Intelligent processing method, system and equipment for high-risk and new-onset definite-diagnosis people of hypertension and medium

Abstract

The invention provides an intelligent processing method, system, equipment and medium for high-risk and newly-onset definite diagnosis crowds of hypertension, and belongs to the technical field of hypertension detection.The method comprises the steps that patient examination data are collected and serve as an original data set; preprocessing the original data; constructing a corresponding parameter group; mapping the family history score into a risk quantification function through a dynamic weight coefficient Wi; summing the product of the dynamic weight coefficient Wi and the quantization function Fi to obtain a total risk score; determining whether the patient is a high-risk population based on the diagnosis code and the total risk score; and pushing the high-risk crowd data to a management center server. According to the method, the data is accurately pushed to a corresponding platform or a storage library, so that safe and effective utilization of the data is guaranteed, and sharing and collaboration of regional medical data are realized.

Description

Technical Field

[0001] This invention belongs to the field of hypertension detection technology, and in particular relates to an intelligent processing method, system, device and medium for high-risk and newly diagnosed hypertension populations. Background Technology

[0002] Hypertension is a common chronic disease that damages vital organs such as the heart, brain, and kidneys, leading to serious complications such as coronary heart disease and stroke, impacting patients' quality of life and life expectancy. Currently, community health checkups can be conducted to assess users' health and collect relevant data, including physiological parameters such as blood pressure and blood sugar levels.

[0003] In related technologies, due to the different information systems used by different medical institutions, the data formats and standards collected vary significantly. For example, the units and representation methods of physiological indicators such as blood pressure and blood glucose levels are inconsistent, diagnostic coding systems are not unified, and there is a large amount of unstructured text data, making data integration difficult. This results in poor data processing effectiveness during data analysis and mining, affecting the analytical and application value of the data. Hypertension screening methods typically rely on blood pressure measurements, lacking a comprehensive consideration of the patient's overall health status and various risk factors, leading to low accuracy in screening results and a high risk of missed or misdiagnosed cases. Furthermore, data from different medical institutions is often independent, lacking effective sharing mechanisms and unified data transmission standards, resulting in information silos and hindering timely data flow and sharing. Summary of the Invention

[0004] This invention provides an intelligent processing method for high-risk and newly diagnosed hypertension populations. The method accurately pushes data to the corresponding platform or repository, which not only ensures the safe and effective use of data, but also realizes the sharing and collaboration of regional medical data.

[0005] The methods include: S101: Collect patient examination data and use it as the raw dataset. The raw dataset includes: patient identification, systolic blood pressure, diastolic blood pressure, body mass index, family history and diagnostic code. S102: Preprocess the raw data; S103: Based on preset age and visit frequency thresholds, the target population is screened through an exclusion rule engine, and a corresponding parameter group P={systolic blood pressure, diastolic blood pressure, body mass index, and family history quantitative parameters} is constructed. S104: Map family history scores to a risk quantification function using dynamic weighting coefficients Wi; S105: The total risk score is obtained by summing the product of the dynamic weighting coefficient Wi and the quantization function Fi; S106: Determine whether the patient is a high-risk individual based on the diagnostic code in step S101 and the total risk score in step S105; S107: High-risk population data will be pushed to the management center server in real time via the MQTT protocol. Newly diagnosed data will be desensitized by AES-256 and pushed to the regional medical platform via the HL7 FHIR standard. The remaining data will be stored in the regional health record database. The push priority is positively correlated with the risk level.

[0006] It should be further noted that step S102 also includes: defining a time series analysis module, constructing a time series feature group T={systolic blood pressure_t, diastolic blood pressure_t, BMI_t}, where t is a daily / weekly time slice, and calculating the rate of change of parameters through a dynamic trend recognition algorithm. ΔT = (Current value - Historical average) / Historical average × 100% When ΔT exceeds the preset fluctuation threshold, an anomaly marker is triggered, and ΔT is fused with the original parameter set P to form an enhanced feature set P'.

[0007] It should be further noted that a dynamic response mechanism is defined in step S107, and a three-level early warning trigger is set; Low risk: Score < 60 points and ΔT ≤ 10%, monitored weekly; Medium risk: 60≤Score<85, and ΔT>15%, requiring daily monitoring; High risk: Score ≥ 85 points. The data will be pushed in real time and reviewed manually before being uploaded to the management center server. When three consecutive warnings are upgraded, the online update of model parameters is triggered; Construct Q(s,a)=Q(s,a)+η×[R(s,a)+γ×maxQ(s',a')-Q(s,a)], where state s=risk level, action a=push strategy, and reward function R=intervention effectiveness score; when the early warning accuracy drops to >5%, execute W_new=W_old+α×(E_target-E_current), α=0.05, and E is the prediction error.

[0008] It should be further noted that, in this method, machine learning algorithms and the same initial model for hypertension risk assessment are deployed on multiple computing nodes respectively. In step S104, after extracting the parameter set of the hypertension risk assessment model, the parameter labels in the parameter set are defined; a list of hypertension prevalence status is constructed based on the parameters corresponding to the parameter labels, and training sample data is formed. Training sample data is sent to each computing node so that each computing node can train the hypertension risk assessment model deployed on its own using the machine learning algorithm deployed on its own. After the management center server receives the model parameters and their values ​​from each computing node, it determines the target values ​​for each model parameter and sends these target values ​​to each computing node. This allows each computing node to adjust the values ​​of the corresponding model parameters in the hypertension risk assessment model to the target values. Then, it jumps to the training sample preparation step to re-prepare the training sample data and train the model until the hypertension risk assessment models of each computing node have converged. It should be further explained that the following steps are performed at each computing node: input the training sample data into the hypertension risk assessment model, use the preset medical knowledge database to predict the hypertension risk assessment data corresponding to the training sample data, and obtain the hypertension risk assessment data output by the model. Using a pre-defined loss function, calculate the loss value between the label and the hypertension risk assessment data; If the loss value is greater than the preset loss threshold, then determine the model parameters to be adjusted and their values, and send the expected model parameters to be adjusted and their values ​​to the management center server. After training and optimizing the hypertension risk assessment model, the optimized model is used to calculate the total risk score in step S105. After calculating the total risk score, the scores are labeled according to the updated dual-channel labeling rules.

