Delivery mode prediction method, electronic device and storage medium
By constructing a delivery method prediction model based on time-lapse ultrasound and clinical data, the problem of insufficient prediction accuracy in the prior art is solved, and more accurate delivery method evaluation and labor management are achieved, reducing the possibility of adverse delivery outcomes.
Patent Information
- Application Number
- CN202510586748.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-08
- Publication Date
- 2025-08-15
AI Technical Summary
The prediction methods of delivery mode in the prior art rely on single-dimensional clinical data, resulting in insufficient prediction accuracy and insufficient evaluation ability, and inability to effectively utilize time-productive ultrasound data.
A delivery method prediction model is constructed based on delivery ultrasound examination data and clinical data. Through training and optimization of various machine learning algorithms, key characteristic variables such as fetal head position and fetal parameters are integrated to generate delivery method prediction results.
It improves the accuracy of delivery method prediction, assists clinicians in optimizing labor decisions, reduces the rate of transit cesarean section, and improves maternal and infant delivery outcomes.
Smart Images
Figure CN120496804A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of medical technology, and in particular to a method for predicting delivery mode, an electronic device, and a storage medium. Background Art
[0002] In recent years, with the in-depth application of artificial intelligence technology in the medical field, machine learning-based prediction of delivery mode has gradually become an important auxiliary tool for obstetric clinical decision-making. Traditional prediction methods mostly rely on empirical clinical evaluation or single-dimensional clinical data analysis, and their prediction accuracy and dynamic evaluation capabilities have significant limitations. In the existing technology, there have been many explorations of prediction models based on clinical data. For example, the Beksac MS team developed a computer-assisted prediction system (Adana System) based on supervised artificial neural networks for predicting vaginal delivery and cesarean section. Similarly, scholars such as De Ramón Fernández A used support vector machines (SVM), multilayer perceptrons (MLP) and random forest (RF) algorithms to predict delivery mode. However, such methods only construct prediction models based on basic clinical information of mothers and newborns, and do not incorporate intrapartum ultrasound parameters. They have the defects of single data dimension and insufficient evaluation accuracy.
[0003] Accordingly, a new delivery mode prediction solution is needed in this field to solve the above problems. Summary of the Invention
[0004] In order to overcome the above-mentioned defects, the present application is proposed to solve or at least partially solve the technical problems of the existing technology such as single data dimension and insufficient evaluation accuracy.
[0005] In a first aspect, a method for predicting mode of delivery is provided, the method comprising: obtaining intrapartum examination data and clinical data of a pregnant woman who is about to give birth, the intrapartum examination data comprising intrapartum ultrasound examination data; inputting the intrapartum examination data and clinical data into a trained mode of delivery prediction model to obtain a mode of delivery prediction result of the pregnant woman who is about to give birth, the mode of delivery prediction result comprising the mode of delivery and the corresponding probability of delivery, wherein the mode of delivery prediction model is constructed based on the intrapartum ultrasound examination data and clinical data.
[0006] In one technical solution of the above-mentioned method for predicting the mode of delivery, the mode of delivery prediction model is constructed based on at least the following steps: obtaining a historical data set of multiple women who have given birth, the historical data set including historical intrapartum ultrasound examination data and historical clinical data, the historical clinical data including basic information of women who have given birth and information of newborns who have been born; constructing a prediction model to be trained based on the historical intrapartum ultrasound examination data and the historical clinical data; using a plurality of preset machine learning algorithms to train the prediction model to be trained, and obtaining a plurality of trained candidate prediction models respectively; and determining the optimal prediction model based on the plurality of trained candidate prediction models as the mode of delivery prediction model.
[0007] In a technical solution of the above-mentioned method for predicting the mode of delivery, the prediction model to be trained is constructed based on the historical intrapartum ultrasound examination data and the historical clinical data, including: normalizing the historical intrapartum ultrasound examination data and the historical clinical data, and dividing them into a training set and a test set; determining key feature variable data based on the training set; constructing the prediction model to be trained based on the key feature variable data; and / or, using a plurality of preset machine learning algorithms to train the prediction model to be trained, and obtaining a plurality of trained candidate prediction models respectively, including: using at least two preset machine learning algorithms among gradient boosting, random forest, logistic regression, Gaussian naive Bayes, support vector machine and K-nearest neighbor, and training the prediction model to be trained through ten-fold cross validation and grid search to optimize hyperparameters, and obtaining a plurality of trained candidate prediction models respectively.
