Pulmonary arterial hypertension risk prediction model based on night hypoxia and cardiac ultrasound indexes and construction method

Abstract

The invention discloses a pulmonary arterial hypertension risk prediction model based on night hypoxia and cardiac ultrasound indexes and a construction method thereof, and the method comprises the steps: collecting cardiac ultrasound examination indexes, right cardiac catheter examination data and night sleep respiration monitoring data of a patient with chronic thromboembolic pulmonary diseases; the method comprises the following steps: according to hemodynamic data measured by right heart catheter examination, grouping into a CTEPD accompanied PH (hereinafter referred to as CTEPH) group and a CTEPD non-accompanied PH group, and carrying out statistical analysis on all collected data; risk factors related to the CTEPH are screened out; and screening CTEPH prediction model variables and constructing a pulmonary arterial hypertension risk prediction model. According to the method, the pulmonary arterial hypertension risk prediction model based on the night hypoxia and cardiac ultrasound indexes is constructed, the CTEPH risk is diagnosed only by combining the hypoxia indexes with the cardiac ultrasound indexes, non-invasive risk assessment is achieved, no harm is caused to patients, operation is easy and convenient, the model performance is good, and the method can be popularized and popularized in hospitals at all levels in China and is wide in applicability.

Description

Technical Field

[0001] The present invention relates to the field of data processing models and methods, and in particular to a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators and a construction method. Background Art

[0002] Chronic thromboembolic pulmonary disease (CTEPD) refers to a disease state in which persistent organized thrombi exist in the pulmonary arteries, leading to pulmonary vascular obstruction. Depending on whether pulmonary hypertension (PH) is present, CTEPD can be divided into two different clinical types, which differ significantly in severity, prognosis, and clinical intervention strategies.

[0003] CTEPD with PH is called chronic thromboembolicpulmonary hypertension (CTEPH), which belongs to the fourth category of PH in clinical classification. In CTEPH, pulmonary vascular resistance increases significantly, which in turn causes an increase in the pressure load of the right ventricle. Long-term excessive pressure load on the right heart can lead to right heart dysfunction or even right heart failure. Symptoms of CTEPH patients usually include progressively worsening dyspnea, decreased exercise tolerance, chest pain, and syncope. Due to the presence of PH, the prognosis of CTEPH is poor. If active treatment is not carried out, the patient's quality of life and survival will be seriously affected. According to the 2022 ESC / ERS guidelines [1] In terms of treatment, there is a big difference between CTEPD with and without PH. CTEPH requires active use of targeted antihypertensive drugs and percutaneous pulmonary balloon angioplasty. Patients who are suitable for surgery need pulmonary endarterectomy. Therefore, identifying patients with PH from CTEPD and early intervention are crucial to the prognosis of CTEPH patients.

[0004] The best existing technology is right heart float catheter examination, but this technology is invasive and has high operating risks, and has high requirements for technical personnel and hospital qualifications. In addition, there are currently limited hospitals in China that can perform right heart float catheter examinations, and they are all tertiary hospitals in first-tier cities. Hospitals in second- and third-tier cities, and even in remote areas, cannot perform this examination. At present, there is no pulmonary hypertension diagnostic model solution that can achieve the diagnostic efficacy of invasive right heart float catheter examination for CTEPH. For this reason, it is necessary to conduct research and design of non-invasive pulmonary hypertension diagnostic models. Summary of the invention

[0005] In view of the above problems, the present invention provides a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators. The nocturnal hypoxia data and cardiac ultrasound indicators of CTEPD patients are used as risk factor input data of the pulmonary hypertension risk prediction model. The predicted risk probability P of the sample CTEPH is calculated by the predicted probability calculation formula. The calculation formula of the risk prediction probability P is: Logit(P) is the logistic regression value of the final risk prediction probability P.

[0006] As a further illustration of the present invention, the SpO2 lower than 90% time percentage parameter T90 in the CTEPD patient's nighttime hypoxia data and the pulmonary artery systolic pressure (PASP) parameter in the cardiac ultrasound index are used as risk factor input values ​​of the pulmonary hypertension risk prediction model, and the calculation formula of the Logit (P) is:

[0007] Logit(P)=C1+C2*T90+C3*PASP

[0008] Among them, C1, C2, and C3 are constant coefficients.

