Diabetes data sharing and analysis system based on cloud platform

Abstract

The present invention relates to the field of medical data processing technology, specifically to a diabetes data sharing and analysis system based on a cloud platform, wherein the system includes a diabetes data collaborative processing module, a patient feature hierarchical management module, a diabetes association path analysis module, and an intelligent semantic reasoning module. In the present invention, the data of diabetic patients in medical institutions are collected through a cloud platform to improve data accuracy and effectiveness, eliminate interference factors and merge data from multiple sources, reduce redundancy and conflict, and improve cross-institutional data integration capabilities. Patient features are hierarchically processed through feature matching and priority sorting to form a hierarchical feature network structure for easy management and analysis, analyze node association strength and path relationships, eliminate weak nodes and expand interactive nodes, improve path analysis depth and research value, match keywords and paths based on semantic reasoning, screen high-frequency nodes and semantic paths, optimize query accuracy and practicality, and enhance data analysis intelligence and management efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a diabetes data sharing and analysis system based on a cloud platform. Background Art

[0002] The field of medical data processing technology includes technical methods for collecting, storing, transmitting, analyzing and sharing medical-related data using information technology. The core content of this technical field is to use computers and data analysis technology to uniformly manage and deeply mine scattered medical data to meet the needs of health management, disease monitoring and medical research. The field of medical data processing technology includes key technical contents such as data standardization, data security protection, distributed storage and big data analysis to ensure the validity, integrity and availability of medical data, while supporting the interconnection and analysis applications of medical data on different platforms.

[0003] Among them, the diabetes data sharing and analysis system refers to a medical data processing system based on information technology, which is specifically used to realize multi-party sharing and efficient analysis of diabetes-related data. The system is aimed at the health management needs of diabetic patients, and mainly covers the collection of patient data information, cross-institutional data sharing and cloud platform-based data processing functions. The system uses cloud platform technology to uniformly format diabetes-related data from different sources and upload them to the cloud. It realizes the secure sharing of data among different users through access control mechanisms, and classifies, correlates and analyzes health data through data analysis methods, and predicts trends, etc., to support multi-dimensional research and application of related data.

[0004] In the collection, storage and analysis of diabetes data, the existing technology has problems such as inconsistent data formats and uneven quality of source data, which makes data integration difficult, and data conflicts and redundancy problems frequently occur, affecting the accuracy and completeness of subsequent analysis. In terms of data sharing, the existing technology relies on a single access control mechanism and fails to flexibly adjust the sharing strategy according to the needs of different users, resulting in low efficiency and even security risks in data sharing. At the data analysis and classification level, the existing technology fails to fully consider the complexity and individual differences of patients' multidimensional characteristics. The classification method is relatively simple and it is difficult to accurately reflect the characteristics of patients' conditions and their dynamic changes. In path association analysis, the existing technology does not have enough depth in mining the node association strength and path interaction relationship, resulting in some potentially valuable paths being ignored. In query analysis and semantic reasoning, the existing technology has low accuracy in matching keywords and path association analysis, and the practicality of the query results is limited. It is difficult to support complex medical research or decision-making needs, which directly limits the depth and breadth of diabetes data sharing and analysis systems in data mining, prediction and application, and cannot meet the efficient needs of precision medicine and cross-institutional data collaboration. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and propose a diabetes data sharing and analysis system based on a cloud platform.

[0006] In order to achieve the above purpose, the present invention adopts the following technical solution: A diabetes data sharing and analysis system based on a cloud platform includes:

[0007] The diabetes data collaborative processing module extracts the patient's blood sugar data, diet records and activity behavior data based on the diabetic patient data of the cloud platform medical institution, analyzes the data batches and multi-dimensional features, filters out noise items and duplicate data, removes interference factors and merges the data to generate encrypted data sets of patient data;

[0008] The patient feature hierarchical management module identifies the patient's diabetes type, drug use record, and blood sugar fluctuation trend based on the patient data encrypted data group, performs feature value matching and classification marking, generates patient feature classification results, calls hierarchical information to prioritize shared feature nodes, and generates a patient hierarchical feature network structure;

[0009] The diabetes association path analysis module analyzes the association strength, interaction records and hierarchical relationship paths between nodes based on the patient hierarchical characteristic network structure, removes low-frequency nodes and weakly associated paths, obtains multidimensional path data, performs extended analysis on the interaction nodes between paths, and generates a diabetes interaction path model;

[0010] Based on the diabetes interaction path model, the intelligent semantic reasoning module analyzes the relationship between query content keywords and model nodes, calls path interaction nodes and path hierarchy information for keyword matching, obtains a set of keyword-associated paths, extracts collaborative data content in combination with data sharing, screens high-frequency nodes that match the query content, and generates diabetes data semantic analysis path results.

[0011] As a further solution of the present invention, the steps of obtaining the patient data encrypted data group are specifically as follows:

[0012] Based on the data of diabetic patients in medical institutions on the cloud platform, the blood sugar data, dietary records and activity behavior data of patients are extracted, and the blood sugar data, dietary records and activity behavior data are preliminarily compared through the data source identification and record timestamp. The three types of data are multi-dimensionally indexed and matched according to the patient ID to generate a multi-dimensional data mapping relationship;

[0013] Based on the multi-dimensional data mapping relationship, noise items and duplicate data are screened and removed, and deviation factors are judged and data weights are adjusted by identifying the deviation behaviors of the three types of data, using the formula: ;

[0014] generating an adjusted feature data relationship;

[0015] in, represents the deviation factor adjustment value, is the feature data weight, is the real-time value of the data, is the median of the standard value of the characteristic data, is the total amount of data, is the data sequence index;

[0016] The adjusted characteristic data relationship is called, and noise is eliminated and multi-source data is merged for the three types of data, namely blood sugar, diet and activity behavior, according to matching features, and encryption is performed through the patient ID to generate a patient data encrypted data group.

[0017] As a further solution of the present invention, the step of obtaining the patient characteristic classification result is specifically as follows:

[0018] Calling the blood sugar record, diet record and medication use record fields in the patient data encrypted data group, analyzing the patient's blood sugar fluctuation range, dietary component intake and medication type changes, performing merging processing on each field data point according to the patient ID, and generating an initial index set of patient characteristics;

[0019] For the blood sugar fluctuation amplitude field in the patient characteristic initial index set, combined with the patient's drug use record and dietary intake, determine whether it meets the diabetes type characteristics, input the characteristic value into the patient classification threshold, and use the formula: ;

[0020] The diabetes type matching value is calculated;

[0021] in, Matching value for diabetes type, is the blood sugar fluctuation weight, is the blood sugar fluctuation data value, is the median value of blood sugar fluctuation, is the blood glucose matching parameter adjustment coefficient, is the dietary intake weight, is the characteristic value of dietary component intake, Adjust parameters for diet;

[0022] The diabetes type matching value is called, the patient's characteristic value is matched to the standard classification template, the patient's diabetes type, drug use record and blood sugar fluctuation trend are classified and marked according to the template classification label, and the patient characteristic classification result is generated.

