System for predicting number of oocytes obtained during ovarian stimulation of object

Abstract

The invention relates to a system for predicting the number of oocytes obtained during ovarian stimulation of an object receiving a standard GnRH antagonist solution for ovulation induction. The system comprises a data acquisition module used for obtaining data of the anti-mullerian hormone (AMH) level, basal follicle stimulating hormone (FSH) level and antrum follicle count (AFC) of the object and a module used for predicting the number of the oocytes obtained during the ovarian stimulation and calculating the information obtained in the data acquisition module to calculate the number of theoocytes (NROs) obtained by the object.

Description

technical field

[0001] The present invention relates to a system and method for predicting the number of oocytes obtained during ovarian stimulation in subjects receiving standard GnRH antagonist regimens for ovulation induction. Background technique

[0002] The number of retrieved oocytes (NROs) is considered a strong surrogate prognostic marker of successful pregnancy in women undergoing controlled ovarian stimulation (COS) and IVF / ICSI cycles. Optimal NROs help to increase the live birth rate (Live-birth-rate, LBR).

[0003] Predicting NROs prior to controlled ovarian stimulation (COS) is the only way to conduct effective and safe treatment. A variety of markers have been used to assess ovarian response, including, for example, age, basal follicle-stimulating hormone (FSH), antral follicle count (AFC), and anti-Müllerian hormone (AMH), which have been widely used Applied to predict ovarian responsiveness. To date, no single marker of ovarian reserve can completely rep...

Examples

Embodiment 1

[0132] Example 1 selects the subjects initially used to construct the model

[0133] Based on the data obtained from patients who were treated at Peking University Third Hospital between January 2017 and December 2017, the model was initially constructed. For patients used for initial model building, basic and clinical characteristics of the patient were collected, including last name, medical record number, serial number, age, BMI index, duration of infertility, previous IVF / ICSI - Number of embryo transfer (IVF / ICSI-ET) attempts, serum basal E 2 levels, FSH levels and LH levels, serum AMH levels, left and right ovarian AFCs, first, second, third, fourth and fifth causes of infertility, conventional or mild ovarian stimulation cycle, type of ovarian stimulation / COS regimen , starting and total dose of recombinant rFSH, duration of rFSH treatment (days), name of rFSH, endometrial thickness on human chorionic gonadotropin (hCG) trigger day, date of oocyte retrieval, and NROs. ...

Embodiment 2

[0151] Embodiment 2 model construction factor selection

[0152] Regression Model Selection

[0153] For the data of the above 1523 patients, the distribution of the number of oocytes obtained was first determined. Since the number of oocytes obtained is count data, Poisson distribution or negative binomial distribution can usually be considered. Utilize JMP Pro v.14 software to carry out the goodness of fit test of Poisson distribution and negative binomial distribution to data in the present embodiment, the result shows as follows figure 1 As shown, the data deviate from the Poisson distribution (χ 2 =7026.46, P2 = 1660.35, P = 0.77). figure 1 Show the fitting situation when using Poisson distribution and negative binomial distribution respectively, from figure 1 The results show that the number of retrieved oocytes (NROs) can be better fitted using the negative binomial distribution. Since the obtained data on the number of oocytes obeys the negative binomial distribut...

Embodiment 3

[0163] Example 3 Build a prediction model

[0164] As mentioned above, negative binomial regression was selected to build a statistical model, and the selection of predictors was carried out using pruned forward method and holdback verification. Using the software JMP Pro v. It is divided into two parts, one part is used as a training set (1066 data, 70%), and the other part is used as a verification set (457 data, 30%).

[0165] First, build the model on the training set and verify the model performance on the validation set. The selection of the prediction model is mainly based on the negative log-likelihood value in the validation set. The lower the negative log-likelihood value in the validation set, the better the model is.

[0166] figure 2 shows the variable screening process of the pruned forward method, specifically, except for the eliminated E 2 In addition, the rest of the above-mentioned single factors were included in the multi-factor negative binomial regress...

technical field

[0001] The present invention relates to a system and method for predicting the number of oocytes obtained during ovarian stimulation in subjects receiving standard GnRH antagonist regimens for ovulation induction. Background technique

[0002] The number of retrieved oocytes (NROs) is considered a strong surrogate prognostic marker of successful pregnancy in women undergoing controlled ovarian stimulation (COS) and IVF / ICSI cycles. Optimal NROs help to increase the live birth rate (Live-birth-rate, LBR).

[0003] Predicting NROs prior to controlled ovarian stimulation (COS) is the only way to conduct effective and safe treatment. A variety of markers have been used to assess ovarian response, including, for example, age, basal follicle-stimulating hormone (FSH), antral follicle count (AFC), and anti-Müllerian hormone (AMH), which have been widely used Applied to predict ovarian responsiveness. To date, no single marker of ovarian reserve can completely rep...

Examples

Embodiment 1

[0132] Example 1 selects the subjects initially used to construct the model

[0133] Based on the data obtained from patients who were treated at Peking University Third Hospital between January 2017 and December 2017, the model was initially constructed. For patients used for initial model building, basic and clinical characteristics of the patient were collected, including last name, medical record number, serial number, age, BMI index, duration of infertility, previous IVF / ICSI - Number of embryo transfer (IVF / ICSI-ET) attempts, serum basal E 2 levels, FSH levels and LH levels, serum AMH levels, left and right ovarian AFCs, first, second, third, fourth and fifth causes of infertility, conventional or mild ovarian stimulation cycle, type of ovarian stimulation / COS regimen , starting and total dose of recombinant rFSH, duration of rFSH treatment (days), name of rFSH, endometrial thickness on human chorionic gonadotropin (hCG) trigger day, date of oocyte retrieval, and NROs. ...

Embodiment 2

[0151] Embodiment 2 model construction factor selection

[0152] Regression Model Selection

[0153] For the data of the above 1523 patients, the distribution of the number of oocytes obtained was first determined. Since the number of oocytes obtained is count data, Poisson distribution or negative binomial distribution can usually be considered. Utilize JMP Pro v.14 software to carry out the goodness of fit test of Poisson distribution and negative binomial distribution to data in the present embodiment, the result shows as follows figure 1 As shown, the data deviate from the Poisson distribution (χ 2 =7026.46, P2 = 1660.35, P = 0.77). figure 1 Show the fitting situation when using Poisson distribution and negative binomial distribution respectively, from figure 1 The results show that the number of retrieved oocytes (NROs) can be better fitted using the negative binomial distribution. Since the obtained data on the number of oocytes obeys the negative binomial distribut...

Embodiment 3

[0163] Example 3 Build a prediction model

[0164] As mentioned above, negative binomial regression was selected to build a statistical model, and the selection of predictors was carried out using pruned forward method and holdback verification. Using the software JMP Pro v. It is divided into two parts, one part is used as a training set (1066 data, 70%), and the other part is used as a verification set (457 data, 30%).

[0165] First, build the model on the training set and verify the model performance on the validation set. The selection of the prediction model is mainly based on the negative log-likelihood value in the validation set. The lower the negative log-likelihood value in the validation set, the better the model is.

[0166] figure 2 shows the variable screening process of the pruned forward method, specifically, except for the eliminated E 2 In addition, the rest of the above-mentioned single factors were included in the multi-factor negative binomial regress...

Images