Ischemic heart disease risk prediction model construction method based on novel inflammation indexes
By using the elastic network regression method to screen new inflammatory indicators and construct a logistic regression model, the problem of insufficient applicability of existing models in Asian populations was solved, and robust and accurate prediction of ischemic heart disease risk was achieved, supporting early intervention.
Patent Information
- Application Number
- CN202511030810.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-25
- Publication Date
- 2025-09-16
AI Technical Summary
Existing ischemic heart disease risk prediction models do not fully consider new inflammatory indicators, especially their lack of applicability in Asian populations, resulting in a lack of effective means for early risk prediction and intervention strategies.
The elastic network regression method was used to screen out new inflammatory indicators related to ischemic heart disease from the inflammatory indicator group. Combined with demographic and lifestyle variables, a logistic regression model was constructed and trained to obtain an ischemic heart disease risk prediction model.
It improves the robustness and accuracy of ischemic heart disease risk prediction, has good clinical applicability, is suitable for Asian populations, and supports early intervention measures.
Smart Images
Figure CN120656738A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of disease prediction models, and specifically relates to a method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory indicator. Background Art
[0002] Ischemic heart disease (IHD), commonly referred to as coronary artery disease (CHD), is primarily caused by the accumulation of atherosclerotic plaque in the coronary arteries. This disease encompasses both acute myocardial infarction (MI) and chronic ischemic forms such as angina pectoris or asymptomatic IHD following MI. Despite significant declines in cardiovascular mortality over the past 50 years, IHD remains a major public health challenge and is projected to cause nearly 20 million deaths annually by 2050, continuing to be the leading cause of death worldwide.
[0003] Current strategies for the primary prevention of IHD primarily rely on traditional cardiovascular risk assessment tools, such as the Framingham risk score and the pooled cohort equation. These tools are not specific for IHD but rather are designed to predict broader outcomes, including atherosclerotic cardiovascular disease. While both incorporate traditional risk factors such as age, sex, blood pressure, cholesterol levels, smoking status, and diabetes, inflammatory markers such as high-sensitivity C-reactive protein (hs-CRP) were not incorporated into the original models due to insufficient early data. Furthermore, the homogeneity of the development population limits the models' applicability to Asian populations and their inadequate integration of inflammatory markers.
[0004] Currently, most research focuses on prognostic assessment for IHD patients, while limited research has focused on developing IHD risk prediction models. Existing risk assessment tools fail to fully consider the potential value of novel inflammatory markers and lack applicability to Asian populations, leaving a critical gap that needs to be filled. Summary of the Invention
[0005] The technical problem to be solved by the present invention is to provide a method for constructing an ischemic heart disease risk prediction model based on a new inflammatory indicator, which can effectively predict the risk of ischemic heart disease and provide a target for early intervention.
[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions: A method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory marker comprises the following steps: Step 10, using the elastic network regression method, screening out new inflammatory indicators related to ischemic heart disease from the inflammatory indicator group as predictors to obtain predictor variables; Step 20: Based on the predictor variables, a logistic regression model is constructed and trained to obtain an ischemic heart disease risk prediction model.
[0007] As a further improvement of the present invention, step 10 specifically includes: Step 101, randomly dividing the ischemic heart disease related dataset into a training set and an internal validation set; Step 102: Based on the training set, the elastic network regression method is used to screen out new inflammatory indicators related to the risk of ischemic heart disease from the inflammatory indicator group; then the screened new inflammatory indicators are verified through the internal validation set to obtain predictive factors related to the risk of ischemic heart disease.
[0008] As a further improvement of the present invention, step 10 further includes: Step 103: Use the predictor as the main predictor variable and adjust to obtain the collaborative predictor variable.
[0009] As a further improvement of the present invention, the predictive factors include the ratio of white blood cells to high-density lipoprotein cholesterol, the ratio of monocytes to lymphocytes, the ratio of monocytes to high-density lipoprotein cholesterol, and the ratio of platelets to lymphocytes.
[0010] As a further improvement of the present invention, the collaborative predictor variables include age, gender, race, marital status, smoking, drinking, hypertension, diabetes, stroke, and kidney disease.