[0023] As a further solution of the present invention, the step of obtaining the patient hierarchical feature network structure is specifically as follows:

[0024] Based on the patient feature classification results, feature nodes are extracted, node-related attributes are classified according to classification labels, nodes are preliminarily grouped according to weight parameters and correlation indicators, unique identifiers are assigned to each group of nodes, and a node grouping attribute set is obtained;

[0025] Based on the node grouping attribute set, the attribute values ​​of the shared feature nodes are extracted, and the node attribute values ​​are matched for relevance in combination with the hierarchical information. The priorities of the shared feature nodes are analyzed using weight parameters and node distribution frequencies, and the priorities are arranged to obtain a hierarchical shared feature priority value table;

[0026] Based on the hierarchical shared feature priority value table, the associations between feature nodes are identified in order of priority, and the priority and group associations between nodes are used as connection parameters to adjust the connection strength in the association to obtain the patient hierarchical feature network structure.

[0027] As a further solution of the present invention, the step of acquiring the multi-dimensional path data is specifically as follows:

[0028] Calling the associated records of each node in the patient hierarchical feature network structure, extracting the interaction strength, interaction times and hierarchical path relationship between nodes, merging the interaction strength cumulative values, calculating the path cumulative interaction weight according to the interaction times, and establishing an initial path relationship diagram between nodes;

[0029] For the initial path relationship graph between the nodes, identify the strength value and association weight of each path, remove low-frequency paths and weakly associated paths based on the path frequency, and use the formula: ;

[0030] Generate optimized path association data set;

[0031] in, is the path strength weight, is the node association weight, is the path interaction strength, is the path interaction frequency, is the path weight adjustment parameter, Adjustment parameters for path frequency;

[0032] The path strength weight in the optimized path association data set is called to match the hierarchical path relationship, and the association weight and path strength value are integrated into the hierarchical path structure to obtain multi-dimensional path data.

[0033] As a further solution of the present invention, the steps for obtaining the diabetes interaction path model are specifically as follows:

[0034] Based on the multi-dimensional path data, the location information and structure of the interaction nodes between the paths are extracted, the geometric characteristics and connection properties of the nodes are analyzed, the direct relationships between the nodes are classified and processed, the interaction nodes are numbered, the structural relationships between the nodes are summarized, and the basic data of the path interaction nodes are obtained;

[0035] Based on the basic data of the path interaction nodes, the number of repeated occurrences of the nodes in the path is identified to obtain the node connection frequency, the direction data of the connection relationship between the paths is counted, the node association strength is analyzed, the node group is screened, and the node association degree distribution relationship is generated by combining the path intersection relationship data;

[0036] Based on the node association degree distribution relationship, the association path connection characteristics in the node group are analyzed, the node weight distribution is parsed according to the path direction data, and the dynamic characteristics between the paths are combined to integrate the dynamic characteristics between the nodes and the paths to generate a diabetes interaction path model.

[0037] As a further solution of the present invention, the step of obtaining the keyword associated path set is specifically:

[0038] Calling the path node data in the diabetes interaction path model, analyzing the query content keywords and extracting key fields, analyzing the semantic relevance of the keywords in the path nodes, matching the semantic weights of the keywords and the path nodes, extracting the node set with a matching degree higher than a threshold, and establishing an initial matching relationship between the keywords and the nodes;

[0039] Based on the initial matching relationship between the keyword and the node, combined with the path interaction strength and the node hierarchical relationship, the matching weight between the keyword and the path is calculated using the formula:

[0040] ;

[0041] Generate a dataset of keyword and path interaction relationships;

[0042] in, Indicates the matching weight between keywords and paths. To query content weight, is the semantic similarity between keywords and path nodes, is the association weight at the path level, is the path interaction weight;

[0043] The keyword-path interaction relationship dataset is called, the keyword matching weight value is integrated with the path interaction strength and the hierarchical relationship, the paths with matching weight values ​​higher than the set threshold are screened, and the associated levels of the paths are grouped and classified to generate a keyword associated path set.

[0044] As a further solution of the present invention, the steps for obtaining the diabetes data semantic analysis path result are specifically as follows:

[0045] Based on the keyword-associated path set, the interactive node information in the path is extracted, the node distribution and connection characteristics are analyzed, the geometric characteristics and relationship weights are analyzed, the path structure data is sorted, and the data sharing content is classified in combination with the logical relationship between the nodes to obtain the keyword path data sharing content;

[0046] Based on the keyword path data sharing content, analyze the association strength according to the frequency of occurrence of nodes in the path, count the path node connection data, locate the path nodes that meet the conditions, screen the high-frequency nodes, classify and summarize them according to the node interaction frequency and position relationship, and generate a high-frequency node matching set;

[0047] Based on the high-frequency node matching set, the association between high-frequency nodes is analyzed, the semantic features of path nodes are extracted, the node connection rules are integrated, the semantic association features of path data are analyzed, and the semantic analysis path results of diabetes data are obtained.

[0048] Compared with the prior art, the advantages and positive effects of the present invention are:

[0049] In the present invention, blood glucose data, dietary records and activity behavior data of diabetic patients in medical institutions are collected based on a cloud platform, and batch analysis, multi-dimensional feature extraction and noise screening are performed on the data, so as to significantly improve the accuracy and effectiveness of the data, ensure that the generated data is unified and convenient for subsequent processing, and construct a high-quality encrypted data group by eliminating interference factors and merging multi-source data, reduce data redundancy and conflict problems, and improve the operability of cross-institutional data integration. Through feature matching, classification and priority sorting based on data, the patient's characteristics are hierarchically processed to form a hierarchical feature network structure, so that the feature association is clearer and convenient for efficient management and analysis. By analyzing the association strength, interaction records and path relationships between nodes, low-frequency nodes and weakly associated paths are eliminated and interactive nodes are expanded, the depth and breadth of path analysis are enhanced, and the research value of diabetes-related data is effectively improved. Based on the interactive path model and semantic reasoning logic, the query content keywords are matched, and the high-frequency nodes and semantically associated paths are screened out in combination with the data sharing mechanism, the accuracy of the query and the practicality of the associated path are optimized, and the scientificity and precision of the data analysis results are further improved. Through multi-dimensional analysis of data and hierarchical relationship mining, the intelligent level of data processing and the efficiency of diabetes data management are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a system flow chart of the present invention;

