Chronic patient follow-up visit sharing and collaborative management system based on Java design

Abstract

The invention relates to the technical field of medical informatization management, in particular to a Java design-based chronic patient follow-up visit sharing and collaborative management system, which comprises a rule base construction module, a node generation module, an association mining module, a network construction module and an information sharing module. According to the method, the structural state nodes for mapping clinical business semantics are constructed, the transverse association topology among multi-dimensional symptoms under the same time sequence is mined by using the co-occurrence support degree algorithm, and the longitudinal treatment failure link is verified and identified in combination with the difference between the evolution trend vector and the intervention behavior; a hierarchical evolution network integrating spatial concurrence characteristics and time evolution tracks is synthesized, a composite abnormal path with long-range propagation characteristics is automatically locked, a collaborative risk score is calculated, key disease course information is converted into a standardized transmission protocol format for real-time sharing, and conversion from single numerical monitoring to complex pathology dynamic tracking is realized. And the combined early warning capability and the cooperative treatment efficiency for chronic disease complications are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of medical information management technology, and in particular to a chronic patient follow-up sharing and collaborative management system based on Java. Background Technology

[0002] The field of medical information management technology aims to achieve the recording, storage, transmission and sharing of medical data through digital means, including the digitization of medical records, the collection of information on the diagnosis and treatment process, the exchange of health data across institutions, the retention of long-term health management information for patients, and the digital expression of doctor-patient interaction records. It also supports medical service processes through information systems, enabling medical services to form a systematic management model in terms of continuous recording, data collection methods, collaborative communication paths and follow-up processing mechanisms.

[0003] The Chronic Patient Follow-up Sharing and Collaborative Management System refers to an information processing approach built around long-term health monitoring of chronic disease patients (covering oral and psychological monitoring, etc.), health data sharing, and collaborative management. This system connects to the regional health information platform to network all electronic records and synchronously updates patients' daily medical information to the electronic records. The system presents different dimensions of information based on different access permissions. For example, administrators can see all information in the city; community hospitals can see service items within the basic public health service standards, including basic information, daily medical information for chronic patients, medication information, follow-up and intervention records; dental clinics can see dental treatment information and the patient's basic information; mental health rehabilitation hospitals can see information and treatment records of managed individuals, including "patients with severe mental disorders"; and outpatient and inpatient departments of general hospitals can see information on all personnel within the region (with the possibility of adding outpatient and inpatient treatment information) and synchronize it with community hospital information in real time. Furthermore, the system supports everyone in accessing health knowledge and health education materials. Doctors' workstations can issue health education prescriptions and vaccination prescriptions for different patients and establish interaction between community hospitals and higher-level hospitals, enabling referrals from lower to higher levels and recommendations from higher to lower levels.

[0004] Current follow-up management often adopts a local independent database and fixed form recording mode, which breaks continuous vital sign data into isolated discrete values. It lacks a structured representation of the endogenous logical relationship between pathological parameters, only supports static numerical comparison, and is difficult to capture the dynamic evolution and concurrent characteristics of the disease over time. The data flow method based on file export or offline handover results in time lag in information synchronization, and it is impossible to build a closed-loop feedback mechanism that reflects the effect of medical intervention and the status of disease response. As a result, risk assessment for complex disease courses lacks multidimensional quantitative basis, which seriously restricts the response speed and intervention accuracy of cross-institutional collaborative diagnosis and treatment. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a Java-based chronic patient follow-up sharing and collaborative management system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a chronic patient follow-up sharing and collaborative management system based on Java, the system comprising: The rule base construction module obtains the vital sign data of chronic patients of the objects to be monitored, defines Java vital sign object entity classes for vital sign data, extracts business status interval metadata from them, and builds a status mapping rule base. The node generation module compares each vital sign data item in the state mapping rule base with the business state interval metadata, encapsulates it into a structured disease state node, and constructs a set of structured disease state nodes. The association mining module, based on the set of structured disease status nodes, calculates the co-occurrence support between structured disease status nodes under the same collection timestamp, filters concurrent association pairs of symptoms, and establishes a horizontal association topology graph of disease status. The network construction module, based on the structured disease state node set, analyzes the evolution trend vector of the structured disease state nodes under continuous collection timestamps, compares it with medical intervention behavior records, and accumulates the horizontal correlation topology graph of the disease to construct a multidimensional disease hierarchical evolution network. The information sharing module traverses the multidimensional disease progression network, calculates the collaborative follow-up risk score for specified structured disease state nodes, and generates collaborative follow-up sharing information for chronic diseases.

[0007] The present invention is improved in that the state mapping rule base includes business state interval metadata defined by Java vital sign object entity classes and clinical business state codes corresponding to the business state interval metadata; the structured disease state node set includes structured disease state nodes encapsulating collection timestamps, physiological parameter type identifiers, physiological parameter values ​​and clinical business state codes; the disease horizontal association topology graph includes concurrent symptom association pairs whose co-occurrence support meets a preset confidence threshold and first-level concurrent association links connecting the concurrent symptom association pairs; the multidimensional disease hierarchical evolution network includes second-level intervention failure links generated by comparing evolution trend vectors with medical intervention behavior records and the first-level concurrent association links; and the chronic disease collaborative follow-up shared information specifically consists of a collaborative follow-up risk score calculated by accumulating preset risk weights of structured disease state nodes in long-chain composite abnormal paths and the long-chain composite abnormal paths used as the scoring basis.

[0008] The present invention is improved in that the rule base construction module includes: The entity object definition submodule obtains the systolic blood pressure, diastolic blood pressure and fasting blood glucose values ​​of the object to be monitored through the IoT medical device interface, constructs a class attribute mapping structure to match the vital signs data of chronic patients, marks the corresponding attribute fields with annotation tags defined for the boundaries of systolic blood pressure, diastolic blood pressure and blood glucose values, and encapsulates the marked attribute fields to generate Java vital signs object entity classes. The interval metadata parsing submodule calls the Java reflection mechanism to parse the Java vital signs object entity class, reads the annotation tag content marked on the attribute field, extracts the systolic blood pressure value boundary, diastolic blood pressure value boundary and fasting blood glucose value boundary from the annotation tag content, classifies each type of boundary information according to physiological parameter type, determines the value range of physiological parameters that exceed the physiological safety threshold and the value range that does not exceed the physiological safety threshold, and obtains the business status interval metadata. The status encoding mapping submodule iterates through the boundary intervals of each value in the business status interval metadata, assigns a unique character sequence as a clinical business status identifier to each independent boundary interval of systolic blood pressure, diastolic blood pressure, and fasting blood glucose, establishes a key-value index relationship between the boundary intervals and the character sequences, and constructs a status mapping rule base.