[0011] As a further improvement of the present invention, step 20 specifically includes: Based on the main predictor variables and collaborative predictor variables, a multivariate logistic regression model was constructed. The estimated values of the model regression coefficients in the training set were obtained by maximizing the likelihood function, and the ischemic heart disease risk prediction model was obtained.
[0012] As a further improvement of the present invention, the expression of the ischemic heart disease risk prediction model is: ;
[0013] Where In WHR represents the logarithm of the ratio of white blood cells to high-density lipoprotein cholesterol; In MLR represents the logarithm of the ratio of monocytes to lymphocytes; In MHR represents the logarithm of the ratio of monocytes to high-density lipoprotein cholesterol; In PLR represents the logarithm of the ratio of platelets to lymphocytes; Age represents age; Sex represents sex; Other Hispanic represents other Hispanic race; Non-Hispanic White represents non-Hispanic white; Non-Hispanic Black represents non-Hispanic black; Other Race represents other race; Widowed / Divorced / Separated represents marital status; Never Married represents never married; Smoking represents smoking; Drinking represents drinking; Diabetes represents diabetes; Kidney Disease represents kidney disease; Hypertension represents hypertension; Stroke Disease represents stroke.
[0014] Compared with the prior art, the technical solution of the present invention has the following beneficial effects: This paper provides a method for constructing an ischemic heart disease risk prediction model based on novel inflammatory markers. Taking into account the association between novel inflammatory markers and ischemic heart disease, the elastic network regression method is used to screen novel inflammatory markers associated with ischemic heart disease as predictors. A logistic regression model is constructed and trained to produce an ischemic heart disease risk prediction model. This method facilitates ischemic heart disease risk prediction and assessment, demonstrating robustness, accuracy, and widespread clinical applicability. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 A flowchart of the method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory indicator provided by the present invention; Figure 2 A flowchart of a specific embodiment of the present invention; Figure 3 Plot the ROC curves for all data sets in the specific examples of the present invention; Figure 4 A calibration curve is drawn for all data sets in a specific embodiment of the present invention; wherein, Figure 4 a represents the calibration of the training set, Figure 4 b represents the calibration of the internal validation set, Figure 4 c represents the calibration of the first external validation set, Figure 4 d represents the calibration of the second external validation set; Figure 5 The DCA curves for all data sets in the specific examples of the present invention are drawn; wherein, Figure 5 a represents the DCA curve of the training set, Figure 5 b represents the DCA curve of the internal validation set, Figure 5 c represents the DCA curve of the first external validation set, Figure 5 d represents the DCA curve of the second external validation set; Figure 6 This is a nomogram of the CABIT model obtained for a specific example of the present invention. DETAILED DESCRIPTION
[0016] The technical solution of the present invention is described in detail below.
[0017] The embodiment of the present invention provides a method for constructing an ischemic heart disease risk prediction model based on a new inflammatory indicator, such as Figure 1 As shown, the following steps are included: Step 10: Using the elastic network regression method, novel inflammatory indicators related to ischemic heart disease are screened out from the inflammatory indicator group as predictors to obtain predictor variables.
[0018] Step 20: Based on the predictor variables, a logistic regression model is constructed and trained to obtain an ischemic heart disease risk prediction model.
[0019] The method in this embodiment uses elastic network regression to identify novel inflammatory markers associated with ischemic heart disease as predictors, constructs a logistic regression model, and trains an ischemic heart disease risk prediction model. This method facilitates ischemic heart disease risk prediction and assessment, demonstrating robustness, accuracy, and widespread clinical applicability.
[0020] As a preferred example, step 10 specifically includes: Step 101: randomly divide the ischemic heart disease related dataset into a training set and an internal validation set.
[0021] This embodiment of the present invention uses the National Health and Nutrition Examination Survey (NHANES) database as a development data source to collect data related to ischemic heart disease. This database contains a rich collection of demographic characteristics and laboratory test information, facilitating multivariate analysis and model training. Specifically, IHD is determined based on MCQ160c (Have you ever been told by a doctor or other health professional that you have coronary heart disease?), MCQ160d (Have you ever been told by a doctor or other health professional that you have angina pectoris, also known as angina?), and MCQ160e (Have you ever been told by a doctor or other health professional that you have heart disease, also known as myocardial infarction?). The collected data is then gradually excluded based on custom criteria (e.g., excluding individuals under 20 years of age, missing key laboratory parameters, or health status), to form an ischemic heart disease-related dataset. Finally, the ischemic heart disease-related dataset is randomly divided into a training set and an internal validation set. Preferably, the random split is a 7:3 ratio.
