Delivery mode prediction method and device, computer equipment and storage medium

Abstract

The invention belongs to the field of medical treatment, and relates to a delivery mode prediction method, which comprises the following steps: obtaining puerpera data including computerized fetal heart monitoring data, ultrasonic examination data and pregnant woman electronic health record data; constructing a multi-modal deep learning model fusing a convolutional neural network and a bidirectional long-short-term memory network according to the puerpera data; and inputting puerpera data monitored in real time into the multi-modal deep learning model to obtain an optimal delivery mode. The invention further provides a delivery mode prediction device, computer equipment and a storage medium. According to the method, a doctor can conveniently plan a delivery scheme in advance, reasonably configure medical resources, reduce unnecessary cesarean delivery rate and reduce maternal and infant complications, the method is of great significance to improvement of maternal and infant health conditions, meanwhile, a new path is developed for further development and application of digital twinborn and related technologies in the field of gynaecology and obstetrics, and the method is worthy of popularization and application. And the field is promoted to stride towards intelligentization and precision.

Description

Technical Field

[0001] The present application relates to the field of medical technology, and in particular to a method, device, computer equipment and storage medium for predicting delivery mode. Background Art

[0002] Globally, maternal and neonatal health issues have always attracted much attention. Every year, a large number of maternal deaths (about 287,000 cases), neonatal deaths (2.4 million cases) and stillbirths (1.9 million cases) occur, of which a considerable proportion (up to 45%) are concentrated in the stages of labor, birth and immediate postpartum period, especially in low- and middle-income countries. As an important mode of delivery, although cesarean section has the potential to save lives, its current utilization rate is unreasonable. Because the underuse of cesarean section will lead to increased maternal and perinatal mortality and morbidity, while overuse is not only associated with no reduction in mortality, but may also cause complications in subsequent pregnancies.

[0003] The process of childbirth is affected by many factors, and traditional clinical methods face many challenges in predicting the mode of delivery. The process of childbirth is affected by the "3P model": the birth canal (maternal pelvis and soft tissue), the fetus (fetal size, fetal position, etc.), and labor force (uterine contractions and maternal efforts during childbirth). Effective clinical decision-making requires the integration of a large amount of data from prenatal examinations, admission to the delivery unit, and during labor. This highlights the need for dynamic and personalized assessment methods, which is the core of precision medicine - tailoring treatment plans for each patient.

[0004] Accurate prediction of delivery mode is essential for optimizing medical resources and ensuring the health of mothers and babies. Technological advances, such as big data analytics, cloud computing, virtual reality, the Internet of Things, and artificial intelligence, have played an important role in promoting this personalized medical approach. These technologies have promoted the development of digital twins (DTs), which are complex simulation technologies that integrate real-time data from different sources such as electronic health records (EHRs), medical devices, and genomic information. With large amounts of patient data collected from multiple sources, digital twins can provide personalized treatment plans and enable predictive analysis and preventive interventions through deep learning algorithms. Digital twin technology has been applied in medical fields such as asthma, diabetes, cancer, and cardiovascular disease, but its potential in delivery care has not yet been fully explored.

[0005] In this context, it is of great significance to develop an effective model for predicting mode of delivery, so as to improve the accuracy of prediction of mode of delivery, provide stronger support for clinical decision-making, and improve maternal and child health outcomes. Summary of the invention

[0006] The purpose of the embodiments of the present application is to propose a method, apparatus, computer device and storage medium for predicting the mode of delivery, the main purpose of which is to improve the efficiency and accuracy of predicting the mode of delivery.

[0007] In order to solve the above technical problems, the present application embodiment provides a method for predicting the mode of delivery, which adopts the following technical solution:

[0008] A method for predicting delivery mode comprises the following steps:

[0009] obtaining maternal data, the maternal data comprising computerized fetal heart rate monitoring data, ultrasound examination data, and maternal electronic health record data;

[0010] Build a multimodal deep learning model that integrates convolutional neural network and bidirectional long short-term memory network based on maternal data;

[0011] Real-time monitored maternal data is input into a multimodal deep learning model to obtain the optimal delivery method.

[0012] Furthermore, the step of obtaining maternal data, wherein the maternal data includes computerized fetal heart rate monitoring data, ultrasound examination data, and electronic health record data of pregnant women, specifically includes:

[0013] Use the digital twin-enabled delivery monitoring system to obtain maternal data in real time.

