Prediction model for risk assessment of fetal growth restriction and construction method and application thereof

Abstract

The application discloses a prediction model for risk assessment of fetal growth restriction and a construction method and application thereof. The application combines various genetic factors, general health indicators and other related clinical parameters to construct an early risk prediction model of FGR, and compared with the prior art, the early prediction accuracy of FGR can be significantly improved, the AUC of the prediction model is 0.82, the specificity is 0.82, the sensitivity is 0.80, the early intervention of doctors can be effectively helped, and the health outcome of newborns is improved.

Description

Technical Field

[0001] This invention relates to the field of disease risk assessment technology, and in particular to a predictive model for assessing the risk of fetal growth restriction, its construction method, and its application. Background Technology

[0002] Fetal growth restriction (FGR) is a common perinatal complication, primarily defined as a fetal weight estimated by ultrasound below the 10th percentile for the corresponding gestational age. The incidence of FGR ranges from 2.75% to 15.53%, and it is the second leading cause of perinatal mortality for both fetuses and newborns. Fetal growth restriction is not only closely associated with adverse pregnancy outcomes but may also lead to long-term adverse consequences, including but not limited to childhood neurodevelopmental delays and cardiovascular and endocrine disorders in adulthood. Currently, there is a lack of effective treatments for FGR, and it is usually only diagnosed after delivery, lacking effective early prediction or diagnostic methods. Therefore, early identification of high-risk FGR fetuses and targeted interventions will be crucial for improving fetal outcomes.

[0003] Some studies have shown that predictive models can be used to improve the early detection rate of fetal growth restriction (FGR). Zhong Haiyan et al. constructed a predictive model by incorporating basic information, serological, and ultrasound indicators from early pregnancy, with an AUC of 0.71. Furthermore, Zhao Tianjiao et al. collected ultrasound indicators from hundreds of pregnant women in early pregnancy and constructed an FGR prediction model including fetal crown-rump length, placental volume, placental blood flow index, vascularization index, and vascularization-blood flow index, with an AUC of 0.762. It can be observed that although these models improved the accuracy of FGR prediction to varying degrees, none of them actually incorporated genetic risk factors, thus neglecting the impact of genetic variations on FGR to some extent. They did not consider the influence of fetal birth weight genetic loci on FGR outcomes, resulting in less comprehensive results and relatively lower predictive accuracy.

[0004] Furthermore, the inclusion factors in existing predictive models involve complex sampling operations, such as collecting data from 24 cross-sections of fetal echocardiography. This significantly increases the difficulty of data collection and the complexity of clinical application, thus affecting the practicality and generalizability of the models. Moreover, existing predictive models lack a comprehensive predictive framework as their foundation, and the included factors often have certain limitations and biases. They fail to comprehensively consider genetic factors, general health indicators, and other relevant clinical parameters from multiple perspectives, thus failing to achieve a comprehensive risk assessment system. This limits the potential of predictive models in medical applications.

[0005] Therefore, it is necessary to construct a FGR risk assessment and prediction model that can significantly improve the ability and accuracy of predicting the occurrence of FGR by integrating multiple biomarkers and genetic factors, thereby enabling early intervention for FGR and improving neonatal health outcomes. The construction of such a model not only helps improve the accuracy of clinical diagnosis but also provides a scientific basis for developing personalized pregnancy management plans and reliable technical support for reducing FGR-related health risks. Summary of the Invention

[0006] This invention aims to at least solve one of the aforementioned technical problems existing in the prior art. Therefore, the purpose of this invention is to provide a predictive model for assessing the risk of fetal growth restriction (FGR), its construction method, and its application. This invention establishes a method for predicting FGR based on clinical risk factors and genetic factors. It comprehensively incorporates a series of candidate risk factors for FGR, including basic clinical factors (maternal BMI, height), biomarkers (white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, neutrophil percentage), and genetic factors (two variant loci associated with fetal birth weight, rs1547643 and rs2430921). The method combines machine learning, Bayesian generalized linear regression, and other methods to construct a predictive model, which shows good predictive ability for the occurrence of FGR. The model constructed based on the method of this invention not only has a very high predictive accuracy for FGR (AUC = 0.82), helping to improve the accuracy of clinical diagnosis, but also provides a scientific basis for developing personalized pregnancy management plans, thereby effectively reducing the health risks associated with FGR.

[0007] In a first aspect, the present invention provides a set of early diagnostic biomarkers, said early diagnostic biomarkers including rs1547643, rs2430921, or a combination thereof.

[0008] A second aspect of the invention provides a set of early diagnostic biomarkers, including one of rs1547643, rs2430921 or a combination thereof, and non-SNP biomarkers.

[0009] In this invention, the term "early diagnosis" refers to diagnosis made when no symptoms are found or present, or when initial symptoms have just appeared, based on conventional technical means or clinical experience. In this invention, "early diagnosis" also refers to diagnosis made during pregnancy when conventional technical means are not yet available for detecting FGR, or when the results of conventional technical means are inaccurate.

[0010] In this invention, the phrase "the results of conventional technical means are not accurate" means that the diagnostic accuracy rate using conventional technical means is equal to or lower than 60%, 50%, 40%, 30%, 20% or 10%; or that the false positive rate is greater than or equal to 30%, 40%, 50%, 60%, 70%, 80%, 90% or 99%.

[0011] In some embodiments of the present invention, the early diagnostic markers are applicable to pregnant women with a gestational age of 30-44 weeks.

[0012] In some embodiments of the present invention, the early diagnostic marker is a small-for-gestational-age early diagnostic marker.

[0013] In some embodiments of the present invention, the non-SNP markers include basic clinical information and / or laboratory test indicators of pregnant women.

[0014] In some embodiments of the present invention, the basic clinical information of the pregnant woman includes at least one of the following: age, pre-pregnancy weight, body mass index, systolic blood pressure, diastolic blood pressure, and height.

[0015] In some embodiments of the present invention, the basic clinical information of the pregnant woman includes at least two of the following: age, pre-pregnancy weight, body mass index, systolic blood pressure, diastolic blood pressure, and height.

[0016] In some embodiments of the present invention, the basic clinical information of the pregnant woman includes: body mass index and pregnant woman's height.

[0017] In some embodiments of the present invention, the laboratory test indicators include at least one of the following: complete blood count test data, liver function test data, kidney function test data, glucose tolerance test data, urine sediment test data, coagulation function test data, and thyroid function test data.

[0018] In some embodiments of the present invention, the laboratory test indicator information includes at least two, three, four, or all of the following: complete blood count test data, liver function test data, kidney function test data, glucose tolerance test data, urine sediment test data, coagulation function test data, and thyroid function test data.

[0019] In some embodiments of the present invention, the laboratory test indicators include: routine blood test data, liver function test data, and kidney function test data.

[0020] In some embodiments of the present invention, the laboratory test indicators are: routine blood test data, liver function test data, and kidney function test data.

