on basic language models such as GPT-4 and PALM 2, and is specifically set by the user for semantic analysis and medical judgment.

[0147] According to an embodiment of the present invention, it also includes:

[0148] Based on the time dimension, the target personnel's pneumonia information is divided into periodic data to obtain multi-period pneumonia data;

[0149] Serializing the multi-cycle pneumonia data based on the time dimension to form target serialized data;

[0150] Based on the basic information of the target person, similarity data is screened from the pneumonia knowledge graph, and the retrieved patient data is obtained;

[0151] Tracing patients and obtaining information based on searching patient data and system database to obtain basic information of similar patients;

[0152] Obtain the periodic pneumonia data of similar patients through the basic information of similar patients, and perform time serialization on the periodic pneumonia data to form similar serialized data;

[0153] Based on the LSTM prediction algorithm, the target serialized data and the similar serialized data are predicted respectively to form the first prediction data and the second prediction data;

[0154] Based on the standard Euclidean distance, the difference between the first prediction data and the second prediction data is calculated and analyzed, and the second prediction data that is highly similar to the first prediction data is screened out through the threshold method and marked as high-similarity prediction data;

[0155] Based on the high similarity prediction data, the patient is traced and marked to obtain the recommended patient information, and the corresponding diagnosis and treatment process information is obtained based on the recommended patient information;

[0156] Extract data values ​​from the first predicted data and the multi-cycle pneumonia data and draw a data change curve based on the time dimension to obtain a lesion data curve graph;

[0157] Draw the numerical change curve of the multi-cycle pneumonia data of the recommended patients and the corresponding predicted data based on the time dimension to form a similar lesion data curve chart;

[0158] Generate a second recommended treatment plan for the target person through the diagnosis and treatment process information;

[0159] The lesion data curve graph, similar lesion data curve graph and the second recommended diagnosis and treatment plan are simultaneously sent to a preset terminal device for display.

[0160] It should be noted that the second recommended diagnosis and treatment plan is obtained based on the analysis of predictive lesion data. The plan is extracted by analyzing similar patients based on the predicted lesion data. Combined with the visualization curve, it can further assist doctors in diagnostic analysis. The analysis angle can be based on similar lesion analysis of historical patients. The basic information of the target person includes the patient's gender, age, occupation and other basic information. Based on the basic information, similar personnel are screened from the knowledge graph. The multi-cycle pneumonia data is the periodic health monitoring data obtained by dividing the time period based on all pneumonia information data of all target persons, including various medical indicator information, from which the pathological rules can be obtained. The similar patients generally include multiple people. Each similar patient corresponds to an independent second prediction data. The pneumonia prediction analysis of the target person corresponds to the first prediction data.

[0161] The present invention discloses a pneumonia analysis and intelligent information recommendation method and system based on a knowledge graph, based on the health big data of historical pneumonia patients and constructing a pneumonia knowledge graph, and obtaining target-related knowledge data related to pneumonia of target personnel; data classification is performed on the target-related knowledge data, and the classification process performs feasibility analysis of pneumonia pathology for each group of data, and multiple classification results are screened based on feasibility to obtain multiple groups of patient-related data; pneumonia TCM diagnosis data data are collected, and a pneumonia TCM diagnosis large model is constructed based on a preset language model, and multiple groups of associated data are imported into the large model for TCM diagnosis schemes, and multiple groups of diagnosis and treatment schemes are formed. The present invention recommends data in multiple dimensions based on the knowledge graph, and the diagnosis and treatment recommendation information has a better reference significance, which can greatly improve the medical auxiliary level of the system.

[0162] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0163] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0164] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0165] A person of ordinary skill in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0166] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0167] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A pneumonia analysis and intelligent information recommendation method based on knowledge graph, characterized in that: include: Collect health big data of historical pneumonia patients, perform text semantic analysis and entity relationship extraction based on NLP on the health big data, and construct a pneumonia knowledge graph; Acquire pneumonia information of target personnel, perform triple information conversion based on the pneumonia information, and extract associated data through the pneumonia knowledge graph to obtain target associated knowledge data; Based on the preset clustering model, the target associated knowledge data is classified. The feasibility of pneumonia pathology of each group of data is analyzed during the classification process. Multiple classification results are screened based on the feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are simulated and integrated with pneumonia patient information to form multiple groups of patient associated data. Collect TCM diagnosis data of pneumonia, and build a TCM diagnosis model of pneumonia based on the preset language model, import multiple groups of related data into the big model for TCM diagnosis, and form multiple groups of diagnosis and treatment plans; Among them, based on the preset clustering model, the target associated knowledge data is classified, the classification process performs feasibility analysis of pneumonia pathology for each group of data, and multiple classification results are screened based on feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are respectively simulated and integrated with pneumonia patient information to form multiple groups of patient associated data, specifically: Construct a clustering model based on DBSCAN, import the target associated knowledge data into the clustering model for clustering and grouping, and obtain multiple initial groups; The knowledge data in each initial group is subjected to feasibility analysis of pneumonia cases. The feasibility analysis process is to determine the data values ​​with pathological abnormalities in the data and mark them, calculate the proportion of abnormal value data, and obtain feasibility based on the proportion value; The initial group whose feasibility meets expectations is marked as the preferred group, and the associated data in the initial group that does not meet expectations is integrated and the clustering grouping and feasibility calculation analysis and judgment are performed cyclically; Finally, all the preferred groups are labeled and multiple groups of related data are obtained; Collect the preset patient data format structure, convert and align the formats of multiple sets of related data, and form multiple sets of patient related data.

