Chronic kidney disease intelligent follow-up visit management system and method based on multi-source data

Abstract

The invention relates to the technical field of medical information, and discloses a chronic kidney disease intelligent follow-up visit management system and method based on multi-source data. According to the method, a multi-source heterogeneous data set is formed by collecting an electronic health record of a patient, real-time wearable equipment monitoring data and active feedback information of the patient. A deep learning model is utilized, and a time sequence data analysis and cross-modal feature alignment module is combined, so that multi-source data collaborative analysis is realized, and the kidney function state of a patient is accurately evaluated. And according to the evaluation result and the individual difference of the patient, a personalized follow-up visit strategy is dynamically generated and comprises a follow-up visit period, a necessary inspection item and a risk intervention priority. The system automatically executes follow-up visit tasks, tracks execution states in real time and feeds back the execution states to the evaluation model, closed-loop iterative optimization is formed, and the chronic kidney disease management efficiency is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical information technology, in particular to a chronic kidney disease intelligent follow-up management system and method based on multi-source data. BACKGROUND

[0002] Traditional chronic kidney disease management relies on regular follow-up by artificial means, and there are problems of data fragmentation and risk assessment lag. Electronic health records, wearable devices and patient complaint data belong to different systems, resulting in a serious information island phenomenon. Existing technologies mostly use single-modal analysis methods, which are difficult to capture dynamic evolution characteristics of kidney function, such as evaluating only through static blood test indicators while ignoring real-time physiological fluctuations and patient behavior data. At the same time, fixed follow-up cycles and uniform examination items cannot adapt to individual differences in disease progression, easily causing over-medical treatment of low-risk patients or insufficient intervention of high-risk patients. In addition, the lack of a closed-loop feedback mechanism makes it difficult to dynamically optimize management strategies according to the execution effect, making it difficult to meet the long-term and heterogeneous management needs of chronic kidney disease. SUMMARY

[0003] In view of the above technical deficiencies, the purpose of the present application is to provide a chronic kidney disease intelligent follow-up management system and method based on multi-source data, which solves the problem of lack of dynamic individualized adjustment and multi-source data collaborative analysis in the prior art.

[0004] To solve the above technical problems, the present application adopts the following technical solutions: In a first aspect, the present application provides a chronic kidney disease intelligent follow-up management method based on multi-source data, which comprises: Collecting electronic health records, real-time wearable device monitoring data and patient active feedback information of patients to form a multi-source heterogeneous data set; The patient active feedback information includes symptom description and medication record; Based on the multi-source heterogeneous data set, a deep learning model is used to evaluate the current kidney function status of the patient, and the deep learning model realizes multi-source data collaborative analysis through a time series data analysis module and a cross-modal feature alignment module; According to the evaluation results and individual differences of the patient, a personalized follow-up strategy is dynamically generated, which includes follow-up cycle, mandatory examination item and risk intervention priority; The personalized follow-up strategy is converted into executable task instructions to automatically trigger follow-up reminders, examination appointments and patient education content push, while real-time tracking of task execution status and feedback to the evaluation model form a closed-loop iterative optimization.

[0005] Preferably, in a possible implementation of the first aspect, the time series data analysis module specifically comprises: A hybrid architecture composed of bidirectional long short-term memory network and temporal convolutional network is constructed to extract time-series features in electronic health records. The real-time wearable device monitoring data stream is segmented by a sliding window mechanism. A self-attention mechanism is introduced to weight and aggregate time-series features, generating a state vector representing the dynamic evolution of kidney function. The state vector is used to quantify the rate of kidney function decline and the probability of acute attack risk.

[0006] Preferably, in one possible implementation of the first aspect, the cross-modal feature alignment module specifically comprises: A feature interaction layer based on cross-attention mechanism is adopted to align the state vector output by the time-series data analysis module with the patient's active feedback information. The semantic correlation of the three modalities of electronic health records, real-time wearable device monitoring data and patient active feedback information is calculated through a learnable weight distribution matrix. A feature mapping layer is used to project heterogeneous data into a unified hidden space, eliminating the scale difference of multi-source data.

