Method, device, electronic device and storage medium for generating coronary heart disease prediction model

Abstract

The present disclosure relates to a method, device, electronic device and storage medium for generating a coronary heart disease prediction model, the method comprising: acquiring a first data set, wherein the first data set comprises clinical data of a plurality of subjects and a diagnosis result of coronary heart disease of each subject; determining key data affecting the predicted risk of coronary heart disease according to the first data set, wherein the key data comprises ceramide; generating a second data set according to the key data affecting the predicted risk of coronary heart disease of each subject and the diagnosis result of coronary heart disease of each subject; inputting the second data set into a regression model to obtain a weight value of each key data in the predicted risk of coronary heart disease; generating a score value in the predicted risk of coronary heart disease for each key data according to each weight value; determining a corresponding relationship between a total score value and a probability of suffering from coronary heart disease; generating a first coronary heart disease prediction model according to the corresponding relationship between the score value of each key data in the predicted risk of coronary heart disease, the total score value and the probability of suffering from coronary heart disease.

Description

Technical Field

[0001] The present disclosure relates to biomedical technology, and more specifically, to a method, device, electronic device and storage medium for generating a coronary heart disease prediction model. Background Art

[0002] Cardiovascular disease, especially coronary heart disease, has become one of the major diseases threatening human health worldwide. According to statistics from the World Health Organization, about 17 million people die from cardiovascular disease each year, accounting for one-third of the global death toll. As an important type of cardiovascular disease, accurate diagnosis of coronary heart disease is crucial for treatment and prevention.

[0003] The current diagnostic methods for coronary heart disease have certain limitations. The gold standard for the diagnosis of coronary heart disease is coronary angiography, but coronary angiography is an invasive examination, and there are risks of complications such as contrast agent allergy, arteriovenous fistula at the puncture site, vascular dissection, and even death during the examination. In addition, coronary angiography requires large equipment and surgeons with interventional experience, which is currently difficult to carry out in some county-level hospitals and all hospitals below the county level. Therefore, the establishment of non-invasive diagnostic tools and methods for coronary heart disease is a difficult problem that needs to be solved in the clinic, and it has great translational application value. At present, doctors mostly rely on the age, gender, and traditional blood pressure, blood lipid and other test results of the patients, combined with personal experience to make judgments on the condition. This method not only lacks standardization, but also makes it difficult to adapt to the precise diagnosis needs of a large number of patients.

[0004] In the existing technology, although there are attempts to improve the accuracy of diagnosis through emerging detection methods such as serological markers, these methods are often complicated to operate, costly, and difficult to popularize and apply in medical institutions with limited resources.

[0005] Therefore, there is a need to provide a new type of cardiovascular test data analysis system to provide doctors with a more standardized and replicable test information analysis method to improve the diagnostic accuracy of coronary heart disease, optimize clinical decision-making, reduce the risk of misdiagnosis and missed diagnosis, and thus provide patients with more accurate medical services. Summary of the invention

[0006] One objective of the present disclosure is to provide a new technical solution for a method for generating a coronary heart disease prediction model.

[0007] According to a first aspect of the present disclosure, a method for generating a coronary heart disease prediction model is provided, comprising:

[0008] Acquire a first data set, wherein the first data set includes clinical data of a plurality of subjects and a coronary heart disease diagnosis result of each subject;

[0009] Determining key data affecting the predicted risk of coronary heart disease based on the first data set; wherein the key data includes ceramide;

[0010] generating a second data set according to the key data affecting the predicted risk of coronary heart disease of each subject and the coronary heart disease diagnosis result of each subject;

[0011] Inputting the second data set into a regression model to obtain weight values ​​of each key data in predicting the risk of coronary heart disease;

[0012] Generating a score value in the predicted risk of coronary heart disease for each key data according to the weight value of each key data in the predicted risk of coronary heart disease;

[0013] Determine the corresponding relationship between the total score and the probability of coronary heart disease, wherein the total score is the sum of the score of each key data in the predicted risk of coronary heart disease;

[0014] A first coronary heart disease prediction model is generated according to the score values ​​of the key data in the coronary heart disease prediction risk, the corresponding relationship between the total score value and the probability of coronary heart disease.

[0015] Optionally, determining key data affecting the predicted risk of coronary heart disease according to the first data set includes:

[0016] Inputting the first data set into a Lasso regression model, and adjusting parameters of the Lasso regression model until an objective function of the Lasso regression model converges or the number of iterations reaches a maximum value;

[0017] When the objective function of the Lasso regression model converges or the number of iterations reaches a maximum value, obtaining the coefficients of each clinical data in the first data set;

[0018] When the coefficient of the clinical data exceeds a preset threshold, the corresponding clinical data is determined as a key data affecting the predicted risk of coronary heart disease.

[0019] Optionally, the step of inputting the second data set into a regression model to obtain weight values ​​of each key data in predicting the risk of coronary heart disease includes:

[0020] Using the key data of each subject that affects the predicted risk of coronary heart disease as the independent variable of the regression model, using the coronary heart disease diagnosis result of each subject as the dependent variable of the regression model, fitting the regression equation in the regression model until the fitting index value reaches a preset threshold;

[0021] Determining the regression coefficients of the respective variables in the regression equation;

[0022] According to the regression coefficients of the respective variables, the weight values ​​of the key data in predicting the risk of coronary heart disease are determined.

[0023] Optionally, the first coronary heart disease prediction model is presented in the form of a nomogram, the nomogram comprising a first nomogram group and a second nomogram group, wherein:

[0024] The first column line group includes column lines corresponding to different results of each key data, and a same score value column line with scales corresponding to different results of all key data.