[0021] Of course, it should be understood that those skilled in the art may also use other types of test data based on the required indicators, including but not limited to the above-mentioned blood routine test data, liver function test data, kidney function test data, glucose tolerance test data, urine sediment test data, coagulation function test data, and thyroid function test data.

[0022] In some embodiments of the present invention, the laboratory test indicators include at least one of the following: hematocrit, hemoglobin concentration, white blood cell count, mean platelet volume, albumin, alkaline phosphatase, globulin, prealbumin, total protein, blood urea nitrogen, creatinine, blood glucose, free T3, percentage of lymphocytes, total number of neutrophils, percentage of neutrophils, human chorionic gonadotropin, and vitamin D.

[0023] In some embodiments of the present invention, the laboratory test indicators include at least two, three, four, or all of the following: hematocrit, hemoglobin concentration, white blood cell count, mean platelet volume, albumin, alkaline phosphatase, globulin, prealbumin, total protein, blood urea nitrogen, creatinine, blood glucose, free T3, percentage of lymphocytes, total number of neutrophils, percentage of neutrophils, human chorionic gonadotropin, and vitamin D.

[0024] In some embodiments of the present invention, the laboratory test indicators include: white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, and neutrophil percentage.

[0025] In some embodiments of the present invention, the laboratory test indicators are: white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, and neutrophil percentage.

[0026] In this invention, the method and source of obtaining the laboratory test index information and the basic clinical information of pregnant women are not limited in any way, and the method and source of obtaining them do not affect the relevant effect of the above-mentioned laboratory test index information and basic clinical information of pregnant women as biomarkers.

[0027] In some embodiments of the present invention, the early diagnostic markers may further include: the pregnant woman's lifestyle, dietary habits, environmental factors, etc.

[0028] In some embodiments of the present invention, the screening methods for other biomarkers are the same as those for basic clinical information or laboratory test indicators of pregnant women.

[0029] A third aspect of the invention provides the use of reagents for detecting the early diagnostic biomarkers described above in the preparation of products for diagnosing and / or assessing and predicting fetal growth restriction.

[0030] In some embodiments of the present invention, the reagents for detecting the early diagnostic biomarkers described above can be obtained based on conventional techniques in the art, including but not limited to: commercially available test kits, synthetic primers, etc.

[0031] In this invention, the detection method for early diagnostic biomarkers is not limited. Depending on actual needs, it can be at the gene level, protein level, or biochemical level. The detection methods include, but are not limited to, sequencing, chemiluminescence, enzyme-linked immunosorbent assay (ELISA), radioimmunoassay, microparticle labeling, and polymerase chain reaction (PCR).

[0032] In some embodiments of the present invention, the diagnosis includes early diagnosis.

[0033] A fourth aspect of the present invention provides a fetal growth restriction diagnostic and / or risk assessment prediction product, the fetal growth restriction diagnostic and / or risk assessment prediction product comprising reagents for detecting the early diagnostic markers described above, and adjuvants.

[0034] In some embodiments of the invention, the excipients include pharmaceutically acceptable excipients.

[0035] In some embodiments of the present invention, the adjuvant includes at least one of the following: buffer solution, solvent or diluent, preservative, surfactant, washing solution, and sample pretreatment reagent.

[0036] In this invention, the term "buffer solution" refers to a reagent used to adjust the pH value of an experimental environment or to simulate a special environment (such as in vivo) to ensure the stability and accuracy of the experiment, including but not limited to phosphate buffer (PBS), Tris-HCl buffer, etc.

[0037] In this invention, the term "solvent" refers to a liquid used to dissolve the corresponding analyte or sample, including but not limited to: organic solvents such as ethanol, methanol, acetonitrile, ethyl acetate, dichloromethane, acetone, petroleum ether, benzene, toluene, diethyl ether, chloroform, and isopropanol; and inorganic solvents such as water.

[0038] In this invention, the term "diluent" refers to a reagent used to dilute the sample to be tested or the original reagent to reduce its concentration so that it can participate in a specific detection reaction, including but not limited to the buffer solution or solvent mentioned above.

[0039] In this invention, the term "preservative" is used to refer to reagents that prevent reagent deterioration, extend the shelf life of reagents, and ensure the stability of reagents during storage and transportation. These include, but are not limited to, benzoic acid, sodium benzoate, p-hydroxybenzoic acid esters (such as their methyl, ethyl, propyl, and butyl esters), sorbic acid and its salts, and o-phenylphenol.

[0040] In this invention, the term "surfactant" also includes dispersants, which are reagents used to increase the surface activity of a solution and help dissolve and disperse components in a sample, including but not limited to: anionic surfactants (such as sodium dodecyl sulfate (SDS), sodium hexadecyl sulfate), cationic surfactants (such as benzalkonium chloride, benzalkonium bromide), nonionic surfactants (such as Tween 20, Tween 40, Tween 80) and amphoteric surfactants (such as lecithin).

[0041] In this invention, the term "cleaning solution" refers to a reagent used to clean the reaction system, instrument tubing and components during the detection reaction, including but not limited to: water and buffer solution.

[0042] In this invention, the term "sample pretreatment reagent" refers to reagents used in the process of sample separation, extraction, purification, and enzyme digestion before formal testing, including but not limited to: extraction solution (extraction agent), purification solution (purification material), dehydration reagent, and digestive enzyme solution.

[0043] In some embodiments of the present invention, the reagents and adjuvants for detecting the early diagnostic markers described above in the fetal growth restriction diagnostic and / or risk assessment prediction products may be individually packaged or co-packaged.

[0044] In some embodiments of the present invention, the fetal growth restriction diagnosis and / or risk assessment prediction product includes a test kit or test kit.

[0045] A fourth aspect of the present invention provides a method for constructing a fetal growth restriction risk assessment and prediction model, comprising the following steps:

[0046] Collect basic clinical information and / or laboratory test indicators of pregnant women, calculate the statistical difference P value between the two and the presence of fetal growth restriction, and screen clinical variables and biomarker variables according to the P value;

[0047] Genome-wide association analysis was performed on the genetic information of pregnant women to target fetal growth restriction, and genetic factor variables were screened based on genome-wide significance thresholds;

[0048] A model for assessing and predicting the risk of fetal growth restriction was developed based on clinical variables, biomarker variables, and genetic factors.

[0049] In this invention, the term "statistical difference P-value" refers to the probability that the difference between samples is due to sampling error, which is obtained by statistical significance testing methods. In this invention, a P-value less than 0.05, 0.01, or 0.001 is considered to indicate a statistical difference.

[0050] In some embodiments of the present invention, a p-value less than 0.05 is considered to indicate a statistically significant difference.

[0051] In some embodiments of the present invention, the basic clinical information of the pregnant woman includes at least one of the following: age, pre-pregnancy weight, body mass index, systolic blood pressure, diastolic blood pressure, and height.

