Evaluation model and method for predicting neonatal adverse outcome based on prenatal examination

Abstract

The invention relates to an evaluation model and method for predicting a neonatal bad outcome based on antenatal examination. The main outcome of the neonatal adverse outcome comprises two parts of disease diagnosis and invasive treatment, and the neonatal adverse outcome is related to death and long-term health risk of children. The evaluation model comprises evaluation variables of weight gain during pregnancy, preeclampsia, umbilical artery S / D, congenital malformation and SGA, wherein the preeclampsia, the congenital malformation and the SGA are independent risk factors of neonatal adverse outcome. According to the scoring method, scoring is carried out based on the evaluation model, when at least one item of preeclampsia, congenital malformation, SGA and maternal congenital heart disease is' YES ', a neonatal adverse outcome high-risk group is prompted, and the higher the score is, the higher the probability of occurrence of the neonatal adverse outcome is. According to the invention, medical professionals can be rapidly helped to more effectively predict the neonatal adverse outcome, so that clinical decision-making is promoted.

Description

Technical Field

[0001] This invention relates to an assessment model and method for predicting adverse neonatal outcomes based on prenatal examinations, which can help professionals more effectively predict adverse neonatal outcomes and make timely referrals. Background Technology

[0002] With changes in birth policies, there has been a significant increase in older mothers having children. The complex demographics and diverse disease spectrum of this population present enormous challenges for obstetrics and neonatology. Serious neonatal diseases may present with mild early symptoms, making early prediction and intervention extremely difficult. Prenatal risk assessment of both mother and fetus can aid in clinical decision-making regarding intrauterine transport. Currently, guidelines for intrauterine transport are typically empirical and lack objective evaluation indicators.

[0003] Adverse neonatal outcomes primarily consist of two parts: disease diagnosis and invasive treatment. The following conditions should be considered when one or more of the above indicators are present: neonatal respiratory distress syndrome, neonatal seizures, intraventricular hemorrhage (grade III-IV), hypoxic-ischemic encephalopathy, cerebral infarction, necrotizing enterocolitis, bronchopulmonary dysplasia, neonatal sepsis, neonatal infectious pneumonia (requiring mechanical ventilation or CPAP), other respiratory diseases (primary atelectasis, respiratory failure), bacterial meningitis; resuscitation, ventilatory support (mechanical ventilation and / or continuous positive airway pressure), pneumothorax requiring intercostal catheter decompression, any intracavitary surgery, central venous or arterial catheterization. Adverse neonatal outcomes are associated with mortality and long-term health risks in children.

[0004] Newborn health assessment at birth is crucial for further monitoring and intervention in obstetrics and neonatology. Preterm infants (<32 weeks) undoubtedly require continued monitoring. For late preterm and full-term infants, the assessment of whether further monitoring is needed primarily relies on the Apgar score. This score, used for over 70 years, still has limitations, such as being affected by maternal sedation or anesthesia, only representing the infant's physiological condition at a specific point in time, and being highly subjective. Using this score alone is not effective in predicting neonatal outcomes. International scholars, such as Holmgren et al., combined fetal heart rate monitoring and maternal factors to predict adverse outcomes in full-term infants (5-minute Apgar score of 7 or umbilical artery pH <7). The model's AUC was 0.710, indicating poor discrimination. The incidence of adverse outcomes in the full-term infant cohort in this study was 0.77%, a very low rate, making prediction even more challenging. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides an assessment model and method for predicting adverse neonatal outcomes based on prenatal examinations. This model can screen and utilize important features to help medical professionals more effectively predict adverse neonatal outcomes, thereby promoting clinical decision-making.

[0006] Therefore, the technical solution adopted by the present invention is: an assessment model for predicting adverse neonatal outcomes based on prenatal examination, characterized in that the assessment model includes assessment variables such as gestational weight gain, early-onset preeclampsia, umbilical artery S / D ratio, congenital malformations, and SGA, wherein early-onset preeclampsia, congenital malformations, and SGA are independent risk factors for adverse neonatal outcomes.

[0007] Preferably, the evaluation variables of the evaluation model also include maternal congenital heart disease, placenta previa, assisted reproductive technology, or parity, wherein maternal congenital heart disease is an independent risk factor for adverse neonatal outcomes.