[0025] The second column line group includes a column line with scales generated based on the total score value and a column line with scales generated based on the probability of coronary heart disease. The column line with scales generated based on the total score value and the column line with scales generated based on the probability of coronary heart disease have a corresponding relationship.

[0026] Optionally, based on each key data, different results have different scoring values ​​in the prediction risk of coronary heart disease.

[0027] Optionally, the key data also include: the subject's age, gender, smoking history, history of hypertension, total cholesterol, glycosylated hemoglobin, and left ventricular end-diastolic diameter.

[0028] Optionally, the method further includes:

[0029] The second data set is used to train the prediction model constructed based on the decision tree machine learning method to obtain a second coronary heart disease prediction model.

[0030] According to a second aspect of the present disclosure, a device for generating a coronary heart disease prediction model is provided, comprising:

[0031] A first data set acquisition module, used for a first data set, wherein the first data set includes clinical data of a plurality of subjects and a coronary heart disease diagnosis result of each subject;

[0032] A key data determination module, used to determine key data affecting the predicted risk of coronary heart disease according to the first data set; wherein the key data includes ceramide;

[0033] The second data set generating module is used to generate a second data set according to the key data affecting the predicted risk of coronary heart disease of each subject and the coronary heart disease diagnosis result of each subject;

[0034] A weight value determination module, used for inputting the second data set into a regression model to obtain a weight value of each key data in the prediction risk of coronary heart disease;

[0035] A scoring value generating module, used for generating a scoring value in the predicted risk of coronary heart disease for each key data according to the weight value of each key data in the predicted risk of coronary heart disease;

[0036] A corresponding relationship determination module, used to determine the corresponding relationship between the total score value and the probability of coronary heart disease, wherein the total score value is the sum of the score values ​​of each key data in the predicted risk of coronary heart disease;

[0037] The first coronary heart disease prediction model generating module is used to generate the first coronary heart disease prediction model according to the score values ​​of the key data in the coronary heart disease prediction risk, the corresponding relationship between the total score value and the probability of coronary heart disease.

[0038] According to a third aspect of the present disclosure, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the computer program is used to control the processor to operate so as to execute the method according to any one of the first aspects of the present disclosure.

[0039] According to a fourth aspect of the present disclosure, a non-volatile computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described in any one of the first aspects is implemented.

[0040] The method for generating a coronary heart disease prediction model provided by the present invention determines key data that affect the predicted risk of coronary heart disease by comprehensively using traditional detection indicators and clinical data composed of an emerging biomarker ceramide, wherein the key data include ceramide, and then uses a regression model to obtain a weight value of each key data in the predicted risk of coronary heart disease. According to the weight value of each key data in the predicted risk of coronary heart disease, a score value in the predicted risk of coronary heart disease is generated for each key data, and then the corresponding relationship between the total score value and the probability of coronary heart disease is determined. According to the corresponding relationship between the score value of each key data in the predicted risk of coronary heart disease, the total score value and the probability of coronary heart disease, a first coronary heart disease prediction model is generated, which helps to improve the diagnostic accuracy of coronary heart disease, optimize clinical decision-making, and reduce the risk of misdiagnosis and missed diagnosis, thereby providing more accurate medical services for subjects.

[0041] Features and advantages of the embodiments of the present specification will become apparent from the following detailed description of exemplary embodiments of the present specification with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the specification and, together with the description, serve to explain the principles of the embodiments of the specification.

[0043] Figure 1 It is a processing flow chart of a method for generating a coronary heart disease prediction model according to an embodiment of the present disclosure.

[0044] Figure 2 It is a schematic diagram of a nomogram corresponding to a first coronary heart disease prediction model according to an embodiment of the present disclosure.

[0045] Figure 3 It is a schematic diagram of the ROC curve corresponding to the first coronary heart disease prediction model according to an embodiment of the present disclosure.

[0046] Figure 4 It is a schematic diagram of a calibration curve corresponding to the first coronary heart disease prediction model according to an embodiment of the present disclosure.

[0047] Figure 5 It is a schematic diagram of a decision curve corresponding to the first coronary heart disease prediction model according to an embodiment of the present disclosure.

[0048] Figure 6 It is a schematic diagram of the ROC curve corresponding to the first coronary heart disease prediction model according to an embodiment of the present disclosure.

[0049] Figure 7 It is a principle block diagram of a device for generating a coronary heart disease prediction model according to an embodiment of the present disclosure.

[0050] Figure 8 It is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0051] Various exemplary embodiments of the present specification will now be described in detail with reference to the accompanying drawings.

[0052] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the embodiments of the present specification and its application or uses.

[0053] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0054] In one embodiment of the present disclosure, a method for generating a coronary heart disease prediction model is provided. Figure 1 As shown, a method for generating a coronary heart disease prediction model in this embodiment may include the following steps S110 to S170.

[0055] Step S110 , obtaining a first data set, wherein the first data set includes clinical data of a plurality of subjects and a coronary heart disease diagnosis result of each subject.

[0056] The clinical data of the subjects included age, gender, body mass index, smoking history, history of hypertension, history of hyperlipidemia, history of diabetes, history of stroke, CKMB (Creatine Kinase MB isoenzyme), triglycerides, total cholesterol, serum creatinine, glycosylated hemoglobin, uric acid, high-sensitivity C-reactive protein, left ventricular ejection fraction, left ventricular end-diastolic diameter and ceramide. For each subject, the diagnosis of coronary heart disease was one of the two: suffering from coronary heart disease and not suffering from coronary heart disease.