[0009] The present invention is improved in that the step of calling the Java reflection mechanism to parse the Java vital signs object entity class includes: Obtain the class type description object of the Java vital signs object entity class in the runtime environment, use the class type description object to call the declaration field retrieval interface, obtain all private and public member variables declared in the Java vital signs object entity class, and encapsulate the obtained variables into a set of attribute fields to be parsed; Iterate through the set of attribute fields to be parsed, and for each attribute field in the set, call the access control overriding method provided by the reflection API to set the access permission flag of the attribute field to be parsed to the enabled state, thereby removing the language-level encapsulation restrictions on the attribute field to be parsed. Check if the modifier list of the attribute field to be parsed contains a type definition reference to the annotation tag; If the detection result confirms the presence of the annotation tag, the dynamic proxy instance object corresponding to the annotation tag is obtained through the reflection context, and the boundary value string stored inside the annotation tag is read through the dynamic proxy instance object; Type conversion is performed on the boundary numerical strings to convert the text-formatted data into double-precision floating-point numbers, thus obtaining the boundaries of systolic blood pressure, diastolic blood pressure, and fasting blood glucose values.

[0010] The present invention is improved in that the node generation module includes: The temporal feature extraction submodule extracts the systolic blood pressure, diastolic blood pressure, fasting blood glucose, and the collection timestamp corresponding to each value from the chronic patient's vital signs data. It identifies the physiological parameter type identifier to which each value belongs and arranges each value and the collection timestamp in chronological order to form a temporal value sequence of vital signs to be processed. The state matching and conversion submodule compares each value in the time-series numerical sequence of vital signs with the boundary interval of each value in the business state interval metadata, determines the range of the boundary interval in which each value falls, indexes the corresponding character sequence in the state mapping rule base according to the range of the boundary interval in which it falls, assigns an identifier representing clinical meaning to each vital sign value, and generates a clinical business state code. The status node encapsulation submodule encapsulates the physiological parameter type identifier, clinical business status code, collection timestamp, and corresponding systolic blood pressure, diastolic blood pressure, or fasting blood glucose value into a structured set of disease status nodes.

[0011] The present invention is improved in that the method of assigning a clinically meaningful identifier to each vital sign value includes: Determine the range of values ​​in the business status interval metadata that each vital sign value currently to be processed in the vital sign time series numerical sequence falls within; Extract the minimum and maximum boundary values ​​of the numerical range, convert the minimum and maximum boundary values ​​into string format and concatenate them to generate the range positioning key for the retrieval state mapping rule base; Iterate through all key-value index relationships stored in the state mapping rule base and lock the target index entry that perfectly matches the numerical range; Access the value range space of the target index entry, read the character sequence stored in the value range space, and temporarily store the character sequence as a clinical status label to be assigned; Compare whether the applicable parameter types defined in the character sequence match the physiological parameter type identifiers corresponding to each vital sign value to be processed; If the matching result is consistent, the character sequence is confirmed as a unique identifier describing the physiological state of each vital sign value to be processed. Establish a unique identifier and a reference association between the memory address of each vital sign value to be processed, and write the unique identifier into the metadata description structure of each vital sign value to be processed.

[0012] The present invention is improved in that the association mining module includes: The node co-occurrence calculation submodule groups the set of structured disease state nodes according to the collection timestamp. For the grouped data under the same collection timestamp, it counts the frequency of the simultaneous occurrence of structured disease state nodes with multiple physiological parameter type identifiers, calculates the probability of the occurrence of a specified combination of structured disease state nodes in the overall dataset, and quantifies the co-occurrence support of structured disease state nodes. The concurrent association filtering submodule compares the co-occurrence support of the structured disease state nodes with a preset co-occurrence confidence threshold, retains the structured disease state node combinations whose co-occurrence support is greater than or equal to the co-occurrence confidence threshold, and establishes a first-level concurrent association link to generate symptom concurrent association pairs. The horizontal topology construction submodule extracts the first-level concurrent association links in the symptom concurrent association pairs, and connects them with structured disease status nodes as vertices and first-level concurrent association links as edges to reflect the graph structure of the mutual influence relationship between multiple symptoms at the same time, thus obtaining the disease horizontal association topology graph.

[0013] The present invention is improved in that the network construction module includes: The evolution trend calculation submodule calculates the slope of the numerical change of structured disease state nodes with the same physiological parameter type identifier in the structured disease state node set in chronological order within the continuous collection timestamps, and fits and generates an evolution trend vector describing the dynamic evolution of a single symptom over time. The intervention response verification submodule acquires medical intervention behavior records that are aligned with the evolution trend vector in time, calculates the numerical deviation between the rate of change of the evolution trend vector and the standard rate of change expected by the medical intervention in the medical intervention behavior records, filters abnormal periods with numerical deviations greater than a preset intervention failure deviation threshold, establishes a secondary intervention failure link connecting relevant nodes within the abnormal period, and generates a set of secondary intervention failure links. The hierarchical network synthesis submodule takes the secondary intervention failure links in the set of secondary intervention failure links as the vertical evolution dimension, adds them to the horizontal correlation topology of the disease, connects the structured disease state nodes of the same symptom at multiple time points, and constructs a multidimensional disease hierarchical evolution network.

[0014] The present invention is improved in that the information sharing module includes: The composite path retrieval submodule retrieves paths in the multidimensional disease progression network that are formed by connecting the first-level concurrent association links in the horizontal direction and the second-level intervention failure links in the vertical direction. It identifies continuous structured disease state node sequences that simultaneously have the characteristics of concurrent symptoms and intervention failure, locks the key abnormal propagation chains in the multidimensional disease progression network, and obtains long chain composite abnormal paths. The risk weight accumulation submodule identifies each structured disease state node in the long chain composite abnormal path, queries the preset risk weight value corresponding to each structured disease state node, accumulates the risk weights of all structured disease state nodes, and calculates the collaborative follow-up risk score. The shared information generation submodule filters target paths whose collaborative follow-up risk scores exceed a preset follow-up priority threshold, extracts chronic patient vital signs data and clinical business status identifiers from the long chain composite abnormal paths associated with the target paths, and converts them into a standardized transmission protocol format to generate chronic disease collaborative follow-up shared information.