[0022] Step 102: Based on the training set, the elastic network regression method is used to screen out new inflammatory indicators related to the risk of ischemic heart disease from the inflammatory indicator group; then the screened new inflammatory indicators are verified through the internal validation set to obtain predictive factors related to the risk of ischemic heart disease.
[0023] Preferably, 20 inflammatory indicators are selected to form an inflammatory indicator group, including high-density lipoprotein cholesterol (HDL-C), white blood cells (WBC), neutrophils (NEU), lymphocytes (LYM), monocytes (MONO), high-sensitivity C-reactive protein (hs-CRP), platelets (PLT) and new inflammatory indicators: white blood cell to high-density lipoprotein cholesterol ratio (WHR=WBC / HDL-C), neutrophil to high-density lipoprotein cholesterol ratio (NHR=NEU / HDL-C), lymphocyte to high-density lipoprotein cholesterol ratio (LHR=LYM / HDL-C), monocyte to high-density lipoprotein cholesterol ratio (MHR=MONO / HDL-C), high-sensitivity C-reactive protein to High-density lipoprotein cholesterol ratio (CHR=hs-CRP / HDL-C), platelet-to-high-density lipoprotein cholesterol ratio (PHR=PLT / HDL-C), neutrophil-to-lymphocyte ratio (NLR=NEU / LYM), monocyte-to-lymphocyte ratio (MLR=MONO / LYM), platelet-to-lymphocyte ratio (PLR=PLT / LYM), high-sensitivity C-reactive protein-to-lymphocyte ratio (CLR=hs-CRP / LYM), systemic inflammatory response index (SIRI=NEU×MONO / LYM), systemic immune inflammatory index (SII=PLT×NEU / LYM), and systemic inflammation value (PIV=NEU×PLT×MONO / LYM).
[0024] The above-mentioned inflammatory indicators are highly correlated with the occurrence and development of ischemic heart disease, and can more comprehensively reflect the body's inflammatory load and immune status. They are more stable and sensitive than single indicators. Moreover, the above-mentioned inflammatory indicators can be directly obtained from routine blood tests, are low-cost, and have the potential for promotion.
[0025] The 20 inflammatory markers in the training and internal validation sets were transformed using natural logarithms to conform to a normal distribution. Elastic net regression was used to screen the 20 inflammatory markers. This method, combining two penalty mechanisms, lasso regression (L1 regularization) and ridge regression (L2 regularization), effectively addresses collinearity between variables and enables robust feature selection for dimensionality reduction. The optimal penalty parameter was selected through 10-fold cross-validation, and novel inflammatory markers associated with IHD risk were preliminarily identified. The robustness of these markers was further tested in the internal validation set, with statistical significance maintained after adjustment for confounding factors (P < 0.05). Finally, the validated novel inflammatory markers were identified as predictors of IHD risk.
[0026] Preferably, the derived predictors include the white blood cell to high-density lipoprotein cholesterol ratio (WHR), the monocyte to lymphocyte ratio (MLR), the monocyte to high-density lipoprotein cholesterol ratio (MHR), and the platelet to lymphocyte ratio (PLR).
[0027] In step 103, the predictor is used as the primary predictor variable, while various demographic, lifestyle, and health-related variables, including age, sex, race, marital status, smoking, alcohol consumption, hypertension, diabetes, stroke, and kidney disease, are adjusted as synergistic predictors. Synergistic predictors are non-core variables that may affect the occurrence of the target outcome event. Unlike the primary predictor variable, these synergistic predictors are not included in the variable screening process. Instead, they are directly included to control for confounding factors, thereby improving the stability and clinical interpretability of the model.
[0028] As a preferred example, step 20 specifically includes: Based on the main predictor variables and collaborative predictor variables, a multivariate logistic regression model was constructed. The estimated values of the model regression coefficients in the training set were obtained by maximizing the likelihood function, and the ischemic heart disease risk prediction model was obtained.
[0029] A multivariate logistic regression model was used to predict the binary outcome (whether IHD developed or not), which has good interpretability, robustness, and variable adjustment capabilities, and has strong predictive ability.