[0007] Preferably, in one possible implementation of the first aspect, the cross-modal feature alignment module deploys a multi-modal fusion strategy and uses a gating mechanism to dynamically adjust the contribution weight of each modal feature to risk assessment. Combined with the prototype contrast learning algorithm, a prototype cluster of patient state is constructed in the hidden space, and the consistency of cross-modal representation is enhanced by minimizing the multi-modal feature distance of patients of the same type and maximizing the feature distance of patients of different types.

[0008] Preferably, in one possible implementation of the first aspect, the deep learning model is specifically implemented as: The time-series data analysis module is used as the feature extraction engine, and the cross-modal feature alignment module realizes feature collaborative optimization through a multi-modal fusion strategy to construct an end-to-end graph neural network architecture. The multi-source heterogeneous data of patients is embedded through graph nodes, and historical cases are connected through graph edges to realize kidney function state assessment based on graph reasoning. Self-supervised pre-training and dynamic weighted loss function are used to jointly optimize the model, where the loss function includes kidney function prediction error, complication classification cross-entropy and cross-modal alignment regularization term.

[0009] Preferably, in one possible implementation of the first aspect, the kidney function state assessment result includes dynamic indicators quantifying the rate of kidney function decline, acute attack risk probability value, complication classification risk level and kidney function state score.

[0010] Preferably, in a possible implementation form of the first aspect, the personalized follow-up strategy generation process comprises: a dynamic optimization model based on a multi-factor decision tree is constructed, quantitative indicators in the renal function state assessment results and patient individual difference characteristics are input, and follow-up strategy parameters are output; The follow-up strategy parameters include a follow-up cycle, a must-check item, and a risk intervention priority; The follow-up cycle is adaptively calculated by an ensemble learning algorithm in combination with a decline rate and a risk probability, the must-check item is dynamically selected by a risk classifier based on a complication probability, and the risk intervention priority is generated by an entropy weight method weighting score model; The patient individual difference includes age, medical history, and drug compliance.

[0011] Preferably, in a possible implementation form of the first aspect, the adaptive calculation of the follow-up cycle specifically comprises: a stacked generalization model is constructed to fuse a random forest and a gradient boosting decision tree, wherein the random forest processes high-dimensional characteristics, and the gradient boosting decision tree optimizes prediction based on patient individual difference characteristics; The stacked generalization model is trained by K-fold cross-validation, and outputs a dynamic follow-up interval.

[0012] Preferably, in a possible implementation form of the first aspect, the entropy weight method weighting score model is specifically implemented as: quantitative renal function decline rates, acute attack risk probability values, and complication risk levels in the renal function state assessment results are extracted, and combined with patient individual difference characteristics as input variables; An information entropy-based weight distribution mechanism is used to calculate variable weight coefficients to reflect the different influences of the variability of the variables on risk contribution; The variables are normalized to eliminate dimensions, and then weighted and fused to generate a comprehensive risk score; An adaptive sorting algorithm is used to dynamically map the score to a risk intervention priority sequence, and the weight coefficients are updated in combination with real-time wearable device data.

[0013] In a second aspect, the present application provides a chronic kidney disease intelligent follow-up management system based on multi-source data, which is used to implement the chronic kidney disease intelligent follow-up management method based on multi-source data in the first aspect, and comprises: A data acquisition module acquires electronic health records of patients, real-time wearable device monitoring data, and patient active feedback information, and forms a multi-source heterogeneous data set; A state assessment module assesses the current renal function state of a patient based on the multi-source heterogeneous data set by using a deep learning model, and the deep learning model realizes multi-source data collaborative analysis through a time series data analysis module and a cross-modal feature alignment module; A strategy generation module dynamically generates a personalized follow-up strategy according to the evaluation result and the individual difference of the patient, the personalized follow-up strategy including a follow-up cycle, a must-check item and a risk intervention priority; A task execution module converts the personalized follow-up strategy into executable task instructions, automatically triggers follow-up reminders, examination appointments and patient education content pushing, simultaneously tracks the task execution state in real time and feeds back to the evaluation model to form a closed-loop iterative optimization.