[0015] The present invention improves upon this invention by locking the key abnormal propagation chains in the multidimensional disease progression network, including: A full-domain scan was performed on the multidimensional disease progression network to identify all connected components in the network topology that are interconnected by horizontal first-level concurrent links and vertical second-level intervention failure links. Extract the continuous structured disease state node sequence corresponding to each connected component, and count the total number of structured disease state nodes in the continuous structured disease state node sequence, which is defined as the path span value. Compare the path span value with the preset minimum threshold for the length of the anomaly propagation chain; If the comparison results show that the path span value is greater than or equal to the minimum threshold of the abnormal propagation chain length, then the continuous structured disease state node sequence is determined to be a sequence with long-range evolution characteristics, and the sequence with long-range evolution characteristics is locked and persistently marked in memory address.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, structured state nodes that map clinical business semantics are constructed, and the co-occurrence support algorithm is used to mine the horizontal correlation topology among multidimensional symptoms in the same time series. The longitudinal treatment failure links are identified by combining the evolution trend vector and the difference verification of intervention behavior. A hierarchical evolution network that integrates spatial concurrency features and temporal evolution trajectory is synthesized, automatically locking composite abnormal paths with long-range propagation characteristics and calculating collaborative risk scores. Key disease information is transformed into a standardized transmission protocol format for real-time sharing, realizing the transformation from single numerical monitoring to complex pathological dynamic tracking, effectively improving the joint early warning capability and collaborative governance efficiency for chronic disease complications. Attached Figure Description

[0017] Figure 1 This is a system module diagram of the present invention; Figure 2 This is a system framework diagram of the present invention; Figure 3 This is a flowchart illustrating the rule base construction module of the present invention; Figure 4 This is a flowchart illustrating the node generation module of the present invention; Figure 5 This is a flowchart illustrating the association mining module of the present invention; Figure 6 This is a flowchart illustrating the network construction module of the present invention; Figure 7 This is a flowchart illustrating the information sharing module of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] Please see Figure 1 This invention provides a technical solution: a chronic patient follow-up sharing and collaborative management system based on Java, the system comprising: The rule base construction module obtains the vital sign data of chronic patients of the objects to be monitored, defines Java vital sign object entity classes for vital sign data, extracts business status interval metadata from them, and builds a status mapping rule base. The node generation module compares each vital sign data in the status mapping rule base with the business status interval metadata, encapsulates it into a structured disease status node, and constructs a set of structured disease status nodes. The association mining module, based on a set of structured disease status nodes, calculates the co-occurrence support between structured disease status nodes under the same collection timestamp, filters concurrent association pairs of symptoms, and establishes a horizontal association topology graph of disease status. The network construction module, based on a set of structured disease status nodes, analyzes the evolution trend vector of the structured disease status nodes under continuous collection timestamps, compares it with medical intervention behavior records, and accumulates the horizontal correlation topology graph of the disease to construct a multidimensional disease hierarchical evolution network. The information sharing module traverses a multidimensional disease progression network, calculates the collaborative follow-up risk score for specified structured disease status nodes, and generates collaborative follow-up sharing information for chronic diseases.

[0020] The state mapping rule base includes business state interval metadata defined by Java vital sign object entity classes, and clinical business state codes corresponding to the business state interval metadata. The structured disease state node set includes structured disease state nodes encapsulating collection timestamps, physiological parameter type identifiers, physiological parameter values, and clinical business state codes. The disease horizontal association topology graph includes concurrent symptom association pairs whose co-occurrence support meets a preset confidence threshold, and first-level concurrent association links connecting concurrent symptom association pairs. The multidimensional disease hierarchical evolution network includes second-level intervention failure links and first-level concurrent association links generated by comparing evolution trend vectors with medical intervention behavior records. The chronic disease collaborative follow-up shared information specifically consists of a collaborative follow-up risk score calculated by accumulating preset risk weights of structured disease state nodes in long-chain composite abnormal paths, and long-chain composite abnormal paths used as the scoring basis.

[0021] Please see Figure 2 and Figure 3 The rule base building module includes: The entity object definition submodule obtains the systolic blood pressure, diastolic blood pressure and fasting blood glucose values ​​of the object to be monitored through the IoT medical device interface, constructs a class attribute mapping structure to match the vital signs data of chronic patients, marks the corresponding attribute fields with annotation tags defined for the boundaries of systolic blood pressure, diastolic blood pressure and blood glucose values, and encapsulates the marked attribute fields to generate Java vital signs object entity classes. The IoT data receiving process is initiated by establishing a low-level binary communication link with the IoT medical device and continuously monitoring incoming data packets. The parser reads the received hexadecimal data stream bit by bit, identifies the frame header and trailer identifiers, and extracts the systolic blood pressure, diastolic blood pressure, and fasting blood glucose values ​​from the payload. The parser performs type validation on the extracted raw values, confirming that the value format conforms to the double-precision floating-point standard and removing abnormal data caused by null values ​​or non-numeric characters. A class structure definition space is allocated in memory, and a corresponding attribute mapping table is created based on the business logic of chronic disease patient vital signs data. In this mapping table, systolic blood pressure, diastolic blood pressure, and blood glucose values ​​are designated as independent key-value pair fields. A pre-built annotation definer is loaded, which dynamically generates annotation tags on each attribute field based on boundary rules set according to medical statistics principles. These boundary rules are based on statistical analysis of historical chronic disease confirmed case data, calculating the numerical distribution characteristics of systolic blood pressure, diastolic blood pressure, and blood glucose at different pathological stages, and selecting numerical points outside the mean plus or minus two standard deviations of the normal distribution curve as critical thresholds. For example, if the mean systolic blood pressure of stage I hypertension patients in the statistical sample is 140 mmHg and the standard deviation is 10, then the upper limit critical threshold is calculated as follows: The calculated critical thresholds are written into the parameter list of the annotation tags, thus enhancing the semantics of the attribute fields. The bytecode generator then intervenes, reading the attribute mapping table and the annotation tags containing the threshold information. Using dynamic bytecode manipulation technology, it assembles and generates the bytecode file of the Java class in the runtime environment. This bytecode file contains private systolic blood pressure, diastolic blood pressure, and blood glucose attribute variables, as well as corresponding public access methods, and solidifies the binding relationship between fields and annotation tags at the bytecode level. The class loader loads the generated bytecode into the Java Virtual Machine memory, completing the instantiation preparation of the vital signs object entity class.

