Heart rupture early warning model training method and device and heart rupture early warning method and device

Abstract

The invention discloses a heart rupture early warning model training method and device and a heart rupture early warning method and device. The training method of the heart rupture early warning model is applied to a data processing terminal, and comprises the following steps: obtaining a training sample set which comprises a plurality of training samples, a feature set of each training sample and a heart rupture warning label, the feature set comprises a plurality of features in population features, social economic features, past medical history features, disease expression features, test and examination features, treatment intervention features, patient life features and medical institution features; and training a preset model by taking the feature set of each training sample as an independent variable and the cardiac rupture warning label of each training sample as a dependent variable to obtain a cardiac rupture early warning model. According to the embodiment of the invention, the heart rupture early warning model with high accuracy can be obtained through training.

Description

Technical Field

[0001] This application belongs to the field of data processing technology, and in particular relates to a method and apparatus for training a cardiac rupture early warning model and for providing early warning of cardiac rupture. Background Technology

[0002] Cardiac rupture is one of the most catastrophic complications following acute myocardial infarction (AMI). It often occurs suddenly, is difficult to diagnose, and has an extremely poor prognosis, resulting in a very high mortality rate and posing a significant challenge in clinical practice. To identify high-risk individuals early and implement proactive preventative measures to reduce the incidence of cardiac rupture, establishing early risk factor assessment and risk prediction mechanisms is crucial. For clinical research on cardiac rupture following acute myocardial infarction or post-stent placement, there is an urgent need to develop cardiac rupture early warning models to provide evidence for optimizing clinical decision-making and implementing personalized precision treatment. However, current cardiac rupture early warning models suffer from limited accuracy due to the relatively singular clinical indicators used in their training samples. Summary of the Invention

[0003] This application provides a method and apparatus for training a cardiac rupture early warning model and for issuing early warnings of cardiac rupture, which can train a cardiac rupture early warning model with high accuracy.

[0004] In a first aspect, embodiments of this application provide a training method for a cardiac rupture early warning model, applied to a data processing terminal, the method comprising: Obtain a training sample set, which includes the plurality of training samples, and the feature set and cardiac rupture alarm label of each training sample. The feature set includes multiple features from demographic features, socioeconomic features, past medical history features, disease manifestation features, laboratory test features, treatment intervention features, patient lifestyle features, and medical institution features. The feature set of each training sample is used as the independent variable, and the heart rupture warning label of each training sample is used as the dependent variable to train the preset model, thereby obtaining the heart rupture early warning model.

[0005] Secondly, embodiments of this application provide a training device for a cardiac rupture early warning model, applied to a data processing terminal, the device comprising: The first acquisition module is used to acquire a training sample set, which includes the plurality of training samples, and the feature set and cardiac rupture alarm label result of each training sample. The cardiac rupture alarm label result is used to characterize whether the training sample has experienced cardiac rupture. The feature set includes multiple features from demographic features, socioeconomic features, past medical history features, disease manifestation features, laboratory test features, treatment intervention features, patient-reported quality of life features, and medical institution features. The training module is used to train a preset model by using the feature set of each training sample as the independent variable and the cardiac rupture warning label of each training sample as the dependent variable, so as to obtain a cardiac rupture early warning model.

[0006] Thirdly, embodiments of this application provide a method for early warning of cardiac rupture, applied to a user terminal, the method comprising: Obtain the feature set of the target user; Input the feature set of the target user into the cardiac rupture early warning model described in any of the above items to determine the cardiac rupture risk warning value of the target user; If the heart rupture risk warning value is greater than or equal to a preset risk threshold, an alarm message is generated, which is used to prompt preventive measures to be taken for the target user.

[0007] Fourthly, embodiments of this application provide an early warning device for cardiac rupture, applied to a user terminal, the device comprising: The second acquisition module is used to acquire the feature set of the target user; The first determining module is used to input the feature set of the target user into the cardiac rupture early warning model as described above, and determine the cardiac rupture risk warning value of the target user. The generation module is used to generate alarm information when the heart rupture risk warning value is greater than or equal to a preset risk threshold. The alarm information is used to prompt preventive measures to be taken for the target user.

[0008] Fifthly, embodiments of this application provide an electronic device, the device comprising: a processor and a memory storing computer program instructions; the processor, when executing the computer program instructions, implements the training method for the cardiac rupture early warning model as described above, or the cardiac rupture early warning method.

[0009] Sixthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the training method for the cardiac rupture early warning model as described above, or the cardiac rupture early warning method.

[0010] In a seventh aspect, embodiments of this application provide a computer program product, wherein instructions in the computer program product, when executed by a processor of an electronic device, cause the electronic device to execute the training method for the heart rupture early warning model as described above, or the heart rupture early warning method.

[0011] The cardiac rupture early warning model training and cardiac rupture early warning method and apparatus of this application embodiment can acquire a training sample set, which includes multiple training samples and feature sets and cardiac rupture alarm labels for each training sample. The feature sets include demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient-reported quality of life characteristics, and medical institution characteristics. The feature sets of each training sample are used as independent variables, and the cardiac rupture alarm labels of each training sample are used as dependent variables to train a preset model, thus obtaining a cardiac rupture early warning model. Therefore, in this application embodiment, compared to existing single-indicator training methods, by integrating demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient-reported quality of life characteristics, and medical institution characteristics as an independent variable pool to train the preset model, the interaction relationships between the various features are learned, resulting in a higher accuracy of the cardiac rupture early warning model. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart illustrating a training method for a cardiac rupture early warning model provided in an embodiment of this application; Figure 2 This is a flowchart illustrating the early warning method for cardiac rupture provided in an embodiment of this application; Figure 3 This is a flowchart illustrating another training method for a cardiac rupture early warning model provided in this application embodiment; Figure 4 This is a schematic diagram of the structure of the training device for the cardiac rupture early warning model provided in this application embodiment; Figure 5 This is a schematic diagram of the structure of the early warning device for heart rupture provided in the embodiments of this application; Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0014] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0015] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0016] Cardiac rupture is one of the catastrophic complications following acute myocardial infarction. It often occurs suddenly, is difficult to diagnose, and has an extremely poor prognosis, with a very high mortality rate, making it a challenging clinical problem. To identify high-risk individuals early and implement proactive preventative measures to reduce the occurrence of cardiac rupture events, establishing early risk factor assessment and risk prediction systems is crucial. For clinical research on cardiac rupture following acute myocardial infarction or post-stent placement, there is an urgent need to develop cardiac rupture early warning models to provide evidence for optimizing clinical decision-making and implementing individualized precision treatment. However, current cardiac rupture early warning models suffer from limited accuracy due to the relatively singular clinical indicators used in their training samples.