[0008] Preferably, the evaluation variables of the assessment model are gestational weight gain, early-onset preeclampsia, umbilical artery S / D ratio, congenital malformations, SGA, and maternal congenital heart disease, among which early-onset preeclampsia, congenital malformations, SGA, and maternal congenital heart disease are independent risk factors for adverse neonatal outcomes.

[0009] As a preferred approach, each assessment variable was assigned a score. A score of ≥6 indicated a high risk of adverse neonatal outcomes, and the higher the score, the greater the likelihood of adverse neonatal outcomes. The specific score values ​​are as follows:

[0010]

[0011]

[0012] This invention also provides a scoring method for predicting adverse neonatal outcomes based on prenatal examinations. The scoring is based on the above assessment model. If any one of the following is "yes": early-onset preeclampsia, congenital malformation, SGA, and maternal congenital heart disease, it indicates that the newborn is in a high-risk group for adverse neonatal outcomes. The higher the score, the greater the likelihood of adverse neonatal outcomes.

[0013] The beneficial effects of this invention are as follows: it collects and screens a large number of risk variables involved in the model building, and finally determines several key risk variables for scoring to determine whether adverse outcomes will occur. After verification, the accuracy of the assessment is very high, which can quickly help medical professionals to more effectively predict adverse outcomes in newborns, thereby promoting clinical decision-making. Attached Figure Description

[0014] Figure 1 This is a flowchart of the evaluation model of the present invention.

[0015] Figure 2 ROC curve for the prenatal prediction model in the internal validation group.

[0016] Figure 3 The ROC curve for the external validation group of the prenatal risk scoring system of this invention.

[0017] Figure 4 This is the calibration curve for the external validation queue of the prenatal risk scoring system of this invention. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0019] refer to Figure 1 The main steps of the evaluation model of this invention are as follows:

[0020] 1. Data Collection: Acquiring relevant patient information. Provided that legal data access permissions are obtained and medical data protection regulations are complied with, this invention can acquire medical records and clinical data through an Electronic Health Record (EHR) system, real-time data through medical devices, or laboratory results data from a laboratory database. This invention does not impose excessive limitations on the specific data acquisition methods, as long as the relevant data can be obtained.

[0021] 2. Data Preprocessing: Inclusion Criteria: 1) Singleton pregnancy; 2) Live birth; 3) Complete clinical data (general information, medical history, pregnancy complications, late-pregnancy fetal ultrasound information, neonatal birth record, and other relevant case data). Exclusion Criteria: 1) Multiple pregnancies, stillbirths, or neonatal deaths; 2) Missing data for key research indicators, such as maternal pregnancy-related illnesses or late-pregnancy fetal umbilical artery blood flow parameters; 3) Abnormal values ​​affecting the assessment.

[0022] 3. Save the data as a database file.

[0023] 4. Divide the data into a training set and a test set.

[0024] 5. Train the model to generate a scoring system.

[0025] 6. Finally, evaluate the model's performance and select the best model for application.

[0026] The following detailed description of the specific model selection and model performance evaluation of the present invention is provided in conjunction with specific embodiments:

[0027] 1. The modeling cohort of this study included 1,651 singleton live births and their mothers who delivered between July 2016 and January 2021 in the prospective cohort of Shanghai Xinhua Hospital – the “Early Life 1000 Days Project”.

[0028] 2. Inclusion criteria: 1) Singleton pregnancy; 2) Live birth; 3) Complete clinical data (general information, medical history, pregnancy complications, fetal ultrasound information in late pregnancy, newborn birth record and other relevant case data).

[0029] 3. Exclusion criteria: 1) Multiple pregnancies, stillbirths, or neonatal deaths; 2) Missing data for key research indicators, such as maternal prenatal conditions or fetal umbilical artery blood flow parameters in late pregnancy; 3) Abnormal values ​​that affect the judgment.

[0030] 4. The risk variables involved in building the model include the following indicators:

[0031] ① General information: pregnant woman's education level, age, BMI, adverse pregnancy history, underlying diseases (hypertension, diabetes, history of congenital heart disease), ART, etc.

[0032] ②Pregnancy data: weight gain during pregnancy, GBS infection, pregnancy complications (preeclampsia, diabetes), placenta previa, congenital malformations.

