Female reproductive aging evaluation method

Abstract

The invention relates to the technical field of female reproductive aging evaluation, and discloses a female reproductive aging evaluation method. The method comprises the following steps: collecting related data of female infertility, and obtaining an accumulated live birth condition after single controlled ovarian stimulation; selecting age, anti-mullerian tube hormone, sinus follicle counting and basic follicle-stimulating hormone as main prediction indexes of accumulated live production by utilizing a random forest model, and verifying related curve characteristics which cannot be linearized; based on a generalized additive model, accumulated live production is well predicted by using the indexes and mapped to the corresponding age, and a reproductive aging evaluation index-reproductive age is obtained. The prediction model is still high in accuracy in an external data set, and the method has universality and generalization performance for distinguishing crowds and evaluating fertility by reproductive age. The method has the advantages that the constructed reproductive age model breaks through a linear framework of traditional time and age, and the accuracy of reproductive aging and female fertility evaluation is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of female reproductive aging assessment, and specifically to a method for assessing female reproductive aging. Background Art

[0002] With the development of social economy, the age of a woman's first childbirth is postponed, and the proportion of advanced maternal age pregnancies increases year by year. However, a woman's fertility gradually declines with age. Accurately assessing a woman's reproductive status and degree of reproductive aging is crucial for the timely diagnosis and treatment of infertility.

[0003] Currently, clinically, ovarian reserve indicators are usually used to evaluate female reproductive aging, such as age, anti-Müllerian hormone, antral follicle count, basal follicle-stimulating hormone, etc. Existing ovarian reserve assessment models mainly predict the risk of diminished ovarian reserve (DOR) and the age of menopause, and the assessment of fertility is relatively limited. In addition, existing clinical assessments mostly rely on the empirical integration of indicators, with subjective biases and a lack of an objective and quantitative comprehensive model. More complexly, age, ovarian reserve indicators, etc. are mostly curvilinearly related to fertility and to each other, and it is impossible to objectively and accurately assess only through simple calculations.

[0004] To solve the above problems, a comprehensive and objective method for assessing female reproductive aging is proposed in this application. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for assessing female reproductive aging to solve the problem in the prior art that there is no reproductive age model with fertility as the core and unable to reasonably integrate multiple indicators as proposed in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A method for assessing female reproductive aging, the method comprising the following steps: Step 1, collect sample data and obtain sample characteristics; Step 1.1, set the sample data collection criteria, patients meeting the criteria are included in the follow-up, and the final outcome - cumulative live birth of a single oocyte retrieval is obtained; Step 1.2, select sample characteristics related to pregnancy outcomes from the sample data according to preset criteria; Step 2, perform transformation and imputation on the sample characteristics described in Step 1.2, and the specific transformation and imputation process is as follows: Step 2.1, transform and encode the sample characteristics: perform standardization and normalization on the obtained data such as age, anti-Müllerian hormone, antral follicle count, basal follicle-stimulating hormone, etc. to make them conform to the input for fitting and training the model; Step 2.2: If there are missing values in some of the sample features, for the null values of different features, use the multiple imputation method to fill in the features with missing values in the sample features. If there are no missing values, there is no need to fill in. Finally, a complete sample without missing values is formed and enters Step 3; Step 3: Use the random forest model to rank the importance of candidate predictive variables for the complete sample formed in Step 2.2, and the key variables selected are entered into Step 4 to construct a prediction model; Step 4: Divide the data of the complete sample into a training set and a test set at a sample ratio of 8:2. Construct a female reproductive aging assessment model based on the generalized additive model for the training set samples formed in Steps 2, 3, and 4: Use the processed data as independent variables and the cumulative live birth rate within 2 years after a single oocyte retrieval as the dependent variable to construct a model. The model formula is , where y is the cumulative live birth, and x i are age, anti-Müllerian hormone, antral follicle count, and basal follicle-stimulating hormone after normalization and standardization, f i is a smooth function, g is a link function, and ε i is the error term; Step 5: Based on the constructed model, calculate the predicted value of the cumulative live birth rate for each patient and map this value to the corresponding age to obtain the reproductive age; Step 6: Use the female reproductive aging assessment model based on the generalized additive model constructed in Steps 4 and 5 and the reproductive age calculation to predict the degree of reproductive aging for the test set; Step 7: Use the female reproductive aging assessment model based on the generalized additive model selected in Step 6 to verify in external samples: Compare the heterogeneity and discrimination of cumulative live birth between the reproductive age and the actual age. If the cumulative live birth heterogeneity of the reproductive age is lower and the discrimination is higher, then the verification model is effective.

