Chronic disease early warning and risk stratification system and method

Abstract

This invention discloses a chronic disease early warning and risk stratification system and method, including a data acquisition and preprocessing module, a chronic disease indicator feature extraction module, a multi-indicator fusion early warning calculation module, a risk stratification determination module, a trend analysis and deterioration early warning module, a complication linkage early warning module, and an early warning intervention and data management module. This invention aggregates multi-dimensional data from clinical tests, personal health records, and lifestyle data through a multi-source data acquisition unit. After cleaning and standardization, data noise and dimensional differences are eliminated. Core feature factors are extracted by combining the disease characteristics of liver disease, diabetes, and hypertension, providing a high-quality and targeted data foundation for subsequent fusion early warning and risk stratification, thus improving the accuracy of early warning and stratification.

Description

Technical Field

[0001] This invention relates to the field of medical big data and chronic disease management technology, specifically to a chronic disease early warning and risk stratification system and method. Background Technology

[0002] Liver disease, diabetes, and hypertension are common chronic diseases in clinical practice, characterized by long course, high insidiousness, and numerous complications, making them important targets for prevention and control in the public health field. Current chronic disease management models in clinical and primary healthcare still rely primarily on traditional indicator recording and regular follow-up visits, which suffer from numerous technical bottlenecks and application pain points, making it difficult to meet the needs of precise and dynamic chronic disease prevention and control.

[0003] Existing chronic disease management systems generally suffer from limitations due to their reliance on single-indicator monitoring. They only record basic indicators such as blood glucose, blood pressure, and liver function independently, lacking integrated analysis and correlation calculations of multiple indicators, thus failing to reflect the overall trend of chronic disease development. Furthermore, there is no unified standardized system for risk stratification of liver disease, diabetes, and hypertension. Different medical institutions use significantly different stratification standards, leading to a lack of comparability in risk assessment results and making it difficult for primary care physicians to quickly and accurately identify high-risk individuals. The systems only provide static displays of indicators, lacking effective trend analysis logic, and cannot predict the deterioration trend of chronic diseases in advance, often resulting in intervention only after disease progression. These three chronic diseases exhibit significant complication linkage effects; for example, diabetes easily leads to hypertension, and liver disease combined with diabetes exacerbates liver function damage. However, existing systems lack a linkage early warning mechanism for complications, easily causing delayed detection and treatment of complications. Simultaneously, existing systems only complete data recording and simple reminders, lacking standardized automatic intervention reminder mechanisms. Early warning information is disconnected from clinical intervention measures, and there is a lack of full lifecycle management of chronic disease data, failing to provide data support for subsequent disease analysis and model optimization.

[0004] In addition, existing chronic disease early warning systems often suffer from the problem of "emphasizing early warning but neglecting stratification". The early warning results lack a refined risk level classification, resulting in unreasonable allocation of medical resources, insufficient intervention resources for high-risk patients, and excessively high management costs for low-risk patients. Summary of the Invention

[0005] The purpose of this invention is to provide a chronic disease early warning and risk stratification system and method to solve the problems existing in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a chronic disease early warning and risk stratification system, comprising a data collection and preprocessing module, a chronic disease indicator feature extraction module, a multi-indicator fusion early warning calculation module, a risk stratification determination module, a trend analysis and deterioration early warning module, a complication linkage early warning module, and an early warning intervention and data management module;

[0007] The data acquisition and preprocessing module is used for the unified acquisition, cleaning, and standardized processing of multi-source chronic disease data; the chronic disease indicator feature extraction module is used to extract core monitoring indicators and feature factors for liver disease, diabetes, and hypertension; the multi-indicator fusion early warning calculation module is used to construct a multi-indicator weighted fusion early warning model and calculate a comprehensive early warning index for chronic diseases; the risk stratification judgment module is used to construct a risk stratification scoring model and realize four-level stratification of chronic disease risk; the trend analysis and deterioration early warning module is used to determine the development trend of the disease and identify the signal of disease deterioration; the complication linkage early warning module is used to construct a complication association rule base and realize the linkage early warning of primary diseases and complications; the early warning intervention and data management module is used to push personalized intervention reminders and complete the full-process data management of chronic diseases.

[0008] Furthermore, the data acquisition and preprocessing module includes a multi-source data acquisition unit, a data cleaning unit, and a data standardization unit;

[0009] The chronic disease-related data collected by the multi-source data acquisition unit are divided into three categories: clinical test data, personal health record data, and lifestyle data. The data cleaning unit uses interpolation, manual review, deduplication, and format conversion to process missing values, outliers, duplicate values, and non-uniform format data. The data standardization unit uses extreme value standardization, dummy variable coding, and rank coding to normalize and quantify the cleaned data.

[0010] Furthermore, the chronic disease indicator feature extraction module includes a disease indicator classification unit, a core indicator screening unit, and a feature factor quantification unit;

[0011] The disease indicator classification unit divides the indicators into two categories: exclusive core indicators and cross-disease related indicators. The core indicator screening unit combines clinical treatment guidelines, Pearson correlation coefficient method, and random forest algorithm to screen core indicators. The feature factor quantification unit transforms the core indicators into disease feature factors with values ​​in the range [0,1], where a larger value indicates a higher risk of the corresponding disease.

[0012] Furthermore, the multi-indicator fusion early warning calculation module includes an indicator weight determination unit, a fusion early warning index calculation unit, and a basic early warning judgment unit; the indicator weight determination unit uses a combination of the analytic hierarchy process (AHP) and clinical expert scoring to determine the feature factor weight coefficients, with the sum of the weight values ​​being 1; the fusion early warning index calculation unit uses a formula... Calculate a comprehensive early warning index for a single chronic disease, where, It is a comprehensive early warning index for a single chronic disease, with a value range of [0,1]. This represents the number of characteristic factors in the set of characteristic factors for this chronic disease. For the first The weight coefficients of each feature factor satisfy the following conditions: ; For the first The quantized values ​​of each feature factor range from [0,1]. Calculate the overall comprehensive early warning index, where , , These are the disease weight coefficients for liver disease, diabetes, and hypertension, respectively. The basic early warning judgment unit has a preset threshold of 0.5, based on... The value of is used to achieve the three-level judgment of basic early warning.

