Postpartum female stress urinary incontinence risk assessment method based on pelvic floor ultrasonic imaging technology

Abstract

The invention belongs to the technical field of medical treatment. The invention discloses a postpartum female stress urinary incontinence risk assessment method based on a pelvic floor ultrasonic imaging technology. The method comprises the following steps that S1, data are acquired and processed to form a tested object cluster; s2, performing SUI follow-up assessment on the tested object cluster; s3, dividing a data set of the tested object cluster; s4, constructing a risk assessment prediction model based on scoring; s5, constructing and visualizing a basic clinical model and a clinical-ultrasound joint model based on risk scores; and S6, performing postpartum SUI risk assessment. The system and the method have the advantages that the SII risk level can be quantitatively evaluated to formulate an individual follow-up visit scheme for implementing hierarchical dynamic management on high-risk groups to form a doctor-patient collaborative dynamic management mode, and the postpartum middle and long-term SII morbidity rate is effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of medical technology, and more specifically, to a method for assessing the risk of postpartum stress urinary incontinence in women based on pelvic floor ultrasound imaging technology. Background Technology

[0002] With socio-economic development and the innovation of perinatal medicine concepts, modern postpartum care has expanded from basic maternal and infant safety to multi-dimensional functional rehabilitation. Postpartum women face a variety of health problems, among which postpartum stress urinary incontinence (SUI) is a common complication. SUI refers to the involuntary leakage of urine from the urethral opening when abdominal pressure increases, such as during sneezing or coughing. This condition not only seriously affects women's quality of life but also places continuous pressure on socio-economic and medical resources.

[0003] Domestic and international research indicates that postpartum succumbing to uterine insufficiency (SUI) management recommendations include a follow-up period of 6 to 12 months to ensure effective monitoring and intervention. However, in actual clinical practice in China, the uneven distribution of medical resources and significant differences in healthcare systems across regions present numerous challenges to follow-up work during this stage, including an incomplete follow-up system, uneven allocation of medical resources, and insufficient maternal compliance. Furthermore, the 6-8 weeks postpartum period is a critical phase in the early postpartum period; effective interventions during this stage can help reduce the incidence of SUI within one year postpartum, representing a crucial window for resolving this clinical contradiction. Therefore, focusing on this critical prevention window in the early postpartum period, integrating multidimensional data such as early postpartum ultrasound indicators and clinical data to quantitatively assess SUI risk levels, and developing individualized follow-up plans to facilitate tiered dynamic management of high-risk groups, thereby constructing a collaborative dynamic management model between doctors and patients to reduce the mid- to long-term incidence of postpartum SUI, has become an urgent technical problem to be solved. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a method for assessing the risk level of postpartum stress urinary incontinence based on pelvic floor ultrasound imaging technology, which can quantify the assessment of SUI risk level, formulate individualized follow-up plans, implement hierarchical dynamic management of high-risk groups, form a dynamic management model of doctor-patient collaboration, and effectively reduce the incidence of postpartum SUI in the mid-to-long term.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] This invention provides a method for assessing the risk of postpartum stress urinary incontinence in women based on pelvic floor ultrasound imaging technology. The method includes the following steps:

[0007] S1. Acquire data and process it to form a cluster of subjects;

[0008] Clinical data and ultrasound data of several vaginal deliveries who underwent early postpartum pelvic floor ultrasound examinations were collected in advance at 6 to 8 weeks postpartum. The data were then processed according to predetermined standards to form a subject cluster.

[0009] S2, Follow-up evaluation of subject cluster SUI;

[0010] According to predetermined standards, a 1-year postpartum follow-up assessment was conducted on the cluster of subjects in step S1 to obtain data on the risk status of SUI.

[0011] S3. Dataset partitioning of the subject cluster;

[0012] Using the risk status data generated by SUI as a stratified variable, the subject cluster in step S2 is divided into a training queue cluster and an internal validation queue cluster according to a predetermined stratified random sampling method.

[0013] S4. Construct a risk assessment and prediction model based on scoring;

[0014] In the subject cluster in step S3, a basic clinical model is constructed based on clinical data and according to preset standards, and a clinical-ultrasound joint model is constructed based on integrated clinical data and ultrasound detection index data and according to preset standards.

[0015] S5. Construction and visualization of basic clinical models and clinical-ultrasound combined models based on risk scores;

[0016] A risk score cumulative score model was constructed using a predetermined standardized scoring method. This model was then embedded into the basic clinical model and the clinical-ultrasound combined model to form a risk score-based basic clinical model and a risk score-based clinical-ultrasound combined model. The risk score-based basic clinical model and the risk score-based clinical-ultrasound combined model were evaluated and calibrated. This resulted in a calibrated clinical-ultrasound combined model. The calibrated clinical-ultrasound combined model was used to perform SUI prediction assessment on a cluster of subjects one year postpartum to obtain SUI risk level assessment results. Based on the SUI risk level assessment results, a trend curve of the total risk score versus the predicted probability of SUI one year postpartum was plotted according to a preset standard to visualize the assessment results.

[0017] S6. Postpartum SUI risk assessment;

[0018] The relevant clinical data collected in advance by postpartum women are substituted into the calibrated clinical-ultrasound combined model in step S5 to quantitatively assess the risk level of SUI, thereby providing data support for dynamic management of doctor-patient collaboration.

[0019] Preferably, step S3 further includes the following step:

[0020] S3.1. Use the pre-set independent samples t-test analysis technique to perform data analysis and processing on the risk status data generated by SUI to obtain risk data of stratified variables;

[0021] S3.2. The data between the corresponding groups in the stratified variable risk data are tested and analyzed using the preset Mann-Whitney U test, chi-square test or Fisher test to obtain the categorical variable of the inter-group risk data. Then, the stratified random sampling method is used to process the categorical variable to divide the subject cluster into the training cohort cluster and the internal validation cohort cluster and express them as percentages.

