Premature delivery prediction method and device, electronic equipment and readable storage medium

Abstract

The invention provides a premature delivery prediction method and device, electronic equipment and a readable storage medium, and relates to the technical field of data analysis. The method comprises the following steps: based on a preset model, scoring acquired comprehensive information of a pregnant woman in an early pregnancy stage; wherein the comprehensive information comprises demographic information, clinical information and biochemical index information; and predicting the premature delivery risk of the pregnant woman according to the scoring result. According to the method, the comprehensive information of the pregnant women at the early pregnancy stage is scored through the preset model, the premature delivery risk of the pregnant women is predicted according to the scoring result, quantitative evaluation of premature delivery risk prediction of the pregnant women at the early pregnancy stage is achieved, and objectivity and certain accuracy are achieved; and the difficulty and the cost of obtaining the comprehensive information of the pregnant women in the early pregnancy are relatively low, the difficulty and the cost of predicting the premature delivery risk of the pregnant women in the early pregnancy are reduced, and medical resources are saved.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a method, device, electronic device and readable storage medium for predicting premature birth. Background Art

[0002] Premature birth generally refers to births between 28 weeks and less than 37 weeks of pregnancy. Newborns born at this time are called premature babies. Premature births are prone to cause corresponding complications in the premature infant's perinatal period, such as respiratory distress syndrome, cerebral hemorrhage and infection, which increase the risk of death for premature infants. Premature birth complications are also the main cause of death in children under 5 years old, which will bring serious economic burdens to families and society.

[0003] In this regard, it is particularly important to prevent premature birth, such as by predicting premature birth so that appropriate intervention measures can be taken to prevent premature birth. At present, some premature birth prediction methods based on prospective cohort studies rely on high-level medical care institutions. Although the prediction performance is good, the cost is high and it is difficult to promote them in primary clinical practice. In addition, the corresponding markers in most premature birth prediction methods are measured in the middle and late pregnancy. When the risk of premature birth is found, the pregnant woman is close to or has become a high-risk pregnant woman, which limits the effectiveness of the intervention measures to be taken or the cost of taking the corresponding intervention measures is extremely high. Summary of the invention

[0004] The problem solved by the invention is: how to reduce the cost of premature birth prediction.

[0005] In order to solve the above problems, the present invention provides a method, device, electronic device and readable storage medium for predicting premature birth.

[0006] In a first aspect, the present invention provides a method for predicting premature birth, comprising:

[0007] Based on the preset model, the comprehensive information of the pregnant women in the early stage of pregnancy is scored; wherein the comprehensive information includes demographic information, clinical information and biochemical index information;

[0008] The risk of premature birth of the pregnant woman is predicted based on the scoring result.

[0009] Optionally, before scoring the acquired comprehensive information of the pregnant woman in the early stage of pregnancy based on the preset model, the method for predicting premature birth further includes:

[0010] Acquiring first data including the comprehensive information of different pregnant women who have completed childbirth;

[0011] determining risk predictors in the first data that are statistically associated with preterm birth;

[0012] Construct the preset model for assigning a score to each of the risk prediction factors.

[0013] Optionally, determining a risk prediction factor in the first data that is statistically associated with premature birth comprises:

[0014] Using univariate logistic regression analysis, screening the first variable in the first data that is statistically associated with premature birth;

[0015] Multivariate logistic regression analysis was used to screen the risk prediction factors that were statistically associated with premature birth among the first variable and the preset second variable related to premature birth.

[0016] Optionally, constructing the preset model for assigning a score to each of the risk prediction factors includes:

[0017] Determine the strength of association of each of the risk predictors with preterm birth;

[0018] Construct the preset model; wherein the preset model is used to assign a score to each of the risk prediction factors based on the association strength.

[0019] Optionally, predicting the risk of premature birth of the pregnant woman according to the scoring result includes:

[0020] Determining whether the sum of the scores of all the risk prediction factors is greater than a preset critical value;

[0021] If so, the risk of premature birth of the pregnant woman is predicted to be medium-high risk; if not, the risk of premature birth of the pregnant woman is predicted to be low risk.

[0022] Optionally, after constructing the preset model, constructing the preset model for assigning a score to each of the risk prediction factors further includes:

[0023] acquiring second data including the comprehensive information of different pregnant women who have completed childbirth;

[0024] Verifying the preset model based on the second data;

[0025] The preset model is optimized based on the verification result.

[0026] Optionally, the verifying the preset model based on the second data includes:

[0027] Based on the second data, determining the accuracy, sensitivity, specificity, positive predictive value and negative predictive value of the preset model;

[0028] The optimizing the preset model based on the verification result includes:

[0029] When at least one of the accuracy, sensitivity, specificity, positive predictive value and negative predictive value does not meet the corresponding preset standard, the preset model is optimized.

[0030] In a second aspect, the present invention provides a device for predicting premature birth, comprising:

[0031] A scoring unit, used to score the acquired comprehensive information of the pregnant woman in the early stage of pregnancy based on a preset model; wherein the comprehensive information includes demographic information, clinical information and biochemical index information;

[0032] The risk prediction unit is used to predict the risk of premature birth of the pregnant woman according to the scoring result.