[0013] Furthermore, the risk stratification determination module includes a risk stratification scoring calculation unit, a risk level determination unit, and a stratification result output unit;

[0014] The risk stratification scoring calculation unit uses the formula Calculate the comprehensive risk score for chronic diseases, among which, The comprehensive risk score for a single chronic disease ranges from [0,2]. This is the comprehensive early warning index for the chronic disease, with a value range of [0,1]. The trend influence coefficient is a fixed constant with a value of 1, reflecting the amplifying effect of the disease trend on the risk score. This is the disease trend coefficient for the chronic disease, with a value range of [0,1]. It is calculated by the trend analysis and deterioration early warning module and transmitted to this unit. A higher value indicates a more pronounced trend of disease progression. This indicates that the disease trend is stable; the risk level determination unit is based on... The value of the score divides the risk of chronic diseases into four levels: low, medium, high, and very high. At the same time, the score is fine-tuned according to the specific characteristics of the three major chronic diseases, and the score after fine-tuning does not exceed 2.0. The hierarchical result output unit visualizes the risk level of a single chronic disease and the overall chronic disease risk level and transmits it to the other modules.

[0015] Furthermore, the trend analysis and deterioration early warning module includes a historical index fitting unit, a disease trend coefficient calculation unit, a deterioration signal identification unit, and a deterioration early warning issuance unit;

[0016] The historical index fitting unit constructs a time series trend fitting curve for the characteristic factors using multinomial fitting and moving average methods; the disease trend coefficient calculation unit calculates the disease trend coefficient using the slope and weight coefficient of the characteristic factors. , The value range is [0,1] and is transmitted to the risk stratification judgment module; the deterioration signal identification unit sets the deterioration signal judgment rules based on the characteristic factor trend, the change of the warning index, the trend coefficient, and the comorbidity; the deterioration warning issuance unit determines the three-level priority of the deterioration warning according to the risk level.

[0017] Furthermore, the complication linkage early warning module includes a complication association rule base unit, a comorbidity risk calculation unit, a complication early warning determination unit, and a linkage early warning issuance unit;

[0018] The complication association rule base unit constructs a rule base for the interaction and complication early warning of the three major chronic diseases based on clinical practice guidelines, and the rule base supports dynamic updates; the comorbidity risk calculation unit calculates the comorbidity risk value using a weighted summation method. , The value range is [0,1]; the complication warning determination unit presets a threshold of 0.6 and determines the complication warning based on the risk level; the linkage warning issuing unit issues a linkage warning that includes the primary disease and complication type, and integrates it with other warnings in terms of priority.

[0019] Furthermore, the early warning intervention and data management module includes an intervention reminder scheme library unit, a personalized intervention push unit, a full-process data storage unit, and a data traceability and model optimization unit;

[0020] The intervention reminder scheme library unit constructs four categories of standardized intervention reminder scheme libraries that match risk levels and warning types, and the scheme libraries support dynamic updates; the personalized intervention push unit combines patient data to generate personalized intervention reminder information and realizes multi-terminal push, supporting manual adjustment of intervention schemes; the full-process data storage unit uses a distributed database to store full-process chronic disease data in a structured and de-identified manner; the data traceability and model optimization unit supports multi-dimensional data traceability and feeds data back to relevant modules to realize model iterative optimization.

[0021] The stratification method for a chronic disease early warning and risk stratification system includes the following steps:

[0022] Step 1: Multi-source data acquisition and preprocessing. The data acquisition and preprocessing module collects multi-source data related to the three major chronic diseases, completes data cleaning and standardization, and forms a standardized chronic disease dataset.

[0023] Step 2: Chronic disease indicator feature extraction. The chronic disease indicator feature extraction module classifies indicators, selects core indicators, and quantifies feature factors in the standardized dataset to form three major chronic disease feature factor sets.

[0024] Step 3: Multi-indicator fusion early warning calculation. The feature factor weights are determined through the multi-indicator fusion early warning calculation module, and the single and overall chronic disease comprehensive early warning index is calculated to complete the basic early warning level three judgment.

[0025] Step 4: Chronic disease risk stratification determination. By combining the comprehensive early warning index and the disease trend coefficient through the risk stratification determination module, a comprehensive chronic disease risk score is calculated to complete the four-level stratification of chronic disease risk.

[0026] Step 5: Trend Analysis and Deterioration Warning. By fitting historical indicator trends through the trend analysis and deterioration warning module, the disease trend coefficient is calculated, deterioration signals are identified, and deterioration warnings are issued according to priority.

[0027] Step 6: Complication linkage early warning. By combining the complication linkage early warning module with the complication association rule base, the comorbidity risk value is calculated, and a complication linkage early warning is issued.

[0028] Step 7: Early warning intervention and data management. Personalized intervention reminders are pushed through the early warning intervention and data management module. Data storage and tracing of the entire chronic disease process are completed, and data feedback is used to optimize the model.

[0029] Furthermore, the comorbidity risk value mentioned in step 6 is obtained by weighted summation of the patient's chronic disease warning status, risk level, and core complication index values. The model optimization mentioned in step 7 uses machine learning algorithms to iteratively update the warning model weight coefficients and risk stratification scoring standards. The disease trend coefficient mentioned in step 9 is obtained by normalizing the slope of the feature factor trend fitting curve and then weighted summation of the feature factor weight coefficients.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] This invention aggregates multi-dimensional data from clinical tests, personal health records, and lifestyles through a multi-source data acquisition unit. After cleaning and standardization, data noise and dimensional differences are eliminated. Core feature factors are extracted by combining the disease characteristics of liver disease, diabetes, and hypertension, providing a high-quality and targeted data foundation for subsequent fusion early warning and risk stratification, thereby improving the accuracy of early warning and stratification.