[0057] Biochemical indicators such as triglycerides, total cholesterol, and high-sensitivity C-reactive protein can be measured by LABOSPECT 008 biochemical analyzer (Hitachi, Tokyo, Japan). Glycated hemoglobin can be measured by Tosoh G8 HPLC analyzer (Tosoh Bioscience, Tokyo, Japan). Left ventricular ejection fraction and left ventricular end-diastolic diameter are measured by echocardiography. Ceramide can be quantitatively measured on a 5500 QTRAP (SCIEX, Framingham, MA) mass spectrometer equipped with an Eksigent 100-XL UHPLC system. In one embodiment, ceramide molecules are Cer (d18: 1 / 16: 0) and ceramide molecules are Cer (d18: 1 / 24: 0) are quantitatively measured respectively. In this embodiment, the clinical data of ceramide is characterized by Cer (d18: 1 / 16: 0) / Cer (d18: 1 / 24: 0).

[0058] Table 1 is the statistical data corresponding to the clinical data of multiple subjects according to one embodiment of the present disclosure. The subjects are divided into two categories, one is subjects with coronary heart disease, and the other is subjects without coronary heart disease. One set of data in Table 1 is the statistical data corresponding to the clinical data of subjects with coronary heart disease, and the other set of data is the statistical data corresponding to the clinical data of subjects without coronary heart disease.

[0059] Taking 1164 subjects with coronary heart disease as an example, the parameters shown in Table 1 are explained.

[0060] The age, body mass index, total cholesterol, blood creatinine, glycosylated hemoglobin, and left ventricular end-diastolic diameter of the subjects with coronary heart disease are shown in the form of mean ± standard deviation. For example, the statistical data corresponding to the age of 1164 subjects with coronary heart disease is 60.1 years ± 9.2 years, where 60.1 years is the mean age of the 1164 subjects with coronary heart disease, and 9.2 years is the standard deviation.

[0061] The number of males, history of hypertension, hyperlipidemia, diabetes, smoking, and stroke among the subjects with coronary heart disease are expressed as the number and the proportion of the number. For example, among the 1,164 subjects with coronary heart disease, there were 892 males, accounting for 76.6%; among the 1,164 subjects with coronary heart disease, there were 730 subjects with hypertension, accounting for 62.7%.

[0062] CKMB, triglycerides, uric acid, high-sensitivity C-reactive protein, left ventricular ejection fraction, ceramide Cer (d18:1 / 16:0) / Cer (d18:1 / 24:0) in subjects with coronary heart disease are shown as median, 25th quantile and 75th quantile. For example, the median CKMB in 1164 subjects with coronary heart disease was 1.5 ng / ml, the 25th quantile was 0.8 ng / ml, and the 75th quantile was 2.0 ng / ml. The median ceramide Cer (d18:1 / 16:0) / Cer (d18:1 / 24:0) in 1164 subjects with coronary heart disease was 0.121, the 25th quantile was 0.105, and the 75th quantile was 0.167.

[0063] It should be noted that Table 1 shows the statistical data of the clinical data of all subjects in the first data set, which is not used as the data in the first data set. The actual clinical data of 1,164 subjects with coronary heart disease and the actual clinical data of 574 subjects without coronary heart disease in Table 1 constitute the first data set, and the actual clinical data of each subject includes the specific data of each parameter shown in Table 1.

[0064] Table 1

[0065]

[0066] Step S120, determining key data affecting the predicted risk of coronary heart disease based on the first data set; wherein the key data includes ceramide.

[0067] In one embodiment, the first data set includes a training set and a test set, 70% of the data in the first data set are randomly selected as the training set of the first coronary heart disease prediction model constructed based on the Lasso regression combined with Logistic regression method, and the remaining 30% of the data are used as the test set of the first coronary heart disease prediction model constructed based on the Lasso regression combined with Logistic regression method. Step S120 specifically includes: inputting the data of the training set in the first data set into the Lasso regression model, adjusting the parameters of the Lasso regression model until the objective function of the Lasso regression model converges or the number of iterations reaches a maximum value; when the objective function of the Lasso regression model converges or the number of iterations reaches a maximum value, obtaining the coefficients of each clinical data in the first data set; when the coefficient of the clinical data exceeds a preset threshold, determining the corresponding clinical data as a key data affecting the risk of coronary heart disease prediction.

[0068] It should be noted that Lasso regression (Least Absolute Shrinkage and Selection Operator, LASSO), also known as lasso regression, is a widely used regression analysis method in statistics. By introducing the adjustment parameter λ and L1 penalty term, the model complexity can be controlled and the variables can be selected, thereby improving the prediction accuracy and explanatory power of the model.

[0069] Specifically, in the first data set, each type of clinical data is used as a variable. Using the LASSO regression model, all clinical data of the first data set are included in the model for fitting. The regularization parameter λ in the LASSO regression is adjusted to control the complexity of the model. The selection of the λ value is usually determined by cross-validation and other methods to optimize the predictive performance of the model. Through the fitting process of LASSO regression, the changes in the coefficients of each variable are observed. As the λ value increases, the coefficients of some variables will gradually decrease to zero. These variables are factors that contribute less to the prediction of coronary heart disease risk, and are not used as key data affecting the prediction of coronary heart disease risk, and can be eliminated. Finally, variables with non-zero coefficients are screened out. These variables are the most valuable factors for predicting the risk of coronary heart disease, as key data affecting the prediction of coronary heart disease risk.