[0051] Figure 2 Flow chart for encrypting a data set for patient data in the present invention;

[0052] Figure 3A flow chart showing the patient characteristics classification results in the present invention;

[0053] Figure 4 A flowchart of the patient stratification feature network structure in the present invention;

[0054] Figure 5 It is a flow chart of multi-dimensional path data in the present invention;

[0055] Figure 6 is a flow chart of the diabetes interaction pathway model in the present invention;

[0056] Figure 7 It is a flow chart of the keyword association path set in the present invention;

[0057] Figure 8 The flowchart of the semantic analysis path results of diabetes data in the present invention. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0059] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0060] See also Figure 1 , the cloud-based diabetes data sharing and analysis system includes:

[0061] The diabetes data collaborative processing module extracts the patient's blood sugar data, diet records and activity behavior data based on the diabetic patient data of the cloud platform medical institution, analyzes the data batches and multi-dimensional features, filters out noise items and duplicate data, removes interference factors and merges the data to generate encrypted data sets of patient data;

[0062] The patient feature hierarchical management module identifies the patient's diabetes type, drug use record, and blood sugar fluctuation trend based on the patient data encrypted data group, performs feature value matching and classification marking, generates patient feature classification results, calls hierarchical information to prioritize shared feature nodes, and generates a patient hierarchical feature network structure;

[0063] The diabetes association path analysis module analyzes the association strength, interaction records and hierarchical relationship paths between nodes based on the hierarchical characteristic network structure of patients, removes low-frequency nodes and weakly associated paths, obtains multi-dimensional path data, and performs extended analysis on the interaction nodes between paths to generate a diabetes interaction path model;

[0064] The intelligent semantic reasoning module is based on the diabetes interaction path model, analyzes the relationship between query content keywords and model nodes, calls path interaction nodes and path hierarchy information for keyword matching, obtains a set of keyword-associated paths, extracts collaborative data content in combination with data sharing, screens high-frequency nodes that match the query content, and generates diabetes data semantic analysis path results.

[0065] The encrypted data group of patient data includes blood sugar data, diet record data, and activity behavior data. The patient feature classification results are specifically diabetes type classification, drug use classification, and blood sugar fluctuation classification. The patient hierarchical feature network structure includes hierarchical feature nodes, hierarchical priority nodes, and shared feature nodes. The multidimensional path data is specifically association strength data, interaction record data, and hierarchical relationship data. The diabetes interaction path model includes high-frequency interaction nodes, path interaction relationships, and extended interaction nodes. The keyword-associated path set includes a keyword node set, a path interaction set, and a hierarchical keyword set. The diabetes data semantic analysis path results include a high-frequency query path, a keyword matching path, and a semantic association path.

[0066] See also Figure 2 , the specific steps for obtaining the patient data encrypted data group are:

[0067] Based on the data of diabetic patients in medical institutions on the cloud platform, the blood sugar data, dietary records and activity behavior data of patients are extracted, and the blood sugar data, dietary records and activity behavior data are preliminarily compared through the data source identification and record timestamp. The three types of data are multi-dimensionally indexed and matched according to the patient ID to generate a multi-dimensional data mapping relationship;

[0068] The patient's blood sugar record, diet record and activity behavior data are obtained by calling the cloud platform database interface. Each data carries a timestamp and patient ID field. The timestamp and ID fields are used as joint conditions to extract data subsets for each patient. For blood sugar data, the original data points collected by the blood sugar monitoring device are used to calibrate the standard equipment to determine the deviation range, determine whether it is valid data, and eliminate invalid data. For diet record data, the diet content uploaded or manually entered by the patient is quantified into specific nutritional data points using the food composition table, such as carbohydrate content, fat content, protein content, etc., and sorted in chronological order. For activity behavior data, the number of steps, exercise time and exercise intensity data collected by the sports tracking device are used to align the time axis and convert units. All data points are indexed according to the patient ID, and multiple types of data are stored in the form of a multidimensional array. By analyzing the intersection range of the time axis, it is determined whether there is an overlapping range of different types of data, and the overlapping part is extracted as the valid intersection range of the data, and finally a multi-dimensional data mapping relationship is generated.

[0069] Based on the multi-dimensional data mapping relationship, noise items and duplicate data are screened and removed. By identifying the deviation behavior of three types of data, deviation factor judgment and data weight adjustment are performed using the formula:

[0070] ;

[0071] generating an adjusted feature data relationship;

[0072] in, represents the deviation factor adjustment value, is the feature data weight, is the real-time value of the data, is the median of the standard value of the characteristic data, is the total amount of data, is the data sequence index;

[0073] The benefit of the formula is that by assigning different weight parameters to the degree of deviation of each type of data , combined with the standard value By adjusting the difference between the two, noise and outliers can be eliminated on the basis of comprehensive consideration of data quality, thus improving the accuracy of feature extraction.

[0074] Represents the adjustment value of the deviation factor, which is used to quantify the result of data feature adjustment. is a weight parameter, which is set according to the importance of different types of data. For example, the weight of blood sugar data can be determined based on the accuracy of the data collection equipment and the fluctuation of the patient's historical data. The actual value of a single data point, such as blood sugar level, carbohydrate content, and exercise steps. is the median of the standard value of the characteristic data, which is calculated by extracting the median of the large sample data distribution. is the total amount of data, is the total number of samples involved in the calculation, Index for data;

[0075] For example, for a feature set containing 5 data , where the weight , median of standard value , the calculation process is:

[0076] Calculate the deviation value: ;

[0077] Calculate the weighted deviation value: ;

[0078] Calculate the sum and divide by the total amount of data:

[0079] ;

[0080] ;

[0081] Square root calculation: ;

[0082] This result shows that the deviation factor adjustment value The overall deviation of the current feature data is quantified. A lower deviation value indicates that the data quality is higher and the feature value is more in line with the standard range. This result can be directly used to generate the adjusted feature data relationship.