[0032] This invention, through multi-indicator fusion early warning calculation and risk stratification scoring model, combined with the real-time early warning status and disease development trend of chronic diseases, has formulated a four-level risk stratification standard applicable to liver disease, diabetes, and hypertension. It has also made fine adjustments for the specific characteristics of each disease, achieving standardization and refinement of risk classification for the three major chronic diseases. This provides a scientific basis for the rational allocation of medical resources and facilitates the rapid identification of high-risk groups by primary healthcare institutions.

[0033] This invention, through time-series fitting and trend coefficient calculation of patients' historical indicator data, can accurately determine the development trend of the disease, identify the signs of disease deterioration in advance, and set early warning priorities according to risk level, realizing the transformation from "post-intervention" to "early prediction", thus buying time for early intervention of chronic diseases and effectively delaying disease progression.

[0034] This invention constructs a complication association rule base based on the clinical correlation patterns of three major chronic diseases, realizing the collaborative monitoring of primary diseases and complications, as well as the linkage early warning of multiple comorbidities. It can promptly identify the comorbidity risk of patients with chronic diseases, provide early warning of clinically common complications, effectively reduce the incidence of complications, and improve the chronic disease management effect for patients with multiple comorbidities.

[0035] This invention constructs a standardized intervention reminder program library that matches risk levels and warning types. It generates personalized intervention reminder information by combining patients' personal data, enabling multi-terminal push and manual adjustment. This allows warning information to be quickly transformed into clinical and lifestyle intervention measures, forming a closed loop of "warning-intervention" and improving the execution and efficiency of chronic disease management. Attached Figure Description

[0036] Figure 1 This is a system module diagram of the present invention;

[0037] Figure 2 This is a schematic diagram of the data acquisition and preprocessing module of the present invention;

[0038] Figure 3 This is a schematic diagram of the chronic disease indicator feature extraction module of the present invention;

[0039] Figure 4 This is a schematic diagram of the multi-indicator fusion early warning calculation module of the present invention;

[0040] Figure 5 This is a schematic diagram of the risk stratification determination module of the present invention;

[0041] Figure 6 This is a flowchart of the method of the present invention. Detailed Implementation

[0042] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0043] Please see Figure 1-6This invention provides a chronic disease early warning and risk stratification system, including a data collection and preprocessing module, a chronic disease indicator feature extraction module, a multi-indicator fusion early warning calculation module, a risk stratification judgment module, a trend analysis and deterioration early warning module, a complication linkage early warning module, and an early warning intervention and data management module.

[0044] The data acquisition and preprocessing module is used to collect multi-source chronic disease data in a unified manner and to clean and standardize the data, providing a high-quality data foundation for subsequent analysis and calculation. The chronic disease indicator feature extraction module is used to extract core monitoring indicators and feature factors for liver disease, diabetes, and hypertension, respectively, to achieve indicator classification and feature quantification. The multi-indicator fusion early warning calculation module is based on a weighted fusion algorithm to construct a multi-indicator fusion early warning model for the three major chronic diseases, calculate the comprehensive early warning index for chronic diseases, and achieve basic early warning judgment. The risk stratification judgment module is based on the comprehensive early warning index and disease feature coefficients to construct a risk stratification scoring model, formulate a unified risk classification standard, and achieve refined risk stratification for liver disease, diabetes, and hypertension. The trend analysis and deterioration early warning module determines the trend of indicator changes, identifies signals of disease deterioration, and issues deterioration early warnings through fitting analysis of historical indicator data. The complication linkage early warning module is based on the complication association rules of the three major chronic diseases to construct a complication linkage early warning model, achieving collaborative monitoring and linkage early warning of primary diseases and complications. The early warning intervention and data management module is used to push standardized intervention reminder information based on early warning results and risk levels, and to complete the storage, traceability, and updating of chronic disease data throughout the entire process, while providing data support for model optimization.

[0045] The data acquisition and preprocessing module is the system's data input layer, specifically including a multi-source data acquisition unit, a data cleaning unit, and a data standardization unit. These three units sequentially complete the data acquisition, purification, and normalization processes to ensure that the data input to subsequent modules has integrity, accuracy, and consistency.

[0046] Multi-source data acquisition unit: This unit collects multi-dimensional chronic disease-related data from patients with liver disease, diabetes, and hypertension. The collected data types are divided into three main categories: clinical testing data, personal health record data, and lifestyle data, covering all dimensions of indicators for the prevention and control of these three chronic diseases. Clinical testing data includes liver function indicators, blood glucose levels, blood pressure indicators, and related indicators such as blood lipids, kidney function, and complete blood count. Personal health record data includes the patient's gender, age, height, weight, date of diagnosis of chronic disease, disease course, past medical history, family medical history, and medication history. Lifestyle data includes the patient's dietary structure, exercise frequency, smoking and drinking history, and sleep patterns. This unit supports batch import of structured data and manual entry of unstructured data, achieving unified aggregation and categorized storage of multi-source data.

[0047] The data cleaning unit is used to clean the collected raw data, eliminating noise and outliers to ensure data integrity. Specific processing methods include: missing value handling (interpolation is used to supplement a small number of randomly missing data points, and large amounts of consecutively missing invalid data are marked and removed); outlier handling (based on clinical medical standards and normal reference ranges for indicators, outlier data exceeding reasonable ranges is identified and removed, and suspected outlier data is marked and returned for manual review); duplicate value handling (duplicate data collected from the same patient at the same testing time is deduplicated, retaining only the unique valid data); and format standardization (standardizing and converting time and numerical data of different formats to achieve data format uniformity).

[0048] Data standardization unit: Used to normalize the cleaned and valid data, eliminating dimensional differences between different indicators and making them comparable and synergistic. For quantitative data of clinical test indicators, extreme value standardization is used to map indicator values ​​to the [0,1] interval; for qualitative data in personal health records and lifestyles, dummy variable coding is used for quantification, such as classifying smoking history into three levels: "none," "occasional," and "long-term," assigned values ​​of 0, 0.5, and 1 respectively; for ordered classification data such as medication history and disease course, hierarchical coding is used for quantification. All standardized data will be classified and stored according to three disease categories: liver disease, diabetes, and hypertension, forming a standardized chronic disease dataset, which will then be transmitted to the chronic disease indicator feature extraction module.

