Construction method of model for evaluating clinical necessity of inverted carrier line PGT-SR detection, system and application thereof

Abstract

The invention discloses a model construction method for evaluating clinical necessity of inverted carrier PGT-SR detection, and a system and application thereof. The invention aims at comprehensively utilizing clinical data and a machine learning technology to provide more accurate PGT-SR detection suggestions for chromosome inversion carriers. By deeply analyzing the influence of clinical factors on a PGT-SR detection result and constructing an efficient prediction model, the method can accurately identify inversion carrier couples really needing PGT-SR assisted pregnancy. This will effectively avoid high-cost treatments on those inversion carriers who do not require PGT-SR treatments, while reducing adverse effects on embryonic development potential. According to the method, the model is constructed to evaluate whether the inversion carrier needs to be subjected to PGT-SR detection or not, and the blank in the aspect of clinical decision support in the prior art is filled.

Description

Technical Field

[0001] The present invention belongs to the technical field of biomedical detection, and particularly relates to a method for constructing a model for evaluating the clinical necessity of PGT-SR detection for inversion carriers, and a system and application thereof. Background Art

[0002] Currently, the expert consensus internationally (ESHRE PGT-SR / PGT-A Working Group, E. Coonen, D. Rubio, et al. ESHRE PGT Consortium good practice recommendations for the detection of structural and numerical chromosomal aberrations, Human reproduction open, (2020) 1-20.) and domestically (Editorial Group of the Expert Consensus on Preimplantation Genetic Diagnosis / Screening. Expert consensus on preimplantation genetic diagnosis / screening technology [J]. Chinese Journal of Medical Genetics, 2018, 35(2): 151-155.) recommends that in couples seeking assisted reproductive treatment, when one or both partners carry chromosomal structural abnormalities, including reciprocal translocation, Robertsonian translocation, inversion, complex translocation, pathogenic microdeletion or microduplication, etc., PGT-SR treatment is recommended. Chromosomal inversion is a form of chromosomal structural abnormality, involving the breakage of a segment of DNA on a chromosome and its reconnection in the opposite direction. In the population with balanced inversion, the genetic material on the chromosome is neither increased nor decreased, only a certain segment of the chromosome has an inversion in position. Because the total amount of genetic information remains unchanged, carriers of balanced inversion usually do not show any abnormal phenotypes. However, although carriers themselves may not be affected, they may face some risks during the reproductive process. During the formation of germ cells, inverted chromosomes may lead to abnormal pairing and crossing over, resulting in chromosomally unbalanced gametes and affecting fertility. Theoretically, the probability of an inversion carrier producing unbalanced gametes is 50%. However, there are many factors affecting the probability of euploidy of gametes in inversion carriers, including the gender of the carrier, the type of inversion (pericentric inversion and paracentric inversion), the size of the segment involved in the inversion, and the complex dynamics during meiosis, etc. (Medical Genetics, 6th Edition, Original work by Lynn B. Jorde, John C. Carey, Michael J. Bamshad, Main translation by Lu Guangxiu). Currently, couples with inversion carriers in each center use PGT-SR in assisted reproductive technology to achieve pregnancy.Research reports that during PGT-SR treatment cycles, the probability of inverted patients obtaining chromosomally balanced embryos is very high. The probabilities of patients with paracentric inversion and pericentric inversion obtaining chromosomally balanced blastocysts are as high as 70.4% and 57.4% respectively, and the proportions of abnormal embryos with chromosomally unbalanced rearrangements caused by genetic effects are only 3.6% and 17.2% respectively (P. Xie, L. Hu, Y. Tan, et al. Retrospecive analysis of meiotic segregation pattern and interchromosomal effects in blastocysts from inversion preimplantation genetic testing cycles, Fertil Steril, 112 (2019) 336-42e333.). A study divided 188 cycles of 165 couples of inversion carriers into two groups. One group was 136 cycles of 125 couples that underwent PGT, and the other group was 52 cycles of 50 couples that did not undergo PGT-SR testing. There were