2. According to claim 1, a pneumonia analysis and intelligent information recommendation method based on knowledge graph is characterized in that: The health big data of historical pneumonia patients is collected, the health big data is subjected to text semantic analysis and entity relationship extraction based on NLP, and a pneumonia knowledge graph is constructed, specifically: Based on the system database, obtain the health big data of historical pneumonia patients; Perform data cleaning, data integration, and data redundancy removal preprocessing on the health big data; The preprocessed health big data is converted into text format, and semantic analysis and triple data extraction are performed on the health big data based on the CNN semantic analysis model, wherein the triple data includes entity, attribute, and relationship data, and triple data is obtained; A pneumonia knowledge graph is constructed based on the triple data.

3. A pneumonia analysis and intelligent information recommendation method based on knowledge graph according to claim 2, characterized in that: The method of collecting health big data of historical pneumonia patients, performing text semantic analysis and entity relationship extraction based on NLP on the health big data, and constructing a pneumonia knowledge graph also includes: Based on the AMIE algorithm, rule extraction and association rule learning are performed from the pneumonia knowledge graph to obtain new rule data; Repeat the association rule learning and the integration of new rule information until the new rule data reaches the preset data volume; The new rule data is analyzed and evaluated for relationship and confidence calculated based on the pneumonia knowledge graph, and threshold screening is performed based on confidence to obtain the optimal relationship rules; Import the optimal relationship rules into the pneumonia knowledge graph as new knowledge data.

4. A method for pneumonia analysis and intelligent information recommendation based on knowledge graph according to claim 3, characterized in that: The target person's pneumonia information is obtained, triple information is converted based on the pneumonia information, and associated data is extracted through the pneumonia knowledge graph to obtain target associated knowledge data, specifically: Obtain pneumonia information of target personnel, perform semantic analysis and triple data conversion based on CNN semantic model, and form target triple data; Perform vector transformation on the target entity in the target triple data and the entity data in the pneumonia knowledge graph based on the vector space model to obtain the first text vector and the second text vector respectively; Calculating semantic similarity between the first text vector and the second text vector based on standard Euclidean distance, screening similarity data based on a threshold method, and obtaining a first similarity text vector and a second similarity text vector; Acquire entity data corresponding to the first similarity text vector and the second similarity text vector and integrate the data to form associated entity data; Based on the associative entity data, knowledge data is acquired from the pneumonia knowledge graph. The acquisition process is to locate the entity data from the knowledge graph based on the associative entity data, and traverse, query and count the relationship data, attribute data and associated entity data corresponding to the entity through the graph structure and the preset associated path distance to obtain the target associated knowledge data.

5. According to claim 1, a pneumonia analysis and intelligent information recommendation method based on knowledge graph is characterized in that: The method collects TCM diagnosis data of pneumonia, builds a TCM diagnosis model of pneumonia based on a preset language model, imports multiple groups of related data into the large model for TCM diagnosis scheme, and forms multiple groups of diagnosis and treatment schemes, specifically: Based on the system database, collect TCM diagnosis data of pneumonia; Construct a large model of TCM diagnosis of pneumonia based on the preset language model; The data is cleaned and converted into text format, and then imported into the pneumonia TCM diagnosis model for training and learning; Import multiple groups of patient-related data into the pneumonia TCM diagnosis model for diagnosis and treatment analysis and generate multiple groups of diagnosis and treatment plans; The multiple groups of diagnosis and treatment plans are displayed through a preset terminal.

6. A pneumonia analysis and intelligent information recommendation system based on knowledge graph, characterized in that: The system includes: a memory and a processor, wherein the memory includes a knowledge graph-based pneumonia analysis and intelligent information recommendation program, and the knowledge graph-based pneumonia analysis and intelligent information recommendation program implements the following steps when executed by the processor: Collect health big data of historical pneumonia patients, perform text semantic analysis and entity relationship extraction based on NLP on the health big data, and construct a pneumonia knowledge graph; Acquire pneumonia information of target personnel, perform triple information conversion based on the pneumonia information, and extract associated data through the pneumonia knowledge graph to obtain target associated knowledge data; Based on the preset clustering model, the target associated knowledge data is classified. The feasibility of pneumonia pathology of each group of data is analyzed during the classification process. Multiple classification results are screened based on the feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are simulated and integrated with pneumonia patient information to form multiple groups of patient associated data. Collect TCM diagnosis data of pneumonia, and build a TCM diagnosis model of pneumonia based on the preset language model, import multiple groups of related data into the big model for TCM diagnosis, and form multiple groups of diagnosis and treatment plans; Among them, based on the preset clustering model, the target associated knowledge data is classified, the classification process performs feasibility analysis of pneumonia pathology for each group of data, and multiple classification results are screened based on feasibility, and finally multiple groups of associated data are obtained. Through the patient data format structure, the multiple groups of associated data are respectively simulated and integrated with pneumonia patient information to form multiple groups of patient associated data, specifically: Construct a clustering model based on DBSCAN, import the target associated knowledge data into the clustering model for clustering and grouping, and obtain multiple initial groups; The knowledge data in each initial group is subjected to feasibility analysis of pneumonia cases. The feasibility analysis process is to determine the data values ​​with pathological abnormalities in the data and mark them, calculate the proportion of abnormal value data, and obtain feasibility based on the proportion value; The initial group whose feasibility meets expectations is marked as the preferred group, and the associated data in the initial group that does not meet expectations is integrated and the clustering grouping and feasibility calculation analysis and judgment are performed cyclically; Finally, all the preferred groups are labeled and multiple groups of related data are obtained; Collect the preset patient data format structure, convert and align the formats of multiple sets of related data, and form multiple sets of patient related data.

Images