[0083] The adjusted feature data relationship is called to remove noise and merge multi-source data for the three types of data, namely blood sugar, diet and activity behavior, based on matching features, and encrypted through the patient ID to generate an encrypted data group of patient data;

[0084] By extracting the adjusted value of blood sugar data, using the timeline and patient ID matching mechanism, the dietary record data and activity behavior data are compared one by one, and the fluctuation of blood sugar data is correlated with the content of carbohydrates, fats, and proteins in the dietary records. The linear fitting model is used to calculate the correlation coefficient value to determine whether the correlation is significant. The significantly correlated eigenvalues ​​are classified into the same data point. For activity behavior data, the adjusted number of steps, exercise time and exercise intensity are extracted, combined with the data of various nutrients in the dietary records, and the calorie metabolism balance value at each time point is calculated through the balance formula of exercise consumption coefficient and intake components. The adjusted data points are encrypted and an asymmetric encryption algorithm is used to encrypt them one by one and output them as an encrypted data group of patient data.

[0085] See also Figure 3 , the specific steps for obtaining the patient characteristic classification results are:

[0086] Call the blood sugar record, diet record and medication record fields in the patient data encryption data group, analyze the patient's blood sugar fluctuation range, dietary intake and medication type changes, perform merging processing on each field data point according to the patient ID, and generate the initial index set of patient characteristics;

[0087] First, the patient ID is used as the primary key to merge the blood sugar fluctuations, dietary intake components and drug use records of each patient. During the merging process, the time axis of each data point is aligned through the timestamp field to ensure that each record corresponds to a unique time point. For blood sugar fluctuation data, the daily average blood sugar value and blood sugar fluctuation range of each patient are extracted, and the abnormal data are eliminated using the error range of the blood sugar testing equipment. For dietary record data, the nutritional data of each meal, including carbohydrate, fat and protein content, are quantified through the food composition table, and the total daily intake is summarized and classified into the patient ID index. For drug use record data, the frequency and dosage of daily medication are quantified, and the main drug types are extracted as one of the classification features. After completing the above processing, all data fields are merged according to the patient ID index to form the initial index set of patient characteristics.

[0088] For the blood sugar fluctuation amplitude field in the initial index of patient characteristics, combined with the patient's medication use record and dietary intake, determine whether it meets the diabetes type characteristics, input the characteristic value into the patient classification threshold, and use the formula:

[0089] ;

[0090] The diabetes type matching value is calculated;

[0091] in, Matching value for diabetes type, is the blood sugar fluctuation weight, is the blood sugar fluctuation data value, is the median value of blood sugar fluctuation, is the blood glucose matching parameter adjustment coefficient, is the dietary intake weight, is the characteristic value of dietary component intake, Adjust parameters for diet;

[0092] The formula is useful in that it introduces parameters for blood sugar fluctuations and dietary component weights. and , and the adjustment factor and , achieving a multi-dimensional comprehensive evaluation of blood sugar fluctuations and dietary characteristics, which helps to improve the accuracy and detail of classification results;

[0093] is the classification matching value, which indicates the matching degree between the feature value and the diabetes type template. is the blood sugar fluctuation weight, which is calculated based on the fluctuation range of blood sugar data. The weight is dynamically adjusted as the abnormal proportion of data changes. For blood sugar fluctuation data, the difference between the daily average blood sugar value and its daily maximum fluctuation is extracted as the data point. is the median value of blood sugar fluctuation, which is calculated based on the distribution of blood sugar fluctuations in the patient population. The adjustment parameters for matching blood sugar data are set according to the patient's historical blood sugar fluctuation data. The dietary component weight is set according to the degree of impact of nutrient intake on blood sugar. The daily dietary nutrient intake value includes carbohydrate, protein and fat intake. Adjust parameters for dietary data, set based on the patient's historical dietary data;

[0094] Given a set of data, the blood sugar fluctuation range of a patient , median blood sugar fluctuation , blood sugar weight , adjustment parameters , dietary intake values (carbohydrates, proteins and fats respectively), dietary weight , adjustment parameters ;

[0095] Blood sugar calculation:

[0096] ;

[0097] ;

[0098] ;

[0099] ;

[0100] Dietary Portion Calculation:

[0101] ;

[0102] ;

[0103] ;

[0104] Comprehensive calculation: ;

[0105] This result shows that the classification matching value It is 4.88, indicating that the patient's characteristic value highly matches a certain type of diabetes classification template, and this matching value can be directly used for subsequent classification result generation.

[0106] Call the diabetes type matching value, match the patient's feature value to the standard classification template, classify and mark the patient's diabetes type, drug use record and blood sugar fluctuation trend according to the template classification label, and generate the patient feature classification result;

[0107] First, determine whether the diabetes type matching value falls into the matching interval of each type of template based on its matching value. For example, if the matching value is 4.88 and falls into the template interval 4-5, the patient is marked as a specific type of diabetes. Then extract the patient's drug usage records, and determine whether the patient uses a specific type of drug through the matching relationship between the drug type label and the drug feature value in the template. If the usage record matches the template feature value, it is marked as meeting the template drug usage characteristics. Finally, extract the patient's blood sugar fluctuation trend, and classify and mark the patient's trend characteristics through the upper and lower limits of daily fluctuations in blood sugar data. For example, if the fluctuation amplitude is within a certain template type feature interval, it is marked as the corresponding trend type. Finally, the patient feature classification results are generated through the above classification and categorization markings.

[0108] See also Figure 4 ,The specific steps for obtaining the patient hierarchical feature network structure are:

[0109] Based on the patient feature classification results, feature nodes are extracted, node-related attributes are classified according to classification labels, nodes are preliminarily grouped according to weight parameters and correlation indicators, unique identifiers are assigned to each group of nodes, and node grouping attribute sets are obtained;

[0110] First, the characteristic information of patient data in the cloud platform is extracted, and the characteristics are classified into corresponding categories through classification labels, such as age, gender, body mass index, blood sugar level, etc. Under each category, the associated attributes of the nodes are further extracted, such as the patient's living habits, family medical history, medication records, etc. The correlation index is calculated by analyzing the node correlation attributes. This index is quantified based on the frequency of data interaction between nodes, the strength of correlation distribution, and the probability of attribute co-occurrence within the patient group. The weight parameter is used to adjust the relative importance of different features. In the preliminary grouping stage, the correlation score of each node on different attributes is calculated, and the nodes are grouped according to the preset weight ratio. The grouping rules are used to ensure that each group of nodes has a certain homogeneity in the classification label. For example, nodes with higher correlation are classified into the same group, and each node after grouping is marked with a unique identifier for subsequent attribute extraction and analysis. A node grouping attribute set is generated for each group of nodes, where each attribute set includes the characteristic information of all nodes in the group and a comprehensive description of the associated attributes.