[0008] In one technical solution of the above-mentioned method for predicting the mode of delivery, determining the key feature variable data based on the training set includes: using the LASSO regression algorithm to perform dimensionality reduction processing on the intrapartum ultrasound examination data and clinical data in the training set, and screening to obtain the key feature variable data, wherein the key feature variable data include the fetal head advancement angle, the fetal head-perineum distance, the fetal head-pubic symphysis distance and the fetal presenting position.
[0009] In a technical solution of the above-mentioned method for predicting the mode of delivery, determining the optimal prediction model based on multiple trained candidate prediction models includes: evaluating multiple trained candidate prediction models based on preset evaluation criteria to obtain the optimal prediction model, wherein the preset evaluation criteria include area under the curve, sensitivity, specificity, accuracy and F1 score.
[0010] In a technical solution of the above-mentioned method for predicting the mode of delivery, the method further includes: using the test set to verify the optimal prediction model, and determining the mode of delivery prediction model based on the verification result.
[0011] In one technical solution of the above-mentioned method for predicting the mode of delivery, the historical intrapartum ultrasound examination data include: fetal position, fetal head direction, fetal head progression angle, fetal head-pubic symphysis distance, fetal head-perineum distance and midline angle; the basic information of the delivered mother includes the size of cervical dilation, height of fetal presenting part, age, height, pre-pregnancy body mass index, weight gain during pregnancy, number of pregnancies, number of parities, history of adverse pregnancy and delivery, mode of delivery, whether analgesia was used during delivery, and delivery outcome; the information of the delivered newborn includes gestational age, birth weight, neonatal score, and whether there is a history of asphyxia rescue.
[0012] In a technical solution of the above-mentioned method for predicting the mode of delivery, the method further includes: using a clinical decision curve and / or a calibration curve to evaluate the clinical practicality of the mode of delivery prediction model; and / or converting the mode of delivery prediction model into a dynamic network nomogram model, and using a model visualization tool to visualize the dynamic network nomogram model, so as to display the corresponding mode of delivery prediction results on the interactive interface based on the data input on the interactive interface.
[0013] In a second aspect, an electronic device is provided, comprising at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program, and when the computer program is executed by the at least one processor, the method described in any one of the technical solutions of the above-mentioned method for predicting the mode of delivery is implemented.
[0014] In a third aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored in the computer-readable storage medium, wherein the program codes are suitable for being loaded and run by a processor to execute the method described in any one of the technical solutions of the above-mentioned method for predicting the mode of delivery.
[0015] The above one or more technical solutions of this application have at least one or more of the following beneficial effects:
[0016] The method for predicting the mode of delivery provided in the present application includes: obtaining the intrapartum examination data and clinical data of the expectant mother, wherein the intrapartum examination data includes intrapartum ultrasound examination data; inputting the intrapartum examination data and clinical data into a trained mode of delivery prediction model to obtain a mode of delivery prediction result of the expectant mother, wherein the mode of delivery prediction result includes the mode of delivery and the corresponding probability of delivery, wherein the mode of delivery prediction model is constructed based on the intrapartum ultrasound examination data and clinical data. The present application integrates the intrapartum ultrasound examination data and clinical data to construct a mode of delivery prediction model, which can accurately evaluate the appropriate mode of delivery based on the intrapartum examination data and clinical data of the expectant mother, thereby helping to improve the clinical utilization rate of the intrapartum ultrasound examination data, and at the same time assisting doctors in optimizing labor process decisions, reducing the rate of transfer to cesarean section, thereby reducing the possibility of adverse outcomes during delivery, and improving maternal and infant delivery outcomes. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The disclosure of this application will become more easily understood with reference to the accompanying drawings. Those skilled in the art will readily appreciate that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this application. Among them:
[0018] Figure 1 This is a flow chart of the main steps of a method for predicting delivery mode according to one embodiment of the present application;
[0019] Figure 2 is a schematic diagram of an interactive interface according to an embodiment of the present application;
[0020] Figure 3 This is a schematic diagram of a detailed process for constructing a delivery mode prediction model according to an embodiment of the present application;
[0021] Figure 4 It is a schematic diagram of the main structure of an electronic device according to an embodiment of the present application.
[0022] Reference numerals:
[0023] 11: Memory; 12: Processor. DETAILED DESCRIPTION
[0024] Some embodiments of the present application are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and are not intended to limit the scope of protection of the present application.