[0009] It should be further explained that step S101 specifically includes: Extract various health-related data features from the raw data; for each data feature, cluster it according to type to form at least one original dataset; In each original dataset, extract key indicator data for hypertension risk assessment; Adjust the data collection frequency and level according to the patient's health status; Using a pre-defined hypertension risk warning model, key indicators are extracted for risk warning.

[0010] It should be further explained that after calculating the total risk score in step S105, a dynamic early warning is determined based on the risk assessment results after classification and the preset response rules. The response rules set different trigger conditions and response methods for different risk levels, including: The trigger condition is that the total risk score is less than the lower threshold. The trigger condition is that the total risk score is between the lower threshold and the upper threshold. The trigger condition is that the total risk score is higher than the upper limit threshold; After completing the dynamic early warning, continue to execute the dual-channel annotation rules in step S106; Based on risk assessment results and dynamic early warning feedback, the effectiveness of current management strategies for high-risk and newly diagnosed hypertension patients is analyzed in real time. When it is found that the current management strategy cannot cope with the risk, the management strategy is updated using a self-learning mechanism and the following self-learning update algorithm. f(px,qx)=(1−lc)f(px, qx)+ lc (r+γmaxQ(py, qy)) Where f(px,qx) represents the expected feedback information of analyzing qx under the current management strategy px for high-risk and newly diagnosed hypertension populations, lc is the learning rate of the management strategy for high-risk and newly diagnosed hypertension populations, r is the immediate feedback information of the current feedback, γ is the feedback discount coefficient used to measure the weight of future feedback information, py is the output state of the next management strategy for high-risk and newly diagnosed hypertension populations, and qy is the information obtained from the next analysis.

[0011] This application also provides an intelligent processing system for high-risk and newly diagnosed hypertension populations, the system comprising: The data acquisition module is used to collect patient examination data and use it as the raw dataset. The raw dataset includes: patient identification identifier, systolic blood pressure, diastolic blood pressure, body mass index, family history and diagnostic code. The preprocessing module is used to preprocess the raw data; The engine screening module is used to filter the target population using preset age thresholds and visit frequency thresholds through exclusion rules, and to construct the corresponding parameter group P={systolic blood pressure, diastolic blood pressure, body mass index, and family history quantitative parameters}. The risk analysis module is used to map family history scores to risk quantification functions using dynamic weighting coefficients Wi; The risk calculation module is used to calculate the sum of the product of the dynamic weight coefficient Wi and the quantization function Fi to obtain the total risk score; Determining whether a patient belongs to a high-risk group based on diagnostic codes and total risk scores; The data push module is used to push high-risk population data to the management center server in real time via the MQTT protocol. Newly diagnosed data is desensitized by AES-256 and pushed to the regional medical platform via the HL7 FHIR standard. The remaining data is stored in the regional health record database. The push priority is positively correlated with the risk level.

[0012] According to another embodiment of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the intelligent processing method for high-risk and newly diagnosed hypertension populations.

[0013] According to another embodiment of this application, a storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the intelligent processing method for high-risk and newly diagnosed hypertension populations.

[0014] As can be seen from the above technical solutions, the present invention has the following advantages: The intelligent processing method for high-risk and newly diagnosed hypertension populations provided in this application automatically executes data collection tasks at preset time intervals. Through standardized data collection specifications and interface protocols, it achieves effective integration and unified collection of multi-source heterogeneous data. The collected raw data undergoes data cleaning and standardization. A hash algorithm is used to generate regionally unique patient identifiers (IDs) to ensure the uniqueness and relevance of patient data; further normalization processing is performed to unify the multi-source heterogeneous data into a standardized structured data format. This effectively solves the problems of inconsistent formats and low structuring levels in multi-source medical data, achieving efficient data integration.

[0015] This application constructs an exclusion rule engine to screen the target population, excluding high-risk and confirmed patients already under management from a large population, thus narrowing the screening scope and improving screening targeting. Then, core parameter sets are extracted, including highly correlated risk factors such as systolic blood pressure, diastolic blood pressure, BMI, and family medical history, and a three-level scoring rule is used to quantify family medical history. The parameter sets are input into a pre-trained hypertension risk assessment model, and a total score is calculated through weighted scoring, comprehensively considering the weights and quantification functions of each parameter to achieve accurate quantitative assessment of hypertension risk. Finally, a dual-channel result determination is performed, classifying and labeling based on the presence of confirmed cases and risk scores to accurately identify high-risk groups and newly diagnosed patients. This application constructs a hierarchical push mechanism. During data transmission, the MQTT protocol is used to achieve efficient real-time data transmission, and the HL7 FHIR standard is used to ensure data standardization and compatibility, enabling regional medical data sharing and collaboration. Simultaneously, measures such as generating unique patient identifiers through hash algorithms and data anonymization are implemented to protect data security. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying 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.