[0111] Based on the node grouping attribute set, the attribute values ​​of the shared feature nodes are extracted, and the node attribute values ​​are matched for relevance in combination with the hierarchical information. The priority of the shared feature nodes is analyzed using the weight parameter and the node distribution frequency, and the priority is arranged to obtain a hierarchical shared feature priority value table.

[0112] First, shared feature nodes are extracted from each node group attribute set, and feature attributes that frequently appear in multiple nodes within each group are selected, such as the blood sugar fluctuation value range, eating habits or exercise frequency shared within a specific group. After the attribute values ​​of the shared feature nodes are extracted, the node attribute values ​​are matched in combination with the hierarchical information of the patient data. The hierarchical information is classified according to the patient's feature type. For example, patients are divided into early, middle and late stages according to the progression stage of diabetes, and the shared feature nodes are mapped according to the hierarchical categories. The correlation between node attribute values ​​is calculated, and the distribution law of shared features between different levels is analyzed. For shared feature nodes, priority is sorted according to their distribution frequency and weight parameters in the hierarchical data. The weight parameters are used to strengthen the focus on high-frequency features and control the degree of participation of low-frequency features. The priority is arranged according to the weighted result of the frequency score and the weight. Finally, a hierarchical shared feature priority value table is generated, in which the priority value of each node clearly identifies its importance and relevance in the hierarchical structure. The priority value table provides accurate basic data for subsequent feature node association analysis.

[0113] Based on the hierarchical shared feature priority value table, the associations between feature nodes are identified in order of priority, and the priority and the group association between nodes are used as connection parameters to adjust the connection strength in the association to obtain the hierarchical feature network structure of patients;

[0114] According to the arrangement order of the priority value table, the relevance of shared feature nodes is identified one by one. By reading the numerical information in the priority value table, the shared feature nodes at the same level or across levels are identified. Through the identification of group association, it is determined which nodes have significant correlation. For each pair of feature nodes, the correlation strength is calculated. The correlation strength is determined based on the matching degree of node attributes, the product of priorities and the co-occurrence frequency between groups. While identifying the association, the connection strength between nodes is adjusted. The adjustment rules are based on the priority value of the shared feature and the distribution characteristics between nodes. For example, nodes with high priorities have higher weights in the network structure, thereby amplifying their connection strength. The connection strength is iteratively updated multiple times through group association parameters to ensure that the final network structure can accurately reflect the correlation of patient characteristics in the hierarchical information. Finally, a patient hierarchical feature network structure is generated, providing precise basic support for further sharing and analysis of diabetes data.

[0115] See also Figure 5 , the specific steps for obtaining multidimensional path data are:

[0116] Call the association record of each node in the patient's hierarchical feature network structure, extract the interaction strength, interaction times and hierarchical path relationship between nodes, merge the cumulative interaction strength value, calculate the path cumulative interaction weight according to the number of interactions, and establish the initial path relationship diagram between nodes;

[0117] First, the interaction times and interaction intensity data between nodes are extracted. The interaction times are obtained by counting the total number of interaction records of each node, and the interaction intensity is obtained by calculating the cumulative interaction weight of each node to its adjacent nodes. All interaction times and interaction weight data need to be normalized according to the time axis to eliminate the influence of different node data volume and time distribution. Combined with the hierarchical relationship of the hierarchical path, the interaction data of each node is classified by hierarchical index information, and the node data on the same hierarchical path are integrated into a group. At the same time, the interaction centrality of each group of data is calculated. The centrality is quantified by the interaction ratio between nodes. The interaction frequency of each node is calculated using time distribution data. The number of interaction records of nodes on each path is counted in segments and the mean of its time distribution is taken to obtain the average number of interactions and its fluctuation range, so as to determine the activity of the node. For low-frequency nodes below a specific threshold, the nodes are eliminated according to the historical distribution of the number of interactions and the interaction frequency, and the interaction data in the node structure is updated to finally generate the initial path relationship diagram between nodes.

[0118] For the initial path relationship graph between nodes, identify the strength value and association weight of each path, and remove low-frequency paths and weakly associated paths based on the path frequency, using the formula:

[0119] ;

[0120] Generate optimized path association data set;

[0121] in, is the path strength weight, is the node association weight, is the path interaction strength, is the path interaction frequency, is the path weight adjustment parameter, Adjustment parameters for path frequency;

[0122] The benefit of the formula is that it associates weights with nodes , path interaction intensity and path interaction frequency By combining the multi-dimensional path weight optimization model, the adjustment parameters and Dynamically control the contribution value of path association, enhancing the rationality of low-frequency path elimination and the granularity of path weight calculation;

[0123] is the path strength weight, which represents the comprehensive weight value of the path association. is the association weight between nodes, which is determined by the ratio of the number of historical interactions between nodes to the total number of interactions. is the path interaction intensity, which is calculated by the cumulative number of interactions between nodes. is the path interaction frequency, which is determined by the time distribution ratio of the path interaction. is the weight adjustment parameter, which adjusts the path interaction strength The influence weight of is the frequency adjustment parameter, which controls the weight of the low-frequency path in the calculation;

[0124] The node association weight of a path , path interaction intensity , path interaction frequency , adjustment parameters , adjustment parameters ;

[0125] Calculate weight and strength:

[0126] ;

[0127] ;

[0128] Calculate the frequency part:

[0129] ;

[0130] ;

[0131] Comprehensive calculation: ;

[0132] This result shows that the path strength weight ,This weight value shows that the path is highly correlated, indicating that it occupies an important position in the multidimensional path structure.

[0133] Call the path strength weight in the optimized path association data set to match the hierarchical path relationship, integrate the association weight and path strength value into the hierarchical path structure, and obtain multi-dimensional path data;

[0134] Firstly, the path weight value is extracted from each row of the path strength matrix, and the relationship between the value and the elimination threshold is compared to eliminate the paths below the threshold. At the same time, the remaining paths are reorganized according to the hierarchical network structure. After the path strength weight value is integrated into the hierarchical path structure, a hierarchical path matching relationship table is generated. The matching table is based on the path association weight and the path strength weight value, and records the specific hierarchical structure of each path. Based on the hierarchical path matching relationship table, the node data of the path is reclassified, and the node interaction intensity, interaction frequency and other characteristics are included in the hierarchical path structure mapping together with the path weight value. The association density between paths is calculated through the distribution statistics of the path strength weight, and the data fields in the path hierarchical structure mapping results are integrated. Finally, multidimensional path data is obtained through multi-dimensional integration of node and path feature weights.