[0052] In some embodiments of the present invention, the basic clinical information of the pregnant woman includes: body mass index and pregnant woman's height.

[0053] In some embodiments of the present invention, the laboratory test indicators include at least one of the following: complete blood count test data, liver function test data, kidney function test data, glucose tolerance test data, urine sediment test data, coagulation function test data, and thyroid function test data.

[0054] In some embodiments of the present invention, the laboratory test indicators include: routine blood test data, liver function test data, and kidney function test data.

[0055] In some embodiments of the present invention, the laboratory test indicators are: routine blood test data, liver function test data, and kidney function test data.

[0056] Of course, it should be understood that those skilled in the art may also use other types of test data based on the required indicators, including but not limited to the above-mentioned blood routine test data, liver function test data, kidney function test data, glucose tolerance test data, urine sediment test data, coagulation function test data, and thyroid function test data.

[0057] In some embodiments of the present invention, the laboratory test indicators include at least one of the following: hematocrit, hemoglobin concentration, white blood cell count, mean platelet volume, albumin, alkaline phosphatase, globulin, prealbumin, total protein, blood urea nitrogen, creatinine, blood glucose, free T3, percentage of lymphocytes, total number of neutrophils, percentage of neutrophils, human chorionic gonadotropin, and vitamin D.

[0058] In some embodiments of the present invention, the laboratory test indicators include: white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, and neutrophil percentage.

[0059] In some embodiments of the present invention, the laboratory test indicators are: white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, and neutrophil percentage.

[0060] In this invention, the method and source of obtaining the laboratory test indicators and basic clinical information of pregnant women are not limited in any way, and the method and source of obtaining them do not affect the relevant effect of the above-mentioned laboratory test indicators and basic clinical information of pregnant women as biomarkers.

[0061] In some embodiments of the present invention, the construction method further includes: classifying the collected basic clinical information of pregnant women and calculating the statistical difference P-value using different statistical methods.

[0062] In some embodiments of the present invention, if the basic clinical information of the pregnant woman is a quantitative clinical variable, a t-test is used; if the basic clinical information of the pregnant woman is an analytical categorical clinical variable, a chi-square test is used.

[0063] In this invention, the term "quantitative clinical variable" refers to a clinical variable that can be defined using a specific quantity. The quantity in this article is not limited by a specific unit of measurement and may include: concentration, number of moles, mass, etc.

[0064] In this invention, the term "analytical categorical clinical variable" or "categorical clinical variable" refers to clinical variables that distinguish the categories of cases in the population based on qualitative principles. These variables mainly describe attributes, including but not limited to: gender, disease type, etc.

[0065] In some embodiments of the present invention, the t-test includes paired-samples t-test and independent-samples t-test (two-sample t-test).

[0066] In some embodiments of the present invention, the t-test is an independent samples t-test (two samples t-test).

[0067] In some embodiments of the present invention, the categorical clinical variables analyzed include: multi-category clinical variables and binary clinical variables.

[0068] In some embodiments of the present invention, the categorical clinical variable being analyzed is a binary clinical variable.

[0069] In some embodiments of the present invention, the construction method further includes: using regression analysis to calculate the statistical difference P-value of the laboratory test index information.

[0070] In some embodiments of the present invention, the regression analysis method includes logistic regression, multinomial regression, and linear regression.

[0071] In some embodiments of the present invention, the regression analysis method is logistic regression.

[0072] In some embodiments of the present invention, when there are continuous variables, the average value is taken.

[0073] In some embodiments of the present invention, for multiple binary variables, a positive result is defined as positive once a positive result is obtained.

[0074] In some embodiments of the present invention, after completing the first screening by regression analysis of laboratory test indicator information, a second screening is further performed using multiple linear regression analysis, and biomarker variables are selected by calculating statistical difference P-values.

[0075] In some embodiments of the present invention, the genetic information of the pregnant woman can be obtained through sequencing technology.

[0076] In some embodiments of the present invention, the genetic information of the pregnant woman is obtained through whole-genome sequencing based on NIPT samples.

[0077] In some embodiments of the present invention, the genetic information of the pregnant woman can also be obtained using high-depth sequencing technology, thereby eliminating the need for further data quality control and improving model accuracy.

[0078] In some embodiments of the present invention, before performing genome-wide association analysis, the genetic information of the pregnant woman is further subjected to data quality control, correction and genotype imputation.

[0079] In some embodiments of the present invention, the data quality control includes: aligning with the hg38 human genome reference genome, removing samples with sequencing depths below 0.05× and alignment rates below 90%, and removing PCR duplicates.

[0080] In some embodiments of the present invention, samtools software is used to remove PCR duplicates. Of course, those skilled in the art can also use other alternative software or algorithms to remove PCR duplicates, including but not limited to samtools.

[0081] In some embodiments of the present invention, the correction includes re-alignment and base quality correction of the indel region.

[0082] In some embodiments of the present invention, GATK software is used for re-alignment and base quality correction of the indel region. Of course, those skilled in the art can also use other alternative software or algorithms for re-alignment and base quality correction of the indel region, including but not limited to GATK.

[0083] In some embodiments of the present invention, the genotype imputation includes: using a Chinese population reference genome as a reference genome for genotype imputation.

[0084] In some embodiments of the present invention, STITCH imputation software is used for genotype imputation. Of course, those skilled in the art can also use other alternative software or algorithms for genotype imputation, including but not limited to STITCH.

[0085] In some embodiments of the present invention, the Chinese population reference genome includes the CHS and CHN genomes provided by the 1000 Genomes Project (1KGP).

[0086] In some embodiments of the present invention, after genotype imputation, mutation detection is used to obtain genetic variation sites.

[0087] In some embodiments of the present invention, Basevar software is used for mutation detection. Of course, those skilled in the art can also use other alternative software or algorithms for mutation detection, including but not limited to Basevar.

[0088] In some embodiments of the present invention, the genome-wide significance threshold is 5e-08.

[0089] In some embodiments of the present invention, the modeling methods include: K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree, Naive Bayes, K-Means, Random Forest, and Adaptive Boosting Algorithm.

[0090] In some embodiments of the present invention, the modeling method is the K-Nearest Neighbor (KNN) algorithm.

[0091] In some embodiments of the present invention, the construction method further includes: performing importance and correlation analysis on genetic factor variables.

[0092] In some embodiments of the present invention, the importance and correlation analysis includes conditional and joint correlation analysis.

[0093] In some embodiments of the present invention, GCTA software is used to perform the conditional and joint association analysis. Of course, those skilled in the art can also use other alternative software or algorithms to perform importance and association analysis, including but not limited to GCTA software.

[0094] In some embodiments of the present invention, the construction method further includes: constructing clinical variables, biomarker variables, and genetic factor variables into matrix data, calculating the coefficients of predictive variables based on an algorithm, and then performing fitting modeling using the modeling methods described above.

[0095] In some embodiments of the present invention, the matrix data is in the form shown in Table 4.