[0017] To address the problems of the prior art, embodiments of this application provide a method and apparatus for training a cardiac rupture early warning model and for issuing early warnings of cardiac rupture. The training method for the cardiac rupture early warning model provided in this application embodiment will be described first below.

[0018] Figure 1 A flowchart illustrating a training method for a cardiac rupture early warning model according to an embodiment of this application is shown. Figure 1 As shown, a training method for a cardiac rupture early warning model is applied to a data processing terminal. The method may include the following steps S101 to S102: S101. Obtain the training sample set, which includes multiple training samples, feature sets of each training sample, and cardiac rupture alarm labels. The feature set includes multiple features from demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient lifestyle characteristics, and medical institution characteristics. S102. Using the feature set of each training sample as the independent variable and the heart rupture warning label of each training sample as the dependent variable, train the preset model to obtain the heart rupture early warning model.

[0019] The training method for the cardiac rupture early warning model in this embodiment is applied to a data processing terminal. It can acquire a training sample set, which includes multiple training samples, feature sets for each sample, and cardiac rupture alarm labels. The feature sets include demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient-reported quality of life characteristics, and medical institution characteristics. The feature sets of each training sample are used as independent variables, and the cardiac rupture alarm labels of each training sample are used as dependent variables to train a preset model, resulting in a cardiac rupture early warning model. Thus, compared to existing single-indicator training methods, this embodiment integrates demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient-reported quality of life characteristics, and medical institution characteristics as an independent variable pool to train the preset model, learning the interaction relationships between various features, thereby improving the accuracy of the cardiac rupture early warning model.

[0020] The training method for the aforementioned cardiac rupture early warning model is primarily applied to a data processing terminal. In this application, the data processing terminal can be a desktop computer, laptop computer, or other terminal device with data processing capabilities.

[0021] The specific implementation methods for each of the above steps are described below.

[0022] In S101, the aforementioned training sample set may include multiple training samples, feature sets for each training sample, and cardiac rupture warning labels. In some embodiments, the cardiac rupture warning label is used to characterize the numerical cardiac rupture risk warning value corresponding to the training sample, a specific, continuous number, such as 0.85, 0.12, 0.63, etc. This number typically represents the "probability of cardiac rupture" or "risk score" predicted by the model; the higher the value, the greater the risk. In other embodiments, the cardiac rupture warning label is used to characterize the cardiac rupture risk warning content description text corresponding to the training sample, a piece of natural language text, such as: "The patient's heart rate is abnormally elevated with slight ST segment depression; continuous monitoring and a repeat electrocardiogram are recommended." This not only includes risk assessment but also explanations of the risk causes and possible recommendations; it is a richer and more readable output format.

[0023] The aforementioned feature set includes multiple features from the following categories: demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient lifestyle characteristics, and medical institution characteristics. Demographic characteristics can describe an individual's basic biological attributes and social markers, such as age, sex, race, ethnicity, height, and weight. Socioeconomic characteristics can describe an individual's socioeconomic status, education level, and living resources, such as marital status, education level, income level, and place of residence. Past medical history characteristics can describe an individual's previously diagnosed diseases and health conditions, such as past illnesses (including coronary heart disease, hypertension, diabetes, hyperlipidemia, chronic kidney disease, atrial fibrillation, myocardial infarction, acute coronary artery disease, heart failure, stroke, ischemic stroke, and hemorrhagic stroke), as well as surgical history, allergy history, and family history. Disease manifestation characteristics can describe the subjective symptoms and objective signs of the patient's current disease onset, such as the type of myocardial infarction, cardiac function classification, and the number of hours from symptom onset to hospital admission. Laboratory test characteristics can be objective quantitative indicators from in vitro diagnostic experiments, such as white blood cell count upon admission, biomarkers, medical institution, electrocardiogram, echocardiography, and other laboratory tests. Treatment intervention characteristics can describe the treatment measures already implemented for the patient, such as whether coronary artery bypass grafting was performed in-hospital, whether angiotensin-converting enzyme inhibitors or angiotensin II receptor antagonists were used upon admission, whether receptor blockers were used upon admission, whether direct percutaneous coronary intervention was performed, and the length of hospital stay. Patient lifestyle characteristics can describe the patient's personal lifestyle habits and behaviors, usually reported by the patient themselves, such as smoking history, alcohol consumption history, exercise habits, and dietary preferences. Medical institution characteristics can describe the attributes of the institution providing medical services, such as hospital level, hospital size, professional capabilities, and geographical location.

[0024] The aforementioned training sample set can, for example, be based on a retrospective study of "Evaluation of Coronary Artery Disease Medical Outcomes and Clinical Translation Research." This study employed a two-stage random sampling design. In the first stage, stratified random sampling was used to determine the hospitals, and in the second stage, systematic sampling was used to determine the study medical records. A cohort of AMI inpatients with national and regional representativeness was established in 162 secondary and tertiary hospitals randomly selected from 31 provinces, autonomous regions, and municipalities across China. Approximately 16,000 inpatients who experienced acute AMI in 2001, 2005, and 2010 were included, among whom approximately 80 experienced in-hospital cardiac rupture.

[0025] In S102, the aforementioned preset model, exemplarily, can be a logistic regression (LR) model or a machine learning model. In the embodiments of this application, the machine learning model can be at least one of the following: Adaptive Boosting (AdaBoost), Category Feature Gradient Boosting (CatBoost), Light Gradient Boosting Machine (LightGBM), Random Forest (RF), Support Vector Machine (SVM), Stochastic Gradient Boosting Tree (SGBT), Extreme Gradient Boosting Tree (XGBT), Neural Network (NNET), Regularized Generalized Linear Model (GLMNET), Naive Bayes (NB).