[0049] The chronic disease indicator feature extraction module is based on the standardized chronic disease dataset output by the data acquisition and preprocessing module. Specifically, it includes a disease indicator classification unit, a core indicator screening unit, and a feature factor quantification unit. Targeting the pathological characteristics and clinical prevention and control priorities of liver disease, diabetes, and hypertension, it completes the classification of indicators, screening of core indicators, and quantification of feature factors, providing a targeted feature indicator set for subsequent multi-indicator fusion calculation.

[0050] Disease Indicator Classification Unit: Based on the specific characteristics of liver disease, diabetes, and hypertension, the indicators in the standardized dataset are divided into two main categories: specific core indicators and cross-disease-related indicators. Specific core indicators are those unique to the diagnosis and monitoring of each chronic disease. For example, specific core indicators for liver disease include liver function indicators such as ALT, AST, TBIL, and albumin (ALB); specific core indicators for diabetes include blood glucose indicators such as FPG, 2hPG, and HbA1c; and specific core indicators for hypertension include blood pressure indicators such as SBP, DBP, and MAP. Cross-disease-related indicators are common indicators that affect all three chronic diseases, including blood lipids (total cholesterol TC, triglycerides TG), body mass index (BMI), age, disease duration, and family history. After classifying the indicators in this unit, a classification system for the indicators of the three chronic diseases is constructed, providing a foundation for the selection of core indicators.

[0051] Core Indicator Screening Unit: Based on clinical medical evidence and statistical analysis methods, core indicators with significant impact on chronic disease early warning and risk stratification are screened from the indicator classification system, while secondary indicators without statistical significance are eliminated to reduce the complexity of subsequent calculations. The specific screening method is as follows: Clinical core indicator sets for each chronic disease are determined by combining clinical treatment guidelines for liver disease, diabetes, and hypertension; the Pearson correlation coefficient method is used to analyze the correlation between indicators, eliminating highly correlated redundant indicators; the random forest algorithm is used to calculate the importance score of each indicator for chronic disease assessment, and indicators with importance scores higher than a preset threshold are selected, ultimately forming the core indicator sets for each of the three major chronic diseases and the cross-disease-related core indicator sets.

[0052] Feature Factor Quantification Unit: This unit transforms the screened core indicators into disease feature factors that can participate in fusion calculations. Based on the disease progression patterns of each chronic disease, it quantifies the core indicators. For example, the HbA1c index for diabetes is divided into different intervals according to clinical standards, corresponding to different blood glucose control feature factors; the SBP and DBP of hypertension are combined to calculate the blood pressure fluctuation feature factor; the ALT / AST ratio of liver disease is used as a liver function impairment feature factor; and BMI, TC, and TG from cross-disease-related indicators are fused into a metabolic abnormality feature factor. The values ​​of each feature factor are mapped to the [0,1] interval, with larger values ​​indicating higher disease risk. This ultimately forms the feature factor set for the three major chronic diseases, which is then transmitted to the multi-indicator fusion early warning calculation module.

[0053] The multi-indicator fusion early warning calculation module is the core early warning calculation layer of the system. Specifically, it includes an indicator weight determination unit, a fusion early warning index calculation unit, and a basic early warning judgment unit. Based on the feature factor set output by the chronic disease indicator feature extraction module, a multi-indicator weighted fusion early warning model is constructed to calculate the chronic disease comprehensive early warning index and realize the basic early warning judgment of liver disease, diabetes, and hypertension. This module introduces Formula 1 as the calculation formula for the chronic disease comprehensive early warning index and embeds it into the fusion early warning index calculation unit.

[0054] The indicator weight determination unit employs a combination of the Analytic Hierarchy Process (AHP) and clinical expert scoring to determine the weight coefficients of each feature factor in the feature factor set. These weight coefficients reflect the degree of influence of each feature factor on chronic disease early warning, with the sum of the weight values ​​being 1. The specific steps are as follows: A hierarchical analysis model is constructed, with chronic disease early warning as the target layer and the feature factors for liver disease, diabetes, and hypertension, along with cross-disease-related feature factors, as the criteria layer. Clinical experts in the field of chronic diseases are invited to conduct pairwise comparisons and scores of the importance of each feature factor, constructing a judgment matrix. A consistency test is performed on the judgment matrix; if the test passes, the weight coefficients of each feature factor are calculated; if the test fails, the scores are readjusted and the judgment matrix is ​​constructed again. For different chronic diseases, the weight coefficients of their specific feature factors and cross-disease-related feature factors are determined separately. For example, the weight of the blood glucose control feature factor for diabetes is higher than that of the cross-disease-related metabolic abnormality feature factor, and the weight of the blood pressure fluctuation feature factor for hypertension is the highest among all feature factors.

[0055] The integrated early warning index calculation unit: Based on the indicator weights, the weight coefficients of the unit output are determined, and a multi-indicator weighted integrated early warning model is constructed. The comprehensive early warning index of chronic diseases such as liver disease, diabetes, and hypertension is calculated using Formula 1. And the overall comprehensive early warning index for the three major chronic diseases. Formula 1:

[0056]

[0057] in, It is a comprehensive early warning index for a single chronic disease (liver disease / diabetes / hypertension), with a value range of [0,1]. This represents the number of characteristic factors in the set of characteristic factors for this chronic disease. For the first The weight coefficients of each feature factor satisfy the following conditions: ; For the first The quantified values ​​of each characteristic factor range from [0,1]. Comprehensive early warning index for the three major chronic diseases. ,in , , The weighting coefficients for liver disease, diabetes, and hypertension are dynamically adjusted based on the patient's diagnosis and severity of chronic diseases. If a patient is diagnosed with only a single chronic disease, the corresponding disease weighting coefficient is 1, and the others are 0. If the patient has multiple comorbidities, the weighting coefficients are adjusted according to the course and severity of each chronic disease to meet the following requirements. The integrated early warning index calculation unit will calculate the obtained... The data is transmitted to the basic early warning judgment unit and the risk stratification judgment module.