[0033] ③ Ultrasound indicators in late pregnancy: gestational age on ultrasound, estimated weight gain (SGA), umbilical artery end-systolic peak / end-diastolic peak (S / D), umbilical artery pulsatility index (PI), and umbilical artery resistance index (RI).

[0034] ④ Delivery information: Method of termination of pregnancy, fetal position, meconium staining, placental adhesion or implantation, gestational age, sex, birth weight, 1-minute and 5-minute Apgar scores of the newborn.

[0035] Variable definitions and groupings:

[0036] 1) Based on the years of education, the education level is divided into three groups: low (≤12 years), medium (12-15 years), and high (>15 years); the age is divided into two groups: ≤35 years and >35 years; the parity is divided into primiparous (0 times) and non-primiparous (≥1 time) based on the delivery history.

[0037] 2) Pre-pregnancy BMI (kg / m): weight / height 2 .

[0038] 3) Adverse pregnancy and childbirth history: Based on whether there has been a history of stillbirth, miscarriage or recurrent miscarriage (≥3 times), it is divided into two groups: none and yes.

[0039] 4) ART: Based on whether assisted reproductive technology was used in this pregnancy, it is divided into two groups: no and yes.

[0040] 5) Pregnancy-related diabetes: According to the 1999 WHO diagnostic criteria for diabetes.

[18] The test was divided into two groups: those with and without glucose. The following criteria were met: ① Fasting plasma glucose ≥ 126 mg / dL (7.0 mmol / L); ② 2-hour plasma glucose ≥ 200 mg / dL (7.0 mmol / L) after a 75g glucose tolerance test; ③ Accompanied by typical hyperglycemia or hyperglycemic crisis, random blood glucose ≥ 200 mg / dL (11.1 mmol / L).

[0041] 6) Preconception hypertension: According to the 2005 "Guidelines for the Prevention and Treatment of Hypertension in China"

[19] Definition: In the absence of antihypertensive medication, systolic blood pressure ≥140 mmHg and / or diastolic blood pressure ≥90 mmHg are divided into two groups: "without" and "with".

[0042] 7) Prenatal congenital heart disease: Based on the presence or absence of congenital heart diseases such as atrial septal defect, ventricular septal defect, pulmonary artery stenosis, and aortic coarctation, it is divided into two groups: none and present.

[0043] 8) Weight gain during pregnancy: Based on whether the weight gain during pregnancy is within the standard range, it is divided into three groups: insufficient weight gain during pregnancy, normal weight gain during pregnancy, and excessive weight gain during pregnancy. The classification refers to the "Range of Weight Gain in Pregnant Women" in the "Monitoring and Evaluation of Weight Gain in Chinese Women during Pregnancy" published by the Chinese Nutrition Society in 2022 (see Appendix Table 1).

[0044] 9) Preeclampsia: According to the Chinese Guidelines for the Diagnosis and Treatment of Hypertensive Disorders in Pregnancy (2015)

[20] After 20 weeks of gestation, a systolic blood pressure / diastolic blood pressure ≥140 / 90 mmHg, accompanied by proteinuria ≥300 mg / h, or a proteinuria / creatinine ratio of [missing information].

[0045] ≥0.3, or random urine protein (+); or although there is no proteinuria, but any of the following organs or systems are involved: vital organs such as the heart, lungs, liver, and kidneys; abnormal changes in the blood system, digestive system, or nervous system; or placental-fetal involvement. Early-onset preeclampsia is defined as having onset before 34 weeks of gestation, and is divided into two groups based on whether or not early-onset preeclampsia is present: none and present.

[0046] 10) Gestational diabetes mellitus: According to the "Guidelines for the Diagnosis and Treatment of Gestational Diabetes Mellitus (2014)" in my country

[21] Diagnostic criteria: Pre-glucose blood glucose ≥5.1 mmol / L (126 mg / dL) or 1-hour glucose level ≥10.0 mg / dL in a 75g glucose tolerance test.

[0047] (180mg / dl) or ≥8.5mmol / L (153mg / dl) in 2 hours, and requiring insulin or oral medication to control blood glucose, are divided into two groups: no and yes.

[0048] 11) Placenta previa: After 28 weeks of gestation, placenta previa is diagnosed by transvaginal ultrasound and divided into two groups: none and present.