[0026] The above-mentioned method uses the feature set of each training sample as the independent variable and the cardiac rupture warning label of each training sample as the dependent variable to train a preset model to obtain a cardiac rupture warning model. For example, it can use the feature set of each training sample as the independent variable and the cardiac rupture warning label of each training sample as the dependent variable to train a preset logistic regression model to obtain a first prediction model; use the feature set of each training sample as the independent variable and the cardiac rupture warning label of each training sample as the dependent variable to train a preset machine learning model to obtain a second prediction model; calculate the discrimination and calibration of the first prediction model and the second prediction model; and determine the cardiac rupture warning model based on the discrimination and calibration of the first prediction model and the second prediction model. The cardiac rupture warning model is the prediction model with higher discrimination or higher calibration among the first prediction model and the second prediction model. Alternatively, the feature set of each training sample can be used as the independent variable, and the cardiac rupture alarm label of each training sample can be used as the dependent variable to train the preset model and obtain the initial warning model; obtain the validation sample set, which includes multiple validation samples and the feature set and cardiac rupture alarm label of each validation sample; input the feature set of each validation sample in the validation sample set into the initial warning model to determine the prediction result of each validation sample; calculate the discrimination and calibration of the initial warning model based on the cardiac rupture alarm label and prediction result of each validation sample in the validation sample set; if the discrimination and calibration of the initial warning model both meet the preset interval, the initial warning model is determined to be the cardiac rupture warning model.

[0027] As one implementation of this application, in order to improve the model's generalization ability and robustness, the method may further include the following before S102: The features in the feature set are filtered to obtain the target feature set, which includes features with a frequency greater than a preset frequency threshold and / or features with a missing rate less than a preset missing rate threshold. Specifically, S102 mentioned above may include: The target feature set of each training sample is used as the independent variable, and the heart rupture warning label of each training sample is used as the dependent variable to train the preset model and obtain the heart rupture early warning model.

[0028] The aforementioned target feature set may include features whose frequency of occurrence is greater than a preset frequency threshold, for example, features whose frequency of occurrence is greater than 1%, and / or features whose missing rate is less than a preset missing rate threshold, for example, features whose missing rate is less than 10%. In this embodiment, the preset frequency threshold and preset missing rate threshold are not limited to 1% and 10%, respectively, and can be set according to actual needs; no specific limitation is made here. Furthermore, for features with a missing rate of less than 10%, multiple imputation methods can be used for imputation to ensure unbiased estimation.

[0029] The above-described filtering of features in the feature set to obtain the target feature set can, for example, directly filter out features whose frequency of occurrence is greater than a preset frequency threshold and / or whose missing rate is less than a preset missing rate threshold. Alternatively, it can filter out features whose frequency of occurrence is greater than a preset frequency threshold and / or whose missing rate is less than a preset missing rate threshold to obtain a candidate feature set, which includes multiple candidate features; calculate the probability of each candidate feature in the candidate feature set; display the probability of each candidate feature on the display operation panel for the operator to view; and, in response to the operator's confirmation operation, determine at least one candidate feature in the candidate feature set as the target feature set.

[0030] In this embodiment, by filtering out features that appear more frequently than a preset frequency threshold and / or features with a missing rate less than a preset missing rate threshold from the feature set, a target feature set is obtained and used in the heart rupture early warning model. This avoids overfitting to rare features, enhances the model's generalization ability, ensures the reliability of features, and improves the prediction confidence and robustness of the model training.

[0031] In some embodiments, the above-mentioned filtering of features in the feature set to obtain the target feature set may specifically include: The features in the feature set are filtered to obtain a candidate feature set, which includes multiple candidate features; Calculate the probability of each candidate feature in the candidate feature set; The probability of each candidate feature is displayed on the operation panel for operators to view; In response to the operator's confirmation input, at least one candidate feature in the candidate feature set is identified as the target feature set.

[0032] The above-mentioned filtering of features in the feature set to obtain a candidate feature set can specifically be done by filtering out features in the feature set whose frequency of occurrence is greater than a preset frequency threshold and / or whose missing rate is less than a preset missing rate threshold.

[0033] The above-mentioned calculation of the probability of each candidate feature in the candidate feature set can, for example, be performed using Lasso regression, or it can be performed using mutual information score and random forest importance. In this embodiment, the calculation method for the probability of each candidate feature in the candidate feature set is not limited to the above, and can be set according to the actual needs of the operator; no specific limitation is made here.

[0034] The personnel described above can be clinical experts with knowledge of cardiac rupture.

[0035] In this embodiment, a candidate feature set is obtained by first screening the features in the feature set, then the probability of each candidate feature in the candidate feature set is calculated, and then at least one candidate feature in the candidate feature set is determined as the target feature set through the confirmation operation of the operator. This dual mechanism of "machine screening + manual confirmation" not only ensures the objectivity of feature selection, but also ensures clinical rationality, thereby improving the accuracy of the target feature set.

[0036] In some embodiments, the demographic characteristics in the target feature set include age and sex; the medical history characteristics include a history of coronary artery disease; the disease manifestation characteristics include the type of myocardial infarction; the laboratory test characteristics include the white blood cell count upon admission; and the treatment intervention characteristics include whether angiotensin-converting enzyme inhibitors or angiotensin II receptor antagonists were used upon admission, and whether direct percutaneous coronary intervention was performed. In some embodiments, S102 may specifically include: The feature set of each training sample is used as the independent variable, and the heart rupture warning label of each training sample is used as the dependent variable. The preset logistic regression model is trained to obtain the first prediction model. The feature set of each training sample is used as the independent variable, and the heart rupture warning label of each training sample is used as the dependent variable to train the preset machine learning model and obtain the second prediction model. Calculate the discrimination and calibration of the first and second prediction models; Based on the discrimination and calibration of the first and second prediction models, a cardiac rupture early warning model is determined. The cardiac rupture early warning model is the prediction model with higher discrimination or higher calibration among the first and second prediction models.

[0037] The aforementioned preset machine learning model can be at least one of the following: Adaptive Boosting (AdaBoost), Category Feature Gradient Boosting (CatBoost), Light Gradient Boosting Machine (LightGBM), Random Forest (RF), Support Vector Machine (SVM), Stochastic Gradient Boosting Tree (SGBT), Extreme Gradient Boosting Tree (XGBT), Neural Network (NNET), Regularized Generalized Linear Model (GLMNET), Naive Bayes (NB).

[0038] The aforementioned discrimination index describes the model's ability to correctly distinguish individuals who experienced an event (positive) from those who did not experience an event (negative). The most commonly used metric for discrimination is the Area Under the ROC Curve (AUC), also known as the C-statistic. The AUC value is obtained by plotting an ROC curve and calculating the area under the curve.