[0096] In some embodiments of the present invention, the genetic factor variables are represented in matrix data as single nucleotide polymorphism (SNP) genotype doses.

[0097] In some embodiments of the present invention, the SNP includes: rs1547643, rs2430921, or a combination thereof.

[0098] In some embodiments of the present invention, the SNPs are rs1547643 and rs2430921.

[0099] In some embodiments of the present invention, the genetic factor variables are represented in matrix data as genotype doses of rs1547643 and rs2430921 (represented as continuous values ​​between 0 and 2).

[0100] In some embodiments of the present invention, the formula for calculating the genotype dose is as follows:

[0101] Genotype dose = Pr(Het|Data) + 2 × Pr(Alt|Data)

[0102] Wherein, Pr(Het|data) represents the posterior probability that the sample has a heterozygous mutation at this locus; Pr(Alt|data) represents the posterior probability that the sample has a homozygous mutation at this locus.

[0103] In some embodiments of the present invention, the algorithms used to calculate the coefficients of the predictor variables include: Bayesian Ridge function, least squares method, random walk method, Levenberg-Marquardt algorithm, etc.

[0104] In some embodiments of the present invention, the specific steps of the method for constructing the fetal growth restriction risk assessment and prediction model include:

[0105] Collect basic clinical information and / or laboratory test indicators of pregnant women, calculate the statistical difference P value between the two and the presence of fetal growth restriction, and screen clinical variables and biomarker variables according to the P value;

[0106] Genome-wide association analysis was performed on the genetic information of pregnant women to target fetal growth restriction, and genetic factor variables were screened based on genome-wide significance thresholds;

[0107] Sample data was extracted based on the selection of clinical variables, biomarker variables, and genetic factors. The extracted sample data was combined with the prevalence of fetal growth restriction to construct a matrix. A linear regression model was used to calculate the predictor coefficients between a single variable and fetal growth restriction. The matrix and predictor coefficients were then input into the modeling algorithm to obtain a fetal growth restriction risk assessment and prediction model.

[0108] In some embodiments of the present invention, the linear regression model is the same as the algorithms used to calculate the coefficients of the predictor variables, including: Bayesian Ridge function, least squares method, random walk method, Levenberg-Marquardt algorithm, etc.

[0109] In some embodiments of the present invention, the modeling algorithm is the same as the modeling method described above, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree, Naive Bayes, K-means, Random Forest and Adaptive Boosting Algorithm.

[0110] A fifth aspect of the present invention provides the application of the fetal growth restriction risk assessment and prediction model constructed by the construction method described above in the preparation of fetal growth restriction risk assessment and prediction products.

[0111] In some embodiments of the present invention, the fetal growth restriction risk assessment and prediction product includes an analysis module.

[0112] In some embodiments of the present invention, the analysis module obtains prediction results by executing the fetal growth restriction risk assessment prediction model constructed by the construction method described above.

[0113] In some embodiments of the present invention, the analysis module outputs a predicted fetal birth length value, and determines whether there is a risk of fetal growth restriction based on the predicted fetal birth weight value.

[0114] In some embodiments of the present invention, the criteria for judging the risk of fetal growth restriction based on predicted fetal birth weight are as follows:

[0115] Compare predicted fetal birth weight with standard values ​​for fetal growth restriction.

[0116] If the predicted birth weight of the fetus is less than the standard value for fetal growth restriction, then the risk of fetal growth restriction is considered high.

[0117] If the predicted birth weight of the fetus is greater than or equal to the standard value for fetal growth restriction, the risk of fetal growth restriction is considered low.

[0118] In some embodiments of the present invention, the fetal growth restriction risk assessment and prediction product further includes at least one of a detection module, a storage module, and a transmission module.

[0119] In some embodiments of the present invention, the fetal growth restriction risk assessment and prediction product further includes at least one input terminal and / or at least one output terminal.

[0120] In some embodiments of the present invention, the input terminal includes, but is not limited to: keyboard, mouse, hard disk, magnetic disk, writable read-only optical disc (CD-ROM), photoelectric paper tape input device, card input device, optical character reader, magnetic tape input device, Chinese character input device, etc.

[0121] In some embodiments of the present invention, the output terminal includes, but is not limited to: mobile client (such as mobile phone, tablet computer, laptop computer), display, printer, light pen, hard disk, magnetic disk, writable read-only optical disc (CD-ROM), etc.

[0122] A sixth aspect of the present invention provides a method for predicting fetal weight, comprising the following steps:

[0123] The genotypes of the pregnant woman's rs1547643 and rs2430921 were converted into genotype doses, and then input together with the pregnant woman's basic clinical information and / or laboratory test indicators into the fetal growth restriction risk assessment and prediction model constructed by the construction method described above, and the predicted fetal weight was calculated.

[0124] In some embodiments of the present invention, the method for changing the genotype dosage is as described above.

[0125] In some embodiments of the present invention, the basic clinical information and / or laboratory test indicators of pregnant women are as described above.

[0126] In this invention, the sources of the pregnant woman's rs1547643 and rs2430921 genotypes, basic clinical information of the pregnant woman, and laboratory test indicators are not limited, and may include data and / or information sources obtained by any conventional means in the art.

[0127] In some embodiments of the present invention, the fetal growth restriction risk assessment and prediction product includes a test kit, a test platform (system), and a test device.

[0128] A seventh aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of calculating a fetal growth restriction risk assessment and prediction model constructed by the construction method described above or the method described above.

[0129] An eighth aspect of the present invention provides a system comprising an analysis module having a computer-readable storage medium as described in the preceding aspect.

[0130] In some embodiments of the present invention, the system further includes an input terminal and an output terminal.

[0131] In some embodiments of the present invention, the input and output terminals are as defined above.

[0132] In some embodiments of the present invention, the system further includes at least one of a detection module, a storage module, and a transmission module.

[0133] In some embodiments of the present invention, the detection module, storage module, and transmission module are described.

[0134] A ninth aspect of the present invention provides an apparatus in which the system described in the above aspects is installed, or which contains the computer-readable storage medium described in the above aspects.

[0135] The beneficial effects of this invention are:

[0136] 1. This invention is the first to discover that the combination of rs1547643 and rs2430921 can serve as an early diagnostic biomarker for FGR. By combining it with other non-SNP biomarkers, it can be used to construct an FGR risk prediction model, providing a new approach for the early diagnosis and analysis of FGR.

[0137] 2. This invention combines hospital electronic medical records, a large amount of genotype data generated during prenatal checkups, basic clinical factors, common biomarkers, and genome-wide association studies of fetal birth weight in a large sample size to screen multiple variables (markers) and establish an FGR prediction method. This method performs better and has higher prediction accuracy than traditional single-factor or two-factor prediction methods, which helps hospitals screen for FGR fetuses, thereby enabling early intervention and improving neonatal health outcomes.