[0033] In a third aspect, the present invention provides an electronic device, including a memory and a processor;

[0034] The memory is used to store computer programs;

[0035] The processor is used to implement the method for predicting premature birth as described in the first aspect when executing the computer program.

[0036] In a fourth aspect, the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, and when the computer program is read and executed by a processor, the method for predicting premature birth as described in the first aspect is implemented.

[0037] Compared with the prior art, the present invention has the following beneficial effects: the present invention scores the comprehensive information of pregnant women in early pregnancy through a preset model, and predicts the risk of premature birth of the pregnant women according to the scoring results. On the one hand, based on the preset model, the quantitative evaluation of the risk prediction of premature birth of pregnant women in early pregnancy is realized, which is objective and accurate to a certain extent; on the other hand, the difficulty and cost of obtaining the comprehensive information of pregnant women in early pregnancy are relatively low, such as through basic medical institutions or health care institutions with relatively limited resources, which reduces the difficulty and cost of predicting the risk of premature birth of pregnant women in early pregnancy, and is conducive to saving medical resources; on the other hand, the comprehensive information includes demographic information, clinical information and biochemical indicator information, so as to more comprehensively evaluate the factors that may affect the premature birth of pregnant women, and improve the accuracy of the risk prediction of premature birth of pregnant women in early pregnancy. In addition, the preset model predicts the risk of premature birth for pregnant women in early pregnancy based on comprehensive information with wide coverage. On the one hand, it reduces the prediction error caused by the missing or abnormal types of information that may affect premature birth, and improves the accuracy, comprehensiveness, stability and robustness of the prediction. On the other hand, since the selected information covers many aspects of pregnant women in early pregnancy, the preset model is highly applicable when dealing with different types of pregnant women. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic diagram of a flow chart of a method for predicting premature birth in an embodiment of the present invention;

[0039] Figure 2 A schematic diagram of a flow chart of a method for predicting premature birth in an embodiment of the present invention;

[0040] Figure 3 1 is a schematic diagram of a sub-flow chart of step 120 in an embodiment of the present invention;

[0041] Figure 4 1 is a schematic diagram of a sub-flow chart of step 130 in an embodiment of the present invention;

[0042] Figure 5 It is a structural block diagram of a device for predicting premature birth in an embodiment of the present invention;

[0043] Figure 6 Schematic diagram of the structure of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0045] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein.

[0046] Combination Figure 1 As shown, an embodiment of the present invention provides a method for predicting premature birth, comprising:

[0047] Step 200: Based on a preset model, score the acquired comprehensive information of the pregnant woman in the early stage of pregnancy; wherein the comprehensive information includes demographic information, clinical information and biochemical indicator information.

[0048] This method is used to quantitatively evaluate the comprehensive information of pregnant women in early pregnancy and predict the risk of premature birth. Specifically, through step 200, a preset premature birth risk prediction model (referred to as the preset model) is used to score the comprehensive information of the pregnant women in early pregnancy according to the comprehensive information of the pregnant women in early pregnancy, and a corresponding scoring result is obtained, so as to reflect the risk of premature birth of the pregnant women in early pregnancy through the scoring result.

[0049] Among them, considering that there are many types of risk factors that may lead to premature birth, the comprehensive information includes demographic information, clinical information and biochemical indicator information, so as to more comprehensively evaluate the factors that may affect the premature birth of pregnant women and provide convenience for effectively identifying the risk of premature birth. Demographic information includes age, household registration, ethnicity or race, education level, marital status, economic status, occupation, parity and gravidity, twin / multiple pregnancy, whether this is assisted reproduction, etc.; clinical information includes adverse pregnancy and delivery history, history of intrauterine operation, high-risk pregnant women, history of immune system diseases, history of hypertension, history of diabetes, thrombophilia, chronic kidney disease, depression / bipolar disorder, epilepsy, etc.; biochemical indicator information includes biomarkers such as urine sugar, platelet count, and alanine aminotransferase. In some embodiments, the comprehensive information of pregnant women in early pregnancy can be obtained when pregnant women in early pregnancy have their first prenatal check-up. For example, medical institutions can collect comprehensive information including demographic information, clinical information and biochemical indicator information of pregnant women in early pregnancy through interviews, physical examinations, laboratory tests, etc. during the first prenatal check-up. In some embodiments, the biochemical indicator information can be selected from the corresponding routine detection indicators (such as urine sugar, platelet count, alanine aminotransferase, etc.), so as to reduce the difficulty and cost of obtaining biochemical indicator information while reducing the difficulty and cost of predicting the risk of premature birth, thereby improving the popularity and practicality of premature birth prediction methods.

[0050] Step 300: predicting the risk of premature birth of the pregnant woman according to the scoring result.