[0022] Preferably, step S4 further includes the following step:

[0023] S4.1. Perform univariate regression initial screening on the training cohort cluster within the subject cluster, i.e., perform univariate logistic regression analysis on the stratified variable risk data of the training cohort cluster according to preset criteria, from...

[0024] The OR values ​​and their 95% CIs for each stratified variable were obtained; where OR is the odds ratio of the statistical index, and CI is...

[0025] Confidence interval;

[0026] S4.2. Set P<0.10 as the predetermined variable retention threshold, and use the preset Wald test method to test, evaluate and statistically process the differences of the stratified variables, thereby selecting 9 candidate variables to enter the multivariate analysis.

[0027] S4.3. The candidate variables are processed for multivariate regression modeling. Based on forward screening (α=0.05) and backward elimination (α=0.10), a risk assessment and prediction model is constructed according to the predetermined standardization scoring method and stepwise regression method. Only clinical data variables are included to construct the basic clinical model; clinical data variables and ultrasound detection index data are integrated to construct the clinical-ultrasound joint model.

[0028] S4.4. The multicollinearity of the basic clinical model and the clinical-ultrasound combined model is evaluated by using a pre-defined variance inflation factor (VIF). Variables with VIF ≥ 5 are removed to ensure the robustness of the two models.

[0029] Preferably, in step S5, based on the results of multifactor logistic regression analysis and referring to the predetermined principles for constructing and processing a simplified risk assessment model, the risk assessment prediction model is evaluated and visualized. The implementation steps are as follows:

[0030] S5.1 Variable preprocessing and reference value (Wij) setting: For continuous variables, grouping is performed according to clinical significance, and the reference value of each group is the midpoint of the group. The midpoint of the first and last groups is determined by the 1st percentile and the 99th percentile, respectively. Categorical variables are coded in binary (0 / 1).

[0031] S5.2, Benchmark Risk Value (W) iREF Determination: Select clinically representative groupings for each risk factor as the baseline risk value W. iREF The risk score for this group is 0. When the observed value is higher than that of the baseline group, a positive score is assigned. The score reflects the change in the risk gradient.

[0032] S5.3 Risk Distance Calculation: Based on Regression Coefficient (β) i ) and reference value (W) ij ), calculate the risk distance between each group and the baseline group = β i ×(W ij -W iREF This indicator quantifies the degree of risk deviation of different levels of risk factors relative to the baseline group;

[0033] S5.4 Standardized Transformation Coefficient (B) Setting: Select clinically significant and measurement-stable variables in the predetermined multivariate logistic regression model, and use the β coefficient value corresponding to their specific range of change as the standard.

[0034] The conversion coefficient (B) is used to establish a risk unit corresponding to 1 point.

[0035] S5.5 Risk Score Conversion: Using a predetermined standardized conversion formula, the score = [β] i ×(W ij -W iREF The calculated value is rounded to the nearest integer to retain its clinical applicability.

[0036] S5.6 Comprehensive Risk Assessment: Obtain the total risk score, which is the sum of the scores corresponding to each risk factor. Obtained through the logistic probability transformation formula:

[0037]

[0038] in ≈ constant term + βi ×W ij + B×total score, calculate the risk probability corresponding to each total score;

[0039] S5.7 Based on the total risk score cumulative score, a risk score cumulative score model is constructed according to a predetermined standardized scoring method and a preset standard. At the same time, the risk score cumulative score model is embedded into the basic clinical model and the clinical-ultrasound joint model respectively, thereby forming a basic clinical model based on risk score and a clinical-ultrasound joint model based on risk score.

[0040] S5.8. Evaluate and calibrate the basic clinical model based on risk score and the clinical-ultrasound combined model based on risk score to obtain the calibrated clinical-ultrasound combined model. Use the calibrated clinical-ultrasound combined model to perform SUI prediction assessment on the clinical data of the subject cluster 1 year postpartum, and then obtain the SUI risk level assessment results of the subject cluster including low risk, medium risk and high risk.

[0041] S5.9 Visual Presentation: Based on the SUI risk level assessment results, a trend curve of the total risk score and the predicted probability of SUI one year postpartum is drawn according to preset standards to visualize the risk assessment results, thereby intuitively showing the cumulative effect of risk.

[0042] Preferably, step S5.8 further includes the following step:

[0043] S5.8.1 Model Discrimination Evaluation

[0044] 1) Using a predetermined receiver operating characteristic curve (ROC) assessment model, ROC curves were plotted for the basic clinical model based on risk scores and the clinical-ultrasound joint model based on risk scores in the training cohort cluster and the internal validation cohort cluster to predict SUI at 1 year postpartum. The area under the curve (AUC) and 95% CI were calculated. The AUC value ranged from 0.5 to 1.0, that is, the closer the value is to 1, the better the model's predictive ability.

[0045] 2) The statistical difference in AUC between the basic clinical model based on risk score and the clinical-ultrasound joint model based on risk score was compared using the predetermined DeLong nonparametric test method, with a significance level of α=0.05. The calibrated clinical-ultrasound joint model was obtained by this method. The goodness of fit of the calibrated clinical-ultrasound joint model was reflected by the cluster AUC of the training cohort, and the generalization ability of the calibrated clinical-ultrasound joint model was reflected by the cluster AUC of the internal validation cohort.

[0046] S5.8.2 Model Calibration Verification

[0047] The Hosmer-Lemeshow goodness-of-fit test was used to evaluate the agreement between the predicted risk probabilities and actual observed values ​​of the calibrated clinical-ultrasound combined model. Based on the SUI stratification variable, the subjects were divided into 10 subgroups. The difference between the predicted positive rate and the actual positive rate in each subgroup was compared using the χ² chi-square test. The null hypothesis was that there was no statistically significant difference between the predicted and actual values. A p-value > 0.05 indicated that the calibrated clinical-ultrasound combined model had good calibration. Calibration curves were plotted for visualization analysis. The ideal calibrated clinical-ultrasound combined model showed a scatter plot close to a 45° diagonal, indicating a high degree of consistency between the predicted risk and the actual incidence.