[0138] 3. The data required for the model in this invention is simple to obtain and can be generated from routine physical examinations of pregnant women, thus not increasing the additional medical burden on pregnant women and making it economically feasible. Moreover, the model algorithm in this invention can be completely completed autonomously by a computer, resulting in accurate results and a short processing time.

[0139] 4. The model in this invention effectively solves the problem of poor clinical operability of existing models. By introducing more interpretable algorithms and model design, it enables doctors to understand the model's decision-making process, thereby developing personalized treatment measures. The model based on this design concept not only improves its clinical application value but also promotes its widespread adoption in clinical practice. Furthermore, the model incorporates genetic variation sites that affect fetal birth weight, addressing the problem of existing models neglecting the impact of genetic variations on FGR and providing a more accurate assessment.

[0140] 5. Compared with existing technologies, the early risk prediction model for neonatal gastrointestinal tract (FGR) in this invention can significantly improve the accuracy of early FGR prediction. The prediction model has an AUC of 0.82, a specificity of 0.82, and a sensitivity of 0.80, which can effectively help doctors intervene early and improve the health outcomes of newborns. Moreover, this invention is the first to combine genetic factors, general health indicators, and other relevant clinical parameters to form a comprehensive risk assessment system, thereby constructing a comprehensive prediction framework. This achieves model generalization and high-accuracy prediction capabilities, filling the gap in the accuracy of individual-level genotype data. Attached Figure Description

[0141] Figure 1 This is a linear regression forest plot showing the relationship between laboratory test indicators and fetal birth weight.

[0142] Figure 2 The results of a genome-wide association analysis of fetal birth weight.

[0143] Figure 3 This is a regional map of the PLCE1 gene.

[0144] Figure 4 This is a regional map of the SLC38A4 gene.

[0145] Figure 5 A plot showing the importance of variable features.

[0146] Figure 6 This is a flowchart illustrating the workflow of the FGR prediction model in this embodiment of the invention.

[0147] Figure 7 The ROC curve is a result of actual measurements using the FGR prediction model in this embodiment of the invention. Detailed Implementation

[0148] The present invention will be further described in detail below through specific embodiments. Unless otherwise specified, the raw materials, reagents, or apparatus used in the embodiments and comparative examples are all available from conventional commercial sources or can be obtained by existing technical methods. Unless otherwise specified, the test or experimental methods are conventional methods in the art.

[0149] Example 1

[0150] This embodiment provides a predictive model for assessing the risk of fetal growth restriction and its construction method. The specific construction process is as follows:

[0151] (1) Selection and inclusion criteria for participants:

[0152] In this embodiment, the subjects were 61,182 pregnant women who underwent routine prenatal checkups at Wuhan Children's Hospital between 2017 and 2020.

[0153] For the selected subjects, a variety of tests were conducted throughout the pregnancy, including basic clinical information, laboratory test indicators, non-invasive prenatal testing (NIPT), oral glucose tolerance test, and ultrasound screening, to collect data on clinical factors, biomarkers, and genetic factors. The specific data collected are shown in Table 1.

[0154] Among them, NIPT screens for fetal chromosomal abnormalities by analyzing cell-free DNA in the mother's blood sample. NIPT is usually performed between 12 and 22 weeks of gestation. While screening for fetal chromosomal abnormalities, gene sequencing data is also obtained based on NIPT analysis.

[0155] Table 1. Collected data and their classification

[0156]

[0157] Informed consent was provided to each selected participant before enrollment.

[0158] Furthermore, all relevant experiments in the embodiments of the present invention have been approved by the Institutional Review Board (IRB) of Wuhan Children's Hospital (2021R062-E03) and the Ethics and Biosafety Committee of BGI (BGI-IRB 21088-T2), and have been authorized by the National Human Genetic Resources Management Office (Approval No.:

[2021] CJ2002).

[0159] In this embodiment, the diagnostic criteria for FGR are as follows: according to the national health industry standard WS / T 800—2022 "Evaluation Standards for Growth of Newborns at Birth at Different Gestational Ages" issued by the National Health Commission and implemented on October 1, 2022, the presence of FGR is determined by comparing the actual measured value of the subject with the value of the tenth percentile of the corresponding gestational week. If the actual measured value is less than the national standard value, FGR is determined to exist.

[0160] (2) Selection of sample information:

[0161] A. Clinical variables:

[0162] For quantitative basic clinical variables, a two-sample t-test was used to determine whether the variable means were equal between the FGR cases and the non-FGR groups (including those with and without FGR).

[0163] For binary clinical variables, the chi-square test is used to examine whether fetal birth weight may be related to the binary variable.

[0164] The clinical variables that were significantly associated with fetal birth weight were selected based on a p-value of less than 0.05.

[0165] The pregnant woman's weight and height were also included as predictive variables.

[0166] In this embodiment, clinical variables were collected through a questionnaire for non-invasive prenatal testing, including the pregnant woman's age, pre-pregnancy weight, body mass index (BMI), systolic blood pressure, diastolic blood pressure, and height. A BMI > 25 kg / m² was considered a threshold. 2 As a threshold for obesity.

[0167] After applying the two-sample t-test and chi-square test as described above to the obtained clinical variables, the results of the candidate risk factor analysis for each basic clinical variable are shown in Table 2.

[0168] Table 2 Regression analysis of basic clinical information and fetal birth weight

[0169]

[0170] Based on the above analysis results, the pregnant woman's body mass index (BMI), pre-pregnancy weight, and height were considered basic clinical risk factors and were included in the predictive variables of the subsequent model construction.

[0171] B. Laboratory test indicators:

[0172] The study examined 80 subcategories of routine laboratory tests, including complete blood count, liver function, kidney function, glucose tolerance test, urine sediment analysis, coagulation function, and thyroid function. These subcategories covered metabolism, liver function, urinalysis, kidney function, complete blood count, protein, hormones, infection, electrolytes, and trace elements. For each individual laboratory test indicator, logistic regression analysis was performed with fetal birth weight as the response variable. Then, based on the p-value of each laboratory test indicator, they were sorted from smallest to largest, with a p-value less than 0.05 used as the screening criterion.

[0173] The laboratory test indicators in this embodiment are all obtained based on conventional testing methods in the field.

[0174] For continuous variables, the average value is used. For multiple binary variables, a positive result is defined as any positive result found.

[0175] The results are as follows Figure 1 As shown.

[0176] Based on the screening results, the 18 biomarkers most significantly correlated with fetal birth weight are: hematocrit, hemoglobin concentration, white blood cell count, mean platelet volume, albumin, alkaline phosphatase, globulin, prealbumin, total protein, blood urea nitrogen, creatinine, blood glucose, free T3, percentage of lymphocytes, total number of neutrophils, percentage of neutrophils, human chorionic gonadotropin, and vitamin D.