[0025] In the description of this application, "module" and "processor" may include hardware, software, or a combination of both. A module may include hardware circuitry, various suitable sensors, communication ports, and memory. It may also include software components, such as program code, or a combination of software and hardware. A processor may be a central processing unit (CPU), a microprocessor, an image processor, a digital signal processor, or any other suitable processor. A processor has data and / or signal processing capabilities. A processor may be implemented in software, hardware, or a combination of both. Computer-readable storage media include any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" refers to all possible combinations of A and B, such as only A, only B, or both A and B. The terms "at least one of A or B" or "at least one of A and B" have similar meanings to "A and / or B" and may include only A, only B, or both A and B. The singular forms "a" and "the" may also include the plural forms.
[0026] The relevant user personal information that may be involved in the various embodiments of this application is strictly in accordance with the requirements of laws and regulations, following the principles of legality, legitimacy and necessity, and based on the reasonable purposes of business scenarios, to process the personal information that users actively provide during the use of products / services or generated due to the use of products / services, as well as the personal information obtained with the user's authorization.
[0027] This application attaches great importance to the security of user personal information and has taken reasonable and feasible security protection measures that comply with industry standards to protect user information and prevent personal information from being accessed, disclosed, used, modified, damaged or lost without authorization.
[0028] In recent years, with the in-depth application of artificial intelligence technology in the medical field, machine learning-based delivery mode prediction has gradually become an important auxiliary tool for obstetric clinical decision-making. Traditional prediction methods often rely on empirical clinical evaluation or single-dimensional clinical data analysis, which has significant limitations in predictive accuracy and dynamic assessment capabilities.
[0029] To this end, the method for predicting the mode of delivery provided in this application includes: obtaining the intrapartum examination data and clinical data of the expectant mother, wherein the intrapartum examination data includes intrapartum ultrasound examination data; inputting the intrapartum examination data and clinical data into a trained mode of delivery prediction model to obtain a mode of delivery prediction result of the expectant mother, wherein the mode of delivery prediction result includes the mode of delivery and the corresponding probability of delivery, wherein the mode of delivery prediction model is constructed based on the intrapartum ultrasound examination data and clinical data. This application integrates the intrapartum ultrasound examination data and clinical data to construct a mode of delivery prediction model, which can accurately evaluate the appropriate mode of delivery based on the intrapartum examination data and clinical data of the expectant mother, thereby helping to improve the clinical utilization rate of intrapartum ultrasound examination data, while assisting doctors in optimizing labor process decisions, reducing the rate of transfer to cesarean section, thereby reducing the possibility of adverse outcomes during delivery, and improving maternal and infant delivery outcomes.
[0030] See attached Figure 1 , Figure 1 FIG. 1 is a flow chart of the main steps of a method for predicting the mode of delivery according to an embodiment of the present application. Figure 1 As shown, the method for predicting the mode of delivery in the embodiment of the present application mainly includes the following steps S101 and S102.
[0031] Step S101: Acquire intrapartum examination data and clinical data of a pregnant woman, wherein the intrapartum examination data includes intrapartum ultrasound examination data.
[0032] In this embodiment, intrapartum examination data and clinical data of the expectant mother are obtained. The intrapartum examination data includes intrapartum ultrasound examination data of the first stage of labor, and the clinical data can be the demographic data, gestational age information, laboratory examination data, etc. of the expectant mother.
[0033] Step S102: Input the intrapartum examination data and clinical data into a trained delivery mode prediction model to obtain a delivery mode prediction result for the expectant mother, wherein the delivery mode prediction result includes the delivery mode and the corresponding delivery probability, wherein the delivery mode prediction model is constructed based on the intrapartum ultrasound examination data and clinical data.
[0034] Based on the method described in the above steps S101 and S102, the present application constructs a delivery mode prediction model based on intrapartum ultrasound examination data and clinical data, and in clinical application, the intrapartum examination data and clinical data of the expectant mother are input into the delivery mode prediction model to obtain the best delivery mode recommendation and the corresponding delivery probability for the expectant mother, thereby improving the utilization rate of intrapartum ultrasound examination data in clinical practice, providing clinicians with a new auxiliary and reference method to optimize labor management decisions, helping to reduce the rate of conversion to cesarean section, thereby avoiding the occurrence of adverse events during delivery, and further improving the delivery outcomes of mothers and babies.