[0014] Furthermore, the step of constructing a multimodal deep learning model integrating a convolutional neural network and a bidirectional long short-term memory network based on the maternal data specifically includes:

[0015] Dividing the maternal data into a training set of a first set ratio and a validation set of a second set ratio;

[0016] Based on the TensorFlow library, a stochastic gradient descent optimization algorithm is adopted, and the cross entropy loss function is used to calculate the loss, and the training set is trained to obtain a multimodal deep learning model of a convolutional neural network and a bidirectional long short-term memory network;

[0017] A five-fold cross-validation set was used to validate and evaluate the multimodal deep learning model.

[0018] Furthermore, the step of inputting the real-time monitored maternal data into the multimodal deep learning model to obtain the optimal delivery method specifically includes:

[0019] The computerized fetal heart monitoring data was input into a convolutional neural network, a bidirectional long short-term memory network, and an attention mechanism for feature extraction;

[0020] Ultrasound examination data and maternal electronic health record data are digitized, normalized, and high-dimensional features are extracted through feature engineering to generate a set of structured feature vectors;

[0021] The extracted computerized fetal heart rate monitoring data features and the structured feature vector are integrated into a multimodal feature vector, which is then input into a fully connected layer. The feature mapping is completed through weight and bias calculation to generate an intermediate classification result. Finally, the output of the fully connected layer is converted into a probability value through a sigmoid activation function, and the mode of delivery is determined according to the probability value.

[0022] Furthermore, the step of determining the delivery mode according to the probability value specifically includes:

[0023] A probability value greater than 0.5 indicates a prediction of cesarean section, and a probability value less than or equal to 0.5 indicates a prediction of vaginal delivery.

[0024] Furthermore, in the multimodal deep learning model integrating the convolutional neural network and the bidirectional long short-term memory network:

[0025] The convolutional neural network is used to extract the spatial features of computerized fetal heart monitoring data, the bidirectional long short-term memory network is used to capture the time series features of computerized fetal heart monitoring data, and the ultrasound examination data and the electronic health record data of pregnant women are input into the multimodal deep learning model together with the computerized fetal heart monitoring data after being standardized for feature fusion.

[0026] Furthermore, the use of a five-fold cross-validation set to validate and evaluate the multimodal deep learning model specifically includes:

[0027] Cross-validation technology was used to optimize the model training process, and the model performance was evaluated and verified through accuracy, sensitivity, specificity, F1 score and Brier score indicators.

[0028] In order to solve the above technical problems, the embodiment of the present application further provides a delivery mode prediction device, which adopts the following technical solution:

[0029] A delivery mode prediction device, comprising:

[0030] An acquisition module, used for acquiring maternal data, wherein the maternal data includes computerized fetal heart rate monitoring data, ultrasound examination data and electronic health record data of pregnant women;

[0031] A building block for constructing a multimodal deep learning model that integrates convolutional neural networks and bidirectional long short-term memory networks based on maternal data;

[0032] The prediction module is used to input the real-time monitored maternal data into the multimodal deep learning model to obtain the optimal delivery method.

[0033] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0034] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the above-mentioned method for predicting delivery mode when executing the computer-readable instructions.

[0035] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0036] A computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for predicting the mode of delivery as described above.

[0037] Compared with the prior art, the embodiments of the present application have the following beneficial effects: the present application helps doctors plan delivery plans in advance, rationally allocate medical resources, reduce unnecessary cesarean section rates, and reduce maternal and infant complications, which is of great significance to improving the health of mothers and infants. At the same time, it also opens up new paths for the further development and application of digital twins and related technologies in the field of obstetrics and gynecology, and promotes this field to move towards intelligence and precision. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0040] Figure 2 A flowchart according to an embodiment of a method for predicting delivery mode of the present application;

[0041] Figure 3 is a structural schematic diagram of an embodiment of a delivery mode prediction device according to the present application;

[0042] Figure 4 It is based on the performance of the deep learning model in this application in predicting the mode of delivery;

[0043] Figure 5It is a graph of training and validation accuracy (Acc) according to the present application;

[0044] Figure 6 is a precision-recall curve diagram of the deep learning model according to the present application when using different data combinations;

[0045] Figure 7 is a schematic diagram of a digital twin-enabled labor monitoring system (DTLMS) according to the present application;

[0046] Figure 8 Example diagram of fetal heart rate monitoring (CTG) during labor according to the present application;

[0047] Fig. 9 It is a schematic diagram of the network architecture of the multimodal deep learning model;

[0048] Fig.10 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0050] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0051] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0052] like Figure 1As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0053] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0054] The terminal devices 101, 102, and 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer II I), MP4 (Moving Picture Experts Group Audio Layer IV), laptop computers, desktop computers, etc.