[0058] Basic Early Warning Judgment Unit: Based on the value of the comprehensive early warning index for chronic diseases, a basic early warning threshold is set to achieve basic early warning judgment for the three major chronic diseases. The preset basic early warning threshold is 0.5. When the comprehensive early warning index E of a single chronic disease is ≤0.5, it is determined that there is no basic early warning signal for that chronic disease; when… When the value of E is greater than 0.5, it is determined that there is a basic warning signal for the chronic disease. Furthermore, based on the value of E, the basic warning is classified as a mild warning (0.5 < 0.5). ≤0.7), moderate warning (0.7 < ≤0.9), severe warning ( The basic early warning judgment results will be transmitted in real time to the risk stratification judgment module, the trend analysis and deterioration early warning module, and synchronized to the early warning intervention and data management module (>0.9).

[0059] The risk stratification determination module is the core layer of the system's risk classification. Specifically, it includes a risk stratification scoring calculation unit, a risk level determination unit, and a stratification result output unit. Based on the chronic disease comprehensive early warning index output by the multi-indicator fusion early warning calculation module, combined with the disease trend coefficient, a risk stratification scoring model is constructed to formulate unified risk classification standards for liver disease, diabetes, and hypertension, thereby achieving refined risk stratification. This module introduces Formula 2 as the calculation formula for the chronic disease comprehensive risk score and embeds it into the risk stratification scoring calculation unit.

[0060] Risk Stratification Scoring Calculation Unit: A comprehensive chronic disease risk scoring model is constructed. Formula 2 is used to calculate the comprehensive chronic disease risk score S for liver disease, diabetes, and hypertension. This comprehensive risk score is the core basis for risk level determination, integrating real-time early warning status and disease progression trends. Formula 2:

[0061]

[0062] in, The comprehensive risk score for a single chronic disease ranges from [0,2]. This is the comprehensive early warning index for the chronic disease, with a value range of [0,1]. The trend influence coefficient is a fixed constant with a value of 1, reflecting the amplifying effect of the disease trend on the risk score. This is the disease trend coefficient for the chronic disease, with a value range of [0,1]. It is calculated by the trend analysis and deterioration early warning module and transmitted to this unit. A higher value indicates a more pronounced trend of disease progression. This indicates a stable disease trend. For patients with multiple comorbidities, after calculating the comprehensive risk score of each individual chronic disease, the maximum value is taken as the patient's overall chronic disease comprehensive risk score, reflecting the patient's overall chronic disease risk level.

[0063] Risk level assessment unit: Establish a unified four-level risk stratification standard applicable to liver disease, diabetes, and hypertension, based on the comprehensive risk score for chronic diseases. The value of is used to classify chronic disease risk into four levels: low risk, medium risk, high risk, and very high risk, thus standardizing the risk classification of the three major chronic diseases. The specific classification standard is as follows: when... A value ≤0.5 is considered low risk, indicating that the patient's chronic disease is well controlled and there is no significant risk; when 0.5 < A value ≤1.0 indicates a medium risk, suggesting that the patient has a mild abnormality in their chronic disease and requires closer monitoring of indicators; a value <1.0 indicates a medium risk. A value ≤1.5 is considered high-risk, indicating significant fluctuations in the patient's chronic disease condition, requiring timely adjustment of the intervention plan; when A score >1.5 is considered extremely high risk, indicating a severe chronic disease with a high probability of exacerbation or complications, requiring intensive clinical intervention. Considering the specific characteristics of liver disease, diabetes, and hypertension, the risk grading criteria are fine-tuned, such as the quantification of liver function impairment characteristic factors in liver disease patients. When the comprehensive risk score is greater than 0.8, the overall risk score is... Increased by 0.2; Quantitative value of HbA1c characteristic factor in diabetic patients When the comprehensive risk score is greater than 0.8, the overall risk score is... Increased by 0.2; Quantitative value of blood pressure fluctuation characteristic factor in hypertensive patients When the comprehensive risk score is greater than 0.8, the overall risk score is... Increase by 0.2, with the adjusted score not exceeding 2.0.

[0064] The stratified results output unit visualizes the risk level of each individual chronic disease and the patient's overall chronic disease risk level. At the same time, it transmits the risk stratification results and comprehensive risk scores to the trend analysis and deterioration warning module, the complication linkage warning module, and the warning intervention and data management module, providing a tiered basis for subsequent deterioration warnings, complication warnings, and intervention reminders.

[0065] The trend analysis and deterioration early warning module is the system's dynamic prediction layer. It specifically includes a historical indicator fitting unit, a disease trend coefficient calculation unit, a deterioration signal identification unit, and a deterioration early warning issuance unit. Through time-series analysis of the patient's historical characteristic factor data, it determines the disease development trend and calculates the disease trend coefficient. It can also identify signs of disease deterioration, issue timely warnings of worsening conditions, and provide a basis for early intervention in chronic diseases.

[0066] Historical Indicator Fitting Unit: This unit retrieves historical characteristic factor data of patients stored in the early warning intervention and data management module, constructs characteristic factor change curves according to time series, and supports selection of time dimensions by week, month, quarter, and year, covering the patient's chronic disease course. A combination of multinomial fitting and moving average methods is used to fit the time series data of characteristic factors, eliminating the influence of short-term fluctuations and obtaining trend fitting curves for each characteristic factor, reflecting the long-term change trend of the characteristic factor. For example, a continuously rising FPG characteristic factor fitting curve in diabetic patients indicates a worsening trend in blood glucose control; a continuously decreasing ALT characteristic factor fitting curve in liver disease patients indicates a easing trend in liver function damage.