[0035] Before implementing the delivery mode prediction method, this application requires pre-constructing a delivery mode prediction model.
[0036] In one embodiment, the delivery mode prediction model is constructed based on at least the following steps: obtaining a historical data set of multiple women who have given birth, the historical data set including historical delivery ultrasound examination data and historical clinical data, the historical clinical data including basic information of women who have given birth and information of newborns who have been born; constructing a prediction model to be trained based on the historical delivery ultrasound examination data and the historical clinical data; using a plurality of preset machine learning algorithms to train the prediction model to be trained, and obtaining a plurality of trained candidate prediction models respectively; and determining an optimal prediction model based on the plurality of trained candidate prediction models as the delivery mode prediction model.
[0037] Specifically, first, delivery data of multiple women who have given birth are collected, and the historical data set includes historical delivery ultrasound examination data and historical clinical data, among which the historical clinical data include basic information of women who have given birth and information of newborns who have been delivered, and the historical delivery ultrasound examination data can be delivery ultrasound examination data of the first stage of labor; and based on the historical data set, a prediction model for predicting the mode of delivery is constructed; and then a variety of preset machine learning algorithms are used to train the prediction model to be trained, and a plurality of trained candidate prediction models are obtained, so as to screen out the optimal prediction model from the multiple trained candidate prediction models and use it as the delivery mode prediction model.
[0038] This application uses different machine learning algorithms to integrate ultrasound examination data and clinical data during the first stage of labor to construct a delivery mode prediction model. The model has good predictive ability for the delivery mode of full-term primiparas or multiparas without contraindications to vaginal delivery. It can provide clinicians with more appropriate clinical references before delivery, reduce the possibility of adverse outcomes during delivery, and improve maternal and infant delivery outcomes.
[0039] In one embodiment, the prediction model to be trained is constructed based on the historical intrapartum ultrasound examination data and the historical clinical data, including: normalizing the historical intrapartum ultrasound examination data and the historical clinical data, and dividing them into a training set and a test set; determining key feature variable data based on the training set; constructing the prediction model to be trained based on the key feature variable data; and / or, using a plurality of preset machine learning algorithms to train the prediction model to be trained, and obtaining a plurality of trained candidate prediction models respectively, including: using at least two preset machine learning algorithms among gradient boosting, random forest, logistic regression, Gaussian naive Bayes, support vector machine and K-nearest neighbor, and training the prediction model to be trained through ten-fold cross validation and grid search to optimize hyperparameters, and obtaining a plurality of trained candidate prediction models respectively.
[0040] Specifically, historical intrapartum ultrasound and clinical data are normalized. Normalization can be determined based on data distribution and task requirements. For example, for ultrasound images, pixel normalization can be used to adjust pixel values to a preset range, while clinical data can be normalized using methods such as one-hot encoding or target encoding. Then, based on a preset ratio, the normalized historical intrapartum ultrasound and clinical data are divided into training and test sets.
[0041] Based on the training set, the key characteristic variable data that best predicts the mode of delivery is screened out, and the prediction model to be trained is constructed based on the key characteristic variable data; then, the preset machine learning algorithms, such as gradient boosting (XGBoost), random forest (RF), logistic regression (LR), Gaussian naive Bayes (GNB), support vector machine (SVM) and K-nearest neighbor (KNN), are used to optimize the hyperparameters using ten-fold cross validation and grid search, and the prediction models to be trained are trained respectively, thereby obtaining multiple trained candidate prediction models.
[0042] In one embodiment, determining the key feature variable data based on the training set includes: using the LASSO regression algorithm to perform dimensionality reduction processing on the intrapartum ultrasound examination data and clinical data in the training set, and screening to obtain the key feature variable data, wherein the key feature variable data include the fetal head advancement angle, the fetal head-perineum distance, the fetal head-pubic symphysis distance and the fetal presenting position.
[0043] Specifically, the intrapartum ultrasound examination data include: fetal position, fetal head direction, angle of progression (AOP), head-symphysis distance (HSD), head-perineum distance (HPD), midline angle (MLA), occiput-spine angle (OSA), direction of the maximum diameter of the fetal head and other related indicators.