[0070] In one embodiment, the key data includes not only ceramide, but also age, gender, smoking history, hypertension history, total cholesterol, glycosylated hemoglobin, and left ventricular end-diastolic diameter. The key data can be determined by the above-mentioned Lasso regression model.

[0071] Step S130, generating a second data set according to the key data affecting the predicted risk of coronary heart disease of each subject and the coronary heart disease diagnosis result of each subject.

[0072] In one embodiment, when the key data include the subject's age, gender, smoking history, history of hypertension, total cholesterol, glycated hemoglobin, left ventricular end-diastolic diameter, and ceramide, the above 8 key data are obtained from the clinical data of each subject in the first data set, and the second data set is generated using the 8 key data corresponding to all subjects.

[0073] Step S140, inputting the second data set into the regression model to obtain the weight value of each key data in the prediction risk of coronary heart disease.

[0074] In one embodiment, step S140 specifically includes: using the key data of each subject that affects the predicted risk of coronary heart disease as the independent variable of the regression model, using the coronary heart disease diagnosis result of each subject as the dependent variable of the regression model, fitting the regression equation in the regression model until the fitting index value reaches a preset threshold; determining the regression coefficient of each variable in the regression equation; and determining the weight value of each key data in the predicted risk of coronary heart disease based on the regression coefficient of each variable.

[0075] The regression model may be a logistic regression model.

[0076] The goodness of fit of the regression equation can be evaluated by at least one of the following indicators: Pearson χ2 (chi-square test), deviance, Hosmer-Lemeshow (HL) index, Akaike information criterion (AIC), and SC (Schwarz Criterion).

[0077] The larger the absolute value of the regression coefficient of the independent variable, the greater the influence of the independent variable on the prediction of coronary heart disease, and the higher the weight value of the corresponding key data in the prediction risk of coronary heart disease.

[0078] Step S150, generating a score value in the predicted risk of coronary heart disease for each key data according to the weight value of each key data in the predicted risk of coronary heart disease.

[0079] In one embodiment, the score value of each key data in the predicted risk of coronary heart disease can be determined by a generation device of a coronary heart disease prediction model. Specifically, a preset total score value is obtained, and the score value of each key data in the predicted risk of coronary heart disease is determined according to the weight value and the total score value of each key data in the predicted risk of coronary heart disease. For example, the preset total score value is 150 points, the weight value of glycated hemoglobin in the predicted risk of coronary heart disease is 30%, then the score value of glycated hemoglobin in the predicted risk of coronary heart disease is 45 points, and the weight value of ceramide Cer (d18: 1 / 16: 0) / Cer (d18: 1 / 24: 0) in the predicted risk of coronary heart disease is 15%, then the score value of ceramide Cer (d18: 1 / 16: 0) / Cer (d18: 1 / 24: 0) in the predicted risk of coronary heart disease is 22.5.

[0080] In another embodiment, after obtaining the weight value of each key data in the predicted risk of coronary heart disease, the weight value of each key data in the predicted risk of coronary heart disease is displayed to a professional, for example, a doctor, so that the professional can assign a score value in the predicted risk of coronary heart disease to each key data according to the weight value of each key data in the predicted risk of coronary heart disease, and return the score value of each key data in the predicted risk of coronary heart disease to a generating device of a coronary heart disease prediction model, so that the generating device of the coronary heart disease prediction model executes subsequent steps according to the score value of each key data in the predicted risk of coronary heart disease.

[0081] It should be noted that the higher the weight of key data in the prediction of coronary heart disease risk, the higher the score of each key data in the prediction of coronary heart disease risk. For example, the weight of total cholesterol, smoking history, gender, age, left ventricular end-diastolic diameter, history of hypertension, ceramide, and glycosylated hemoglobin in the prediction of coronary heart disease risk increases in turn, and the score of total cholesterol, smoking history, gender, age, left ventricular end-diastolic diameter, history of hypertension, ceramide, and glycosylated hemoglobin in the prediction of coronary heart disease risk increases in turn.

[0082] Based on each key data, different results can be divided in the prediction risk of coronary heart disease.

[0083] Taking the key data including the subject's age, gender, smoking history, history of hypertension, total cholesterol, glycosylated hemoglobin, left ventricular end-diastolic diameter, and ceramide as an example, the different results corresponding to each key data are explained.

[0084] Age results were divided into greater than 65 years and less than or equal to 65 years. Gender was divided into male and female. Smoking history was divided into smoking and non-smoking. Hypertension history was divided into hypertension and no hypertension. Total cholesterol was divided into greater than 5.2mmol / L and less than or equal to 5.2mmol / L. Glycated hemoglobin was divided into greater than 6% and less than or equal to 6%. Left ventricular end-diastolic diameter was divided into greater than 52mm and less than or equal to 52mm. Ceramide Cer (d18:1 / 16:0) / Cer (d18:1 / 24:0) was divided into greater than 0.118 and less than or equal to 0.118.

[0085] Based on each key data, different results have different scoring values ​​in predicting the risk of coronary heart disease.