[0055] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0056] It should be noted that the delivery mode prediction method provided in the embodiment of the present application is generally executed by a server, and correspondingly, the delivery mode prediction device is generally arranged in a terminal device.

[0057] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0058] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for predicting a delivery mode according to the present application. The method for predicting a delivery mode comprises the following steps:

[0059] Step S201, obtaining maternal data, wherein the maternal data includes computerized fetal heart rate monitoring data (cCTG), ultrasound examination data (US) and maternal electronic health record data (EHR).

[0060] In this embodiment, the delivery monitoring system based on digital twin empowerment developed by the applicant is used to obtain maternal data in real time. The system has been used in the Department of Obstetrics and Gynecology of the First Affiliated Hospital of Jinan University. Figure 7 shown.

[0061] The delivery monitoring system based on digital twin empowerment covers physical space, DT data module, digital space and DT intelligent module. Physical space refers to the actual scene and related medical conditions of pregnant women and fetuses in the real environment. This includes the patient's physical condition, the hospital or clinic environment, and medical equipment and sensors used for real-time monitoring and data collection. The DT data module integrates dynamic multi-source data on maternal and child health, including electronic health records (EHR), ultrasound (US) images and physiological signals (such as fetal heart rate monitoring); the digital space realizes 3D visualization of maternal and child structures based on intrapartum ultrasound examination; the DT intelligent module is the core prediction model that can identify risk factors, predict adverse events or delivery methods, and assist clinical decision-making. The system can monitor multiple patients in multiple wards at the same time, and display electronic medical records, real-time electronic fetal monitoring signals, and 3D virtual delivery processes based on intrapartum ultrasound imaging for each patient.

[0062] In addition, the delivery monitoring system based on digital twin empowerment has the ability to process multiple data formats. In China's mixed use of electronic systems and paper medical records, it can automatically identify and extract information from paper reports to ensure effective use of data. This embodiment screened the collected data from cases in the Department of Obstetrics and Gynecology of the First Affiliated Hospital of Jinan University within a specific time period, excluding data on multiple pregnancies, planned cesarean sections, antenatal fetal deaths, major fetal abnormalities, and premature births, and finally determined the maternal data set for research. The data of 105 maternal births (76 vaginal births and 29 cesarean sections) were strictly screened for research, and the data collection process was exempted by the institutional ethics review committee.

[0063] like Figure 7 As shown in the figure, the delivery monitoring system enabled by digital twins can monitor multiple patients in multiple wards at the same time. For a specific patient, the system can display electronic medical records, real-time electronic fetal heart rate monitoring signals, and three-dimensional virtual delivery progress based on labor ultrasound imaging. By integrating this diverse information, the system expects to use artificial intelligence technology to assist clinical decision-making. Figure 8 The fetal heart rate (FHR) is shown to coincide with uterine contractions (UC), which are tracked by continuous graphic recording.

[0064] Step S202: construct a multimodal deep learning model integrating a convolutional neural network and a bidirectional long short-term memory network based on the maternal data.

[0065] In this embodiment, the training process of the multimodal deep learning model integrating the convolutional neural network and the bidirectional long short-term memory network is as follows:

[0066] First, the maternal data is divided into a training set of a first set ratio and a validation set of a second set ratio;

[0067] Then, based on the TensorFlow library, a stochastic gradient descent (momentum = 0.9) optimization algorithm was adopted, and the cross entropy loss function was used to calculate the loss. The training set was trained to obtain a multimodal deep learning model of a convolutional neural network and a bidirectional long short-term memory network.

[0068] In this embodiment, the training is performed for 120 cycles with an initial learning rate of 0.01. The learning rate is adjusted with a decay factor of 10 after 15 and 90 cycles, and L2 regularization (factor 0.0001) is applied to alleviate overfitting.

[0069] Among them, the convolutional neural network is used to extract the spatial features of computerized fetal heart monitoring data, the bidirectional long short-term memory network is used to capture the time series features of computerized fetal heart monitoring data, and the ultrasound examination data and maternal electronic health record data are standardized and input into the multimodal deep learning model together with the computerized fetal heart monitoring data for feature fusion.

[0070] Finally, a five-fold cross-validation set was used to verify and evaluate the multimodal deep learning model.