[0022] The interval metadata parsing submodule calls the Java reflection mechanism to parse the Java vital signs object entity class, read the annotation tag content marked on the attribute field, extract the systolic blood pressure value boundary, diastolic blood pressure value boundary and fasting blood glucose value boundary from the annotation tag content, classify each type of boundary information according to physiological parameter type, determine the value range of physiological parameters that exceed the physiological safety threshold and the value range that does not exceed the physiological safety threshold, and obtain the business status interval metadata. The process of resolving Java physical characteristic object entity classes by invoking Java reflection includes: Obtain the class type description object of the Java vital signs object entity class in the runtime environment, use the class type description object to call the declaration field retrieval interface, obtain all private and public member variables declared in the Java vital signs object entity class, and encapsulate the obtained variables into a set of attribute fields to be parsed; Iterate through the set of attribute fields to be parsed, and for each attribute field in the set, call the access control overriding method provided by the reflection API to set the access permission flag of the attribute field to be parsed to the enabled state, thereby removing the language-level encapsulation restrictions on the attribute field to be parsed. Check if the modifier list of the attribute field to be parsed contains a type definition reference to the annotation tag; If the detection result confirms the presence of the annotation tag, the dynamic proxy instance object corresponding to the annotation tag is obtained through the reflection context, and the boundary value string stored inside the annotation tag is read through the dynamic proxy instance object; Type conversion is performed on the boundary numerical strings to convert the text format data into double-precision floating-point numbers, thus obtaining the boundaries of systolic blood pressure, diastolic blood pressure, and fasting blood glucose values. The Java Reflection API is used to obtain a class object reference of the Java vital signs entity class generated by the entity object definition submodule. The field retrieval methods of the class object are called to retrieve all member variables declared in the class definition, including private variables, and these variable objects are stored in a list of fields to be processed. This field list is traversed, and for each field object, an access control override operation is performed to set the field's accessibility flag to the enabled state. The annotation query method of the field object is called to check if the current field contains a boundary definition annotation of the specified type. When an annotation is detected, a proxy instance of the annotation is obtained through the reflection context. The boundary value string stored inside the annotation is extracted by calling the attribute read method of the proxy instance. This string contains the critical threshold determined statistically in the previous steps. A string parser is initialized to split the extracted boundary value string according to a specific delimiter, and a double-precision floating-point conversion function is called to convert the text-formatted numerical value into a computer-calculateable floating-point value, thereby obtaining the precise upper and lower limit boundary values ​​of systolic blood pressure, diastolic blood pressure, and blood glucose. Based on these boundary values, the numerical space is divided into three possible intervals: the first is the range below the lower limit, representing low-risk values; the second is the range between the lower and upper limits, representing a safe range; and the third is the range above the upper limit, representing high-risk values. For example, if the analysis yields a lower limit of systolic blood pressure of 90 and an upper limit of 140, then the interval is constructed as follows: , and These interval information and their corresponding security status identifiers are encapsulated into interval metadata objects and stored in a hash mapping structure indexed by physiological parameter type.

[0023] The status coding mapping submodule iterates through the boundary intervals of each value in the business status interval metadata, assigns a unique character sequence as a clinical business status identifier to each independent boundary interval of systolic blood pressure, diastolic blood pressure, and fasting blood glucose, establishes a key-value index relationship between the boundary intervals and the character sequences, and constructs a status mapping rule base. The system receives the interval metadata object generated by the interval metadata parsing submodule and initiates a traversal program for the business state interval metadata. It sequentially reads the list of boundary intervals for systolic blood pressure, diastolic blood pressure, and blood glucose stored in the hash mapping structure. For each independent numerical interval, a sequence generator is invoked, which assigns a unique character sequence according to preset encoding rules. The encoding rules are based on the standardization of clinical diagnostic terminology, statistically analyzing the frequency of commonly used clinical diagnostic descriptive terms, and mapping frequently occurring pathological state descriptions to short alphanumeric combinations. A bidirectional index table is constructed in memory, using the boundary interval object as the key and the assigned character sequence as the value, storing it in a key-value pair set. A conflict detection algorithm is executed to ensure that no two different intervals are assigned the same character sequence. After completing the encoding assignment for all intervals, the mapping table containing all key-value index relationships is serialized and stored as a state mapping rule library. For example, for systolic blood pressure in the interval... In the case of internal mapping, the key-value pairs stored in the mapping rule base are " "; For blood sugar within the range In the case of internal storage, the stored key-value pairs are " This rule base establishes a mapping from numerical ranges to character encodings.

[0024] Please see Figure 2 and Figure 4 The node generation module includes: The temporal feature extraction submodule extracts the systolic blood pressure, diastolic blood pressure, fasting blood glucose, and the collection timestamp corresponding to each value from the vital signs data of chronic patients. It identifies the physiological parameter type identifier to which each value belongs, and arranges each value and the collection timestamp in chronological order to form a temporal value sequence of vital signs to be processed. Connect to the original chronic disease patient vital signs data storage area and batch read raw data records containing systolic blood pressure, diastolic blood pressure, and fasting blood glucose. Start the field separation program to identify the physiological parameter values ​​and corresponding data collection timestamps for each record. Based on the metadata tags in the data records, identify the physiological parameter type to which the current value belongs. Initialize a time series container, assembling the extracted values, collection timestamps, and physiological parameter type identifiers into a time series data object. Place all assembled time series data objects into a sorting list and call the sorting algorithm, using the collection timestamp as the sort key, to sort the objects in the list in ascending order. During the sorting process, perform time alignment operations and check the time intervals between adjacent data points. If data is found to be missing at certain time points, perform linear interpolation based on the rate of change of data at adjacent times. The principle of linear interpolation is based on the average rate of change of the patient's historical data. By calculating the time difference and numerical difference between two valid data points, the theoretical value of the missing time point is estimated. For example, if the value at time point 10 is 120 and the value at time point 20 is 130, and the value at time point 15 needs to be supplemented, the calculation process is as follows: After alignment and completion, a time series of vital signs values ​​is output, strictly arranged in chronological order.

[0025] The state matching and conversion submodule compares each value in the vital sign time series with the boundary interval of each value in the business state interval metadata, determines the range of the boundary interval in which each value falls, indexes the corresponding character sequence in the state mapping rule base according to the range of the boundary interval in which it falls, assigns an identifier representing the clinical meaning to each vital sign value, and generates a clinical business state code. Assigning clinically meaningful labels to each vital sign value includes: Determine the range of values ​​in the business status interval metadata that each vital sign value currently to be processed in the vital sign time series numerical sequence falls within; Extract the minimum and maximum boundary values ​​of the numerical range, convert the minimum and maximum boundary values ​​into string format and concatenate them to generate the range positioning key for the retrieval state mapping rule base; Iterate through all key-value index relationships stored in the state mapping rule base and lock the target index entry that perfectly matches the numerical range; Access the value range space of the target index entry, read the character sequence stored in the value range space, and temporarily store the character sequence as a clinical status label to be assigned; Compare whether the applicable parameter types defined in the character sequence match the physiological parameter type identifiers corresponding to each vital sign value to be processed; If the matching result is consistent, the character sequence is confirmed as a unique identifier describing the physiological state of each vital sign value to be processed. Establish a unique identifier and a reference association between the memory address of each vital sign value to be processed, and write the unique identifier into the metadata description structure of each vital sign value to be processed. For each vital sign value in the sequence, its physiological parameter type identifier is identified. Then, the business status interval metadata loaded in memory is accessed to retrieve a list of all boundary intervals corresponding to that parameter type. Boundary intervals in the list are extracted one by one, and a numerical inclusion judgment operation is performed, comparing the current vital sign value with the minimum and maximum boundary values ​​of the interval. If the current value is greater than or equal to the minimum boundary value and less than the maximum boundary value, it is determined that the value falls within the current interval. The minimum and maximum boundary values ​​of the interval are extracted, and a string concatenation function is called to convert these two values ​​into strings and combine them with a specific connector to generate a unique interval location key. Using this location key, a search is performed in the hash index of the status mapping rule base to locate the target index entry that perfectly matches the value interval. The value domain content stored under this entry, i.e., the previously allocated character sequence, is read. The applicable parameter type defined by this character sequence is compared to the parameter type identifier of the current vital sign value. After successful verification, the character sequence is confirmed as a unique identifier describing the physiological state of the current vital sign value, i.e., the clinical business status code. New storage space is allocated in memory, and a reference association is established between this unique identifier and the memory address of the current vital sign value object. For example, the current value is 145, which matches the range. If the corresponding character sequence is found to be CodeA according to the rule base, then CodeA will be used as the status identifier for that value.