[0086] The score values ​​of different results corresponding to each key data are explained. When the age is greater than 65 years old, the corresponding score value is the score value determined based on the weight value of age in the predicted risk of coronary heart disease. When the age is less than or equal to 65 years old, the corresponding score value is 0. When the gender is male, the corresponding score value is the score value determined based on the weight value of gender in the predicted risk of coronary heart disease. When the gender is female, the corresponding score value is 0. When the smoking history is smoking, the corresponding score value is the score value determined based on the weight value of the smoking history in the predicted risk of coronary heart disease. When the smoking history is not smoking, the corresponding score value is 0. When the history of hypertension is hypertension, the corresponding score value is the score value determined based on the weight value of the hypertension history in the predicted risk of coronary heart disease. When the history of hypertension is not hypertension, the corresponding score value is 0. When the total cholesterol is greater than 5.2mmol / L, the corresponding score value is the score value determined based on the weight value of the total cholesterol in the predicted risk of coronary heart disease. When the total cholesterol is less than or equal to 5.2mmol / L, the corresponding score value is 0. When glycated hemoglobin is greater than 6%, the corresponding score is determined based on the weight of glycated hemoglobin in the risk prediction of coronary heart disease. When glycated hemoglobin is less than or equal to 6%, the corresponding score is 0. When the left ventricular end-diastolic diameter is greater than 52 mm, the corresponding score is determined based on the weight of the left ventricular end-diastolic diameter in the risk prediction of coronary heart disease. When the left ventricular end-diastolic diameter is less than or equal to 52 mm, the corresponding score is 0. When ceramide Cer (d18:1 / 16:0) / Cer (d18:1 / 24:0) is greater than 0.118, the corresponding score is determined based on the weight of ceramide in the risk prediction of coronary heart disease. When ceramide is less than or equal to 0.118, the corresponding score is 0.

[0087] Step S160, determining the corresponding relationship between the total score and the probability of coronary heart disease, wherein the total score is the sum of the score of each key data in the predicted risk of coronary heart disease.

[0088] The correspondence between the total score value and the probability of coronary heart disease can be one total score value corresponding to one probability of coronary heart disease, or one total score value segment corresponding to one probability of coronary heart disease.

[0089] In one embodiment, the correspondence between the total score value and the probability of suffering from coronary heart disease can be determined by a generating device of a coronary heart disease prediction model.

[0090] In another embodiment, the correspondence between the total score and the probability of coronary heart disease is determined by a professional. Specifically, after the professional determines the correspondence between the total score and the probability of coronary heart disease, the correspondence between the total score and the probability of coronary heart disease is returned to the generating device of the coronary heart disease prediction model, so that the generating device of the coronary heart disease prediction model performs subsequent steps according to the correspondence between the total score and the probability of coronary heart disease.

[0091] Step S170, generating a first coronary heart disease prediction model according to the score values ​​of the key data in the coronary heart disease prediction risk, the corresponding relationship between the total score value and the probability of coronary heart disease.

[0092] In one embodiment, the first coronary heart disease prediction model is displayed in the form of a nomogram, so that the first coronary heart disease prediction model is visualized. The nomogram includes a first nomogram group and a second nomogram group. The first nomogram group includes nomograms corresponding to different results of each key data, and the same score value nomogram with scales corresponding to different results of all key data.

[0093] The second column line group includes a column line with scales generated based on the total score value and a column line with scales generated based on the probability of coronary heart disease. The column line with scales generated based on the total score value and the column line with scales generated based on the probability of coronary heart disease have a corresponding relationship.

[0094] Figure 2 It is a schematic diagram of a nomogram corresponding to a first coronary heart disease prediction model according to an embodiment of the present disclosure.

[0095] like Figure 2As shown, the first group of ranking lines includes ranking lines corresponding to different results based on age, ranking lines corresponding to different results based on gender, ranking lines corresponding to different results based on ceramide, ranking lines corresponding to different results based on history of hypertension, ranking lines corresponding to different results based on total cholesterol, ranking lines corresponding to different results based on smoking history, ranking lines corresponding to different results based on left ventricular end-diastolic diameter, ranking lines corresponding to different results based on glycated hemoglobin, and the same scoring value ranking line with scales corresponding to different results of all key data. The scoring values ​​corresponding to different results of each key parameter can be determined by the same scoring value ranking line with scales corresponding to different results of all key data. When the age is greater than 65 years old, the corresponding scoring value is 17.5, and when the age is less than or equal to 65 years old, the corresponding scoring value is 0. When the gender is male, the corresponding scoring value is 10, and when the gender is female, the corresponding scoring value is 0. When ceramide Cer(d18:1 / 16:0) / Cer(d18:1 / 24:0) is greater than 0.118, the corresponding score is 22.5, and when ceramide is less than or equal to 0.118, the corresponding score is 0. When the history of hypertension is hypertension, the corresponding score is 20, and when the history of hypertension is not hypertension, the corresponding score is 0. When total cholesterol is greater than 5.2mmol / L, the corresponding score is 7.5, and when total cholesterol is less than or equal to 5.2mmol / L, the corresponding score is 0. When the smoking history is smoking, the corresponding score is 9.5, and when the smoking history is non-smoking, the corresponding score is 0. When the left ventricular end-diastolic diameter is greater than 52mm, the corresponding score is 18, and when the left ventricular end-diastolic diameter is less than or equal to 52mm, the corresponding score is 0. When glycosylated hemoglobin is greater than 6%, the corresponding score is 45, and when glycosylated hemoglobin is less than or equal to 6%, the corresponding score is 0.

[0096] like Figure 2 As shown, the second column line group includes a column line generated based on the total score value and a column line generated based on the probability of coronary heart disease. The column line generated based on the total score value and the column line generated based on the probability of coronary heart disease have a corresponding relationship. The corresponding relationship is that one total score value corresponds to one probability of coronary heart disease. Figure 2 As shown, the total score is 150. When the total score determined based on the above 8 key data is 0, the probability of coronary heart disease is not 0, because in addition to the influence of these 8 key data, the influence of other data needs to be considered.

