A method and system for dynamically updating the risk of kawasaki disease based on full-disease-cycle follow-up data

Abstract

The present application belongs to the technical field of medical data processing, and particularly relates to a Kawasaki disease risk dynamic updating method and system based on whole disease cycle follow-up data. The method constructs a time series data structure by acquiring clinical data at multiple time points in the acute phase and the follow-up stage, calculates an initial risk assessment result based on the acute phase data using a risk assessment model, and when new follow-up data is acquired, the new data and historical data are used as joint input, the existing risk assessment result is iteratively updated based on a preset updating rule, and risk change trend analysis is performed based on the risk assessment results associatedly stored at each time point. The present application solves the problem in the prior art that risk assessment relies on single time point data and lacks a dynamic updating mechanism, and realizes continuous risk assessment covering the whole disease cycle.

Description

Technical Field

[0001] This invention belongs to the field of medical data processing technology, and in particular relates to a method and system for dynamically updating the risk of Kawasaki disease based on follow-up data throughout the entire disease cycle. Background Technology

[0002] Kawasaki disease is an acute systemic vasculitis in children characterized primarily by inflammation of small and medium-sized arteries, with coronary artery injury being its most serious complication. Studies have shown that without timely treatment, approximately 15% to 25% of affected children may develop coronary artery dilation or aneurysms, significantly impacting their long-term cardiovascular prognosis. Therefore, accurate risk assessment associated with Kawasaki disease is of paramount importance in clinical practice.

[0003] In recent years, with the development of echocardiography and laboratory testing technologies, more and more studies have attempted to use clinical data to establish risk assessment models. These models collect clinical manifestations, laboratory test indicators, and imaging results of children in the acute phase, and then construct predictive models based on statistical analysis methods, risk scoring systems, or machine learning algorithms to predict the risk of coronary artery injury in children with Kawasaki disease.

[0004] However, most existing Kawasaki disease risk assessment methods are based on clinical data from a single time point, such as acute-phase clinical manifestations or laboratory indicators. These methods typically perform a single risk assessment at the child's initial diagnosis or during the acute phase, and the results are primarily used to support acute-phase treatment decisions, exhibiting a clear static assessment characteristic. In actual clinical management, children with Kawasaki disease usually undergo acute-phase treatment and long-term follow-up. New clinical data is continuously acquired during follow-up, reflecting disease progression and risk changes. However, existing technologies lack a solution to integrate follow-up data from multiple time points and continuously update risk assessment results. This makes it difficult to dynamically update the output of existing risk assessment models based on new data acquired during follow-up, thus failing to accurately reflect changes in risk throughout the disease's progression.

[0005] Patent CN106339593B discloses a Kawasaki disease classification and prediction method based on medical data modeling. This method collects basic patient information, clinical symptoms, laboratory test indicators, and imaging results, and uses machine learning algorithms to build a classification model for disease state prediction. However, this method primarily builds its prediction model based on clinical data collected at a single point in time, making it difficult to integrate multi-time-point clinical data acquired during follow-up. Furthermore, this method only performs a risk assessment once during model building or initial prediction. When new follow-up data is acquired subsequently, it is difficult to dynamically update the existing risk assessment results, thus hindering continuous risk management during long-term follow-up.

[0006] Patent CA2955214A1 / EP2116618A1 proposes a method for monitoring the diagnosis and treatment of Kawasaki disease based on biomarker detection. This method diagnoses the disease by detecting the levels of biomarkers such as IL-17F and sCD40L, and monitors the treatment effect by detecting related biomarkers before and after treatment. However, this method primarily relies on detecting the levels of one or a few biomarkers, making its assessment results highly dependent on a single indicator and unable to comprehensively reflect the patient's disease status. Furthermore, its technical purpose mainly focuses on disease diagnosis and efficacy monitoring, rather than predicting the risk of complications such as coronary artery damage. In addition, this method typically relies on biomarker levels detected at a single point in time, lacking a systematic integration of data from different stages of the disease, making it difficult to reflect changes in the patient's risk throughout the disease process, and unable to achieve dynamic risk updates based on multi-timepoint data.