[0014] The present application has the beneficial effects that: through multi-source heterogeneous data fusion and a deep learning model, the precise capture of the dynamic evolution characteristics of renal function is realized. The time series data analysis module combines bidirectional LSTM and time convolution network, effectively quantifies the renal function decline rate and acute attack risk; the cross-modal feature alignment module eliminates the data scale difference through the cross-attention mechanism, and enhances the consistency of multi-modal representation.

[0015] The system can dynamically generate a personalized strategy including a follow-up cycle, a must-check item and a risk priority, and the closed-loop iterative mechanism continuously optimizes the model through the task execution state feedback, so that the complication prediction accuracy is improved. The scheme significantly improves the efficiency of chronic kidney disease management and reduces the incidence of acute exacerbation. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0017] Figure 1 A flow chart of a chronic kidney disease intelligent follow-up management method based on multi-source data is provided for the present application.

[0018] Figure 2 A system structure diagram of a chronic kidney disease intelligent follow-up management system based on multi-source data is provided for the present application.

[0019] Marked with a figure: 1-data acquisition module, 2-state evaluation module, 3-strategy generation module, 4-task execution module. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0021] Embodiment One: As shown, the present application provides a chronic kidney disease intelligent follow-up management method based on multi-source data, comprising: Figure 1 Collecting the electronic health records of patients, real-time wearable device monitoring data and patient active feedback information to form a multi-source heterogeneous data set.

[0022] In this embodiment, the electronic health records are obtained through the hospital information system interface, including patient history test reports, image data, medication records and outpatient medical history texts, and the data format covers structured tables and unstructured texts. Real-time wearable device monitoring data is accessed to the system through Bluetooth and Internet of Things protocols, dynamically collecting physiological indicators such as patient blood pressure, heart rate, blood oxygen saturation and daily steps, and the sampling frequency is set according to clinical needs. Patient active feedback information is collected by a mobile terminal application, and patient subjective symptom description and medication records are obtained through a pre-set questionnaire form.

[0023] For the heterogeneity problem of multi-source data, the system first performs data preprocessing. For unstructured text in electronic health records, named entity recognition technology is used to extract key medical terms and convert them to standardized codes. Wearable device data stream is divided into time segments by sliding window mechanism, and abnormal values are calibrated by Kalman filter. Patient active feedback information is extracted by natural language processing model, and converted into structured feature vector.

[0024] The processed multi-source data is stored in the database after timestamp alignment, forming a multi-source heterogeneous data set with patient ID as the primary key.

[0025] Based on the multi-source heterogeneous data set, a deep learning model is used to evaluate the current kidney function status of the patient. The deep learning model realizes collaborative analysis of multi-source data through a time series data analysis module and a cross-modal feature alignment module.

[0026] In this embodiment, the time series data analysis module constructs a hybrid architecture composed of bidirectional long short-term memory network and time convolution network as the feature extraction engine. The historical test indicators, medication records and disease course texts in the electronic health records are input into the hybrid architecture. The bidirectional long short-term memory network captures long-term dependencies, and the time convolution network extracts local time series patterns through an inflation convolution kernel.

[0027] For continuous monitoring data stream generated by real-time wearable devices, a sliding window mechanism is used to divide it into fixed length segments, and each segment is standardized and sent to the hybrid architecture for processing.

[0028] ​In the feature aggregation stage, the extracted time-series features are dynamically weighted using a self-attention mechanism to generate a state vector representing the dynamic evolution of kidney function. This state vector contains two key quantitative indicators: first, the rate of kidney function decline based on the first derivative of the state vector, which represents the linear or nonlinear change trend of kidney function; second, the acute episode risk probability value calculated by the fully connected layer and the Sigmoid activation function, which reflects the possibility of short-term kidney function deterioration.