[0097] The method for generating a coronary heart disease prediction model provided by the present invention determines key data that affect the predicted risk of coronary heart disease by comprehensively using traditional detection indicators and clinical data composed of an emerging biomarker ceramide, wherein the key data include ceramide, and then uses a regression model to obtain a weight value of each key data in the predicted risk of coronary heart disease. According to the weight value of each key data in the predicted risk of coronary heart disease, a score value in the predicted risk of coronary heart disease is generated for each key data, and then the corresponding relationship between the total score value and the probability of coronary heart disease is determined. According to the corresponding relationship between the score value of each key data in the predicted risk of coronary heart disease, the total score value and the probability of coronary heart disease, a first coronary heart disease prediction model is generated, which helps to improve the diagnostic accuracy of coronary heart disease, optimize clinical decision-making, and reduce the risk of misdiagnosis and missed diagnosis, thereby providing more accurate medical services to patients.

[0098] Based on the premise that the key data only include the difference between ceramide and ceramide, two first coronary heart disease prediction models are generated, namely the first coronary heart disease prediction model a and the first coronary heart disease prediction model b. Based on the first coronary heart disease prediction model a and the first coronary heart disease prediction model b, sensitivity is used as the vertical axis to represent the true positive rate, and 1-specificity is used as the horizontal axis to represent the false positive rate. ROC curves (Receiver Operating Characteristic Curve) are drawn respectively, as shown in Figure 3 As shown. Then, the area under the ROC curve (AUC) is calculated. The AUC value is an indicator used to evaluate the performance of the first coronary heart disease prediction model. The higher the AUC value, the better the prediction ability of the model. Figure 3 As shown, the AUC value corresponding to the first coronary heart disease prediction model a (0.839) is greater than the AUC value corresponding to the first coronary heart disease prediction model b (0.773), P < 0.05. It can be concluded that the prediction performance of the first coronary heart disease prediction model a is better than that of the first coronary heart disease prediction model b.

[0099] Next, the calibration curve was used to evaluate the calibration of the first coronary heart disease prediction model a for predicting the risk of coronary heart disease, that is, the degree of deviation between the predicted value and the actual value. The results showed that the calibration of the first coronary heart disease prediction model a was good and there was no overfitting. Figure 4 . Figure 4 The horizontal axis represents the predicted value of the probability of coronary heart disease, which is predicted by using the first coronary heart disease prediction model a, and the vertical axis represents the actual value of the probability of coronary heart disease. Figure 4 Three calibration curves are shown, namely, curve 1, curve 2 and curve 3, where curve 1 is the Apparent calibration curve, curve 2 is the Bias-corrected calibration curve, and curve 3 is the Ideal curve. Figure 4It can be seen that curve 1 (Apparent correction curve), curve 2 (Bias-corrected correction curve) and curve 3 (Ideal curve) are very close, indicating that the prediction accuracy of the first coronary heart disease prediction model a is better.

[0100] In addition, the clinical decision curve was used to evaluate the clinical application value of the first coronary heart disease prediction model a for predicting the risk of coronary heart disease, that is, to evaluate the net benefit of the difference between the benefit of correctly identifying positive subjects and the loss of incorrectly identifying positive subjects by the first coronary heart disease prediction model a. The results showed that the clinical net benefit of the first coronary heart disease prediction model a was better, as can be seen in detail. Figure 5 . Figure 5 The horizontal axis represents the predicted value of the probability of coronary heart disease, which is predicted using the first coronary heart disease prediction model a, and the vertical axis represents the net benefit. Figure 5 Three curves are shown, namely, curve 1, curve 2 and curve 3. Curve 1 is the net income curve, curve 2 is the All curve, and curve 3 is the None curve. Figure 5 It can be seen that the first coronary heart disease prediction model a has a good clinical net benefit and is of practical value.

[0101] Finally, the first coronary heart disease prediction model was verified in a randomly selected validation set containing 30% of the samples in the first data set, and the ROC curve was drawn to verify the prediction performance and robustness of the first coronary heart disease prediction model. The results showed that in the validation set cohort, the AUC value (0.825) corresponding to the first coronary heart disease prediction model a was greater than the AUC value (0.783) corresponding to the first coronary heart disease prediction model b, P < 0.05, which shows that the first coronary heart disease prediction model a has good and robust prediction performance and is better than the first coronary heart disease prediction model b. Figure 6 .

[0102] In one embodiment, the method for generating a coronary heart disease prediction model further includes: using a second data set to train a prediction model constructed based on a decision tree machine learning method to obtain a second coronary heart disease prediction model.

[0103] In one embodiment, 70% of the second data set is extracted as a training set of a prediction model constructed based on a decision tree machine learning method, and the remaining 30% is used as a test set of a prediction model constructed based on a decision tree machine learning method. The training set is input into the prediction model constructed based on the decision tree machine learning method for training. In the training process, first, nodes are generated, and the corresponding nodes are marked as leaf nodes according to the statistical analysis results, and then feature selection is performed based on information gain (ID3), gain rate (C4.5) and Gini index (CART). When the instance set of a certain node has the same feature vector, the growth of the decision tree is stopped, or when the decision tree reaches a preset height, the growth of the decision tree is stopped, and the training is stopped to obtain a prediction model constructed based on the decision tree machine learning method after training. Using the test set, the prediction model constructed based on the decision tree machine learning method after training is verified. When the accuracy of the output result of the prediction model constructed based on the decision tree machine learning method after training exceeds the preset threshold, it is determined that the verification is passed, that is, when the accuracy of the coronary heart disease results of each subject output by the prediction model constructed based on the decision tree machine learning method after training exceeds the preset threshold, it is determined that the verification is passed. If the verification is passed, the trained prediction model constructed based on the decision tree machine learning method will be used as the second coronary heart disease prediction model.