[0051] Specifically, through step 300, according to the obtained scoring result, the risk of premature birth corresponding to the scoring result (such as low risk, medium-high risk, etc.) is determined to achieve the prediction of the risk of premature birth of pregnant women in early pregnancy. In some embodiments, the prediction result can be used to assist medical workers in assessing the risk of premature birth of pregnant women, so as to timely discover the possible risk of premature birth of pregnant women in early pregnancy, and formulate corresponding preventive measures as soon as possible to reduce the incidence of premature birth, and avoid the situation that when the risk of premature birth is discovered in the middle and late pregnancy, the pregnant woman may have been close to or become a high-risk pregnant woman, and the cost of taking corresponding intervention measures is extremely high.

[0052] In this way, the method of this embodiment scores the comprehensive information of pregnant women in early pregnancy through a preset model, and predicts the risk of premature birth of pregnant women according to the scoring results. On the one hand, based on the preset model, the quantitative evaluation of the risk prediction of premature birth of pregnant women in early pregnancy is realized, which is objective and accurate to a certain extent; on the other hand, the difficulty and cost of obtaining the comprehensive information of pregnant women in early pregnancy are relatively low, such as through basic medical institutions or health care institutions with relatively limited resources, which can realize the collection of comprehensive information, reduce the difficulty and cost of the risk prediction of premature birth of pregnant women in early pregnancy, and help save medical resources; on the other hand, the comprehensive information includes demographic information, clinical information and biochemical index information, so as to more comprehensively evaluate the factors that may affect the premature birth of pregnant women, and improve the accuracy of the risk prediction of premature birth of pregnant women in early pregnancy. In addition, the preset model predicts the risk of premature birth of pregnant women in early pregnancy based on comprehensive information with a wide and comprehensive information coverage. On the one hand, it reduces the prediction error caused by the lack or abnormality of individual information types that affect premature birth, and improves the accuracy, comprehensiveness, stability and robustness of the prediction; on the other hand, because the selected information covers many aspects of pregnant women in early pregnancy, the preset model has strong applicability when dealing with different types of pregnant women.

[0053] Optionally, combined Figure 1 , Figure 2 As shown, before step 200, the premature birth prediction method further includes:

[0054] Step 110: Acquire first data including comprehensive information of different pregnant women who have completed childbirth;

[0055] Step 120, determining a risk prediction factor in the first data that is statistically associated with premature birth;

[0056] Step 130: construct a preset model for assigning a score to each risk prediction factor.

[0057] By pre-constructing a preset model, the efficiency of subsequent premature birth prediction based on the comprehensive information of pregnant women in early pregnancy is improved. In order to ensure the applicability and versatility of the preset model, the construction of the preset model is based on the first data obtained including the comprehensive information of different pregnant women who have completed childbirth, that is, the first data includes the comprehensive information of multiple different pregnant women who have completed childbirth, wherein the comprehensive information includes demographic information, clinical information and biochemical index information. Specifically, through step 110, the first data is obtained, and these data can be obtained from multiple prenatal examinations, medical records, laboratory tests and other channels of pregnant women who have completed childbirth. Thereafter, through step 120, the first data is analyzed using an analysis method such as a statistical method (such as univariate and multivariate logistic regression analysis), and the factors (referred to as risk prediction factors) that are significantly statistically associated with premature birth in the first data are screened out, and these factors may significantly affect the probability of premature birth. Afterwards, through step 130, a preset model is constructed for scoring each screened risk prediction factor. In some embodiments, when the preset model assigns scores to risk prediction factors, different weights can be set according to the degree of influence of different risk prediction factors, so that the comprehensive information of each pregnant woman can be quantitatively scored through the preset model to reflect the risk of premature birth.

[0058] Optionally, combined Figure 2 , Figure 3 As shown, step 120 includes:

[0059] Step 121: Use univariate logistic regression analysis to screen the first variable in the first data that is statistically associated with premature birth.

[0060] Univariate logistic regression analysis is a statistical analysis method used to evaluate the association between a single independent variable (risk factor) and a binary dependent variable (such as premature birth or not). Specifically, through step 121, each type of information in all types of information in the comprehensive information in the first data is used as a potential risk factor (independent variable), and logistic regression analysis is performed separately to calculate the statistical association between each independent variable and the occurrence of premature birth.

[0061] For example, the statistical significance between the independent variable and the dependent variable is reflected by the p-value; the calculation formula of the p-value is:

[0062] p = 2 × (1-Φ (|z|)),

[0063] Where Φ is the cumulative distribution function of the standard normal distribution, and the z value is calculated as follows:

[0064] z = β / SE,

[0065] Among them, β is the regression coefficient of the independent variable, which indicates the direction and magnitude of the independent variable's influence on the dependent variable; SE is the standard error of the regression coefficient, which reflects the uncertainty of the estimated β value. A smaller p value (such as p less than 0.05) indicates that there is a significant statistical association between the independent variable and the dependent variable. The independent variable with a p value less than 0.05 in the first data is selected as the first variable.

[0066] In some embodiments, whether or not a premature birth corresponds to a low risk of premature birth, a high risk of premature birth (or a medium-high risk of premature birth), respectively.

[0067] Step 122: using multivariate logistic regression analysis, screening risk prediction factors that are statistically associated with premature birth in the first variable and a preset second variable related to premature birth.