no differences in biochemical pregnancy, clinical pregnancy, ongoing pregnancy, miscarriage, and live birth rate in the first transfer cycle between the two groups, nor in the cumulative live birth rate per cycle. Among the 136 cycles that underwent PGT testing, there was no difference in the proportion of euploid embryos obtained by pericentric inversion and paracentric inversion carriers (60.71% vs 50.54%). Similarly, the proportion of euploid embryos in male carriers was not significantly higher than that in female carriers (61.2% vs 56.1%). Therefore, this study proposed that in the Han Chinese population, the PGT-SR technique has no obvious benefit for inversion carriers (Y. Shao, J. Li, J. Lu, et al. Clinical outcomes of Preimplantation genetic testing (PGT) application in couples with chromosomal inversion, a study in the Chinese Han population, Reprod Biol Endocrinol, 18 (2020) 79.).There are also studies showing that the incidence of aneuploidy originating from the inverted chromosome in the embryos of inversion carrier couples is only 32.9%, much lower than the theoretical value of 50% (J. Tong, J. Jiang, Y. Niu, et al. Do chromosomal inversion carriers really need preimplantation genetic testing?, J Assist Reprod Genet, 39 (2022) 2573-9.). In addition, the aneuploidy rate of inversion carrier embryos is not significantly higher than that of the age-matched control group (i.e., conventional in vitro fertilization cycle) (D. Young, D. Klepacka, M. McGarvey, W. B. Schoolcraft, M. G. Katz-Jaffe. Infertility patients with chromosome inversions are not susceptible to an inter-chromosomal effect. J Assist Reprod Genet, 36 (2019) 509-16.). In summary, the latest research results suggest that some inversion patients may not really benefit from PGT-SR treatment. The cost of PGT-SR treatment may be twice or even higher than that of traditional assisted reproductive technologies (such as in vitro fertilization-embryo transfer, IVF-ET). This treatment requires biopsying about 5-10 cells from the blastocyst for genetic analysis. Although this sampling process is very delicate, it may still cause physical damage to the embryo, thereby affecting its growth and development. In addition, the impact of this technology on children's health still needs further research and monitoring. Although the proportion of abnormal embryos produced by chromosomal inversion balanced carriers undergoing PGT-SR treatment is not as high as the theoretical value, the abnormal pairing and crossover of inverted chromosomes during germ cell formation produce chromosomally abnormal gametes, which can lead to problems such as infertility and recurrent miscarriage in carriers, causing great harm to the physical and mental health of patients. Therefore, for inversion carrier couples seeking assisted reproductive treatment, tools that can accurately evaluate the necessity of PGT-SR treatment will have important clinical value.The number of inversion carriers who undergo PGT-SR-assisted pregnancy both at home and abroad is small. For example, the number of inversion carrier couples undergoing PGT-SR testing investigated by the Colorado Center for Reproductive Medicine was only 52 cases (D. Young, D. Klepacka, M. McGarvey, W. B. Schoolcraft, M. G. Katz-Jaffe. Infertility patients with chromosome inversions are not susceptible to an inter-chromosomal effect. J Assist Reprod Genet, 36 (2019) 509-16.), and the number of inversion carriers enrolled in the study of Shanghai Renji Hospital for PGT-SR-assisted pregnancy over the years was 57 cases (J. Tong, J. Jiang, Y. Niu, et al. Do chromosomal inversion carriers really need preimplantation genetic testing?, J Assist Reprod Genet, 39 (2022) 2573-9.). Our unit has collected 108 cycles from 91 patient couples in the past ten years, which is a relatively large sample study for this population. Since the number of inversion carriers seeking assisted reproductive technology (ART) is small, the important clinical question of whether this relatively rare population can benefit from PGT-SR testing has often been overlooked. Currently, as long as inversion carriers seek ART, reproductive geneticists only directly arrange PGT-SR testing based on the opinions of expert consensus.

[0003] Do inversion carriers with chromosomal structural abnormalities really need to undergo PGT-assisted pregnancy? PGT involves invasive biopsy of embryos. Is it overtreatment for inversion carrier couples? Which characteristics of inversion carrier couples can truly benefit from PGT technology? Currently, there is no established tool to evaluate whether inversion carrier couples need to undergo PGT-SR testing when seeking ART.