[0044] Clinical data include: maternal cervical dilation size, fetal presenting part height, age, height, pre-pregnancy body mass index (BMI), weight gain during pregnancy, number of pregnancies, number of parities, adverse pregnancy and delivery history, mode of delivery (natural labor / induced labor), whether analgesia was used during delivery, delivery outcome (natural vaginal delivery / surgical delivery) and other basic information as well as neonatal data, including gestational age, birth weight, neonatal 1′ Apgar score, neonatal 5′ Apgar score, neonatal 10′ Apgar score, and whether there was a history of asphyxia rescue.
[0045] The LASSO regression algorithm (Least Absolute Shrinkage and Selection Operator) was used to reduce the dimensionality of the original features in the intrapartum ultrasound examination data and clinical data. Key feature variables, including the angle of progression (AOP), head-perineal distance (HPD), head-symphysis distance (HSD), and fetal presenting position (MLA), were screened out. A prediction model was constructed using the key feature variable data.
[0046] In one embodiment, determining the optimal prediction model based on multiple trained candidate prediction models includes: evaluating multiple trained candidate prediction models based on preset evaluation criteria to obtain the optimal prediction model, wherein the preset evaluation criteria include area under the curve, sensitivity, specificity, accuracy and F1 score.
[0047] Specifically, the performance of multiple trained candidate prediction models is comprehensively evaluated using the area under the curve (AUC), sensitivity, specificity, accuracy, and F1 score to determine the optimal prediction model with the best performance. The area under the curve (AUC), the area under the ROC curve, is used to evaluate the discrimination of different models. Sensitivity analyzes the sensitivity of the output to changes in input parameters or variables. Specificity evaluates the model's ability to identify negative samples (true negatives), that is, whether the model can correctly avoid misclassifying samples that are actually "negative" as "positive." Accuracy is the most intuitive performance metric, representing the proportion of correctly classified samples among all samples. The F1 score is used to measure model performance. The F1 score comprehensively considers the model's precision and recall, and is the harmonic mean of the two. The F1 value ranges from 0 to 1, with higher values indicating better model performance.
[0048] In this example, the area under the curve (AUC) was used as the optimal evaluation indicator, and the model performance was comprehensively evaluated in combination with indicators such as sensitivity, specificity, accuracy, and F1 score.
[0049] In one embodiment, the method further includes: using the test set to verify the optimal prediction model, and determining a delivery mode prediction model based on the verification result.
[0050] Specifically, the data in the test set is input into the optimal prediction model, and the predicted results and probability of the delivery mode are output. The predicted results are compared with the actual delivery results to verify the prediction performance of the optimal prediction model. If it meets the preset standards, the optimal prediction model will be used as the delivery mode prediction model.
[0051] In one embodiment, the historical intrapartum ultrasound examination data include: fetal position, fetal head direction, fetal head progression angle, fetal head-pubic symphysis distance, fetal head-perineum distance and midline angle; the basic information of the delivered mother includes the size of cervical dilation, fetal presenting part, age, height, pre-pregnancy body mass index, weight gain during pregnancy, number of pregnancies, number of parities, adverse pregnancy and delivery history, mode of delivery, whether analgesia was used during delivery, and delivery outcome; the information of the delivered newborn includes gestational age, birth weight, neonatal score, and whether there is a history of asphyxia rescue.
[0052] Specifically, historical intrapartum ultrasound examination data include: fetal position, fetal head direction, angle of head progression (AOP), head-pubic symphysis distance (HSD), head-perineum distance (HPD), midline angle (MLA) and other related indicators.
[0053] The basic information of the mother who has delivered includes the size of the cervical dilation, the height of the fetal presenting part, age, height, pre-pregnancy body mass index (BMI), weight gain during pregnancy, number of pregnancies, number of parities, adverse pregnancy and delivery history, mode of delivery (natural labor / induced labor), whether analgesia was used during delivery, and delivery outcome (natural vaginal delivery / surgical delivery).
[0054] The information of delivered newborns includes gestational age, birth weight, newborn 1′ Apgar score, newborn 5′ Apgar score, newborn 10′ Apgar score, whether there is a history of asphyxia rescue, etc.
[0055] In one embodiment, the method further includes: using a clinical decision curve and / or a calibration curve to evaluate the clinical practicality of the delivery mode prediction model; and / or converting the delivery mode prediction model into a dynamic network nomogram model, and using a model visualization tool to visualize the dynamic network nomogram model, so as to display the corresponding delivery mode prediction results on the interactive interface based on the data input on the interactive interface.