[0068] Multivariate logistic regression is a statistical analysis method used to simultaneously evaluate the independent associations between multiple independent variables and binary dependent variables. Specifically, through step 122, the first variable selected in step 121 (the variable with p<0.05 in the univariate logistic regression analysis) is subjected to multivariate logistic regression analysis together with the preset second variable related to premature birth. Variable selection methods such as stepwise regression (such as forward selection, backward elimination or stepwise method) can be combined to select variables that are still significantly associated with premature birth after controlling the influence of other variables as risk prediction factors.

[0069] For example, a multi-factor logistic regression model is first constructed: the first variable and the second variable are simultaneously included in the model to form a regression equation of the following form:

[0070] logit(P(Y=1))=β0+β1X1+β2X2+...+β n X n ,

[0071] Among them, Y is a binary dependent variable (such as premature birth or not); X1, X2, ..., X n is the independent variable (including the first variable and the second variable), β0 is the intercept term, β1, β2, ..., β n is the regression coefficient of each variable, indicating the degree and direction of the influence of the independent variable on the dependent variable;

[0072] Next, stepwise regression methods (such as forward selection, backward elimination, or stepwise method) are used for variable screening. These methods can help identify independent variables that are still statistically associated with preterm birth after controlling for the influence of other variables, thereby optimizing the predictive ability of the pre-set model;

[0073] Then, the regression coefficient β, standard error SE and p-value are calculated for each variable; the calculation of p-value is the same as that of univariate logistic regression analysis, and the formula is:

[0074] p = 2 × (1-Φ (|z|)),

[0075] Among them, z = β / SE. A smaller p value (such as p < 0.05) indicates that the variable is still significantly associated with the risk of premature birth after controlling for the influence of other factors;

[0076] Finally, variables with a p-value less than 0.05 in the multivariate logistic regression model were screened out from all independent variables (the first variable and the second variable). These independent variables will serve as risk prediction factors significantly associated with premature birth, providing an important basis for premature birth prediction and corresponding clinical decision-making.

[0077] In this way, through the screening of step 121 and step 122, the risk prediction factor is finally obtained. Specifically, the univariate logistic regression analysis in step 121 is used to preliminarily identify the independent variables related to premature birth in the first data, and quickly screen out the risk factors with statistical significance as the first variable; and in step 122, based on the first variable and the second variable, the influence of other variables is further controlled by multivariate logistic regression analysis, so as to ensure the independence between the selected (or obtained) risk prediction factors. In this way, the accuracy and reliability of the risk prediction factor are improved, and a more scientific basis is provided for the prevention of premature birth and clinical decision-making. Among them, the preset variables related to premature birth include the variables related to premature birth disclosed in the prior art (such as research reports) (which are risk factors identified in existing studies), which are recorded as the second variable. By introducing the second variable as reported in the research, on the one hand, the existing research results are fully utilized to ensure that the preset model can cover a wider range of influencing factors, thereby improving the comprehensiveness and accuracy of premature birth prediction; on the other hand, by combining the existing research results (second variable) with the new data (first variable), new perspectives and directions can be provided for subsequent research, and further exploration and development in related fields can be promoted.

[0078] Optionally, combined Figure 2 , Figure 4 As shown, step 130 includes:

[0079] Step 131, determine the strength of association between each risk prediction factor and preterm birth.

[0080] Specifically, in order to simultaneously consider multiple independent variables (risk prediction factors) and simultaneously evaluate the impact of each risk prediction factor on preterm birth, it is preferred to determine the strength of association between each risk prediction factor and preterm birth based on multivariate logistic regression analysis.

[0081] For example, based on multivariate logistic regression analysis, the strength of association between each risk prediction factor and the occurrence of premature birth can be quantified by calculating the regression coefficient β of each risk prediction factor. Among them, β represents the degree and direction of the influence of the independent variable on the dependent variable. Specifically, the value of β reflects how the log-odds of the dependent variable changes for each unit change of the independent variable. For example, the larger the absolute value of β, the stronger the association between the independent variable and the dependent variable. If β is positive, it indicates that an increase in the independent variable will increase the risk of premature birth (i.e., the probability of premature birth is higher); if β is negative, it indicates that an increase in the independent variable will reduce the risk of premature birth.

[0082] The strength of the association between each risk predictor and the occurrence of premature birth can also be quantified by calculating the odds ratio (OR) of each risk predictor. The odds ratio indicates the relative risk change of premature birth when the risk predictor changes by one unit, while controlling for the influence of other variables. For example, OR>1 indicates that the risk predictor is positively correlated with the risk of premature birth, while OR<1 indicates a negative correlation. The corresponding calculation method is: OR=e β , where β is the regression coefficient of the corresponding risk prediction factor.

[0083] Step 132: construct a preset model; wherein the preset model is used to assign a score to each risk prediction factor based on the strength of association.