[0004] Inversion is a chromosomal structural abnormality. Although inversion carriers usually do not exhibit any abnormal phenotypes, during germ cell formation, abnormal gametes caused by abnormal pairing and crossing-over of inverted chromosomes may lead to problems such as infertility and recurrent miscarriages in carrier couples, causing great harm to the physical and mental health of patients. Expert consensus both internationally and domestically recommends that couples with chromosomal structural abnormalities (including inversion carriers) undergo further PGT-SR testing when seeking assisted reproductive technology. However, the latest research results show that the effects of PGT-SR treatment and non-PGT-SR treatment in inverted couples seeking assisted reproductive treatment are similar, and the proportion of embryos with chromosomal abnormalities in the embryos of inversion carrier couples is lower than the theoretical value, indicating that some inverted patients seeking assisted reproductive treatment may not need PGT-SR testing. The PGT-SR technology is not only expensive, but also the biopsy operation involved is an invasive surgery for embryos, and its long-term impact on children's health remains to be further studied and monitored. Therefore, currently, there is a lack of a practical tool to accurately evaluate whether inversion carrier couples seeking assisted reproductive treatment need PGT-SR testing. Summary of the Invention

[0005] Object of the Invention: The technical problem to be solved by the present invention is to provide a method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers.

[0006] Another technical problem to be solved by the present invention is to provide a system obtained by the above construction method.

[0007] Finally, the technical problem to be solved by the present invention is to provide the application of the above construction method or the above system in the preparation of genetic counseling tools for inversion carriers or the optimization of the allocation of reproductive medicine resources.

[0008] Technical Solution: To solve the above technical problems, the present invention provides a method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers, including the following steps:

[0009] (1) Classify the test results of inversion carriers who have received PGT-SR treatment, and screen the data related to clinical factors from the data of the population who need PGT-SR testing;

[0010] (2) Extract the characteristic parameters related to the necessity of PGT-SR testing from the data related to the above clinical factors;

[0011] (3) Based on the above characteristic parameters, construct a prediction model for evaluating the clinical necessity of PGT-SR testing for inversion carriers.

[0012] Among them, the basis for classification in step (1) is that inversion carriers with aneuploid embryo proportion ≥ 50% are the population that needs to undergo PGT-SR testing, and inversion carriers with aneuploid embryo proportion < 50% are the population recommended not to undergo PGT-SR testing.

[0013] Among them, the clinical factor-related data described in step (1) include: one or more of the female age, female body mass index, male age, male semen density, normal sperm morphology rate, forward sperm proportion, non-forward sperm proportion, immotile sperm proportion, couple infertility type, infertility duration, history of adverse pregnancy and childbirth, carrier inversion type, carrier gender, proportion of the inversion segment in the chromosome where it is located, number of antral follicles in the left ovary of the female, number of antral follicles in the right ovary of the female, basal FSH, basal LH, basal E2, basal T, female ovulation induction medication regimen, starting dose of Gn, total amount of Gn, total number of days of Gn, E2 on the HCG day, LH on the HCG day, P on the HCG day, number of follicles on the HCG day, total number of retrieved eggs, number of mature eggs, maturity, number of fertilized eggs, number of biopsied blastocysts, and biopsied blastocyst rate.

[0014] Among them, the method for extracting the characteristic parameters described in step (2) includes: using L1 regularization regression (Lasso regression) to screen key features. Specifically, Lasso regression compresses the model coefficients by introducing an L1 regularization term into the objective function, thereby achieving the dual goals of variable selection and model complexity optimization. The specific algorithm formula is as follows:

[0015]

[0016] Among them, N is the total number of samples, β is the coefficient vector, xi is the clinical factor of the i-th sample, yi is the response variable of the i-th sample, ‖β‖1 is the L1 norm of the coefficient vector, and λ is the regularization parameter used to control the penalty intensity on the coefficients. Among them, i is a natural number from 1, 2, 3, 4... N.

[0017] Among them, the characteristic parameters related to the necessity of PGT-SR testing described in step (2) include history of adverse pregnancy and childbirth, proportion of the inversion segment in the chromosome where it is located, number of follicles on the HCG day, total amount of Gn, and total number of days of Gn.