[0049] 12) Placental adhesion or implantation: Based on the diagnosis of placental adhesion or implantation by obstetric ultrasound in late pregnancy, it is divided into two groups: none and present. 13) GBS infection: Based on whether the vaginal secretion culture is positive, it is divided into two groups: none and present.

[0050] 14) Estimated gestational age (SGA): Based on the Intergrowth-21st estimated gestational age standard, infants are divided into two groups: those with and those without, depending on whether their gestational age is below the 10th percentile of the average estimated gestational age.

[0051] 15) SGA: Based on the Intergrowth-21st birth weight standard, it is divided into two groups: none and present, depending on whether it is below the 10th percentile of the birth weight of infants of the same sex and gestational age.

[0052] 16) Congenital malformations: Based on the presence of severe congenital heart disease, gastrointestinal atresia, hydrocephalus, diaphragmatic hernia, visceral bulging, etc., on ultrasound, they are divided into two groups: none and present.

[0053] 17) Delivery method: Based on whether it is vaginal delivery or abdominal cesarean section, it is divided into two groups: vaginal delivery and cesarean section.

[0054] 18) Fetal position: Based on whether the fetus is in the head-down position at the time of delivery, it is divided into two groups: head-down and non-head-down.

[0055] 19) Meconium-stained amniotic fluid: Based on the degree of meconium staining (I° staining is light green, II° staining is yellow-green, III ... yellow-green, III° staining is yellow-green, III° staining is yellow-green, III° staining is yellow

[0056] The pollution is brownish-yellow and is divided into two groups: none or I°-II° and III°.

[0057] 20) Gestational age: Divided into three groups: full-term infants (≥37 weeks), late preterm infants (34-37 weeks), and preterm infants (<34 weeks). Sex is divided into male and female groups based on the sex of the newborn.

[0058] 5. Primary Outcomes: Based on the presence of any of the following diseases and invasive treatments after birth, the newborn is classified into the poor outcome group: Disease diagnosis: neonatal respiratory distress syndrome, neonatal seizures, intraventricular hemorrhage (grade III-IV), hypoxic-ischemic encephalopathy, cerebral infarction, necrotizing enterocolitis, bronchopulmonary dysplasia, neonatal sepsis, neonatal infectious pneumonia (requiring mechanical ventilation or CPAP), other respiratory diseases (primary atelectasis, respiratory failure), bacterial meningitis; resuscitation, ventilation support (mechanical ventilation and / or continuous positive airway pressure), pneumothorax requiring intercostal catheter decompression, any body cavity surgery, central venous or arterial catheter placement.

[0059] 6. The prenatal prediction model used the presence of adverse outcomes in newborns as the dependent variable. Based on univariate analysis and a p-value of <0.20, 12 variables were selected as candidate variables, including parity, ART, pre-gestational diabetes, pre-gestational congenital heart disease, gestational weight gain, early-onset preeclampsia, placenta previa, placenta accreta or accreta, estimated fetal weight, umbilical artery S / D ratio, umbilical artery PI, umbilical artery RI, and congenital malformations. Five-fold cross-validation was used, and stepwise regression analysis was employed to select variables associated with adverse outcomes, resulting in five combined models. Gestational weight gain, early-onset preeclampsia, umbilical artery S / D ratio, congenital malformations, and SGA were included in all models. Models 1 and 4 had more variables than the other models (see Table 1).

[0060] Table 1. Training of the 5-fold cross-validation prenatal model

[0061]

[0062] 7. The above models were evaluated. Based on predictive performance and the number of variables in the internal validation set, Model 2 was ultimately selected as the best model. The area under the ROC curve for Model 2 was 0.85, with a 95% CI (0.70–1.00), a sensitivity of 88.5%, a specificity of 71.4%, a positive predictive value of 12.2%, and a negative predictive value of 99.3%. The Hosmer-Lemeshow test results showed χ² = 4.508 and P = 0.809, indicating that this predictive model has good discriminative validity and predictive calibration (see Table 2).

[0063] Figure 2 The ROC curves for the test sets of each model are shown. It can be seen that among the prediction models with 6 variables, Model 2 performs better.

[0064] Table 2 Predictive performance of the prenatal model on the corresponding validation set.

[0065]

[0066] Note: Models 2, 3, and 5 each have 6 variables, while Models 1 and 4 each have 7 variables.

[0067] *The final model was selected.