[0009] Furthermore, the constant coefficients C1, C2, and C3 in the Logit (P) calculation formula are -4.042, 0.082, and 0.085, respectively. The Logit (P) calculation formula is:

[0010] Logit(P)=Logit(P)=-4.042+0.082*T90+0.085*PASP.

[0011] Furthermore, the pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators intuitively displays the risk prediction probability P of the CTEPH patient through a visual chart, and the SpO2 lower than 90% time percentage parameter T90 and the pulmonary artery systolic pressure parameter PASP correspond to indicator scores, and the risk prediction probability P of the patient is obtained by adding up each of the indicator scores.

[0012] On the other hand, the present invention also provides a method for constructing a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac echocardiography indicators, which specifically includes the following steps:

[0013] S1, collect cardiac ultrasound examination indexes, right cardiac catheter examination data and nocturnal sleep respiratory monitoring data of CTEPD patients, and group the data into two groups, CTEPH group and CTEPD without PH, according to the hemodynamic data measured by right cardiac catheter examination. Perform statistical analysis on all collected data; set binary variables for statistical analysis, CTEPH is set as 1, and CTEPD without PH is set as 0;

[0014] S2, screen out risk factors associated with CTEPH;

[0015] S3, screening CTEPH risk prediction model variables and constructing a pulmonary hypertension risk prediction model;

[0016] S4, evaluating the results of the pulmonary hypertension risk prediction model;

[0017] S5, performing internal and external validation on the pulmonary hypertension risk prediction model;

[0018] S6. Determine the final pulmonary hypertension risk prediction model and apply it.

[0019] Furthermore, in step S2, univariate logistic regression analysis is used to screen out risk factors associated with CTEPH.

[0020] Furthermore, in step S3, a logistic prediction model is applied to construct the pulmonary hypertension risk prediction model.

[0021] Furthermore, when the prediction model screens out different variables through different variable screening methods, different pulmonary hypertension risk prediction models are constructed.

[0022] Furthermore, in step S3, ROC analysis is applied to compare the effectiveness of different pulmonary hypertension risk prediction models in diagnosing CTEPH in the CTEPD population.

[0023] Furthermore, the present invention also provides a CTEPH risk prediction device based on nocturnal hypoxia combined with cardiac ultrasound indicators, including one or more processors and a memory, wherein the memory is used to store computer program code, and the computer program code is run by the processor to construct the pulmonary hypertension risk prediction model as described above.

[0024] Beneficial effects of the present invention:

[0025] The present invention constructs a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators. It only needs to use a portable sleep apnea monitor to monitor the patient's nocturnal hypoxia, and uses the hypoxia indicator combined with the cardiac ultrasound indicator to jointly diagnose the CTEPH risk. This is a non-invasive risk assessment, does not cause any harm to the patient, is easy to operate, and has good model performance. It can be popularized and promoted in hospitals at all levels across the country and has wide applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flow chart of a method for constructing a pulmonary hypertension risk prediction model according to an embodiment of the present invention;

[0027] Figure 2 Prediction model variable screening for the embodiment of the present invention Figure 1 ;

[0028] Figure 3 Prediction model variable screening for the embodiment of the present invention Figure 2 ;

[0029] Figure 4 This is a diagram showing the diagnostic efficacy of Model I and Model II for CTEPH using the ROC curve comparison model according to an embodiment of the present invention;

[0030] Figure 5 This is a diagram showing the diagnostic efficacy of Model I and PASP for CTEPH using ROC curve comparison in an embodiment of the present invention;

[0031] Figure 6 It is a nomogram of a CTEPH risk prediction model constructed based on nocturnal hypoxia and cardiac ultrasound indicators in an embodiment of the present invention;

[0032] Figure 7 It is a result evaluation calibration curve diagram of the CTEPH risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators constructed in an embodiment of the present invention;

[0033] Figure 8 The figure is an internal validation ROC curve diagram of the CTEPH risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators constructed in an embodiment of the present invention;

[0034] Figure 9 It is an external validation ROC curve diagram of the CTEPH risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators constructed in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following is a detailed description of the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0036] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside", "outside", "first", "second", etc., indicating directions or positions or sequential relationships, are based on the directions or positions or sequential relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0037] The invention discloses a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators. The nocturnal hypoxia data and cardiac ultrasound indicators of CTEPD patients are used as risk factor input data of the CTEPH risk prediction model. The predicted risk probability P of the sample CTEPH patients is calculated by the predicted probability calculation formula. The calculation formula of the risk prediction probability P is: Logit(P) is the logistic regression value of the final risk prediction probability P.