[0067] Disease trend coefficient calculation unit: Based on the trend fitting curve of the feature factors, the change rate of each feature factor is calculated, and combined with the weight coefficient of each feature factor, the disease trend coefficient of a single chronic disease is calculated. and will The value is transmitted to the risk stratification scoring calculation unit of the risk stratification determination module. Disease trend coefficient. The calculation logic is as follows: for the trend fitting curve of each feature factor, calculate its slope. , A value greater than 0 indicates that the characteristic factor is on an upward trend (increased risk). <0 indicates that the characteristic factor is showing a downward trend (risk reduction). This indicates that the trend of the characteristic factor is stable; the slope of each characteristic factor is... Normalized to the [0,1] interval, the trend quantification values ​​of each feature factor are obtained. The disease trend coefficient was calculated using the weighted summation method. ,in These are the weighting coefficients for each feature factor, consistent with the weighting coefficients in the multi-indicator fusion early warning calculation module. The higher the value, the more obvious the worsening trend of the patient's chronic disease.

[0068] Deterioration signal identification unit: Defines the criteria for determining disease deterioration, based on the characteristic factor trend fitting curve and the disease trend coefficient. This identifies signs of worsening chronic diseases. Specific criteria include: the slope of the trend fitting curve for a single core feature factor. The trend quantification value is positive for three consecutive time periods and after normalization. >0.7; Chronic Disease Comprehensive Early Warning Index The disease trend coefficient increased for two consecutive time periods, with an increase of ≥0.1; >0.7, and chronic disease comprehensive risk score >1.0; the disease trend coefficient of two or more chronic diseases in patients with multiple comorbidities. >0.6. Meeting any of the above rules is considered a sign of worsening condition.

[0069] The deterioration warning unit: When the deterioration signal recognition unit detects a sign of worsening condition, it immediately issues a deterioration warning. This warning is linked to the basic warning and risk level, with priority determined based on the risk level: deterioration warnings for low-risk and medium-risk patients have general priority, those for high-risk patients have higher priority, and those for very high-risk patients have the highest priority. The deterioration warning information includes the type of chronic disease being warned, deterioration characteristic factors, trend changes, and risk level, and is transmitted to the complication linkage warning module and the warning intervention and data management module.

[0070] The complication linkage early warning module is the system's collaborative early warning layer. Specifically, it includes a complication association rule base unit, a comorbidity risk calculation unit, a complication early warning judgment unit, and a linkage early warning issuance unit. Based on the clinical complication association patterns of liver disease, diabetes, and hypertension, it constructs a complication association rule base to achieve collaborative monitoring of primary diseases and complications, as well as linkage early warning for multiple comorbidities, thus overcoming the limitations of existing systems that only provide early warning for single chronic diseases.

[0071] Complication Association Rule Base Unit: This unit constructs a complication association rule base based on clinical practice guidelines and evidence-based medicine. It outlines the interrelationships between three major chronic diseases: liver disease, diabetes, and hypertension, as well as early warning rules for common complications and associated diseases of each chronic disease. The rule base supports dynamic updates based on clinical research progress. Core association rules include: a bidirectional association rule between diabetes and hypertension (poor glycemic control in diabetic patients increases the risk of hypertension, and blood pressure fluctuations in hypertensive patients exacerbate abnormal glycemic levels in diabetic patients); an association rule between diabetes and liver disease (diabetic metabolic disorders worsen liver function damage, and patients with viral hepatitis are prone to developing abnormal glucose metabolism); an association rule between hypertension and liver disease (hepatic vascular lesions caused by hypertension worsen liver function damage, and patients with cirrhosis are prone to portal hypertension); and early warning rules for complications of various chronic diseases, such as diabetes combined with hyperlipidemia easily leading to diabetic nephropathy and diabetic retinopathy, hypertension combined with high BMI easily leading to cardiovascular and cerebrovascular diseases, and severe liver function damage in liver disease easily leading to hepatic encephalopathy and ascites. The rule base clearly defines the core indicator thresholds and risk triggering conditions for the occurrence of each complication.

[0072] Comorbidity Risk Calculation Unit: For patients diagnosed with a single chronic disease, this unit calculates the comorbidity risk value for developing other chronic diseases or complications based on a comorbidity association rule base. Comorbidity risk value The value ranges from [0,1], with higher values ​​indicating a higher risk of comorbidities or complications. The calculation logic for the comorbidity risk value is based on the patient's current chronic disease comprehensive early warning index. Comprehensive risk score Comorbidity risk values ​​are calculated using a weighted summation method, combining the quantified values ​​of core indicators in the comorbidity association rules; for example, the comorbidity risk value of diabetic patients. It integrates quantitative values ​​and weighting coefficients of glycemic control characteristic factors, blood pressure fluctuation characteristic factors, and metabolic abnormality characteristic factors. For patients with multiple comorbidities, it calculates the risk values ​​of mutual influence between various chronic diseases, reflecting the risk of disease progression under comorbid conditions.

[0073] Complication early warning judgment unit: based on comorbidity risk value The system uses risk triggering conditions in the comorbidity rule base to determine comorbidity warnings. The preset comorbidity risk threshold is 0.6. When the value is >0.6, a complication warning is issued based on the patient's chronic disease risk level: low-risk and medium-risk patients. A complication warning is issued when the criterion is greater than 0.7, indicating high-risk or very high-risk patients. A complication warning is issued when the score is greater than 0.6. Simultaneously, when the patient's single chronic disease comprehensive risk score... When the risk level is greater than 1.5 (extremely high risk), an early warning for complications of the chronic disease will be automatically triggered, with a focus on monitoring common clinical complications.

[0074] Linked Early Warning Issuing Unit: When the complication early warning determination unit determines that a complication early warning needs to be issued, a linked early warning for complications is issued. The linked early warning includes the type of primary disease, the type of complication to be warned, the comorbid risk value, the core triggering indicators, and the direction of intervention recommendations. The linked early warning is integrated with the deterioration early warning and the basic early warning, with the complication early warning for extremely high-risk patients having the highest priority. It is synchronously transmitted to the early warning intervention and data management module to achieve unified push of early warning information.