[0177] Multivariate linear regression analysis (multiple linear regression analysis) was performed on the above 18 biomarkers to further highlight the association between the relevant biomarkers and FGR. The indicators that remained significant were: white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, and neutrophil percentage. Therefore, white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, and neutrophil percentage were used as risk factors for laboratory test indicators and included in the predictive variables of the subsequent model construction.

[0178] C. Genetic variables:

[0179] The maternal genotype data obtained from high-throughput non-invasive prenatal genetic testing (NIPT) were preprocessed to obtain genetic factor variables. The specific steps were as follows: Ultra-low depth whole-genome sequencing was performed on samples collected in NIPT to obtain ultra-low depth sequencing fragments. These fragments were then aligned to the hg38 human genome reference genome using BWA software. Samples with sequencing depths below 0.05× and alignment rates below 90% were removed. PCR repeats were removed using samtools software, and indel region re-alignment and base quality correction were performed using GATK software. Genotype imputation was performed using STITCH software with CHS and CHN provided by the 1000 Genomes Project (1KGP) as reference genomes, and variant detection was performed using Basevar software. Finally, genome-wide association analysis (GWAS) was performed using PLINK software to identify sequence variations at the whole-genome level using biostatistical methods. A genome-wide significance threshold of 5e-08 was set to screen for loci significantly associated with fetal birth weight. Furthermore, considering the principle of linkage disequilibrium, conditional and co-association analyses are needed in GCTA software for single nucleotide polymorphisms (SNPs) that are significant across the entire genome in order to identify independent important signals.

[0180] The results of the genome-wide association analysis are as follows: Figure 2 As shown in Table 3, the results of conditional and joint association analysis are presented.

[0181] Table 3 Results of Conditional and Joint Association Analysis

[0182]

[0183] Genome-wide association analysis showed that four genes on chromosome 10 (PLCE1, NT5C2, CNNM2, and AS3MT) and the gene on chromosome 12 (SLC38A4) were associated with FGR.

[0184] Further analysis of the loci in the aforementioned genes, combined with conditional and joint association analysis results, revealed that the risk gene loci rs1547643 (chr10:94252108, PLCE1, G / T) and rs2430921 (chr12:46765413, SLC38A4, T / A) in PLCE1 and SLC38A4 were selected as predictive variables for genetic factors in model construction. Among these, PLCE1 and SLC38A4... Figure 3 and Figure 4 As shown.

[0185] (4) Feature importance analysis:

[0186] SHAP (SHapley Additive exPlanations) values ​​are a game theory-based method used to interpret the output of machine learning models, quantifying the contribution of each feature to the prediction result. In this embodiment, to better demonstrate the importance and correlation of the aforementioned related predictor variables, the inventors used SHAP values ​​to perform feature importance analysis on these related predictor variables.

[0187] The results are as follows Figure 5 As shown.

[0188] Based on the SHAP values, the importance of FGR features, ranked from highest to lowest, is as follows: total protein (TP), pregnant woman's BMI, pregnant woman's height (pHeight), globulin (GLB), percentage of neutrophils (NEU_pct), rs2430921 (chr12:46765413), prealbumin (PA), creatinine (CR), total number of neutrophils (NEU), white blood cell count (WBC), and rs1547643 (chr10:94252108).

[0189] (5) Model building:

[0190] Based on the data requirements (using the dropna function in Python to remove all samples with missing values) and information selection in the above steps, 5219 individuals were finally selected from 61182 samples for subsequent training modeling, validation correction and effect testing.

[0191] Specifically, the proportions used for subsequent training modeling (training set), validation correction (validation set), and performance testing (test set) are as follows: 20% of the 5219 individuals are randomly selected as the test set. Then, the 5219 individuals are resampled using the `resample` function in Python to obtain a balanced dataset. The `KFold` cross-validation function (10×, i.e., repeated 10 times, each time selecting a different subset as the validation set) is used to divide the balanced dataset into training and validation sets. The specific strategy of KFold cross-validation is as follows: the dataset is divided into 10 subsets using KFold cross-validation; 9 of these subsets (90% of the balanced dataset) are used as the training set, and the remaining 1 subset (10% of the balanced dataset) is used as the validation set.

[0192] Due to the nature of KFold cross-validation, there is no overlap between the training and validation sets. Furthermore, validation data is not included in the training set during each iteration, ensuring the fairness and accuracy of model evaluation.

[0193] Ultimately, the training set used to build the model used information from 3757 individuals, the information from 418 individuals was used to validate the built model (validation set), and the information from another 1044 individuals was used to test the performance of the built model (test set).

[0194] The predictive variable information obtained in step (2) above (total protein, pregnant woman's BMI, pregnant woman's height, globulin, percentage of neutrophils, rs2430921, prealbumin, creatinine, total number of neutrophils, white blood cell count, rs1547643) are used as columns and each sample is used as a row to construct a data information matrix.

[0195] In this embodiment, an example of the input matrix obtained based on the above content is shown in the table below.

[0196] Table 4 Example of Input Matrix

[0197]

[0198]

[0199] The first 12 columns of the matrix represent the subject's anonymity number, maternal BMI, maternal height, white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, neutrophil percentage, rs1547643, and rs2430921, respectively. Due to the insufficient coverage depth of the low-depth whole-genome data, genotyping using STITCH software is required. Therefore, the individual genotypes rs1547643 and rs2430921 in columns 11 and 12 are genotype doses (represented as continuous values ​​between 0 and 2). Genotype dose (Dosage) represents the expected non-reference sequence base dose at a given locus in a sample, and its value always falls between 0 and 2. The formula for calculating genotype dose is:

[0200] Dosage=Pr(Het|Data)+2×Pr(Alt|Data)

[0201] Wherein, Pr(Het|data) represents the posterior probability that a sample has a heterozygous mutation at that locus, and Pr(Alt|data) represents the posterior probability that a sample has a homozygous mutation at that locus. Alternatively, an individual's genotype can be encoded as a value of 0, 1, or 2, where 0 indicates homozygosity with the REF allele, 1 indicates heterozygosity with the REF allele, and 2 indicates homozygosity with the ALT allele.

[0202] The BayesianRidge function in Python is used to calculate the coefficients of predictors. The input parameter `formula` is:

[0203] FGR~pBMI+pHeight+WBC+GLB+PA+TP+CR+NEU+NEU_pct+rs1547643+rs2430921.

[0204] Among them, pBMI represents the pregnant woman's BMI, pHeight represents the pregnant woman's height, WBC represents the white blood cell count, GLB represents globulin, PA represents prealbumin, TP represents total protein, NEU represents the total number of neutrophils, and NEU_pct represents the percentage of neutrophils.

[0205] In the embodiments, the coefficients β of the linear regression fitting model with the best parameters are shown in Table 5, and the model is constructed using these coefficients β.