[0048] S5.8.3 Clinical Practicality Analysis

[0049] The clinical net benefit of the calibrated clinical-ultrasound combined model was evaluated using a predetermined decision curve analysis (DCA) method according to pre-defined standards. Specifically, a decision curve coordinate system was established: the horizontal axis represented the risk threshold probability (0%-100%), and the vertical axis represented the standardized net benefit. Decision curves were plotted between the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores, and the baselines for "all intervention" and "no intervention." The net benefit value at each risk threshold point was calculated according to pre-defined standards, and the clinical utility of the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores was quantitatively compared. If the decision curve consistently exceeded the baseline within a specific threshold range, it indicated that the model had clinical application value in that risk interval. Based on this, the potential clinical net benefit of the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores in guiding early intervention decisions for postpartum SUI was evaluated.

[0050] S5.8.4 Establishment of a Calibrated Clinical-Ultrasound Combined Model

[0051] Based on training and internal validation cohort data, a risk score-based clinical-ultrasound joint model was established to predict postpartum SUI risk at 1 year by using 90% sensitivity and 90% specificity as criteria, respectively. The stability of the risk score-based clinical-ultrasound joint model in predicting the risk stratification level of postpartum SUI was evaluated by comparing the consistency of the cutoff values ​​of the two datasets. A calibrated clinical-ultrasound joint model was then developed to assess the risk of postpartum SUI in the subject cluster at 1 year.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] In this invention, clinical data and ultrasound data of several vaginal deliveries who underwent early postpartum pelvic floor ultrasound examinations were collected 6 to 8 weeks postpartum. The data were then processed according to predetermined standards to form a subject cluster. A one-year follow-up assessment was conducted on the subject cluster at a predetermined time to obtain SUI risk status data. Using the SUI risk status data as a stratification variable, the subject cluster was divided into a training cohort cluster and an internal validation cohort cluster according to predetermined proportions using a predetermined stratified random sampling method. Based on the clinical data, a basic clinical model was constructed according to predetermined standards. Based on the integrated clinical data and ultrasound data, a clinical-ultrasound combined model was constructed according to predetermined standards. A risk score cumulative score model was constructed using a predetermined standardized scoring method and embedded into the basic clinical model and the clinical-ultrasound combined model to form a risk score-based basic clinical model and a risk score-based clinical-ultrasound combined model. The risk score-based basic clinical model and the risk score-based clinical-ultrasound combined model were then evaluated. The clinical-ultrasound combined model is evaluated and calibrated to obtain a calibrated clinical-ultrasound combined model. This model is then used to predict and assess postpartum juvenile dysfunction (SUI) in a cluster of subjects one year postpartum to obtain SUI risk level assessment results. Based on the SUI risk level assessment results, a trend curve of the total risk score versus the predicted probability of SUI one year postpartum is plotted according to preset standards to visualize the assessment results. Pre-collected clinical data from postpartum women is substituted into the calibrated clinical-ultrasound combined model to quantitatively assess SUI risk levels, thus providing data support for medical staff to develop individualized follow-up plans and implement hierarchical dynamic management. This facilitates hierarchical dynamic management of high-risk groups by medical staff to form a reliable doctor-patient collaborative dynamic management model, thereby reducing the incidence of postpartum SUI in the mid-to-long term through precise control. Therefore, this invention has the advantages of being able to quantitatively assess SUI risk levels to develop individualized follow-up plans for hierarchical dynamic management of high-risk groups, forming a doctor-patient collaborative dynamic management model, and effectively reducing the incidence of postpartum SUI in the mid-to-long term. Attached Figure Description

[0054] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0055] Figure 1 This is a flowchart of a method for assessing the risk of postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology, as described in this invention.

[0056] Figure 2This is a trend curve of the total risk score and the predicted probability of SUI 1 year postpartum, based on a risk score-based basic clinical model of a risk assessment method for postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology, as described in this invention.

[0057] Figure 3 This is a trend curve of the total risk score and the predicted probability of SUI 1 year postpartum, based on a clinical-ultrasound combined model of risk score-based risk assessment method for postpartum female stress urinary incontinence using pelvic floor ultrasound imaging technology, as described in this invention. Detailed Implementation

[0058] See Figures 1 to 3 As shown, this embodiment provides a method for assessing the risk of postpartum stress urinary incontinence in women based on pelvic floor ultrasound imaging technology. The method includes the following steps:

[0059] S1. Acquire data and process it to form a cluster of subjects;

[0060] Clinical data and ultrasound data of several vaginal deliveries who underwent early postpartum pelvic floor ultrasound examinations were collected in advance at 6 to 8 weeks postpartum. The data were then processed according to predetermined standards to form a subject cluster.

[0061] S2, Follow-up evaluation of subject cluster SUI;

[0062] According to predetermined standards, a 1-year postpartum follow-up assessment was conducted on the cluster of subjects in step S1 to obtain data on the risk status of SUI.

[0063] S3. Dataset partitioning of the subject cluster;

[0064] Using the risk status data generated by SUI as a stratified variable, the subject cluster in step S2 is divided into a training queue cluster and an internal validation queue cluster according to a predetermined stratified random sampling method.

[0065] S4. Construct a risk assessment and prediction model based on scoring;

[0066] In the subject cluster in step S3, a basic clinical model is constructed based on clinical data and according to preset standards, and a clinical-ultrasound joint model is constructed based on integrated clinical data and ultrasound detection index data and according to preset standards.

[0067] S5. Construction and visualization of basic clinical models and clinical-ultrasound combined models based on risk scores;

[0068] A risk score cumulative score model was constructed using a predetermined standardized scoring method. This model was then embedded into the basic clinical model and the clinical-ultrasound combined model to form a risk score-based basic clinical model and a risk score-based clinical-ultrasound combined model. The risk score-based basic clinical model and the risk score-based clinical-ultrasound combined model were evaluated and calibrated. This resulted in a calibrated clinical-ultrasound combined model. The calibrated clinical-ultrasound combined model was used to perform SUI prediction assessment on a cluster of subjects one year postpartum to obtain SUI risk level assessment results. Based on the SUI risk level assessment results, a trend curve of the total risk score versus the predicted probability of SUI one year postpartum was plotted according to a preset standard to visualize the assessment results.