[0007] In the data processing step, it also includes using age, anti-Müllerian hormone, antral follicle count, and basal follicle-stimulating hormone after normalization and standardization as independent variables, using fresh embryo transfer live birth / cumulative live birth or not as the dependent variable, and using the adjusted odds ratio to compare the efficacy of different outcome indicators in evaluating aging.

[0008] Before the model construction step, it also includes analyzing the correlation between age, anti-Müllerian hormone, antral follicle count, basal follicle-stimulating hormone and the cumulative live birth rate, confirming that it is curvilinear correlation, and it cannot be linearized by exponential function, power function, logarithmic function, reciprocal transformation, Box-Cox method: The correlation before and after processing is evaluated by the generalized additive model, and the degrees of freedom (edf values) are all greater than 1.5, and the significance values (p values) are all less than 0.05.

[0009] After the data acquisition step, it further includes calculating the quartile / median ratio of the anti-Müllerian hormone, antral follicle count, and basal follicle-stimulating hormone indicators for women within 1 year of actual age. If the ratio is greater than 25%, it indicates strong ovarian reserve heterogeneity.

[0010] In the model verification step, the cumulative live birth heterogeneity and discrimination between reproductive age and actual age are compared. If, relative to the actual age, the interquartile range of the expected live births within 1 year of reproductive age is reduced by half, and the discrimination of advanced age for live births is higher, then the model is determined to have robustness and extrapolability.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The reproductive age model constructed by the present invention breaks through the limitations of the traditional chronological age linear framework. For the first time, the age corresponding to the expected live birth rate is used as the measurement dimension, which can accurately and intuitively describe the reproductive aging process, and effectively improves the accuracy of the assessment of reproductive aging and female fertility.

[0012] 2. The present invention integrates multi-dimensional data such as age, ovarian reserve, and cumulative live birth rate, and is supported by external verification data, providing a more scientific and objective basis for clinicians to evaluate the reproductive health of women of childbearing age with fertility needs, helping to diagnose and treat infertility in a timely manner, and having important significance for improving the global fertility dilemma. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic flowchart of a method for assessing female reproductive aging according to the present invention; Figure 2 It is a model diagram of a method for assessing female reproductive aging according to the present invention; Figure 3 It is a comparison of the efficacy of different fertility outcomes of a method for assessing female reproductive aging according to the present invention; Figure 4 It is a schematic diagram of the curve correlation of a method for assessing female reproductive aging according to the present invention; Figure 5 It is a schematic diagram of the difference in ovarian reserve indicators of women of the same age (within 1 year) of a method for assessing female reproductive aging according to the present invention; Figure 6 It is a schematic diagram of the heterogeneity of ovarian reserve and the difference in pregnancy outcomes during reproductive aging of a method for assessing female reproductive aging according to the present invention; Figure 7 It is a schematic diagram of the fitted cumulative live birth rate of a method for assessing female reproductive aging according to the present invention; Figure 8 It is a schematic diagram of the number and cumulative live birth rate of women with ovarian reserve decline, good ovarian reserve but poor live birth, and good ovarian reserve and good live birth of a method for assessing female reproductive aging according to the present invention; Figure 9 Schematic diagram of the predictive value of the generalized additive model of a female reproductive aging assessment method of the present invention for cumulative live births in internal and external datasets; Figure 10 Schematic diagram of cumulative live birth fitting and mapping of a female reproductive aging assessment method of the present invention; Figure 11 Schematic diagram of the distinguishing features of reproductive aging and ovarian reserve decline defined by reproductive age in a female reproductive aging assessment method of the present invention; Figure 12 Schematic diagram of the heterogeneity of cumulative live births with a reproductive age lower than the actual age in external data validation of a female reproductive aging assessment method of the present invention; Figure 13 Schematic diagram of the discrimination of cumulative live births with a reproductive age higher than the actual age in external data validation of a female reproductive aging assessment method of the present invention. Detailed implementation manners

[0014] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0015] Please refer to Figures 1-13 , the present invention provides a technical solution: a female reproductive aging assessment method, and this method includes the following steps: Step 1, collect sample data and obtain sample characteristics; Step 1.1, set the sample data collection criteria, patients meeting the criteria are included in the follow-up, and the final outcome - cumulative live births of a single oocyte retrieval is obtained.