[0018] Among them, the construction of the prediction model in step (3) also includes the training of the model. The influence of different regularization parameters λ on the model performance is evaluated through cross-validation. Specifically, 10-fold cross-validation is used to adjust the hyperparameters, and the model performance is evaluated through the area under the ROC curve.

[0019] Among them, the formula of the prediction model in step (3) is as follows: Comprehensive score = a × presence or absence of adverse pregnancy and childbirth history + b × proportion of the inverted segment in the chromosome where it is located + c × number of follicles on the HCG day + d × total amount of Gn + e × total number of Gn days + intercept. When the comprehensive score > the threshold, it is determined as category 1, that is, PGT-SR detection is performed; when the comprehensive score ≤ the threshold, it is determined as category 0, that is, PGT-SR detection is not performed.

[0020] Among them, a is 3.224836e-01, b is 7.207817e-05, c is -4.346183e-02, d is 5.953810e-05, e is 2.909721e-02, the intercept is 1.723857e-01, and the threshold is 0.024.

[0021] Among them, in step (3), when the comprehensive score of the prediction model > 0.024, it is determined as category 1, that is, PGT-SR detection is performed; when the comprehensive score ≤ 0.024, it is determined as category 0, that is, PGT-SR detection is not performed

[0022] The content of the present invention further includes a system for evaluating the clinical necessity of PGT-SR detection for inversion carriers, including:

[0023] A data acquisition module for obtaining data related to the clinical factors of inversion carriers;

[0024] A feature extraction module for screening feature parameters related to the necessity of PGT-SR detection;

[0025] A model construction module for generating a prediction model based on the feature parameters;

[0026] A result output module for displaying the evaluation result of clinical necessity.

[0027] The content of the present invention further includes the application of the described construction method or the described system in the preparation of genetic counseling tools for inversion carriers or the optimization of the allocation of reproductive medicine resources.

[0028] Based on the test results of inversion carrier couples undergoing PGT-SR treatment, they were divided into two groups according to the proportion of chromosomally abnormal embryos, with ≥50% and <50% respectively. A retrospective analysis was conducted on 34 clinical information of the patient couples. Then, a Lasso regression model (Least Absolute Shrinkage and Selection Operator regression model) was used for data modeling. The Lasso algorithm used in model training compresses the coefficients by introducing the L1 regularization term, while achieving variable selection and optimizing model complexity, thereby improving the prediction accuracy and interpretability of the model. During the modeling process, a 10-fold cross-validation method was used to optimize the model parameters: the influence of different regularization parameters λ on the model performance was evaluated through cross-validation, and the λ value (0.08513205) corresponding to the minimum average error during the cross-validation process was selected as the final hyperparameter of the model, and a total of 5 effective clinical factors were obtained. A machine learning prediction model was constructed using these selected clinical factors, and the clinical application value of the model in evaluating the necessity for patients to further undergo PGT-SR testing was verified in clinical patients. Finally, a prediction tool that can accurately evaluate the clinical necessity of inversion carriers undergoing PGT-SR treatment was obtained.

[0029] Advantages: Compared with the prior art, the present invention has the following advantages: The present invention aims to comprehensively utilize clinical data and machine learning technology to provide more accurate PGT-SR testing suggestions for inversion carrier couples. By deeply analyzing the influence of clinical factors on PGT-SR test results and constructing an efficient prediction model, the present invention can accurately identify inversion carrier couples who truly need PGT-SR assisted reproduction. This will effectively avoid high-cost treatment for those inversion carrier couples who do not require PGT-SR treatment, while reducing the adverse effects on embryonic developmental potential. The present invention constructs a model to evaluate whether inversion carriers need to undergo PGT-SR testing, filling the gap in clinical decision-making support in the prior art. Description of the Drawings

[0030] Figure 1 It is a flowchart for model construction and performance verification; Event 0: The proportion of abnormal embryos <50%, suggesting the population not recommended for PGT-SR testing (non PGT-SR); Event 1: The proportion of abnormal embryos ≥50%, suggesting the population recommended for PGT-SR testing (PGT-SR);

[0031] Figure 2 Establish a Lasso regression model; (A) Lasso variable screening, the abscissa is the logarithm of λ, and the ordinate is the variable coefficient; (B) Determine the optimal λ value of the model; (C) Determine the optimal threshold of the model;

[0032] Figure 3ROC curve for the training set; (A) ROC curve incorporating 5 clinical factors; (B) ROC curve for individual clinical factors; APH (adverse pregnancy history).