[0069] S6. Postpartum SUI risk assessment;

[0070] The relevant clinical data collected in advance by postpartum women are substituted into the calibrated clinical-ultrasound combined model in step S5 to quantitatively assess the risk level of SUI, thereby providing data support for dynamic management of doctor-patient collaboration.

[0071] In this embodiment, step S3 further includes the following step:

[0072] S3.1. Use the pre-set independent samples t-test analysis technique to perform data analysis and processing on the risk status data generated by SUI to obtain risk data of stratified variables;

[0073] S3.2. The data between the corresponding groups in the stratified variable risk data are tested and analyzed using the preset Mann-Whitney U test, chi-square test or Fisher test to obtain the categorical variable of the inter-group risk data. Then, the stratified random sampling method is used to process the categorical variable to divide the subject cluster into the training cohort cluster and the internal validation cohort cluster and express them as percentages.

[0074] In this embodiment, step S4 further includes the following step:

[0075] S4.1. Perform univariate regression screening on the training cohort cluster in the subject cluster, that is, perform univariate logistic regression analysis on the risk data of the stratified variables of the training cohort cluster according to the preset criteria, so as to obtain the OR value and 95% CI of each stratified variable.

[0076] S4.2. Set P<0.10 as the predetermined variable retention threshold, and use the preset Wald test method to test, evaluate and statistically process the differences of the stratified variables, thereby selecting 9 candidate variables to enter the multivariate analysis.

[0077] S4.3. The candidate variables are processed for multivariate regression modeling. Based on forward screening (α=0.05) and backward elimination (α=0.10), a risk assessment and prediction model is constructed according to the predetermined standardization scoring method and stepwise regression method. Only clinical data variables are included to construct the basic clinical model; clinical data variables and ultrasound detection index data are integrated to construct the clinical-ultrasound joint model.

[0078] S4.4. The multicollinearity of the basic clinical model and the clinical-ultrasound combined model is evaluated by using a pre-defined variance inflation factor (VIF). Variables with VIF ≥ 5 are removed to ensure the robustness of the two models.

[0079] In this embodiment, in step S5, based on the results of multi-factor logistic regression analysis and referring to the predetermined principles for constructing and processing a simplified risk assessment model, the risk assessment prediction model is evaluated and visualized. The implementation steps are as follows:

[0080] S5.1 Variable preprocessing and reference value (Wij) setting: For continuous variables, grouping is performed according to clinical significance, and the reference value of each group is the midpoint of the group. The midpoint of the first and last groups is determined by the 1st percentile and the 99th percentile, respectively. Categorical variables are coded in binary (0 / 1).

[0081] S5.2, Benchmark Risk Value (W) iREF Determination: Select clinically representative groupings for each risk factor as the baseline risk value W. iREF The risk score for this group is 0. When the observed value is higher than that of the baseline group, a positive score is assigned. The score reflects the change in the risk gradient.

[0082] S5.3 Risk Distance Calculation: Based on Regression Coefficient (β) i ) and reference value (W) ij ), calculate the risk distance between each group and the baseline group = β i ×(W ij -W iREF This indicator quantifies the degree of risk deviation of different levels of risk factors relative to the baseline group;

[0083] S5.4 Standardized Transformation Coefficient (B) Setting: Select clinically significant and measurement-stable variables in the predetermined multivariate logistic regression model, and use the β coefficient value corresponding to their specific range of change as the standard.

[0084] The conversion coefficient (B) is used to establish a risk unit corresponding to 1 point.

[0085] S5.5 Risk Score Conversion: Using a predetermined standardized conversion formula, the score = [β] i ×(W ij -W iREF The calculated value is rounded to the nearest integer to retain its clinical applicability.

[0086] S5.6 Comprehensive Risk Assessment: Obtain the total risk score, which is the sum of the scores corresponding to each risk factor. Obtained through the logistic probability transformation formula:

[0087]

[0088] in ≈ constant term + β i ×W ij + B×total score, calculate the risk probability corresponding to each total score;

[0089] S5.7 Based on the total risk score cumulative score, a risk score cumulative score model is constructed according to a predetermined standardized scoring method and a preset standard. At the same time, the risk score cumulative score model is embedded into the basic clinical model and the clinical-ultrasound joint model respectively, thereby forming a basic clinical model based on risk score and a clinical-ultrasound joint model based on risk score.

[0090] S5.8. Evaluate and calibrate the basic clinical model based on risk score and the clinical-ultrasound combined model based on risk score to obtain the calibrated clinical-ultrasound combined model. Use the calibrated clinical-ultrasound combined model to perform SUI prediction assessment on the clinical data of the subject cluster 1 year postpartum, and then obtain the SUI risk level assessment results of the subject cluster including low risk, medium risk and high risk.

[0091] S5.9 Visual Presentation: Based on the SUI risk level assessment results, a trend curve of the total risk score and the predicted probability of SUI one year postpartum is drawn according to preset standards to visualize the risk assessment results, thereby intuitively showing the cumulative effect of risk.

[0092] In this embodiment, step S5.8 further includes the following step:

[0093] S5.8.1 Model Discrimination Evaluation

[0094] 1) Using a predetermined receiver operating characteristic curve (ROC) assessment model, ROC curves were plotted for the basic clinical model based on risk scores and the clinical-ultrasound joint model based on risk scores in the training cohort cluster and the internal validation cohort cluster to predict SUI at 1 year postpartum. The area under the curve (AUC) and 95% CI were calculated. The AUC value ranged from 0.5 to 1.0, that is, the closer the value is to 1, the better the model's predictive ability.

[0095] 2) The statistical difference in AUC between the basic clinical model based on risk score and the clinical-ultrasound joint model based on risk score was compared using the predetermined DeLong nonparametric test method, with a significance level of α=0.05. The calibrated clinical-ultrasound joint model was obtained by this method. The goodness of fit of the calibrated clinical-ultrasound joint model was reflected by the cluster AUC of the training cohort, and the generalization ability of the calibrated clinical-ultrasound joint model was reflected by the cluster AUC of the internal validation cohort.