[0007] In summary, existing technologies lack a technical solution capable of constructing a time-series data structure based on follow-up data from multiple time points and dynamically updating risk assessment results when new follow-up data is acquired. How to achieve continuous updating and iterative calculation of risk assessment results based on follow-up data throughout the entire disease cycle has become a pressing technical problem to be solved in this field. Summary of the Invention

[0008] To address the aforementioned issues, the first technical solution of this application discloses a method for dynamically updating the risk of Kawasaki disease based on full-disease-cycle follow-up data, comprising:

[0009] Acquire Kawasaki disease-related clinical data of the target object at multiple time points, and construct a time series data structure based on the clinical data at the multiple time points. The clinical data includes data from the acute phase and data from the follow-up phase. The time series data structure associates the clinical data at different time points through a unified target object identifier.

[0010] Based on the acute phase data, an initial risk assessment result is calculated using a risk assessment model, and the initial risk assessment result is stored in association with the corresponding time tag.

[0011] When new follow-up phase data is acquired, the new follow-up phase data and historical time series data are used as joint inputs. The existing risk assessment results are iteratively updated based on preset update rules to generate updated risk assessment results. The updated risk assessment results are then associated with and stored with the corresponding time tags.

[0012] Based on the risk assessment results of associated storage at each time point, conduct risk change trend analysis;

[0013] The update rules include: updating model input variables based on newly added follow-up data; and calculating updated risk assessment results jointly based on historical risk assessment results and newly added follow-up data.

[0014] Furthermore, the construction of the time series data structure includes: standardizing clinical data at different time points to generate a unified data structure, and configuring time labels for data at each time point.

[0015] Furthermore, the risk assessment results include risk value, risk level, or a combination of risk value and risk level.

[0016] Furthermore, the risk assessment model includes a statistical analysis model, a machine learning model, or a scoring model.

[0017] Furthermore, the risk change trend analysis includes: calculating the difference, the rate of change, or analyzing the trend of risk assessment results at adjacent time points.

[0018] The second technical solution of this application discloses a dynamic risk update system for Kawasaki disease based on full disease cycle follow-up data, including:

[0019] The data acquisition module is used to acquire Kawasaki disease-related clinical data of the target object at multiple time points and construct a time series data structure;

[0020] The risk assessment module is used to calculate the initial risk assessment results based on data from the acute phase using a risk assessment model.

[0021] The risk update module is used to iteratively update the existing risk assessment results based on preset update rules when new follow-up phase data is acquired, using the new follow-up phase data and historical time series data as joint inputs.

[0022] The risk output module is used to output the risk assessment results at each time point and to perform risk change trend analysis based on the risk assessment results stored in association at each time point.

[0023] Furthermore, it also includes a data processing module, which is used to standardize clinical data at different time points, generate a unified data structure, and configure time tags for data at each time point.

[0024] Furthermore, the risk update module updates the model input variables based on the new follow-up phase data, and corrects or recalculates the risk assessment results based on historical risk assessment results and the new follow-up phase data.

[0025] And a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the method described in the first technical solution.

[0026] Compared with existing technologies, the Kawasaki disease risk dynamic update method and system based on full disease cycle follow-up data provided by this invention has at least the following beneficial effects:

[0027] (1) Achieve dynamic updating of risk assessment results

[0028] This invention triggers a risk update process when new follow-up data is acquired. By jointly calculating the new data and historical data, the risk assessment results are continuously updated, overcoming the problem in the prior art that the risk assessment results are based on data from a single point in time and are difficult to update.

[0029] (2) Establish a continuous risk assessment record covering the entire disease cycle.

[0030] This invention associates and stores risk assessment results generated at different time points with corresponding time tags, thereby forming a continuous risk assessment record throughout the acute and follow-up phases, enabling the risk assessment results to reflect the changes over time.

[0031] (3) Support for the comprehensive processing of multidimensional clinical data

[0032] This invention improves the completeness of data utilization by integrating multi-dimensional data such as clinical symptom information, laboratory test indicators, and imaging examination results obtained at different stages and processing and calculating them under a unified data structure.