[0038] Specifically, as a pulmonary hypertension risk prediction model constructed in practical applications, the SpO2 lower than 90% time percentage parameter T90 in the CTEPD patient's nighttime hypoxia data and the pulmonary artery systolic pressure parameter PASP in the cardiac ultrasound index are used as risk factor input values ​​of the pulmonary hypertension risk prediction model, and the calculation formula of the Logit (P) is:

[0039] Logit(P)=C1+C2*T90+C3*PASP

[0040] Among them, C1, C2, and C3 are constant coefficients.

[0041] For details, see the following Table 1:

[0042] Table 1: Statistical model details table

[0043] Item Estimate Std.Error Z.value P.value (Intercept) -4.042 1.022 -3.953 0 T90 0.082 0.035 2.337 0.019 PASP 0.085 0.021 4.109 0

[0044] Table 1 gives a calculation table of related variables of a regression model; Estimate in the table is the coefficient in the prognostic evaluation model formula, that is, C1 to C3 described in the formula; a P value less than 0.05 indicates that the corresponding variable is statistically significant (that is, it is considered to be meaningful for prognosis prediction and can be included in the model). It can be seen from the table that T90 and PASP are both statistically significant. At the same time, the constants C1, C2, and C3 in the table are -4.042, 0.082, and 0.085, respectively; that is, the calculation formula of the above Logit (P) is specifically:

[0045] Logit(P)=Logit(P)=-4.042+0.082*T90+0.085*PASP.

[0046] According to the pulmonary hypertension risk prediction model and its risk prediction probability calculation formula, in actual application, if a CTEPD patient has a T90 of 20% and a PASP of 35 mmHg, then: Logit(P) = -4.042 + 0.082*20 + 0.085*35 = 0.573, convert Logit(P) to P, and use the logic function to substitute the value of Logit(P) into the formula, The calculated P value is 0.6398, P = 63.98%. For this patient, PASP is less than 40 mmHg. From the results of cardiac ultrasound, PH can be ruled out, but from the current model, the probability of this CTEPD patient with PH (or diagnosed with CTEPH) is 63.98%, suggesting that we still need to be vigilant about the possibility of CTEPH.

[0047] In a preferred embodiment, the pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators can intuitively display the risk prediction probability P of the CTEPD patient through a visual chart, and the SpO2 is less than 90% of the time percentage parameter T90 and the pulmonary artery systolic pressure parameter PASP correspond to the index score, and each of the index scores is added to correspond to the risk prediction probability P of the patient. Figure 6 As shown, in this embodiment, a visualization chart is constructed for the risk assessment model. The nomogram is similar to a visualization chart of a scoring scale. After each patient obtains the specific T90 and PASP values, the scores of each indicator can be added up to obtain the probability of clinical deterioration events in this patient. Each patient has these indicators, which can be respectively corresponded to the scales on the right side of each indicator, and then the vertical alignment columns are used to see the scores corresponding to the top Points (risk score); the scores corresponding to each indicator are added up and located at the TotalPoints (total risk score) below, and then the vertical alignment columns are used to see the Risk (risk level) at the bottom, which is the predicted probability of the patient's outcome, which can intuitively reflect the probability of the patient's disease).

[0048] On the other hand, the present invention also provides a method for constructing a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators, which specifically includes the following steps:

[0049] S1. Cardiac ultrasound examination indices, right cardiac catheter examination data, and nocturnal sleep respiratory monitoring data of CTEPD patients were collected, and the data were grouped into two groups: CTEPH group and CTEPD without PH according to the hemodynamic data measured by right cardiac catheter examination. All collected data were statistically analyzed; binary variables were set for statistical analysis, with CTEPH set as 1 and CTEPD without PH set as 0. In practical application, in order to construct the pulmonary hypertension risk prediction model of the embodiment, a total of 120 patients with a clear diagnosis of CTEPD were collected. The demographic and clinical data such as arterial oxygen saturation (SaO2), 6-min walk distance (6MWD), N-terminal pro-brain natriuretic peptide (NT-proBNP), and lung function indicators such as forced expiratory volume in 1s (FEV1), forced vital capacity (FVC), FEV1%, FEV1 / FVC%, residual volume (RV), total lung capacity (TLC), TLC%, RV / TLC%, single-breath carbon monoxide diffusing capacity (DLco-SB), DLco-SB%, and diffusing capacity divided by the alveolar volume (diffusing capacity divided by the alveolar volume) were collected. The average nighttime oxygen saturation (Peripheral Capillary Oxygen Saturation, SpO2) < 90% was defined as nocturnal hypoxia; the percentage of time when SpO2 was below 90% for more than 10% was defined as T90%, which was used to assess the severity of nocturnal hypoxia.Nocturnal minimal SpO2 (min SpO2), sleep apnea index and hypoxia index were collected; cardiac ultrasound indices such as PASP, tricuspid annular plane systolic excursion (TAPSE) and right atrial area (RA) were collected; baseline hemodynamic parameters including mPAP, PVR, right atrial pressure (RAP), cardiac output (CO) and cardiac index were collected. According to the above data, CTEPD patients were divided into two groups according to the hemodynamic data measured by right cardiac catheterization examination: CTEPD with (or called CTEPH) (n=80) or without PH (n=40), and all collected data were statistically analyzed. Statistical analysis set binary variables, CTEPH was set to 1, and CTEPD without PH was set to 0.

[0050] S2. Screen out risk factors associated with CTEPH. In this embodiment, by applying univariate logistic regression analysis, the risk factors associated with CTEPH were screened out as follows: meanSpO2, minSpO2, T90, SaO2, NT-proBNP, mPAP, PASP, RA area, TAPSE, 6MWD, RV and RV / TLC%.

[0051] S3, screening CTEPH risk prediction model variables and constructing a pulmonary hypertension risk prediction model. In this embodiment, when using the logistic prediction model to construct a pulmonary hypertension risk prediction model to screen the prediction model variables, see the attached Figure 2 As shown, according to the univariate analysis, the optimal subset ( Figure 2 ) The variables screened out are T90 and PASP; see Appendix Figure 3 As shown in Table 2 below:

[0052] Table 2: List of lasso regression screening variable results

[0053] <![CDATA[Variable. number ]]> Variable.name Coefficient lambda.type lambda.value 0 (intercept) 14.776 lambda.min 0.021 1 <![CDATA[meanSpO2]]> NA lambda.min 0.021 10 RV 0.324 lambda.min 0.021 11 RV / TLC5 NA lambda.min 0.021 2 <![CDATA[minSpO2]]> -0.017 lambda.min 0.021 3 T90 0.039 lambda.min 0.021 4 <![CDATA[SaO2]]> -0.169 lambda.min 0.021 5 <![CDATA[ NT-proBNP ]]> NA lambda.min 0.021 6 PASP 0.067 lambda.min 0.021 7 RA area NA lambda.min 0.021 8 TAPSE -0.52 lambcla.min 0.021 9 6MWD NA lambcla.min 0.021 0 (Intercept) 6.785 lambda.lse 0.064 1 meanSpO2 NA lambda.lse 0.064 10 RV NA lambda.lse 0.064 11 RV / TLC NA lambda.1se 0.064 2 minSpO2 NA lambda.lse 0.064 3 T90 0.017 lambda.lse 0.064 4 SaO2 -0.089 lambda.lse 0.064 5 <![CDATA[ NTproBNP ]]> NA lambda.lse 0.064 6 PASP 0.043 lambda.lse 0.064 7 Area NA lambda.lse 0.064 8 TAPSE NA lambda.lse 0.064 9 6hBVD NA lambda.lse 0.064

[0054] The variables selected by stepwise regression and lasso regression were T90, SaO2 and PASP.

[0055] According to the screening of the above two prediction model variables, this embodiment constructs two pulmonary hypertension risk prediction models Model I (T90 and PASP) and Model II (T90, SaO2 and PASP), and compares the advantages and disadvantages of the two results by applying ROC analysis. The ROC curve (Receiver Operating Characteristic Curve) is a statistical indicator used to evaluate the advantages and disadvantages of classification models. The basic idea is: by adjusting the discrimination threshold of the classification model, different sensitivities (Sensitivity) (true positive rate) and specificities (Specificity) complements (false positive rate, i.e. 1-specificity) can be obtained; the true positive rate and false positive rate under these different thresholds are plotted in the coordinate system to form an ROC curve. The larger the area under the ROC curve (AUC, area under the curve), the better the effect of the classification model.