[0096] S5.8.2 Model Calibration Verification

[0097] The Hosmer-Lemeshow goodness-of-fit test was used to evaluate the agreement between the predicted risk probabilities and actual observed values ​​of the calibrated clinical-ultrasound combined model. Based on the SUI stratification variable, the subjects were divided into 10 subgroups. The difference between the predicted positive rate and the actual positive rate in each subgroup was compared using the χ² chi-square test. The null hypothesis was that there was no statistically significant difference between the predicted and actual values. A p-value > 0.05 indicated that the calibrated clinical-ultrasound combined model had good calibration. Calibration curves were plotted for visualization analysis. The ideal calibrated clinical-ultrasound combined model showed a scatter plot close to a 45° diagonal, indicating a high degree of consistency between the predicted risk and the actual incidence.

[0098] S5.8.3 Clinical Practicality Analysis

[0099] The clinical net benefit of the calibrated clinical-ultrasound combined model was evaluated using a predetermined decision curve analysis (DCA) method according to pre-defined standards. Specifically, a decision curve coordinate system was established: the horizontal axis represented the risk threshold probability (0%-100%), and the vertical axis represented the standardized net benefit. Decision curves were plotted between the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores, and the baselines for "all intervention" and "no intervention." The net benefit value at each risk threshold point was calculated according to pre-defined standards, and the clinical utility of the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores was quantitatively compared. If the decision curve consistently exceeded the baseline within a specific threshold range, it indicated that the model had clinical application value in that risk interval. Based on this, the potential clinical net benefit of the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores in guiding early intervention decisions for postpartum SUI was evaluated.

[0100] S5.8.4 Establishment of a Calibrated Clinical-Ultrasound Combined Model

[0101] Based on training and internal validation cohort data, a risk score-based clinical-ultrasound joint model was established to predict postpartum SUI risk at 1 year by using 90% sensitivity and 90% specificity as criteria, respectively. The stability of the risk score-based clinical-ultrasound joint model in predicting the risk stratification level of postpartum SUI was evaluated by comparing the consistency of the cutoff values ​​of the two datasets. A calibrated clinical-ultrasound joint model was then developed to assess the risk of postpartum SUI in the subject cluster at 1 year.

[0102] In this embodiment, OR is the ratio of statistical indicators, and CI is the confidence interval.

[0103] In this embodiment, the simplified risk assessment model can be constructed by referring to the Framinghan cardiovascular risk assessment model as needed.

[0104] In this embodiment, firstly, clinical data and ultrasound data of several vaginal deliveries who underwent early postpartum pelvic floor ultrasound examinations are collected 6 to 8 weeks postpartum. The data are then processed according to predetermined standards to form a subject cluster. Secondly, a 1-year follow-up assessment is conducted on the subject cluster at a predetermined time to obtain SUI risk status data. Using the SUI risk status data as a stratification variable, a predetermined stratified random sampling method is used to divide the subject cluster into a training cohort cluster and an internal validation cohort cluster according to a predetermined ratio. Based on the clinical data, a basic clinical model is constructed according to predetermined standards. Based on the integrated clinical data and ultrasound data, a clinical-ultrasound combined model is constructed according to predetermined standards. Then, a risk score cumulative score model is constructed using a predetermined standardized scoring method and embedded into the basic clinical model and the clinical-ultrasound combined model to form a risk score-based basic clinical model and a risk score-based clinical-ultrasound combined model. The study involved evaluating and calibrating a basic clinical model based on risk scores and a combined clinical-ultrasound model based on risk scores to obtain a calibrated combined clinical-ultrasound model. This calibrated model was then used to predict and assess postpartum juvenile dysfunction (SUI) in a cluster of subjects one year after delivery, resulting in SUI risk level assessments. Based on these SUI risk level assessments, a trend curve was plotted between the total risk score and the predicted probability of SUI one year postpartum, according to pre-defined standards, to visualize the assessment results. Finally, pre-collected clinical data from postpartum women was substituted into the calibrated combined clinical-ultrasound model to quantitatively assess SUI risk levels, classifying them into low, medium, and high risk levels. This provides data support for healthcare professionals to develop individualized follow-up plans and implement tiered dynamic management, facilitating tiered dynamic management of high-risk groups and forming a reliable doctor-patient collaborative dynamic management model. Ultimately, this precise control helps reduce the incidence of postpartum SUI in the medium to long term.

[0105] This example illustrates a postpartum 1-year 1SUI1 risk assessment case:

[0106] Data analysis in this study was performed using SPSS 25.0 (IBM, USA) and Stata 18.0 (Stata Corp, USA). Two-tailed tests were used, with a p-value <0.05 considered statistically significant.

[0107] 1. Enrollment and baseline data

[0108] Study Subject Cluster: Women who underwent standardized early postpartum (6-8 weeks postpartum) pelvic floor ultrasound examination at our hospital between September 2022 and February 2024 via vaginal delivery were recruited. Clinical data and ultrasound measurements were collected, and a standardized SUI (Self-Induced Ulcer Indication) follow-up assessment was conducted one year postpartum. Details are as follows:

[0109] A total of 243 postpartum women completed the entire study process. Upon evaluation, 97 women (39.9%) developed SUI within one year postpartum. Using stratified random sampling (stratification variable being SUI within one year postpartum), the study subjects were randomly divided into a training cohort cluster (training group) (n=169, including 68 SUI patients) and an internal validation cohort cluster (validation group) (n=74, including 29 SUI patients) at a ratio of 7:3.

[0110] 2. Construction and visualization of a scoring-based risk assessment and prediction model

[0111] Univariate regression for initial screening, multivariate regression for model construction:

[0112] Table 2-1 Results of univariate and multivariate analyses of the training group

[0113]

[0114] Note: BMI is Body Mass Index; BND is Bladder Neck Mobility; LHA is Ligament Anal Hiatus Area.