[0033] (4) Improve the consistency of multi-time point data processing

[0034] This invention improves the comparability and processing consistency of data at different time points by standardizing data at different time points and performing risk calculations based on a unified data structure.

[0035] (5) It has good scalability

[0036] This invention does not limit the specific data type or risk assessment model form, and can be applied to different statistical models, machine learning models or scoring models, making it easy to expand according to actual application needs. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1A flowchart illustrating a method for dynamically updating the risk of Kawasaki disease based on full disease cycle follow-up data, provided in an embodiment of the present invention;

[0039] Figure 2 This is a schematic diagram of the structure of a dynamic risk update system for Kawasaki disease based on full disease cycle follow-up data, provided in an embodiment of the present invention. Detailed Implementation

[0040] To make the technical problems solved, the technical solutions, and the beneficial effects 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 of the invention and are not intended to limit the invention.

[0041] The first embodiment of this application provides a method for dynamically updating the risk of Kawasaki disease based on full-disease-cycle follow-up data. For example... Figure 1 As shown, this method acquires clinical data from multiple time points in the acute and follow-up phases of children with Kawasaki disease, constructs a time-series data structure, and iteratively updates existing risk assessment results based on new follow-up data, thereby achieving dynamic assessment and continuous monitoring of the risk of complications such as coronary artery injury.

[0042] Step 1: Obtain full disease cycle follow-up data and construct a time series data structure

[0043] The term "whole disease cycle" refers to the complete disease management process for children with Kawasaki disease, from the acute phase of onset, through the subacute phase, recovery phase, and long-term follow-up phase. The "whole disease cycle follow-up data" described in this application encompasses clinical data collected at each of the aforementioned stages, and is not limited to single data points from the acute phase.

[0044] The "time series data structure" refers to a data structure organized chronologically, where each time node corresponds to a set of clinical data records, and nodes are linked through a unified target object identifier. This structure supports data retrieval, appending, and updating operations along the time dimension.

[0045] Acquire Kawasaki disease-related clinical data from the target subjects (i.e., children with Kawasaki disease) at multiple time points. The clinical data includes at least acute phase data and data from one or more follow-up phases. The acute phase typically refers to the period from the onset of the disease to the initial treatment (e.g., within 1-2 weeks after onset); the follow-up phases include, but are not limited to, the subacute phase (e.g., 2-4 weeks after onset), the recovery phase (e.g., 1-3 months after onset), and long-term follow-up (e.g., 6 months, 1 year, or longer after onset).

[0046] The clinical data includes: basic patient information (such as age, gender, weight, number of days since onset), clinical symptom information (such as duration of fever, rash, conjunctival congestion, changes in oral mucosa, changes in the extremities, lymphadenopathy, etc.), laboratory test indicators (such as white blood cell count, neutrophil percentage, C-reactive protein, erythrocyte sedimentation rate, platelet count, hemoglobin, albumin, alanine aminotransferase, etc.), imaging examination results (such as coronary artery diameter measured by echocardiography, coronary artery Z-value, presence of coronary artery dilatation or coronary artery aneurysm, etc.), and treatment-related information (such as the duration and dosage of intravenous immunoglobulin administration, aspirin regimen, and glucocorticoid use, etc.).

[0047] In an extended implementation, the clinical data may further include genetic data (such as gene polymorphism information related to susceptibility to Kawasaki disease), proteomics data, metabolomics data, or wearable device monitoring data (such as heart rate variability, activity level, etc.) to expand the input dimensions of the risk assessment model.

[0048] It should be noted that the clinical data described in this application can be set according to the type and needs of the Kawasaki disease risk assessment model. For example, when the model is a gamma globulin-unresponsive Kawasaki disease risk assessment model, the clinical data may include coronary artery lesions, limb edema, albumin, gender, total bilirubin, systemic immune inflammation index, etc.; when the model is a Kawasaki disease coronary artery aneurysm risk assessment model, the clinical data may include gender, age, albumin, C-reactive protein, intravenous immunoglobulin resistance, duration of fever, platelet count, hemoglobin, baseline Z-score, erythrocyte sedimentation rate, etc.