[0017] Figure 1 Flowchart of intelligent management methods for high-risk and newly diagnosed hypertension populations; Figure 2A schematic diagram of an intelligent processing system for high-risk and newly diagnosed hypertension patients; Figure 3 This is a schematic diagram of an electronic device. Detailed Implementation

[0018] The following details the specific steps of the intelligent processing method for high-risk and newly diagnosed hypertension populations involved in this application. For illustrative purposes and not for limitation, specific details such as particular system structures and technologies are presented to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details.

[0019] It should be understood that, when used in this specification, the term "comprising" indicates the presence of the described feature, integral, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0020] It should be understood that "one or more" as mentioned in this application refers to one, two, or more, and "multiple" as mentioned in this application refers to two or more. In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0021] To facilitate a clear description of the technical solutions of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" do not necessarily imply that they are different.

[0022] The terms "one embodiment" or "some embodiments" used in this application mean that one or more embodiments of this application include the specific features, structures, or characteristics described in that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this application do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized.

[0023] In embodiments of the present invention, computer program code for performing the operations of the present disclosure can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include, but are not limited to, object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer.

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figure 1 The diagram shows a flowchart of an intelligent processing method for high-risk and newly diagnosed hypertension populations in a specific embodiment. The method includes: The methods include: S101: Collect patient examination data and use it as the raw dataset. The raw dataset includes: patient identification, systolic blood pressure, diastolic blood pressure, body mass index, family history, and diagnostic code.

[0026] In this embodiment, data interface technology is used to extract the required data from different data sources periodically or in real time according to the established data format and transmission protocol, and the data is aggregated to form an original dataset, providing basic data support for subsequent data processing and analysis.

[0027] In this way, by connecting with the information systems of various medical institutions, such as hospital electronic medical record systems and physical examination center report systems, patient examination data can be obtained. The data collection process follows standardized data collection specifications to ensure data accuracy and consistency. The collected data includes patient identification identifiers to uniquely identify each patient; diagnostic codes, especially those related to hypertension, used for subsequent analysis to determine if the patient already has a hypertension diagnosis record.

[0028] S102: Preprocess the raw data.

[0029] This embodiment uses data cleaning and standardization algorithms to check and transform the raw data one by one according to pre-set medical knowledge rules and data processing standards, thereby improving data quality.

[0030] The preprocessing in this embodiment can begin with data cleaning to identify and correct erroneous, missing, and duplicate values ​​in the data. For example, abnormal blood pressure values ​​that exceed the range of medical common sense are checked and corrected; for missing body mass index (BMI) data, if height and weight data are available, they are calculated and supplemented; otherwise, reasonable estimations are made based on the characteristics of similar population data. Next, data standardization is performed, converting data of different units and magnitudes into a standardized format.

[0031] Optionally, the units for systolic and diastolic blood pressure can be standardized, and body mass index (BMI) can be normalized according to a certain standard range. Simultaneously, patient identification identifiers can be encrypted to ensure patient privacy and security.

[0032] As one implementation embodiment of this application, preprocessing is performed on the raw data to ensure data consistency and integrity. From the preprocessed data, data such as blood pressure trends, body mass index fluctuation ranges, and specific family medical history can be extracted. For each data feature, clustering is performed according to its type to form at least one feature set. Within each feature set, key indicator data for hypertension risk assessment are further extracted.

[0033] This embodiment allows for adjustment of data collection frequency and levels based on the patient's health status. The health status in this embodiment can be low-risk, medium-risk, or high-risk.

[0034] When an abnormally high risk or other abnormal situation is detected, timely measures should be taken, such as increasing the monitoring frequency or suspending certain non-critical tasks, to ensure the stable operation of the system and the validity of the data.

[0035] This embodiment utilizes a pre-defined hypertension risk warning model to extract key indicator data for risk warning. This hypertension risk warning model can be a model in the training phase or a model in the application phase. In the model application phase, key indicator data is input into the hypertension risk warning model so that the model can output risk warning results based on this data. For each warning in the risk warning results, a tracking task is set up to display risk analysis data and solution suggestions.

[0036] It should be noted that during the model training phase, labels are assigned to key indicator data to form training sample data; this training sample data is then distributed to multiple training nodes. Each training node uses a different machine learning algorithm to train its own hypertension risk warning model. While the nodes share the model, they do not share sensitive data, and collaboratively optimize model parameters through federated learning to ensure data security and privacy.

[0037] This embodiment receives model parameters and values ​​from each training node, calculates the average value of the same parameter, and uses it as the target value for that parameter. The target value is then sent to each training node, allowing it to adjust its model parameters until the models of all nodes converge. Using a historical case database and an expert knowledge base, the fault prediction model can push relevant information to patients after detecting hypertension risk, and simultaneously store it in the database.

[0038] S103: Based on preset age and visit frequency thresholds, the target population is screened through an exclusion rule engine, and a corresponding parameter group P={systolic blood pressure, diastolic blood pressure, body mass index, and family history quantification parameters} is constructed.

[0039] In this embodiment, an exclusion rule engine is implemented by writing program code to logically filter data based on set thresholds. Using data mining and modeling techniques, the intrinsic relationship between these key parameters and the risk of hypertension is sought based on the filtered data, and a mathematical model capable of assessing the risk of hypertension is constructed.