[0084] Specifically, a preset model is constructed for assigning points to each risk prediction factor based on the strength of association. The construction of the preset model aims to use the strength of association of the determined risk prediction factors to assign points to each risk prediction factor, so that when the comprehensive information of new pregnant women in the early stage of pregnancy is obtained, the preset model can be used to quickly score and improve the efficiency of premature birth prediction. In some embodiments, the regression coefficient can be mapped to a score by setting different scoring criteria (such as linear or nonlinear) to form a comprehensive risk assessment tool. For example, the risk prediction factor can be divided into multiple levels according to the size of the regression coefficient or the odds ratio, and each level corresponds to a different score.

[0085] Optionally, the preset model is also used to determine the values ​​corresponding to risk prediction factors of different degrees based on the regression coefficient or odds ratio of each risk prediction factor.

[0086] Specifically, the risk prediction factor can be directly mapped to a corresponding score according to the size of the regression coefficient β or the OR value. For example, the natural logarithm of the OR value of the risk prediction factor is multiplied by 10 to obtain the assignment of each risk prediction factor. In this way, the larger the OR value, the higher the assigned score, indicating that the risk prediction factor contributes more to the risk of premature birth. Moreover, considering that the same type of risk prediction factors may affect premature birth to varying degrees, such as platelet counts of 100-300 are assigned (or scored) 0 points, platelet counts less than 100 are assigned 3 points, platelet counts greater than 300 and less than 400 are assigned 2 points, and platelet counts greater than or equal to 400 are assigned 3 points.

[0087] For ease of understanding, the following risk prediction factors include systolic blood pressure, diastolic blood pressure, history of intrauterine operation, urine sugar, platelet count, alanine aminotransferase, provincial household registration, high-risk pregnant women, autoimmune diseases, epilepsy, age, parity, gravidity, body mass index, history of miscarriage, premature birth, stillbirth, multiple pregnancy, assisted reproduction, hypertension, diabetes, thrombophilia, chronic kidney disease, and severe mental illness as examples, combined with the following Table 1:

[0088] Table 1

[0089]

[0090]

[0091] Among them, the entry column represents the risk prediction factor, and the assignment column represents the assignment corresponding to the risk prediction factor of different degrees.

[0092] Optionally, step 200 includes:

[0093] The values ​​corresponding to all risk prediction factors in the comprehensive information of pregnant women in early pregnancy that have been obtained are used as scores.

[0094] Specifically, based on the assignments corresponding to the determined risk prediction factors of different degrees, the assignments corresponding to all risk prediction factors in the comprehensive information of pregnant women in early pregnancy that have been obtained are used as scores, and the corresponding scoring results of pregnant women in early pregnancy are obtained in combination to predict the risk of premature birth of pregnant women.

[0095] Optionally, the sum of the values ​​corresponding to all risk prediction factors in the comprehensive information of the pregnant woman in the early pregnancy that has been obtained is used as the total score, and the risk of premature birth is determined by comparing the total score with the preset critical value. In some embodiments, a low risk of premature birth means that the risk of premature birth is low and premature birth is not likely to occur; a high risk of premature birth or a medium-high risk of premature birth means that the risk of premature birth is high and premature birth is relatively easy to occur.

[0096] Optionally, if the total score range needs to be set so that the total scores corresponding to all scoring results are within the total score range, the values ​​corresponding to all risk prediction factors can be normalized or standardized, and the values ​​corresponding to all risk prediction factors can be mapped to the total score range (such as 0-60 points). In this way, it is convenient to unify the standards, maintain the discrimination of the preset model, and facilitate clinical use. Exemplarily, in combination with Table 1, the maximum value of the sum of the values ​​corresponding to all risk prediction factors is 120 points (the same risk prediction factor only takes the value corresponding to one degree to avoid conflicts). If the total score range needs to be set to 0-60 points, the values ​​corresponding to the risk prediction factors can be divided by 2 based on Table 1 as new values.

[0097] Optionally, step 300 includes:

[0098] Determine whether the sum of the scores of all risk prediction factors is greater than a preset critical value;

[0099] If so, the risk of premature birth of the pregnant woman is predicted to be medium-high risk; if not, the risk of premature birth of the pregnant woman is predicted to be low risk.

[0100] Specifically, the preset critical value can be set by analyzing historical data (such as the first data) to ensure that it can statistically effectively distinguish between low-risk and medium- and high-risk pregnant women; the setting of this value can be subject to model training and verification to ensure accuracy.

[0101] According to the preset model, the sum of the scores (as a comprehensive score) of all risk prediction factors in the comprehensive information of pregnant women in early pregnancy that have been obtained is determined to determine whether the sum is greater than the preset critical value. When the sum of the scores of all risk prediction factors is greater than the preset critical value, the preset model predicts that the risk of premature birth of the pregnant woman is medium-high risk; conversely, if the sum of the scores is lower than the critical value, the prediction is low risk. In this way, on the one hand, by performing a simple comparison operation on the sum of the scores, a prediction of the risk of premature birth can be made quickly, which greatly improves the convenience of clinical application; on the other hand, it is convenient to quickly distinguish between low risk of premature birth and medium-high risk of premature birth, so that medical workers can quickly identify potential pregnant women with medium-high risk of premature birth according to the comprehensive score, and intervene in time.