[0049] Data collected at different time points are associated through a unified target object identifier (e.g., patient unique identifier, medical record number, or hash code), and a time series data structure is constructed in chronological order. The time series data structure can be stored using relational database tables, time series databases, or structured files (e.g., JSON, XML format), where each record corresponds to data collected at a specific time point and includes a timestamp (e.g., collection date, number of days since onset, or disease stage label) for that time point.

[0050] The "unified target object identifier" is used to uniquely identify the coded information of the same patient, ensuring that data from different time points and different sources can be accurately associated with the same target object. This identifier can be a patient ID in the hospital information system, or a de-identified identifier generated based on privacy protection requirements.

[0051] Step 2: Standardization of follow-up data

[0052] The time series data is standardized to generate a unified data structure adapted to the input requirements of the risk assessment model. This eliminates data differences from different data sources, acquisition devices, or time points, improving the consistency and comparability of subsequent calculations. The standardization process includes:

[0053] Data field mapping: Mapping raw data fields from different medical information systems (such as HIS, LIS, PACS systems) to a unified standard field name;

[0054] Data format standardization: unify the conversion of date formats, numeric formats, text encoding, etc.;

[0055] Unit or dimension consistency processing: Convert different units of measurement for laboratory indicators (such as mg / L and mg / dL of C-reactive protein) into standard units, and perform normalization or standardized scaling processing on continuous variables;

[0056] Missing data handling: For missing values, interpolation, mean imputation, multiple imputation, or model-based prediction imputation are used to handle them.

[0057] After standardization, time labels (or stage labels) are generated for the data at each time point, and a unified data structure is constructed for subsequent calculations.

[0058] Step 3: Constructing the initial risk assessment results

[0059] Based on data from the acute phase, an initial risk assessment result is obtained by using a risk assessment model to calculate the input data.

[0060] The risk assessment model described in this application (a risk assessment model used to construct initial risk assessment results) refers to a mathematical model or algorithmic model used for disease risk prediction or classification. This model can be any available risk assessment model that already exists in the prior art, is publicly available in the field of medical data analysis, or has been established or deployed. This includes, but is not limited to: statistical analysis models (such as Logistic Regression models, Cox proportional hazards models, and linear discriminant analysis models), machine learning models (such as Random Forest models, Support Vector Machine models, Gradient Boosting Tree models, and various neural network models), or scoring models (such as risk scoring systems based on clinical indicator weighting, improved scoring models, etc.).

[0061] In this invention, the risk assessment model can be a Kawasaki disease coronary artery injury prediction model established in previous studies by medical institutions, a publicly available prediction model, or a third-party model pre-trained based on historical sample data. The technical solution of this invention does not limit the specific algorithm form, training data source, or internal parameter settings of the model, as long as it can map the input clinical data to risk values ​​or risk levels. This invention uses multi-time-point follow-up data throughout the entire disease cycle to dynamically update and iteratively calculate the initial risk assessment results output by the model.

[0062] The initial risk assessment results are associated with and stored with corresponding time tags (such as "acute phase" or specific collection date), and the storage format includes database records, structured files or object instances in memory.

[0063] In a preferred embodiment, the risk assessment model supports parameter updates or input variable updates based on new follow-up data. For example, when using online learning or incremental learning mechanisms, the model can adaptively adjust model parameters based on new data; when using a static model, the model parameters can be kept unchanged by updating input variables (feature values), thus enabling recalculation of results.

[0064] In extended implementations, different types of prediction models can also be used for risk calculation. For example, when using recurrent neural networks or long short-term memory networks (LSTM), time series data can be directly used as input, and the model's time series modeling capabilities can be used to automatically learn the time dependencies of the data, thereby simplifying the design of explicit iterative update rules.