[0016] Step 1.2, select sample characteristics related to pregnancy outcomes in the sample data according to the preset criteria; Step 2, perform transformation and imputation on the sample characteristics described in step 1.2, and the specific transformation and imputation process is as follows: Step 2.1, transform and encode the sample characteristics: perform standardization and normalization processing on the obtained data such as age, anti-Müllerian hormone (AMH), antral follicle count (AFC), basal follicle-stimulating hormone (bFSH), etc., so that it conforms to the input for fitting and training the model; Step 2.2, if there are missing values in some of the sample characteristics, for the null values of different characteristics, use the method of multiple imputation to fill the characteristics with missing values in the sample characteristics. If there are no missing values, there is no need to fill, and finally a complete sample without missing values is formed and enters step 3; Step 3: Use the random forest model to rank the importance of candidate predictive variables for the complete sample formed in Step 2.2, and select the key variables obtained through screening to enter Step 4 for constructing a prediction model; Step 4: Divide the data of the complete sample into a training set and a test set at a sample ratio of 8:2. Construct a female reproductive aging assessment model based on the generalized additive model for the training set samples formed in Steps 2, 3, and 4: Use the processed data as independent variables and the cumulative live birth rate within 2 years after a single oocyte retrieval as the dependent variable to construct a model. The model formula is , where y is the cumulative live birth, and x i are age, anti-Müllerian hormone (AMH), antral follicle count (AFC), and basal follicle-stimulating hormone (bFSH) after normalization and standardization, f i is a smooth function, g is a link function, and ε i is the error term; Step 5: Based on the constructed model, calculate the predicted value of the cumulative live birth rate for each patient, and map this value to the corresponding age to obtain the reproductive age; Step 6: Use the female reproductive aging assessment model and reproductive age calculation based on the generalized additive model constructed in Steps 4 and 5 to predict the degree of reproductive aging for the test set; Step 7: Use the female reproductive aging assessment model based on the generalized additive model selected in Step 6 to verify in an external sample: Compare the heterogeneity and discrimination of cumulative live birth between the reproductive age and the actual age. If the cumulative live birth heterogeneity of the reproductive age is lower and the discrimination is higher, then the validation model is effective.

[0017] Furthermore, in the data processing step, it also includes using age, anti-Müllerian hormone (AMH), antral follicle count (AFC), and basal follicle-stimulating hormone (bFSH) after normalization and standardization as independent variables, using the live birth / cumulative live birth of fresh embryo transfer as the dependent variable, and using the adjusted odds ratio to compare the efficacy of different outcome indicators in evaluating aging.

[0018] Furthermore, before the model construction step, it also includes analyzing the correlation between age, anti-Müllerian hormone (AMH), antral follicle count (AFC), basal follicle-stimulating hormone (bFSH), and the cumulative live birth rate, confirming that it is a curvilinear correlation, and it cannot be linearized through exponential function, power function, logarithmic function, reciprocal transformation, and Box-Cox method: The correlation before and after processing is evaluated by the generalized additive model, and the degrees of freedom (edf values) are all greater than 1.5, and the significance values, as shown in the appendix Figure 4 shown, also known as p-values, are all less than 0.05.

[0019] Further, after the data acquisition step, it further includes calculating the quartile / median ratio of anti-Müllerian hormone (AMH), antral follicle count (AFC), and basal follicle-stimulating hormone (bFSH) in women within 1 year of actual age. If the ratio is greater than 25%, it indicates strong ovarian reserve heterogeneity.

[0020] Further, in the model validation step, the cumulative live birth heterogeneity and discrimination between reproductive age and actual age are compared. Relative to the actual age, the interquartile range of expected live births within 1 year of reproductive age is reduced by half, and the discrimination of advanced age on live birth is higher, then it is determined that the model has robustness and extrapolability.

[0021] Example Data collection: According to the population inclusion criteria, relevant data of eligible patients in Shandong University Affiliated Reproductive Hospital and Shanghai Renji Hospital from 2015 to 2021 were collected, including age, anti-Müllerian hormone (AMH), antral follicle count (AFC), basal follicle-stimulating hormone (bFSH), Gn dosage, HCG dosage, number of eggs retrieved, transplantation status, pregnancy outcome, etc. Data processing: The collected data of anti-Müllerian hormone (AMH), antral follicle count (AFC), basal follicle-stimulating hormone (bFSH), and age were normalized and standardized to make them comparable. Calculate the corrected OR value to evaluate the discrimination of commonly used clinical reproductive aging indicators on the live birth rate of fresh embryo transfer and cumulative live birth rate. Use the generalized additive model to analyze the curve correlation between commonly used clinical reproductive aging indicators and cumulative live birth rate, and try various methods for linearization processing and evaluation. Calculate the dispersion of ovarian reserve indicators in women of similar age to evaluate ovarian reserve heterogeneity. Model construction and validation: Select the generalized additive model, with the normalized and standardized age, anti-Müllerian hormone (AMH), antral follicle count (AFC), and basal follicle-stimulating hormone (bFSH) as independent variables and cumulative live birth as the dependent variable to construct a model for curve fitting the cumulative live birth rate. Obtain the fitted cumulative live birth of each patient according to the model, and get the corresponding reproductive age (rAGE) through mapping the data set. Use external unit data to externally validate the model and evaluate the cumulative live birth heterogeneity and discrimination of reproductive age.