[0056] Specifically, the clinical decision curve analysis (DCA) is a method used to evaluate the value of a prediction model in clinical decision-making. It evaluates the potential benefits of the model in actual clinical applications by calculating the "net benefit" (NB) under different threshold probabilities. This embodiment uses the clinical decision curve to evaluate the clinical practicality of the delivery mode prediction model.
[0057] A dynamic network nomogram visualization model is used, that is, a nomogram model website is constructed, and data is input in real time in the interactive interface of the website. The interactive interface feedbacks the mode of delivery as vaginal delivery or cesarean section, and the corresponding probability of delivery. The real-time interactive visualization of the mode of delivery prediction model is achieved through the dynamic network nomogram, which facilitates the promotion and application of the mode of delivery prediction model in clinical practice.
[0058] The above steps S101 and S102 are further described below.
[0059] With respect to step S101 , intrapartum examination data and clinical data of the expectant mother are obtained, wherein the intrapartum examination data includes intrapartum ultrasound examination data.
[0060] Specifically, the intrapartum ultrasound examination data and clinical data of pregnant women are obtained. The intrapartum ultrasound examination data can include data such as fetal position, fetal head direction, fetal head advancement angle, fetal head-pubic symphysis distance, fetal head-perineum distance and midline angle; the clinical data can include data such as the size of cervical dilation, height of fetal presenting part, age, height, pre-pregnancy body mass index, weight gain during pregnancy, number of pregnancies, number of parities, and adverse pregnancy and delivery history.
[0061] For step S102, the intrapartum examination data and clinical data are input into a trained delivery mode prediction model to obtain a delivery mode prediction result for the expectant mother, wherein the delivery mode prediction result includes the delivery mode and the corresponding delivery probability, wherein the delivery mode prediction model is constructed based on the intrapartum ultrasound examination data and clinical data.
[0062] Specifically, the intrapartum ultrasound examination data and clinical data of the expectant mother are input into the delivery mode prediction model, and the delivery mode prediction model outputs the delivery mode prediction result of the expectant mother, which includes vaginal delivery or cesarean section, as well as the corresponding delivery probability.
[0063] In a specific embodiment, the intrapartum ultrasound examination data of a pregnant woman are angle of progression (AOP): 125°, head-perineum distance (HPD): 29.4mm, head-symphysis pubis distance (HSD): 21.5mm, clinical data include cervical dilation 7cm, fetal presenting position +2, fetal parameters are single cephalic presentation at 38 weeks of gestation, and ultrasound estimated weight is 3200g.
[0064] The four data of fetal presenting position, angle of progression (AOP), head-perineum distance (HPD) and head-symphysis pubis distance (HSD) are input into the delivery mode prediction model through the interactive interface. Figure 2 As shown; the delivery mode prediction model outputs the delivery mode prediction result that the delivery mode prediction result of the expectant mother is normal delivery, and the delivery probability is 97%, that is, the success rate of vaginal trial delivery is about 97% (the probability calculated by the prediction model is the probability of surgical delivery), and vaginal trial delivery is encouraged.
[0065] In this way, this application uses a machine learning algorithm to construct a delivery mode prediction model based on intrapartum ultrasound examination data and clinical data, and uses a dynamic nomogram model to quickly and accurately obtain delivery mode prediction results, which helps to manage the labor process more accurately and improve the application value of intrapartum ultrasound examination data in delivery management. At the same time, it provides auxiliary and reference methods for clinicians to help clinicians judge the delivery mode more accurately, solves the judgment differences caused by traditional methods that rely solely on the doctor's clinical experience, effectively reduces the incidence of unplanned cesarean sections, and improves maternal and child prognosis.
[0066] See attached Figure 3 , Figure 3 FIG. 1 is a flowchart of detailed steps for constructing a delivery mode prediction model according to an embodiment of the present application; FIG. Figure 3 As shown, the construction of the delivery mode prediction model in this embodiment includes the following steps:
[0067] Step S201: Start.
[0068] Step S202: Collect the mother's historical intrapartum ultrasound examination data and historical clinical data.
[0069] Specifically, historical intrapartum ultrasound examination data and historical clinical data of 259 primiparas or multiparas without contraindications to vaginal trial of labor were collected.
[0070] Step S203: exclude data of 10 maternal cases with missing information exceeding 30%.
[0071] Step S204: normalize the remaining maternal data and divide them into a training set and a test set in proportion.