[0115] 3. Construction of basic clinical models and clinical-ultrasound combined models based on risk scores

[0116] Based on the results of multivariate logistic regression analysis, a standardized scoring method was used to integrate the basic clinical model and...

[0117] The clinical-ultrasound combined model was transformed into a basic clinical model based on risk score and a clinical-ultrasound combined model based on risk score, which are nested with a risk score cumulative score model.

[0118] The basic clinical model based on risk scoring is constructed as follows (see Table 2-2 for details): According to the regression coefficient β of each variable, the scoring criteria for each indicator are as follows: ①BMI: <24 kg / m 2 0 points, ≥24 kg / m 2 ① Assign 3 points; ② Age at delivery: <30 years old, 0 points; 30-34 years old, 1 point; ≥35 years old, 2 points; ③ Urinary incontinence during pregnancy: no history of the condition, 0 points; history of the condition, 2 points. The total score of this scoring system is the sum of the scores of each indicator, with a theoretical score range of 0-7 points, where 7 points is the highest risk threshold.

[0119] Table 2-2 Risk Score Results Based on Basic Clinical Model of Risk Score

[0120]

[0121] Note 1) Reference value (W) ij = Between-group median;

[0122] 2) *Reference values ​​for BMI and delivery age (W) for the first and last groups ij Determined using the 1st percentile and the 99th percentile respectively;

[0123] 3) B=5β, age=5×0.12=0.6.

[0124] Based on a risk score-based clinical-ultrasound combined model (see Tables 2-3 for details): In addition to clinical indicators, [the model was] incorporated...

[0125] After ultrasound parameters were obtained, the scoring criteria for each variable were adjusted as follows: ①BMI: <24 kg / m 2 0 points, ≥24 kg / m 2 ① 5 points; ② Urinary incontinence during pregnancy: 0 points for none, 4 points for present; ③ BND: <25mm: 0 points, 25-34mm: 3 points, 35-44mm: 5 points, ≥45mm: 7 points; ④ Infundibulum sign of the internal urethral meatus: 0 points for not detected, 6 points for detected. The theoretical total score range of this combined model is extended to 0-22 points, where 22 points corresponds to the highest risk threshold.

[0126] Table 2-3 Risk score results of the clinical-ultrasound combined model based on risk scores

[0127]

[0128] Note 1) Reference value (W) ij = Between-group median;

[0129] 2) *Reference values ​​for the first and last groups of BMI and BND (W) ij Determined using the 1st percentile and the 99th percentile respectively;

[0130] 3) B=5β, BND =5×0.06=0.3.

[0131] 4. Visualization of risk score results from the basic clinical model and the combined clinical-ultrasound model based on risk scores.

[0132] Trend curves were plotted between the total risk score and the predicted probability of SUI at 1 year postpartum using a basic clinical model based on risk scores (see appendix for details). Figure 2 (See attached chart for trend curves of total risk score and predicted probability of SUI at 1 year postpartum using a combined clinical-ultrasound model based on risk scores) Figure 3This study visualized the risk score results of both the basic clinical model and the combined clinical-ultrasound model based on risk scores. The results showed that the predictive probabilities of both models increased. In the basic clinical model based on risk scores, the predictive probability of postpartum SUI corresponding to baseline risk (0 points) was 12.59%, and when the score rose to the highest risk threshold (7 points), the corresponding predictive probability was 90.57%. The combined clinical-ultrasound model based on risk scores, constructed by incorporating quantitative parameters of pelvic floor ultrasound, achieved more optimized clinical applicability. Its predictive probability for baseline risk (0 points) decreased to 4.33%, while the predictive probability corresponding to the highest risk threshold (22 points) increased to 97.08%. This indicates that the combined clinical-ultrasound model based on risk scores can more accurately distinguish between different risk levels in postpartum SUI risk assessment one year after delivery.

[0133] 5. Performance evaluation of the basic clinical model and clinical-ultrasound combined model of SUI based on risk scores 1 year postpartum

[0134] 1) Model discrimination evaluation results

[0135] The efficacy of the risk score-based basic clinical model and the combined clinical-ultrasound model in predicting postpartum SUI at 1 year was compared using ROC curves. In the training group, the AUC of the risk score-based basic clinical model was 0.723 (95% CI: 0.649–0.789), while the AUC of the risk score-based combined clinical-ultrasound model increased to 0.825 (95% CI: 0.760–0.879). The difference in AUC between the two models was statistically significant according to the DeLong test (P = 0.005). In the validation group, the AUC of the risk score-based basic clinical model was [missing data].

[0136] The AUC of the basic clinical model based on risk scores was 0.722 (95% CI: 0.605-0.820), while the AUC of the clinical-ultrasound combined model based on risk scores remained at a high level (0.818, 95% CI: 0.711-0.898). The AUCs of both models were very close in both the training and validation groups, suggesting good model stability. However, the AUC values ​​of the basic clinical model based on risk scores were below 0.75 in both datasets, indicating relatively limited overall discriminative power.

[0137] 2) Model calibration verification results

[0138] In the training group, the Hosmer-Lemeshow goodness-of-fit test showed that the χ² test statistic for the basic clinical model based on risk scores was 5.30 (P=0.380), and for the combined clinical-ultrasound model based on risk scores, it was 12.67 (P=0.124), with no statistically significant difference (P>0.05). This suggests that the predicted probabilities of both the basic clinical model and the combined clinical-ultrasound model based on risk scores are in good agreement with the actual observed outcome distributions. The validation group analysis showed a similar trend; the HL test results for the basic clinical model based on risk scores (χ²=2.12, P=0.832) and the combined clinical-ultrasound model based on risk scores (χ²=4.79, P=0.686) showed no statistically significant difference (P>0.05), confirming that the model calibration ability remained stable between the training and validation groups.