[0206] Table 5. Fit coefficients of the linear regression model

[0207] Predictors coefficient β BMI (pBMI) for pregnant women -0.01570 Height of pregnant women (pHeight) -0.01263 White blood cell count (WBC) -0.00077 Globulin (GLB) -0.00992 Prealbumin (PA) -0.00247 Total protein (TP) 0.01767 Creatinine (CR) 0.00128 Total neutrophil count (NEU) 0.00123 Neutrophil percentage (NEU_pct) -0.00775 rs1547643 -0.00054 rs2430921 -0.00298 intercept 0.06155

[0208] Then, the FGR predicted value (y value) for each individual is determined by linearly combining the above predictor Xi with the coefficient βi according to the following formula.

[0209] y = β0 + β1x1 + ... + β i x i .

[0210] The FGR prediction value is reflected in the fetal birth weight, which is a continuous value. The presence of FGR is further determined by comparing the predicted fetal birth weight with the standard value in WS / T 800—2022 "Standards for Growth Evaluation of Newborns at Birth at Different Gestational Ages".

[0211] The construction process in this embodiment is as follows: Figure 6 As shown.

[0212] Example 2

[0213] In this embodiment, the inventors first randomly selected a portion of data from the test set to validate the obtained model. The input matrix formed by the selected data is shown in the table below.

[0214]

[0215] Based on the input matrix shown in Table 6 and the model in the above embodiments, the FGR prediction results for the above data are shown in the following table.

[0216] Table 7 Model Prediction Results

[0217] Subject Anonymity Number Prediction results The real situation 10 normal normal 22 normal normal 23 normal normal 24 normal normal 30 normal normal 55 normal normal 78 normal normal 98 FGR FGR 100 FGR FGR 200 normal normal

[0218] The results showed that the model constructed in this embodiment had the expected accuracy in predicting FGR.

[0219] Further, following the method described above, the predictor variable information for each sample in the test set is input, and the FGR incidence value of the disease is output to predict the predicted value of the test set. This predicted value is then compared with the true value to calculate the AUC (Area Under the Curve). The AUC is calculated based on the BayesianRidge function in Python software, using the Bayesian generalized linear model to estimate the model coefficients.

[0220] The results are as follows Figure 7 As shown.

[0221] The results show that, following the above embodiments, the model obtained after 10-fold cross-validation and KNN (K-Nearest Neighbors) modeling (setting n_neighbors=5 and using the KNeighborsClassifie function in Python for KNN training) can be effectively used for FGR risk prediction, with high accuracy, AUC of 0.82, specificity of 0.82, and sensitivity of 0.80, and can be used to accurately predict FGR risk.

[0222] Example 3

[0223] In this embodiment, a fetal growth restriction risk assessment and prediction product is provided, which includes an input terminal, an analysis module, and an output terminal.

[0224] The input terminal is used for inputting data or information. In this embodiment, for example, the input terminal receives information such as total protein, pregnant woman's BMI, pregnant woman's height, globulin, neutrophil percentage, rs2430921, prealbumin, creatinine, total neutrophil count, white blood cell count, and rs1547643, and then transmits this information to the analysis module. The input terminal can optionally include, depending on the usage scenario, a keyboard, mouse, hard disk, magnetic disk, writable read-only optical disc (CD-ROM), photoelectric paper tape input device, card input device, optical character reader, magnetic tape input device, Chinese character input device, etc.

[0225] The analysis module contains a storage medium on which the fetal growth restriction risk assessment and prediction model described in the above embodiments is written, so that the fetal growth restriction risk assessment and prediction model can be executed in the analysis module. By receiving information from the input terminal, the fetal growth restriction risk assessment and prediction model is substituted into it for calculation to obtain the prediction result. The prediction result is exported from the analysis module to the output terminal. The analysis module can be optionally configured according to the needs of the application scenario, including: general-purpose computers, cloud computing networks, microcontrollers, embedded systems (such as smartphones, home appliances, etc.), dedicated processors (such as graphics processors and tensor processors), algorithm processors (such as neural computers and array processors), etc.

[0226] The output end is used for outputting data or information. Depending on the needs of the usage scenario, it may include: mobile clients (such as mobile phones, tablets, and laptops), monitors, printers, light pens, hard drives, magnetic disks, writable read-only optical discs (CD-ROMs), etc.

[0227] Depending on the user's needs, the product may further include:

[0228] Detection module: Used to receive samples for testing and output the test results as input information.

[0229] Storage module: Used to store information or results from each module.

[0230] Transmission module: Used to transmit information output by other modules, and can transmit one or more pieces of information to one or more receiving ends simultaneously.

[0231] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

[0232] References

[0233] 1. Abuhamad, A., Martins, JG and Biggio, JR, 2021. Diagnosis and management of fetal growth restriction: the SMFM guideline and comparison with the ISUOG guideline. Ultrasound in Obstetrics & Gynecology, 57(6), pp.880-883.

[0234] 2. Chen Ruixin, Qi Hongbo, Liu Xinghui, 2021. Interpretation of the 2021 American College of Obstetricians and Gynecologists Guidelines on Fetal Growth Restriction. Practical Journal of Obstetrics and Gynecology, 37(12):907-909.

[0235] 3. Yu, M., Liu, Y. and Wang, L., 2022. Analysis of Causes and Results of Fetal Growth in UteroCaused by Genetic Factors Detected by Ultrasound. Contrast Media&Molecular Imaging, 2022(1), p.3703132.

[0236] 4. Yu,

[0237] 5.Lees,C.C.,Stampalija,T.,Baschat,A.A.,da Silva Costa,F.,Ferrazzi,E.,Figueras,F.,Hecher,K.,Poon,L.C.,Salomon,L.J.and Unterscheider,J.,2020.ISUOGPractice Guidelines:diagnosis and management of small-for-gestational-agefetus and fetal growth restriction.Ultrasound in Obstetrics&Gynecology,56(2),pp.298-312.

[0238] 6.Chawla,D.,2021.Fetal growth restriction and neurodevelopmentaloutcome.Indian Journal ofPediatrics,88(6),pp.538-539.

[0239] 7.Yener,G.,Kavak,S.B.,Gul,Y.,Kavak,E.C.,Bulmus,F.G.,Sanli,C.,Batmaz,I.and Bulu,G.,2023.Elabela levels in pregnancies with intrauterine growthretardation.Ginekologia Polska,94(2),pp.113-118.

[0240] 8.MacDonald,T.M.,Hui,L.,Robinson,A.J.,Dane,K.M.,Middleton,A.L.,Tong,S.and Walker,S.P.,2019.Cerebral–placental–uterine ratio as novel predictor oflate fetal growth restriction:prospective cohort study.Ultrasound inObstetrics&Gynecology,54(3),pp.367-375.

[0241] 9.Pini,N.,Lucchini,M.,Esposito,G.,Tagliaferri,S.,Campanile,M.,Magenes,G.and Signorini,MG,2021.A machine learning approach to monitor theemergence of late intrauterine growthrestriction.Frontiers in ArtificialIntelligence,4,p.622616.