[0039] The aforementioned calibration degree can be used to describe the consistency between the risk probability predicted by the model and the actual risk. The most commonly used calculation methods are the calibration plot and the Brier score, where a lower Brier score indicates a higher calibration degree.

[0040] The aforementioned cardiac rupture early warning model is the prediction model with higher discrimination or higher calibration among the first and second prediction models.

[0041] The above-mentioned determination of the cardiac rupture early warning model based on the discrimination and calibration of the first and second prediction models can be exemplified as follows: if the discrimination indices of the first and second prediction models are similar, the prediction model with the better calibration index is preferentially identified as the cardiac rupture early warning model; if the discrimination index of the first prediction model is significantly better than that of the second prediction model, the prediction model with the better discrimination index is preferentially identified as the cardiac rupture early warning model. The optimal model is one with both high discrimination and high calibration.

[0042] In this embodiment, by using the feature set of each training sample as the independent variable and the cardiac rupture warning label of each training sample as the dependent variable, a preset logistic regression model and a machine learning model are trained to obtain a first prediction model and a second prediction model. The discrimination and calibration of the first prediction model and the second prediction model are calculated to evaluate the model. Then, the prediction model corresponding to the higher discrimination or higher calibration is determined as the cardiac rupture warning model, which can ensure that the trained cardiac rupture warning model has better performance.

[0043] In some embodiments, the above-described S102 may specifically include: The feature set of each training sample is used as the independent variable, and the heart rupture warning label of each training sample is used as the dependent variable to train the preset model and obtain the initial warning model. Obtain a verification sample set, which includes multiple verification samples and the feature set and heart rupture alarm label for each verification sample; The feature sets of each validation sample in the validation sample set are input into the initial early warning model to determine the prediction results of each validation sample; Based on the cardiac rupture alarm labels and prediction results of each validation sample in the validation sample set, the discrimination and calibration of the initial early warning model are calculated. If the discrimination and calibration of the initial warning model both meet the preset range, the initial warning model is determined to be a cardiac rupture warning model.

[0044] The aforementioned validation sample set may include multiple validation samples, each with its own feature set and cardiac rupture alarm label. It should be noted that the validation sample set is independent of the training sample set. The method for obtaining the validation sample set is the same as that for obtaining the training sample set, and will not be repeated here.

[0045] The aforementioned preset interval can be an interval set by the user according to actual needs, and is not limited to a fixed interval. This application embodiment does not impose any specific limitations. It should be noted that the discrimination and calibration can correspond to different preset intervals.

[0046] In this embodiment, an initial warning model is first trained using a training sample set to obtain an initial warning model. Then, the initial warning model is verified using a verification sample set, and its discrimination and calibration are calculated. Only when both discrimination and calibration meet the preset range are the initial warning model determined as a heart rupture warning model, which can ensure the reliability of the trained heart rupture warning model.

[0047] As another implementation of this application, in order to provide subsequent early warning of cardiac rupture to the patient, after S102 above, the method may further include: The weights of each feature are determined based on the model coefficients of each feature in the cardiac rupture early warning model. Establish multiple risk levels and corresponding risk score ranges for each risk level.

[0048] The weight of each feature is determined based on the model coefficients of each feature in the cardiac rupture early warning model. For example, the model coefficients of each feature in the cardiac rupture early warning model can be linearly amplified into integer weights, and the total weight of each feature in the cardiac rupture early warning model is 100.

[0049] The aforementioned establishment of multiple risk levels and corresponding risk scoring ranges for each risk level can, for example, involve establishing three risk levels: high, medium, and low. The risk scoring range for the high-risk level could be (80-100), the risk scoring range for the medium-risk level could be (60-80), and the risk scoring range for the low-risk level could be [0-60]. Of course, in this embodiment, the multiple risk levels and corresponding risk scoring ranges are not limited to this and can be set according to the user's actual needs; no specific limitations are made here.

[0050] In this embodiment of the application, the weight of each feature is determined based on the model coefficient of each feature in the cardiac rupture early warning model, and multiple risk levels and risk score intervals corresponding to each risk level are established so as to provide early warning of cardiac rupture for patients in the future.

[0051] Figure 2 A schematic flowchart of a method for early warning of cardiac rupture according to an embodiment of this application is shown. Figure 2 As shown, a method for early warning of cardiac rupture, applied to a user terminal, may include the following steps S201 to S203: S201. Obtain the feature set of the target user; S202. Input the feature set of the target user into the cardiac rupture early warning model of any of the above items to determine the cardiac rupture risk warning value of the target user. S203. When the risk warning value for heart rupture is greater than or equal to the preset risk threshold, an alarm message is generated. The alarm message is used to prompt preventive measures to be taken for the target user.

[0052] The cardiac rupture early warning method of this application embodiment is applied to a user terminal. It can acquire the feature set of the target user and input it into the cardiac rupture early warning model to accurately obtain the cardiac rupture risk warning value of the target user. Then, when the cardiac rupture risk warning value is greater than or equal to a preset risk threshold, alarm information is generated in a timely manner, thereby prompting preventive measures to be taken for the target user.

[0053] The aforementioned early warning method for cardiac rupture is primarily applied to user terminals. In this application, the user terminal can be a terminal device that users directly contact and use to obtain services or complete tasks, such as a clinical medical terminal within a hospital, a patient and home terminal, or a telemedicine terminal.

[0054] The specific implementation methods for each of the above steps are described below.

[0055] In S201, the aforementioned target user can be the user who received the heart rupture warning in this instance.

[0056] The aforementioned acquisition of the target user's feature set can be exemplified by manually collecting the target user's feature set and inputting it into the user terminal, or by retrieving the target user's feature set from a database connected to the user terminal's data.

[0057] In S202, the aforementioned cardiac rupture early warning model can be obtained from the training method of the aforementioned cardiac rupture early warning model.

[0058] The above-mentioned input of the target user's feature set into any of the above-mentioned cardiac rupture early warning models to determine the target user's cardiac rupture risk warning value can be exemplified by directly inputting the target user's feature set into any of the above-mentioned cardiac rupture early warning models to determine the target user's cardiac rupture risk warning value; or, it can also be that the target user's target feature set is input into any of the above-mentioned cardiac rupture early warning models to determine the target user's cardiac rupture risk warning value.