[0040] It should be noted that the age threshold in this embodiment can be preset to 30 years old, or other ages. The frequency threshold for outpatient visits is set according to different medical institutions and research experience, and is set to at least 2 visits per year. The exclusion rule engine filters out patient data from the original dataset based on the threshold, selecting patients who are 30 years of age or older and whose outpatient visit frequency meets or exceeds the threshold. For family history, quantitative rules are used to convert it into quantitative parameters of family history, such as assigning 3 points to a direct relative with hypertension, 2 points to a collateral relative with hypertension, and 0 points to no family history. Using the filtered data, combined with machine learning algorithms or statistical analysis methods, a hypertension risk assessment model is constructed. During the model construction process, systolic blood pressure, diastolic blood pressure, body mass index, and quantitative parameters of family history are identified as key parameters for subsequent risk assessment calculations.

[0041] S104: Map family history scores to a risk quantification function using dynamic weighting coefficients Wi; This embodiment continuously adjusts the dynamic weight coefficients through a weight optimization algorithm, and transforms the family history score into a quantitative value related to the risk of hypertension based on the principle of mathematical functions, providing a basis for the subsequent calculation of the total risk score.

[0042] It should be noted that, based on the preset age threshold and the frequency threshold for medical visits, the target population is screened by the exclusion rule engine, and a parameter group P={systolic blood pressure, diastolic blood pressure, body mass index, and quantitative family history parameter} is constructed. The quantitative family history parameter is generated by the three-level scoring rule in step S104: 3 points for confirmed diagnosis of immediate family members, 2 points for confirmed diagnosis of collateral family members, and 0 points for no family history. The model parameter weights are dynamically calibrated through historical mislabeling rate.

[0043] In this embodiment, the family history parameter of model parameter group P in S103 directly references the output value of the three-level scoring rule in S104, making the model construction process strictly dependent on the feature quantization results of S104. The model weight calibration rule in S103 is based on the historical labeled data in S104, forming a closed loop of feature quantization, model construction, and feedback optimization. The family history scoring rule in step S104 is automatically synchronized to the parameter group P definition in S103, realizing bidirectional linkage between feature definition and model parameters. The deep coupling of the family history scoring rule and the model parameter weight calibration allows the genetic risk weight to be dynamically adjusted according to the labeled feedback.

[0044] This embodiment also extracts a parameter set from the constructed parameter set, including quantitative parameters such as systolic blood pressure, diastolic blood pressure, body mass index, and family history. Dynamic weighting coefficients are continuously adjusted and optimized based on model training and actual data validation to reflect the relative importance of each parameter in assessing hypertension risk.

[0045] For the quantitative parameter of family history, a specific functional relationship is used, combined with a dynamic weighting coefficient Wi, to map the family history score to a risk quantification function. For example, if the quantitative parameter of family history is 3 points, the corresponding risk quantification value is calculated based on the dynamic weighting coefficient and the set function. This value reflects the degree to which family history factors contribute to the risk of hypertension. S105: The total risk score is obtained by summing the product of the dynamic weighting coefficient Wi and the quantization function Fi.

[0046] Based on the principle of mathematical weighted summation, this embodiment calculates a total risk score that reflects the degree of hypertension risk of a patient by comprehensively calculating the importance of each parameter (reflected by the dynamic weight coefficient Wi) and its quantitative function Fi on hypertension risk.

[0047] For each parameter in the parameter set, there is a corresponding quantification function Fi. For example, systolic blood pressure has a quantification function set according to different blood pressure ranges, and the same applies to diastolic blood pressure and body mass index (BMI). The dynamic weighting coefficient Wi corresponding to each parameter is multiplied by the corresponding quantification function Fi, and then all the products are summed to obtain the total risk score. For example, the dynamic weighting coefficient for systolic blood pressure is W1, and the quantification function is F1; the dynamic weighting coefficient for diastolic blood pressure is W2, and the quantification function is F2; ​​the dynamic weighting coefficient for BMI is W3, and the quantification function is F3; the dynamic weighting coefficient for family history is W4, and the quantification function is F4. The total risk score is: Score = W1 × F1 + W2 × F2 + W3 × F3 + W4 × F4.

[0048] It should be noted that the three-level quantitative value of the family history score extracted in step S104 is used as the input parameter of the Fi function in step S105, and the dynamic adjustment rule of the weight α is driven by the historical labeled data in step S104. The defined β weight parameter can be a high-salt diet marker, which is coupled with the systolic blood pressure fluctuation compensation factor in S105. When the daily fluctuation of SBP is >20 mmHg, the β value is automatically reduced by 15%. The timely data of the treatment response in step S105 triggers the re-optimization of the family history weight α in S104, and the α gain = historical mislabeling rate × 0.5.

[0049] It can be seen that step S104 defines the structured processing rules for the parameter group; step S105 outputs the total risk score based on the parameters of S104, and optimizes the weight parameters through real-time feedback, forming a closed loop of parameter extraction, dynamic scoring, and feedback optimization.

[0050] S106: Determine whether the patient is a high-risk individual based on the diagnostic code from step S101 and the total risk score from step S105.

[0051] Specifically, if an ICD-10 coded I10 category diagnostic record exists, the patient is marked as a newly diagnosed case. If there is no diagnostic record and the Score is greater than or equal to the dynamic threshold, the patient is marked as a high-risk individual.

[0052] In this embodiment, data query and comparison techniques are used to determine the patient's diagnosis and risk score based on pre-set diagnostic coding rules and risk threshold judgment standards, thereby classifying and labeling the patient. The system searches the patient's diagnostic coding data for ICD-10 category I10 diagnostic records. If a record exists, the patient is directly labeled as a newly diagnosed patient. For patients without diagnostic records, their calculated total risk score (Score) is compared with a dynamic threshold. The dynamic threshold is dynamically adjusted based on extensive clinical data and model validation results, combined with characteristics of different regions and populations. If the Score is greater than or equal to the dynamic threshold, the patient is labeled as a high-risk individual.