[0102] In some embodiments, corresponding critical values ​​may be further set to further refine the medium and high risks, for example, a critical value between medium risk and high risk may be set, so as to provide a more refined premature birth intervention strategy based on the refined risk classification.

[0103] Optionally, after building the preset model, building the preset model for assigning a score to each risk prediction factor further includes:

[0104] acquiring second data including comprehensive information of different pregnant women who have completed childbirth;

[0105] Verifying the preset model based on the second data;

[0106] The preset model is optimized based on the verification results.

[0107] Specifically, second data including comprehensive information of different pregnant women who have completed childbirth, which is different from the above-mentioned first data, is obtained. For example, comprehensive information of pregnant women who have completed childbirth other than the pregnant women corresponding to the first data can be obtained to verify the preset model. The second data is different from the first data used to construct the preset model and is independent of each other to ensure that the verification process of the preset model does not depend on the training data (i.e., the first data), that is, to ensure that the verification process of the preset model is not affected by the training data, thereby improving the external validity and generalization ability of the preset model; and by using new data of pregnant women who have completed childbirth to verify the preset model, the performance of the preset model in actual clinical applications can be more accurately evaluated, ensuring that the preset model is not only effective for the first data, but also can maintain the accuracy and stability of prediction in different populations or other clinical environments. Moreover, by distinguishing between training data and verification data (i.e., the second data), overfitting of the preset model can be avoided, while also ensuring that an effective reference can be provided for the subsequent optimization of the preset model.

[0108] By using the second data to independently verify the preset model, the deficiencies in the preset model can be identified and reflected in the verification results, providing an effective basis for the subsequent optimization of the preset model. Based on the deficiencies of the preset model in the verification results, the preset model can be optimized in a targeted manner to improve the applicability and stability of the preset model in different populations and clinical environments, thereby further improving the external validity and generalization ability of the preset model.

[0109] Optionally, verifying the preset model based on the second data includes:

[0110] Based on the second data, the accuracy, sensitivity, specificity, positive predictive value and negative predictive value of the preset model are determined.

[0111] Specifically, based on the currently obtained preset model, the risk prediction factor part of the comprehensive information in the second data is input into the preset model to obtain the corresponding scoring results, and predictions are made according to the scoring results. The predicted results are compared with the actual situation (since the second data comes from pregnant women who have completed delivery, whether they have premature births is known and included in the second data). The accuracy, sensitivity, specificity, positive predictive value and negative predictive value of the preset model can be determined.

[0112] Among them, accuracy measures the proportion of correct predictions made by the preset model, that is, the proportion of the total number of correct predictions of premature birth by the preset model to all predictions. By evaluating the overall accuracy of the preset model, it can intuitively reflect the frequency of correct predictions made by the preset model in actual use, thereby providing a reliable basis for premature birth prediction; sensitivity, also known as recall rate, measures the ability of the preset model to identify real premature birth cases, that is, the proportion of true positive cases to all actual positive cases. A preset model with high sensitivity can effectively capture pregnant women at high risk of premature birth (or medium and high risk pregnant women), which is helpful for the implementation of early intervention and preventive measures, thereby reducing the risk of missed diagnosis in clinical practice; specificity measures the ability of the preset model to correctly identify non-preterm cases, that is, the proportion of true negative cases to all actual negative cases. A preset model with high specificity can reduce Misdiagnosis can be avoided to avoid unnecessary medical intervention for non-high-risk pregnant women and reduce the waste of medical resources; the positive predictive value reflects the probability of premature birth when the preset model predicts high risk (or medium-high risk), that is, the proportion of cases predicted to be positive that are actually positive. A high positive predictive value can improve the accuracy of clinical decision-making and ensure that the preset model has a high possibility of premature birth when it predicts high risk (or medium-high risk), thereby helping medical workers to carry out targeted intervention; the negative predictive value reflects the probability of premature birth not actually occurring when the preset model predicts low risk, that is, the proportion of cases predicted to be negative that are actually negative. A high negative predictive value means that the preset model is more reliable in excluding the risk of premature birth, which helps to reduce excessive intervention for low-risk pregnant women and enhance the confidence of medical workers in low-risk predictions.

[0113] For example, if in the second data, the number of samples correctly predicted as positive by the preset model is a, the number of samples incorrectly predicted as positive by the preset model is b, the number of samples incorrectly predicted as negative by the preset model is c, the number of samples correctly predicted as negative by the preset model is d, and the total number of samples predicted by the preset model is a+b+c+d, then:

[0114] Sensitivity = a / (a+c)*100%,

[0115] Specificity = d / (b+d)*100%,

[0116] Positive predictive value = a / (a+b)*100%,

[0117] Negative predictive value = d / (c+d)*100%,

[0118] Accuracy = (a+d) / (a+b+c+d)*100%.

[0119] Verifying the accuracy, sensitivity, specificity, positive predictive value and negative predictive value of the preset model based on the second data can comprehensively evaluate the performance of various aspects of the preset model and ensure its predictive ability and stability in different scenarios. It can not only improve the application effect of the preset model in actual clinical practice, but also provide a clear reference basis for the subsequent optimization of the preset model, thereby further enhancing the accuracy and reliability of the prediction and providing a scientific basis for the precise prevention and management of premature birth.