[0056] For details, see the attached Figure 4 The results showed that there was no significant difference in the area under the curve (AUC) between the two models (P = 0.211). When using ROC analysis to analyze the efficacy of Model I and PASP in diagnosing CTEPH in the CTEPD population, see Appendix Figure 5 As shown in the figure, the Model I AUC of this embodiment is 0.911, which is significantly greater than PASP (AUC = 0.860), and the difference between the two is statistically significant (P = 0.019). Therefore, it can be seen that compared with the cardiac ultrasound index PASP, the Model I of this embodiment has a higher diagnostic efficacy for CTEPH. As mentioned above, the Model I diagnostic model is constructed using the training set, and the risk prediction model nomogram is constructed using the logistic prediction model as shown in the attached figure. Figure 6 shown.

[0057] S4, evaluating the results of the pulmonary hypertension risk prediction model; in this embodiment, the results of the above-mentioned Model I diagnosis model are evaluated: a. Discrimination: The model R is calculated by fitting the data of the example. 2 The model C-index (Concordance index) is 0.904, which means the discrimination of the model is 0.904. b. Calibration: Hoslem Test: X-squared = 10.001, P = 0.265. c. Calibration curve is shown in the attached Figure 7 It should be noted that R 2 It is an indicator of the goodness of fit or amount of explained variation of the regression model. R 2The value range is 0-1. The larger the value, the better the independent variable explains the variation of the dependent variable, and the better the model fit. 2 =0.566, indicating that the independent variables of nocturnal hypoxia and cardiac ultrasound indicators can predict 56.6% of the risk type variation of the dependent variable, and the fitting effect is acceptable. C-index is an indicator for evaluating the distinguishing ability of the classification model, which is expressed as the proportion of the number of combinations in which the model has correct classification results among all possible data pairs. The C-index range is 0-1, and the larger the value, the stronger the distinguishing ability of the model. In this scenario, the C-index is 0.904, indicating that the CTEPH risk prediction model can correctly distinguish 90.4% of high-risk and low-risk data pairs, and has a strong distinguishing ability.

[0058] S5, the pulmonary hypertension risk prediction model is internally and externally validated. Specifically, the internal validation is to use a test set (24 cases, including 13 cases of CTEPH) to verify that the AUC value of the diagnostic model constructed in this embodiment for diagnosing CTEPH reaches 0.898, and the 95% confidence interval is (0.843, 0.954), see Appendix Figure 8 As shown, 0.898 meets the confidence condition. External validation: All subjects of the external validation study were from the Department of Pulmonary Circulation, Shanghai Pulmonary Hospital Affiliated to Tongji University. Inpatients who met the criteria were selected, and the clinical data of the patients were obtained, including nocturnal hypoxia data obtained from portable sleep apnea monitoring, cardiac ultrasound indicators, hemodynamic data and other comprehensive information samples, a total of 48 cases. The number of cases obtained in the validation were: 1 (CTEPH) 30 cases; 0 (CTEPD without PH) 18 cases. The ROC curve is shown as follows Figure 9 As shown in , the calculated AUC is 0.876, which is close to the internal validation result and meets the confidence condition, indicating that the prognostic evaluation model performs well.

[0059] S6. Finally, after the above internal verification and external verification, a pulmonary hypertension risk prediction model (i.e., CTEPH risk assessment model) based on the above risk assessment model formula of this embodiment is obtained. In the practical application of the pulmonary hypertension risk prediction model of this embodiment, a chronic thromboembolic pulmonary hypertension risk prediction device based on nocturnal hypoxia combined with cardiac ultrasound indicators can be formed based on this risk prediction model, including one or more processors and memories, the memory is used to store computer program code, and the processor runs the computer program code to construct the pulmonary hypertension risk prediction model as described above. In actual operation, it is only necessary to use a portable sleep apnea monitor to monitor the patient's nocturnal hypoxia, and use hypoxia indicators combined with cardiac ultrasound indicators to jointly diagnose CTEPH risks. It is a non-invasive risk assessment, does not cause any harm to patients, is easy to operate, and has good model performance. It can be popularized and promoted in hospitals at all levels across the country and has wide applicability.

[0060] The above only describes the preferred embodiments of the present invention, but it cannot be understood as limiting the claims. The present invention is not limited to the above embodiments, and its specific structure is allowed to be changed. In short, all changes made within the protection scope of the independent claims of the present invention are within the protection scope of the present invention.