[0033] Figure 4 ROC curve for the validation set; (A) ROC curve incorporating 5 clinical factors; (B) ROC curve for individual clinical factors; APH (adverse pregnancy history). Detailed implementation manner

[0034] Explanation of technical terms in the present invention:

[0035] PGT-SR (Preimplantation Genetic Testing for Structural Rearrangements): Preimplantation genetic testing for structural rearrangements

[0036] FSH (Follicle Stimulating Hormone): Follicle stimulating hormone

[0037] LH (Luteinizing Hormone): Luteinizing hormone

[0038] E2 (Estradiol): Estradiol

[0039] T (Testosterone): Testosterone

[0040] P (Progesterone): Progesterone

[0041] Gn (Gonadotropin): Gonadotropin

[0042] HCG (Human Chorionic Gonadotropin): Human chorionic gonadotropin

[0043] Basal FSH, basal LH, basal E2, basal T: Basal hormone levels, which are measured by taking peripheral blood on the 2nd - 4th day of the menstrual cycle. The determination of basal hormone levels is a standard method for evaluating ovarian function and endocrine status.

[0044] E2 on the HCG day, LH on the HCG day, P on the HCG day: Hormone levels in peripheral blood measured on the day of injecting HCG to trigger ovulation, which is a key link in assisted reproductive therapy and can provide important clinical information for doctors.

[0045] Definition of adverse pregnancy history: The couple has a history of more than 2 biochemical pregnancies, natural miscarriages, fetal arrests, teratomas, or giving birth to children with chromosomal abnormalities.

[0046] Biochemical pregnancy two or more times: That is, early pregnancy is confirmed by blood HCG test, but does not develop into clinical pregnancy.

[0047] Spontaneous abortion: Pregnancy ends naturally before 20 weeks.

[0048] Fetal arrest: The embryo or fetus stops developing in the uterus.

[0049] Teratoid fetus: The fetus has structural abnormalities.

[0050] Giving birth to a child with chromosomal abnormalities: It is clear that the fetus or child has chromosomal structural or numerical abnormalities.

[0051] Ratio (%) of the inverted segment to the chromosome where it is located = Size of the inverted segment (Mb) / Size of the chromosome where the inverted segment is located (Mb) * 100

[0052] Number of follicles on the HCG day: On the day of injecting human chorionic gonadotropin (HCG), the number of follicles with a diameter ≥ 14 mm in the ovary observed by ultrasound examination.

[0053] Total amount of Gn: In the entire ovarian stimulation cycle, the total dose of gonadotropin (Gn) used by the patient, expressed in "IU" (International Unit).

[0054] Total number of days of Gn: The total number of days from the start of using gonadotropin (Gn) to injecting HCG.

[0055] Since whether PGR-SR testing is required is based on the overall detection and evaluation model of inversion carrier couples, therefore, the inversion carriers described in the present invention all refer to inversion carrier couples.