[0026] The status node encapsulation submodule encapsulates physiological parameter type identifiers, clinical business status codes, collection timestamps, and corresponding systolic blood pressure, diastolic blood pressure, or fasting blood glucose values ​​into objects, forming a structured set of disease status nodes. Create a class instance of a structured patient condition status node. This instance has member variables for storing physiological parameter type identifiers, clinical status codes, acquisition timestamps, and specific vital sign values. Extract all relevant information for the current time point from the data stream processed by the state matching and transformation submodule. This includes the type code identifying systolic blood pressure, diastolic blood pressure, or blood glucose, the clinical status code obtained after matching and transformation, the original acquisition timestamp, and the corresponding floating-point value. Call the node instance's attribute setting method to assign these data items to the corresponding member variables. During encapsulation, perform a data consistency check to ensure that the correspondence between timestamps, values, and status codes is not incorrect. After populating the attributes of a single node, add the node instance to a dynamic list, which constitutes a set of structured patient condition status nodes. Construct a multi-level index structure based on timestamps and parameter types within the set. Assign a globally unique memory ID to each node, which will be used as a vertex identifier in subsequent graph structure construction.

[0027] Please see Figure 2 and Figure 5 The association mining module includes: The node co-occurrence calculation submodule groups the set of structured disease state nodes according to the collection timestamp. For the grouped data under the same collection timestamp, it counts the frequency of the simultaneous occurrence of structured disease state nodes with multiple physiological parameter type identifiers, calculates the probability of the occurrence of a specified combination of structured disease state nodes in the overall dataset, and quantifies the co-occurrence support of structured disease state nodes. Grouping is performed on the structured set of disease status nodes. All nodes in the set are traversed, and the data collection timestamp attribute of each node is read. Nodes with the same timestamp are grouped into the same time slice group. The occurrence of various clinical business status codes in each time slice group is identified, and a binary existence vector of the status codes on the time series is constructed. A weighted association strength algorithm based on cosine similarity is introduced to calculate the correlation strength between any two status codes. and Weighted correlation strength between The calculation process is performed according to the following formula: In the formula, This represents the final calculated weighted association strength, used to measure the co-occurrence tightness of two symptoms over time. This represents the total number of time slice groups, and its value comes from the statistical results of the count of independent time points collected in the dataset. Let `<loop_variable>` represent the loop term being processed. Indexes of time slice groups. Represents status code In the The existence of states in a time slice group, if the time group contains a status code. The node, then Assign a value of 1, otherwise assign a value of 0. Represents status code In the The existence state in each time slice group, if it exists. Assign a value of 1, otherwise assign a value of 0. This represents the clinical importance coefficient, which is based on the disease severity grading standard. It is a weighted score calculating the impact of different combinations of complications on patient prognosis; a higher value indicates a more significant negative impact of the combination on the condition. Here, the normalized average severity score is used as the coefficient. For example, the statistical impact of hypertension and hyperglycemia on prognosis is 0.8, and this coefficient is set after normalization and weighting adjustment. . This is a numerical smoothing factor used to prevent the denominator from being zero and to smooth low-frequency noise. Its setting is based on the minimum precision error of the floating-point arithmetic operations in the computing system, taking 10 times the upper limit of the error as a safety value, for example, set to 0.001. Assume there are a total of... The existence vector of status code CodeA is given by time slices. The existence vector of status code CodeB is The calculation process for the numerator is as follows: First, calculate the product of each term. The calculation process for the denominator is as follows: Calculate the sum and square root of the elements of vector A. Calculate the elements and square roots of vector B. Multiply the two together and add a smoothing factor, that is (Considering floating-point precision, in practice) Finally, the weighted association strength is calculated. This result quantifies the closeness of the two symptoms over time, serving as the co-occurrence support of structured disease state nodes.

[0028] The concurrent association filtering submodule compares the co-occurrence support of structured disease status nodes with the preset co-occurrence confidence threshold, retains the structured disease status node combinations whose co-occurrence support is greater than or equal to the co-occurrence confidence threshold, and establishes a first-level concurrent association link to generate symptom concurrent association pairs. The predefined co-occurrence confidence threshold is read, and this threshold is set based on frequent itemset mining analysis of a large-scale historical medical record. The co-occurrence frequency of symptoms in patients diagnosed with complications in the historical database is selected, and its average value is calculated and subtracted by a standard deviation as the baseline for determining a valid association. For example, if historical data shows that the average co-occurrence frequency of hypertension and diabetes is 0.15 and the standard deviation is 0.02, then the co-occurrence confidence threshold is calculated as follows: The process iterates through all candidate state node combinations, comparing the co-occurrence support of each combination with a threshold. The comparison result has two possibilities: first, if the co-occurrence support is greater than or equal to the co-occurrence confidence threshold, a significant concurrent association is determined between the combinations, and they are retained; second, if the co-occurrence support is less than the co-occurrence confidence threshold, it is considered accidental co-occurrence and is filtered out. For example, if the calculated co-occurrence support is 0.799, while the threshold is 0.13, since... If the condition is met, the combination is determined to constitute a valid association. For the retained combinations, a first-level concurrent association link is established, that is, an undirected edge is defined between the two state nodes, and the weight of the edge is set to the support value.