Claims

1. A risk prediction model for pulmonary hypertension based on nocturnal hypoxia and cardiac ultrasound indicators, characterized by: The nighttime hypoxia data and cardiac ultrasound indexes of patients with chronic thromboembolic pulmonary disease were used as the risk factor input data of the pulmonary hypertension risk prediction model. The predicted probability P of the sample chronic thromboembolic pulmonary disease patients with pulmonary hypertension was calculated by the predicted probability calculation formula. The calculation formula of the risk prediction probability P is: Logit(P) is the logistic regression value of the final risk prediction probability P.

2. The pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators according to claim 1, characterized in that: The SpO2 lower than 90% time percentage parameter T90 in the nighttime hypoxia data of the chronic thromboembolic pulmonary disease patients and the pulmonary artery systolic pressure parameter (PASP) in the cardiac ultrasound index are used as the risk factor input values ​​of the pulmonary hypertension risk prediction model, and the calculation formula of the Logit (P) is: Logit(P)=C1+C2*T90+C3*PASP Among them, C1, C2, and C3 are constant coefficients.

3. The pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators according to claim 2, characterized in that: The constant coefficients C1, C2, and C3 in the Logit (P) calculation formula are -4.042, 0.082, and 0.085, respectively. The Logit (P) calculation formula is: Logit(P)=Logit(P)=-4.042+0.082*T90+0.085*PASP.

4. The pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators according to claim 2, characterized in that: The pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators intuitively displays the risk prediction probability P of the patient with chronic thromboembolic pulmonary hypertension through a visual chart, and the SpO2 is lower than 90% of the time percentage parameter T90 and the pulmonary artery systolic pressure parameter PASP correspond to an index score, and each of the index scores is added together to correspond to the risk prediction probability P of the patient.

5. A method for constructing a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators, characterized in that: The following steps are involved: S1, collect cardiac ultrasound examination indexes, right cardiac catheter examination data and nocturnal sleep breathing monitoring data of patients with chronic thromboembolic pulmonary disease, and group the data into two groups according to the hemodynamic data measured by right cardiac catheter examination: CTEPH chronic thromboembolic pulmonary hypertension and chronic thromboembolic pulmonary disease without pulmonary hypertension. Perform statistical analysis on all collected data; set binary variables for statistical analysis, chronic thromboembolic pulmonary hypertension is set as 1, and chronic thromboembolic pulmonary disease without pulmonary hypertension is set as 0; S2, screen for risk factors associated with chronic thromboembolic pulmonary hypertension; S3, screening the variables of the risk prediction model for chronic thromboembolic pulmonary hypertension and constructing a risk prediction model for pulmonary hypertension; S4, evaluating the results of the pulmonary hypertension risk prediction model; S5, performing internal and external validation on the pulmonary hypertension risk prediction model; S6. Determine the final pulmonary hypertension risk prediction model and apply it.

6. The method for constructing a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac ultrasound indicators according to claim 5, characterized in that: In step S2, univariate logistic regression analysis is used to screen out risk factors associated with chronic thromboembolic pulmonary hypertension.

7. The method for constructing a pulmonary hypertension risk prediction model based on nocturnal hypoxia and echocardiography indices according to claim 5, characterized in that: In step S3, a logistic prediction model is used to construct the pulmonary hypertension risk prediction model.

8. The method for constructing a pulmonary hypertension risk prediction model based on nocturnal hypoxia and cardiac echocardiography indices according to claim 7, characterized in that: When the prediction model screens out different variables through different variable screening methods, different pulmonary hypertension risk prediction models are constructed.

9. The method for constructing a pulmonary hypertension risk prediction model based on nocturnal hypoxia and echocardiography indices according to claim 8, characterized in that: In step S3, ROC analysis is used to compare the effectiveness of different pulmonary hypertension risk prediction models in diagnosing chronic thromboembolic pulmonary hypertension in a population with chronic thromboembolic pulmonary disease.

10. A device for predicting the risk of chronic thromboembolic pulmonary hypertension based on nocturnal hypoxia combined with cardiac hyperindicators, characterized in that: It includes one or more processors and memories, wherein the memories are used to store computer program codes, and the computer program codes are run by the processors to construct the pulmonary hypertension risk prediction model described in any one of claims 1-9.

Images