[0056] Construction of the model in Example 1

[0057] 1. Data collection, grouping, and screening: First, collect the clinical data of inversion carrier couples who received PGT-SR treatment in the Reproductive Medicine Department of Nanjing Drum Tower Hospital from January 2015 to November 2024. A total of 108 clinical data information of 91 couples' assisted reproductive ovulation induction cycles were collected. Inversions in the heterochromatic region of chromosome 9, inv(9)(p12q13) and inv(9)(p11q13), have been excluded because these inversions are generally considered polymorphic phenomena in the population. The collected clinical information covers a wide range of physiological and treatment parameters, providing a rich data basis for subsequent analysis. Second, define inversion carrier couples as the experimental group and the control group based on the PGT-SR test results, and group them according to the proportion of aneuploid embryos. The proportion of aneuploid embryos (%) = (number of PGT-SR tests - number of PGT-SR-diagnosed transferable embryos) / number of PGD-SR tests * 100. Inversion carrier couples with a proportion of aneuploid embryos ≥ 50% are divided into the experimental group (i.e., the population that needs PGT-SR testing), and inversion carrier couples with a proportion of aneuploid embryos < 50% are divided into the control group (i.e., the population recommended not to undergo PGT-SR testing). Then, screen the effect factors affecting patient outcomes from 34 clinical factors: female age, female body mass index, male age, male semen density, normal sperm morphology rate, forward sperm ratio, non-forward sperm ratio, immotile sperm ratio, couple infertility type, infertility duration, history of adverse pregnancy and childbirth, carrier inversion type, carrier gender, proportion of the inverted segment in the chromosome, number of antral follicles in the left ovary of the female, number of antral follicles in the right ovary of the female, basal FSH, basal LH, basal E2, basal T, female ovulation induction medication regimen, Gn starting dose, total Gn dose, total Gn days, E2 on HCG day, LH on HCG day, P on HCG day, number of follicles on HCG day, total number of retrieved oocytes, number of mature oocytes, maturity, number of fertilized eggs, number of biopsied blastocysts, and biopsy blastocyst rate.

[0058] 2. Data division: This invention analyzed 108 assisted reproductive ovulation induction cycles of 91 couples with chromosomal inversions in the hospital from 2015 to 2024. The patients were divided into two groups according to the embryo aneuploidy rate: the group with a proportion of aneuploid embryos < 50% (Event 0) had a total of 53 cycles, and the group with a proportion of aneuploid embryos ≥ 50% (Event 1) had a total of 55 cycles. Detailed clinical data and treatment parameters of the patient couples were collected, and the data were divided into a training set and a validation set according to an 8:2 ratio using the random stratified sampling method. Among them, the training set contained 86 cycles (41 in Event 0 group and 45 in Event 1 group), and the validation set contained 22 cycles (12 in Event 0 group and 10 in Event 1 group)( Figure 1 ). This grouping method ensures that the training set and the validation set are representative in terms of event distribution, laying a reliable data foundation for subsequent establishment of a prediction model.

[0059] 3. Model Training: In the training set, we used the Lasso regression model (Least Absolute Shrinkage and Selection Operator regression model) for data modeling. The Lasso algorithm used in model training compresses the coefficients by introducing the L1 regularization term, while achieving variable selection and optimizing the model complexity, thereby improving the prediction accuracy and interpretability of the model. The calculation formula of the algorithm is as follows:

[0060]

[0061] where N is the total number of samples, β is the coefficient vector, xi is the clinical factor of the i-th sample, and yi is the response variable of the i-th sample, that is, the classification of the proportion of aneuploid embryos (Event 0 or Event 1). is the L1 norm, which is used to achieve sparsity (that is, making some coefficients zero). λ is the regularization parameter, which controls the penalty intensity for the coefficients. Among them, i is a natural number from 1, 2, 3, 4... N.

[0062] The Lasso variable screening process is as shown in Figure 2 Figure A. The abscissa is the logarithm of the regularization parameter λ (log(λ)), and the ordinate is the standardized coefficient of each clinical variable. As the value of λ increases, the Lasso regression gradually compresses the coefficients of the variables through L1 regularization, and the coefficients of some variables are compressed to 0, thereby achieving variable selection. Each colored curve in the figure represents the coefficient change trajectory of a clinical variable. Finally, 5 key variables (whether there is a history of adverse pregnancy and childbirth, the proportion of the inverted segment in the chromosome, the number of follicles on the HCG day, the total amount of Gn, and the total number of days of Gn) are screened. These variables still maintain non-zero coefficients when the value of λ is small, indicating that they have significant predictive value for the outcome.

[0063] In the process of constructing the model of the present invention, the 10-fold cross-validation method was used to optimize the model parameters: the influence of different regularization parameters λ on the model performance was evaluated through cross-validation, and the λ value (0.08513205) corresponding to the minimum average error during the cross-validation process was selected as the best regularization parameter of the model. The left dotted line in the figure indicates the position of the best λ (see Figure 2 Figure B). At this time, the model reaches the best balance between the fitting ability and the generalization performance, thereby improving the stability and prediction accuracy of the model.