[0135] See also Figure 6 ,The specific steps for obtaining the diabetes interaction path model are:

[0136] Based on multi-dimensional path data, the location information and structure of the interaction nodes between paths are extracted, the geometric characteristics and connection properties of the nodes are analyzed, the direct relationships between nodes are classified and processed, the interaction nodes are numbered, the structural relationships between nodes are summarized, and the basic data of the path interaction nodes are obtained;

[0137] Multidimensional path data was loaded from the cloud platform database of diabetes data. The geometric position of each node on the path and the path structure of its connection were extracted by parsing the interactive node information in the path. The geometric properties included the coordinate information of the node and its relative position on the path, while the connection attributes reflected the association between nodes in different paths, such as whether the same node belonged to the intersection of different paths at the same time. The direct relationship between nodes was classified. The classification criteria included whether the node was a continuous node of a single path, whether it was an intersection or branch point of the path, etc. The interactive nodes were numbered with a unique identifier. The identifier format was generated by combining the node position, path classification and geometric attributes to ensure that each node number was unique and traceable. The connection relationship of all nodes in the path was further summarized and sorted. By analyzing the node interaction mode between paths, the direct and indirect structural relationships between nodes were recorded, and finally the basic data of path interaction nodes were obtained to provide clear node information for subsequent data analysis.

[0138] Based on the basic data of path interaction nodes, the number of repeated occurrences of nodes in the path is identified to obtain the node connection frequency, the direction data of the connection relationship between paths is counted, the node association strength is analyzed, the node group is screened, and the node association degree distribution relationship is generated by combining the path intersection relationship data;

[0139] By analyzing the basic data of path interaction nodes, the number of times each interaction node appears repeatedly in all paths is calculated to obtain the node connection frequency. The connection frequency reflects the importance of a node in the path. For example, high-frequency nodes are located at the intersection of paths or key branch points. By counting the connection direction data between paths, the connection direction characteristics between nodes in different paths are analyzed. For example, it is recorded whether the connection direction of the path nodes from the starting point to the end point is consistent or has an opposite relationship. Combined with the connection frequency and direction data, the association strength between nodes is quantified. The association strength is obtained by comprehensive calculation of the number of shared connection relationships between nodes, direction consistency and node connection frequency. Node groups with high association are screened out, and isolated or weakly associated nodes are eliminated. Combined with the path intersection relationship data, the node groups are further analyzed for association to generate the node association distribution relationship. The association distribution relationship table clearly shows the association strength of each node and its importance in the path network, laying the foundation for the dynamic analysis of the interaction path.

[0140] Based on the node association degree distribution relationship, the association path connection characteristics in the node group are analyzed, the node weight distribution is analyzed according to the path direction data, and the dynamic characteristics between the paths are combined to integrate the dynamic characteristics between the nodes and the paths to generate the diabetes interaction path model;

[0141] According to the node association distribution relationship, the connection path information of the high-association node group is extracted, and the connection characteristics of the path in the interactive network are analyzed, including the length of the path, the node distribution density and the direct and indirect connection patterns between the interactive nodes. By parsing the path direction data, the weight distribution of each node is calculated. The size of the weight reflects the importance of the node in the path interaction. For example, the starting and end nodes of the path have higher weight values, while the weight of the intermediate node is comprehensively calculated based on its connection frequency and association strength. In the analysis of dynamic characteristics between paths, the change patterns between paths in different time periods are recorded, such as the increase and decrease of node groups and the adjustment of connection directions, and the dynamic characteristics between nodes and paths are integrated. Finally, a diabetes interaction path model is constructed based on the node weight distribution and the dynamic characteristics of the path. The model comprehensively reflects the path network relationship and its dynamic change law of diabetic patient data in the cloud platform, and provides a panoramic path structure diagram for the sharing and multidimensional analysis of diabetes data.

[0142] See also Figure 7 , the steps for obtaining the keyword associated path set are as follows:

[0143] Call the path node data in the diabetes interaction path model, analyze the query content keywords and extract key fields, analyze the semantic relevance of keywords in the path nodes, match the semantic weights of keywords and path nodes, extract the node set with a matching degree higher than the threshold, and establish the initial matching relationship between keywords and nodes;

[0144] The semantic labels, interaction records and path level information of each node in the model are extracted, the keywords in the query content are decomposed into semantic units and mapped to the semantic label set of the model nodes, and the relevance of the keywords in the path nodes is calculated through semantic matching. In the matching degree calculation process, the relevance score is quantified according to the semantic distance between the keyword and the node semantic label (calculated through the predefined semantic vector space), and the nodes with semantic distance below the preset threshold are screened out. The matching weight of the node is adjusted in combination with the frequency data of the node interaction record. For example, in the node with high semantic relevance but low interaction record frequency, the time weighting coefficient is used to correct its matching weight. Through this process, the initial node matching weight set corresponding to each keyword is obtained, and the weight value is used as the node importance indicator for sorting, and finally the initial matching relationship between the keyword and the node is obtained.

[0145] Based on the initial matching relationship between keywords and nodes, combined with the path interaction strength and node hierarchical relationship, the matching weight between keywords and paths is calculated using the formula:

[0146] ;

[0147] Generate a dataset of keyword and path interaction relationships;

[0148] in, Indicates the matching weight between keywords and paths. To query content weight, is the semantic similarity between keywords and path nodes, is the association weight at the path level, is the path interaction weight;

[0149] The benefit of the formula is that by querying the weight , semantic similarity , path level weight and interaction weights The combined effect of dynamic adjustment of the matching weight between keywords and paths enhances the model's ability to parse the semantic relationship between keywords and paths, and takes into account the impact of path hierarchy and interaction weight on matching results;

[0150] Indicates the matching weight between keywords and paths, and measures the comprehensive correlation between keywords and paths. The query content weight is determined by calculating the frequency and semantic strength of the keywords in the query content. is the semantic similarity between keywords and path nodes, calculated by the semantic vector model. is the association weight of the path level, which is determined by the depth or relative position of the path level. is the path interaction weight, which is obtained by calculating the interaction frequency of path nodes;

[0151] The query content weight of a keyword , semantic similarity , path level weight , path interaction weight ;

[0152] Calculate the query content weight and semantic similarity: ;

[0153] Calculate the path level weight and interaction weight part: ;

[0154] Comprehensive calculation: ;

[0155] This result shows that the matching weight of keywords and paths , which shows that the keywords have high semantic relevance and interactive matching weight in this path, indicating that its path matching results have high relevance.