[0029] The cross-modal feature alignment module is responsible for fusing the state vector output by the time-series module and the patient's active feedback information. First, a feature interaction layer based on cross-attention mechanism is used to realize multi-modal feature alignment: the time-series state vector is used as the query vector, and the patient's active feedback information is used as the key-value vector. The semantic correlation between the two is calculated through a learnable weight distribution matrix to generate a feedback feature vector aligned with the time feature.

[0030] Meanwhile, the cross-modal feature alignment module calculates the global semantic correlation matrix of the three modalities of electronic health records, real-time monitoring data and patient active feedback information through another parallel branch, quantifying the consistency of the contribution of different modalities to the same clinical representation. Subsequently, the feature mapping layer projects the heterogeneous data into a unified hidden space, and uses a multi-layer perceptron to eliminate the scale difference of different modalities of data.

[0031] To further optimize the multi-modal fusion effect, a gating mechanism is used to dynamically adjust the feature weights of each modality: a differentiable gating unit is used to calculate the contribution coefficients of electronic health records, monitoring data and feedback information in real time according to the context information of the current patient state, and to suppress the influence of noise modalities. In addition, prototype contrast learning algorithm is used to enhance the consistency of the representation: K patient state prototype clusters are constructed in the hidden space (in this embodiment, stable period, slow progression period and high risk period are used), and the distance between multi-modal features of patients of the same type (in this embodiment, Euclidean distance is used) is minimized and the distance between features of patients of different types is maximized, to drive the model to learn the essential state representation independent of the modality.

[0032] The overall architecture of the deep learning model uses an end-to-end graph neural network to realize kidney function state assessment. The time-series data analysis module is used as the feature extraction engine, and the cross-modal feature alignment module uses a multi-modal fusion strategy to realize feature collaborative optimization. The patient's multi-source heterogeneous data is embedded as a graph node, and each node represents the comprehensive state of a patient. Historical cases are connected through the graph edges: the feature similarity between the current patient node and the historical case node is calculated by cosine similarity, and the Top-K similar nodes are retained to construct a dynamic subgraph.

[0033] The graph reasoning process adopts a graph attention network to aggregate the state information of neighboring nodes (in this embodiment, the historical complication development trajectory and intervention effect) to generate an enhanced assessment result of the kidney function state. The result includes four dimensions of quantitative output: first, a dynamic indicator of the rate of kidney function decline, outputting a specific numerical value; second, an acute attack risk probability value, ranging from 0 to 1; third, a complication classification risk level, outputting the probability distribution of complications such as diabetic nephropathy and cardiovascular events through a Softmax layer and mapping to low, medium and high levels; and fourth, a comprehensive kidney function state score, generated by a fully connected layer integrating the first three indicators into a continuous value of 0 to 100 points.

[0034] The model training strategy adopts a two-stage optimization method. In the first stage, self-supervised pre-training is performed: a mask recovery task is designed using unlabeled historical data to learn general representations through reconstruction loss. In the second stage, a dynamic weighted loss function is introduced for supervised fine-tuning. The loss function includes three terms: the Huber loss function is used for the prediction error of kidney function to provide robust constraints for the regression tasks of the decline rate and state score; the cross-entropy loss of complication classification optimizes the multi-label classification performance; and the cross-modal alignment regularization term constrains the consistency of features in the hidden space of different modalities by maximizing mutual information. During training, an adaptive weight scheduling algorithm is used to dynamically adjust the loss term weights based on the performance of each task on the validation set, ensuring balanced convergence in multi-objective optimization.

[0035] In this embodiment, when the patient adds new electronic medical records or wearable device data, the time series data analysis module starts incremental calculation to update the state vector; after the patient submits feedback information, the cross-modal feature alignment module responds and refreshes the fused features. The graph neural network automatically reconstructs the patient subgraph every 24 hours to include the latest similar case information to optimize the evaluation results.