[0075] The early warning intervention and data management module is the system's output and data storage layer. Specifically, it includes an intervention reminder plan library unit, a personalized intervention push unit, a full-process data storage unit, and a data traceability and model optimization unit. It completes the visualization of early warning information, pushes standardized intervention reminders, and stores, traces, and updates the full-process data of chronic diseases, realizing the closed-loop management of the system. At the same time, it provides data support for the continuous optimization of the early warning model and risk stratification model.

[0076] The intervention reminder protocol library unit constructs a standardized intervention reminder protocol library that matches the risk levels and warning types of liver disease, diabetes, and hypertension. The library is divided into four categories: basic monitoring intervention, lifestyle intervention, medication adjustment intervention, and clinical visit intervention. Each intervention protocol is precisely matched with low, medium, high, and very high risk levels, as well as basic warning, deterioration warning, and complication-related warning. The library content is developed by clinical experts in the field of chronic diseases and supports dynamic updates based on clinical treatment guidelines. Specifically, the basic monitoring intervention protocol specifies the monitoring frequency for indicators for patients at different risk levels; the lifestyle intervention protocol provides personalized recommendations on diet, exercise, sleep, smoking cessation, and alcohol cessation; the medication adjustment intervention protocol provides clinicians with a reference direction for medication adjustments but does not involve specific medication guidance; and the clinical visit intervention protocol specifies the timing and recommended departments for patients with different warning types and risk levels.

[0077] Personalized Intervention Push Unit: Based on the basic early warning, deterioration warning, and complication-related early warning results received from the early warning intervention and data management module, as well as the risk stratification results, this unit retrieves matching intervention plans from the intervention reminder plan library. Combined with the patient's personal health record and lifestyle data, it generates personalized intervention reminder information. The intervention reminder information can be pushed through multiple platforms, including system platform visualization, SMS reminders, and WeChat official account reminders. The target audience includes patients and their corresponding attending physicians / primary care managers. Simultaneously, this unit allows physicians to manually adjust the intervention reminder plan according to the patient's actual situation, generating a personalized intervention plan.

[0078] The end-to-end data storage unit provides structured storage for the entire process of early warning and risk stratification for liver disease, diabetes, and hypertension. Stored data includes raw collected data, preprocessed standardized data, feature factor sets, comprehensive early warning indices, comprehensive risk scores, risk levels, early warning results, intervention reminder plans, and patient intervention implementation status. Data storage utilizes a distributed database, categorized and stored according to each patient's unique identifier, enabling full lifecycle management of patient chronic disease data. Simultaneously, it strictly adheres to relevant medical data privacy protection regulations, de-identifying patients' personal information to ensure data security.

[0079] The data traceability and model optimization unit supports multi-dimensional traceability of patients' chronic disease data throughout the entire process, including time, indicator, early warning, and intervention dimensions. Doctors and health managers can query relevant data using criteria such as patient unique identifier, chronic disease type, and time range, providing data support for clinical diagnosis and chronic disease management. Simultaneously, this unit periodically feeds the stored end-to-end data back to the multi-indicator fusion early warning calculation module and the risk stratification judgment module. Through machine learning algorithms, the weight coefficients of the early warning model and the scoring criteria of the risk stratification model are iteratively optimized, ensuring that the model's early warning accuracy and risk stratification accuracy continuously improve with data accumulation, achieving system self-optimization.

[0080] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A chronic disease early warning and risk stratification system, characterized by: It includes modules for data acquisition and preprocessing, chronic disease indicator feature extraction, multi-indicator fusion early warning calculation, risk stratification judgment, trend analysis and deterioration early warning, complication linkage early warning, and early warning intervention and data management. The data acquisition and preprocessing module is used for the unified acquisition, cleaning and standardization of multi-source chronic disease data; the chronic disease indicator feature extraction module is used to extract core monitoring indicators and feature factors for liver disease, diabetes and hypertension; the multi-indicator fusion early warning calculation module is used to construct a multi-indicator weighted fusion early warning model and calculate the comprehensive early warning index for chronic diseases; the risk stratification judgment module is used to construct a risk stratification scoring model and realize four-level stratification of chronic disease risk. The trend analysis and deterioration early warning module is used to determine the disease development trend and identify signals of disease deterioration. The complication linkage early warning module is used to build a rule base for complication association and realize the linkage early warning of primary diseases and complications; the early warning intervention and data management module is used to push personalized intervention reminders and complete the data management of the entire chronic disease process.

2. The chronic disease early warning and risk stratification system according to claim 1, characterized in that: The data acquisition and preprocessing module includes a multi-source data acquisition unit, a data cleaning unit, and a data standardization unit. The chronic disease-related data collected by the multi-source data acquisition unit are divided into three categories: clinical test data, personal health record data, and lifestyle data. The data cleaning unit uses interpolation, manual review, deduplication, and format conversion to process missing values, outliers, duplicate values, and non-uniform format data. The data standardization unit uses extreme value standardization, dummy variable coding, and rank coding to normalize and quantify the cleaned data.

3. The chronic disease early warning and risk stratification system according to claim 1, characterized in that: The chronic disease indicator feature extraction module includes a disease indicator classification unit, a core indicator screening unit, and a feature factor quantification unit. The disease indicator classification unit divides the indicators into two categories: exclusive core indicators and cross-disease related indicators. The core indicator screening unit combines clinical treatment guidelines, Pearson correlation coefficient method, and random forest algorithm to screen core indicators. The feature factor quantification unit transforms the core indicators into disease feature factors with values ​​in the range [0,1], where a larger value indicates a higher risk of the corresponding disease.