[0059] In S203, the aforementioned alarm information can be used to prompt preventative measures to be taken against the target user.

[0060] The above-mentioned alarm information is generated when the heart rupture risk warning value is greater than or equal to the preset risk threshold. For example, when the heart injury risk warning value is greater than or equal to the preset risk threshold, the alarm information can be generated and sounded on the clinical medical care terminal in the hospital; or the alarm information can be generated and displayed on the display and operation interface of the patient and home terminal.

[0061] As another implementation of this application, in order to further enhance the early warning capability for cardiac rupture, the method may further include the following before S203: Based on the target user's feature set and the preset weights of each feature, the risk score of the target user is calculated. The risk score is mapped to a risk score range corresponding to multiple preset risk levels to determine the target risk level of the target user. Specifically, S203 mentioned above may include: An alarm message is generated when the risk warning value for cardiac rupture is greater than or equal to the preset risk threshold, and / or the target risk level meets the preset level.

[0062] The weights of the above features can be determined based on the model coefficients of each feature in the cardiac rupture early warning model.

[0063] The target risk level of the aforementioned target user is the risk level corresponding to the interval in which the target user's risk score falls within the risk score interval corresponding to multiple risk levels.

[0064] The aforementioned preset level, exemplarily, can be a high-risk level. Of course, the preset levels in this embodiment are not limited to this, and can be set according to the user's actual needs, which is not specifically limited here.

[0065] In this embodiment, the risk score of the target user can be calculated based on the feature set of the target user and the preset weight of each feature, and mapped to the risk score interval corresponding to multiple preset risk levels to determine the target risk level of the target user. The warning is then issued by two methods: whether the heart rupture risk warning value is greater than or equal to the preset risk threshold and whether the target risk level meets the preset level, thereby further enhancing the warning capability of heart rupture.

[0066] To facilitate understanding of the training method of the cardiac rupture early warning model in the embodiments of this application, such as Figure 3 As shown, the practical application process of the training method for this cardiac rupture early warning model is explained in detail below: This application provides a system for constructing a cardiac rupture prediction model, applied to a data information processing terminal (i.e., the aforementioned data processing terminal), and applicable to the training method of the aforementioned cardiac rupture early warning model. The system for constructing the cardiac rupture prediction model includes the following major modules: 1. Data Acquisition Module: Connected to the central control module, the data acquisition module executes the data acquisition and variable definition steps in S301, receives the instruction form connected to the central control module, first sends the request information to the switch, the switch sends it to the router, the router sends it to the DNS server, the DNS calls the vertical search engine to establish a search engine in the Internet, and finds the data information to be collected through network radar and automatic indexing technology, and then calls information tracking and automatic sorting to classify the information.

[0067] 2. Prediction Module: Connected to the central control module, this module predicts the collected sample data (equivalent to the training sample set mentioned above). The prediction module executes steps S302-S304, receiving data from the central control module and inputting it into the prediction program. The prediction program contains a preset prediction algorithm, and the data is processed by the algorithm to obtain the prediction result, as detailed below: 1) Predictive Factor Screening Module: This module executes the predictive factor selection steps in S302, primarily for initial screening of predictive factors. This initial screening is based on clinical significance and a missing value rate <10%, converting continuous variables into categorical variables. Risk factors for cardiac rupture are then identified through univariate analysis and multivariate logistic regression. Predictive factors (equivalent to the aforementioned candidate feature set) are initially selected via stepwise logistic regression, and the probability of each candidate predictive factor is calculated using Lasso regression. Finally, clinical expert knowledge is incorporated for manual verification. This dual mechanism of "machine screening + manual verification" ensures both the objectivity of feature selection and clinical rationality, thus improving the factor selection for predicting the risk of cardiac rupture in AMI patients.

[0068] In this application, the initial screening of predictive factors can also include potential predictive factors that are clinically significant, reliable, easy to collect, statistically significant, and have a frequency of more than 1%, and will be selected as predictive factors.

[0069] The selected factors (equivalent to the target feature set mentioned above) may, for example, include: age, gender, history of coronary artery disease, type of myocardial infarction (STEMI), white blood cell count upon admission, use of ACEI / ARB drugs upon admission, and whether primary PCI was performed.

[0070] This application integrates information such as demographic characteristics, socioeconomic characteristics, past medical history, treatment status, patient-reported quality of life, laboratory tests, biomarkers, medical institution characteristics, electrocardiogram characteristics, echocardiography, and coronary intervention as a variable pool. Variables with more than 10% missing data or a frequency of less than 1% are directly removed. For variables with less than 10% missing data, multiple imputation methods are used to imputate them to ensure the unbiasedness of the estimation.

[0071] For example, potential predictors were identified based on clinical relevance and reliability, with all variables having a frequency exceeding 1%. Candidate predictors included the following dimensions: patient demographic characteristics (age, sex, marital status, smoking status, hospitalization costs); medical history (myocardial infarction, acute coronary artery disease, percutaneous coronary intervention, coronary artery bypass grafting, heart failure, atrial fibrillation, ischemic stroke, hemorrhagic stroke, hypertension, diabetes, dyslipidemia, chronic kidney disease); and in-hospital clinical characteristics (time from symptom onset to admission, Killip functional classification, duration of chest discomfort, duration of ischemia, complete left bundle branch block, thrombolytic therapy, heart rate, blood pressure, body mass index, heart failure with reduced ejection fraction, refusal of PCI). The following criteria were considered: refusal of CABG treatment, failure to undergo direct PCI, ST-segment elevation myocardial infarction, and length of hospital stay; medication use (aspirin, clopidogrel, or other antiplatelet drugs used before admission; in-hospital treatment within 24 hours: aspirin, beta-blockers, clopidogrel, angiotensin-converting enzyme inhibitors / angiotensin II receptor blockers, statins; continuous in-hospital treatment: ACEI / ARB, statins, aspirin, beta-blockers, clopidogrel); and laboratory tests (white blood cell count, hemoglobin, estimated glomerular filtration rate, blood glucose, creatinine, blood urea nitrogen, high-sensitivity C-reactive protein, low-density lipoprotein, troponin T, and troponin I).

[0072] Missing data in the training dataset are imputed using the R language's 'mice' package.

[0073] Most continuous variables have been converted to categorical format for easier clinical application. Candidate predictors after imputation are used to screen the final predictors of the predictive model.