[0029] The horizontal topology construction submodule extracts the first-level concurrent association links in the concurrent association pairs of symptoms, and connects them with structured disease status nodes as vertices and first-level concurrent association links as edges to reflect the graph structure of the mutual influence relationship between multiple symptoms at the same time, thus obtaining the disease horizontal association topology graph. Initialize an empty graph data structure with structured disease state nodes as vertices. Traverse the set of concurrent symptom association pairs generated by the concurrent association filtering submodule. For each association pair, extract the two state nodes and the first-level concurrent association link connecting them. Check if these two nodes already exist in the graph structure; if not, create the corresponding vertex. Add an edge connecting these two vertices to the graph; this edge represents the first-level concurrent association link. Repeat this process until all filtered association pairs are mapped to the graph structure. Generate a subgraph for each time slice containing concurrent associations, or distinguish different lateral slices in the global graph using timestamps as an attribute. The final generated disease lateral association topology graph shows the entangled relationships of various pathological states in the patient at a specific time. This topology graph not only contains nodes and edges but also includes edge weight information, enabling subsequent analysis to quantify the strength of this lateral influence.

[0030] Please see Figure 2 and Figure 6 The network building blocks include: The evolution trend calculation submodule calculates the slope of the numerical change of structured disease state nodes with the same physiological parameter type identifier in the structured disease state node set in chronological order within the continuous collection timestamps, and fits and generates an evolution trend vector describing the dynamic evolution of a single symptom over time. From the structured set of disease status nodes, a subset of nodes with the same physiological parameter type identifier is selected. This subset is then strictly sorted by the data collection timestamp. A sliding time window is defined, and linear regression calculations are performed within the window. The timestamp value of each node within the window is extracted as the independent variable. Extract the corresponding vital sign values ​​as dependent variables. Using the least squares method, the slope of the fitted line for this set of data is calculated. The formula for calculating the slope is: calculate the difference between the time and the average time for each point, multiply it by the difference between the numerical value and the average value, sum these products, and then divide by the sum of the squares of the differences between the time and the average time for each point. For example, if three consecutive data points are collected... The average time is 2, and the average value is 102. The numerator is calculated as follows: The denominator is calculated as follows: The final slope is: The calculated slope value represents the rate of change of the symptom within the current time period, and is categorized into three possibilities: positive (upward trend), negative (downward trend), or zero (stationary). An evolution trend vector is constructed, which consists of these continuous slope values.

[0031] The intervention response verification submodule acquires medical intervention behavior records that are aligned with the evolution trend vector in time, calculates the numerical deviation between the rate of change of the evolution trend vector and the standard rate of change of the expected medical intervention in the medical intervention behavior records, filters abnormal periods with numerical deviations greater than the preset intervention failure deviation threshold, establishes secondary intervention failure links connecting relevant nodes within the abnormal periods, and generates a set of secondary intervention failure links. Access the database interface of the medical management system to obtain the medical intervention behavior records of the monitored object. These records include medication time, drug type, dosage, and expected onset time window. Align the time axis of the evolution trend vector with the time axis of the medical intervention records. Read the expected standard rate of change for each intervention behavior. This standard rate of change is derived from pharmacokinetic data accumulated during the drug clinical trial phase. For example, if a drug is expected to cause an indicator to change at a rate of 0.5 decrease per unit time, then the expected standard rate of change is -0.5. Calculate the numerical deviation between the actual rate of change in the corresponding time period in the evolution trend vector and this standard rate of change. The deviation is calculated using the absolute difference method. Read the preset intervention failure deviation threshold. This threshold is set based on the distribution of the rate of change deviation in a large number of cases of ineffective treatment with similar drugs, selecting the 25th percentile value of the distribution curve as the threshold. For example, statistically analyze the deviation data of ineffective cases and determine that the 25th percentile is 0.3. Compare the calculated deviation value with this threshold. The comparison results are categorized into two possibilities: if the deviation value is greater than the preset intervention failure deviation threshold, the period is determined to be an abnormal period; if the deviation value is less than or equal to the threshold, it is determined to be a normal response. For example, if the actual rate of change is 0.1 and the expected rate is -0.5, the deviation is... .because If the intervention fails, it is determined to be an intervention failure. During the abnormal period, a secondary intervention failure link is established between the status node at the current time point and the status node of the same type at the next time point.

[0032] The hierarchical network synthesis submodule takes the secondary intervention failure links in the set of secondary intervention failure links as the vertical evolution dimension, adds them to the horizontal correlation topology of the disease, connects the structured disease state nodes of the same symptom at multiple time points, and constructs a multidimensional disease hierarchical evolution network. A horizontal correlation topology graph of the disease condition is loaded, reflecting the connections between different symptoms at the same time. A set of secondary intervention failure links is introduced, representing longitudinal evolution anomalies in the time dimension. Each link in the set of secondary intervention failure links is traversed, and the starting node (time point) connected by the link is identified. ) and termination node (time point) In the horizontal correlation topology of the disease, the two corresponding nodes are located, and a vertical edge is added between them. The originally isolated time slices are then connected through vertical failure links, forming a three-dimensional network structure. In this network, nodes are connected not only to other symptom nodes at the same time through first-level concurrent correlation links, but also to nodes of the same type of symptom at the previous or next time point through second-level intervention failure links. This structure is defined as a multidimensional disease hierarchical evolution network. The network is then validated to ensure all links are correctly connected, and attribute labels are added to the edges to distinguish between "concurrent correlation" and "intervention failure."

[0033] Please see Figure 2 and Figure 7 The information sharing module includes: The composite path retrieval submodule retrieves paths in the multidimensional disease progression network that are formed by connecting the first-level concurrent links in the horizontal direction and the second-level intervention failure links in the vertical direction. It identifies continuous structured disease state node sequences that simultaneously have the characteristics of concurrent symptoms and intervention failure, locks the key abnormal propagation chains in the multidimensional disease progression network, and obtains long chain composite abnormal paths. Key anomalous propagation chains identified in the multidimensional disease progression network include: A full-domain scan was performed on the multidimensional disease progression network to identify all connected components in the network topology that are interconnected by horizontal first-level concurrent links and vertical second-level intervention failure links. Extract the continuous structured disease state node sequence corresponding to each connected component, and count the total number of structured disease state nodes in the continuous structured disease state node sequence, which is defined as the path span value. Compare the path span value with the preset minimum threshold for the length of the anomaly propagation chain; If the comparison results show that the path span value is greater than or equal to the minimum threshold of the abnormal propagation chain length, then the continuous structured disease state node sequence is determined to be a sequence with long-range evolution characteristics, and the sequence with long-range evolution characteristics is locked and persistently marked in memory address. A search algorithm is executed within a multidimensional disease progression network. The search objective is to find paths consisting of alternating horizontal first-level concurrent links and vertical second-level intervention failure links. All connected components meeting the criteria are identified, and their corresponding continuous structured disease state node sequences are extracted. For each extracted sequence, the total number of nodes is counted and defined as the path span value. A preset minimum threshold for the length of an abnormal propagation chain is read, which is set with reference to the average number of disease progression steps from the appearance of initial symptoms to the development of serious complications in chronic diseases. For example, if statistical data shows that an average of 5 key state nodes are involved from a single symptom to multi-organ complications, the threshold is set to 5. The calculated path span value is compared numerically with this threshold. The comparison result has two possibilities: first, if the path span value is greater than or equal to the minimum threshold for the length of the abnormal propagation chain, the sequence is judged to have long-term evolution characteristics; second, if the path span value is less than the threshold, it is considered a short-term fluctuation and is not highlighted. For example, if the retrieved path contains 6 nodes, since... If the path is not found to be a long-chain complex abnormal path, then the path is identified as such. These key sequences are locked in memory, their memory addresses are marked, and they are serialized and persistently stored.