[0072] Specifically, after excluding the data of 10 maternal cases with more than 30% missing individual information, we finally obtained the data of 249 maternal cases. After normalization, the 249 maternal data were randomly divided into training set and test set in an 8:2 ratio. The training set included 199 maternal data and the test set included 50 maternal data.
[0073] Step S205: Based on the training set, construct multiple prediction models to be trained.
[0074] Specifically, based on the training set, the LASSO regression method is applied to perform dimensionality reduction on the data in the training set, the key feature variable data is screened, and the prediction model to be trained is constructed based on the key feature variable data.
[0075] Step S206: using a preset machine learning algorithm, the prediction model to be trained is trained by optimizing hyperparameters through ten-fold cross validation and grid search, and a plurality of trained candidate prediction models are obtained.
[0076] Specifically, the preset machine learning algorithms include gradient boosting algorithm (XGBoost), random forest (RF), logistic regression (LR), Gaussian naive Bayes (GNB), support vector machine (SVM) and K-nearest neighbor (KNN).
[0077] Step S207: Based on a preset evaluation standard, multiple trained candidate prediction models are evaluated to obtain an optimal prediction model.
[0078] Specifically, the area under the curve (AUC) was used as the main evaluation criterion, combined with sensitivity, specificity, accuracy and F1 score for comprehensive evaluation to screen out the optimal prediction model.
[0079] Step S208: Using the test set, verify the optimal prediction model, and determine the delivery mode prediction model based on the verification results.
[0080] Step S209: draw a calibration curve and a clinical decision curve.
[0081] Specifically, a calibration curve was drawn to evaluate the accuracy of the delivery mode prediction model, and a clinical decision curve was drawn to evaluate the clinical practicality of the delivery mode prediction model.
[0082] Step S210: Dynamic nomogram visualization of the delivery mode prediction model.
[0083] Specifically, by using a dynamic network nomogram visualization model, that is, building a nomogram model website, real-time operations on the website can predict the mode of delivery and the corresponding probability of vaginal delivery or cesarean section in real time, which is convenient for subsequent promotion and application in clinical practice.
[0084] Step S211: End.
[0085] It should be pointed out that although the various steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effect of the present application, different steps do not have to be performed in such an order. They can be performed simultaneously (in parallel) or in other orders. These adjusted solutions are equivalent to the technical solutions described in this application, and therefore will also fall within the scope of protection of this application.
[0086] It will be understood by those skilled in the art that all or part of the processes in the method for implementing the above embodiment of the present application can also be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the above-mentioned various method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file or some intermediate form. The computer-readable storage medium may include: any entity or device, medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium, etc. that can carry the computer program code.
[0087] Another aspect of the present application provides an electronic device.
[0088] In an embodiment of an electronic device according to the present application, the electronic device may include at least one processor; and a memory in communication with the at least one processor; wherein the memory stores a computer program, and when the computer program is executed by the at least one processor, the method described in any of the above embodiments is implemented. The electronic device described in the present application may include a driving device, a smart car, a robot, and the like. See the attached Figure 4 , Figure 4 exemplarily shows that the memory 11 and the processor 12 are communicatively connected via a bus.
[0089] The electronic device described in this application may be, but is not limited to, a mobile phone, a tablet computer, a desktop computer, a laptop computer, a handheld computer, a notebook computer, a vehicle-mounted device, an ultra-mobile personal computer (UMPC), a netbook, a personal digital assistant (PDA), an augmented reality (AR) or virtual reality (VR) device, etc., and the embodiments of this application are not limited to this.
[0090] Another aspect of the present application provides a computer-readable storage medium.
[0091] In an embodiment of a computer-readable storage medium according to the present application, the computer-readable storage medium can be configured to store a program for executing the method for predicting the mode of delivery of the above-mentioned method embodiment. The program can be loaded and executed by a processor to implement the above-mentioned method for predicting the mode of delivery. For ease of explanation, only the parts related to the embodiment of the present application are shown. For specific technical details not disclosed, please refer to the method section of the embodiment of the present application. The computer-readable storage medium can be a storage device formed by various electronic devices. Optionally, the computer-readable storage medium in the embodiment of the present application is a non-transitory computer-readable storage medium.
[0092] Thus far, the technical solution of the present application has been described in conjunction with an embodiment shown in the accompanying drawings. However, it is readily understood by those skilled in the art that the scope of protection of the present application is obviously not limited to these specific embodiments. Without departing from the principles of the present application, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present application.