[0065] Step 4: Dynamic risk updates based on follow-up data

[0066] When new follow-up data is acquired, the new data and historical time series data are used as joint inputs to the risk update calculation unit. The risk update calculation unit iteratively calculates the existing risk assessment results based on preset update rules (which define the calculation logic for updating existing risk assessment results when new data is acquired. The core of these rules is to utilize the joint information of historical risk values ​​and new data, rather than simply replacing historical results). The preset update rules include at least:

[0067] (a) Update the model input variables based on the new follow-up data: Input the updated feature values ​​into the model using laboratory indicators, imaging results, etc. from the new follow-up data;

[0068] (b) Correcting or recalculating model calculations based on historical risk assessment results and new data: Using the risk assessment results from the previous time point as the reference benchmark for the current calculation, the risk value is corrected by combining new data. For example, a weighted fusion method can be used to fuse historical risk values ​​with independent risk values ​​calculated based on new data according to preset weights; or a difference correction method can be used to adjust the risk value upwards or downwards based on the magnitude of changes in key indicators at adjacent time points (such as changes in coronary Z-scores or the rate of decrease in C-reactive protein).

[0069] Through the aforementioned iterative updates (where "iterative update" refers to the generation of risk assessment results as a progressively iterative process, with the risk value calculation at each time point based on the results of the previous time point and the currently added data, forming a continuous risk assessment chain), updated risk assessment results are generated and output. Simultaneously, the updated results are associated with and stored with the corresponding time tags. Thus, the risk assessment result at each time point is calculated based on all available data at that time point and before, forming a continuously updated risk value sequence that progresses with the disease.

[0070] In an extended implementation, the risk update mechanism may also adopt a periodic update method, such as automatically triggering risk recalculation according to a preset follow-up plan (e.g., 2 weeks, 1 month, 3 months, 6 months, 1 year after onset); or adopt a hybrid mode, triggering additional updates when abnormal changes in key indicators are detected on the basis of fixed periodic updates, thereby realizing the update of risk assessment results.

[0071] Step 5: Risk Result Output and Trend Calculation

[0072] Based on the time series risk assessment results, the risk assessment values ​​of the target object at different time points are output. The risk assessment results include risk values ​​(such as probability values ​​between 0 and 1), risk levels (such as low risk, medium risk, high risk), or combinations thereof.

[0073] Furthermore, by calculating the difference, rate of change, or trend analysis of risk assessment results at adjacent time points, data on risk change trends can be obtained. For example, the absolute difference or relative rate of change between the current time point and the previous time point can be calculated, or a risk change curve can be fitted based on a risk value sequence from multiple time points to determine whether the risk is rising, falling, or stable.

[0074] In an extended implementation, the risk assessment results can also be visualized through risk curves, risk trend graphs, risk heatmaps, risk warning prompts (such as triggering a warning notification when the risk value exceeds a threshold or the rate of increase is too fast), or risk comparison reports, to assist clinical decision-making.

[0075] The "risk change trend analysis" refers to extracting the regular characteristics of risk changes over time based on risk assessment results at multiple time points using mathematical methods. The analysis may include: calculating the first difference between risk values ​​at adjacent time points to determine the direction of change; calculating the rate of change to assess the speed of change; and using moving averages or curve fitting methods to smooth short-term fluctuations and reveal long-term trends.

[0076] The second embodiment of this application discloses a dynamic risk update system for Kawasaki disease based on full disease cycle follow-up data. For example... Figure 2 As shown, the system includes:

[0077] The data acquisition module is used to acquire clinical data collected from target objects at different time points from data sources such as hospital information systems, laboratory information systems, image archiving, and communication systems, and to construct time-series data based on a unified target object identifier. This module supports real-time data interface integration or batch data import.

[0078] Data processing module: This module is used to standardize the raw data output by the data acquisition module, including data field mapping, format unification, unit consistency, and handling of missing values, and to generate a unified data structure and time label.

[0079] Risk assessment module: Based on the acute phase standardized data output by the data processing module, it calls the preset risk assessment model to calculate the initial risk assessment result and stores the result in association with the corresponding time tag.