[0030] Preferably, the expression of the ischemic heart disease risk prediction model is: ;
[0031] Where CABIT Score represents the linear prediction score for ischemic heart disease; In WHR represents the logarithm of the ratio of white blood cells to high-density lipoprotein cholesterol; In MLR represents the logarithm of the ratio of monocytes to lymphocytes; In MHR represents the logarithm of the ratio of monocytes to high-density lipoprotein cholesterol; In PLR represents the logarithm of the ratio of platelets to lymphocytes; Age represents age (specific numerical value); Sex represents sex (if male, Sex = 1; if female, Sex = 0).
[0032] With Mexican American and Married / Living with partner as the reference groups, the following race and marriage variables are set: Other Hispanic represents other Hispanic race (if other Hispanic, Other Hispanic=1, otherwise Other Hispanic=0); Non-Hispanic White represents non-Hispanic white (if non-Hispanic white, Non-Hispanic White=1, otherwise Non-HispanicWhite=0); Non-Hispanic Black represents non-Hispanic black (if non-Hispanic black, Non-Hispanic Black=1, otherwise Non-Hispanic Black=0); Other Race represents other race (if none of the four races mentioned above, Other Race=1, otherwise Other Race=0).
[0033] Widowed / Divorced / Separated indicates marriage (if any of the statuses are widowed, divorced, or separated, Widowed / Divorced / Separated = 1, otherwise Widowed / Divorced / Separated = 0); Never Married indicates unmarried (if unmarried, Never Married = 1, otherwise Never Married = 0).
[0034] The remaining variables indicate whether there is a history of corresponding behaviors or diseases: Smoking represents smoking (if smoking, Smoking=1, if not smoking, Smoking=0); Drinking represents drinking (if drinking, Drinking=1, if not drinking, Drinking=0); Diabetes represents diabetes (if having diabetes, Diabetes=1, if not having diabetes, Diabetes=0); Kidney Disease represents kidney disease (if having kidney disease, Kidney Disease=1, if not having kidney disease, Kidney Disease=0); Hypertension represents hypertension (if having hypertension, Hypertension=1, if not having hypertension, Hypertension=0); Stroke Disease represents stroke (if having a history of stroke, StrokeDisease=1, if not having a history of stroke, StrokeDisease=0).
[0035] Preferably, when applying the expression to Chinese or other Asian populations, ethnicity variables can be grouped under the "OtherRace" category to achieve reasonable mapping of model variables within the local population. Furthermore, the model was independently externally validated in two prospective cohorts established in China to confirm its clinical applicability across ethnic groups and regions.
[0036] A specific example is provided below.
[0037] like Figure 2 As shown, data from three waves of the NHANES database, collected from 2015–2016, 2017–2020, and 2021–2023, initially recruited 37,464 participants aged 20 years or older. After phased exclusions, 11,840 participants were included in the final analysis. Among these 11,840 participants, the prevalence of IHD was 9.09%. IHD was primarily found in older men, who were more likely to be divorced / separated / widowed, smoke, and have a history of hypertension, diabetes, stroke, or kidney disease. Furthermore, IHD patients generally had elevated levels of biochemical parameters, including white blood cell count (WBC), neonates (NEU), mononuclear oxidase (MONO), and hs-CRP, while HDL-C, lysozymes (LYM), platinum lipoprotein (PLT), and pleural leukocyte antigen (PLL) levels were decreased.
[0038] 11,840 participants were randomly divided into training set ( n =8323) and the internal validation set ( n=3517). Based on the training set, 20 new inflammatory markers were screened using an elastic network regression model. These markers were then validated using an internal validation set. Ultimately, four consistent predictors associated with IHD risk were identified. These synergistic predictors were then adjusted to obtain the predictors shown in Table 1.