[0053] S107: High-risk population data will be pushed to the management center server in real time via the MQTT protocol. Newly diagnosed data will be desensitized by AES-256 and pushed to the regional medical platform via the HL7 FHIR standard. The remaining data will be stored in the regional health record database. The push priority is positively correlated with the risk level.

[0054] In this embodiment, data from individuals labeled as high-risk is transmitted to the management center server in real-time using the MQTT protocol, enabling the management center to promptly implement health management measures for these individuals. For data from newly diagnosed patients, sensitive information is first anonymized using the AES-256 encryption algorithm to ensure patient privacy. Then, the data is pushed to the regional healthcare platform according to the HL7 FHIR (Health Level Seven Fast Healthcare Interoperability Resources) standard for further diagnosis and treatment by medical institutions. Data from individuals who are neither high-risk nor newly diagnosed patients is stored in the regional health record database as part of the medical big data for subsequent medical research and public health analysis. Push priority is based on risk level, with high-risk group data receiving the highest priority, followed by newly diagnosed patient data, and all other data receiving the lowest priority.

[0055] In some specific embodiments, step S105 calculates the total risk score based on the parameter set P = {systolic blood pressure (SBP), diastolic blood pressure (DBP), body mass index (BMI), family history of hypertension (FHX), and high-salt diet marker (SFD)} by summing the product of a dynamic weight coefficient Wi and a quantization function Fi that includes a blood pressure fluctuation compensation factor and a family history risk mapping rule. The weight Wi is dynamically optimized based on historical annotation accuracy and response timeliness. If an ICD-10 coded I10 category diagnostic record exists, the patient is labeled as a "newly diagnosed patient"; if no diagnostic record exists and the score ≥ dynamic threshold T, the patient is labeled as a "high-risk group," where ΔT = historical false alarm rate × risk sensitivity coefficient. The differentiated push based on the annotation results shows that the scoring result of step S105 directly drives the annotation decision in step S106, and the annotation confidence of step S106 is fed back to step S105 to trigger weight re-optimization. The push response timeliness in step S107 triggers a hot update of the model parameters in S105. In this embodiment, step S105 covers data modeling, step S106 makes decision rules, and step S107 executes the closed loop. The adaptive adjustment of Wi and the threshold sensitivity coefficient constitute technical improvements.

[0056] Based on the above embodiments, in order to further improve the reliability of the intelligent processing method for high-risk and newly diagnosed hypertension populations provided in the above embodiments, the following is an implementable method: In step S102, a time series analysis module is embedded to construct a time series feature group T={systolic blood pressure_t, diastolic blood pressure_t, BMI_t} (t is a daily / weekly time slice), and the parameter change rate is calculated through a dynamic trend recognition algorithm: ΔT = (current value - historical average) / historical average × 100%.

[0057] When ΔT exceeds a preset fluctuation threshold, an anomaly marker is triggered, and ΔT is merged with the original parameter set P to form an enhanced feature set P'. For example, daily fluctuations in systolic blood pressure > 20 mmHg.

[0058] In step S107, a dynamic response mechanism is defined, and a three-level early warning trigger is set; Low risk: Score < 60 points and ΔT ≤ 10%, monitored weekly; Medium risk: 60 ≤ Score < 85 points and ΔT > 15%, monitored daily; High risk: Score ≥ 85 points, pushed in real time, combined with manual review, and the reviewed data is uploaded to the management center server.

[0059] In this embodiment, Q(s,a) = Q(s,a) + η × [R(s,a) + γ × maxQ(s',a') - Q(s,a)] is constructed, where state s = risk level, action a = push strategy, and reward function R = intervention effectiveness score. When the early warning accuracy drops > 5%, W_new = W_old + α × (E_target - E_current) is executed, α = 0.05, and E is the prediction error.

[0060] In this embodiment, ΔT can be used as an input feature for the risk score in step S105, automatically increasing the score weight when blood pressure fluctuations are abnormal. This embodiment integrates the ΔT feature, based on the family history score from step S104 and the newly added spatial / temporal features, and improves the model's adaptability through dynamic weight adjustment.

[0061] As one embodiment of this application, different machine learning algorithms and the same initial model for hypertension risk assessment are deployed on multiple computing nodes. For example, the first computing node uses a support vector machine-based hypertension risk feature classification algorithm, the second computing node uses a random forest-based hypertension risk factor combination analysis algorithm, and the third computing node uses a logistic regression-based hypertension risk quantification prediction algorithm. These algorithms can be used for subsequent training and optimization of the hypertension risk assessment model. In this embodiment, after extracting the parameter group P={SBP, DBP, BMI, FHX, High-Salt Diet Marker SFD} in step S104, labels are assigned to these parameter data. The labels are a list of hypertension prevalence states based on clinical diagnostic criteria for hypertension and expert experience. This list records the corresponding disease analysis data for each prevalence state, such as the range of abnormal blood pressure fluctuations, the degree of influence of family genetic factors, treatment plans for hypertension patients, and health management recommendations for high-risk groups, forming training sample data. In this embodiment, the training sample data is sent to each computing node so that each computing node can use its locally deployed machine learning algorithm to train its locally deployed hypertension risk assessment model.

[0062] The hypertension risk assessment model is based on a patient's blood pressure, BMI, family history, etc., and uses machine learning algorithms to quantify the probability or risk level of future hypertension-related hypertension diagnosis and complications, providing data support for early clinical intervention, health management and risk stratification.