[0071] Among them, the cross-validation technology was used to optimize the model training process, and the model performance was evaluated and verified by evaluation indicators such as accuracy, sensitivity, specificity, area under the receiver operating characteristic curve (AUC), F1 score (F1Score) and Brier score (Brier Score). These indicators measure the ability of the model to predict the mode of delivery from different dimensions, such as accuracy reflects the overall prediction accuracy, AUC measures the ability of the model to distinguish different modes of delivery, F1 score balances the relationship between precision and recall, and Brier score evaluates the accuracy of probability prediction. The EHR data, US examination data and cCTG recording data of 105 parturients were analyzed for features, including maternal clinical characteristics (such as BMI, premature rupture of membranes, etc.), ultrasound characteristics (such as fetal position) and fetal heart monitoring characteristics (such as FHR changes, etc.), and the performance of the model under different data combinations was compared to comprehensively evaluate the model performance and demonstrate its advantages.

[0072] Fig. 9 The multimodal deep learning model network architecture is demonstrated.

[0073] Step S203, inputting the real-time monitored maternal data into a multimodal deep learning model to obtain the optimal delivery method.

[0074] The multimodal deep learning model is used to predict the mode of delivery. It adopts a multimodal deep learning framework, combines three input modalities of computerized fetal heart rate monitoring data (cCTG), ultrasound examination data (US) and maternal electronic health record data (EHR), processes the input features separately, and finally performs classification prediction through fusion.

[0075] The input data comes from three modalities. The computerized fetal heart rate monitoring data (cCTG) signal is a dynamic time series data that reflects the changes in fetal heart rate (FHR) and uterine contraction signal (UC), and is mainly used to describe the physiological dynamics during delivery. The electronic health record data (EHR) of pregnant women contains static health records of the mother, such as pregnancy history, physical data (age, height, weight, BMI, etc.), number of births, previous delivery methods, etc., which are closely related to the background factors of the delivery method. Ultrasound examination data (US) mainly provides ultrasound imaging characteristics of the fetus, such as fetal position, fetal head circumference, biparietal diameter (BPD) and other anatomical information. These indicators are particularly important when evaluating the delivery method.

[0076] First, the computerized fetal heart rate monitoring data (cCTG) is input into the convolutional neural network (CNN-BiLSTM) module for feature extraction. First, the local features of the signal are extracted through the convolutional layer (Conv) to capture the important patterns in the fetal heart rate or uterine contraction signal; batch normalization (Batch Normalization) improves the training stability, and the nonlinear ability is introduced through the ReLU activation function to enhance the model expression. Subsequently, the dimension is further reduced through the maximum pooling (Max Pooling) to retain the most important features and reduce the computational complexity. The features extracted by CNN then enter the bidirectional long short-term memory network (BiLSTM) module, which is used to capture the long-term time series dependency characteristics of computerized fetal heart rate monitoring data (cCTG). BiLSTM contains two LSTM layers, forward and backward, which can extract information from the two time directions of the sequence, so as to more comprehensively understand the dynamic characteristics in computerized fetal heart rate monitoring data (cCTG). In order to further improve the efficiency of using computerized fetal heart rate monitoring data (cCTG) features, the model introduces an attention mechanism after CNN-BiLSTM, dynamically adjusts the weights of features according to the contribution of different signal segments, and makes the model more focused on features that have a significant impact on prediction.

[0077] Ultrasound examination data (US) and maternal electronic health record data (EHR) are preprocessed and feature extracted through the feature engineering module. EHR and US data are digitized, normalized, and high-dimensional features are extracted through feature engineering to generate a set of structured feature vectors.

[0078] In the feature fusion module (Concat), the features of computerized fetal heart rate monitoring data (cCTG) are integrated with the features of maternal electronic health record data (EHR) and ultrasound examination data (US) into a unified multimodal feature vector. The fused high-dimensional feature vector enters the fully connected layer (Full Connection), and further features are mapped through a series of weight and bias calculations to generate intermediate classification results. Finally, the output of the fully connected layer is converted into a probability value through the sigmoid activation function. The model output result is the predicted probability of delivery mode, where a probability value greater than 0.5 indicates a cesarean section, and a probability value less than or equal to 0.5 indicates a vaginal delivery.