Claims
1. A method for predicting delivery mode, characterized in that: The method comprises: Acquiring intrapartum examination data and clinical data of a pregnant woman, wherein the intrapartum examination data includes intrapartum ultrasound examination data; The intrapartum examination data and clinical data are input into a trained delivery mode prediction model to obtain a delivery mode prediction result for the expectant mother, wherein the delivery mode prediction result includes the delivery mode and the corresponding delivery probability, wherein the delivery mode prediction model is constructed based on the intrapartum ultrasound examination data and clinical data.
2. The method for predicting delivery mode according to claim 1, wherein: The delivery mode prediction model is constructed based on at least the following steps: Acquire a plurality of historical data sets of women who have given birth, wherein the historical data sets include historical intrapartum ultrasound examination data and historical clinical data, wherein the historical clinical data includes basic information of women who have given birth and information of newborns who have been born; constructing a prediction model to be trained based on the historical intrapartum ultrasound examination data and the historical clinical data; Using a plurality of preset machine learning algorithms to train the prediction model to be trained, and obtaining a plurality of trained candidate prediction models respectively; Based on the multiple trained candidate prediction models, the optimal prediction model is determined as the delivery mode prediction model.
3. The method for predicting delivery mode according to claim 2, wherein: The step of constructing a prediction model to be trained based on the historical intrapartum ultrasound examination data and the historical clinical data comprises: Performing normalization processing based on the historical intrapartum ultrasound examination data and the historical clinical data, and dividing the data into a training set and a test set; Based on the training set, determining key feature variable data; Constructing the prediction model to be trained based on the key feature variable data; And / or, using multiple preset machine learning algorithms to train the prediction model to be trained, and obtaining multiple trained candidate prediction models respectively includes: At least two preset machine learning algorithms among gradient boosting, random forest, logistic regression, Gaussian naive Bayes, support vector machine and K-nearest neighbor are used to train the prediction model to be trained through ten-fold cross validation and grid search to optimize hyperparameters, and obtain multiple trained candidate prediction models.
4. The method for predicting delivery mode according to claim 3, wherein: Determining key feature variable data based on the training set includes: The LASSO regression algorithm was used to perform dimensionality reduction processing on the intrapartum ultrasound examination data and clinical data in the training set, and key feature variable data were screened to obtain key feature variable data, wherein the key feature variable data included the fetal head advancement angle, fetal head-perineum distance, fetal head-pubic symphysis distance, and fetal presenting position.
5. The method for predicting delivery mode according to claim 3, wherein: The determining of the optimal prediction model based on the plurality of trained candidate prediction models includes: Based on preset evaluation criteria, multiple trained candidate prediction models are evaluated to obtain the optimal prediction model, wherein the preset evaluation criteria include area under the curve, sensitivity, specificity, accuracy and F1 score.
6. The method for predicting delivery mode according to claim 5, wherein: The method further comprises: The optimal prediction model is verified using the test set, and a delivery mode prediction model is determined based on the verification results.
7. The method for predicting delivery mode according to claim 2, wherein: The historical intrapartum ultrasound examination data include: fetal position, fetal head direction, fetal head advancement angle, fetal head-pubic symphysis distance, fetal head-perineum distance and midline angle; The basic information of the mother who has delivered includes the size of cervical dilation, height of fetal presentation, age, height, body mass index before pregnancy, weight gain during pregnancy, number of gravidities, number of parities, adverse pregnancy and delivery history, mode of delivery, whether analgesia was used during delivery, and delivery outcome; The information of the delivered newborn includes gestational age, birth weight, neonatal score, and whether there is a history of asphyxia rescue.
8. The method for predicting delivery mode according to claim 1, wherein: The method further comprises: Using clinical decision curves and / or calibration curves to evaluate the clinical utility of the delivery mode prediction model; and / or, The delivery mode prediction model is converted into a dynamic network nomogram model, and the dynamic network nomogram model is visualized using a model visualization tool to display the corresponding delivery mode prediction results on the interactive interface based on the data input on the interactive interface.
9. An electronic device comprising at least one processor and at least one memory, wherein the memory is adapted to store a plurality of program codes, wherein: The program code is suitable for being loaded and run by the processor to execute the delivery mode prediction method according to any one of claims 1 to 8.
10. A computer-readable storage medium storing a plurality of program codes, characterized in that: The program code is suitable for being loaded and run by a processor to execute the method for predicting the delivery mode according to any one of claims 1 to 8.