[0063] Hypertension risk assessment models can employ RNN recurrent neural network models, XGBoost / LightGBM gradient boosting tree models, or Bayesian network models.

[0064] Further, at each computing node, the following steps are performed: Training sample data is input into the hypertension risk assessment model. The model uses a pre-defined database to predict the hypertension risk assessment data corresponding to the training sample data, obtaining the hypertension risk assessment data output by the model. A pre-defined loss function, such as the mean squared error loss function, is used to measure the difference between the predicted risk assessment data and the actual disease status label, calculating the loss value between the label and the hypertension risk assessment data. If this loss value is greater than a pre-defined loss threshold (set based on experimental and clinical experience) used to determine whether the model's prediction accuracy meets the standard, the desired model parameters and their values ​​are determined and sent to the device where the management center server is located. After the management center server receives the model parameters and their values ​​from each computing node, it determines the target value for each model parameter and sends this target value to each computing node. This allows each computing node to adjust the corresponding model parameter in its hypertension risk assessment model to the target value. The system then proceeds to the training sample preparation step, re-preparing the training sample data and training the model until the hypertension risk assessment models on all computing nodes have converged. Furthermore, upon receiving the model parameters and their values ​​from each computing node, if the same model parameter has multiple values, the average value for that parameter is calculated and used as the target value for that parameter. Another implementation in this embodiment combines the screening results of the optimized model with the training and optimization of the hypertension risk assessment model. The optimized model is then used to calculate the total risk score in step S105. After calculating the total risk score, it is labeled according to the updated dual-channel labeling rules. If an ICD-10 coded I10 category diagnostic record exists, the patient is labeled as a newly diagnosed case. If there is no diagnostic record and the Score calculated according to the optimized model is ≥ the warning threshold T, the patient is labeled as a high-risk individual. Because the model has been trained and optimized, the accuracy of hypertension risk assessment is improved, making the labeling results more reliable. This embodiment implements a tiered real-time push mechanism based on the updated screening and labeling results. High-risk population data is pushed to the management center's server-side early warning dashboard via the MQTT protocol, along with high-risk factors derived from optimized model analysis and targeted health management recommendations. Newly diagnosed data, after being anonymized, is pushed to the regional medical platform via the HL7 FHIR standard, along with personalized treatment plan recommendations based on the optimized model and historical case analysis. When other data is stored in the regional health record database, it is also linked to record the risk assessment process data based on the optimized model, providing higher-quality data support for subsequent hypertension prevention and control research and strategy development.

[0065] In one exemplary embodiment, after calculating the total risk score in step S105, the system automatically determines whether to trigger a dynamic early warning based on the graded risk assessment results and preset response rules. The response rules set different triggering conditions and response methods for different risk levels. This embodiment can implement the following response rules. Low-risk response rule: triggered when the total risk score is less than the lower threshold. This can generate low-priority prompts, which can be displayed through background logging, low-frequency message alerts, or by marking the system as pending observation. Medium-risk response rule: Triggered when the total risk score falls between the lower and upper thresholds. This generates a medium-priority alert, sends a reminder to relevant personnel via messaging tools, and marks the system as alerted. High-risk response rule: Triggered when the total risk score exceeds the upper limit threshold. The response involves generating a high-priority alert, sending an emergency alarm to relevant personnel via phone, message, etc., and simultaneously highlighting the warning in the system, requiring immediate intervention. The response rules in this embodiment determine the form and intensity of notifications not only based on the risk level but also in conjunction with subsequent regulatory action plans. According to the regulatory action plan, the response rules dynamically adjust the frequency of notifications, the notification channels, and the response level. When a low-risk warning is issued, it will only be monitored periodically. When the total risk score detected is lower than the lower limit threshold, it will only indicate that there is a potential risk. When a medium-risk warning is issued, the risk source will be monitored regularly, and the monitoring frequency will be increased as the risk persists or increases. When a high-risk warning is issued, the warning is monitored in real time, and the status of the risk source is continuously assessed until the risk is eliminated. In this embodiment, if a dynamic early warning is triggered, corresponding risk response suggestions are automatically generated and provided based on the risk level and regulatory action plan. These suggestions include strengthening monitoring, adjusting resource allocation, and implementing emergency measures to ensure that relevant personnel can take effective action based on the prompts. After completing the dynamic early warning and response, the dual-channel labeling rule of the original step S106 is executed. That is, if there is an ICD-10 code I10 category diagnostic record, the patient is labeled as a newly diagnosed patient; if there is no diagnostic record and the Score is greater than or equal to the early warning threshold T, the patient is labeled as a high-risk group. At the same time, the dynamic early warning information is associated with the labeling results and recorded for subsequent analysis and processing. This embodiment analyzes the effectiveness of current management strategies for high-risk and newly diagnosed hypertension populations in real time based on risk assessment results and dynamic early warning feedback. When it is found that the current management strategy cannot cope with the risk, the management strategy is updated using a self-learning mechanism and the following self-learning update algorithm. f(px,qx)=(1−lc)f(px, qx)+ lc (r+γmaxQ(py, qy)) Where f(px,qx) represents the expected feedback information of analyzing qx under the current management strategy px for high-risk and newly diagnosed hypertension populations, lc is the learning rate of the management strategy for high-risk and newly diagnosed hypertension populations, r is the immediate feedback information of the current feedback, γ is the feedback discount coefficient used to measure the weight of future feedback information, py is the output state of the next management strategy for high-risk and newly diagnosed hypertension populations, and qy is the information obtained from the next analysis.