[0079] The architecture design of this multimodal deep learning model fully utilizes the complementarity of multimodal data: computerized fetal heart rate monitoring data (cCTG) signals are extracted through CNN-BiLSTM to extract time series features, maternal electronic health record data (EHR) and ultrasound examination data (US) are used to generate background features through feature engineering, the attention mechanism dynamically weights important features, and the feature fusion module integrates multimodal data, and finally achieves accurate prediction of delivery mode through fully connected layers and Sigma ID. This architecture combines time series analysis and static feature modeling capabilities, while optimizing the synergy of different modal data in the prediction process, providing a reliable support tool for clinical prediction of delivery mode.

[0080] In this embodiment, when only computerized fetal heart rate monitoring data (cCTG) is used, the AUC of the model is 0.4642±0.0320 (95% confidence interval: 0.2509-0.6775). After combining the maternal electronic health record data (EHR) and computerized fetal heart rate monitoring data (cCTG) signals, the classification performance is improved to AUC of 0.6447±0.1014 (95% confidence interval: 0.4400-0.8494). After integrating all data sources, namely: computerized fetal heart rate monitoring data (cCTG), ultrasound examination data (US) and maternal electronic health record data (EHR), the model performance reaches the best, with an AUC of 0.9710±0.0211 (95% confidence interval: 0.8992-1.0428). In summary, the fusion of multimodal data (especially the combination of cCTG, EHR and US) is the key to improving the performance of the prediction model. Figure 4As shown, the receiver operating characteristic curve (ROC curve) of the deep learning model when using different data combinations is shown.

[0081] This trend is also reflected in other evaluation indicators, such as accuracy (ACC), sensitivity (SEN), specificity (SPE), F1 score and Brier score, as shown in Table 1. The training and validation accuracy curves and precision-recall curves of the deep learning model are shown in the Supplementary Figure 6-7 .

[0082] Table 1 Performance of deep learning models using different data combinations

[0083]

[0084]

[0085] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0086] Further references Figure 3 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a device for predicting the mode of delivery, and the device embodiment is Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0087] like Figure 4 As shown, the delivery mode prediction device 300 described in this embodiment includes: an acquisition module 301, a construction module 302 and a prediction module 303. Among them:

[0088] The acquisition module 301 is used to acquire maternal data, which includes computerized fetal heart monitoring data, ultrasound examination data and electronic health record data of pregnant women.

[0089] The construction module 302 is used to construct a multimodal deep learning model that integrates a convolutional neural network and a bidirectional long short-term memory network based on maternal data.

[0090] The prediction module 303 is used to input the real-time monitored maternal data into the multimodal deep learning model to obtain the optimal delivery method.

[0091] This application helps doctors plan delivery plans in advance, rationally allocate medical resources, reduce unnecessary cesarean section rates, and reduce maternal and infant complications. It is of great significance to improve the health of mothers and infants. At the same time, it also opens up new paths for the further development and application of digital twins and related technologies in the field of obstetrics and gynecology, and promotes this field to move towards intelligence and precision.

[0092] To solve the above technical problems, the present application also provides a computer device. Fig.10 , Fig.10 This is a basic structural block diagram of the computer device in this embodiment.

[0093] The computer device 100 includes a memory 1001, a processor 1002, and a network interface 1003 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 100 having components 1001-1003, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), digital signal processors (DSP), embedded devices, etc.

[0094] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0095] The memory 1001 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 1001 can be an internal storage unit of the computer device 100, such as a hard disk or memory of the computer device 100. In other embodiments, the memory 1001 can also be an external storage device of the computer device 100, such as a plug-in hard disk equipped on the computer device 100, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 1001 can also include both the internal storage unit of the computer device 100 and its external storage device. In this embodiment, the memory 1001 is generally used to store the operating system and various application software installed on the computer device 100, such as computer-readable instructions of the method for predicting the delivery mode, etc. In addition, the memory 1001 can also be used to temporarily store various data that have been output or will be output.

[0096] The processor 1002 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 1002 is generally used to control the overall operation of the computer device 6. In this embodiment, the processor 1002 is used to run the computer-readable instructions stored in the memory 1001 or process data, such as running the computer-readable instructions of the delivery mode prediction method.

[0097] The network interface 1003 may include a wireless network interface or a wired network interface. The network interface 1003 is generally used to establish a communication connection between the computer device 100 and other electronic devices.

[0098] This application helps doctors plan delivery plans in advance, rationally allocate medical resources, reduce unnecessary cesarean section rates, and reduce maternal and infant complications. It is of great significance to improve the health of mothers and infants. At the same time, it also opens up new paths for the further development and application of digital twins and related technologies in the field of obstetrics and gynecology, and promotes this field to move towards intelligence and precision.