[0034] The risk weight accumulation submodule identifies each structured disease status node in the long chain of complex abnormal paths, queries the preset risk weight value corresponding to each structured disease status node, accumulates the risk weights of all structured disease status nodes, and calculates the collaborative follow-up risk score. Read the locked long-chain complex anomaly path and traverse each structured disease state node contained in the path. Access the built-in risk weight dictionary, which is constructed based on the severity scores of different pathological states given by the medical expert system. For example, the weight corresponding to Code A state is 0.5, and the weight corresponding to Code B state is 0.8. The specific values ​​of these weights are determined through a Delphi method expert survey, which statistically analyzes the scores of multiple experts and takes the average. Initialize an accumulator to accumulate the risk weight values ​​of all nodes in the path one by one. For example, if the path contains 3 nodes with weights of 0.5, 0.8, and 0.7 respectively, the collaborative follow-up risk score is calculated as follows: This score comprehensively reflects the degree of continuous deterioration of the patient's current condition over time and the severity of complications in the spatial dimension. The score result is then correlated with the corresponding long-chain complex anomaly path object.

[0035] The shared information generation submodule filters target paths whose collaborative follow-up risk scores exceed the preset follow-up priority threshold, extracts chronic patient vital signs data and clinical business status identifiers from long chain composite abnormal paths associated with the target paths, and converts them into a standardized transmission protocol format to generate chronic disease collaborative follow-up shared information. The system receives the calculated collaborative follow-up risk score and compares it with a preset follow-up priority threshold. This threshold is set based on the allocation capacity of medical resources and the principle of hierarchical medical treatment. The distribution of historical follow-up score data within the jurisdiction is statistically analyzed, and the top 10% of the scores are set as high-priority thresholds. For example, if the 90th percentile of historical score data is 1.8, then the follow-up priority threshold is set to 1.8. The risk score is compared with the threshold, and the result is divided into two possibilities: first, the score exceeds or equals the threshold, indicating that the patient corresponding to the target pathway is in an extremely high-risk state; second, the score is below the threshold, indicating a normal risk. For example, if the calculated score is 2.0, since... This triggers a high-priority decision. Core data is extracted from the long, complex abnormal paths associated with these target paths, including raw chronic disease patient vital signs data, generated clinical operational status identifiers, and identified intervention failure points. Following international standard transmission protocols, the extracted data is encapsulated into a standardized message format. The generated message includes the patient's identification, a detailed temporal description of the abnormal path, a risk score, and recommended intervention measures.

[0036] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A chronic patient follow-up sharing and collaborative management system based on Java, characterized in that, The system includes: The rule base construction module obtains the vital sign data of chronic patients of the objects to be monitored, defines Java vital sign object entity classes for vital sign data, extracts business status interval metadata from them, and builds a status mapping rule base. The node generation module compares each vital sign data item in the state mapping rule base with the business state interval metadata, encapsulates it into a structured disease state node, and constructs a set of structured disease state nodes. The association mining module, based on the set of structured disease status nodes, calculates the co-occurrence support between structured disease status nodes under the same collection timestamp, filters concurrent association pairs of symptoms, and establishes a horizontal association topology graph of disease status. The network construction module, based on the structured disease state node set, analyzes the evolution trend vector of the structured disease state nodes under continuous collection timestamps, compares it with medical intervention behavior records, and accumulates the horizontal correlation topology graph of the disease to construct a multidimensional disease hierarchical evolution network. The information sharing module traverses the multidimensional disease progression network, calculates the collaborative follow-up risk score for specified structured disease state nodes, and generates collaborative follow-up sharing information for chronic diseases.

2. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 1, characterized in that, The state mapping rule base includes business state interval metadata defined by Java vital signs object entity classes and clinical business state codes corresponding to the business state interval metadata. The structured disease state node set includes structured disease state nodes encapsulating collection timestamps, physiological parameter type identifiers, physiological parameter values, and clinical business state codes. The disease horizontal association topology graph includes concurrent symptom association pairs whose co-occurrence support satisfies a preset confidence threshold and first-level concurrent association links connecting the concurrent symptom association pairs. The multidimensional disease hierarchical evolution network includes second-level intervention failure links generated by comparing evolution trend vectors with medical intervention behavior records and the first-level concurrent association links. The chronic disease collaborative follow-up shared information specifically consists of a collaborative follow-up risk score calculated by accumulating preset risk weights of structured disease state nodes in long-chain composite abnormal paths and the long-chain composite abnormal paths used as the scoring basis.

3. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 1, characterized in that, The rule base construction module includes: The entity object definition submodule obtains the systolic blood pressure, diastolic blood pressure and fasting blood glucose values ​​of the object to be monitored through the IoT medical device interface, constructs a class attribute mapping structure to match the vital signs data of chronic patients, marks the corresponding attribute fields with annotation tags defined for the boundaries of systolic blood pressure, diastolic blood pressure and blood glucose values, and encapsulates the marked attribute fields to generate Java vital signs object entity classes. The interval metadata parsing submodule calls the Java reflection mechanism to parse the Java vital signs object entity class, reads the annotation tag content marked on the attribute field, extracts the systolic blood pressure value boundary, diastolic blood pressure value boundary and fasting blood glucose value boundary from the annotation tag content, classifies each type of boundary information according to physiological parameter type, determines the value range of physiological parameters that exceed the physiological safety threshold and the value range that does not exceed the physiological safety threshold, and obtains the business status interval metadata. The status encoding mapping submodule iterates through the boundary intervals of each value in the business status interval metadata, assigns a unique character sequence as a clinical business status identifier to each independent boundary interval of systolic blood pressure, diastolic blood pressure, and fasting blood glucose, establishes a key-value index relationship between the boundary intervals and the character sequences, and constructs a status mapping rule base.

4. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 3, characterized in that, The process of invoking Java reflection to parse the Java vital signs object entity class includes: Obtain the class type description object of the Java vital signs object entity class in the runtime environment, use the class type description object to call the declaration field retrieval interface, obtain all private and public member variables declared in the Java vital signs object entity class, and encapsulate the obtained variables into a set of attribute fields to be parsed; Iterate through the set of attribute fields to be parsed, and for each attribute field in the set, call the access control override method provided by the reflection API to set the access permission flag of the attribute field to be parsed to the enabled state, thereby removing the language-level encapsulation restrictions on the attribute field to be parsed. Check if the modifier list of the attribute field to be parsed contains a type definition reference of the annotation tag; If the detection result confirms the annotation tag, then obtain the dynamic proxy instance object corresponding to the annotation tag through the reflection context, and read the boundary value string stored inside the annotation tag through the dynamic proxy instance object; Type conversion is performed on the boundary numerical strings to convert the text-formatted data into double-precision floating-point numbers, thus obtaining the boundaries of systolic blood pressure, diastolic blood pressure, and fasting blood glucose values.

5. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 1, characterized in that, The node generation module includes: The temporal feature extraction submodule extracts the systolic blood pressure, diastolic blood pressure, fasting blood glucose, and the collection timestamp corresponding to each value from the chronic patient's vital signs data. It identifies the physiological parameter type identifier to which each value belongs and arranges each value and the collection timestamp in chronological order to form a temporal value sequence of vital signs to be processed. The state matching and conversion submodule compares each value in the time-series numerical sequence of vital signs with the boundary interval of each value in the business state interval metadata, determines the range of the boundary interval in which each value falls, indexes the corresponding character sequence in the state mapping rule base according to the range of the boundary interval in which it falls, assigns an identifier representing clinical meaning to each vital sign value, and generates a clinical business state code. The status node encapsulation submodule encapsulates the physiological parameter type identifier, clinical business status code, collection timestamp, and corresponding systolic blood pressure, diastolic blood pressure, or fasting blood glucose value into a structured set of disease status nodes.

6. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 5, characterized in that, The identifiers assigned to each vital sign value to represent its clinical meaning include: Determine the range of values ​​in the business status interval metadata that each vital sign value currently to be processed in the vital sign time series numerical sequence falls within; Extract the minimum and maximum boundary values ​​of the numerical range, convert the minimum and maximum boundary values ​​into string format and concatenate them to generate the range positioning key for the retrieval state mapping rule base; Iterate through all key-value index relationships stored in the state mapping rule base and lock the target index entry that perfectly matches the numerical range; Access the value range space of the target index entry, read the character sequence stored in the value range space, and temporarily store the character sequence as a clinical status label to be assigned; Compare whether the applicable parameter types defined in the character sequence match the physiological parameter type identifiers corresponding to each vital sign value to be processed; If the matching result is consistent, the character sequence is confirmed as a unique identifier describing the physiological state of each vital sign value to be processed. Establish a unique identifier and a reference association between the memory address of each vital sign value to be processed, and write the unique identifier into the metadata description structure of each vital sign value to be processed.

7. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 1, characterized in that, The association mining module includes: The node co-occurrence calculation submodule groups the set of structured disease state nodes according to the collection timestamp. For the grouped data under the same collection timestamp, it counts the frequency of the simultaneous occurrence of structured disease state nodes with multiple physiological parameter type identifiers, calculates the probability of the occurrence of a specified combination of structured disease state nodes in the overall dataset, and quantifies the co-occurrence support of structured disease state nodes. The concurrent association filtering submodule compares the co-occurrence support of the structured disease state nodes with a preset co-occurrence confidence threshold, retains the structured disease state node combinations whose co-occurrence support is greater than or equal to the co-occurrence confidence threshold, and establishes a first-level concurrent association link to generate symptom concurrent association pairs. The horizontal topology construction submodule extracts the first-level concurrent association links in the symptom concurrent association pairs, and connects them with structured disease status nodes as vertices and first-level concurrent association links as edges to reflect the graph structure of the mutual influence relationship between multiple symptoms at the same time, thus obtaining the disease horizontal association topology graph.

8. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 1, characterized in that, The network construction module includes: The evolution trend calculation submodule calculates the slope of the numerical change of structured disease state nodes with the same physiological parameter type identifier in the structured disease state node set in chronological order within the continuous collection timestamps, and fits and generates an evolution trend vector describing the dynamic evolution of a single symptom over time. The intervention response verification submodule acquires medical intervention behavior records that are aligned with the evolution trend vector in time, calculates the numerical deviation between the rate of change of the evolution trend vector and the standard rate of change expected by the medical intervention in the medical intervention behavior records, filters abnormal periods with numerical deviations greater than a preset intervention failure deviation threshold, establishes a secondary intervention failure link connecting relevant nodes within the abnormal period, and generates a set of secondary intervention failure links. The hierarchical network synthesis submodule takes the secondary intervention failure links in the set of secondary intervention failure links as the vertical evolution dimension, adds them to the horizontal correlation topology of the disease, connects the structured disease state nodes of the same symptom at multiple time points, and constructs a multidimensional disease hierarchical evolution network.

9. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 1, characterized in that, The information sharing module includes: The composite path retrieval submodule retrieves paths in the multidimensional disease progression network that are formed by connecting the first-level concurrent links in the horizontal direction and the second-level intervention failure links in the vertical direction. It identifies continuous structured disease state node sequences that simultaneously have the characteristics of concurrent symptoms and intervention failure, locks the key abnormal propagation chains in the multidimensional disease progression network, and obtains long-chain composite abnormal paths. The risk weight accumulation submodule identifies each structured disease state node in the long chain composite abnormal path, queries the preset risk weight value corresponding to each structured disease state node, accumulates the risk weights of all structured disease state nodes, and calculates the collaborative follow-up risk score. The shared information generation submodule filters target paths whose collaborative follow-up risk scores exceed a preset follow-up priority threshold, extracts chronic patient vital signs data and clinical business status identifiers from the long chain composite abnormal paths associated with the target paths, and converts them into a standardized transmission protocol format to generate chronic disease collaborative follow-up shared information.

10. The chronic patient follow-up sharing and collaborative management system based on Java design according to claim 9, characterized in that, The key abnormal propagation chains in the multidimensional disease progression network include: A full-domain scan was performed on the multidimensional disease progression network to identify all connected components in the network topology that are interconnected by horizontal first-level concurrent links and vertical second-level intervention failure links. Extract the continuous structured disease state node sequence corresponding to each connected component, and count the total number of structured disease state nodes in the continuous structured disease state node sequence, which is defined as the path span value. Compare the path span value with the preset minimum threshold for the length of the anomaly propagation chain; If the comparison results show that the path span value is greater than or equal to the minimum threshold of the abnormal propagation chain length, then the continuous structured disease state node sequence is determined to be a sequence with long-range evolution characteristics, and the sequence with long-range evolution characteristics is locked and persistently marked in memory address.

Images