[0156] Call the keyword and path interaction relationship data set, integrate the keyword matching weight value with the path interaction strength and hierarchical relationship, filter the paths with matching weight values ​​higher than the set threshold, group and classify the path association levels, and generate a keyword association path set;

[0157] The matching weight of each keyword and path is extracted, and the paths with matching weight values ​​higher than the set threshold are screened out by combining the path interaction strength and path hierarchy data. The paths are grouped and classified according to the association hierarchy, and the node interaction data is further extracted from the grouped path set. The semantic consistency between the semantic label of the node and the hierarchical path is analyzed, and the path association weight of the node is adjusted according to the consistency data. The association information of the keyword and the path is integrated to finally form a keyword-associated path set, which contains the complete matching results of the keyword and all relevant nodes and path information in the model path.

[0158] See also Figure 8 ,The specific steps for obtaining the results of the diabetes data semantic analysis path are:

[0159] Based on the keyword-related path set, the interactive node information in the path is extracted, the node distribution and connection characteristics are analyzed, the geometric characteristics and relationship weights are analyzed, the path structure data is sorted, and the data sharing content is classified according to the logical relationship between the nodes to obtain the keyword path data sharing content;

[0160] The interactive node information is extracted from the diabetes data path set in the cloud platform. First, the nodes contained in each path are scanned and recorded, and the geometric characteristics of the nodes are extracted, including the relative position of the nodes, the position order in the path, and the distance between the nodes. When parsing the node distribution, the shared node characteristics between the paths are combined to analyze whether the nodes have a cross-path distribution pattern, for example, whether a node is an intersection point or a key node of multiple paths. For each node, the relationship weight with the adjacent nodes is calculated. The relationship weight is obtained by the frequency of node connection and the connection stability. For example, the weight of a node is given a higher value due to its high-frequency connection characteristics. When sorting the path structure data, the geometric characteristics of the nodes, the relationship weight and the path identifier are combined to generate structured data records, and the path data is classified according to the logical relationship between the nodes. For example, the shared content is sorted and classified according to the time dimension or diabetes feature category, and finally the keyword path data sharing content is formed, which lays a data foundation for the subsequent high-frequency node screening.

[0161] Based on the shared content of keyword path data, the association strength is analyzed according to the frequency of occurrence of nodes in the path, the path node connection data is counted, the path nodes that meet the conditions are located, the high-frequency nodes are screened, and the nodes are classified and summarized according to the interaction frequency and position relationship of the nodes to generate a high-frequency node matching set;

[0162] The frequency of node occurrence in each path data is counted. The statistical process not only considers the frequency of nodes in a single path, but also the number of repetitions in multiple paths. The calculation of association strength is based on the total frequency of node occurrence and the direct connection relationship with the node. For example, the association strength of a node is significantly increased due to its high-frequency connection and cross-path distribution characteristics. When counting the path node connection data, the direct and indirect connection relationship between each node and the node is recorded, including the connection direction and the number of connections. When locating path nodes that meet the conditions, the nodes that meet the specific frequency threshold are screened, and the nodes are combined with the geometric characteristics and connection weights to confirm whether they are key nodes. For the high-frequency nodes screened out, they are classified in combination with their interaction frequency and position relationship, and nodes with similar interaction patterns are grouped together to form a high-frequency node matching set. The matching set contains the interaction attributes of each node and its functional positioning in the path network, which provides basic data support for analyzing the association between high-frequency nodes.

[0163] Based on the high-frequency node matching set, analyze the association between high-frequency nodes, extract the semantic features of path nodes, integrate the node connection rules, analyze the semantic association features of path data, and obtain the semantic analysis path results of diabetes data;

[0164] The connection relationship between nodes is extracted through the high-frequency node matching set, and the association characteristics of nodes in the path network are analyzed, including the connection strength, connection direction and multi-node interaction mode of the nodes. In order to extract the semantic characteristics of the path nodes, the attribute information of the high-frequency nodes is associated with the path label data. For example, semantic labels are assigned to nodes based on the patient's age level, living habits and other dimensions. By integrating the node connection rules, the logical relationship in the path is analyzed, and the priority and strength relationship of the connection between high-frequency nodes are clarified. For example, the connection rules of nodes with high association intensity are calculated preferentially in the path network, and their position weights in the path network are adjusted. When further analyzing the semantic association characteristics of the path data, the number of shared semantics between nodes and the semantic similarity of the associated paths are calculated to finally form the results of the semantic analysis path of diabetes data. The results show the semantic association pattern between nodes in the path network, and provide a visualization and quantitative basis for the sharing and semantic hierarchical analysis of diabetes data on the cloud platform.

[0165] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A diabetes data sharing and analysis system based on a cloud platform, characterized in that: The system comprises: The diabetes data collaborative processing module extracts the patient's blood sugar data, diet records and activity behavior data based on the diabetic patient data of the cloud platform medical institution, analyzes the data batches and multi-dimensional features, filters out noise items and duplicate data, removes interference factors and merges the data to generate encrypted data sets of patient data; The patient feature hierarchical management module identifies the patient's diabetes type, drug use record, and blood sugar fluctuation trend based on the patient data encrypted data group, performs feature value matching and classification marking, generates patient feature classification results, calls hierarchical information to prioritize shared feature nodes, and generates a patient hierarchical feature network structure; The diabetes association path analysis module analyzes the association strength, interaction records and hierarchical relationship paths between nodes based on the patient hierarchical characteristic network structure, removes low-frequency nodes and weakly associated paths, obtains multidimensional path data, performs extended analysis on the interaction nodes between paths, and generates a diabetes interaction path model; The steps for obtaining the multi-dimensional path data are specifically as follows: Calling the associated records of each node in the patient hierarchical feature network structure, extracting the interaction strength, interaction times and hierarchical path relationship between nodes, merging the interaction strength cumulative values, calculating the path cumulative interaction weight according to the interaction times, and establishing an initial path relationship diagram between nodes; For the initial path relationship graph between the nodes, identify the strength value and association weight of each path, remove low-frequency paths and weakly associated paths based on the path frequency, and use the formula: ； Generate optimized path association data set; in, is the path strength weight, is the node association weight, is the path interaction strength, is the path interaction frequency, is the path weight adjustment parameter, Adjustment parameters for path frequency; Calling the path strength weight in the optimized path association data set, matching the hierarchical path relationship, integrating the association weight and the path strength value into the hierarchical path structure, and obtaining multi-dimensional path data; Based on the diabetes interaction path model, the intelligent semantic reasoning module analyzes the relationship between query content keywords and model nodes, calls path interaction nodes and path hierarchy information for keyword matching, obtains a set of keyword-associated paths, extracts collaborative data content in combination with data sharing, screens high-frequency nodes that match the query content, and generates diabetes data semantic analysis path results.