[0064] In the implementation process of the present invention, the prediction model constructed based on the Lasso regression model calculates the comprehensive score of the sample through the following formula:

[0065] Score = 3.224836e-01 * Presence or absence of adverse pregnancy and childbirth history + 7.207817e-05 * Proportion of the inverted segment in the chromosome (%) - 4.346183e-02 * Number of follicles on the HCG day + 5.953810e-05 * Total amount of Gn + 2.909721e-02 * Total number of days of Gn + 1.723857e-01.

[0066] By assigning different weights to key factors, this model comprehensively considers the influence of relevant factors on the prediction result. Among them, the weight coefficients of each variable are optimized through the Lasso regression algorithm, as follows: the weight coefficient of the presence or absence of adverse pregnancy and childbirth history is 3.224836e-01; the weight coefficient of the proportion of the inverted segment in the chromosome (%) is 7.207817e-05; the weight coefficient of the number of follicles on the HCG day is -4.346183e-02; the weight coefficient of the total amount of Gn is 5.953810e-05; the weight coefficient of the total number of days of Gn is 2.909721e-02; the intercept of the model is 1.723857e-01.

[0067] 4. Threshold evaluation: To determine the optimal classification threshold of the model, the present invention uses the Youden index as the evaluation criterion. The Youden index determines the optimal classification threshold by maximizing the sum of sensitivity and specificity, thereby achieving the optimal balance of classification performance. In the training set, based on the predicted probability scores output by the model, the optimal threshold is determined to be 0.024 in combination with the Youden index. Classify and discriminate the samples according to this threshold, and the accuracy and reliability of the result prediction are the best ( Figure 2 C).

[0068] Example 2 Evaluate the efficacy of the model in determining whether PGT-SR testing is required in inversion patients in the training set

[0069] In the training set, the present invention uses the optimal threshold to discriminate the samples: when the comprehensive score (Score) of the sample > 0.024, it is determined as class 1; when the comprehensive score (Score) ≤ 0.024, it is determined as class 0. Through evaluation, as Figure 3 , the area under the curve (AUC) of this model in the training set is 0.695, which is better than the model performance using only the presence or absence of adverse pregnancy and childbirth history as the prediction index (AUC = 0.629), indicating that the multi-variable joint prediction model of the present invention has higher prediction accuracy and clinical practicability. The analysis results show that when the threshold is set to 0.024, the performance of the model is the best. At this time, the sensitivity is 75.6%, the specificity is 63.4%, and the accuracy is 69.8%, which is significantly better than the model performance using only the presence or absence of adverse pregnancy and childbirth history (sensitivity is 60.0%, specificity is 65.9%, and accuracy is 37.2%).

[0070] Example 3: Evaluate the performance of the model in the validation set to confirm whether PGT-SR testing is required for inversion carriers

[0071] To verify the robustness and generalization ability of the Lasso regression model constructed in the present invention, the same discrimination criteria as those in the training set were applied in the validation set. Specifically, the λ value and the optimal classification threshold determined in the training set were directly applied to the validation set to evaluate the classification performance of the model on the validation set. The verification results are as Figure 4 shown. The area under the curve (AUC) of the model in the validation set was 0.650, the sensitivity was 80.0%, the specificity was 50.0%, and the accuracy was 63.6%, which was also significantly better than the performance of the model using only a history of adverse pregnancy and childbirth as a prediction index (AUC = 0.617, sensitivity = 40.0%, specificity = 83.3%, accuracy = 63.6%). This verification step not only confirmed the consistency of the model's performance in the training set but also further tested the reliability and stability of the model in practical applications. Through the double verification of the training set and the validation set, the present invention ensured the effectiveness of the Lasso regression model in variable selection, prediction accuracy, and generalization ability.

Claims

1. A method for constructing a model to evaluate the clinical necessity of PGT-SR testing for inversion carriers, characterized in that, It includes the following steps: (1) Classify the test results of inversion carriers who receive PGT-SR treatment, and screen the clinical factor-related data from the data of the population who need PGT-SR test; (2) Extract the characteristic parameters related to the necessity of PGT-SR test from the clinical factor-related data; (3) Construct a prediction model for evaluating the clinical necessity of PGT-SR test for inversion carriers based on the characteristic parameters.