Claims

1. A method for assessing female reproductive aging, characterized in that: The method comprises the following steps: Step 1: Collect sample data and obtain sample characteristics; Step 1.1: Set the sample data collection criteria, and patients who meet the criteria will be included in the follow-up to obtain the final outcome - cumulative live births from a single oocyte retrieval; Step 1.2, select sample features related to pregnancy outcomes in the sample data according to preset criteria; Step 2: Convert and interpolate the sample features described in step 1.

2. The specific conversion and interpolation process is as follows: Step 2.1, convert and encode sample features: standardize and normalize the acquired data such as age, anti-Mullerian hormone, antral follicle count, basal follicle stimulating hormone, etc. to make them consistent with the input of fitting and training models; Step 2.2: If some of the sample features have missing values, then use the multiple interpolation method to fill in the missing values ​​of different features. If there are no missing values, no filling is required. Finally, a complete sample with no missing values ​​is formed and proceed to step 3. Step 3: Use the random forest model to sort the importance of candidate predictor variables for the complete sample formed in step 2.2, and select the key variables to enter step 4 to build a prediction model; Step 4: Split the data of the complete sample into a training set and a test set at a sample ratio of 8:

2. Use the training set samples formed in steps 2, 3, and 4 to construct a female reproductive aging assessment model based on the generalized additive model: Use the processed data as the independent variable, and use the cumulative live birth rate within 2 years after a single egg retrieval as the dependent variable to construct the model. The model formula is: , where y is the cumulative live births, x is i The normalized and standardized age, anti-Mullerian hormone, antral follicle count, and basal follicle-stimulating hormone, f i is a smooth function, g is a connection function, ε i is the error term; Step 5: Based on the constructed model, the predicted cumulative live birth rate of each patient is calculated and mapped to the corresponding age to obtain the reproductive age; Step 6, using the female reproductive aging assessment model based on the generalized additive model constructed in steps 4 and 5 and the reproductive age calculation to predict the degree of reproductive aging of the test set; Step 7. Use the female reproductive aging assessment model based on the generalized additive model selected in step 6 to verify it on external samples: compare the cumulative live birth heterogeneity and discrimination between reproductive age and actual age. If the cumulative live birth heterogeneity of reproductive age is lower and the discrimination is higher, the verification model is valid.

2. A method for assessing female reproductive aging according to claim 1, characterized in that: The data processing step also includes using normalized and standardized age, anti-Mullerian hormone, antral follicle count, and basal follicle-stimulating hormone as independent variables, and whether or not fresh embryo transfer is a live birth / cumulative live birth as a dependent variable, and using the adjusted odds ratio to compare the effectiveness of different outcome indicators in evaluating aging.

3. A method for assessing female reproductive aging according to claim 1, characterized in that: Before the model building step, the correlation between age, anti-Mullerian hormone, antral follicle count, basal follicle-stimulating hormone and cumulative live birth rate was analyzed, and it was confirmed that it was a curvilinear correlation and could not be linearized by exponential function, power function, logarithmic function, inverse transformation, and Box-Cox method: the correlation before and after treatment was evaluated by generalized additive model, and the degrees of freedom were all greater than 1.5, and the significance values ​​were all less than 0.

05.

4. A method for assessing female reproductive aging according to claim 1, characterized in that: After the data acquisition step, the quartile / median ratio of anti-Mullerian hormone, antral follicle count, and basal follicle-stimulating hormone indexes of females within 1 year of actual age is calculated. If the ratio is greater than 25%, it indicates that the ovarian reserve heterogeneity is strong.

5. A method for assessing female reproductive aging according to claim 1, characterized in that: In the model validation step, by comparing the heterogeneity and discrimination of cumulative live births between reproductive age and actual age, if the interquartile range of cumulative live births within 1 year of reproductive age is reduced by half compared with actual age, and older age has a higher discrimination for cumulative live births, the model is judged to be robust and extrapolatable.

Images