[0120] Optimizing the preset model based on the verification results includes:

[0121] When at least one of the accuracy, sensitivity, specificity, positive predictive value and negative predictive value does not meet the corresponding preset standard, the preset model is optimized.

[0122] Determine whether the preset model needs to be optimized according to the verification results. Specifically, accuracy, sensitivity, specificity, positive predictive value and negative predictive value are key indicators for measuring the performance of the preset model. Each indicator reflects the performance of the preset model in different aspects. When at least one of the accuracy, sensitivity, specificity, positive predictive value and negative predictive value does not meet the corresponding preset standard, such as a certain indicator is low, it means that it performs poorly in the corresponding prediction task, and the current preset model has a strong predictive ability for the first data, but performs poorly in the second data, and cannot effectively adapt to different populations or clinical environments, indicating that the current preset model has deficiencies in the corresponding aspects, such as the influence of specific risk factors that may not be captured, or the generalization ability when processing new data is insufficient. In this regard, the preset model can be optimized in a targeted (or comprehensive) manner according to the corresponding deficiencies in the verification results to improve the overall performance and clinical applicability of the preset model, and ensure that it can provide reliable prediction results in practical applications. In some embodiments, the method of optimizing the preset model includes redefining the risk prediction factor weight, adding new risk prediction factors, etc.

[0123] Optionally, when setting the preset critical value, the F1 value is maximized based on the principle, where the F1 value is the harmonic mean of the precision and the recall, and the specific relationship is as follows:

[0124] F1=2×Precision×Recall / (Precision+Recall),

[0125] Among them, precision (Precision) is the positive prediction value, Precision = a / (a+b)*100%; recall rate (Recall) is the sensitivity, Recall = a / (a+c)*100%.

[0126] In this way, the F1 value is the harmonic mean of precision and recall, which comprehensively considers the accuracy and recognition ability of the preset model. The higher the F1 value, the better the preset model performs in balancing precision and recall.

[0127] For example, assuming that for the second data, if the preset critical value is set to 12 points, the number of samples correctly predicted as positive by the preset model is a=50, the number of samples incorrectly predicted as positive by the preset model is b=10, the number of samples incorrectly predicted as negative by the preset model is c=5, and the number of samples correctly predicted as negative by the preset model is d=35, then:

[0128] Sensitivity = 50 / (50+5)×100%=90%,

[0129] Specificity = 35 / (10+35)×100% = 77.78%,

[0130] Positive predictive value = 50 / (50+10)×100% = 83.33%,

[0131] Negative predictive value = 35 / (5+35)×100% = 87.50%,

[0132] Accuracy = (50+35) / (50+10+5+35)×100%=85%,

[0133] F1≈2×0.8333×0.9 / (0.8333+0.9)≈86.96%,

[0134] That is, the preset critical value of 12 points makes the F1 value reach 86.96%, indicating that the preset model performs well in maintaining high precision and recall.

[0135] Combination Figure 5 As shown, another embodiment of the present invention provides a premature birth prediction device, comprising:

[0136] A scoring unit, used to score the acquired comprehensive information of the pregnant woman in the early stage of pregnancy based on a preset model; wherein the comprehensive information includes demographic information, clinical information and biochemical index information;

[0137] The risk prediction unit is used to predict the risk of premature birth of pregnant women based on the scoring results.

[0138] The premature birth prediction device of this embodiment is used to implement the premature birth prediction method described above. Its advantages over the prior art are the same as the advantages of the premature birth prediction method described above over the prior art, and will not be repeated here.

[0139] Optionally, the premature birth prediction device also includes a modeling unit, which is used to: obtain first data including comprehensive information of different pregnant women who have completed childbirth; determine risk prediction factors in the first data that are statistically associated with premature birth; and construct a preset model for assigning scores to each risk prediction factor.

[0140] Optionally, the modeling unit is specifically used to: use univariate logistic regression analysis to screen a first variable in the first data that is statistically associated with premature birth; use multivariate logistic regression analysis to screen the first variable and a preset second variable related to premature birth for a risk prediction factor that is statistically associated with premature birth.

[0141] Optionally, the modeling unit is specifically used to: determine the strength of association between each risk prediction factor and premature birth; and construct a preset model; wherein the preset model is used to assign a score to each risk prediction factor based on the strength of association.

[0142] Optionally, the risk prediction unit is specifically used to: determine whether the sum of the scores obtained by assigning all risk prediction factors is greater than a preset critical value; if so, predict that the risk of premature birth of the pregnant woman is medium-high risk; if not, predict that the risk of premature birth of the pregnant woman is low risk.

[0143] Optionally, the modeling unit is further used to: after constructing the preset model, obtain second data including comprehensive information of different pregnant women who have completed childbirth; verify the preset model based on the second data; and optimize the preset model based on the verification result.