2. The method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers according to claim 1, wherein In step (1), the basis for classification is that inversion carriers with aneuploid embryo ratio ≥ 50% are the population who need PGT-SR test, and inversion carriers with aneuploid embryo ratio < 50% are recommended not to have PGT-SR test.

3. The method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers according to claim 1, wherein, The clinical factor-related data in step (1) includes one or more of the following: female age, female body mass index, male age, male semen density, normal sperm morphology rate, forward sperm ratio, non-forward sperm ratio, immotile sperm ratio, type of couple infertility, infertility duration, history of adverse pregnancy and childbirth, carrier inversion type, carrier gender, proportion of inversion segment in the chromosome where it is located, number of antral follicles in the left ovary of the female, number of antral follicles in the right ovary of the female, basal FSH, basal LH, basal E2, basal T, ovulation induction drug regimen of the female, starting dose of Gn, total amount of Gn, total days of Gn, E2 on the day of HCG, LH on the day of HCG, P on the day of HCG, number of follicles on the day of HCG, total number of retrieved oocytes, number of mature oocytes, maturity, number of fertilized eggs, number of biopsied blastocysts or biopsy blastocyst rate.

4. The method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers according to claim 1, wherein, The extraction method of the characteristic parameters in step (2) includes: using L1 regularization regression to screen key features. Preferably, the Lasso algorithm used in the extraction method of the characteristic parameters in step (2) introduces an L1 regularization term to compress the coefficients. The specific algorithm formula is as follows: Among them, N is the total number of samples, β is the coefficient vector, xi is the clinical factor of the i-th sample, yi is the response variable of the i-th sample, ‖β‖1 is the L1 norm of the coefficient vector, λ is the regularization parameter, which controls the penalty intensity for the coefficients. Among them, i is a natural number from 1, 2, 3, 4... N.

5. The method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers according to claim 1, wherein, The characteristic parameters related to the necessity of PGT-SR test in step (2) include history of adverse pregnancy and childbirth, proportion of inversion segment in the chromosome where it is located, number of follicles on the day of HCG, total amount of Gn, and total days of Gn.

6. The method for constructing a model for evaluating the clinical necessity of PGT-SR detection in inversion carriers according to claim 1, wherein The construction of the prediction model in step (3) includes training the prediction model, evaluating the influence of different regularization parameters λ on the model performance through cross-validation. Specifically, 10-fold cross-validation is used to adjust the hyperparameters, and the model performance is evaluated through the area under the ROC curve.

7. The method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers according to claim 1, wherein The formula of the prediction model in step (3) is as follows: Comprehensive score = a × history of adverse pregnancy and childbirth + b × proportion of inversion segment in the chromosome where it is located + c × number of follicles on the day of HCG + d × total amount of Gn + e × total days of Gn + intercept. When the comprehensive score > threshold, it is determined as category 1, that is, PGT-SR test is performed; when the comprehensive score ≤ threshold, it is determined as category 0, that is, PGT-SR test is not performed.

8. The method for constructing a model for evaluating the clinical necessity of PGT-SR testing for inversion carriers according to claim 7, wherein Where a is 3.224836e-01, b is 7.207817e-05, c is -4.346183e-02, d is 5.953810e-05, e is 2.909721e-02, the intercept is 1.723857e-01, and the threshold of the comprehensive score is 0.

024. When the comprehensive score of the prediction model > 0.024, it is determined as category 1, that is, PGT-SR detection is performed; when the comprehensive score ≤ 0.024, it is determined as category 0, that is, PGT-SR detection is not performed.

9. A system for evaluating the clinical necessity of performing PGT-SR testing on inversion carriers, characterized in that, Including: A data acquisition module for obtaining data related to the clinical factors of inversion carriers; A feature extraction module for screening feature parameters related to the necessity of PGT-SR detection; A model construction module for generating a prediction model based on the feature parameters; A result output module for displaying the evaluation results of clinical necessity.

10. Use of the construction method according to any one of claims 1 to 8 or the system according to claim 9 in the preparation of a genetic counseling tool for inversion carriers or the optimization of the allocation of reproductive medicine resources.

Images