[0074] In the training dataset, a structured approach was used for variable selection: First, highly collinear variables were excluded based on the criteria of Pearson correlation coefficient > 0.75 or variance inflation factor > 10; second, variables significantly associated with the outcome (P-value < 0.05) were selected using univariate logistic regression; third, significant variables from the univariate analysis were further screened using the LASSO method; fourth, a stepwise multivariate logistic regression model was established, with a variable inclusion P-value threshold of 0.2 and a exclusion threshold of 0.1; finally, clinical expert knowledge was incorporated for manual verification, and clinically important variables were added to the final predictor list to ensure clinical significance.

[0075] 2) Cardiac Rupture Prediction Model Construction Module: This module can execute the steps of building the heart rupture prediction model in S303. It will use Logistic Regression (LR) and machine learning algorithms to build the prediction model. Specific machine learning algorithms include Adaptive Boosting (AdaBoost), CatBoost, Light Gradient Boosting Machine (LightGBM), Random Forest (RF), Support Vector Machine (SVM), Stochastic Gradient Boosting Tree (SGBT), Extreme Gradient Boosting Tree (XGBT), Neural Network (NNET), Regularized Generalized Linear Model (GLMNET), Naive Bayes (NB), etc. Among them, SVM classifies by finding the optimal hyperplane, suitable for linear and nonlinear problems, and excels at high-dimensional data but is sensitive to noise; RF is an ensemble method that outputs results through voting among multiple decision trees, is resistant to overfitting, and can assess feature importance; AdaBoost trains weak classifiers by iteratively adjusting sample weights, suitable for simple models but sensitive to outliers; NNet simulates neuron structures and can learn complex patterns but requires a large amount of data and parameter tuning; GLM extends linear regression and supports various distributions (such as logistic regression); LR is used for binary classification, and its output probability has strong interpretability; SGBT and XGBT are used for gradient-optimized decision tree ensembles, are efficient, and widely used in competitions; NB is based on Bayes' theorem, suitable for text classification but assumes feature independence. Each method has its advantages, and the choice should be based on the model's performance.

[0076] In addition, the study will allocate integer scores based on the weighting of model coefficients to establish a risk scoring system and conduct corresponding risk stratification.

[0077] 3) Evaluation module for cardiac rupture prediction model: This module evaluates and compares the discriminant and calibration scores of at least two established cardiac rupture prediction models using methods such as the area under the receiver operating characteristic curve, C-index, and Brier score, comparing parameters such as discriminant AUC and calibration score.

[0078] 4) Cardiac rupture prediction model validation module: Given the low incidence of endpoint events and limited sample size, this module can perform the cardiac rupture prediction model validation steps in S304. It is primarily used for internal validation of the prediction model, which will be evaluated and the corrected AUC value calculated through 1000 Bootstrap sampling simulations (equivalent to the aforementioned validation sample set). The study will conduct cross-validation and externally validate the model performance in an independent population cohort.

[0079] 3. The central control module connects to the data collection module, prediction module, model building module, cloud storage module, local storage module, and data encryption module. The main controller acquires external signals, analyzes and processes them, and outputs them to the output channel. When an analog output is required, the system converts it into a standard electrical signal via a D / A converter for output to control the normal operation of each module.

[0080] 4. Storage module, connected to the central control module, is used to store the collected data. The data connection between the two modules is established through an SQL interface, and the data is stored using an SQL service database. The data acquisition and reading module packages the converted data and sends it to the central control module. The central control module then performs tag operations to transmit the data to the SQL service database and the storage module.

[0081] These four modules enable the construction of a cardiac rupture prediction model. This model facilitates the development and application of the risk score in step S305, identifying key factors that may influence clinical outcomes of cardiac rupture, quantifying clinical outcome risks, and performing risk stratification and classification management. This provides evidence for optimizing clinical decision-making and implementing personalized precision treatment. Against the backdrop of rapid development in artificial intelligence analysis technology, exploring strategies for predicting and assessing cardiac rupture risks not only helps to overcome the limitations of traditional closed-loop clinical diagnosis and treatment models but also fills the research gap in the field of cardiac rupture risk prediction in China, providing new insights for improving cardiovascular disease management.

[0082] Based on the training method for the cardiac rupture early warning model provided in the above embodiments, this application also provides a specific implementation of the training device for the cardiac rupture early warning model. Please refer to the following embodiments.

[0083] like Figure 4 As shown, the training device 400 for the heart rupture early warning model provided in this application embodiment is applied to a data processing terminal and may include the following modules: a first acquisition module 401 and a training module 402.

[0084] The first acquisition module 401 is used to acquire a training sample set, which includes multiple training samples, a feature set of each training sample, and a cardiac rupture alarm label result. The cardiac rupture alarm label result is used to characterize whether the training sample has experienced cardiac rupture. The feature set includes multiple features from demographic features, socioeconomic features, past medical history features, disease manifestation features, laboratory test features, treatment intervention features, patient-reported quality of life features, and medical institution features. Training module 402 is used to train a preset model by using the feature set of each training sample as the independent variable and the heart rupture warning label of each training sample as the dependent variable, so as to obtain a heart rupture early warning model.

[0085] The training device for the cardiac rupture early warning model in this embodiment is applied to a data processing terminal and can acquire a training sample set. The training sample set includes multiple training samples, a feature set for each training sample, and a cardiac rupture alarm label. The feature set includes demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient-reported quality of life characteristics, and medical institution characteristics. The feature set of each training sample is used as the independent variable, and the cardiac rupture alarm label of each training sample is used as the dependent variable to train a preset model, thus obtaining the cardiac rupture early warning model. Therefore, in this embodiment, compared to existing single-indicator training methods, by integrating demographic characteristics, socioeconomic characteristics, past medical history characteristics, disease manifestation characteristics, laboratory test characteristics, treatment intervention characteristics, patient-reported quality of life characteristics, and medical institution characteristics as an independent variable pool to train the preset model, the interaction relationships between the various features are learned, resulting in a higher accuracy of the cardiac rupture early warning model.

[0086] As one implementation of this application, in order to improve the model's generalization ability and robustness, the above-mentioned device 400 may further include: The filtering module is used to filter features in the feature set to obtain a target feature set. The target feature set includes features with a frequency greater than a preset frequency threshold and / or features with a missing rate less than a preset missing rate threshold. The aforementioned training module 402 is specifically used to train a preset model by taking the target feature set of each training sample as the independent variable and the heart rupture warning label of each training sample as the dependent variable, so as to obtain a heart rupture early warning model.