[0039] Table 1 Predictor variables Variable (95%) -value -value In WHR 0.2149 1.2397(0.6801, 2.2599) 0.7015 0.4830 In MLR 0.9173 2.5026(1.4945, 4.1907) 3.4875 4.90E-04 In MHR 0.2707 1.3109(0.6607, 2.6011) 0.7745 0.4390 In PLR -0.8213 0.4399(0.2988, 0.6476) -4.1621 3.91E-05 Age 0.0587 1.0604(1.0514, 1.0696) 13.3988 1.62E-40 Sex 0.3798 1.4620 (1.1090, 1.9273) 2.6940 7.07E-03 Other Hispanic 0.5028 1.6534(0.9710, 2.8154) 1.8516 0.0641 Non-Hispanic White 0.7535 2.1244(1.3522, 3.3378) 3.2689 1.08E-03 Non-Hispanic Black 0.5316 1.7017(1.0340, 2.8007) 2.0915 0.0365 Other Race 0.5782 1.7828(1.0260, 3.0978) 2.0511 0.0403 Divorced / separated / widowed 0.3112 1.3650 (1.0695, 1.7422) 2.4996 0.0125 Never married 0.3394 1.4040 (0.5177, 3.8078) 0.6667 0.5050 Smoking 0.7375 2.0906(1.6329, 2.6766) 5.8493 5.12E-09 Drinking -0.2366 0.7893(0.5438, 1.1457) -1.2444 0.2130 Diabetes 0.6959 2.0055(1.5636, 2.5723) 5.4796 4.39E-08 Kidney Disease 0.3218 1.3796(0.9555, 1.9920) 1.7169 0.0860 Hypertension 0.5044 1.6560 (1.2292, 2.2309) 3.3173 9.13E-04 Stroke 0.7330 2.0812(1.4723, 2.9421) 4.1501 3.36E-05 Based on these predictor variables, a logistic regression model was constructed. The CABIT model was constructed by maximizing the likelihood function to estimate the model regression coefficients for the training set. The final model included In WHR, In MLR, In MHR, In PLR, age, sex, and race; marital status, smoking, alcohol consumption, diabetes, kidney disease, hypertension, and stroke.
[0040] The expression is as follows: ;
[0041] The discriminative ability of the CABIT model obtained in this example is evaluated below.
[0042] External validation samples were collected from a physical examination cohort at the Second Affiliated Hospital of Nanjing Medical University from July 2018 to November 2019 and a cardiovascular population cohort from October 2015 to May 2020. The external validation cohorts were identified based on clinical diagnoses recorded in hospital medical records during follow-up. The first external validation cohort (external validation set 1) consisted of 168 hospital employees undergoing routine physical examinations from July 2018 to November 2019, with a median follow-up of 6.0 years. Participants with a history of IHD within the past 6 months, those who refused medical examinations, valvular or pericardial disease, severe liver or kidney dysfunction, malignant tumors, or recent surgery or trauma were excluded. The second external validation set (external validation set 2) consisted of 49 patients who presented with cardiovascular disease symptoms from October 2015 to May 2020, with a median follow-up of 6.7 years. Patients with angiographically confirmed severe coronary artery stenosis, coronary artery spasm, and severe autoimmune or infectious diseases were excluded.
[0043] In order to evaluate the discriminative ability of the CABIT model obtained in this example, the AUC and ROC curves were calculated based on the CABIT model, as shown in Figure 3As shown, the predictive ability of the model for IHD was assessed using the area under the receiver operating characteristic (ROC) curve (AUC). In the training set and internal validation set of this example, the AUC was 0.838 (95% CI: 0.825-0.851) and 0.823 (95% CI: 0.802-0.845), respectively, demonstrating strong discriminative ability. The CABIT model maintained good performance in the first external validation set (External Validation Set 1) and the second external validation set (External Validation Set 2), with AUCs of 0.831 (95% CI: 0.760-0.902) and 0.702 (95% CI: 0.550-0.855), respectively.
[0044] To evaluate the calibration of the CABIT model obtained in this example, the expected number of IHD events was compared with the number of actually observed IHD events. The expected-to-observed ratio (EO ratio) was calculated and the calibration was evaluated by plotting the ratio of the average predicted event count to the average observed event count (calibration curve). The EO ratio for the training set was 0.98, indicating excellent calibration. Figure 4 (a). Acceptable calibration was observed in the validation cohort, supporting the robustness of the model in the population, as shown in Figure 4 (b)~(d).