[0036] According to the evaluation results and individual differences of the patient, personalized follow-up strategies are dynamically generated, including follow-up cycle, mandatory items and risk intervention priority.

[0037] In this embodiment, the strategy generation process is realized by constructing a multi-factor decision tree dynamic optimization model. The model takes the quantitative indicators in the kidney function state assessment results as input, including the rate of kidney function decline, the acute attack risk probability value, and the complication classification risk level, while integrating patient individual difference features, including patient age, medical history and drug adherence. Through the hierarchical judgment logic of the multi-factor decision tree, the follow-up cycle, mandatory item list and risk intervention priority sequence are dynamically output as three types of strategy parameters.

[0038] The adaptive calculation of follow-up cycle is implemented by a stacked generalization model. This model combines random forest algorithm and gradient boosting decision tree algorithm to form a two-level prediction architecture. The first level random forest model processes high-dimensional medical features, such as analyzing the non-linear relationship between the rate of kidney function decline and the probability of acute episode. When the annual decline of glomerular filtration rate of a patient exceeds 5 and the probability of acute episode is greater than 0.3, the model automatically identifies it as a high-risk mode.

[0039] The second level gradient boosting decision tree is optimized and corrected based on the individual difference features of the patient, further shortening the follow-up interval based on the output of the random forest. The stacked generalization model is trained by K-fold cross-validation, uses early stopping mechanism to prevent overfitting, and finally generates a follow-up cycle that is negatively correlated with the risk level. The follow-up interval of high-risk patients is compressed to 7-14 days, the standard cycle of 30 days is maintained for medium-risk patients, and the cycle of low-risk patients is extended to 60-90 days, and the clinical rationality of interval adjustment is ensured by regularization constraint.

[0040] The dynamic selection of mandatory items is completed by the risk classifier module. This module generates a list of mandatory items based on the risk level of complications. For example, when the model detects that the risk level of diabetic nephropathy has risen to medium, it adds urine protein creatinine ratio detection and fundus examination; if the probability of cardiovascular event risk exceeds the threshold, electrocardiogram and myocardial enzyme spectrum detection are included. The risk classifier combines real-time wearable device data stream for incremental update, such as patients with abnormal blood pressure monitoring for 3 consecutive days, even if the original complication risk level has not increased, 24-hour ambulatory blood pressure monitoring will be temporarily added.

[0041] The risk intervention priority generation adopts an entropy weight method weighted scoring model. First, extract the rate of kidney function decline, acute episode risk probability value, and complication risk level three core indicators, and combine them with patient individual difference features to form an input variable matrix. Through information entropy weight distribution mechanism, the weight coefficient of each variable is calculated: the greater the variation, the higher the weight, reflecting its difference contribution to the comprehensive risk.

[0042] After all variables are normalized by maximum and minimum to eliminate dimensional differences, the weight coefficient is weighted to generate a comprehensive risk score of 0 to 100. Finally, through an adaptive sorting algorithm, the score is mapped to an intervention priority sequence: patients with a score greater than or equal to 80 are included in the emergency intervention group, patients with a score greater than or equal to 60 but less than 80 are included in the priority processing group, and patients with a score less than 60 are included in the routine follow-up group. The model updates the weight coefficient every 24 hours combined with the latest wearable device data to ensure the dynamic accuracy of the priority sorting.

[0043] The personalized follow-up strategy is converted into executable task instructions, automatically triggering follow-up reminders, examination appointments, and patient education content push, while real-time tracking of task execution status and feedback to the evaluation model form a closed-loop iterative optimization.

[0044] In this embodiment, the personalized follow-up strategy output by the strategy generation module is converted into a structured instruction set via a task parsing engine. The follow-up cycle parameters trigger a scheduled task scheduler, which pushes a follow-up reminder message to the patient's mobile terminal at a preset time. The message content includes the appointment location, doctor information, and precautions. The mandatory examination item parameters drive the hospital information system interface to automatically generate an examination order, match the optimal examination time slot using a reservation algorithm, and synchronize the appointment certificate to the patient's terminal and department terminal.