[0242] 10. Zhong Haiyan, Luo Wenbin, Wang Dongju, Xiao Xiaomin, 2018. Risk assessment and prediction of fetal growth restriction by combined early and mid-pregnancy screening. International Journal of Obstetrics and Gynecology, 45(6), 4.

[0243] 11. Zhao Tianjiao, Zhao Zhizhi. 2023. Prediction of fetal growth restriction by three-dimensional energy Doppler ultrasound combined with color Doppler in early pregnancy. Chinese Journal of Family Planning, 31(4), 964-967.

Claims

1. A set of early diagnostic biomarkers, characterized in that, The early diagnostic biomarkers include rs1547643, rs2430921, or combinations thereof; Preferably, the early diagnostic biomarkers further include: non-SNP biomarkers; Preferably, the early diagnostic marker is an early diagnostic marker for fetal growth restriction; Preferably, the non-SNP markers include basic clinical information and / or laboratory test indicators of the pregnant woman.

2. The early diagnostic biomarker according to claim 1, characterized in that, The basic clinical information of the pregnant woman includes at least one of the following: age, pre-pregnancy weight, body mass index, systolic blood pressure, diastolic blood pressure, and height; and / or The laboratory testing indicators include at least one of the following: complete blood count data, liver function test data, kidney function test data, glucose tolerance test data, urine sediment test data, coagulation function test data, and thyroid function test data; Preferably, the basic clinical information of the pregnant woman includes: body mass index and pregnant woman's height; Preferably, the laboratory testing indicators include: routine blood test data, liver function test data, and kidney function test data.

3. The early diagnostic biomarker according to claim 1 or 2, characterized in that, The laboratory test indicators include at least one of the following: hematocrit, hemoglobin concentration, white blood cell count, mean platelet volume, albumin, alkaline phosphatase, globulin, prealbumin, total protein, blood urea nitrogen, creatinine, blood glucose, free T3, percentage of lymphocytes, total number of neutrophils, percentage of neutrophils, human chorionic gonadotropin, and vitamin D. Preferably, the laboratory test indicators include: white blood cell count, globulin, prealbumin, total protein, creatinine, total neutrophil count, and neutrophil percentage.

4. Use of the reagent for detecting the early diagnostic biomarkers according to any one of claims 1-3 in the preparation of products for fetal growth restriction diagnosis and / or risk assessment and prediction; Preferably, the diagnosis includes early diagnosis.

5. A product for diagnosing and / or predicting the risk of fetal growth restriction, characterized in that, The fetal growth restriction diagnostic and / or risk assessment prediction product includes a reagent for detecting the early diagnostic markers of any one of claims 1-3, and an adjuvant; preferably, the adjuvant includes at least one of: buffer solution, solvent or diluent, preservative, surfactant, washing solution, and sample pretreatment reagent.

6. A method for constructing a fetal growth restriction risk assessment and prediction model, comprising the following steps: Collect basic clinical information and / or laboratory test indicators of pregnant women, calculate the statistical difference P value between the two and the presence of fetal growth restriction, and screen clinical variables and biomarker variables according to the P value; Genome-wide association analysis was performed on the genetic information of pregnant women to target fetal growth restriction, and genetic factor variables were screened based on genome-wide significance thresholds; A model for assessing and predicting the risk of fetal growth restriction was developed based on clinical variables, biomarker variables, and genetic factors.

7. The construction method according to claim 6, characterized in that, The basic clinical information of the pregnant woman includes at least one of the following: age, pre-pregnancy weight, body mass index, systolic blood pressure, diastolic blood pressure, and height; and / or The laboratory test indicators include at least one of the following: hematocrit, hemoglobin concentration, white blood cell count, mean platelet volume, albumin, alkaline phosphatase, globulin, prealbumin, total protein, blood urea nitrogen, creatinine, blood glucose, free T3, percentage of lymphocytes, total number of neutrophils, percentage of neutrophils, human chorionic gonadotropin, and vitamin D.

8. The construction method according to claim 6 or 7, characterized in that, The construction method further includes: classifying the collected basic clinical information of pregnant women, and calculating the statistical difference P-value using different statistical methods; and / or The statistical difference P-value was calculated using regression analysis on the laboratory test indicators. Preferably, if the basic clinical information of the pregnant woman is a quantitative clinical variable, a t-test is used; If the basic clinical information of the pregnant woman is a categorical clinical variable for analysis, the chi-square test should be used. Preferably, the regression analysis method includes logistic regression, multinomial regression, and linear regression.

9. The construction method according to claim 6, characterized in that, The genetic factor variables are represented in the matrix data as single nucleotide polymorphism (SNP) genotype doses; Preferably, the formula for calculating the genotype dose is: Genotype dose = Pr(Het|Data) + 2 × Pr(Alt|Data) Wherein, Pr(Het|data) represents the posterior probability that the sample has a heterozygous mutation at this locus; Pr(Alt|data) represents the posterior probability that the sample has a homozygous mutation at that locus.

10. The application of the fetal growth restriction risk assessment and prediction model constructed by the construction method according to any one of claims 6-9 in the preparation of fetal growth restriction risk assessment and prediction products; Preferably, the fetal growth restriction risk assessment and prediction product includes an analysis module, which obtains prediction results by executing the fetal growth restriction risk assessment and prediction model constructed by the construction method according to any one of claims 6-9.

11. The application according to claim 10, characterized in that, The analysis module outputs a predicted fetal birth weight value, and determines whether there is a risk of fetal growth restriction based on the predicted fetal birth weight value. Preferably, the criteria for assessing the risk of fetal growth restriction based on predicted fetal birth weight are as follows: Compare predicted fetal birth weight with standard values ​​for fetal growth restriction. If the predicted birth weight of the fetus is less than the standard value for fetal growth restriction, then the risk of fetal growth restriction is considered high. If the predicted birth weight of the fetus is greater than or equal to the standard value for fetal growth restriction, the risk of fetal growth restriction is considered low.

12. A method for predicting fetal weight, comprising the following steps: The genotypes of the pregnant woman's rs1547643 and rs2430921 are converted into genotype doses, and then input together with the pregnant woman's basic clinical information and / or laboratory test indicators into the fetal growth restriction risk assessment and prediction model constructed by the construction method of any one of claims 6-9 to calculate the predicted fetal weight. Preferably, the method for changing the genotype dosage is as described in claim 9; The pregnant woman's basic clinical information and / or laboratory test results are as described in claim 2 or 3.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, performs the steps of calculating the fetal growth restriction risk assessment and prediction model constructed by the construction method of any one of claims 6-9 or the method of claim 12.

14. A system, characterized in that, The system includes an analysis module, which contains the computer-readable storage medium of claim 13; Preferably, the system further includes an input terminal and an output terminal.

15. An apparatus, characterized in that, The device is equipped with the system of claim 14, or contains the computer-readable storage medium of claim 13.

Images