[0087] In some embodiments, the above-mentioned filtering module may specifically include: The filtering unit is used to filter the features in the feature set to obtain a candidate feature set, which includes multiple candidate features. The first calculation unit is used to calculate the probability of each candidate feature in the candidate feature set; The display unit is used to show the probability of each candidate feature on the display operation panel for the operator to view; The first confirmation unit is used to determine at least one candidate feature in the candidate feature set as the target feature set in response to the confirmation operation input by the operator.

[0088] In some embodiments, the above target features include age and sex for demographic characteristics, history of coronary heart disease for medical history, type of myocardial infarction for disease manifestation characteristics, white blood cell count upon admission for laboratory tests, and treatment intervention characteristics for admission including whether angiotensin-converting enzyme inhibitors or angiotensin II receptor antagonists were used or whether direct percutaneous coronary intervention was performed.

[0089] In some embodiments, the training module 402 described above may specifically include: The first training unit is used to train a preset logistic regression model by taking the feature set of each training sample as the independent variable and the heart rupture warning label of each training sample as the dependent variable, so as to obtain the first prediction model. The second training unit is used to train the preset machine learning model by taking the feature set of each training sample as the independent variable and the heart rupture warning label of each training sample as the dependent variable, and to obtain the second prediction model. The second calculation unit is used to calculate the discrimination and calibration of the first prediction model and the second prediction model. The second determining unit is used to determine the cardiac rupture early warning model based on the discrimination and calibration of the first prediction model and the second prediction model. The cardiac rupture early warning model is the prediction model with higher discrimination or higher calibration among the first prediction model and the second prediction model.

[0090] In some embodiments, the training module 402 described above may specifically include: The third training unit is used to train the preset model by taking the feature set of each training sample as the independent variable and the heart rupture warning label of each training sample as the dependent variable, so as to obtain the initial warning model. The acquisition unit is used to acquire the verification sample set, which includes multiple verification samples and the feature set and heart rupture alarm label of each verification sample. The third determining unit is used to input the feature set of each verification sample in the verification sample set into the initial warning model and determine the prediction result of each verification sample. The third calculation unit is used to calculate the discrimination and calibration of the initial early warning model based on the cardiac rupture alarm labels and prediction results of each validation sample in the validation sample set. The fourth determining unit is used to determine the initial warning model as a heart rupture warning model when both the discrimination and calibration of the initial warning model meet the preset range.

[0091] As another implementation of this application, in order to provide subsequent early warning of cardiac rupture to the patient, the device 400 may further include: The second determination module is used to determine the weight of each feature based on the model coefficients of each feature in the cardiac rupture early warning model. The module is used to establish multiple risk levels and the corresponding risk score ranges for each risk level.

[0092] like Figure 5As shown, the heart rupture early warning device 500 provided in this application embodiment is applied to a user terminal and may include the following modules: a second acquisition module 501, a first determination module 502 and a generation module 503.

[0093] The second acquisition module 501 is used to acquire the feature set of the target user; The first determining module 502 is used to input the feature set of the target user into the cardiac rupture early warning model of any of the above items, and determine the cardiac rupture risk warning value of the target user. The generation module 503 is used to generate alarm information when the heart rupture risk warning value is greater than or equal to a preset risk threshold. The alarm information is used to prompt preventive measures to be taken for the target user.

[0094] The heart rupture early warning device of this application embodiment is applied to a user terminal. It can acquire the feature set of the target user and input it into the heart rupture early warning model terminal to accurately obtain the heart rupture risk warning value of the target user. Then, when the heart rupture risk warning value is greater than or equal to a preset risk threshold, alarm information is generated in a timely manner, thereby prompting the target user to take preventive measures.

[0095] As another implementation of this application, in order to further enhance the early warning capability for cardiac rupture, the aforementioned device 500 may further include: The calculation module is used to calculate the risk score of the target user based on the target user's feature set and the preset weights of each feature; The third determination module is used to map the risk score to a risk score range corresponding to multiple preset risk levels, and determine the target risk level of the target user. The aforementioned generation module 503 is specifically used to generate alarm information when the heart rupture risk warning value is greater than or equal to a preset risk threshold, and / or the target risk level meets the preset level.

[0096] Figure 6 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0097] An electronic device may include a processor 601 and a memory 602 storing computer program instructions.

[0098] Specifically, the processor 601 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0099] Memory 602 may include mass storage for data or instructions. For example, and not limitingly, memory 602 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 602 may include removable or non-removable (or fixed) media. Where appropriate, memory 602 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 602 is non-volatile solid-state memory.

[0100] In a particular embodiment, memory 602 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Thus, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this disclosure.

[0101] The processor 601 reads and executes computer program instructions stored in the memory 602 to implement any of the training methods for the cardiac rupture early warning model or the early warning method for cardiac rupture in the above embodiments.

[0102] In one example, the electronic device may also include a communication interface 603 and a bus 610. For example, Figure 6 As shown, the processor 601, memory 602, and communication interface 603 are connected through bus 610 and complete communication with each other.

[0103] The communication interface 603 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0104] Bus 610 includes hardware, software, or both, that couples components of an electronic device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 610 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0105] The electronic device can execute the training method of the cardiac rupture early warning model or the cardiac rupture early warning method in the embodiments of this application, thereby achieving a combination of Figure 1 and Figure 4 The training method and apparatus for the cardiac rupture early warning model described above, or the above-mentioned Figure 2 and Figure 5 The method and device described are for early warning of cardiac rupture.

[0106] Furthermore, in conjunction with the training method for the cardiac rupture early warning model and the cardiac rupture early warning method in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any one of the cardiac rupture early warning model training methods or cardiac rupture early warning methods in the above embodiments.

[0107] This application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the training methods for a cardiac rupture early warning model or a cardiac rupture early warning method in the above embodiments.

[0108] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0109] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0110] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0111] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0112] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for training a heart rupture early warning model, characterized in that, The method is applied to a data processing terminal and comprises the following steps: Obtaining a training sample set, wherein the training sample set comprises a plurality of training samples, and a feature set and a heart rupture warning label of each training sample, and the feature set comprises a plurality of features such as demographic characteristics, socioeconomic characteristics, previous medical history characteristics, disease manifestation characteristics, laboratory examination characteristics, treatment intervention characteristics, patient life characteristics and medical institution characteristics; Training a preset model by taking the feature set of each training sample as an independent variable and the heart rupture warning label of each training sample as a dependent variable to obtain a heart rupture early warning model.