[0139] 3) Results of clinical applicability analysis

[0140] DCA curves were used to compare the clinical utility of the baseline clinical model and the combined clinical-ultrasound model based on risk scores. In the training group, DCA analysis showed that the decision curves of both models were above the reference line representing extreme decision strategies, with the risk score-based combined clinical-ultrasound model exhibiting a superior net benefit advantage within the threshold probability range of 0.2–1.0. The validation group results further confirmed that the clinical applicability of the risk score-based combined clinical-ultrasound model was superior to that of the baseline clinical model: when the risk threshold probability increased to 0.7, the net benefit of the risk score-based baseline clinical model was already below the baseline level, while the risk score-based combined clinical-ultrasound model did not show a similar situation until the threshold probability reached 0.8, suggesting that the risk score-based combined clinical-ultrasound model maintains stable efficacy across a wider range of clinical decision thresholds.

[0141] 6. Establishment of dual cutoff values ​​for the clinical-ultrasound combined model based on risk scores

[0142] Based on training and validation data, with a sensitivity of 90% and a specificity of 90% as optimization targets, a dual cutoff value was set for the clinical-ultrasound joint model based on risk score to predict the risk of SUI 1 year postpartum. Thus, a calibrated clinical-ultrasound joint model was obtained to assess the risk of SUI 1 year postpartum in the subject cluster.

[0143] Table 2-4 Dual cutoff values ​​of the clinical-ultrasound joint model based on risk scores in the training and validation groups.

[0144]

[0145] In summary, the present invention employs the above-mentioned method, which has the advantages of being able to quantitatively assess the risk level of SUI in order to formulate individualized follow-up plans for implementing hierarchical dynamic management of high-risk groups, forming a dynamic management model of doctor-patient collaboration, and effectively reducing the incidence of SUI in the mid-to-long term after childbirth.

[0146] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this invention.

Claims

1. A method for assessing the risk of postpartum stress urinary incontinence in women based on pelvic floor ultrasound imaging technology, characterized in that, The method includes the following steps: S1. Acquire data and process it to form a subject cluster: Collect clinical data and ultrasound test data of several vaginal deliveries who have undergone early postpartum pelvic floor ultrasound examinations 6 to 8 weeks postpartum. Process the data according to preset standards to form a subject cluster. S2. Follow-up assessment of SUI in subject cluster: Follow-up assessment of the subject cluster in step S1 was carried out 1 year postpartum according to predetermined standards to obtain data on the risk status of SUI. S3. Data set division of the subject cluster: Using the risk status data generated by SUI as the stratification variable, the subject cluster in step S2 is divided into a training queue cluster and an internal validation queue cluster according to a predetermined stratified random sampling method. S4. Construct a risk assessment and prediction model based on scoring: In the subject cluster of step S3, a basic clinical model is constructed based on clinical data and according to preset standards, and a clinical-ultrasound joint model is constructed based on integrated clinical data and ultrasound detection index data and according to preset standards. S5. Construction and Visualization of Basic Clinical Model and Clinical-Ultrasound Combined Model Based on Risk Score: A cumulative risk score model is constructed using a predetermined standardized scoring method. This model is then embedded into the basic clinical model and the clinical-ultrasound combined model to form a basic clinical model and a clinical-ultrasound combined model based on risk score. The basic clinical model and the clinical-ultrasound combined model based on risk score are evaluated and calibrated. A calibrated clinical-ultrasound combined model is obtained. The calibrated clinical-ultrasound combined model is used to perform SUI prediction assessment on a cluster of subjects one year postpartum to obtain SUI risk level assessment results. Based on the SUI risk level assessment results, a trend curve of the total risk score and the predicted probability of SUI one year postpartum is plotted according to a preset standard to visualize the assessment results. S6. Postpartum SUI Risk Assessment: Substitute the relevant clinical data collected in advance by postpartum women into the calibrated clinical-ultrasound combined model in step S5 to quantitatively assess the SUI risk level, thereby providing data support for dynamic management of doctor-patient collaboration.

2. The method for assessing the risk of postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology according to claim 1, characterized in that, Step S3 also includes the following steps: S3.

1. Use the pre-set independent samples t-test analysis technique to perform data analysis and processing on the risk status data generated by SUI to obtain risk data of stratified variables; S3.

2. The data between corresponding groups in the stratified variable risk data are tested and analyzed using the preset Mann-Whitney U test, chi-square test or Fisher test to obtain the categorical variable of the inter-group risk data. Then, the stratified random sampling method is used to process the categorical variable to divide the subject cluster into the training cohort cluster and the internal validation cohort cluster and express them as percentages.

3. The method for assessing the risk of postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology according to claim 2, characterized in that, Step S4 also includes the following steps: S4.

1. Perform univariate regression screening on the training cohort cluster in the subject cluster, that is, perform univariate logistic regression analysis on the risk data of the stratified variables of the training cohort cluster according to the preset criteria, so as to obtain the OR value and 95% CI of each stratified variable. S4.

2. Set P<0.10 as the predetermined variable retention threshold, and use the preset Wald test method to test, evaluate and statistically process the differences of the stratified variables, thereby selecting 9 candidate variables to enter the multivariate analysis. S4.

3. The candidate variables are processed for multivariate regression modeling. Based on forward screening (α=0.05) and backward elimination (α=0.10), a risk assessment and prediction model is constructed according to the predetermined standardization scoring method and stepwise regression method. Only clinical data variables are included to construct the basic clinical model; clinical data variables and ultrasound detection index data are integrated to construct the clinical-ultrasound joint model. S4.

4. The multicollinearity of the basic clinical model and the clinical-ultrasound combined model is evaluated by using a pre-defined variance inflation factor (VIF). Variables with VIF ≥ 5 are removed to ensure the robustness of the two models.