[0066] When completing step S107, the tiered real-time push mechanism is implemented, combined with optimized management strategies. When high-risk population data is pushed to the management center's server-side early warning dashboard via the MQTT protocol, optimized management strategy suggestions for high-risk populations are simultaneously pushed, enabling the management center's server to intervene based on the latest strategies. When newly diagnosed data, after being anonymized, is pushed to the regional medical platform via the HL7 FHIR standard, information on adjustments to management strategies for newly diagnosed patients is also pushed, assisting the regional medical platform in improving subsequent treatment plans. When other data is stored in the regional health record database, dynamic early warning information and management strategy optimization data from this risk assessment process will also be linked and recorded.

[0067] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0068] The following are embodiments of the intelligent processing system for high-risk and newly diagnosed hypertension populations provided in this disclosure. This system and the intelligent processing methods for high-risk and newly diagnosed hypertension populations in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the intelligent processing system for high-risk and newly diagnosed hypertension populations, please refer to the embodiments of the intelligent processing methods for high-risk and newly diagnosed hypertension populations described above.

[0069] like Figure 2 As shown, the system includes: The data acquisition module is used to collect patient examination data and use it as the raw dataset. The raw dataset includes: patient identification identifier, systolic blood pressure, diastolic blood pressure, body mass index, family history and diagnostic code. The preprocessing module is used to preprocess the raw data; The engine screening module is used to filter the target population using preset age thresholds and visit frequency thresholds through exclusion rules, and to construct the corresponding parameter group P={systolic blood pressure, diastolic blood pressure, body mass index, and family history quantitative parameters}. The risk analysis module is used to map family history scores to risk quantification functions using dynamic weighting coefficients Wi; The risk calculation module is used to obtain the total risk score by summing the product of the dynamic weighting coefficient Wi and the quantization function Fi. Determining whether a patient belongs to a high-risk group based on diagnostic codes and total risk scores; The data push module is used to push high-risk population data to the management center server in real time via the MQTT protocol. Newly diagnosed data is desensitized by AES-256 and pushed to the regional medical platform via the HL7 FHIR standard. The remaining data is stored in the regional health record database. The push priority is positively correlated with the risk level.

[0070] like Figure 3 As shown, this application also provides an electronic device, including a display module 103, a memory 102, a processor 101, and a computer program stored in the memory and executable on the processor 101. When the processor 101 executes the program, it implements the steps of the method for parsing and loading a GIM model of a power transmission project.

[0071] In embodiments of the present invention, electronic devices include, but are not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments described and / or claimed herein.

[0072] In this embodiment, processor 101 may be implemented using at least one of an Application-Specific Integrated Circuit (ASIC), a Programmable Logic Device (PLD), a Field-Programmable Gate Array (FPGA), a processor, a controller, a microcontroller, a microprocessor, or an electronic unit designed to perform the functions described herein. In some cases, such implementations may be implemented within a controller. For software implementations, implementations such as processes or functions may be implemented with separate software modules that allow the performance of at least one function or operation. The software code may be implemented by a software application (or program) written in any suitable programming language, and the software code may be stored in memory and executed by the controller.

[0073] The display module 103 is used to display information input by the user or information provided to the user. The display module 103 may include a display panel, which may be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), or the like.

[0074] The memory 102 can be used to store software programs and various data. The memory 102 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0075] This application also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the intelligent processing method for high-risk and newly diagnosed hypertension populations.

[0076] The storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0077] In a storage medium, a readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0078] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligently managing high-risk and newly diagnosed hypertension populations, characterized in that, The methods include: S101: Collect patient examination data and use it as the raw dataset. The raw dataset includes: patient identification, systolic blood pressure, diastolic blood pressure, body mass index, family history and diagnostic code. S102: Preprocess the raw data; S103: Based on preset age and visit frequency thresholds, the target population is screened through an exclusion rule engine, and a corresponding parameter group P={systolic blood pressure, diastolic blood pressure, body mass index, and family history quantitative parameters} is constructed. S104: Map family history scores to a risk quantification function using dynamic weighting coefficients Wi; S105: The total risk score is obtained by summing the product of the dynamic weighting coefficient Wi and the quantization function Fi; S106: Determine whether the patient is a high-risk individual based on the diagnostic code in step S101 and the total risk score in step S105; S107: High-risk population data will be pushed to the management center server in real time via the MQTT protocol. Newly diagnosed data will be desensitized by AES-256 and pushed to the regional medical platform via the HL7 FHIR standard. The remaining data will be stored in the regional health record database. The push priority is positively correlated with the risk level.

2. The intelligent processing method for high-risk and newly diagnosed hypertension populations according to claim 1, characterized in that, Step S102 further includes: defining a time series analysis module, constructing a time series feature group T={systolic blood pressure_t, diastolic blood pressure_t, BMI_t}, where t is a daily / weekly time slice, and calculating the parameter change rate through a dynamic trend recognition algorithm. ΔT = (Current value - Historical average) / Historical average × 100% When ΔT exceeds the preset fluctuation threshold, an anomaly marker is triggered, and ΔT is fused with the original parameter set P to form an enhanced feature set P'.