[0068] Multivariate logistic regression results showed that pre-pregnancy congenital heart disease, preeclampsia (early onset), predicted birth weight (SGA), congenital malformations, and umbilical artery blood flow (S / D) were all independent risk factors for adverse neonatal outcomes (p<0.05), as detailed in Table 3.

[0069] Table 3 Multivariate Regression Analysis of Prenatal Risk Prediction Model

[0070]

[0071] Note: *Odds ratio adjusted based on included variables such as parity, ART, pre-gestational diabetes, placenta previa, placenta accreta, or placenta implantation.

[0072] 8. To further realize the clinical application of this model, this study used the method described by Sullivan et al. to create a simple integer risk index. The prenatal risk prediction scoring scale has 6 scoring items, with a total score of 0-30. The intrapartum risk prediction scoring scale also has 6 scoring items, with a total score of 0-31. See Table 4 for details.

[0073] Table 4. Risk Prediction Scoring Table for Adverse Outcomes in Neonates

[0074]

[0075]

[0076] External validation: The external validation cohort collected data from 1,598 pregnant women and their newborns with complete clinical data from Jiaxing and Wuxi Maternal and Child Health Hospitals from January to March 2022. The incidence of adverse neonatal outcomes was 4.8%.

[0077] Inclusion criteria: 1) Singleton pregnancy; 2) Live birth; 3) Complete clinical data (general information, medical history, pregnancy complications, fetal ultrasound information in late pregnancy, newborn birth record and other relevant case data).

[0078] Exclusion criteria: 1) Multiple pregnancies, stillbirths, or neonatal deaths; 2) Missing data for key research indicators, such as maternal prenatal conditions or fetal umbilical artery blood flow parameters in late pregnancy; 3) Abnormal values ​​that affect the judgment.

[0079] The predictive ability of the intrapartum risk scoring system for adverse neonatal outcomes was analyzed using the area under the ROC curve. The results showed an AUC of 0.79 (95% CI 0.73-0.85). Figure 3 ).

[0080] The scoring threshold was selected using the Youden index maximization method. When the score was ≥6, the sensitivity was 45.3%, the specificity was 97.5%, the positive predictive value was 80.6%, and the negative predictive value was 96.2%. A calibration curve was plotted (e.g., ...). Figure 4 The actual curve fits the ideal curve reasonably well (P = 0.202).

[0081] The results in summary show that the evaluation model of this invention exhibits good discrimination and calibration.

[0082] It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

Claims

1. An assessment model for predicting adverse neonatal outcomes based on prenatal examinations, characterized in that... The assessment model includes the following assessment variables: gestational weight gain, early-onset preeclampsia, umbilical artery S / D ratio, congenital malformations, and SGA. Among these, early-onset preeclampsia, congenital malformations, and SGA are independent risk factors for adverse neonatal outcomes.

2. The assessment model for predicting adverse neonatal outcomes based on prenatal examinations according to claim 1, characterized in that... The assessment variables in the assessment model also include maternal congenital heart disease, placenta previa, assisted reproductive technology, or parity, among which maternal congenital heart disease is an independent risk factor for adverse neonatal outcomes.

3. The assessment model for predicting adverse neonatal outcomes based on prenatal examinations according to claim 2, characterized in that... The assessment variables in the assessment model are gestational weight gain, early-onset preeclampsia, umbilical artery S / D ratio, congenital malformations, SGA, and maternal congenital heart disease. Among these, early-onset preeclampsia, congenital malformations, SGA, and maternal congenital heart disease are independent risk factors for adverse neonatal outcomes.

4. The assessment model for predicting adverse neonatal outcomes based on prenatal examinations according to claim 3, characterized in that... Each assessment variable was assigned a score. A score of ≥6 indicated a high risk of adverse neonatal outcomes, and the higher the score, the greater the likelihood of adverse neonatal outcomes. The specific score values ​​are as follows:

5. A scoring method for predicting adverse neonatal outcomes based on prenatal examinations, wherein the scoring is based on the assessment model described in claim 4, and if any one of early-onset preeclampsia, congenital malformation, SGA, and maternal congenital heart disease is "yes", it indicates that the newborn is in a high-risk group for adverse neonatal outcomes, and the higher the score, the greater the likelihood of adverse neonatal outcomes.

Images