[0099] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the method for predicting the mode of delivery as described above.

[0100] This application helps doctors plan delivery plans in advance, rationally allocate medical resources, reduce unnecessary cesarean section rates, and reduce maternal and infant complications. It is of great significance to improve the health of mothers and infants. At the same time, it also opens up new paths for the further development and application of digital twins and related technologies in the field of obstetrics and gynecology, and promotes this field to move towards intelligence and precision.

[0101] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0102] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for predicting delivery mode, characterized in that: The steps include: obtaining maternal data, the maternal data comprising computerized fetal heart rate monitoring data, ultrasound examination data, and maternal electronic health record data; Build a multimodal deep learning model that integrates convolutional neural network and bidirectional long short-term memory network based on maternal data; Real-time monitored maternal data is input into a multimodal deep learning model to obtain the optimal delivery method.

2. The method for predicting delivery mode according to claim 1, characterized in that: The step of obtaining maternal data, wherein the maternal data includes computerized fetal heart rate monitoring data, ultrasound examination data, and electronic health record data of pregnant women, specifically includes: Use the digital twin-enabled delivery monitoring system to obtain maternal data in real time.

3. The method for predicting delivery mode according to claim 1, characterized in that: The step of constructing a multimodal deep learning model integrating a convolutional neural network and a bidirectional long short-term memory network according to the maternal data specifically includes: Dividing the maternal data into a training set of a first set ratio and a validation set of a second set ratio; Based on the TensorFlow library, a stochastic gradient descent optimization algorithm is adopted, and the cross entropy loss function is used to calculate the loss, and the training set is trained to obtain a multimodal deep learning model of a convolutional neural network and a bidirectional long short-term memory network; A five-fold cross-validation set was used to validate and evaluate the multimodal deep learning model.

4. The method for predicting delivery mode according to claim 1, characterized in that: The step of inputting the real-time monitored maternal data into the multimodal deep learning model to obtain the optimal delivery method specifically includes: The computerized fetal heart monitoring data was input into a convolutional neural network, a bidirectional long short-term memory network, and an attention mechanism for feature extraction; Ultrasound examination data and maternal electronic health record data are digitized, normalized, and high-dimensional features are extracted through feature engineering to generate a set of structured feature vectors; The extracted computerized fetal heart monitoring data features and the structured feature vector are integrated into a multimodal feature vector, the multimodal feature vector is input into the fully connected layer, the feature mapping is completed through weight and bias calculation, and the intermediate classification result is generated. Finally, the output of the fully connected layer is converted into a probability value through the Sigmoid activation function, and the mode of delivery is determined according to the probability value.

5. The method for predicting delivery mode according to claim 4, characterized in that: The step of determining the delivery mode according to the probability value specifically includes: A probability value greater than 0.5 indicates a prediction of cesarean section, and a probability value less than or equal to 0.5 indicates a prediction of vaginal delivery.

6. The method for predicting delivery mode according to claim 3, characterized in that: In the multimodal deep learning model integrating convolutional neural network and bidirectional long short-term memory network: The convolutional neural network is used to extract the spatial features of computerized fetal heart monitoring data, the bidirectional long short-term memory network is used to capture the time series features of computerized fetal heart monitoring data, and the ultrasound examination data and the electronic health record data of pregnant women are input into the multimodal deep learning model together with the computerized fetal heart monitoring data after being standardized for feature fusion.

7. The method for predicting delivery mode according to claim 3, characterized in that: The use of a five-fold cross-validation set to validate and evaluate the multimodal deep learning model specifically includes: Cross-validation technology was used to optimize the model training process, and the model performance was evaluated and verified using accuracy, sensitivity, specificity, F1 score, and Brier score indicators.

8. A device for predicting delivery mode, characterized in that: include: An acquisition module, used for acquiring maternal data, wherein the maternal data includes computerized fetal heart rate monitoring data, ultrasound examination data and electronic health record data of pregnant women; A building block for constructing a multimodal deep learning model that integrates convolutional neural networks and bidirectional long short-term memory networks based on maternal data; The prediction module is used to input the real-time monitored maternal data into the multimodal deep learning model to obtain the optimal delivery method.

9. A computer device, comprising a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the method for predicting a delivery mode according to any one of claims 1 to 6 when executing the computer-readable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for predicting the mode of delivery according to any one of claims 1 to 6.

Images