[0104] By comprehensively using traditional detection indicators and clinical data on the composition of the emerging biomarker ceramide, the key data that affect the predicted risk of coronary heart disease are determined. The key data include ceramide. Then, the second data set including the key data of ceramide and the coronary heart disease diagnosis results of each subject is generated and input into the prediction model constructed based on the decision tree machine learning method for training. The second coronary heart disease prediction model is obtained, which helps to improve the diagnostic accuracy of coronary heart disease.

[0105] An embodiment of the present disclosure provides a device for generating a coronary heart disease prediction model, such as Figure 7 The device 400 for generating a coronary heart disease prediction model comprises a first data set acquisition module 410, a key data determination module 420, a second data set generation module 430, a weight value determination module 440, a score value generation module 450, a corresponding relationship determination module 460, and a first coronary heart disease prediction model generation module 470.

[0106] The first data set acquisition module 410 is used for the first data set, wherein the first data set includes clinical data of multiple subjects and the coronary heart disease diagnosis result of each subject.

[0107] The key data determination module 420 is used to determine the key data that affects the predicted risk of coronary heart disease based on the first data set; wherein the key data includes ceramide.

[0108] The second data set generating module 430 is used to generate a second data set according to the key data affecting the predicted risk of coronary heart disease of each subject and the coronary heart disease diagnosis result of each subject.

[0109] The weight value determination module 440 is used to input the second data set into the regression model to obtain the weight value of each key data in the prediction risk of coronary heart disease.

[0110] The scoring value generating module 450 is used to generate a scoring value in the predicted risk of coronary heart disease for each key data according to the weight value of each key data in the predicted risk of coronary heart disease.

[0111] The corresponding relationship determination module 460 is used to determine the corresponding relationship between the total score value and the probability of coronary heart disease, wherein the total score value is the sum of the score values ​​of each key data in the predicted risk of coronary heart disease.

[0112] The first coronary heart disease prediction model generating module 470 is used to generate a first coronary heart disease prediction model according to the corresponding relationship between the score value of each key data in the coronary heart disease prediction risk, the total score value and the probability of coronary heart disease.

[0113] In one embodiment, the key data determination module 420 is used to input the first data set into the Lasso regression model and adjust the parameters of the Lasso regression model until the objective function of the Lasso regression model converges or the number of iterations reaches a maximum value; when the objective function of the Lasso regression model converges or the number of iterations reaches a maximum value, the coefficients of each clinical data in the first data set are obtained; when the coefficient of the clinical data exceeds a preset threshold, the corresponding clinical data is determined as a key data affecting the predicted risk of coronary heart disease.

[0114] In one embodiment, the weight value determination module 440 is used to use the key data of each subject that affects the predicted risk of coronary heart disease as the independent variable of the regression model, and the coronary heart disease diagnosis result of each subject as the dependent variable of the regression model, to fit the regression equation in the regression model until the fitting index value reaches a preset threshold; determine the regression coefficient of each variable in the regression equation; and determine the weight value of each key data in the predicted risk of coronary heart disease based on the regression coefficient of each variable.

[0115] In one embodiment, the first coronary heart disease prediction model is displayed in the form of a nomogram, which includes a first nomogram group and a second nomogram group, wherein the first nomogram group includes nomograms corresponding to different results of each key data, and the same score value nomogram with scales corresponding to different results of all key data, and the second nomogram group includes nomograms with scales generated based on the total score value and nomograms with scales generated based on the probability of coronary heart disease, and the nomograms with scales generated based on the total score value and the nomograms with scales generated based on the probability of coronary heart disease have a corresponding relationship.

[0116] In one embodiment, based on each key data, different results have different scoring values ​​in predicting the risk of coronary heart disease.

[0117] In one embodiment, the key data also include: age, gender, smoking history, hypertension history, total cholesterol, glycosylated hemoglobin, and left ventricular end-diastolic diameter of the subject.

[0118] In one embodiment, the device for generating a coronary heart disease prediction model further includes a second coronary heart disease prediction model generation module. The second coronary heart disease prediction model generation module is used to train the prediction model constructed based on the decision tree machine learning method using the second data set to obtain the second coronary heart disease prediction model.

[0119] An embodiment of the present disclosure provides an electronic device, such as Figure 8 The electronic device 500 includes a memory 520 and a processor 510. The memory 520 stores a computer program, and the computer program is used to control the processor 510 to operate to execute the method for generating a coronary heart disease prediction model in any of the above embodiments.

[0120] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. For the device embodiment, its related parts can be referred to the partial description of the method embodiment.

[0121] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0122] The embodiments of the present specification may be systems, methods and / or computer program products. The computer program product may include a computer-readable storage medium carrying computer instructions for causing a processor to implement various aspects of the embodiments of the present specification.