4. The method for assessing the risk of postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology according to claim 3, characterized in that, In step S5, based on the results of multifactor logistic regression analysis and referring to the predetermined principles for constructing and processing a simplified risk assessment model, the risk assessment prediction model is evaluated and visualized. The implementation steps are as follows: S5.1 Variable preprocessing and reference value (Wij) setting: For continuous variables, grouping is performed according to clinical significance, and the reference value of each group is the midpoint of the group. The midpoint of the first and last groups is determined by the 1st percentile and the 99th percentile, respectively. Categorical variables are coded in binary (0 / 1). S5.2, Benchmark Risk Value (W) iREF Determination: Select clinically representative groupings for each risk factor as the baseline risk value W. iREF The risk score for this group is 0. When the observed value is higher than that of the baseline group, a positive score is assigned. The score reflects the change in the risk gradient. S5.3 Risk Distance Calculation: Based on Regression Coefficient (β) i ) and reference value (W) ij ), calculate the risk distance between each group and the baseline group = β i ×(W ij -W iREF This indicator quantifies the degree of risk deviation of different levels of risk factors relative to the baseline group; S5.4 Standardized Transformation Coefficient (B) Setting: Select clinically significant and measurement-stable variables in the predetermined multivariate logistic regression model, and use the β coefficient value corresponding to their specific range of change as the standard. The conversion coefficient (B) is used to establish a risk unit corresponding to 1 point. S5.5 Risk Score Conversion: Using a predetermined standardized conversion formula, the score = [β] i ×(W ij -W iREF The calculated value is rounded to the nearest integer to retain its clinical applicability. S5.6 Comprehensive Risk Assessment: Obtain the total risk score, which is the sum of the scores corresponding to each risk factor. Obtained through the logistic probability transformation formula: in ≈ constant term + β i ×W ij + B×total score, calculate the risk probability corresponding to each total score; S5.7 Based on the total risk score cumulative score, a risk score cumulative score model is constructed according to a predetermined standardized scoring method and a preset standard. At the same time, the risk score cumulative score model is embedded into the basic clinical model and the clinical-ultrasound joint model respectively, thereby forming a basic clinical model based on risk score and a clinical-ultrasound joint model based on risk score. S5.

8. Evaluate and calibrate the basic clinical model based on risk score and the clinical-ultrasound combined model based on risk score to obtain the calibrated clinical-ultrasound combined model. Use the calibrated clinical-ultrasound combined model to perform SUI prediction assessment on the clinical data of the subject cluster 1 year postpartum, and then obtain the SUI risk level assessment results of the subject cluster including low risk, medium risk and high risk. S5.9 Visual Presentation: Based on the SUI risk level assessment results, a trend curve of the total risk score and the predicted probability of SUI one year postpartum is drawn according to preset standards to visualize the risk assessment results, thereby intuitively showing the cumulative effect of risk.

5. The method for assessing the risk of postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology according to claim 4, characterized in that, Step S5.8 further includes the following steps: S5.8.1 Model Discrimination Evaluation 1) Using a predetermined receiver operating characteristic (ROC) assessment model, ROC curves were plotted for predicting SUI at 1 year postpartum in the training cohort cluster and the internal validation cohort cluster using the basic clinical model based on risk score and the clinical-ultrasound joint model based on risk score, respectively. The area under the curve (AUC) and 95% CI were calculated. 2) The statistical difference in AUC between the basic clinical model based on risk score and the clinical-ultrasound joint model based on risk score was compared using the predetermined DeLong nonparametric test method, with a significance level of α=0.

05. The calibrated clinical-ultrasound joint model was obtained by this method. The goodness of fit of the calibrated clinical-ultrasound joint model was reflected by the cluster AUC of the training cohort, and the generalization ability of the calibrated clinical-ultrasound joint model was reflected by the cluster AUC of the internal validation cohort. S5.8.2 Model Calibration Verification The Hosmer-Lemeshow goodness-of-fit test was used to evaluate the agreement between the predicted risk probabilities and actual observed values ​​of the calibrated clinical-ultrasound combined model. Based on the SUI stratification variable, the subjects were divided into 10 subgroups. The difference between the predicted positive rate and the actual positive rate in each subgroup was compared using the χ² chi-square test. The null hypothesis was that there was no statistically significant difference between the predicted and actual values. A p-value > 0.05 indicated that the calibrated clinical-ultrasound combined model had good calibration. Calibration curves were plotted for visualization analysis. The ideal calibrated clinical-ultrasound combined model showed a scatter plot close to a 45° diagonal, indicating a high degree of consistency between the predicted risk and the actual incidence. S5.8.3 Clinical Practicality Analysis The clinical net benefit of the calibrated clinical-ultrasound combined model was evaluated using a predetermined decision curve analysis (DCA) method according to pre-defined standards. Specifically, a decision curve coordinate system was established: the horizontal axis represented the risk threshold probability (0%-100%), and the vertical axis represented the standardized net benefit. Decision curves were plotted between the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores, and the baselines for "all intervention" and "no intervention." The net benefit value at each risk threshold point was calculated according to pre-defined standards, and the clinical utility of the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores was quantitatively compared. If the decision curve consistently exceeded the baseline within a specific threshold range, it indicated clinical application value within that risk interval. Based on this, the potential clinical net benefit of the basic clinical model based on risk scores and the clinical-ultrasound combined model based on risk scores in guiding early intervention decisions for postpartum SUI was evaluated. S5.8.4 Establishment of a Calibrated Clinical-Ultrasound Combined Model Based on training and internal validation cohort data, a risk score-based clinical-ultrasound joint model was established to predict postpartum SUI risk at 1 year by using 90% sensitivity and 90% specificity as criteria, respectively. The stability of the risk score-based clinical-ultrasound joint model in predicting the risk stratification level of postpartum SUI was evaluated by comparing the consistency of the cutoff values ​​of the two datasets. A calibrated clinical-ultrasound joint model was then developed to assess the risk of postpartum SUI in the subject cluster at 1 year.

6. The method for assessing the risk of postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology according to claim 5, characterized in that, In step S5.8.1, the area under the curve (AUC) ranges from 0.5 to 1.0, meaning that the closer the value is to 1, the better the model's predictive ability.

7. The method for assessing the risk of postpartum female stress urinary incontinence based on pelvic floor ultrasound imaging technology according to claim 5, characterized in that, OR stands for odds ratio, and CI stands for confidence interval.

Images