[0045] Risk intervention priority instructions activate a multi-level response mechanism. High-priority tasks trigger immediate voice alerts and text message notifications, medium-priority tasks are integrated into the patient's daily to-do list, and low-priority tasks are incorporated into periodic health education plans. Patient education content is dynamically generated using a knowledge graph engine. Based on current risk assessment results and historical patient feedback preferences, it automatically combines video explanations, illustrated manuals, and interactive Q&A materials, and delivers them to mobile devices in batches according to risk level.

[0046] The task execution status is tracked in real time through a distributed log system, recording events such as instruction issuance time, patient confirmation time, examination completion rate, and educational content opening rate, and annotating abnormal status codes such as overdue appointments, unread reminders, or missing data. The tracking data is cleaned by the stream processing engine to form an execution performance report, which is then injected into the incremental training data set of the status evaluation module through the feedback interface. When the same patient has two consecutive tasks overdue or key physiological indicators deviate from the predicted trajectory, the system automatically triggers the recalculation process of the evaluation model, using the latest execution status data to dynamically update the patient status evaluation results, completing the closed-loop optimization iteration.

[0047] Example 2: Figure 2 As shown, the present invention provides an intelligent follow-up management system for chronic kidney disease based on multi-source data, comprising: Data collection module 1 collects patients' electronic health records, real-time wearable device monitoring data, and patient-initiated feedback information to form a multi-source heterogeneous data set; Status Assessment Module 2: Based on multi-source heterogeneous datasets, it uses a deep learning model to assess the patient's current renal function status. The deep learning model implements collaborative analysis of multi-source data through a time series data analysis module and a cross-modal feature alignment module. Strategy generation module 3 dynamically generates personalized follow-up strategies based on assessment results and individual patient differences. The personalized follow-up strategies include follow-up cycles, mandatory inspection items, and risk intervention priorities. Task execution module 4 converts personalized follow-up strategies into executable task instructions, automatically triggering follow-up reminders, examination appointments, and patient education content push. At the same time, it tracks the task execution status in real time and feeds it back to the evaluation model to form a closed-loop iterative optimization.

[0048] It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application cover modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

Claims

1. An intelligent follow-up management method for chronic kidney disease based on multi-source data, characterized in that: The method comprises: Collect patients' electronic health records, real-time wearable device monitoring data, and patient-initiated feedback information to form a multi-source heterogeneous data set; The patient's active feedback information includes symptom description and medication records; Based on the multi-source heterogeneous dataset, a deep learning model is used to evaluate the patient's current renal function status. The deep learning model realizes multi-source data collaborative analysis through a time series data analysis module and a cross-modal feature alignment module; Dynamically generate personalized follow-up strategies based on assessment results and individual patient differences, including follow-up cycles, mandatory inspection items, and risk intervention priorities; The personalized follow-up strategy is converted into executable task instructions, which automatically trigger follow-up reminders, examination appointments and patient education content push. At the same time, the task execution status is tracked in real time and fed back to the evaluation model to form a closed-loop iterative optimization.

2. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 1, characterized in that: The time series data analysis module specifically includes: Constructing a hybrid architecture consisting of a bidirectional long short-term memory network and a temporal convolutional network to extract temporal features from electronic health records; Split the real-time wearable device monitoring data stream through a sliding window mechanism; A self-attention mechanism is introduced to perform weighted aggregation of temporal features to generate a state vector representing the dynamic evolution of renal function. The state vector is used to quantify the rate of renal function decline and the probability of acute attack risk.

3. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 2, characterized in that: The cross-modal feature alignment module specifically includes: A feature interaction layer based on the cross-attention mechanism is used to align the state vector output by the time series data analysis module with the patient's active feedback information; The semantic relevance of three modalities, namely electronic health records, real-time wearable device monitoring data, and patient-initiated feedback information, is calculated through a learnable weight assignment matrix. The feature mapping layer is used to project heterogeneous data into a unified latent space to eliminate the scale differences of multi-source data.

4. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 3, characterized in that: The cross-modal feature alignment module deploys a multimodal fusion strategy and uses a gating mechanism to dynamically adjust the contribution weight of each modal feature to risk assessment; Combined with the prototype contrastive learning algorithm, a prototype cluster of patient states is constructed in the latent space, which enhances the consistency of cross-modal representation by minimizing the multimodal feature distance of similar patients and maximizing the feature distance of heterogeneous patients.

5. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 4, characterized in that: The deep learning model is specifically implemented as follows: Using the time series data analysis module as the feature extraction engine, the cross-modal feature alignment module achieves feature collaborative optimization through a multimodal fusion strategy, building an end-to-end graph neural network architecture. By embedding patients' multi-source heterogeneous data through graph nodes and connecting historical cases using graph edges, we can achieve graph-based reasoning-based kidney function status assessment. The model was optimized jointly using self-supervised pre-training and a dynamic weighted loss function, where the loss function included renal function prediction error, complication classification cross entropy, and cross-modal alignment regularization term.

6. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 5, characterized in that: The renal function status assessment results include: dynamic indicators that quantify the rate of renal function decline, acute attack risk probability values, complication classification risk levels, and renal function status scores.

7. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 6, characterized in that: The personalized follow-up strategy generation process includes: Constructing a dynamic optimization model based on a multi-factor decision tree, inputting the quantitative indicators in the renal function status assessment results and the individual difference characteristics of the patients, and outputting follow-up strategy parameters; Follow-up strategy parameters include follow-up period, mandatory inspection items, and risk intervention priority; The follow-up period is adaptively calculated using an integrated learning algorithm that combines the decay rate and risk probability. Mandatory inspection items are dynamically selected by a risk classifier based on complication probability, and the risk intervention priority is generated using an entropy-based weighted scoring model. Individual patient differences include age, medical history, and medication compliance.

8. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 7, characterized in that: The adaptive calculation of the follow-up period specifically includes: A stacked generalization model was constructed by fusing random forests and gradient boosted decision trees, where random forests process high-dimensional features and gradient boosted decision trees optimize predictions based on individual patient differences. The stacked generalization model is trained via K-fold cross-validation and outputs dynamic follow-up intervals.

9. The intelligent follow-up management method for chronic kidney disease based on multi-source data according to claim 7, characterized in that: The entropy weighted scoring model is specifically implemented as follows: Extracting the quantitative renal function decline rate, acute attack risk probability value and complication risk level from the renal function status assessment results, combined with individual patient differences as input variables; A weight allocation mechanism based on information entropy is used to calculate the weight coefficient of each variable to reflect the differential impact of its variability on risk contribution; The variables are normalized to eliminate the dimension and then weighted fusion is performed to generate a comprehensive risk score; The scores are dynamically mapped into a risk intervention priority sequence through an adaptive sorting algorithm, and the weight coefficients are updated in combination with real-time wearable device data.

10. An intelligent follow-up management system for chronic kidney disease based on multi-source data, characterized in that: The system is used to implement the intelligent follow-up management method for chronic kidney disease based on multi-source data as described in claims 1 to 9, comprising: The data collection module collects patients' electronic health records, real-time wearable device monitoring data, and patient-initiated feedback information to form a multi-source heterogeneous data set; A status assessment module, which assesses the patient's current renal function status based on the multi-source heterogeneous dataset using a deep learning model, wherein the deep learning model implements multi-source data collaborative analysis through a time series data analysis module and a cross-modal feature alignment module; A strategy generation module dynamically generates personalized follow-up strategies based on assessment results and individual patient differences. The personalized follow-up strategies include follow-up cycles, mandatory inspection items, and risk intervention priorities. The task execution module converts the personalized follow-up strategy into executable task instructions, automatically triggers follow-up reminders, examination appointments and patient education content push, and simultaneously tracks the task execution status in real time and feeds it back to the evaluation model to form a closed-loop iterative optimization.

Images