4. The chronic disease early warning and risk stratification system according to claim 1, characterized in that: The multi-indicator fusion early warning calculation module includes an indicator weight determination unit, a fusion early warning index calculation unit, and a basic early warning judgment unit. The indicator weight determination unit uses a combination of the analytic hierarchy process (AHP) and clinical expert scoring to determine the feature factor weight coefficients, with the sum of the weight values ​​being 1. The fusion early warning index calculation unit uses a formula... Calculate a comprehensive early warning index for a single chronic disease, where, It is a comprehensive early warning index for a single chronic disease, with a value range of [0,1]. This represents the number of characteristic factors in the set of characteristic factors for this chronic disease. For the first The weight coefficients of each feature factor satisfy the following conditions: ; For the first The quantized values ​​of each feature factor range from [0,1]. Calculate the overall comprehensive early warning index, where , , These are the disease weight coefficients for liver disease, diabetes, and hypertension, respectively. The basic early warning judgment unit has a preset threshold of 0.5, based on... The value of is used to achieve the three-level judgment of basic early warning.

5. The chronic disease early warning and risk stratification system according to claim 1, characterized in that: The risk stratification determination module includes a risk stratification scoring calculation unit, a risk level determination unit, and a stratification result output unit. The risk stratification scoring calculation unit uses the formula Calculate the comprehensive risk score for chronic diseases, among which, The comprehensive risk score for a single chronic disease ranges from [0,2]. This is the comprehensive early warning index for the chronic disease, with a value range of [0,1]. The trend influence coefficient is a fixed constant with a value of 1, reflecting the amplifying effect of the disease trend on the risk score. This is the disease trend coefficient for the chronic disease, with a value range of [0,1]. It is calculated by the trend analysis and deterioration early warning module and transmitted to this unit. A higher value indicates a more pronounced trend of disease progression. This indicates that the disease trend is stable; the risk level determination unit is based on... The value of the score divides the risk of chronic diseases into four levels: low, medium, high, and very high. At the same time, the score is fine-tuned according to the specific characteristics of the three major chronic diseases, and the score after fine-tuning does not exceed 2.

0. The hierarchical result output unit visualizes the risk level of a single chronic disease and the overall chronic disease risk level and transmits it to the other modules.

6. The chronic disease early warning and risk stratification system according to claim 1, characterized in that: The trend analysis and deterioration early warning module includes a historical index fitting unit, a disease trend coefficient calculation unit, a deterioration signal identification unit, and a deterioration early warning issuance unit; The historical index fitting unit uses multinomial fitting and moving average methods to construct the time series trend fitting curve of the feature factors. The disease trend coefficient calculation unit calculates the disease trend coefficient using the feature factor slope and weight coefficient. , The value range is [0,1] and is transmitted to the risk stratification determination module; The deterioration signal identification unit sets deterioration signal judgment rules based on feature factor trends, early warning index changes, trend coefficients, and comorbidities; the deterioration early warning issuing unit determines the three-level priority of deterioration early warning according to the risk level.

7. The chronic disease early warning and risk stratification system according to claim 1, characterized in that: The complication linkage early warning module includes a complication association rule base unit, a comorbidity risk calculation unit, a complication early warning determination unit, and a linkage early warning issuance unit; The complication association rule base unit constructs a rule base for the interaction and complication early warning of the three major chronic diseases based on clinical practice guidelines, and the rule base supports dynamic updates; the comorbidity risk calculation unit calculates the comorbidity risk value using a weighted summation method. , The value range is [0,1]; the complication warning determination unit presets a threshold of 0.6 and determines the complication warning based on the risk level; the linkage warning issuing unit issues a linkage warning that includes the primary disease and complication type, and integrates it with other warnings in terms of priority.

8. The chronic disease early warning and risk stratification system according to claim 1, characterized in that: The early warning intervention and data management module includes an intervention reminder scheme library unit, a personalized intervention push unit, a full-process data storage unit, and a data traceability and model optimization unit; The intervention reminder plan library unit constructs four categories of standardized intervention reminder plans that match risk levels and warning types, and the plan library supports dynamic updates; the personalized intervention push unit combines patient data to generate personalized intervention reminder information and realizes multi-terminal push, and supports manual adjustment of intervention plans; The full-process data storage unit uses a distributed database to store the full-process chronic disease data in a structured and de-identified manner; the data traceability and model optimization unit supports multi-dimensional data traceability and feeds the data back to relevant modules to achieve iterative optimization of the model.

9. The stratification method of the chronic disease early warning and risk stratification system according to any one of claims 1-8, characterized in that: Includes the following steps: Step 1: Multi-source data acquisition and preprocessing. The data acquisition and preprocessing module collects multi-source data related to the three major chronic diseases, completes data cleaning and standardization, and forms a standardized chronic disease dataset. Step 2: Chronic disease indicator feature extraction. The chronic disease indicator feature extraction module classifies the standardized dataset, selects core indicators, and quantifies feature factors to form three major chronic disease feature factor sets. Step 3: Multi-indicator fusion early warning calculation. The feature factor weights are determined through the multi-indicator fusion early warning calculation module, and the single and overall chronic disease comprehensive early warning index is calculated to complete the basic early warning level three judgment. Step 4: Chronic disease risk stratification determination. By combining the comprehensive early warning index and the disease trend coefficient through the risk stratification determination module, a comprehensive chronic disease risk score is calculated to complete the four-level stratification of chronic disease risk. Step 5: Trend Analysis and Deterioration Warning. By fitting historical indicator trends through the trend analysis and deterioration warning module, the disease trend coefficient is calculated, deterioration signals are identified, and deterioration warnings are issued according to priority. Step 6: Complication linkage early warning. By combining the complication linkage early warning module with the complication association rule base, the comorbidity risk value is calculated, and a complication linkage early warning is issued. Step 7: Early warning intervention and data management. Personalized intervention reminders are pushed through the early warning intervention and data management module. Data storage and tracing of the entire chronic disease process are completed, and data feedback is used to optimize the model.

10. The stratification method of the chronic disease early warning and risk stratification system according to claim 9, characterized in that: The comorbidity risk value mentioned in step 6 is obtained by weighted summation of the patient's chronic disease warning status, risk level and complication core indicator quantification values. The model optimization mentioned in step 7 is to iteratively update the warning model weight coefficient and risk stratification scoring standard through machine learning algorithm. The disease trend coefficient mentioned in step 9 is obtained by weighted summation of the slope of the feature factor trend fitting curve after normalization and combined with the feature factor weight coefficient.

Images