[0045] To evaluate the clinical utility of the CABIT model derived in this example, decision curve analysis (DCA), clinical impact curve (CIC), and net reduction of intervention curve (NR) were performed for all sets. Decision curve analysis is a method for directly evaluating the clinical value of a prediction model. It calculates the net clinical benefit (NB) of the allocated intervention based on the model. NB is calculated as follows: , TP and FP represent the number of true positives and false positives respectively, n represents the sample size of the study, P t It indicates the acceptable level of predicted probability of adverse events that warrants intervention. Clinical value can also be expressed as the net reduction (NR) of intervention, which is the number of unnecessary interventions guided by the prediction model compared to the "universal intervention" strategy. The NR is calculated as: , where NB Model NB, NB representing the CABIT model All NB represents the "universal intervention" strategy.
[0046] The results of DCA showed that the model produced positive net benefits compared with the whole population intervention strategy and the whole population no intervention strategy within the 45% risk threshold probability range, supporting its potential value, such as Figure 5In the training set, CIC showed that the estimated number of high-risk patients at a risk threshold probability of 10% was close to the actual number of events, while the NR curve indicated that the use of this model could reduce unnecessary interventions compared with the whole population strategy.
[0047] A nomogram was developed from the CABIT model and visualized using the combined dataset from all cohorts to improve clarity and robustness, as Figure 6 shown.
[0048] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art will appreciate that the present invention is not limited to the specific embodiments described above. The specific embodiments and descriptions in the specification are intended only to further illustrate the principles of the present invention. The basic principles, main features, and advantages of the present invention are shown and described above without departing from the spirit and scope of the present invention. Those skilled in the art will appreciate that various changes and modifications may be made, and such changes and modifications are intended to fall within the scope of the invention as claimed.
Claims
1. A method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory indicator, characterized in that: The following steps are involved: Step 10, using the elastic network regression method, screening out new inflammatory indicators related to ischemic heart disease from the inflammatory indicator group as predictors to obtain predictor variables; Step 20: Based on the predictor variables, a logistic regression model is constructed and trained to obtain an ischemic heart disease risk prediction model.
2. The method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory indicator according to claim 1, characterized in that: The step 10 specifically includes: Step 101, randomly dividing the ischemic heart disease related dataset into a training set and an internal validation set; Step 102: Based on the training set, the elastic network regression method is used to screen out new inflammatory indicators related to the risk of ischemic heart disease from the inflammatory indicator group; then the screened new inflammatory indicators are verified through the internal validation set to obtain predictive factors related to the risk of ischemic heart disease.
3. The method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory indicator according to claim 2, characterized in that: The step 10 further includes: Step 103: Use the predictor as the main predictor variable and adjust to obtain the collaborative predictor variable.
4. The method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory indicator according to claim 1, characterized in that: The predictors include the leukocyte to high-density lipoprotein cholesterol ratio, the monocyte to lymphocyte ratio, the monocyte to high-density lipoprotein cholesterol ratio, and the platelet to lymphocyte ratio.
5. The method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory index according to claim 3, characterized in that: The co-predictor variables included age, sex, race, marriage, smoking, drinking, hypertension, diabetes, stroke, and kidney disease.
6. The method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory index according to claim 3, characterized in that: Step 20 specifically includes: Based on the main predictor variables and collaborative predictor variables, a multivariate logistic regression model was constructed. The estimated values of the model regression coefficients in the training set were obtained by maximizing the likelihood function, and the ischemic heart disease risk prediction model was obtained.
7. The method for constructing an ischemic heart disease risk prediction model based on a novel inflammatory index according to claim 1, characterized in that: The expression of the ischemic heart disease risk prediction model is: ； Where In WHR represents the logarithm of the ratio of white blood cells to high-density lipoprotein cholesterol; In MLR represents the logarithm of the ratio of monocytes to lymphocytes; In MHR represents the logarithm of the ratio of monocytes to high-density lipoprotein cholesterol; In PLR represents the logarithm of the ratio of platelets to lymphocytes; Age represents age; Sex represents sex; Other Hispanic represents other Hispanic race; Non-Hispanic White represents non-Hispanic white; Non-Hispanic Black represents non-Hispanic black; Other Race represents other race; Widowed / Divorced / Separated represents marital status; Never Married represents never married; Smoking represents smoking; Drinking represents drinking; Diabetes represents diabetes; Kidney Disease represents kidney disease; Hypertension represents hypertension; Stroke Disease represents stroke.