3. The intelligent processing method for high-risk and newly diagnosed hypertension populations according to claim 2, characterized in that, In step S107, a dynamic response mechanism is defined, and a three-level early warning trigger is set; Low risk: Score < 60 points and ΔT ≤ 10%, monitored weekly; Medium risk: 60≤Score<85, and ΔT>15%, requiring daily monitoring; High risk: Score ≥ 85 points. The data will be pushed in real time and reviewed manually before being uploaded to the management center server. When three consecutive warnings are upgraded, the online update of model parameters is triggered; Construct Q(s,a)=Q(s,a)+η×[R(s,a)+γ×maxQ(s',a')-Q(s,a)], where state s=risk level, action a=push strategy, and reward function R=intervention effectiveness score; when the early warning accuracy drops to >5%, execute W_new=W_old+α×(E_target-E_current), α=0.05, and E is the prediction error.

4. The intelligent processing method for high-risk and newly diagnosed hypertension populations according to claim 1, characterized in that, In this method, machine learning algorithms and the same initial model for hypertension risk assessment are deployed on multiple computing nodes respectively; In step S104, the parameter labels in the parameter group are defined; A list of hypertension prevalence states is constructed based on the parameters corresponding to the parameter labels, and training sample data is generated. Training sample data is sent to each computing node so that each computing node can train the hypertension risk assessment model deployed on its own using the machine learning algorithm deployed on its own. After the management center server receives the model parameters and their values ​​from each computing node, it determines the target values ​​for each model parameter and sends these target values ​​to each computing node. This allows each computing node to adjust the values ​​of the corresponding model parameters in the hypertension risk assessment model to the target values. Then, it jumps to the training sample preparation step to re-prepare the training sample data and train the model until the hypertension risk assessment models of each computing node have converged.

5. The intelligent processing method for high-risk and newly diagnosed hypertension populations according to claim 4, characterized in that, The following steps are performed on each computing node: input the training sample data into the hypertension risk assessment model, use the preset medical knowledge database to predict the hypertension risk assessment data corresponding to the training sample data, and obtain the hypertension risk assessment data output by the model. Using a pre-defined loss function, calculate the loss value between the label and the hypertension risk assessment data; If the loss value is greater than the preset loss threshold, then determine the model parameters to be adjusted and their values, and send the expected model parameters to be adjusted and their values ​​to the management center server. After training and optimizing the hypertension risk assessment model, the optimized model is used to calculate the total risk score in step S105. After calculating the total risk score, the scores are labeled according to the updated dual-channel labeling rules.

6. The intelligent processing method for high-risk and newly diagnosed hypertension populations according to claim 1 or 2, characterized in that, Step S101 specifically includes: Extract various health-related data features from the raw data; for each data feature, cluster it according to type to form at least one original dataset; In each original dataset, extract key indicator data for hypertension risk assessment; Adjust the data collection frequency and level according to the patient's health status; Using a pre-defined hypertension risk warning model, key indicators are extracted for risk warning.

7. The intelligent processing method for high-risk and newly diagnosed hypertension populations according to claim 2, characterized in that, After completing step S105 to calculate the total risk score, determine whether to trigger a dynamic early warning based on the risk assessment results after classification and the preset response rules. The response rules set different trigger conditions and response methods for different risk levels, including: The trigger condition is that the total risk score is less than the lower threshold. The trigger condition is that the total risk score is between the lower threshold and the upper threshold. The trigger condition is that the total risk score is higher than the upper limit threshold; After completing the dynamic early warning, continue to execute the dual-channel annotation rules in step S106; Based on risk assessment results and dynamic early warning feedback, the effectiveness of current management strategies for high-risk and newly diagnosed hypertension patients is analyzed in real time. When it is found that the current management strategy cannot cope with the risk, the management strategy is updated using a self-learning mechanism and the following self-learning update algorithm. f(px,qx)=(1−lc)f(px, qx)+ lc (r+γmaxQ(py, qy)) Where f(px,qx) represents the expected feedback information of analyzing qx under the current management strategy px for high-risk and newly diagnosed hypertension populations, lc is the learning rate of the management strategy for high-risk and newly diagnosed hypertension populations, r is the immediate feedback information of the current feedback, γ is the feedback discount coefficient used to measure the weight of future feedback information, py is the output state of the next management strategy for high-risk and newly diagnosed hypertension populations, and qy is the information obtained from the next analysis.

8. An intelligent processing system for high-risk and newly diagnosed hypertension populations, characterized in that, The system is used to implement the intelligent processing method for high-risk and newly diagnosed hypertension populations as described in any one of claims 1 to 7; The system includes: The data acquisition module is used to collect patient examination data and use it as the raw dataset. The raw dataset includes: patient identification identifier, systolic blood pressure, diastolic blood pressure, body mass index, family history and diagnostic code. The preprocessing module is used to preprocess the raw data; The engine screening module is used to filter the target population using preset age thresholds and visit frequency thresholds through exclusion rules, and to construct the corresponding parameter group P={systolic blood pressure, diastolic blood pressure, body mass index, and family history quantitative parameters}. The risk analysis module is used to map family history scores to risk quantification functions using dynamic weighting coefficients Wi; The risk calculation module is used to obtain the total risk score by summing the product of the dynamic weight coefficient Wi and the quantization function Fi; Determining whether a patient belongs to a high-risk group based on diagnostic codes and total risk scores; The data push module is used to push high-risk population data to the management center server in real time via the MQTT protocol. Newly diagnosed data is desensitized by AES-256 and pushed to the regional medical platform via the HL7 FHIR standard. The remaining data is stored in the regional health record database. The push priority is positively correlated with the risk level.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the intelligent processing method for high-risk and newly diagnosed hypertension populations as described in any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent processing method for high-risk and newly diagnosed hypertension populations as described in any one of claims 1 to 7.

Images