[0080] Risk Update Module: When new follow-up data is acquired, it receives the newly added standardized data output by the data processing module, extracts the historical time series data of the target object and the risk assessment results of the previous time point from the historical database, performs iterative calculations based on preset update rules, generates updated risk assessment results, and stores them.

[0081] Risk Output Module: Used to extract risk assessment results of target objects at various time points from the database, output risk values ​​and / or risk levels, and calculate risk change trends, presenting them in the form of charts or reports.

[0082] In a specific implementation, the above modules are executed by computer programs on a server or medical data processing system, and are implemented by the processor calling program instructions stored in the storage medium. The modules interact with each other through standardized data interfaces (such as RESTful APIs, message queues, or database sharing). The system can be deployed on a local hospital server, a cloud-based medical platform, or an edge computing device.

[0083] In an extended implementation, the method and system can be implemented through a Clinical Decision Support System (CDSS), an Electronic Medical Record (EMR) plugin, standalone medical data analysis software, or a mobile healthcare application. In a distributed implementation, the data acquisition module can be deployed at the hospital, the risk assessment and update module can be deployed on a cloud server, and the risk output module can be presented through a web interface or a mobile app.

[0084] The technical solution and technical effects of this application will be described in detail below through specific embodiments.

[0085] Example 1: Dynamic update of Kawasaki disease risk based on full disease cycle follow-up data

[0086] (1) Risk assessment model

[0087] This embodiment uses the Logistic regression risk assessment model as an example model.

[0088] The model input variables include one or more of the following: age, gender, clinical symptom information, laboratory test indicators, and imaging test indicators.

[0089] This example uses C-reactive protein (CRP), platelet count (PLT), and coronary artery Z-score as sample input variables for illustration.

[0090] The model outputs a risk probability value between 0 and 1, which is used to represent the risk level of coronary artery injury in the target subject.

[0091] Example model is as follows: ;

[0092] in: .

[0093] The above model is only an example model in the embodiments of the present invention. The present invention is not limited to using this model, and random forest model, support vector machine model, neural network model, scoring model or other risk assessment model can also be used.

[0094] (2) Initial risk assessment in the acute phase

[0095] The clinical data obtained by target subject A during the acute phase are shown in Table 1:

[0096] Table 1

[0097] .

[0098] After inputting the above data into the risk assessment model:

[0099] .

[0100] Therefore, the initial risk assessment results are generated:

[0101] The risk value for T0 (acute phase) is 0.66.

[0102] The system associates the risk value of 0.66 with the time stamp "T0" and stores it to form an initial risk assessment record.

[0103] (3) Risk update at the first follow-up

[0104] Follow-up was conducted in the third week after the onset of illness, and the newly obtained clinical data are shown in Table 2:

[0105] Table 2

[0106] .

[0107] The system correlates the T0 acute phase data with the T1 follow-up phase data in chronological order to construct a time series feature matrix, and uses the time series feature matrix as input to the risk update calculation unit.

[0108] The risk value was then recalculated using the new data:

[0109] .

[0110] This embodiment adopts the following risk update rules:

[0111] .

[0112] in, This indicates the results of historical risk assessments. This indicates the risk assessment result recalculated based on the newly added follow-up data.

[0113] It should be noted that the above risk update rules are only used to illustrate the implementation of the present invention, and the present invention is not limited to this specific calculation formula. Any technical solution that can update the existing risk assessment results based on new follow-up data and generate updated risk assessment results can be used as an implementation of the present invention.

[0114] In this example: .

[0115] Therefore: T1 risk value = 0.68

[0116] The system associates the risk value of 0.68 with the time tag "T1" and stores it.

[0117] (4) Risk update during the second follow-up visit

[0118] Long-term follow-up was conducted in the third month after the onset of the disease, and the newly obtained data are shown in Table 3:

[0119] Table 3

[0120] .

[0121] The system further integrates the clinical indicators corresponding to the T0, T1, and T2 time points in chronological order.

[0122] The updated time series feature matrix is ​​generated and used as input to the risk update calculation unit.

[0123] Recalculate using the new data:

[0124] .

[0125] Continue to use the risk update rules:

[0126] .