[0123] A computer-readable storage medium may be a tangible device that can hold and store computer instructions used by a computer instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which computer instructions are stored, and any suitable combination of the above. The computer-readable storage medium used herein is not to be interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0124] The computer instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device through a network layer, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network layer may include copper transmission cables, optical fiber transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network layer adapter card or network layer interface in each computing / processing device receives the computer instructions from the network layer and forwards the computer instructions for storage in a computer-readable storage medium in each computing / processing device.

[0125] The flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to the multiple embodiments of this specification. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a computer instruction, and a part of a module, a program segment or a computer instruction contains one or more executable computer instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented by a dedicated hardware-based system that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions. It is well known to those skilled in the art that it is equivalent to implement it by hardware, implement it by software, and implement it by combining software and hardware.

[0126] The embodiments of the present specification have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art without departing from the scope of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for generating a coronary heart disease prediction model, characterized in that: include: Acquire a first data set, wherein the first data set includes clinical data of a plurality of subjects and a coronary heart disease diagnosis result of each subject; Determining key data affecting the predicted risk of coronary heart disease based on the first data set; wherein the key data includes ceramide; generating a second data set according to the key data affecting the predicted risk of coronary heart disease of each subject and the coronary heart disease diagnosis result of each subject; Inputting the second data set into a regression model to obtain weight values ​​of each key data in predicting the risk of coronary heart disease; Generating a score value in the predicted risk of coronary heart disease for each key data according to the weight value of each key data in the predicted risk of coronary heart disease; Determine the corresponding relationship between the total score and the probability of coronary heart disease, wherein the total score is the sum of the score of each key data in the predicted risk of coronary heart disease; According to the score values ​​of the key data in the risk prediction of coronary heart disease, the corresponding relationship between the total score value and the probability of coronary heart disease, a first coronary heart disease prediction model is generated; wherein, The step of inputting the second data set into a regression model to obtain weight values ​​of each key data in predicting the risk of coronary heart disease includes: Using the key data of each subject that affects the predicted risk of coronary heart disease as the independent variable of the regression model, using the coronary heart disease diagnosis result of each subject as the dependent variable of the regression model, fitting the regression equation in the regression model until the fitting index value reaches a preset threshold; Determining the regression coefficients of the respective variables in the regression equation; According to the regression coefficients of the respective variables, the weight values ​​of the key data in predicting the risk of coronary heart disease are determined.

2. The method according to claim 1, characterized in that Determining key data affecting the predicted risk of coronary heart disease based on the first data set includes: Inputting the first data set into a Lasso regression model, and adjusting parameters of the Lasso regression model until an objective function of the Lasso regression model converges or the number of iterations reaches a maximum value; When the objective function of the Lasso regression model converges or the number of iterations reaches a maximum value, obtaining the coefficients of each clinical data in the first data set; When the coefficient of the clinical data exceeds a preset threshold, the corresponding clinical data is determined as a key data affecting the predicted risk of coronary heart disease.

3. The method according to claim 1, characterized in that The first coronary heart disease prediction model is presented in the form of a nomogram, and the nomogram includes a first nomogram group and a second nomogram group, wherein: The first column line group includes column lines corresponding to different results of each key data, and a same score value column line with scales corresponding to different results of all key data. The second column line group includes a column line with scales generated based on the total score value and a column line with scales generated based on the probability of coronary heart disease. The column line with scales generated based on the total score value and the column line with scales generated based on the probability of coronary heart disease have a corresponding relationship.

4. The method according to claim 3, characterized in that Based on each key data, different results have different scoring values ​​in predicting the risk of coronary heart disease.

5. The method according to claim 1, characterized in that The key data also include: the subject's age, gender, smoking history, history of hypertension, total cholesterol, glycosylated hemoglobin, and left ventricular end-diastolic diameter.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: The second data set is used to train the prediction model constructed based on the decision tree machine learning method to obtain a second coronary heart disease prediction model.

7. A device for generating a coronary heart disease prediction model, characterized in that: include: A first data set acquisition module, used for a first data set, wherein the first data set includes clinical data of a plurality of subjects and a coronary heart disease diagnosis result of each subject; A key data determination module, used to determine key data affecting the predicted risk of coronary heart disease according to the first data set; wherein the key data includes ceramide; The second data set generating module is used to generate a second data set according to the key data affecting the predicted risk of coronary heart disease of each subject and the coronary heart disease diagnosis result of each subject; A weight value determination module, used for inputting the second data set into a regression model to obtain a weight value of each key data in the prediction risk of coronary heart disease; A scoring value generating module, used for generating a scoring value in the predicted risk of coronary heart disease for each key data according to the weight value of each key data in the predicted risk of coronary heart disease; A corresponding relationship determination module, used to determine the corresponding relationship between the total score value and the probability of coronary heart disease, wherein the total score value is the sum of the score values ​​of each key data in the predicted risk of coronary heart disease; The first coronary heart disease prediction model generation module is used to generate the first coronary heart disease prediction model according to the score value of each key data in the coronary heart disease prediction risk, the corresponding relationship between the total score value and the probability of coronary heart disease; wherein, The weight value determination module is used to use the key data of each subject that affects the predicted risk of coronary heart disease as the independent variable of the regression model, and the coronary heart disease diagnosis result of each subject as the dependent variable of the regression model, and fit the regression equation in the regression model until the fitting index value reaches a preset threshold; Determining the regression coefficients of the respective variables in the regression equation; According to the regression coefficients of the respective variables, the weight values ​​of the key data in predicting the risk of coronary heart disease are determined.

8. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and the computer program is used to control the processor to operate so as to perform the method according to any one of claims 1 to 6.

9. A non-volatile computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images