2. The method of claim 1, wherein, Before the step of training the preset model by taking the feature set of each training sample as an independent variable and the heart rupture warning label of each training sample as a dependent variable to obtain the heart rupture early warning model, the method further comprises the following steps: Screening features in the feature set to obtain a target feature set, wherein the target feature set comprises features with a frequency greater than a preset frequency threshold and / or features with a missing rate less than a preset missing rate threshold; The step of training the preset model by taking the feature set of each training sample as an independent variable and the heart rupture warning label of each training sample as a dependent variable to obtain the heart rupture early warning model comprises the following steps: Training the preset model by taking the target feature set of each training sample as an independent variable and the heart rupture warning label of each training sample as a dependent variable to obtain the heart rupture early warning model.

3. The method of claim 2, wherein, The step of screening features in the feature set to obtain the target feature set comprises the following steps: Screening features in the feature set to obtain a candidate feature set, wherein the candidate feature set comprises a plurality of candidate features; Calculating the probability of each candidate feature in the candidate feature set; Displaying the probability of each candidate feature on a display operation panel for an operator to view; In response to a confirmation operation input by the operator, determining at least one candidate feature in the candidate feature set as the target feature set.

4. The method of claim 2, wherein, The demographic characteristics in the target feature set comprise age and gender, the previous medical history characteristics comprise previous coronary heart disease history, the disease manifestation characteristics comprise heart infarction type, the laboratory examination characteristics comprise admission white blood cell count, and the treatment intervention characteristics comprise whether to use angiotensin converting enzyme inhibitor or angiotensin II receptor antagonist drug and whether to perform direct percutaneous coronary intervention treatment.

5. The method of claim 1, wherein, The step of training the preset model by taking the feature set of each training sample as an independent variable and the heart rupture warning label of each training sample as a dependent variable to obtain the heart rupture early warning model comprises the following steps: Training a preset logistic regression model by taking the feature set of each training sample as an independent variable and the heart rupture warning label of each training sample as a dependent variable to obtain a first prediction model; Training a preset machine learning model by taking the feature set of each training sample as an independent variable and the heart rupture warning label of each training sample as a dependent variable to obtain a second prediction model; Calculating the discrimination and calibration of the first prediction model and the second prediction model; and determine a heart rupture early warning model according to the discrimination and calibration of the first prediction model and the second prediction model, the heart rupture early warning model being the prediction model corresponding to higher discrimination or higher calibration.

6. The method of claim 1, wherein, The method further comprises: training a preset model according to the feature set of each training sample as an independent variable and the heart rupture alarm label of each training sample as a dependent variable to obtain an initial early warning model; obtaining a verification sample set, the verification sample set comprising a plurality of verification samples, a feature set of each verification sample, and a heart rupture alarm label of each verification sample; inputting the feature set of each verification sample in the verification sample set into the initial early warning model to determine a prediction result of each verification sample; calculating the discrimination and calibration of the initial early warning model according to the heart rupture alarm label and the prediction result of each verification sample in the verification sample set; determining the initial early warning model as the heart rupture early warning model when the discrimination and calibration of the initial early warning model both satisfy a preset interval.

7. The method of claim 1, wherein, The method further comprises: determining the weight of each feature according to the model coefficient of each feature in the heart rupture early warning model; establishing a plurality of risk levels and a risk score interval corresponding to each risk level.

8. A method of alerting a heart rupture, characterized by, The method applied to a user terminal comprises: obtaining a feature set of a target user; inputting the feature set of the target user into the heart rupture early warning model to determine a heart rupture risk early warning value of the target user; generating an alarm information when the heart rupture risk early warning value is greater than or equal to a preset risk threshold, the alarm information being used to prompt taking preventive measures for the target user.

9. The method of claim 8, wherein, The method further comprises: calculating a risk score of the target user according to the feature set of the target user and the preset weight of each feature; mapping the risk score in a risk score interval corresponding to a target risk level of the target user to determine the target risk level of the target user; The method further comprises: generating an alarm information when the heart rupture risk early warning value is greater than or equal to a preset risk threshold and / or the target risk level satisfies a preset level. 10.A device for training a heart rupture early warning model, characterized in that, The device applied to a data processing terminal comprises: The first obtaining module is configured to obtain a training sample set, wherein the training sample set comprises a plurality of training samples, a feature set of each of the training samples, and a heart rupture alarm label result of each of the training samples, the heart rupture alarm label result is used to represent whether heart rupture occurs in the training sample, and the feature set comprises a plurality of features in a demographic feature, a socio-economic feature, a previous medical history feature, a disease manifestation feature, a laboratory examination feature, a treatment intervention feature, a patient-reported quality of life feature, and a medical institution feature; The training module is configured to train a preset model by taking the feature set of each of the training samples as an independent variable and taking the heart rupture alarm label of each of the training samples as a dependent variable, to obtain a heart rupture early warning model.

11. A heart rupture early warning device, characterized by, The device is applied to a user terminal and comprises: The second obtaining module is configured to obtain a feature set of a target user. The first determining module is configured to input the feature set of the target user into the heart rupture early warning model according to any one of claims 1-7 to determine a heart rupture risk early warning value of the target user. The generating module is configured to generate an alarm information in a case where the heart rupture risk early warning value is greater than or equal to a preset risk threshold, and the alarm information is used to prompt a preventive measure for the target user.

12. An electronic device, comprising: The device comprises a processor and a memory storing computer program instructions, and the processor implements the training method of the heart rupture early warning model according to any one of claims 1-7 or the early warning method of the heart rupture according to any one of claims 8-9 when executing the computer program instructions.

13. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer program instructions, and the computer program instructions are executed by the processor to implement the training method of the heart rupture early warning model according to any one of claims 1-7 or the early warning method of the heart rupture according to any one of claims 8-9.

14. A computer program product, characterised in that, The instructions in the computer program product are executed by the processor of the electronic device, so that the electronic device executes the training method of the heart rupture early warning model according to any one of claims 1-7 or the early warning method of the heart rupture according to any one of claims 8-9.

Images