[0127] Therefore: T2 risk value ≈ 0.72

[0128] The system associates and stores the risk value of 0.72 with the time stamp "T2".

[0129] (5) Calculation of risk change trend

[0130] The system generates the following continuous risk assessment records, as shown in Table 4:

[0131] Table 4

[0132] .

[0133] Further calculate the rate of change of risk:

[0134] .

[0135] Based on this, the system generates risk change trend information and outputs risk change curves.

[0136] (6) Comparison of technical effects

[0137] If the existing single static assessment method is used, and only acute phase data is used for a single risk assessment, the risk value will always remain at 0.66, which cannot reflect the subsequent disease progression.

[0138] By employing the method of this invention, the system can continuously update the risk assessment results as new follow-up data is acquired, and generate a risk value sequence: .

[0139] While the inflammatory marker CRP gradually decreased, the coronary Z-score continued to rise. Static assessment could not reflect the impact of changes in the coronary Z-score on risk status. The dynamic update mechanism can identify the increased risk resulting from the progression of coronary artery disease and continuously correct the risk assessment results, thereby achieving continuous recording and dynamic reflection of the risk status changes of the target subjects. Therefore, this invention can dynamically update existing risk assessment results based on new follow-up data, forming a continuous risk assessment record covering the entire disease cycle and improving the ability of risk assessment results to reflect the disease progression status.

[0140] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for dynamically updating the risk of Kawasaki disease based on full-disease-cycle follow-up data, characterized in that, include: Acquire Kawasaki disease-related clinical data of the target object at multiple time points, and construct a time series data structure based on the clinical data at the multiple time points. The clinical data includes data from the acute phase and data from the follow-up phase. The time series data structure associates the clinical data at different time points through a unified target object identifier. Based on the acute phase data, an initial risk assessment result is calculated using a risk assessment model, and the initial risk assessment result is stored in association with the corresponding time tag. When new follow-up phase data is acquired, the new follow-up phase data and historical time series data are used as joint inputs. The existing risk assessment results are iteratively updated based on preset update rules to generate updated risk assessment results. The updated risk assessment results are then associated with and stored with the corresponding time tags. Based on the risk assessment results of associated storage at each time point, conduct risk change trend analysis; The update rules include: updating model input variables based on newly added follow-up data; And the updated risk assessment results are calculated jointly based on historical risk assessment results and newly added follow-up data.

2. The method according to claim 1, characterized in that, The construction of the time series data structure includes: standardizing clinical data at different time points to generate a unified data structure, and configuring time labels for data at each time point.

3. The method according to claim 1, characterized in that, The risk assessment results include risk value, risk level, or a combination of risk value and risk level.

4. The method according to claim 1, characterized in that, The risk assessment model includes statistical analysis models, machine learning models, or scoring models.

5. The method according to claim 1, characterized in that, The risk assessment model can update parameters and / or input variables based on new follow-up data.

6. The method according to claim 1, characterized in that, The risk change trend analysis includes: calculating the difference, rate of change, or trend analysis of the risk assessment results at adjacent time points.

7. A dynamic risk update system for Kawasaki disease based on full disease cycle follow-up data, characterized in that, include: The data acquisition module is used to acquire Kawasaki disease-related clinical data of the target object at multiple time points and construct a time series data structure; The risk assessment module is used to calculate the initial risk assessment results based on data from the acute phase using a risk assessment model. The risk update module is used to iteratively update the existing risk assessment results based on preset update rules when new follow-up phase data is acquired, using the new follow-up phase data and historical time series data as joint inputs. The risk output module is used to output the risk assessment results at each time point and to perform risk change trend analysis based on the risk assessment results stored in association at each time point.

8. The system according to claim 7, characterized in that, It also includes a data processing module, which is used to standardize clinical data at different time points, generate a unified data structure, and configure time labels for data at each time point.

9. The system according to claim 7, characterized in that, The risk update module updates the input variables of the model based on the new follow-up phase data, and corrects or recalculates the risk assessment results based on the historical risk assessment results and the new follow-up phase data.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 5.

Images