[0144] Optionally, the modeling unit is also used to: determine the accuracy, sensitivity, specificity, positive predictive value and negative predictive value of the preset model based on the second data; when at least one of the accuracy, sensitivity, specificity, positive predictive value and negative predictive value does not meet the corresponding preset criteria, optimize the preset model.

[0145] Combination Figure 6 As shown, another embodiment of the present invention provides an electronic device, including a memory 601 and a processor 602;

[0146] Memory 601, used for storing computer programs;

[0147] The processor 602 is used to implement the above-mentioned method for predicting premature birth when executing the computer program.

[0148] In other words, an electronic device includes a memory 601 and a processor 602 coupled to the memory 601; the memory 601 is configured to store a computer program; and the processor 602 is configured to perform the following operations when executing the computer program:

[0149] Based on the preset model, the comprehensive information of the pregnant women in early pregnancy is scored; the comprehensive information includes demographic information, clinical information and biochemical index information;

[0150] Based on the scoring results, the risk of premature birth of pregnant women is predicted.

[0151] The electronic device of this embodiment can be used to implement the above-mentioned premature birth prediction method. Its advantages over the prior art are the same as the advantages of the above-mentioned premature birth prediction method over the prior art, which will not be repeated here.

[0152] Another embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is read and executed by a processor, the above-mentioned method for predicting premature birth is implemented.

[0153] In other words, a non-volatile computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to perform the following operations:

[0154] Based on the preset model, the comprehensive information of the pregnant women in early pregnancy is scored; the comprehensive information includes demographic information, clinical information and biochemical index information;

[0155] Based on the scoring results, the risk of premature birth of pregnant women is predicted.

[0156] The technical solution of the embodiment of the present invention, in essence or in other words, the part that contributes to the prior art, or the whole or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for making a computer device (which can be a personal computer, server, or network device, etc.) or a processor perform all or part of the steps of the method of the embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

[0157] The computer-readable storage medium of this embodiment can be used to implement the above-mentioned premature birth prediction method. Its advantages over the prior art are the same as the advantages of the above-mentioned premature birth prediction method over the prior art, and will not be repeated here.

[0158] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A method for predicting premature birth, characterized in that: include: Based on the preset model, the comprehensive information of the pregnant women in the early stage of pregnancy is scored; wherein the comprehensive information includes demographic information, clinical information and biochemical index information; The risk of premature birth of the pregnant woman is predicted based on the scoring result.

2. The method for predicting premature birth according to claim 1, wherein: Before scoring the acquired comprehensive information of the pregnant woman in the early stage of pregnancy based on the preset model, the method for predicting premature birth further includes: Acquiring first data including the comprehensive information of different pregnant women who have completed childbirth; determining risk predictors in the first data that are statistically associated with preterm birth; Construct the preset model for assigning a score to each of the risk prediction factors.

3. The method for predicting premature birth according to claim 2, wherein: Determining the risk prediction factor statistically associated with premature birth in the first data comprises: Using univariate logistic regression analysis, screening the first variable in the first data that is statistically associated with premature birth; Multivariate logistic regression analysis was used to screen the risk prediction factors that were statistically associated with premature birth among the first variable and the preset second variable related to premature birth.

4. The method for predicting premature birth according to claim 2, wherein: The construction of the preset model for assigning a score to each of the risk prediction factors includes: Determine the strength of association of each of the risk predictors with preterm birth; Construct the preset model; wherein the preset model is used to assign a score to each of the risk prediction factors based on the association strength.

5. The method for predicting premature birth according to claim 2, wherein: The predicting of the risk of premature birth of the pregnant woman according to the scoring result includes: Determining whether the sum of the scores of all the risk prediction factors is greater than a preset critical value; If so, the risk of premature birth of the pregnant woman is predicted to be medium-high risk; if not, the risk of premature birth of the pregnant woman is predicted to be low risk.

6. The method for predicting premature birth according to claim 4, wherein: After constructing the preset model, constructing the preset model for assigning a score to each of the risk prediction factors further includes: acquiring second data including the comprehensive information of different pregnant women who have completed childbirth; Verifying the preset model based on the second data; The preset model is optimized based on the verification result.

7. The method for predicting premature birth according to claim 6, wherein: The verifying the preset model based on the second data includes: Based on the second data, determining the accuracy, sensitivity, specificity, positive predictive value and negative predictive value of the preset model; The optimizing the preset model based on the verification result includes: When at least one of the accuracy, sensitivity, specificity, positive predictive value and negative predictive value does not meet the corresponding preset standard, the preset model is optimized.

8. A device for predicting premature birth, characterized in that: include: A scoring unit, used to score the acquired comprehensive information of the pregnant woman in the early stage of pregnancy based on a preset model; wherein the comprehensive information includes demographic information, clinical information and biochemical index information; The risk prediction unit is used to predict the risk of premature birth of the pregnant woman according to the scoring result.

9. An electronic device, characterized in that: including memory and processor; The memory is used to store computer programs; The processor is configured to implement the method for predicting premature birth according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is read and executed by a processor, the method for predicting premature birth according to any one of claims 1 to 7 is implemented.

Images