2. The cloud platform-based diabetes data sharing and analysis system according to claim 1, characterized in that: The steps for obtaining the patient data encrypted data group are specifically as follows: Based on the data of diabetic patients in medical institutions on the cloud platform, the blood sugar data, dietary records and activity behavior data of patients are extracted, and the blood sugar data, dietary records and activity behavior data are preliminarily compared through the data source identification and record timestamp. The three types of data are multi-dimensionally indexed and matched according to the patient ID to generate a multi-dimensional data mapping relationship; Based on the multi-dimensional data mapping relationship, noise items and duplicate data are screened and removed, and deviation factors are judged and data weights are adjusted by identifying the deviation behaviors of the three types of data, using the formula: ； generating an adjusted feature data relationship; in, represents the deviation factor adjustment value, is the feature data weight, is the real-time value of the data, is the median of the standard value of the characteristic data, is the total data volume, is the data sequence index; The adjusted characteristic data relationship is called, and noise is eliminated and multi-source data is merged for the three types of data, namely blood sugar, diet and activity behavior, according to matching features, and encryption is performed through the patient ID to generate a patient data encrypted data group.

3. The cloud platform-based diabetes data sharing and analysis system according to claim 2, characterized in that: The steps for obtaining the patient characteristic classification results are specifically as follows: Calling the blood sugar record, diet record and medication use record fields in the patient data encrypted data group, analyzing the patient's blood sugar fluctuation range, dietary component intake and medication type changes, performing merging processing on each field data point according to the patient ID, and generating an initial index set of patient characteristics; For the blood sugar fluctuation amplitude field in the patient characteristic initial index set, combined with the patient's drug use record and dietary intake, determine whether it meets the diabetes type characteristics, input the characteristic value into the patient classification threshold, and use the formula: ； The diabetes type matching value is calculated; in, Matching value for diabetes type, is the blood sugar fluctuation weight, is the blood sugar fluctuation data value, is the median value of blood sugar fluctuation, is the blood glucose matching parameter adjustment coefficient, is the dietary intake weight, is the characteristic value of dietary component intake, Adjust parameters for diet; The diabetes type matching value is called, the patient's characteristic value is matched to the standard classification template, the patient's diabetes type, drug use record and blood sugar fluctuation trend are classified and marked according to the template classification label, and the patient characteristic classification result is generated.

4. The cloud platform-based diabetes data sharing and analysis system according to claim 3, characterized in that: The steps for obtaining the patient stratified feature network structure are specifically as follows: Based on the patient feature classification results, feature nodes are extracted, node-related attributes are classified according to classification labels, nodes are preliminarily grouped according to weight parameters and correlation indicators, unique identifiers are assigned to each group of nodes, and a node grouping attribute set is obtained; Based on the node grouping attribute set, the attribute values ​​of the shared feature nodes are extracted, and the node attribute values ​​are matched for relevance in combination with the hierarchical information. The priorities of the shared feature nodes are analyzed using weight parameters and node distribution frequencies, and the priorities are arranged to obtain a hierarchical shared feature priority value table; Based on the hierarchical shared feature priority value table, the associations between feature nodes are identified in order of priority, and the priority and group associations between nodes are used as connection parameters to adjust the connection strength in the association to obtain the patient hierarchical feature network structure.

5. The cloud platform-based diabetes data sharing and analysis system according to claim 1, characterized in that: The steps for obtaining the diabetes interaction path model are specifically as follows: Based on the multi-dimensional path data, the location information and structure of the interaction nodes between the paths are extracted, the geometric characteristics and connection properties of the nodes are analyzed, the direct relationships between the nodes are classified and processed, the interaction nodes are numbered, the structural relationships between the nodes are summarized, and the basic data of the path interaction nodes are obtained; Based on the basic data of the path interaction nodes, the number of repeated occurrences of the nodes in the path is identified to obtain the node connection frequency, the direction data of the connection relationship between the paths is counted, the node association strength is analyzed, the node group is screened, and the node association degree distribution relationship is generated by combining the path intersection relationship data; Based on the node association degree distribution relationship, the association path connection characteristics in the node group are analyzed, the node weight distribution is parsed according to the path direction data, and the dynamic characteristics between the paths are combined to integrate the dynamic characteristics between the nodes and the paths to generate a diabetes interaction path model.

6. The cloud platform-based diabetes data sharing and analysis system according to claim 5, characterized in that: The steps for obtaining the keyword associated path set are specifically as follows: Calling the path node data in the diabetes interaction path model, analyzing the query content keywords and extracting key fields, analyzing the semantic relevance of the keywords in the path nodes, matching the semantic weights of the keywords and the path nodes, extracting the node set with a matching degree higher than a threshold, and establishing an initial matching relationship between the keywords and the nodes; Based on the initial matching relationship between the keyword and the node, combined with the path interaction strength and the node hierarchical relationship, the matching weight between the keyword and the path is calculated using the formula: ； Generate a dataset of keyword and path interaction relationships; in, Indicates the matching weight between keywords and paths. To query content weight, is the semantic similarity between keywords and path nodes, is the association weight at the path level, is the path interaction weight; The keyword-path interaction relationship dataset is called, the keyword matching weight value is integrated with the path interaction strength and the hierarchical relationship, the paths with matching weight values ​​higher than the set threshold are screened, and the associated levels of the paths are grouped and classified to generate a keyword associated path set.

7. The cloud platform-based diabetes data sharing and analysis system according to claim 6, characterized in that: The steps for obtaining the diabetes data semantic analysis path results are specifically as follows: Based on the keyword-associated path set, the interactive node information in the path is extracted, the node distribution and connection characteristics are analyzed, the geometric characteristics and relationship weights are analyzed, the path structure data is sorted, and the data sharing content is classified in combination with the logical relationship between the nodes to obtain the keyword path data sharing content; Based on the keyword path data sharing content, analyze the association strength according to the frequency of occurrence of nodes in the path, count the path node connection data, locate the path nodes that meet the conditions, screen the high-frequency nodes, classify and summarize them according to the node interaction frequency and position relationship, and generate a high-frequency node matching set; Based on the high-frequency node matching set, the association between high-frequency nodes is analyzed, the semantic features of path nodes are extracted, the node connection rules are integrated, the semantic association features of path data are analyzed, and the semantic analysis path results of diabetes data are obtained.

Images